Sei sulla pagina 1di 172

U n i v e r s i t y

o f
A n t w e r p
F a c u l t y
o f
M e d i c i n e
Department of General Practice

Antibiotics for coughing in general practice


exploring, describing and optimising prescribing
Antibiotica voor hoestklachten in de huisartspraktijk
exploreren, beschrijven en optimaliseren van het voorschrijven

Dissertation for the degree of doctor in Medical Science


at the University of Antwerp - Universitaire Instelling Antwerpen
to be defended by

Samuel COENEN

Paul Van Royen


Joke Denekens

Antwerp, 2003

Exam commission
Supervisors:
Prof. Dr. J. Denekens, University of Antwerp
Prof. Dr. P. Van Royen, University of Antwerp
Doctoral commission:
Prof. Dr. M. De Broe (chair) , University of Antwerp
Prof. Dr. W. De Backer, University of Antwerp
Prof. Dr. A. Meheus, University of Antwerp
External members:
Prof. Dr. G-J. Dinant, University of Maastricht
Prof. Dr. P. Little, University of Southampton

Antibiotics for coughing in general practice: exploring, describing and


optimising prescribing.
Samuel Coenen
ISBN 90-5728-039-6
Department of General Practice
University of Antwerp
Universiteitsplein 1
BE 2610 Antwerp
Belgium
E-mail: samuel.coenen@ua.ac.be

Table of contents
Chapter I

Introduction

Chapter II

A qualitative decision analysis


Fam Pract 2000;17:380-5
Huisarts Nu 2001;30:390-7
A questionnaire study to quantify and condense the
reasons for prescribing
BMC Family Practice 2002;3:16 (10p)
Huisarts Nu 2003;32:180-9
GPs perception of patients requests determines
prescription behaviour

13

A clinical practice guideline


Huisarts Nu 2002;31;391-411
Cluster randomised controlled trial of a tailored
professional intervention to optimise prescribing

73

Chapter III

Chapter VI

Chapter V
Chapter VI

29

51

79

Chapter VII

Patients views on respiratory symptoms and antibiotics 109

Chapter VIII

General discussion

121

Summary
Samenvatting

133
145

Curriculum vitae

159

List of Publications

161

Dankwoord

167

I
Antibiotics for coughing in general practice:
General introduction
The research activities described in this dissertation were performed in general
practice and with general practitioners (GPs). For a good understanding this
introduction (Chapter I of this dissertation) will consider some of the
particular characteristics and the interrelation of this setting, coughing, and
antibiotics.

General practice
GPs can play a key role in the organisation of health care. Dealing with a large
variety of reasons for encounter they decide whether the patient needs health
care or not, and in the former case whether this can be provided in primary care
or not. This so-called filter-function , also referred to as gatekeeping ,
therefore can result in a rational use of resources.1 2 On the other hand the
means to perform as a GP are limited. After all primary care is characterised by
low technology and low cost. Furthermore, for the many problems encountered
in primary care the evidence base is limited. As a result it is unavoidable that
GPs have to deal with uncertainty in daily practice. Establishing an operational
diagnosis for every day complaints often is difficult, as is deciding which
patient will benefit from a specific treatment.

Chapter I
Guidelines form a potentially valuable bridge between evidence-based
medicine and clinical practice. Adherence to guidelines can improve quality of
care, reduce inappropriate variations in practice and improve cost
effectiveness.3 However studies in general practice indicate poor compliance
with guidelines or limited effects upon patient outcomes.4 If guidelines are to
address variability, we need to understand how and why variation occurs in
general practice and to develop reliable ways of identifying unacceptable
variations. These topics will be covered in this thesis.

Coughing
Complaints about coughing are a good example of an every day reason for
encounter which the GP has to manage as a gatekeeper, ensuring rational use of
resources, despite limited means and lacking evidence.
In general practice, medical decisions are prompted most often by complaints
about coughing, especially for respiratory tract infections (RTIs). The
frequency of coughing as a reason for encounter at the start of a new illness
episode was found to be 5.3% by Jan De Maeseneer in 1989 for Belgium,5 and
stable at 5.4% by Lambrechts et al. for over a decade in the Netherlands.6 For
Belgium recent figures linking patients complaints to diagnoses are not
available. In the Netherlands however per 1000 patients visiting their GP in the
last year 168.9 consult with coughing as (one of) the most prominent
complaint(s). Irrespective of their age more than three quarters of the final
diagnoses are RTIs.6 Using the International Classification of Primary Care,7
upper RTIs, including laryngitis/tracheitis and sinusitis, represent nearly half of
the final diagnoses. Acute bronchitis is the final diagnosis in a quarter of the
consultations. Proven influenza and pneumonia each only account for about
2% of the final diagnoses. These final diagnoses are part of the top 10 of most
frequent final diagnoses for new illness episodes with coughing as reason for
encounter. Asthma can be found in this top 10 too. Notwithstanding the
difficulties to distinguish between respiratory infections and not infectious
causes of coughing such as for example asthma, the validity of the diagnostic
criteria used to classify respiratory infections is questionable. It is hard to
distinguish upper from lower RTIs, and history and clinical examination do not
allow to differentiate between acute bronchitis and pneumonia.8 Therefore in
this thesis all research activities depart from complaints about coughing and

General introduction
not from doubtful diagnoses. Furthermore, technical investigations such as
blood or sputum analysis, or X-ray examination are not considered suitable
and/or feasible in general practice for the differential diagnoses above, nor to
discriminate between viral and bacterial infections.9 10 Finally, and this is more
important, there is no evidence allowing GPs to identify patients with acute
bronchitis or acute cough who might benefit from antibiotics.11 12

Antibiotics
Antibiotic use for coughing in general practice, finally, is a good example of
misuse of available resources, resulting in sub-optimal patient care.
The discovery of antibiotics (penicillin) by Alexander Fleming in 1928
triggered enormous progress in the field of medicine. However, as early as
1944 Fleming observed that some bacteria were able to destroy penicillin, and
he warned that the misuse of antibiotics could lead to selection of resistant
bacteria, making antibiotics loose their effectiveness in the treatment of lifethreatening infections. This warning was lost in the first flush of the discovery
of increasing numbers of antibiotics and the success of these medicines,
especially in the treatment of RTIs.
A decade ago the increase of bacterial resistance to antimicrobial agents was
declared a crisis.13 At the same time the Alexander Project was established in
Europe and the USA to examine the antimicrobial susceptibility of communityacquired lower RTIs bacterial pathogens. Up to now, the causative agent of
the most frequent life-threatening bacterial infection, Streptococcus
pneumoniae - or pneumococcus for short -, has become even less sensitive to
penicillin and other antibiotics world-wide.14 15 According to the most recent
data from the Belgian Streptococcus pneumoniae Reference Lab (Jan
Verhaegen, Leuven) more than 30% of pneumococci isolated are resistant to
erytromycin and tetracycline, whereas more than 5% show full penicillin
resistance (Figure 1).
The increase in bacterial resistance is associated with the increased use of
antibiotics, both in animals and in humans. In addition, both agricultural use
(50%) and human use (50%) of antibiotics is inappropriate.16 In the case of
humans, 80% is used in the community, this means outside the hospital,
especially in general practice and for RTI. Evidence for the misuse of

Chapter I
antibiotics can be found in the variation in outpatient antibiotic consumption in
Europe (Figure 2).17 Especially the pronounced difference between
neighbouring countries such as Belgium and the Netherlands, 26.7 vs. 8.9
Daily Defined Doses (DDD) per 1000 inhabitants per day, is unlikely to be
caused by differences in frequency of bacterial infections. Furthermore,
although differences in health care system, cultural and social factors, and
maybe in physicians and patients attitudes to antibiotics might explain these
international differences, even in the Netherlands with the lowest antibiotic
consumption in Europe it was estimated that antibiotics for respiratory
complaints were most probably indicated for only half of those prescribed
one.18 Furthermore, the variation in the use of different kinds of antibiotics
between countries most likely is associated with inappropriate prescribing too.
Data from the National Sickness and Invalidity Insurance Institute (NSIII):
Health Care Service, in Belgium are very similar compared to the data
presented by Cars and colleagues. These data only relate to reimbursed drugs
prescribed in ambulant care, that is for patients outside the hospital, and
delivered by the pharmacist. In cost and in volume (DDD) over 80% of these
drugs are prescribed by GPs. The same goes for antibiotics of which about 20
DDDs per 1000 inhabitants per day are prescribed by certified GPs in Belgium.
Since 1997 prescribing of both antibiotics in general and different classes of
antibiotics has been rather stable according to the currently available NSIII
data (Figure 3). These data also provide more details about combinations of
antibiotics compared to the data presented by Cars and colleagues. The most
striking finding in this regard is the volume of penicillins, including betalactamase inhibitor (Anatomic Therapeutic Chemical Classification (ATC):
J01CR), c.q. the combination of amoxicillin and clavulanic acid, prescribed in
Belgium: 4.1 DDDs per 1000 inhabitants per day, or over 20% of all antibiotics
prescribed by GPs in Belgium. The variation in the use of different kinds of
antibiotics most likely is associated with inappropriate prescribing within one
country as well. After all the share different kinds of antibiotics have in the
total antibiotic prescribing (in DDD) shows a large variation between certified
GPs in Belgium (Figure 4). And there is no evidence that this variation can be
explained entirely by differences in frequency or aetiology of the encountered
infectious diseases, nor by differences in antimicrobial resistance of the
causative microbes.

General introduction

Figure 1 Penicillin, tetracycline and erythromycin susceptibility of Belgian isolates of


Streptococcus pneumonia: percentage of non susceptible isolates in the period 1986(1)
2002.

40%
35%

Erythromycin
Tetracycline

30%

Penicillin

25%
20%
15%
10%
5%

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

1990

1989

1988

1987

1986

0%

(1) Data from the Belgian Streptococcus pneumoniae Reference Lab (Jan Verhaegen, Leuven). For 2002
36,2 % of the isolates are resistant to Erythromycin (Macrolide), 30,9 % to Tetracycline, 15,2 % to
Penicillin, and not presented in the figure 0,5 % to Ofloxacine (Quinolone).

Chapter I

Figure 2 The volumes of non-hospital antibiotic sales (Anatomic Therapeutic


Chemical (ATC) J01) in 1997 in 15 member states of the European Union, expressed
(1)
as defined daily doses (DDD) per 1000 inhabitants per day.

40

Defined daily dose per 1000 inhabitants per day

Others (2)
Macrolides and lincosamides J01F

35

Quinolones J01M
Trimethroprim J01EA

30

Tetracyclines J01A
Cephalosporins J01D
Penicillinase resistant penicillins J01CF

25

Narrow-spectrum penicillins J01CE


Broad-spectrum penicillins J01CA

20

15

10

The Netherlands

Denmark

Sweden

Germany

Austria

UK

Ireland

FInland

Greece

Italy

Luxembourg

Belgium

Portugal

Spain

France

(1) Data from Cars O, Mlstad S, Melander S. Variation in antibiotic use in the European Union. Lancet
2001; 357:1851-1853. The WHO ATC-DDD-classification of 1997 is used.
(2) Includes amphenicols (J01BA), aminoglycosides (J01GB), sulphonamides (J01E), some combinations
(J01CR) within beta-lactam antibacterials, penicillins (J01C) and the entire J01X (glycopeptides (J01XA),
polymyxines (J01XB), steroid antimicrobials (J01XC) and other antimicrobials (J01XX)), but not
nitrofuran derivates (J01XE).

General introduction

Defined daily dose per 1000 inhabitants per day

Figure 3 The volumes of dispensed antibiotics (Anatomic Therapeutic Chemical


(ATC) J01) in the period 1997-2001 in Belgium, prescribed by certified general
(1)
practitioners, expressed as defined daily doses (DDD) per 1000 inhabitants per day.

Others (2)

25

Macrolides and lincosamides J01F


Quinolones J01M

20

Sulphonamides and trimethoprim J01E

15

Tetracyclines J01A
Cephalosporins J01D

10

Combinations of penicillins, incl. betalactamase inhibitors J01CR


Penicillinase-resistant penicillins J01CF

Narrow-spectrum penicillins J01CE


Broad-spectrum penicillins J01CA

0
1997

1998

1999

2000

2001

(1) Data from the National Sickness and Invalidity Insurance Institute (NSIII): Health Care Service, in
Belgium. The WHO ATC-DDD classification of January 2003 is used.
(2) Includes amphenicols (J01BA), aminoglycosides (J01GB), glycopeptides (J01XA), polymyxines (J01XB),
steroid antimicrobials (J01XC) and other antimicrobials (J01XX), but not nitrofuran derivatives (J01XE)

Chapter I

Figure 4 The share four kinds of antibiotics have in the total antibiotic prescribing (in
DDD) of all GPs in Belgium:(1) box and whiskers plot.(2)

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Amoxicillin and Macrolides/Total Quinolones/Total
clavulanic
antibiotics
antibiotics
acid/Total
antibiotics

Amoxicillin
syrup/Total
antibiotics

(1) Data from the Conjoint Sickness Funds Data Agency. The WHO ATC-DDD-classification of 2000 is
used.
(2) Bar represents 50th percentile, box 25th and 75th percentile, and whiskers 1st and 99th percentile
respectively.

General introduction
It has been shown that community antibiotic sales mirror patterns of antibiotic
resistance in Western Europe.19 Others have observed that changes in antibiotic
use may be followed by changes in antibiotic resistance.20 21 Consequently, the
best way of preserving the effectiveness of antibiotics is to use them more
appropriately, i.e. in cases where patients will actually benefit. The alternative
of continuing to develop new antibiotics will only solve the problem of
antibiotic resistance if the principles of judicious use of antibiotics are
implemented at the same time. The growth in resistance is progressing faster
than the development of new antibiotics.

Aims
This dissertation aims to contribute to the development of effective strategies
for a more appropriate use of antibiotics. Since coughing is one of the most
common complaints in general practice, and most antibiotics are prescribed by
GPs and for this condition, the appropriate use of antibiotics to treat coughing
is a key area of action in order to tackle the resistance problem. Consequently,
by describing, exploring and optimising the prescription of antibiotics for
coughing we can safeguard a major development in the field of medicine, i.e.
the use of antibiotics in the treatment of life-threatening infections.
In the first part of the dissertation we explore the way GPs manage patients
who consult them with complaints about coughing to contribute to the
necessary understanding of the complex prescribing decision. In a qualitative
study (Chapter II), we explored the diagnostic and therapeutic decisions by
Flemish general practitioners regarding adult patients who consult them
complaining about a cough as well as the determinants of their decisions by
means of focus groups. A questionnaire (Chapter III) was used to quantify
and condense the determinants of the antibiotic prescribing decision generated
in the focus group study. In order to validate the focus group and questionnaire
findings and to obtain a valid estimate of the effect of these determinants on
GPs prescription of antibiotics we recorded their management of patients
consulting with acute cough as one of the most prominent complaints
(Chapter IV).
The second part of this dissertation provides recommendations for changing
current practices and specifically for optimising the use of antibiotics for acute

Chapter I
cough in general practice. According to a standardized methodology defined by
the Scientific College of Flemish General Practitioners (WVVH) the guideline
for good clinical practice: acute cough was developed (Chapter V). In a
prospective, cluster-randomised, controlled, before-and-after study we
assessed the effect of an educational intervention, implementing a clinical
practice guideline by means of academic detailing (Chapter VI).
The third part addresses the perspective of the patient, the end consumer of all
medical treatment, in this case antibiotics. For a better understanding of
patients'views about frequent respiratory symptoms, cough, earache and sore
throat, and antibiotic treatment, we performed a postal questionnaire study with
patients in Belgium, in the UK and in the Netherlands. For this dissertation the
emphasis will be laid on the results for Belgium (Chapter VII).
Finally, we summarise and discuss the results of this dissertation, and conclude
with opportunities for further research (Chapter VIII).

References
1. Dixon J, Holland P, Mays N. Primary care: core values. Developing primary
care: gatekeeping, commissioning, and managed care. BMJ 1998;317:125-8.
2. Coenen S, Avonts D, Van Royen P, Denekens J. Chronic obstructive
pulmonary disease: don t forget the gatekeeper [letter]. Lancet 1998;352:649.
3. Grimshaw J, Russell I. Effect of clinical guidelines on medical practice: a
systematic review of rigorous evaluations. Lancet 1993;342:1317-22.
4. Worrall G, Chaulk P, Freake D. The effects of clinical practice guidelines on
patient outcomes in primary care: a systematic review. CMAJ 1997;156:170512.
5. De Maeseneer J. Huisartsgeneeskunde: een verkenning [General practice: an
exploration]. Proefschrift [Dissertation] Rijksuniversiteit Gent, 1989.
6. Okkes I, Oskam S, Lamberts H. Van klacht naar diagnose [From complaint
to diagnosis]. Bussum: Coutinho, 1998.

10

General introduction
7. Lamberts H, Wood M. ICPC. International classification of Primary Care.
Oxford: Oxford University Press, 1987.
8. Metlay J, Kapoor W, Fine M. Does This Patient Have Community-Acquired
Pneumonia? Diagnosing Pneumonia by History and Physical Examination.
JAMA 1997;278:1440-5.
9. Jonsson J, Sigurdsson J, Kristinsson K, Gudnadttir M, Magnusson S. Acute
bronchitis in adults. How close do we come to its aetiology in general
practice? Scand J Prim Health Care 1997;15:156-60.
10. Johnson P, Macfarlane J, Humphreys H. How is sputum microbiology used
in general practice? Resp Med 1996;90:87-8.
11. Fahey T, Stocks N, Thomas T. Quantitative systematic review of
randomised controlled trials comparing antibiotic with placebo for acute cough
in adults. BMJ 1998;316:906-10.
12. Smucny J, Becker L, Glazier R, McIsaac W. Are Antibiotics Effective
Treatment for Acute Bronchitis? A Meta-Analysis. J Fam Pract 1998;47:45360.
13. Neu H. Crisis of antibiotic resistance. Science 1992;257:1064-73.
14. Felmingham D, Feldman C, Hryniewicz W, Klugman K, Kohno S, Low D,
et al. Surveillance of resistance in bacteria causing community-acquired
respiratory tract infections. Clin Microbiol Inf 2002;8:12-42.
15. Schito GC, Debbia EA, Marchese A. The evolving threat of antibiotic
resistance in Europe: new data from the Alexander Project. J Antimicrob
Chemother 2000;46:3-9.
16. Wise R, Hart T, Cars O, Streulens M, Helmuth R, Huovinen P, Sprenger
M. Antimicrobial resistance. Is a major threat to public health [editorial]. BMJ
1998;317:609-10.
17. Cars O, Mlstad S, Melander S. Variation in antibiotic use in the European
Union. Lancet 2001;357:1851-3.

11

Chapter I
18. De Melker R. Efficacy of antibiotics in frequently occurring airway
infections in Family Practice. Ned Tijdschr Geneeskd 1998;142:452-6.
19. Bronzwaer S, Cars O, Buchholz U, Molstad S, Goettsch W, Veldhuijzen I,
et al. A European study on the relationship between antimicrobial use and
antimicrobial resistance. Emerg Infect Dis 2002;8:278-8.
20. Seppala H, Klaukka T, Vuopio-Varkila J, Muotiala A, Helenius H, Lager
K, et al. The Effect of Changes in the Consumption of Macrolide Antibiotics
on Erythromycin Resistance in Group A Streptococci in Finland. NEJM
1997;337:441-6.
21. Molstad S, Cars O. Major change in the use of antibiotics following a
national programme: Swedish Strategic Programme for the Rational Use of
Antimicrobial Agents and Surveillance of Resistance (STRAMA). Scand J
Infect Dis 1999;31:191-5.

12

II
Antibiotics for coughing in general practice:
A qualitative decision analysis
Introduction
Medical decision analysis regarding respiratory tract infections (RTIs) mainly
focuses on the differential diagnosis between viral and bacterial RTIs, between
upper and lower RTIs and between different clinical syndromes such as
bronchitis and pneumonia.1 2 And establishing the appropriate treatment is
directly linked to the diagnosis of RTIs.3 But researchers such as Melbye et al.
have recognised that there is no single yardstick for these diagnoses.4 Thus, it
may be questioned whether general practitioners (GPs) can work this way in
actual practice.
More than 80% of the excessive use of antibiotics in RTIs is caused by GPs
prescription behaviour.5 This does not only result in a medicalising effect, but
also has enormous financial implications and contributes significantly to an
increase in bacterial resistance.6
In general practice complaints about coughing very often constitute the starting
point of medical decision-making regarding RTIs. Therefore, this study
examines the diagnostic decisions of GPs based on complaints about coughing
by adult patients. The determinants of these decisions were derived from an
exploratory, descriptive focus group investigation.

Published in Fam Pract 2000;17:380-5.

13

Chapter II
Methods
In order to generate a discussion of the topic among GPs on the basis of their
own experience, a focus group investigation was set up.7 This research method
generates the rich details of complex experiences and the reasoning behind
actions, beliefs, perceptions and attitudes of people.
By means of snowball sampling8 139 GPs from the Antwerp area in Belgium
were recruited. They were contacted by telephone in order to find out whether
they would be willing to participate. Eventually, a purposeful sample of sixty
GPs were invited to the university campus in order to collect the determinants
from a broad spectrum of GPs, in terms of sex, general practice experience as
well as university of graduation, instead of just from a representative sample.
The number of focus groups was determined by content saturation, i.e. the
moment at which answers to the questions provided did not contain any new
elements. For the purpose of flexibility, six focus groups were put forth, but in
the end four groups proved sufficient. For each focus group, a maximum of
twelve GPs received a written invitation in order to ensure the participation of
four to eight GPs.
In order not to hinder the free exchange of views, the composition of the
groups was homogeneous in terms of sex as well as general practice experience
(Table 1). Each group also had a moderator (IH) and an observer (SC). Each
semi-structured discussion was guided by the moderator, lasted 90 minutes and
centred on the following questions:
1.You are consulted by one of your adult patients who complains about
coughing. Which diagnoses come to mind?
2. How do you differentiate between the various possibilities in your patient?
3. You suspect an infection of the respiratory tract. Do you differentiate in any
way? Which distinctions do you make?
4. How do you differentiate between the various possibilities in your patient?

14

A qualitative decision analysis


Non-verbal information regarding the discussion was logged by the observer
on a specially designed scoring sheet9 10 and afterwards conferred with the
moderator. She, a psychologist, was familiar with the principles of general
practice medical decision-making.11
The recordings were transcribed and subsequently analysed by two researchers,
i.e. SC and EV, independently and in accordance with the principles of
qualitative content analysis.12 All codes with their labels and definitions and
the transcriptions were imported into QSR NUD*IST software for
computerised analysis.13
Interpretation of the coded texts enabled a classification of the codes and the
establishment of relationships between the various codes or categories. This
resulted in hypotheses on GPs decision-making regarding complaints about
coughing.

Results
In March 1998 four focus groups met in which 24 of the 48 invited GPs
participated. The only excuse given for not participating was lack of time. The
participants did not differ from those who declined in terms of age, sex and
university of graduation (Table 1).
The interpretation of the coded texts enabled a classification of the codes. The
categories to which the most relevant codes for determinants were assigned,
are: epidemiology, prior knowledge, history, clinical examination, doctor- and
patient-related factors.
The process of generating hypotheses based on the relationship between these
categories and the GPs decisions is illustrated by representative answers to the
four key questions (Table 2).

15

Chapter II

Possible diagnoses for adult patients with complaints about coughing


Independent of prior knowledge of the patient the first answer and diagnoses
that comes to mind was RTI in all focus groups (Table 2: text 1). Other
possible hypotheses which emerged in all focus groups were obstructive lung
diseases (COPD and asthma), allergy, (gastro-oesophageal) reflux, cardiac
decompensation, pulmonary oedema, smoking and other irritations, side effects
of ACE-I and tumours. Other hypotheses mentioned in three of the focus
groups were psychogenic cough, pulmonary embolism and foreign body. These
diagnoses were made based on prior knowledge of the patient (Table 2:2).

Decision-making in complaints about coughing


In cases where these GPs had reason to suspect hypotheses other than
infectious coughing, they had been able to check these hypotheses, but the
questions asked were far from routine questions. The GPs believed that
infectious coughing had the highest probability (Table 2:1) and, as a result,
they ask routine questions to confirm this hypothesis (Table 2:3). GPs stated
that they were better able to confirm certain hypotheses than to rule them out
explicitly (Table 2:4).

Possible diagnoses and decision-making regarding complaints about


coughing and suspected RTI
In all focus groups, GPs made a distinction between upper and lower RTIs,
viral and bacterial RTIs, chronic and acute RTIs, and between different clinical
syndromes (e.g. bronchitis and pneumonia).
In the analysis of the different texts coded as determinants, a distinction was
made between those determining the probability of a particular condition, e.g.
clinical signs and symptoms and those only influencing the action thresholds,
i.e. Pauker s Testing and Test-treatment thresholds.14
Determining the probability. GPs tried to make a distinction between the
different types of RTIs on the basis of medical history and clinical
examination, e.g. sputum colour (Table 2:5). Furthermore it was argued that
this can only lead to a suspected distinction between viral and bacterial RTIs,
while it is difficult to distinguish between bronchitis and pneumonia (Table
2:6). The value or feasibility of technical investigations such as blood analysis,
sputum examination (Table 2:5) or medical imaging was questioned.

16

A qualitative decision analysis

Table 1 Composition of focus groups according to number, age, sex, university of


graduation and the number of single- and duo practices of participants, compared to
the dropouts for age, sex, university of graduation.

Invited
GPs

Participants

Dropouts

Groups

Total
1

Total

Total

Number

48

24

24

Mean age

35

45

29

28

40

35

35

(limits)

(26-63)

(39-63) (26-32) (26-29) (37-44) (26-63) (26-50)

Man/Woman

24/24

7/0

0/6

4/0

0/7

11/13

13/11

UA(1)/other

29/19

3/4

4/2

1/3

6/1

14/10

15/9

8/24

(2)

(2)

Solo-practices

3/7

1/6

0/4

4/7

Duo-practices

3/7(2)

3/6(2,3)

4/4(3)

3/7

(2)

13/24(2,3)

(1) University of Antwerp


(2) 3 General practitioners working in a group practice participated.
(3) 5 General Practitioners in Professional Training participated (GPPT): a duo-practice then means a solopractice with a GPPT.

17

Chapter II
Table 2 Representative extracts.
1. A: The first thing that comes to mind is a common, simple, ordinary infection of the upper or
lower respiratory tract.
2. MO: Is that all you think of when someone complains about coughing ? Is there anything else
?
K: No, I dont think all these things come to mind; it rather depends on the person that comes in.
3. P. Consider whether there is an atopical constitution, in the family or whether they have a
history of hay fever, asthma, or eczema. [Several participants agree]
MO: Do you always ask these questions or... ?
P: No, it is directed you know, if your intuition, if you have a feeling, then you will ask
questions about it, definitely not routine questions.
MO: Which are the routine questions you ask a patient with complaints about coughing - an
adult patient ?
P: How long have you had this cough ? Do you cough up something, is it a dry cough, is it a
cough that is productive ? Are there any other symptoms, such as a fever?
4. MO: How do know that it is not, that it is not asthma.
R: That it is not ? [Laughter]
MO: Yes, I always say
R: If, for example, the complaint is infectious, yes, then it can still be asthma but that is not your
first diagnosis, you know. Yes, on the whole I can tell when it is asthma.
5. W: If you read about it, or hear about is, you cant tell in advance that green is bacterial and
white is viral, although you get the impression in general practice that it is possible to say this.

P: This is a rough division you make, because as a general practitioner you dont have the
possibility to say could you spit in this pot please and I dont immediately have a culture, so
you have to use the means which you have at your disposal and otherwise, yes. You are
somewhat limited.

G: But if there are coloured sputa, youre going to take it seriously ?


P: That is for instance a stronger argument than fever of course, for me at least [agreement].
6. MO: But can you tell the difference between bacterial or whether it is a virus ? [Several
participants shake their heads]
MO: This is not possible ?
K: Only suspect, you know.
MO: Is there always a clear difference between bronchitis and pneumonia ? [Several
participants: no.]

18

A qualitative decision analysis


A: No, this involves a bit of guesswork.
7. S: Often also things patients say, Oh, with me it sinks very quickly [K agrees: yes].
S: Or I always take antibiotics or My other general practitioner prescribed antibiotics and it
works well. Then you already know, well, if I dont prescribe antibiotics, then he will call back
within two days.
N: Or never again.
8. E: Nowadays, there are also adults who ask give me an antibiotic because I have to work.
K: You are not going to prescribe antibiotics, are you ?
9. E: And probably also a bit [depending] on your experience in the period before. For instance,
I once missed a pneumonia. Then you realise, you are not going to wait for those three weeks
anymore, but more rapidly [agreement] and then it eases off again.
10. P: Usually you have to say, I m going to start something [antibiotics] here or Im not
going to start anything. You cant say, Im going to wait until something is cultured, because this
will take three days. So, you have to rely on something in order to possibly start something.
Well, this is not only the colour of the sputa, but a number of elements taken together which will
push you across a certain threshold whether or not to prescribe antibiotics.
11. G: And tracheitis, then the pain is situated low in the neck, right here, above the windpipe.
MO: Yes.
P: These are welcome diagnoses?
G: If they tell you this, you are really satisfied.
P: Yes. In such a complaint you are certain that you dont have to prescribe antibiotics and it will
be all right.
12. P: You dont know this in bronchitis, you know. And you will conclude more easily that you
have to prescribe something here. This is what I meant earlier when I said that antibiotics are
frequently prescribed when not really necessary.
13. W: If it is bacterial, you have to prescribe antibiotics, if it is viral it doesnt really matter
whether you prescribe them or not.
K: Eventually, you give too much.
14. M: Yes, they blame you, yes. You are blamed for not prescribing antibiotics when
necessary. I heard a colleague of mine saying: I regret a couple of things, that is that I did not
prescribe antibiotics, it was at a lecture Someone from the emergency unit said that we
prescribed far too many antibiotics. Yet, try not prescribing antibiotics and then having to find
out afterwards that there was something.
15. K: We try to differentiate. We are already satisfied if we can establish the difference
between bacterial and viral. Several participants agree.
M: We explain why we dont prescribe antibiotics.
K agrees: Explain why we dont. But to differentiate between all these different viruses,
personally, I cant do that.

19

Chapter II
Determining the action thresholds. Also other determinants play a role, such as
patients expectations, time pressure during consultation or fear of loosing
patients (Table 2:7). A distinction can be made between patient-related (for
instance the patient s willingness to take medicines) and GP-related factors (for
instance recent experiences) (Table 2:8-9). These factors are determinants of
the decision whether or not to prescribe antibiotics (Table 2:7-8). GPs
suggested that in the end they make this (therapeutic) decision in complaints
about coughing and suspected RTI (Table 2:10).

Eventually the decision-making process is related to GPs diagnostic


(un)certainty. Pain in the trachea for instance was regarded as a sure diagnosis
of trachetis and this argument provided certainty as to whether or not to
prescribe antibiotics (Table 2:11). In the case of bronchitis, however, GPs were
less certain of the diagnosis (Table 2:6), which caused uncertainty in
prescription behaviour, and in a number of cases antibiotics were unnecessarily
prescribed (Table 2:14). Such a decision in favour of antibiotics can be
explained by qualitative decision analysis:15 prescribing antibiotics
unnecessarily is considered less inappropriate (Table 2:13) i.e. caused GPs
less chagrin than inappropriately not prescribing antibiotics (Table 2:14).
From the above it may be concluded that the decision to prescribe antibiotics is
better explained by both types of determinants than by conventional diagnostic
groups of RTIs. Although, GPs explain this prescribing decision to the patient
by refering to the diagnosis (Table 2:15).
In light of this interpretation of the texts, hypotheses were generated which
constitute the actual results of this kind of investigation (Table 3).

Discussion
Qualitative research in general and focus group research in particular is not
often used in medical research, where there is a clear preference for
randomised controlled trials. Clinical researchers have a problem with the fact
that qualitative methods replace testing hypotheses by generating hypotheses,

20

A qualitative decision analysis


that measurements are replaced by explanations and generalisations by
interpretations.
The creation of an evidence-based medical culture will depend, however, on
contributions from both quantitative and qualitative traditions. Qualitative
methods allow the examination of areas inaccessible to quantitative methods.
They are more suited to understand complex topics than to show their
relevance. As a result, these methods are very useful to investigate medical
decision-making by exploring the explained as well as the implicit routines and
rules adhered to by GPs.16 Focus group research yields data more quickly than
participant observation. The interaction during discussions affords a better
insight into the development of knowledge and ideas than in-depth interviews.
The weakness of qualitative research concerns bias and generalisation.
Compared to quantitative research, the methods are more valid but less
reliable.17 In order to ensure the trustworthiness of the results, the data were
analysed by two researchers, who worked blind and independently of each
other. This made it possible to reach a consensus about the code book and the
assignment of codes to the texts. Furthermore, the hypotheses are supported by
the data. Presenting the results orally to Flemish and European GPs and
researchers, their feedback confirmed our interpretation of the texts.18 19 For
this written report, only the most representative texts have been translated.
In March 1998 there was an increase in the number of consultations for acute
RTIs.20 For some GPs the increased workload associated with this epidemic
was probably the reason for not participating. As far as the participants were
concerned, however, this epidemic created suitable conditions for this survey,
and yielded valid information on the complex decision-making processes by
participating GPs.
Our sampling method, composition of groups and working towards saturation,
provided a broad range of data.7 Non-verbal information showed all group
members were actively involved and clearly stated their opinions and
disagreements. The latter concerned the importance of determinants for the
prescribing decision. After content analysis, however, the evidence for two
distinct categories of determinants emerged from all focus groups.
The survey was both exploratory and descriptive. Although the results do not
represent the norm, it is possible on the basis of data in the literature to design
an evidence-based decision-making model, which closely relates to the GPs
21

Chapter II
way of thinking. As a result of the selection bias and the non-statistical nature
of the sample, the results cannot be generalised. Hence, the results have to be
quantified formally.
In order to test the validity of the hypotheses, they were compared against
results of other research methods, a process commonly referred to as
triangulation. It is clear that GPs only explicitly work on diagnoses, which
seem plausible, while collecting fewer arguments for less evident diagnoses.
(Table 3: hypotheses 1-2). It seems as if they can only confirm diagnoses.
Gatekeepers though, are mainly expected to be good at excluding diagnoses.
Indeed, GP assessment is a relatively powerful excluder in patients suffering
from RTIs.1 This apparent contradiction may be explained by the fact that
the determinants to rule out hypotheses such as a GP s judgement could not
be made sufficiently explicit by means of the method used. (Table 2:5)
According to the participants, the differentiation between RTIs was based on a
low degree of certainty (Table 3:3). In coughing and suspected RTI, GPs can
only provide weak arguments for the diagnosis of RTI e.g. pneumonia on
the basis of medical history and clinical examination.1 GPs question the value
and/or feasibility of technical investigations such as blood or sputum analysis,
or X-ray examination.2 According to Kassirer, the fact of aiming for diagnostic
certainty results in excessive testing, whereas certainty is not a precondition for
good therapeutic decisions. Dealing with diagnostic uncertainty is related to the
therapies available.21 If GPs consider antibiotics highly effective and almost
risk free (Table 2:15), it is logical that they will decide to use them in
treatments even if there is a certain degree of uncertainty.
In the prescription of antibiotics, also other factors such as patients
expectations play a role (Table 3:4). If there is diagnostic uncertainty, this is
almost unavoidable.22 In addition, Butler s research has shown that irrational
prescription behaviour regarding sore throats can be explained by the desire to
avoid straining the doctor-patient relationship.23 The organisation of healthcare
in Belgium, where there is no official relationship between doctors and patients
and where doctors are paid fee for service, may also account for the excessive
use of antibiotics.24 All these factors fall within Feinstein s Chagrin factor.15
GPs considered it less appropriate not to have prescribed antibiotics when this
proved to be necessary (Table 2:18), than having prescribed antibiotics when
not necessary (Table 2:17). The latter caused less chagrin to the GPs. Then,
when necessary does not only mean necessary to cure patients, but also

22

A qualitative decision analysis

Table 3 Hypotheses on GPs decision-making regarding complaints about coughing


and on the determinants underlying their decisions.

1. The first diagnosis that comes to a GPs mind is respiratory tract infection (RTI).
This diagnosis is reached independent of the patient. Other hypotheses emerge
only if they are considered plausible as a result of prior knowledge of the patient.

2. GPs ask routine questions to confirm only the most likely diagnoses. Explicitly
ruling out other diagnoses is less often used in decision-making.

3. In suspected RTI, GPs want to make a distinction between clinical syndromes


such as bronchitis and pneumonia, viral and bacterial RTI and upper and lower
RTI. This cannot be achieved with certainty on the basis of medical history and
clinical examination. Dealing with diagnostic uncertainty, GPs decisions are
directed at whether or not to prescribe antibiotics.

4. For this (therapeutic) decision, also doctor- and patient-related factors play a role.
These factors give rise to a shift in the action thresholds in favour of antibiotics, a
phenomenon explained by the Chagrin factor. The decision to prescribe
antibiotics is better explained by both types of determinants than by the
conventional diagnostic groups of RTIs.

23

Chapter II
necessary to function as adequately as possible as a GP without losing
patients as a result of unfulfilled expectations or undetected serious diseases.
Finally our results are in line with Howie s hypothesis: although GPs
therapeutic decisions are normally described using a diagnostic label, in reality
it is often better to view them in terms of symptoms and signs and influenced
by factors (Table 3:4).25 A diagnosis is then formulated as a justification for a
therapeutic decision (Table 2:17).

In patients with complaints about coughing GPs need manageable arguments to


select patients who may or may not benefit from antibiotics. These data have to
be collected in further research and lead to answers to the following questions
derived from the hypotheses:
1. Does ruling out less likely diagnoses add something to merely trying to
diagnose a (certain) RTI?
2. Do GPs prescribe antibiotics to a lesser degree and more adequately when
they have stronger clinical evidence to support their decisions?
3. Are diagnostic syndromes on the basis of this strong clinical evidence more
manageable in general practice than classical syndromes?
So far, such clinical evidence is unknown both for patients with coughs and for
bronchitis. Meta-analyses also show that in most cases antibiotics do not offer
any benefits, which outweigh the possible side effects.3 26 As a result, research
into a more effective use of antibiotics has to pay special attention to the
doctor- and patient-related factors in the relationship and communication
between GPs and patients.
The authors are developing an educational intervention that builds on these
findings, aiming to reduce antibiotic use and cost, while preserving patient
outcomes.

24

A qualitative decision analysis


References
1. Metlay J, Kapoor W, Fine M. Does This Patient Have Community-Acquired
Pneumonia? Diagnosing Pneumonia by History and Physical Examination.
JAMA 1997;278:1440-5.
2. Jonsson J, Sigurdsson J, Kristinsson K, Gudnadttir M, Magnusson S. Acute
bronchitis in adults. How close do we come to its aetiology in general practice?
Scand J Prim Health Care 1997;15:156-60.
3. Becker L, Glazier R, McIsaac W, Smucny J. Antibiotics for Acute
Bronchitis (Cochrane Review). In: The Cochrane Library, Issue 4, 1999.
Oxford: Update Software.
4. Melbye H, Straume B, Aasebo U, Dale K. Diagnosis of pneumonia in adults
in general practice. Relative importance of typical symptoms and abnormal
chest signs evaluated against a radiographic reference standard. Scand J Prim
Health Care 1992;10:226-33.
5. Wise R, Hart T, Cars O, Streulens M, Helmuth R, Huovinen P, Sprenger M.
Antimicrobial resistance. Is a major threat to public health [editorial]. BMJ
1998;317:609-10.
6. Butler C, Rollnick S, Kinnersley P, Jones A, Stott N. Reducing antibiotics
for respiratory tract symptoms in primary care: consolidating '
why' and
considering '
how'
. Br J Gen Pract 1998;48:1865-70.
7. Morgan D, Krueger R. The Focus group Kit, Volumes 1-6. London: Sage
Publications, 1998.
8. Marshall M. Sampling for qualitative research. Fam Pract 1996;13:522-5.
9. Douglas T. Groepswerk in de praktijk [Groupwork in practice]. '
s
Gravenhage: VUGA, 1979.
10. Alblas G. Groepsprocessen: het functioneren in taakgerichte groepen
[Groupprocesses: functioning in task-oriented groups]: Van Loghum-Slaterus,
1983.

25

Chapter II
11. Van den Ende J, Derese A, Debaene L, de Bthune X, Lemiengre M, Van
Puymbroek H, et al. Medische besliskunde: een nieuw accent in de Vlaamse
huisartsgeneeskunde [Medical decision analysis: a new accent in Flemish
general practice]. Huisarts Nu 1996;9:281-344.
12. Morgan D. Qualitative content analysis: a guide to paths not taken. Qual
Health Res 1993;3:112-1.
13. Q.R.S.NUD.IST 4 user guide. London: Sage Publications Software, 1997.
14. Pauker S, Kassirer J. The threshold approach to clinical decision making.
NEJM 1980;302:1109-17.
15. Feinstein A. The '
Chagrin Factor'and Qualitative Decision Analysis. Arch
Intern Med 1985;145:1257-9.
16. Britten N, Jones R, Murphy E, Stacy R. Qualitative research methods in
general practice and primary care. Fam Pract 1995;12:104-14.
17. Britten N, Fisher B. Qualitative research and general practice [editorial]. Br
J Gen Pract 1993;43:270-1.
18. Coenen S, Van Royen P, Vermeire E, Hermann I, Denekens J. Met welke
argumenten voert de huisarts zijn diagnostisch beleid bij de contactreden
'
hoesten'[General practitioners arguments to make diagnostical decisions
when coughing is the reason for encounter]. VHI Referatendag [Lecture-day of
the Flemish General Practitioners Institute]; 1999; Gent.
19. Coenen S, Van Royen P, Vermeire E, Hermann I, Denekens J. What
determines medical decision-making in patients with coughing as the reason
for encounter? Focus group research with general practitioners. First European
Networks Open Conference WONCA '
99; 1999; Palma de Mallorca.
20. Ducoffre G. Surveillance van Infectieuze Aandoeningen door een Netwerk
van Laboratoria voor Microbiologie 1997 + Epidemiologische Trends 19831996 [Surveillance of Infectious Disease by a Network of Laboratories for
Microbiology 1997 + Epidemiological Trends 1983-1996]. Brussel: Ministerie
van Sociale Zaken, Volksgezondheid en Leefmilieu, Wetenschappelijk
Instituut Volksgezondheid-Louis Pasteur [Ministry of Social Affairs, Public

26

A qualitative decision analysis


Health and Environment, Scientific Institute of Public Health-Louis Pasteur],
1998.
21. Kassirer J. Our stubborn quest for diagnostic certainty. A cause of
excessive testing. NEJM 1989;320:1489-91.
22. Fahey T. Antibiotics for respiratory tract symptoms in general practice
[editorial]. Br J Gen Pract 1998;48:1815-6.
23. Butler CC, Rollnick S, Pill R, Maggs-Rapport F, Stott N. Understanding
the culture of prescribing: qualitative study of general practitioners'and
patients'perceptions of antibiotics for sore throats. BMJ 1998;317:637-42.
24. Grol R, De Maeseneer J, Whitfield M, Mokkink H. Disease-Centred Versus
Patient-Centred Attitutes: Comparison of General Practitioners in Belgium,
Britain and The Netherlands. Fam Pract 1990;7:100-3.
25. Howie J. Diagnosis, the Achilles heel? J R Coll Gen Pract 1972;22:310-5.
26. Fahey T, Stocks N, Thomas T. Quantitative systematic review of
randomised controlled trials comparing antibiotic with placebo for acute cough
in adults. BMJ 1998;316:906-10.

27

28

III
Antibiotics for coughing in general practice:
A questionnaire study to quantify and condense
the reasons for prescribing
Introduction
Antibiotics are being overprescribed in ambulant care,1 especially for
respiratory tract infections (RTIs).2 For this prescribing decision different types
of determinants are already highlighted. 3-6 However, gaining insight into the
actual reasons for context specific prescribing remains important to design
effective strategies to optimise antibiotic prescribing.7
In general practice, medical decisions (concerning RTIs) are prompted most
often by complaints about coughing: 169 times per 1000 patients per year for a
new illness episode.8 Since there is no evidence base for the prescription of
antibiotics for coughing in case of suspected RTI,9 and since antibiotic
prescribing results in financial costs to the patient and society, adverse effects
and development of bacterial resistance,10 we explored the diagnostic and
therapeutic decisions of Flemish general practitioners (GPs) regarding adult
patients who consult them with complaints about coughing by means of focus
groups.11 We found medical as well as non-medical reasons for antibiotic
prescriptions in case of suspected RTI.12 Our hypotheses on Flemish GPs'
decisions were in line with previous research. The differentiation between

Published in BMC Fam Pract 2002;3:16 (10p).

29

Chapter III
RTIs, e.g. acute bronchitis and pneumonia, could not be achieved with
certainty on the basis of medical history and clinical examination:13 i.e.
medical reasons. Dealing with this diagnostic uncertainty, GPs'decisions were
directed at whether or not to prescribe antibiotics.14 Determinants playing an
important role in this decision are physician related, e.g. having missed
pneumonia once, or patient related, e.g. patient expectations:15 i.e. non-medical
reasons.
Since it is time for action,16 besides a better understanding of the actual
determinants for context specific prescribing of antibiotics, we also have to
make them operational for the design of an intervention. Therefore, we aimed
to quantify and to condense the determinants generated in the focus group
study. By means of this postal questionnaire study in Flemish general practice,
we assessed to what extent Flemish GPs consider those determinants in
decision making in case of suspected RTI in a coughing patient and how
strongly the determinants support or counter antibiotic treatment.

Methods

Design
We performed an explanatory study comparing GPs'responses from a selfadministered questionnaire based upon focus group findings.12

Setting and sample


We approached Flemish GPs who were willing to participate in previous
studies of our research unit.12 17 The questionnaire was sent to this selected
group by mail early September 1999. A reminder was sent to all nonresponders two weeks later. Responses were accepted until the end of
September 1999. The survey was pilot tested.
Of the 316 GPs originally selected to be in the sample, 7 were no longer
practising, 5 returned surveys with more than 20% of items unanswered, 116
failed to respond before the end of September 1999, leaving 188 GPs who
completed the survey. The overall response rate was 59.5%.

30

A questionnaire study to quantify and condense


the reasons for prescribing

Instrument
To assess the importance of determinants for the antibiotic prescribing decision
the GPs were sent one questionnaire in two parts, one (Q1) assessing to what
extent they consider these determinants in decision making in case of suspected
RTI in a coughing patient, and another (Q2) assessing how strongly these
determinants support or counter antibiotic treatment. The response set for each
item was a 5-point verbal rating scale (VRS). For Q1 the VRS ranged from 1
(never), 2 (seldom), over 3 (sometimes), to 4 (often) and 5 (always), for Q2
from 2 (strongly in favour), 1 (in favour), over 0 (neutral), to -1 (against) and 2 (strongly against). '
See addendum for the original questionnaire(in Dutch)
used in this study'
Focus groups, exploring the determinants of GPs'diagnostic and therapeutic
decisions in adult patients with complaints about coughing, provided the items
proposed in the questionnaire.12 The main categories at which we ordered the
determinants were: epidemiology, e.g. an influenza epidemic, prior knowledge
of the patient, e.g. he/she is smoking, symptoms, e.g. a sputum producing
cough, clinical signs, e.g. a normal lung auscultation, patient related nonmedical determinants, e.g. patient'
s demand for antibiotics, physician related
non-medical determinants, e.g. having missed pneumonia once. There were
also: first impression, e.g. patient looks very ill, laboratory results, e.g. normal
erythrocyte sedimentation rate, and natural course, e.g. the illness is
worsening.
For the questionnaire (Q1 and Q2) items were chosen from all the above
categories to include a meaningful selection of all the issues mentioned in the
focus group study. Concerning their content the items were to be manageable
in daily practice and to determine the decision whether or not to prescribe
antibiotics. Therefore no issues about laboratory results were included: blood
or sputum analysis, or X-ray examination are seldom performed in Flemish
[Coenen, unpublished] general practice in case of suspected RTI.8
Determinants only used to confirm other possible diagnoses with coughing as
principal complaint, e.g. risk factors for pulmonary oedema, were also
excluded. The items were to allow clear and brief formulation as well. Five
physicians staffing our Centre, including two authors of the report (PVR and
SC) selected the issues in the decision to prescribe in consensus and rephrased
them into questionnaire items faithful to the original formulation.

31

Chapter III

Analysis
All statistical analyses were performed using Statistica 5.1 (StatSoft, Inc.,
Tulsa, OK, USA). To make the selected issues operational for an intervention
trial, exploratory factor analyses on the questionnaire'
s items in Q1 and Q2
were performed, using the principal axis method and varimax normalised
rotation. The relative importance of the operational factors yielded was
assessed using Wilcoxon Matched Pairs test. For comparison of ordinal
variables between two groups, the Mann-Whitney test was used.

Results
The mean age (SD) of the GPs was 42.8 years (7.7), 65.9% were men. 46.9 %
of the GPs worked single-handed. GPs were predominantly rewarded by fee for
service; 24.9% had more than 120 patient encounters per week.

Considering the determinants in decision making


Assessing to what extent GPs consider the determinants for antibiotic
prescribing in decision making (Q1), on average GPs considered all 42 items.
Factor analysis suggests groups of variables whose values are similar, in this
case GPs'responses to Q1 items. Factor analysis of all items from Q1 yielded
three factors, i.e. groups of items which GPs considered similarly, explaining
33 % of the variance (Figure 1). Factor 1 included all the items relating to the
lung auscultation. Factor 2 included only items relating to non-medical reasons,
either patient or physician related. Factor 3 included items determining whether
or not there is something unusual happening. Each factor grouping had good
internal consistency, with Cronbach equal to .90 for factor 1, .86 for factor 2
and .87 for factor 3.

32

A questionnaire study to quantify and condense


the reasons for prescribing
(1)

Figure 1 Items from part 1 of the questionnaire (Q1) : distribution of scores and
factor loadings per yielded factor.

Items from Q1(1)

Factor loadings(2)
F1
F2
F3

Crepitations at lung auscultation

0,84 -0,06

0,13

Ronchi at lung auscultation

0,92

0,01

0,17

Wheezing at lung auscultation

0,79

0,09

0,14

Reduced vesicular breathing

0,74

0,01

0,20

Patient asks for medication in general

0,05

0,68

0,32

Patient needs quick recovery for work

0,15

0,70

0,05

Patient asks for antibiotics

0,07

0,81

0,16

Patient expects antibiotics according to you

0,07

0,78

0,07

You will be blamed not having prescribed antibiotics, if it


subsequently appears to be necessary

0,05

0,68

0,14

Patient will already reconsult within two days if not better


and not prescribed antibiotics

0,14

0,51

0,22

-0,02

0,52

0,00

Dyspnoea

0,38

0,10

0,55

Flu-like complaints

0,21

0,01

0,53

Complaint existing less than three days

0,08

0,24

0,49

Deteriorating general condition

0,14

0,08

0,70

Fever getting higher

0,16

0,09

0,66

Complaints improving spontaneously

0,19

0,27

0,61

Respiration rate is to high

0,23

0,12

0,50

Hoarseness

0,28

0,19

0,55

Localised thoracic pain

0,10

0,05

0,61

Mainly lying in bed

0,09

0,11

0,73

You work under pressure of time

Never

Distribution of scores(3)
2
3
4

Sometimes
Seldom
Often

Always

(1) Part 1 of the questionnaire (Q1) is assessing to what extent the questionnaire items are considered in decision making in case
of suspected RTI in a coughing patient. Only items with factor loading > .40 to the yielded factor are presented.
(2) Factor loadings to the yielded factors are presented. Factor 1 (F1) includes all the items relating to the lung auscultation. Factor
2 (F2) includes only items relating to non-medical reasons, either patient or physician related. Factor 3 (F3) includes items
determining whether or not there is something unusual happening.
(3) The small box represents the median, the larger box the interquartile range, the wiskers the scoring range.

33

Chapter III
The median (interquartile range) scores as defined by factor analysis were 5.0
(from 5.0 to 5.0) for factor 1, lung auscultation, 3.0 (form 2.8 to 4.0) for factor
2, non-medical reasons, and 4.0 (from 4.0 to 5.0) for factor 3, unusual or not.
Using Wilcoxon Matched Pairs test to compare the scores of the factors
scores of factor 1 did not approximate a normal distribution the differences
between all three factors are significant at P < 0.001 (Figure 2). Since the
differences between the scores approximate a normal distribution, this test is
almost as powerful as the t-test.

Figure 2 Factors from part 1 of the questionnaire: comparing scores as defined by


factor analysis.

Factor 1 *

Factor 2 *

Factor 3 *

Max
Min
75%
25%
Never

Seldom

Sometimes

Often

Always

Median

Part 1 of the questionnaire (Q1) is assessing to what extent the questionnaire items are considered in
decision making in case of suspected RTI in a coughing patient.
Factor 1 includes all the items relating to the lung auscultation.
Factor 2 includes only items relating to non-medical reasons, either patient or physician related.
Factor 3 includes items determining whether or not there is something unusual happening.
Asterisk means that the factor'
s score significantly differs from the other factors'scores (p < .001).

34

A questionnaire study to quantify and condense


the reasons for prescribing
Of course, GPs also considered items from Q1 not presented in figure 1 (factor
loading .40 to the yielded three factors); always (median = 5) whether the
patient has fever, is coughing up sputum and whether the sputum is coloured,
whether the patient is looking ill and whether he/she has a medical history of
COPD or smoking; often (median = 4) whether the coughing is frequent or
started suddenly and whether the patient consults for the first time with this
complaint, is saying he/she is feeling ill, is older than 60 years of age, tried
self-management first, is known to you or has a red throat, as well as whether
there is an RTI epidemic and whether the patient rapidly consults and will
reconsult if not better; sometimes (median = 3) whether the patient is
compliant, or is recovering slowly even under antibiotic treatment; seldom
(median = 2) whether the patient is visited at home or that you make the patient
reconsult anyway after 3 to 4 days. For most items the interquartile range was
1.

In favour or against antibiotics


Assessing how strongly the determinants for antibiotic prescribing support or
counter antibiotic treatment (Q2), none of the 63 items is strongly in favour or
against antibiotic treatment. Factor analysis of all items from Q2 yielded two
factors, i.e. groups of items which according to the GPs support antibiotic
treatment similarly. The factors included items expressing a need for antibiotic
treatment, and no need for antibiotic treatment respectively. This confirmed our
construction of Q2.

In favour
Factor analysis of all 37 items that support antibiotic treatment according to
their mean and sumscore, yielded two factors, i.e. groups of items which
according to the GPs are equally in favour of antibiotic treatment, explaining
24% of the variance (Figure 3). Factor 1 only included items relating to
medical reasons, either from the lung auscultation or determining whether or
not there is something unusual happening, factor 2 only included items relating
to non-medical reasons, either patient or physician related. Each factor
grouping had good internal consistency, with Cronbach equal to .82 for
factor 1, .83 for factor 2.

35

Chapter III
Figure 3 Items in favour of antibiotic treatment from part 2 of the questionnaire (Q2)(1):
distribution of scores and factor loadings per yielded factor.
Items Q2
in favour of antibiotic treatment(1)

Factor loadings(2)
F1
F2

Deteriorating general condition

0,55

0,06

Percussion dullness

0,41

0,08

More than three days of fever

0,52

0,15

More than three days in bed with fever

0,60

0,10

Ronchi at lung auscultation

0,57

0,18

Looking ill

0,61

0,17

A child getting higher fever

0,42

0,22

High fever (> 38.5C)

0,51

0,09

Patient says he/she is feeling ill

0,60

0,25

Wheezing at lung auscultation

0,53

0,14

Swollen cervical lymph nodes

0,41

0,12

Coughing up sputum

0,42

0,19

You are not easy about it

0,23

0,48

Patient needs quick recovery for work

0,24

0,49

You will be blamed not having prescribed antibiotics, if it


subsequently appears to be necessary

0,25

0,69

You wont see the patient again, if not recovering

0,14

0,56

A childs parent pressure you

0,17

0,72

Patient expects antibiotics according to you

0,14

0,73

Patient asks for antibiotics

0,21

0,71

You work under pressure of time

-0,06

0,41

-2

Distribution of scores(3)
-1
0
1

Strongly
Neutral
Strongly
against
in favour
Against
In favour
(1) Part 2 of the questionnaire (Q2) is assessing how strong the questionnaire items argue in favour or against antibiotic treatment
in case of suspected RTI in a coughing patient. Only items that on average argue in favour of antibiotic treatment, with factor
loading > .40 to only one of the yielded factors are presented.
(2) Factor loadings to the yielded factor are presented. Factor 1 (F1) only includes items relating to medical reasons, either from the
lung auscultation or determining whether or not there is something unusual happening, factor 2 (F2) only includes items relating
to non-medical reasons, either patient or physician related.
(3) The small box represents the median, the larger box the interquartile range, the wiskers the scoring range.

36

A questionnaire study to quantify and condense


the reasons for prescribing
The median (interquartile range) scores as defined by factor analysis were 1.0
(from 0.5 to 1.0) for factor 1, medical reasons and 0.0 (from 0.0 to 1.0) for
factor 2, non-medical reasons. Using Wilcoxon Matched Pairs test the scores
of the two factors differed significantly at P < 0.001 (Figure 4).

Figure 4 Factors from part 2 of the questionnaire: comparing scores as defined by


factor analysis.

IN FAVOUR
Factor 1 *

Factor 2 *

AGAINST
Max
Min

Factor 1 *

75%
25%
Strongly
against
-

Against

Neutral

In favour

-1

Strongly
in favour

Median

Part 2 of the questionnaire (Q2) is assessing how strong the questionnaire items argue in favour or against
antibiotic treatment in case of suspected RTI in a coughing patient.
Factors '
in favour'include items which according to their mean and sumscore support antibiotic treatment.
Factor 1 only includes items relating to medical reasons, either from the lung auscultation or determining
whether or not there is something unusual happening, factor 2 only includes items relating to non-medical
reasons, either patient or physician related.
Factors '
against'include items which according to their mean and sumscore fail to support antibiotic
treatment. Factor 1 only includes items expressing no need for antibiotic treatment, either medical or nonmedical.
Asterisk means that the factor'
s score significantly differs from the other factors'scores (p < 0.001)

37

Chapter III

Against
Factor analysis of all 26 items that fail to support antibiotic treatment according
to their mean and sumscore, yielded only one factor, i.e. group of items which
according to the GPs are equally against antibiotic treatment, explaining 17%
of the variance (Figure 5). The factor only included items expressing no need
for antibiotic treatment, either medical or non-medical. Factor grouping had
good internal consistency, with Cronbach equal to .80.
The median (interquartile range) score as defined by factor analysis was -1.0
(from -1.0 to -0.5). Using Wilcoxon Matched Pairs test the score of this factors
differed significantly at P < 0.001 form the scores of the two factors in favour
of antibiotics (Figure 4).
Items from Q2 not presented in figure 3, and figure 5 respectively (factor
loading
.40 to the yielded three factors) support or counter antibiotic
treatment as well. In favour (median = 1) are crepitations at lung auscultation,
medical history of COPD, onset of new complaints in a viral syndrome,
consulting for the second time, dyspnoea, tachypnoea, localised thoracic pain,
painful teeth or sinuses, coloured sputum, haemoptysis, reduced vesicular
breathing, red throat with exudates on the tonsils, the patient being older than
60 years of age and not consulting rapidly. Neutral (median = 0) are smoking,
home visit, frequent coughing, no swollen cervical lymph nodes, no localised
thoracic pain, medication demand, as well as an RTI epidemic, a dry cough, a
red throat without exudates on the tonsils, the patient is known to you, that you
make the patient reconsult anyway after 3 to 4 days, that without antibiotic
treatment the patient will already reconsult within two days, if not better and
bad compliance with antibiotics. Against (median = -1) are consulting rapidly,
influenza-like symptoms, no worsening after two days and not wanting
antibiotic treatment. For most items the interquartile range was 1.
No relation between the response groups characteristics and the scores as
defined by factor analyses of Q1 and Q2 was found to be relevant and
significant.

38

A questionnaire study to quantify and condense


the reasons for prescribing
Figure 5 Items against antibiotic treatment from part 2 of the questionnaire (Q2)(1):
distribution of scores and factor loadings per yielded factor.
Items from Q2
(1)
against antibiotic treatment

Factor loadings
F1

Hoarseness

0,50

No smoker

0,40

First consult

0,51

Younger than 60 years of age

0,40

You will see the patient again, if not recovering

0,52

No Chronic Obstructive Pulmonary Disease

0,59

White sputum

0,42

Complaint existing less than three days

0,57

You rest easy about it

0,60

Not looking ill

0,56

Normal lung auscultation, no crepitations nor ronchi

0,41

Improving under own (home) medication

0,62

(2)

-2

Distribution of scores
-1
0
1

(3)

Strongly
Neutral
Strongly
against
in favour
Against
In favour
(1) Part 2 of the questionnaire is assessing how strong the questionnaire items argue in favour or against antibiotic treatment in case of
suspected RTI in a coughing patient. Only items that on average argue against antibiotic treatment, with factor loading > .40 to the
yielded factor are presented.
(2) Factor loadings to the yielded factor are presented. Factor 1 (F1)only includes items expressing no need for antibiotic treatment, either
medical or non-medical.
(3) The small box represents the median, the larger box the interquartile range, the wiskers the scoring range.

39

Chapter III
Discussion
This questionnaire study with adequate response18 enabled us to quantify and
condense the focus group determinants and confirmed our focus group finding,
that GPs'decisions to prescribe antibiotics are determined by both medical and
non-medical reasons.12
Neither the internal validity nor the reliability of the questionnaire was
formally assessed, but was assumed acceptable: the questionnaire was
developed based upon our focus group study and, notwithstanding other factor
loadings, the factor analysis of a sample of all GPs in professional training in
June 1999 yielded the same results [Coenen, unpublished]. The results of the
factor analysis thus seem independent of selection bias. The quantification
results though may be biased due to the recruitment and non-response of GPs.
The response group characteristics however approximate that of all Flemish
GPs, and for a postal survey of general practitioners a response rate of 59.5% is
good according to the literature.18 Self report also might have limited our data
by underestimating the importance of the non-medical reasons. Nevertheless
our data show that non-medical reasons determine antibiotic prescribing as
well.
In their decision making Flemish GPs seem to consider all the determinants
included in the questionnaire. Since the complexity of the prescribing decision,
we were not surprised the yielded factors explained only little variance.
Nevertheless, the GPs almost always consider the operational factor '
lung
auscultation'
, often '
whether or not there is something unusual happening'
both medical reasons and to a lesser extent '
non-medical reasons'
, either
patient or physician related. According to the GPs non-medical as well as
medical reasons are in favour of antibiotic treatment, the non-medical reasons
to a lesser extent.
Yet, for patients with acute (productive) cough the benefit from antibiotics is
limited: antibiotics do not influence the duration of productive cough, nor that
of limitation in work or activities; out of every 10 patients with acute
(productive) cough more than 8 will be clinically improved after 711 days
regardless the use of antibiotics; less than one patient extra will be improved
due to antibiotics, but as many patients will experience the side effects of
treatment.9 19 And although there is a strong association between focal chest
signs and radiographic pneumonia, which suggests presence of focal chest
40

A questionnaire study to quantify and condense


the reasons for prescribing
signs may be an important medical reason for antibiotic prescribing, there are
no clinical criteria to identify subsets of patients who are most likely to benefit
form antibiotic treatment.20 The presence of focal chest signs however is
associated with antibiotic prescribing.21
Also non-medical reasons such as patient expectations have been shown to
affect prescribing behaviour of GPs for both upper22 and lower5 RTIs. And, it
has been suggested that GPs'perception of patient expectations may be the
strongest determinant for antibiotic prescribing.23 24 In addition there is little
agreement between patient expectations and GPs'perception of these.24 25 And,
for as long as it is difficult in the primary care setting to identify patients for
whom antibiotics will be beneficial, these non medical reasons will inevitably
keep on playing an important role in the decision to prescribe antibiotics.26
Hence good clinical practice guidelines and interventions to optimise antibiotic
prescribing for acute cough in Flemish general practice have to take nonmedical reasons into account.
In the Flemish guideline for acute cough, for example, we recommend a
clinical and stepwise approach to assess the cause of acute cough. First,
possibly life-threatening, treatable conditions such as life-threatening
pneumonia should be ruled out.27-29 Although this first step may automatically
and quickly be undertaken, we like to explicitly stress its importance. Next, we
would like awareness of other not immediately life threatening conditions.
Asthma, postnasal drip or gastro-oesophageal reflux are not as prevalent as an
RTI, but require specific treatment.27 Even though such conditions may not be
obvious in a first encounter, it is worthwhile to take them into account. If
finally an RTI seems to be the most likely cause, there are no clinical criteria to
determine '
which patient will benefit from antibiotics'
.13 Nevertheless, this is
12
the question GPs are confronted with. Because the benefits from antibiotics
are outweighed by their harm side effects, financial cost and bacterial
resistance we promote reassurance, information and treatment without
antibiotics in case of suspicion of an RTI. To support this treatment decision,
we recommend to involve the patient and to make the non-medical reasons
explicit.
In a cluster randomised controlled trial we evaluate whether an educational
intervention based on the Flemish guideline for acute cough optimises
antibiotic prescribing for acute cough, i.e. achieves the goals of the Belgian

41

Chapter III
public campaign: "Antibiotics, use them less often, but better," (www.redantibiotica.org/english/index.html) without affecting the patients'symptom
resolution.30 We focus on the non-medical reasons for prescribing, more
specifically on the GP'
s perception of patient expectations. Using the baseline
data of this controlled before and after study we will also validate the
importance of determining whether there is something unusual happening, the
lung auscultation, and non-medical reasons in the prescribing decision of
Flemish GPs.

Conclusions
This study assessed the importance for the antibiotic prescribing decision of the
determinants of previous focus group research and confirmed it'
s findings.12
According to the GPs non-medical as well as medical reasons can argue in
favour of antibiotic treatment. Good clinical practice guidelines and
interventions to optimise antibiotic prescribing have to take non-medical
reasons for antibiotic prescribing into account.

References
1. Cars O, Mlstad S, Melander S. Variation in antibiotic use in the European
Union. Lancet 2001;357:1851-3.
2. Wise R, Hart T, Cars O, Streulens M, Helmuth R, Huovinen P, Sprenger M.
Antimicrobial resistance. Is a major threat to public health [editorial]. BMJ
1998;317:609-10.
3. Howie JG. Clinical judgement and antibiotic use in general practice. BMJ
1976;2:1061-4.
4. Kuyvenhoven M, de Melker R, van der Velden K. Prescription of antibiotics
and prescribers'characteristics. A study into prescription of antibiotics in upper
respiratory tract infections in general practice. Fam Pract 1993;10:366-70.

42

A questionnaire study to quantify and condense


the reasons for prescribing
5. Macfarlane J, Holmes W, Macfarlane R, Britten N. Influence of patients'
expectations on antibiotic management of acute lower respiratory tract illness
in general practice: questionnaire study. BMJ 1997;315:1211-4.
6. Coenen S, Kuyvenhoven M, Butler C, Van Royen P, Verheij T. Variation in
European antibiotic use [letter]. Lancet 2001;358:1272.
7. De Sutter AI, De Meyere MJ, De Maeseneer JM, Peersman WP. Antibiotic
prescribing in acute infections of the nose or sinuses: a matter of personal
habit? Fam Pract 2001;18:209-13.
8. Okkes I, Oskam S, Lamberts H. Van klacht naar diagnose [From complaint
to diagnosis]. Bussum: Coutinho, 1998.
9. Fahey T, Stocks N, Thomas T. Quantitative systematic review of
randomised controlled trials comparing antibiotic with placebo for acute cough
in adults. BMJ 1998;316:906-10.
10. Butler C, Rollnick S, Kinnersley P, Jones A, Stott N. Reducing antibiotics
for respiratory tract symptoms in primary care: consolidating '
why' and
considering '
how'
. Br J Gen Pract 1998;48:1865-70.
11. Coenen S, van Royen P, Denekens J. Reducing antibiotics for respiratory
tract symptoms in primary care: '
why'only sore throat, '
how'about coughing?
[letter]. Br J Gen Pract 1999;49:400-1.
12. Coenen S, Van Royen P, Vermeire E, Hermann I, Denekens J. Antibiotics
for coughing in general practice: a qualitative decision analysis. Fam Pract
2000;17:380-5.
13. Metlay J, Kapoor W, Fine M. Does This Patient Have CommunityAcquired Pneumonia? Diagnosing Pneumonia by History and Physical
Examination. JAMA 1997;278:1440-5.
14. Kassirer J. Our stubborn quest for diagnostic certainty. A cause of
excessive testing. NEJM 1989;320:1489-91.
15. Butler CC, Rollnick S, Pill R, Maggs-Rapport F, Stott N. Understanding
the culture of prescribing: qualitative study of general practitioners'and
patients'perceptions of antibiotics for sore throats. BMJ 1998;317:637-42.

43

Chapter III
16. Huovinen P, Cars O. Control of antimicrobial resistance: time for action
[editorial]. BMJ 1998;317:613-4.
17. Michiels B, Avonts D, Van Royen P, Denekens J, Vander Auwera J-C.
Lower incidence of the upper respiratory tract infections among general
practitioners as compared to their patients. Eur J Epidemiol 2001;17:1059-61.
18. McAvoy B, Kaner E. General practice postal surveys: a questionnaire too
far? BMJ 1996; 313: 732-733. BMJ 1996;313:732-3.
19. Smucny J, Fahey T, Becker L, Glazier R, McIsaac W. Antibiotics for acute
bronchitis (Cochrane Review). In: The Cochrane Library, Issue 4, 2002.
Oxford: Update Software.
20. Macfarlane J, Holmes W, Gard P, Macfarlane R, Rose D, Weston V, et al.
Prospective study of the incidence, aetiology and outcome of adult lower
respiratory tract illness in the community. Thorax 2001;56:109-14.
21. Macfarlane J, Lewis SA, Macfarlane R, Holmes W. Contemporary use of
antibiotics in 1089 adults presenting with acute lower respiratory tract illness in
general practice in the U.K.: implications for developing management
guidelines. Respir Med 1997;91:427-34.
22. Little P, Williamson I, Warner G, Gould C, Gantley M, Kinmonth AL.
Open randomised trial of prescribing strategies in managing sore throat [see
comments]. BMJ 1997;314:722-7.
23. Cockburn J, Pit S. Prescribing behaviour in clinical practice: Patients'
expectations and doctors'perceptions of patients'expectations - a questionnaire
study. BMJ 1997;315:520-3.
24. Britten N, Ukoumunne O. The influence of patients'hopes of receiving a
prescription on doctors' perceptions and the decision to prescribe: a
questionnaire survey [see comments]. BMJ 1997;315:1506-10.
25. Dosh S, Hickner J, Mainous AI, Ebell M. Predictors of antibiotic
prescribing for nonspecific upper respiratory tract infections, acute bronchitis,
and acute sinusitis. J Fam Pract 2000;49:407-14.

44

A questionnaire study to quantify and condense


the reasons for prescribing
26. Fahey T. Antibiotics for respiratory tract symptoms in general practice. Br
J Gen Pract 1998;48:1815-6.
27. Managing Cough as a Defense Mechanism and as a Symptom. A
Consensus Panel Report of the American College of Chest Physicians. CHEST
1998;114(suppl):133S-181S.
28. Huchon G, Woodhead M. Management of adult community-acquired lower
respiratory tract infections. Eur Respir Rev 1998;8:391-426.
29. BTS Guidelines for the Management of Community Acquired Pneumonia
in Adults. Thorax 2001;56(Suppl 4):1iv-64iv.
30. Coenen S, Van Royen P, Michiels B, Denekens J. Promotion of rational
antibiotic use in Flemish general practice: implementation of a guideline for
acute cough [abstract]. Prim Care Respir J 2002;11:56.

45

Chapter III
Addendum: The original questionnaire (in Dutch) used in this study.
Een aantal persoonlijke gegevens, die anoniem verwerkt worden.
In te vullen of schrappen wat niet past
Postcode praktijkadres:

Geboortedatum:

//19

Geslacht:

vrouw / man

Jaar van promotie als arts:

19

Universiteit van promotie als arts:

KUL / RUG / UIA / VUB / andere:

Aantal jaren praktijkervaring als huisarts:

jaar

Gemiddeld aantal patintencontacten per week:

patintencontacten per week

Ik krijg hoofdzakelijk een vast loon:

Ja / Neen

Ik word hoofdzakelijk betaald per prestatie:

Ja / Neen / andere:

Totaal aantal huisartsen in uw praktijk vorig semester:

waarvan HIBO(s)

Huisarts of HIBO:

Huisarts / HIBO

46

Huisarts: Hoeveel jaren heb je als HIBO gewerkt:

0 / 1 (= huidige 7de jaar arts) / 2 / 3 /

HIBO:

1e (= huidige 7de jaar arts) / 2e / 3e /

Hoeveelste jaar als HIBO:

A questionnaire study to quantify and condense


the reasons for prescribing
Beschouw een patint met hoestklachten, waarbij je een luchtweginfectie vermoedt.
In welke mate ga je volgende punten na?

Voorbeeld: Als je wilt aangeven dat je weinig frequent let op de kleur van de ogen van de patint, antwoord je:
Nooit
Weinig
Soms
Vaak
Welke kleur de ogen van de patint hebben.
X
In te vullen:

Nooit

Weinig

Soms

Vaak

Altijd

Altijd

Of het plots begonnen is.


Of je de patint op huisbezoek ziet.
Of er een epidemie van luchtweginfecties is.
Of de patint ouder is dan 60 jaar.
Of de patint vlug naar de dokter gaat.
Of de patint bekend is met COPD.
Of de patint rookt.
Of de patint zich hiermee voor het eerst aanbiedt.
Of de patint er ziek uitziet.
Of de patint vertelt zich ziek te voelen.
Of de patint frequent hoest.
Of er sputum wordt opgehoest.
Of het sputum gekleurd is.
Of de klacht minder dan drie dagen bestaat.
Of er koorts is.
Of de klacht gepaard gaat met heesheid.
Of er gelokaliseerde thoraxpijn is.
Of de klachten griepaal zijn, d.w.z. spierpijn, algemene malaise
Of de patint kortademig is.
Of de patint een rode keel heeft.
Of de patint zijn ademhalingsfrequentie te snel is.
Of er bij longauscultatie verminderd ademgeruis is.
Of er bij longauscultatie wheezing te horen is.
Of er bij longauscultatie crepitaties te horen zijn.
Of er bij longauscultatie vochtige ronchi te horen zijn.
Of de patint zijn klachten spontaan gebeterd zijn.
Of de patint hoofdzakelijk in bed ligt.
Of de patint zijn algemene toestand achteruitgaat.
Of de patint meer koorts krijgt.
Of de patint eerst eigen middeltjes probeerde.
Of de patint onder antibiotica niet snel geneest.
Of de patint vraagt naar medicatie in het algemeen.
Of de patint antibiotica vraagt.
Of de patint snel beter moet zijn voor zijn werk.
Of de patint wel terugkomt als het niet betert.
Of de patint therapietrouw is.
Of je de patint kent.
Of het je kwalijk zal genomen worden, als je geen antibiotica hebt
gegeven waar het achteraf nodig bleek.
Dat je de patint toch laat terugkomen na 3 4 dagen.
Of je onder grote tijdsdruk staat.
Of de patint volgens jou antibiotica verwacht.

47

Chapter III
Beschouw een patint met hoestklachten, waarbij je een luchtweginfectie vermoedt.
Hoe sterk vr of tegen pleiten de volgende gegevens om te behandelen
met een antibioticum?
Voorbeeld: Als voor jou het hebben van blauwe ogen, sterk voor het behandelen m et een antibioticum pleit, antwoord je:
Sterk voor
Voor
N eutraal
Tegen
Sterk tegen
i.v.m . het behandelen m et een antibioticum
D at de patint blauwe ogen heeft.
X

In te vullen:

D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at
D at

48

je de patint op huisbezoek ziet.


er een epidem ie van luchtweginfecties is.
de patint jonger is dan 60 jaar.
de patint ouder is dan 60 jaar.
de patint vlug naar de dokter gaat.
de patint niet vlug naar de dokter gaat.
de patint geen CO P D heeft.
de patint bekend is m et CO P D .
de patint niet rookt.
de patint rookt.
de patint zich hierm ee voor het eerst aanbiedt.
de patint zich hierm ee voor de tweede keer aanbiedt.
de patint er niet ziek uitziet.
de patint er ziek uitziet.
de patint vertelt zich ziek te voelen.
de patint frequent hoest.
de patint een droge hoest heeft.
de klacht m inder dan drie dagen bestaat.
er sputum wordt opgehoest.
het sputum wit is.
het sputum gekleurd is.
de klacht gepaard gaat met hem optoe of bloedfluimen.
de klacht gepaard gaat met heesheid.
er wel gelokaliseerde thoraxpijn is.
er geen gelokaliseerde thoraxpijn is.
de klachten griepaal zijn, d.w.z. spierpijn, algem ene m alaise
er bij het hoesten ook hoge koorts (> 38.5C) is.
er naast het hoesten ook pijn is t.h.v. tanden, sinussen.
het hoesten gepaard gaat met meer dan drie dagen koorts.
de patint geen gezwollen cervicale lymfeklieren heeft.
de patint gezwollen cervicale lymfeklieren heeft.
de patint kortademig is.
de patint een rode keel heeft zonder witte stippen.
de patint een rode keel heeft m et witte stippen.
de patint zijn ademhalingsfrequentie te snel is.

Sterk voor
V oor
N eutraal
Tegen
Sterk tegen
i.v.m. het behandelen met een antibioticum

A questionnaire study to quantify and condense


the reasons for prescribing

Idem, in te vullen:

Sterk voor
Voor
Neutraal
Tegen
Sterk tegen
i.v.m. het behandelen met een antibioticum

Dat de longauscultatie normaal is, d.w.z. geen ronchi of crepitaties.


Dat er bij longauscultatie verminderd ademgeruis is.
Dat er bij longauscultatie wheezing te horen is.
Dat er bij longauscultatie crepitaties te horen zijn.
Dat er bij longauscultatie vochtige ronchi te horen zijn.
Dat er bij percussie demping is.
Dat het na twee dagen niet verergerd is.
Dat een kind meer koorts krijgt.
Dat het met eigen (thuis)medicatie al beter ging.
Dat de patint nieuwe klachten ontwikkelt bij een griepaal
syndroom.
Dat de patint antibiotica slecht inneemt.
Dat de patint medicatie in het algemeen vraagt.
Dat de patint geen antibiotica wil.
Dat de patint antibiotica vraagt.
Dat de ouders van een kind druk uitoefenen.
Dat je de patint kent.
Dat je gerust bent
Dat je je niet gerust voelt.
Dat als de patint niet geneest, je de patint wel zal terugzien.
Dat als de patint niet geneest, je de patint niet meer terugziet.
Dat je de patint toch doet terugkomen na 3 4 dagen.
Dat de patint al drie dagen in bed ligt met koorts.
Dat de algemene toestand van de patint achteruitgaat.
Dat de patint snel beter moet zijn voor zijn werk.
Dat het je zal kwalijk genomen worden, als je geen antibiotica hebt
gegeven, waar het achteraf nodig bleek.
Dat je onder grote tijdsdruk staat.
Dat de patint volgens jou antibiotica verwacht.
Dat zonder antibiotica de patint binnen twee dagen al zal terug
komen, als er geen beterschap is.

Nogmaals bedankt voor uw inspanning en uw tijd, nodig om deze vragenlijst in te vullen.

49

50

IV
Antibiotics for coughing in general practice:
GPs perception of patients requests determines
prescription behaviour
Introduction
In general practice, medical decisions are prompted most often by complaints
about coughing: 169 times per 1000 patients per year for a new illness episode,
especially for acute respiratory tract infections (RTIs).1 Despite the lack of
evidence to support the prescription of antibiotics for coughing in case of
suspected RTIs,2 3 antibiotics are being overprescribed for this condition,
especially in primary care.4 5 This results in an unnecessary financial burden on
both the patient and society, as well as adverse effects and development of
bacterial resistance.6 Because it is difficult to establish accurate diagnoses of
RTIs in general practice, we chose to study this acute problem in primary care
based on the most frequent symptom, i.e. coughing.7 When we explored the
diagnostic and therapeutic decisions of Flemish general practitioners (GPs)
regarding adult coughing patients by means of focus groups,8 we found
medical as well as non-medical reasons for antibiotic prescriptions in case of a
suspected RTI.9 Our hypotheses on Flemish GPs'decisions were in line with
previous research. In the case of suspected RTI, there is a low degree of
certainty in the differentiation between RTIs, e.g. between acute bronchitis and
pneumonia.10 11 Clinical signs and symptoms, i.e. medical reasons, often leave

51

Chapter IV
GPs with diagnostic uncertainty. In the end, the decision whether or not to
prescribe antibiotics is taken.12 Prescription behaviour is also determined by
doctor- and patient-related factors, e.g. patient expectations and the GP s
perception of them.13 These are clearly non-medical reasons.
A questionnaire study assessing the importance of the focus group
determinants for the decision to prescribe antibiotics confirmed that nonmedical as well as medical reasons may give rise to antibiotic treatment.14
Patients'request for an antibiotic was mentioned in the focus groups and scored
high in the questionnaire study as a non-medical reason for antibiotic
prescribing.
To validate this focus group and questionnaire study finding and to contribute
to the necessary understanding of the complex prescription decision, this study
aims to establish a valid estimate of the effect by GPs perception of patients
request for an antibiotic on their prescription of antibiotics, specifically for
patients consulting with acute cough as one of the most prominent complaints.

Methods

Design
We performed an explanatory analysis of data collected during a cluster
randomized controlled trial (cRCT)15 to obtain a valid estimate of the
relationship between the GPs perception of the patients request for an
antibiotic and their prescription of antibiotics.

Setting and sample


We approached 149 Flemish GPs not reluctant to take part in further study on
this topic at the time of the postal questionnaire study.14 Eighty-five GPs
agreed to participate in the cRCT by returning a completed and signed form to
collect their characteristics (Figure 1 and Table 1). These GPs were asked to
include 20 consecutive adult acute cough patients (inclusion criteria, see box 1)
in the periods February-April 2000 and 2001.

52

GPs perception of patients requests determines


prescription behaviour

Figure 1. Flow of general practitioners (GPs) through the study.

GPs responding in questionnaire study (n=193)

GPs reluctant to take part in further study on


this topic (n=44)

GPs approached for this study (n=149)

GPs not willing to take part in this study


(n=64)

GPs recruited for this study (n=85)

GPs not including patients for this study (n=13)

GPs including patients for this study (n=72)

Box 1. Inclusion criteria of adult acute cough patients

18 - 65 years old
No compromised immunity
New or worsening cough
- present for less than 30 days
- as (one of the) most prominent complaint(s)
- as reason for first encounter at the GPs practice

53

Chapter IV
Table 1 Characteristics of participating general practitioners (1).

Variable

N = 72
Figures are numbers (percentages)

(2)

MEN

53

(74)

UA GRADUATES

32

(44)

PROFFESIONAL TRAINING

19

(28)

FEE FOR SERVICE

69

(96)

NOT SINGLE-HANDED

35

(49)

16

(22)

13

(19)

GPPTs IN PRACTICE

(3)

PART TIME
PROVINCE
Antwerp

44

(61)

Brussels

(3)

Limburg

(7)

West Flanders

12

(17)

(13)

PEAKFLOW METER

East Flanders

67

(93)

SPIROMETER

16

(22)

TRAINING PRACTICE

24

(33)

ACADEMIC LINK

23

(32)

RECORDS OF HOME VISITS

59

(82)

COMPUTERIZED RECORDS

51

(71)

(3)

COMPLEMENTARY MEDICINE
Figures are means (standard deviation)
AGE AT START OF THIS STUDY

44

(8)

104

(42)

HOME VISITS PER WEEK

34

(19)

MEDICAL REPRESENTATIVES PER MONTH

15

(10)

15

(7)

(1)

PATIENT ENCOUNTERS PER WEEK

ATC J COST RATIO

(4)

ATC J VOLUME RATIO (4)

(1) General practitioners including the 1448 patients eligible for analysis. Their characteristics do not
differ from the characteristics of the other GPs who agreed to participate in this study (n=13).
(2) Denominators vary due to missing values
(3) General Practitioners in Professional Training
(4) The ratios the gross amount for antimicrobials for systemic use (ATC J)/ the gross amount for all
pharmaceutical specialties and the volume (Daily Defined Dosage) DDD)) of ATC J/ the volume for
all pharmaceutical specialties are both expressed as percentages on individual prescribing feedback
from the National Sickness and Invalidity Insurance Institution (NSIII) to GPs. We asked the
participating GPs for these percentages and calculated their mean.

54

GPs perception of patients requests determines


prescription behaviour

Data Collection and Measures


The GPs were asked to use pre-printed forms to record patient demographics,
the presence of co-morbidity and risk, symptoms, signs, as well as the
circumstances of the consultation, the tests ordered, and the prescription of a
follow up visit, of a referral and of an antibiotic (Table 2). For co-morbidity
and risk, symptoms, signs, and for the circumstances requests for antibiotics
and requests for other medication the three possible answer categories were
yes, no, and don t know. For the circumstances workload and impression
the three categories were low, high, and very high, and not ill, ill, and very ill,
respectively. For tests ordered and prescriptions , there were only two
possible answer categories, viz. yes and no.

Statistical Analyses
We calculated frequency distributions of individual variables and assessed
univariate associations between each variable and the prescription of
antibiotics. Continuous variables were converted to categorical data for the
univariate analyses but treated as continuous data for the multivariate analysis.
We developed a model to obtain a valid estimate of the relationship between
the GPs perception of the patients request for an antibiotic and their
prescription of antibiotics. We used a hierarchical backwards elimination
procedure described by Kleinbaum,16 taking clustering of the data into account
(see box 2). Before starting the procedure, all categorical variables were
dichotomised. Since we aimed to estimate the effect of the perception of a
request for an antibiotic, and since we wanted to control the above relationship
only for the presence of the other covariates, we dichotomised co-morbidity
and risk, symptoms, signs, and the circumstance requests antibiotics and
requests other medication by recoding don t know into no, and the
circumstances workload and impression by recoding very high into high,
and very ill into ill, respectively. To deal with collinearity, the variables
reduced breathing sounds , wheezing , ronchi , and crepitations were
replaced by the variable lung auscultation , representing the number of
abnormal auscultatory findings, while a new dichotomous variable higher risk
was created based upon Fine s prediction rule to identify patients with
community-acquired pneumonia at low risk for mortality or complications
(Table 2).17

55

Chapter IV
Statistical analyses were performed with SAS statistical software.18

Results
72 GPs participated in the study (Fig. 1), with 1448 patients eligible for
analysis. According to these GPs 218 (15%) asked for an antibiotic and 500
(35%) were prescribed an antibiotic (Table 2).

Univariate analysis
The prescription of antibiotics was associated with medical and non-medical
information. The categorical variables and conversion of the continuous
variables into categorical data reveals that antibiotics were prescribed more
often for the oldest patients, in cases of prolonged coughing and more
abnormal auscultatory findings. There is also a higher incidence of antibiotic
prescription among younger GPs, those with the highest number of patient
encounters and home visits on a weekly basis, GPs seeing a high number of
medical representatives per month, and GPs prescribing relatively more
antimicrobials for systemic use (ATC J) in cost and in volume (ATC J cost
ratio and ATC J volume ratio respectively) (Table 3). In the event of a medical
history of COPD, a patient reporting a feeling of sickness, symptoms like
sputum, fever and shortness of breath, or signs such as percussion dullness,
antibiotics were prescribed more often too (Table 2). Likewise, there was a
strong link between the GP s perception of a patient s request for an antibiotic
and the prescription of one (OR = 4.64 (95% CI: 2.96-7.26)) (Table 2).
For these associations, the dependence of a pair of responses belonging to the
same cluster was highly significant, with the intra-cluster correlation
coefficient being 0.20 on average.

56

GPs perception of patients requests determines


prescription behaviour
Box 2 Hierarchical backwards elimination procedure and cluster data
To obtain a valid estimate of the relationship between the GPs perception of the
patients request for an antibiotic (E) and the prescription of antibiotics by GPs (X), we
estimated a logistic model which contained all covariates (Vs) as possible
confounders and all E*V interaction terms as possible effect modifiers: logit P(X) = +
E+
V +
E*V. The covariates were the other information the GPs recorded
about the patients, as well as their characteristics. We also added the interaction
terms of gender and age, and gender and year of birth, respectively, as well as the
variables year, group and year*group to control for the cRCT design. If some of the Vs
or E*Vs dropped out of the starting model due to collinearity, confounding and
interaction were evaluated in a stratified analysis of E versus X controlling for each V
separately.
First, interactions were assessed by eliminating one by one the interaction term with
least significant type 3 score statistics. Only significant E*Vs were retained in a gold
standard model. P-values of E*V parameter estimates would be considered significant
if smaller than 0.01 instead of smaller than 0.05 only if necessary for a clear
interpretation of the effect of E on X.
Second, the confounding effect of all Vs not in significant E*Vs in the full model was
assessed, followed by precision considerations. We looked for a subset of V'
s for
which the model gave roughly the same parameter estimates for E and the significant
E*Vs, but with narrower confidence intervals.
We adjusted logistic regression estimates for clustering within our data (patients are
19
nested within GPs). We used alternating logistic regression (ALR), a technique
closely related to generalized estimating equations (GEE).20 With ALR, estimating
equations are specified for marginal and association parameters. The association
between pairs of responses is measured by log odds ratios( ), instead of correlations
as with ordinary GEE. The advantage of ALR over GEE is that the association
between measurements can be modelled and that uncertainty measures are attached
to the estimated dependence parameters. Using GEE we can also provide an order of
21
magnitude for the intra-cluster correlation coefficient ( ) which is related to in ALR.
22-25

The marriage of GEE or ALR with goodness-of-fit (GOF) is not an easy one.
Our
approach to fit a broader model with interactions and to test whether the additional
25
terms are significant is regarded as an appropriate way to determine the fit as well.
Furthermore, an extension of the Hosmer and Lemeshow GOF statistic to marginal
regression models for repeated binary responses was used to determine whether the
model fits the data.25 In order to assess the significance of the proposed GOF statistic
score statistics were used. A significant score statistic indicates that the proposed
model leaves a substantial amount of variability in the data not taken into account. The
original Hosmer and Lemeshow GOF statistic for ordinary logistic regression by
assessing agreement between predicted and observed risk by decile of predicted risk
26
was also used.

57

58
(75)
(33)
(62)
(57)
(36)
(61)
(28)
(38)
(44)
(34)
(52)
(42)
(2)

274
295
125
253
374
166
306
281
175
299
138
185
215
164
255
212
9

HIGH WORKLOAD
IMPRESSION OF SICKNESS
REQUEST FOR ANTIBIOTICS
REQUEST FOR MEDICATION

DURATION OF COUGHING
SPUTUM
FEVER
RUNNY NOSE
HEADACHE
MUSCLE ACHE
SOAR THROAT
WHEEZING
SHORT OF BREATH
CHESTPAIN
LOSS OF APPETITE
LIMITED ACTIVITY

13
69

REFERRAL
FOLLOW UP CONTACT

Prescriptions

21
7
17
6
11

RADIOGRAPH
SPUTUM ANALYSIS
INFECTION PARAMETERS
SEROLOGY
OTHER

Test ordering

Higher risk(4)
Number of abnormal auscultatory finding(5)
PERCUSSION DULNESS

Signs

Symptoms

Circumstances

21
68

(3)
(14)

(2)
(8)

(2)
(1)
(1)
(1)
(3)

22
8
13
12
28

(4)
(1)
(4)
(1)
(2)

(34)
(1)

323

(49)
(26)
(62)
(48)
(33)
(56)
(15)
(23)
(29)
(22)
(41)

(58)
(35)
(10)
(51)

538
319
93
480
458
243
581
446
301
525
136
216
263
194
374

(9)
(6)
(2)
(1)
(5)
(30)

(43)

87
57
18
8
50
281

391

(55)
(61)
(25)
(52)

(10)
(11)
(3)
(1)
(3)
(37)

51
54
17
6
17
183

ASTHMA
COPD (CARA)
ACE-INHIBITOR
ASPIRATION RISK
TROMBO-EMBOLIC RISK
SMOKING

Co-morbidity & Risk

(47)

220

AGE
MEN

(1,03 to 1,07)
(2,32 to 3,97)
(1,17 to 1,89)
(0,79 to 1,37)
(1,12 to 1,71)
(1,01 to 1,59)
(0,77 to 1,19)
(1,78 to 3,07)
(1,85 to 3,05)
(1,61 to 2,52)
(1,69 to 2,87)
(1,47 to 2,52)

1,05
3,04
1,49
1,04
1,38
1,27
0,96
2,34
2,38
2,02
2,20
1,92

(1,49 to 6,03)
(0,97 to 8,69)
(1,37 to 10,11)
(0,54 to 5,49)
(0,73 to 3,06)
1,71 (0,98 to 3,00)
3,06 (1,99 to 4,69)

2,99
2,90
3,72
1,72
1,50

1,38 (1,11 to 1,71)


2,53 (2,02 to 3,18)
6,15 (0,71 to 53,21)

(0,88 to 1,47)
(2,29 to 4,04)
(2,96 to 7,26)
(0,74 to 1,26)

(0,97 to 1,76)
(1,57 to 3,64)
(1,01 to 4,49)
(1,87 to 4,34)
(0,69 to 1,64)
(1,25 to 1,95)

1,14
3,04
4,64
0,97

1,31
2,39
2,13
2,85
1,07
1,56

1,01 (1,00 to 1,02)


1,07 (0,90 to 1,28)

(0,99 to 1,10)
(1,62 to 3,92)
(0,67 to 1,83)
(0,60 to 1,53)
(1,03 to 2,37)
(0,48 to 1,17)
(0,77 to 1,96)
(0,34 to 1,17)
(0,68 to 2,37)
(1,07 to 2,65)
(0,74 to 2,02)
(0,87 to 3,42)

(0,68 to 1,59)
(1,47 to 3,53)
(8,86 to 48,99)
(0,62 to 1,54)

(0,21 to 1,26)
(0,35 to 1,69)
(0,26 to 4,19)
(1,24 to 176,71)
(0,15 to 2,08)
(0,59 to 1,68)

(0,58 to 10,06)
(0,07 to 2,82)
(2,86 to 274,54)
(0,01 to 1,31)
(0,56 to 5,56)
0,04 (0,01 to 0,14)
2,18 (0,81 to 5,85)

2,41
0,45
28,04
0,11
1,76

1,21 (0,78 to 1,87)


3,04 (2,03 to 4,54)
1,46 (0,15 to 13,80)

1,04
2,52
1,11
0,96
1,57
0,75
1,22
0,63
1,27
1,68
1,22
1,73

1,04
2,28
20,83
0,98

0,52
0,77
1,05
14,79
0,55
1,00

1,02 (1,00 to 1,04)


2,20 (0,73 to 6,59)

Table 2 Predictors of the prescription of antibiotics by Flemish general practitioners (GPs) in adults acute cough patients.
Figures are numbers (percentage) of adults(1)
Not prescribed
Crude odds ratio
Prescribed
Adjusted odds ratio(2)
(95% CI)(3)
(95% CI)(3)
antibiotics
antibiotics
Demographics

Chapter IV

Antwerp
Brussels
Limburg
West Flanders
East Flanders

269
41
35
98
57
483
93
137
126
376
333
8

198
111
497
199
66
59
(54)
(8)
(7)
(20)
(11)
(97)
(19)
(27)
(25)
(75)
(67)
(2)

(40)
(24)
(99)
(41)
(13)
(12)

Prescribed
antibiotics

(51)
(43)
(43)

(71)
(3)
(4)
(10)
(13)
(95)
(32)
(37)
(36)
(85)
(70)
(3)

(46)
(29)
(98)
(48)
(21)
(21)
(1,16 to 9,84)
(0,86 to 5,89)
(1,25 to 3,43)
(0,60 to 2,11)
(0,98 to 2,96)
(0,37 to 0,99)
(0,46 to 1,13)
(0,46 to 1,11)
(0,35 to 1,07)
(0,65 to 1,80)
(0,40 to 0,70)
(0,99 to 1,03)
(1,00 to 1,01)
(1,01 to 1,03)
(0,98 to 1,04)
(1,02 to 1,08)
(1,19 to 1,53)

(0,86 to 2,02)
(0,53 to 1,33)
(0,56 to 1,29)
(2,64 to 6,23)
(0,48 to 1,15)
(0,40 to 1,07)
(0,32 to 0,86)

1,76 (1,30 to 2,38)


1,00 (1,00 to 1,01)
1,01 (1,00 to 1,01)

0,64 (0,52 to 0,80)


0,99 (0,63 to 1,55)
0,57 (0,40 to 0,82)

1
3,38
2,24
2,07
1,12
1,70
0,61
0,72
0,71
0,61
1,08
0,53
1,01
1,01
1,02
1,01
1,05
1,35

1,31
0,84
0,85
4,06
0,74
0,65
0,52

Crude odds ratio


(95% CI)(3)

(4,85 to 176,71)
(0,88 to 8,63)
(1,60 to 9,76)
(0,37 to 9,50)
(0,25 to 46,94)
(0,21 to 0,90)
(0,61 to 2,85)
(0,67 to 3,10)
(0,68 to 2,81)
(0,60 to 2,72)
(0,04 to 1,15)
(1,01 to 1,47)
(0,99 to 1,02)
(1,00 to 1,06)
(0,98 to 1,10)
(0,84 to 1,06)
(0,97 to 2,57)

0,23 (0,13 to 0,41)


0,98 (0,96 to 1,01)
0,90 (0,76 to 1,06)

(0,04 to 4,93E+6)
(0,71 to 4,09)

(0,10 to 0,42)
(0,10 to 26,09)
(0,18 to 1,44)
(0,28 to 4,36)
(0,35 to 5,16)

0,74 (0,42 to 1,28)


0,75 (0,32 to 1,78)
0,72 (0,23 to 2,29)

1
29,20
2,75
3,95
1,88
3,43
0,44
1,32
1,44
1,38
1,28
0,20
1,22
1,00
1,03
1,04
0,94
1,58

456,00
1,70
0,20
1,63
0,51
1,11
1,35

Adjusted odds ratio(2)


(95% CI)(3)

(1) Denominators vary due to missing values


(2) Adjusted for all variables in the final model
(3) Score test using Alternative Logistic Regression
means p<0.05
,
(4) Patient age > 50 or patient has congestive heartfailure, cerebrovascular disease, liver disease, kidney disease o r neoplastic disease, or has altered consciousness, pulse rate>125/'
respiratory rate>30/'
, temperature>38C or systolic blood pressure>90 mmHg
(5) Less vesicular breathing, wheezing, ronchi or crepitations
(6) The ratios of the gross amount for a ntimicrobials for systemic use (ATC J)/the gross amount for all pharmaceutical specialties and the volume (Daily Defined Dosage (DDD)) of ATC J/the
volume for all pharmaceutical specialties are both expressed as percentages on individual prescribing feedba ck from the National Sickness and Invalidity Institution (NSII) to GPs.

481
406
206

670
28
37
93
120
904
303
355
342
808
663
29

433
259
926
451
197
198

Not prescribed
antibiotics

190 (38)
REGISTRATION YEAR
227 (45)
REGISTRATION GROUP
78 (41)
YEAR*GROUP
Interaction terms final model
REQUEST FOR ANTIBIOTICS*Number of abnormal auscultatory findings
PATIENT GENDER*AGE
GP'
S GENDER*YEAR OF BIRTH

Design

PEAKFLOW METER
SPIROMETER
TRAINING PRACTICE
ACADEMIC LINK
RECORDS OF HOME VISITS
COMPUTERISED RECORDS
COMPLEMENTARY MEDICIN
YEAR OF BIRTH
AVERAGE NUMBER OF PATIENT ENCOUNTER PER WEEK
AVERAGE NUMBER OF HOME VISITS PER WEEK
AVERAGE NUMBER OF MEDICAL REPRESENATIVES/MONTH
ATC J COST RATIO(6)
ATC J VOLUME RATIO(6)

MEN
UA GRADUATE
PROFESSIONAL TRAINING
FEE FOR SERVICE
NOT SINGLE-HANDED
GPPTs IN PRACTICE
PART TIME
PROVINCE

GPs characteristics

Table 2 continued.

GPs perception of patients requests determines


prescription behaviour

59

Chapter IV

Multivariate analysis
Some variables dropped out of the model due to collinearity. Of these, only the
GPs year of birth, the number of patient encounters per week and ATC J cost
ratio were significant effect modifiers of the univariate relation between the
GPs perception of a patient s request for antibiotics and a prescription of
antibiotics. For younger GPs (year of birth 65 vs. 45), GPs with fewer patient
encounters per week (80 vs. 150) and GPs with a higher ATC J cost ratio (.20
vs. .10), the patients requests were even more strongly associated with
antibiotic prescription.
This resulted in a model containing 7 interaction terms (patient age, smoking,
number of abnormal auscultatory findings, GPs university of graduation, parttime working status, registration group (control vs. intervention) and
registration year (2000 vs. 2001)) (p<0.05). After eliminating interaction terms
with a p-value greater than 0.01, only one interaction term was retained in the
model. In order to obtain a comprehensible and valid estimate of the effect of
GPs perception of patients requests on their prescribing antibiotics, the final
model controls for this interaction term and all possible confounders (Table 2).
This model fits the data well (GOFHorton = 0.71;GOFHosmer-Lemeshow = 0.72). For
the patients in the final models (n=819) the univariate association between the
GP s perception of the patients request for antibiotics and the prescription of
an antibiotic (OR = 4.60 (2.59-8.17)) was very similar to that for all patients
(n=1448).
Because of the introduction of GPs characteristics in this model the
dependence for a pair of responses belonging to the same cluster was no longer
significant, the intra-cluster correlation coefficient being 0.02.
From the final model we learn that the GPs perception of the patients request
for antibiotics is still significantly associated with the prescription of
antibiotics. This association is independent of the other information the GPs
recorded and of their characteristics (Table 2). Significant confounders of this
association are aspiration risk (ORadj = 14.79 (1.24-176.71)), an impression of
sickness (ORadj = 2.28 (1.47-3.53)), the presence of sputum (ORadj = 2.52
(1.62-3.92)), of a headache (ORadj = 1.57 (1.03-2.37)), of thoracic pain (ORadj =
1.68 (1.07-2.65)), the number of abnormal auscultatory findings ((ORadj = 3.04
(2.03-5.54)), investigating infection parameters (ORadj = 28.04 (2.86-274.54)),

60

GPs perception of patients requests determines


prescription behaviour
patient referral (ORadj = 0.04 (0.01-0.14)), the age of the GPs at the start of this
study (ORadj = 1.22 (1.01-1.47)), previous professional training (ORadj = 0.20
(0.10-0.42)), having a spirometer in the practice (ORadj = 0.44 (0.21-0.90)), and
the practice location (Figure 2).
However, the effect of patients requesting antibiotics on GP s prescriptions of
antibiotics depends on the outcome of the lung auscultation. When a patient is
perceived to be requesting antibiotics, they were prescribed significantly more
often when the lung auscultation results were normal (ORadj = 20.83 (8.8648.99) (Figure 2), or in case of only one abnormal auscultatory finding (4.79
(2.16 to 10.60)).
In case of a normal lung auscultation the adjusted predicted probability for an
antibiotic prescription is 0.09 (95% CI 0.02-0.30) if no request for antibiotics is
perceived compared to 0.84 (0.52-0.97) if a request is perceived. If only one
abnormal auscultatory finding is present these probabilities are 0.16 (0.030.53), 0.98 (0.92-1.00), respectively. If there is more than one abnormal
auscultatory finding there is no relevant and significant difference between a
request for an antibiotic and the absence of a request.

Discussion
The GPs perception of patient requests for antibiotics is significantly
associated with the prescription of antibiotics to adult acute cough patients,
even when controlling for the other information the GPs recorded and for their
characteristics. Antibiotics were prescribed significantly more often when a
patient was perceived to request an antibiotic and the lung auscultation was
normal or revealed only one abnormal finding. Aspiration risk, an impression
of sickness, the presence of sputum, of headache, and of thoracic pain,
abnormal auscultatory findings, investigating infection parameters and a
younger GP were also associated with increased antibiotic prescribing.
Prescribing depended on the GP s practice location as well. Patient referral,
previous professional training and having a spirometer in the practice were
associated with less prescribing.

61

62
Demographics
18-35
36-45
46-55
56-65

216
128
98
49
9

190
115
81
27
87

176
128
97
95

(43)
(26)
(20)
(10)
(2)

(38)
(23)
(16)
(5)
(17)

(35)
(26)
(20)
(19)

74-65
71 (14)
64-55
210 (42)
54-45
202 (40)
44-25
44-25
17
(3)
AVERAGE NUMBER OF PATIENT ENCOUNTER PER WEEK
AVERAGE NUMBER OF PATIENT ENCOUNTER PER WEEK (3)
<51
56 (11)
51-100
182 (36)
101-150
189 (38)
151-200
48 (10)
25
(5)
>200

YEAR OF BIRTH
YEAR OF BIRTH (3)

GPs characteristics

Number of abnormal auscultatory finding(4)


0
1
2
3
4

Signs

1-3
4-6
7-9
10-12
>12

DURATION OF COUGHING
DURATION OF COUGHING (3)

Symptoms

AGE
AGE (3)

Prescribed
antibiotics

(10)
(46)
(38)
(6)

(10)
(56)
(23)
(9)
(2)

98
527
219
86
18

(77)
(14)
(6)
(3)
(0)

(48)
(24)
(13)
(4)
(10)

(41)
(25)
(19)
(15)

98
432
364
54

729
129
60
27
3

451
232
126
41
98

385
239
178
141

Not prescribed
antibiotics

1,01
0,88
1
0,67
0,74
0,46
1,01
1,34
1
0,62
1,36
1,26
2,35

2,53
1
3,49
6,85
11,80
25,73

1,05
1,22
1
1,28
1,80
1,76
2,12

1,01
1,11
1
1,16
1,19
1,40

(0,38 to 1,03)
(0,76 to 2,43)
(0,58 to 2,71)
(1,59 to 3,49)

(0,41 to 1,09)
(0,45 to 1,21)
(0,28 to 0,74)
(1,00 to 1,01)
(1,07 to 1,69)

(0,99 to 1,03)
(0,70 to 1,11)

(2,50 to 4,87)
(4,26 to 11,02)
(6,06 to 22,99)
(1,00 to 660,11)

(2,02 to 3,18)

(1,01 to 1,64)
(1,32 to 2,46)
(1,00 to 3,10)
(1,51 to 2,98)

(1,03 to 1,07)
(1,12 to 1,32)

(0,94 to 1,44)
(0,92 to 1,52)
(1,05 to 1,86)

(1,00 to 1,02)
(1,02 to 1,21)

Crude odds ratio


(95% CI)(2)

Table 3 Predictors of the prescription of antibiotics by Flemish general practitioners (GPs) in adults acute
cough patients.
Categorical variables and continuous variables converted into categorical data.
Figures are numbers (percentage) of adults(1)

Chapter IV

Prescribed
antibiotics

(12)
(20)
(36)
(14)
(18)

(11)
(27)
(17)
(26)
(19)

(11)
(30)
(29)
(14)
(16)

(17)
(41)
(25)
(7)
(9)

105
180
328
130
168

104
244
155
233
171

94
266
257
127
141

146
358
220
61
82

Not prescribed
antibiotics
1,02
1,38
1
1,21
1,26
2,42
3,14
1,01
1,21
1
1,69
1,49
2,62
2,23
1,05
1,34
1
2,08
2,88
3,92
3,96
1,35
1,44
1
2,46
2,40
5,27
5,42
(1,23 to 4,94)
(1,10 to 5,24)
(2,28 to 12,16)
(2,74 to 10,71)

(0,93 to 4,63)
(1,28 to 6,45)
(1,73 to 8,86)
(1,91 to 8,21)
(1,19 to 1,53)
(1,24 to 1,66)

(0,77 to 3,69)
(0,70 to 3,15)
(1,18 to 5,82)
(0,85 to 5,84)
(1,02 to 1,08)
(1,15 to 1,56)

(0,56 to 2,61)
(0,65 to 2,46)
(1,14 to 5,17)
(1,54 to 6,43)
(0,98 to 1,04)
(1,01 to 1,46)

(1,01 to 1,03)
(1,17 to 1,63)

Crude odds ratio


(95% CI)(2)

(5) The ratios of the gross amount for antimicrobials for systemic use (ATC J)/the gross amount for all pharmaceutical specialties and the volume
(Daily Defined Dosage (DDD)) of ATC J/the volume for all pharmac eutical specialties are both expressed as percentages on individual prescribing
feedback from the National Sickness and Invalidity Institution (NSII) to GPs.

(2) Score test using Alternative Logistic Regression


means p<0.05
(3) Categorical variable or continuous variable converted to categorical variable analysed as an ordinal variable not as a class variable
(4) Less vesicular breathing, wheezing, ronchi or crepitations

(1) Denominators vary due to missing values

AVERAGE NUMBER OF HOME VISITS PER WEEK


AVERAGE NUMBER OF HOME VISITS PER WEEK (3)
31
(6)
<11
11-20
62 (13)
21-30
123 (25)
31-40
83 (17)
>40
193 (39)
AVERAGE NUMBER OF MEDICAL REPRESENATIVES/MONTH
AVERAGE NUMBER OF MEDICAL REPRESENATIVES/MONTH (3)
0(!) -5
0(!)-5
28
(6)
5-10
95 (21)
11-15
58 (13)
16-20
183 (41)
>20
84 (19)
ATC J COST RATIO(5)
ATC J COST RATIO(3) (5)
20
(4)
<8,1
94 (20)
8,1-12
128 (28)
12,1-16
125 (27)
16,1-20
95 (21)
>20
ATC J VOLUME RATIO(5)
ATC J VOLUME RATIO(3) (5)
<2,1
31
(7)
2,1-3
140 (32)
3,1-4
111 (25)
68 (16)
4,1-5
>5
87 (20)

GPs characteristics

Table 3 continued.

GPs perception of patients requests determines


prescription behaviour

63

Chapter IV

Figure 2 The relation between the effect of physicians perception of patients


requests on antibiotic prescribing for acute cough and the significant confounders of
this relation. Adjusted odds ratios and 95% confidence limits.
Investigating infection parameters
Request for antibiotics
Aspiration risk
# abnormal ausculatory findings
Sputum
Impression of sickness
Chestpain
Headache
Physician'
s year of birth
Physician'
s practice location
Antwerp (reference)
Brussels
Limburg
West Flanders
East Flanders
Spirometer
Professional training
Referral
0,001

64

0,01

0,1

10

100

1000

GPs perception of patients requests determines


prescription behaviour

Study limitations
It is possible that the results are biased due to the recruitment, non-response
and response quality of the GPs. The GPs recruited, however, did not differ
from the other 64 GPs approached for this study nor from the other 108 GPs
responding in the questionnaire study in terms of age, university of graduation,
the number of GPs and general practitioners in professional training (GPPTs)
in the practice, the number of GPs rewarded fee for service, and the average
number of patient encounters per week. Although among the participants there
was a higher proportion of males (63/85 vs. 36/64: p=0.02, vs. 64/108:p=0.03
respectively), their age and gender distribution is similar to national averages.
Furthermore, in terms of all recorded characteristics, the GPs who included the
1448 patients eligible for analysis did not differ from the other GPs willing to
participate in the cRCT (n=13) nor did those with differ from those without
patients in the final model.
Patient characteristics might also influence the results of this study. Face-toface interviews of two samples of about 1000 people representative of the
population over fourteen also that showed about fifteen percent of the
respondents would request the GP for an antibiotic in Belgium
(www.health.fgov.be/antibiotics/cabn.htm). There were no differences in the
patients characteristics between those records included in the analysis
(n=1448) and those records eligible for the cRCT15 (but excluded here because
of incomplete data for the GP s perception of the patient s request for an
antibiotic), or between those records included and those left out of the final
model. Furthermore, the crude odds ratio (95 % CI) for all eligible patients (N
= 1448) and that of all patients with complete information (N = 819) are 4.64
(2.96-7.26) and 4.60 (2.59-8.17), respectively, suggesting there is no bias due
to the selection of patients with complete information in the final model.
There was no formal assessment of either the internal validity or the reliability
of the pre-printed forms; rather, they were assumed acceptable as they had
been developed based upon our focus group study9 and questionnaire study.14
Self-reporting might have limited our data as well, by underestimating the
importance of the non-medical reasons. Still, our data show that a non-medical
reason is independently associated with increased prescription of antibiotics.

65

Chapter IV

Acute cough
Knowledge about the determinants that play an important role in clinicians
decisions to prescribe antibiotics for respiratory infections is useful in
designing interventions to decrease inappropriate antibiotic prescribing.27 For
this study, we chose to include patients with acute cough. Although it is the
most frequent complaint in general practice there is much uncertainty as to the
diagnosis and treatment of patients with an acute cough. Surprisingly little
evidence is available to support decisions concerning these patients in daily
practice.28 It is therefore not surprising that diagnostic labels such as acute
bronchitis are used inconsistently in general practice.29 To avoid
misclassification, rather broad definitions such as lower respiratory tract
illness are used in patients with an acute cough.30 When applying the criteria
described by Hopstaken31 to our data, it was found that 624 acute cough
patients (43%) had a lower respiratory tract infection (LRTI). Even more
patients met the criteria described by Holmes and MacFarlane30 32 (n=1112
(77%). In about 60% of patients with a LRTI requesting for an antibiotic the
lung auscultation was normal or revealed only one abnormal finding. And the
effect of requesting an antibiotic on the prescription of antibiotics did not differ
between patients with or without a LRTI. This suggests not only that our
findings also apply to patients with a LRTI as defined in the literature, but also
that choosing to study patients with acute cough is more than justified. After
all, there is only little agreement between the classification of patients with
acute cough when applying the above criteria to define patients with a LRTI.

Non-medical reasons
Other non-medical reasons have been shown to affect prescribing behaviour of
GPs as well, such as patient expectations 33 34 and, to a greater extent, GPs
perception of patient expectations.35 36 However, when adjusting for medical
reasons associated with the prescription of antibiotics, non-medical reasons
have not until now been shown to determine GPs prescribing of antibiotics.27
In this study we aimed to obtain a valid estimate of the effect of GPs
perception of patients requests for an antibiotic on their prescribing antibiotics
to adult acute cough patients. We used a hierarchical backwards elimination
procedure, starting with interaction assessment.16 Only one interaction term
was retained in the final model for a clear interpretation of the estimate of X.
The more conservative model contained seven interaction terms. It would seem

66

GPs perception of patients requests determines


prescription behaviour
that the only meaningful description of these results is that the lower the
number of auscultatory findings the more a perceived request favours antibiotic
prescribing. In previous studies we also found that a positive lung auscultation
in itself was associated with antibiotic prescribing9 and that diagnostic
uncertainty influenced the complex decision-making process.14 The results of
this study support these findings and contribute to the understanding of the
prescription decision. After all, it sounds sensible that more abnormal
auscultatory findings provide GPs with more certainty regarding the need of an
antibiotic, and vice versa. Antibiotics are thus prescribed for medical reasons if
these are available, but when GPs have to deal with diagnostic uncertainty nonmedical reasons favour antibiotic prescribing. This can also be attributed to the
so-called chagrin factor.9 37 In the present study GPs consider it less appropriate
not to prescribe antibiotics when a patient requests them since this causes more
chagrin, even if the evidence shows that the limited benefit for patients with
acute (productive) cough is outweighed by the side effects,2 3 and the
medicalising effect, the financial costs and the effect on antimicrobial
resistance.6

Optimising the prescription of antibiotics


A good clinical practice guideline and an intervention to optimise antibiotic
prescribing for acute cough in Flemish general practice have taken non-medical
reasons into account. We performed a multifaceted intervention, including
educational outreach visits (academic detailing),38 to implement a guideline for
acute cough,15 since this might be most effective.39 40 Planning other clusterrandomised trials on this topic one should take into account an intra-cluster
correlation coefficient of 0.20 to adjust power calculations.

Conclusion
This study enabled us to obtain a valid estimate of the effect of GPs perception
of patients requests for antibiotics on the prescription of antibiotics and
confirmed our focus group study and questionnaire study findings, i.e. that
GPs decisions to prescribe antibiotics are determined by both medical and
non-medical reasons.9 14 Hence, good clinical practice guidelines and
interventions to optimise antibiotic prescribing have to take into account nonmedical reasons in the prescription of antibiotics.

67

Chapter IV
References
1. Okkes I, Oskam S, Lamberts H. Van klacht naar diagnose [From complaint
to diagnosis]. Bussum: Coutinho, 1998.
2. Fahey T, Stocks N, Thomas T. Quantitative systematic review of
randomised controlled trials comparing antibiotic with placebo for acute cough
in adults. BMJ 1998;316:906-10.
3. Smucny J, Fahey T, Becker L, Glazier R, McIsaac W. Antibiotics for acute
bronchitis. In: The Cochrane Library, Issue 4, 2002. Oxford: Update Software.
4. Wise R, Hart T, Cars O, Streulens M, Helmuth R, Huovinen P, Sprenger M.
Antimicrobial resistance. Is a major threat to public health [editorial]. BMJ
1998;317:609-10.
5. Koninklijke Academie voor Geneeskunde van Belgi [Belgian Royal
Academy for Medicine]. Advies inzake het overgebruik van antibiotica
[Advise concerning the overuse of antibiotics]. Tijdschr Geneesk 1999;55:1734.
6. Butler C, Rollnick S, Kinnersley P, Jones A, Stott N. Reducing antibiotics
for respiratory tract symptoms in primary care: consolidating '
why' and
considering '
how'
. Br J Gen Pract 1998;48:1865-70.
7. Coenen S, Van Royen P, Denekens J. Diagnosis of Acute Bronchitis [letter;
see reply]. J Fam Pract 1999;48:741-2.
8. Coenen S, van Royen P, Denekens J. Reducing antibiotics for respiratory
tract symptoms in primary care: '
why'only sore throat, '
how'about coughing?
[letter]. Br J Gen Pract 1999;49:400-1.
9. Coenen S, Van Royen P, Vermeire E, Hermann I, Denekens J. Antibiotics
for coughing in general practice: a qualitative decision analysis. Fam Pract
2000;17:380-5.
10. Metlay J, Kapoor W, Fine M. Does This Patient Have CommunityAcquired Pneumonia? Diagnosing Pneumonia by History and Physical
Examination. JAMA 1997;278:1440-5.

68

GPs perception of patients requests determines


prescription behaviour
11. Zaat J, Stalman W, Assendelft W. Hoort, wie klopt daar? Een
systematische literatuurstudie naar de waarde van anamnese en lichamelijk
onderzoek bij verdenking op een pneumonie [Listen, who is knocking? A
systematic review on the value of history and physical examination in case of
suspected pneumonia]. Huisarts en Wetenschap 1998;41:461-9.
12. Kassirer J. Our stubborn quest for diagnostic certainty. A cause of
excessive testing. NEJM 1989;320:1489-91.
13. Butler CC, Rollnick S, Pill R, Maggs-Rapport F, Stott N. Understanding
the culture of prescribing: qualitative study of general practitioners'and
patients'perceptions of antibiotics for sore throats. BMJ 1998;317:637-42.
14. Coenen S, Michiels B, Van Royen P, Van der Auwera J-C, Denekens J.
Antibiotics for coughing in general practice: a questionnaire study to quantify
and condense the reasons for prescribing. BMC Fam Pract 2002;3:16 (10p).
15. Coenen S, Van Royen P, Michiels B, Denekens J. Promotion of rational
antibiotic use in Flemish general practice: implementation of a guideline for
acute cough [abstract]. Prim Care Respir J 2002;11:56.
16. Kleinbaum D. Logistic regression: A self-learning text. New York:
Springer-Verlag Publisher, 1994.
17. Fine M, Auble T, Yealy D, Hanusa B, Weisfeld L, Singer D, et al. A
prediction rule to identify low-risk patients with community-acquired
pneumonia. NEJM 1997;336:243-50.
18. SAS System for Windows [program]. 8.02 version. Cary, NC: SAS Institute
Inc., 2001.
19. Wears R. Advanced statistics: Statistical methods for analyzing cluster and
cluster-randomized data. Acad Emerg Med 2002;9:330-41.
20. Liang K, Zeger S. Longitudinal data analysis using generalized linear
models. Biometrika 1986;73:13-22.
21. Ridout M, Demtrio C, Firht D. Estimating intraclass correlation for binary
data. Biometrics 1999;55:137-48.

69

Chapter IV
22. Rotnitzky, Jewell. Hypothesis testing of regression parameters in
semiparametric generalized linear models for cluster correlated data.
Biometrika 1990;77:485-97.
23. Pan W. GOF tests for GEE with correlated binary data. Scan J Stat
2002;29:101-10.
24. Barnhart, Williamson. GOF tests for GEE modelling with binary responses.
Biometrics 1998;54:720-9.
25. Horton. GOF for GEE: an example with mental health service utilisation.
Stat Med 1999;18:213-22.
26. Hosmer DJ, Lemeshow S. Applied logistic regression. New York, NY:
John Wiley & Sons, 1989.
27. Dosh S, Hickner J, Mainous AI, Ebell M. Predictors of antibiotic
prescribing for nonspecific upper respiratory tract infections, acute bronchitis,
and acute sinusitis. J Fam Pract 2000;49:407-14.
28. Verheij T. Diagnosis and prognosis of lower respiratory tract infections: a
cough is not enough. Br J Gen Pract 2001;51:174-5.
29. Hueston W, Mainous Ar, Dacus E, Hopper J. Does acute bronchitis really
exist? J Fam Pract 2000;49:401-6.
30. Holmes W, Macfarlane J, Macfarlane R, Hubbard R. Symptoms, signs, and
prescribing for acute lower respiratory tract illness. Br J Gen Pract
2001;51:177-181.
31. Hopstaken R, Nelemans P, Stobberingh E, Muris J, Rinkens P, Dinant G. Is
roxithromycin better than amoxicillin in the treatment of acute lower
respiratory tract infections in primary care? A double-blind randomized
controlled trial. J Fam Pract 2002;51:329-36.
32. Macfarlane J, Holmes W, Gard P, Macfarlane R, Rose D, Weston V, et al.
Prospective study of the incidence, aetiology and outcome of adult lower
respiratory tract illness in the community. Thorax 2001;56:109-114.

70

GPs perception of patients requests determines


prescription behaviour
33. Little P, Williamson I, Warner G, Gould C, Gantley M, Kinmonth AL.
Open randomised trial of prescribing strategies in managing sore throat [see
comments]. BMJ 1997;314:722-7.
34. Macfarlane J, Holmes W, Macfarlane R, Britten N. Influence of patients'
expectations on antibiotic management of acute lower respiratory tract illness
in general practice: questionnaire study. BMJ 1997;315:1211-4.
35. Cockburn J, Pit S. Prescribing behaviour in clinical practice: Patients'
expectations and doctors'perceptions of patients'expectations - a questionnaire
study. BMJ 1997;315:520-523.
36. Britten N, Ukoumunne O. The influence of patients'hopes of receiving a
prescription on doctors' perceptions and the decision to prescribe: a
questionnaire survey [see comments]. BMJ 1997;315:1506-10.
37. Feinstein A. The '
Chagrin Factor'and Qualitative Decision Analysis. Arch
Intern Med 1985;145:1257-9.
38. Coenen S, Van Royen P, Michels J, et al. Aanbeveling voor goede
medische praktijkvoering: Acute hoest [Good Clinical Practice Guideline:
Acute Cough]. Huisarts Nu 2002;31:391-411.
39. Wensing M, van der Weijden T, Grol R. Implementation guidelines and
innovations in general practice: which interventions are effective? Br J Gen
Pract 1998;48:991-7.
40. Gross P, Pujat D. Implementing practice guidelines for appropriate
antimicrobial usage: a systematic review. Med Care 2001;39(8 Suppl 2):II5569.

71

72

V
Antibiotics for coughing in general practice:
A clinical practice guideline
In Belgium, clinical practice guidelines are developed and disseminated since
1996. Clinical practice guidelines are systematically developed statements to
assist practitioner and patient decisions about appropriate health care for
specific clinical circumstances .1 The main goal of guidelines is to improve the
practice and outcome of medical care by reducing inappropriate variations in
practice.2 Up to now the Scientific College for Flemish General Practitioners
(WVVH) has produced 16 guidelines, including the guideline for acute cough.
These guidelines are available on the WVVH website (www.wvvh.be). In this
chapter we will present the standardized methodology defined by the WVVH
according to which we developed the guideline for acute cough as well as the
key messages of the resulting guideline.
The development of clinical practice guidelines by the WVVH follows a stepby-step methodology.1 3 First, the topic of the guideline is chosen and
described. Developing guidelines is time and money consuming. The available
means therefore are reserved primarily for the most relevant topics, namely the
major causes of morbidity and mortality in the population. Since the guidelines
of the WVVH are primarily meant to assist general practitioners, topics
relevant for primary care are chosen. Furthermore, the aim of guideline
development is to improve quality of care, reduce inappropriate variations in

Guideline Acute Cough is published in Huisarts Nu 2002;31:391-411.

73

Chapter V
practice and improve cost effectiveness. The description of the topic entails a
clear description of the targeted problem, and of the objectives of the guideline.
Second, a literature search is performed to collect the available evidence on the
problem. The search for evidence must be systematic and the results critically
appraised. The first step is looking for existing guidelines. Afterwards metaanalyses, systematic literature reviews (Cochrane reviews) and controlled
studies are looked for.
Third, an author group is composed. Guidelines are ideally developed by a
multidisciplinary team, which can also include patients, if necessary. A team
manager keeps an eye on the group'
s process while the different authors can
concentrate on the development of the guideline itself. This author group make
a draft guideline in which the guideline recommendations are formulated
clearly and precisely. For each of the key messages the corresponding level of
evidence is stated. Three levels of evidence are described for that purpose, with
level 1 for definite evidence, level 2 for likely evidence, and level 3 for an
indication for evidence.4 The level of evidence depends on the quality of the
individual studies and it is applicable for management decisions other than
treatment.
Fourth, the draft guideline is evaluated by external experts, i.e. not part of the
author group. Specialists and GP s can be asked to fulfil this important task.
These experts look at the guideline from a scientific point of view and they can
formulate remarks if these are evidence based and reference to the evidence is
provided. After this evaluation procedure, all authors come together and follow
a strict consensus procedure to decide about all comments and remarks.
Fifth, the feasibility of the guideline is evaluate in local groups of GPs. It is a
peer review process with patient cases, critical reading of the guideline and
feedback from the participating GPs own practical experience. A local
moderator as well as some of the authors should be present. Also after this
evaluation procedure, the comments and remarks are taken into account into a
final version, after a consensus procedure.
Sixth, the current version of the guideline is validated. The validation
committee, which consists of representatives of the four Flemish universities,
the SCGFP, and local groups of GPs, gives the final evaluation and validation
of the guideline. The task is limited to the appraisal of the guideline with

74

A clinical practice guideline


respect to the described methodology. The international AGREE-instrument is
a good basis for this appraisal.5
The final version of the guideline is published in the journal of the WVVH,
Huisarts Nu, and, together with annual follow-up reports, available on the
WVVH website.
To improve the quality of care, reduce inappropriate variation in practice and
improve cost-effectiveness, we decided to make a guideline for the
management of acute cough in general practice. Antibiotics are being over
prescribed in general practice, especially for respiratory tract infections. Acute
cough is one of the most frequent reasons for consulting in general practice (for
respiratory tract infections). Overuse of antibiotics for respiratory tract
infections wastes resources (both for the unnecessary drugs themselves and the
subsequent visits6) and increases resistance.7 8 Whether to prescribe an
antibiotic for acute cough or other respiratory complaints is a common
dilemma in primary care. The precise diagnosis is often unclear.9-11
Hence for the guideline for acute cough we aimed to formulate
recommendations for the diagnosis of the causes of acute cough and for the
treatment in case of respiratory infections with acute (productive) cough,
pneumonia excluded. Levels of evidence are provided for each of the
recommendations.

75

Chapter V

Diagnosis
The guideline used for the intervention recommends a clinical and stepwise
approach to diagnose the cause of acute cough in patients aged greater than 12
years who complain about acute cough with or without purulent sputum that
have not been treated in the preceding week with antibiotics, not patients
known to have chronic obstructive pulmonary disease or a chronic cough (=
more than 30 days).
First conditions such as pneumonia, pulmonary embolism, left ventricular
failure (pulmonary oedema), pneumothorax, aspiration and irritation by toxic
agents should be ruled out by means of history and clinical examination.
Although these are not frequent conditions, and although acute cough may not
be the most prominent complaint, these conditions are treatable, and possibly
life-threatening. They should not be missed. (level of evidence 3)
In case of clinical suspicion of pneumonia patient at low risk for mortality or
complications can be identified by means of history and clinical examination.
This risk determines the place of treatment (level 2). Treating these patients at
home with antibiotics is justified, ideally this decision is documented with a
positive chest X-ray. (level 3)
If another cause than a respiratory infection is present (for example asthma,
gastro-esophagial reflux disease, ACE-inhibitors) management needs to be
adjusted accordingly. Even though such conditions may not be obvious in a
first encounter, it is worthwhile to take them into account. (level 3)
If finally a respiratory infection seems to be the most likely cause, it is not
feasible to distinguish between viral and bacterial infections. (level 2)
Furthermore this differentiation is not meaningful for the therapeutic decision.
(level 3)

76

A clinical practice guideline

Treatment
In case of respiratory infections with acute (productive) cough, pneumonia
excluded, antibiotics have no effect on the (duration of the) productive cough
and limitation of work or other activities. Of each ten patients after seven to
eleven days more than eight are clinically improved regardless the use of an
antibiotic. Less than one patient extra improves due to the antibiotic, but as
many patients experience the side effects. (level 1)The possible benefits of
antibiotics are outweighed by their harm. Antibiotics are justified only in case
of compromised immunity. (level 3)
We recommend to explicitate patient expectations, to reassure patients and
inform them about the cause and duration of the complaints, to explain why
antibiotics are not necessary, and to instruct patients when they should
reconsult. (level 3)
The effectiveness of over-the-counter medicines is unclear. For the
symptomatic treatment an antitussivum (dextromethorphan) or an expectorans
(guaifenesine) can be prescribed. (level 3)

77

Chapter V
References
1. Buntinx F. Een standaard der standaarden [A guideline for guidelines].
Huisarts Nu 1996;25:13-6.
2. Van Royen P. Aanbevelingen in de praktijk. Bedreiging voor vrijheid van
diagnose en therapie [Guidelines in practice. A threat to freedom of diagnosis
and treatment]? Huisarts Nu 1998;27:207-10.
3. Van Royen P, Coenen S, Denekens J, Dieleman P, Michels J. From practice
guidelines to implementation of good clinical practice. A Flemish guideline for
acute cough and rational antibiotic use in general practice [abstract]. Eur J Gen
Pract 2001;8:104.
4. Van Royen P. Niveaus van bewijskracht: levels of evidence. Huisarts Nu
2002;31:54-7.
5. The AGREE Collaboration. AGREE Instrument. Available at:
http://www.agreecollaboration.org. Accessibility verified December 1,2001.
6. Little P, Gould C, Williamson I, Warner G, Gantley M, Kinmonth AL.
Reattendance and complications in a randomised trial of prescribing strategies
for sore throat: the medicalising effect of prescribing antibiotics [see
comments]. BMJ 1997;315:350-2.
7. Butler C, Rollnick S, Kinnersley P, Jones A, Stott N. Reducing antibiotics
for respiratory tract symptoms in primary care: consolidating '
why' and
considering '
how'
. Br J Gen Pract 1998;48:1865-70.
8. Coenen S, van Royen P, Denekens J. Reducing antibiotics for respiratory
tract symptoms in primary care: '
why'only sore throat, '
how'about coughing?
[letter]. Br J Gen Pract 1999;49:400-1.
9. Coenen S, Van Royen P, Vermeire E, Hermann I, Denekens J. Antibiotics
for coughing in general practice: a qualitative decision analysis. Fam Pract
2000;17:380-5.
10. Coenen S, Van Royen P, Denekens J. Diagnosis of acute bronchitis [letter;
see reply]. J Fam Pract 1999;48:471-2.
11. Hueston W, Mainous Ar, Dacus E, Hopper J. Does acute bronchitis really
exist? J Fam Pract 2000;49:401-6.

78

VI
Antibiotics for coughing in general practice:
Cluster randomised controlled trial of a tailored
professional intervention to optimise prescribing
Introduction
In primary care antibiotics are being overprescribed, especially for respiratory
infections.1 2 This is also true for the Netherlands3 with the lowest antibiotic
consumption in the European Union: 9 defined daily doses per 1000
inhabitants per day.4 The problem is particularly important for countries such
as the UK with a consumption being twice, or Belgium with a consumption of
nearly three times that of the Netherlands.4 After all antibacterial resistance is
linked with the antibiotic consumption,5 and it is time for action.6
Decreasing the use of antibiotics has been among the most targeted issues of
different strategies to improve the use of medicines. Regulatory/financial
measures,7 organisational interventions,8-10 and professional interventions can
be distinguished. Professional interventions use primarily evidence-based
arguments on effectiveness, safety, cost and sometimes applicability for
changing professional practice. Implementing evidence-based guidelines is one
of the most known, and best studied examples of this approach. Most studies
however were done in the US, and many targeted hospital prescribing rather
than primary care prescribing.11 Nevertheless, in some studies a considerable

79

Chapter VI
decrease of antibiotic use was seen using a multifaceted intervention to
implement guidelines for appropriate antimicrobial usage.12
Specific barriers to change occur at the level of the social context and at the
broader context of the health care structure and culture. In Belgium most GPs
are paid fee for service. There is a plethora of mostly solo practicing GPs,
competition for patients and open access to secondary care. In contrast to the
Netherlands pharmacotherapy is not discussed on a regular basis in local
groups involving pharmacists. Furthermore, there are cultural differences and
different attitudes towards respiratory symptoms and antibiotics between
Belgium and the Netherlands. All of which may partly explain the large
variation in antibiotic consumption between both countries.13 14
Other barriers to change relate to the credibility of the guideline and to the
individual prescriber. Though most GPs are aware of the problem of antibiotic
resistance at population level, there may be lack of awareness of the effect of
overprescribing in individuals on the probability of future infections with
resistant pathogens. Internal barriers within the prescriber relate to knowledge,
attitude, but also to the decision process when prescribing an antibiotic. We
looked at the antibiotic prescribing decision of Flemish GPs in patients with
complaints about coughing, not patients with acute bronchitis.15 Using
qualitative and quantitative research methods we found non-medical reasons
played an important role in the prescribing decision, especially in case of
diagnostic uncertainty.16 17 Moreover, they favoured antibiotic prescribing.
Despite lacking evidence for their effectiveness18 19 GPs anticipated more
'
chagrin'over not prescribing antibiotics than over prescribing them anyway,
since antibiotics might prevent them from loosing patients as a result of
unfulfilled patient expectations or undetected serious disease.
To optimise antibiotic prescribing in our country, we developed a context
specific evidence based guideline for acute cough. The main recommendation
is that most patients with acute cough do not need antibiotics. Although no
single combination of approaches is clearly better than the other to implement
guidelines, we preferred the individual approach of academic detailing, and
tailored the intervention to identified barriers within GPs.

80

Cluster randomised controlled trial of a tailored


professional intervention to optimise prescribing
A cluster randomised controlled trial was conducted to assess the effectiveness
of a tailored professional intervention to support the implementation of a
clinical practice guideline for the management of acute cough on antibiotic
prescribing in adult patients with acute cough in general practice.

Methods

Design
We tested the main hypothesis with a cluster randomised pre-test-post-test
controlled trial (cRCT). General practitioners (GPs) were randomised before
the pre-test in an intervention group and a control group (Figure 1). Our
intervention was preceded by a national public campaign.20 21 Only the GPs in
the intervention group received a tailored intervention to support implementing
a clinical practice guideline for the management of acute cough in adult
patients. Pre-test data were collected during a three month period in 2000, posttest data one year later, in 2001, after both interventions.

Participants
In total 149 GPs not reluctant to take part in further study on the topic at the
time of the postal questionnaire study17 were sent a letter and questionnaire
inviting them to join the study. Overall, 85 GPs agreed to participate and were
randomised. After sorting the questionnaire data by gender, university of
promotion, and age all even records were allocated to one group, all uneven to
another. After making sure GPs from the same practice ended up in the same
group, tossing a coin decided which group would serve as intervention group.
We included consultations for acute cough if they concerned immunocompetent patients, 18-65 years, with new or worsening coughing, present for
less than 30 days as (one of) the most important complaint(s) and as the reason
for first encounter at the GPs practice.
The involvement of GPs and patients in the trial is summarised in Figure 1.

81

Chapter VI

Interventions
Apart from our intervention all participating GPs received booklets and leaflets
of a public campaign initiated in Belgium in November 2000 and continued
until December 2000 (Figure 1).20 21 This campaign also included TV spots and
radio messages informing the public on overconsumption and misuse of
antibiotics, the resulting resistance problem and the self-limiting character of
most frequent infections in the community. All GPs were invited to participate
in the cRCT before the pre-test, reminded of the trial before the post-test by
mail, and received a fee of
24.79 or
61.97 after each study period
depending on their response. After an appointment was made by telephone they
received the material and instructions for data-collection by means of a practice
visit and a reminder phone call at the start of each registration period.
Before the post-test period GPs in the intervention group received our tailored
professional intervention (Figure 1), consisting of a clinical practice guideline
for the management of acute cough in general practice, an educational outreach
visit to GPs based on the principles of academic detailing,22 and a postal
reminder of the key messages (Box 1).
An author group of GPs developed a clinical practice guideline according to a
standardised methodology defined by the Scientific College of Flemish
General Practitioners and in line with the AGREE criteria.23 Fine tuning for the
specific context of Flemish GPs was based on previous descriptive studies on
the management of acute cough and the determinants of antibiotic
prescribing.16 17 The guideline for the intervention was reviewed by a
multidisciplinary panel of experts. An educational package was developed in
accordance with this guideline and key messages were formulated (Box 1).
Before the post-test all GPs in the intervention group received the guideline by
mail and a telephone call. Each time GPs were asked to read the guideline in
anticipation of an outreach visit at their practice delivered by one of two
trained facilitators. The facilitators, not GPs, combined the educational visit
with the material and instructions delivery. They rephrased the information in
the guideline using simple overheads and emphasising the key messages. The
educational element of this method was a dialogue on perceived barriers to
adhering to the guideline.

82

Cluster randomised controlled trial of a tailored


professional intervention to optimise prescribing

Figure 1 GP and patient flow, and design of the study

149 GPs invited to join the study


64 declined
85 GPs agreed to participate

randomised

42 GPs in intervention group

43 GPs in control group

Pre-Test Period: February-April 2000


Consultation data (patients)
485 eligible for recruitment (100%) a
445 included (92%) a
365 eligible for analysis c (75%) a

Consultation data (GPs)


36 GPs respond
no response from 6 GPs
c size 10(4,15) a
median (ICQ) cluster

Consulation data (GPs)


35 GPs respond
no response from 8 GPs b
median (ICQ) cluster size 13(9,18) a

Patient diaries
243/365 patients responded (67%) a

Consultation data (patients)


574 eligible for recruitment (100%) a
531 included (93%) a
445 eligible for analysis c (78%) a

Patient diaries
278/445 patients responded (62%) a

Belgian public campaign: Decembre 2000 (http://www.red-antibiotica.org/english/index.html)


Intervention: January 2001
Post-Test Period: February-April 2001
Consultation data (patients)
398 eligible for recruitment (100%) a
356 included (89%) a
292 eligible for analysis c (73%) a

Patient diaries
208/292 patients responded (71%) a

Consultation data (GPs)


27 GPs respond d
no response from 15 GPs e
median (ICQ) cluster size 10(5,16) a

Consulation data (GPs)


32 GPs respond d
no response from 11 GPs e
median (ICQ) cluster size 14(7,16) a

Consultation data (patients)


521 eligible for recruitment (100%) a
468 included (90%) a
401 eligible for analysis c (77%) a

Patient diaries
280/401 patients responded (70%) a

a: The proportions of patients eligible for recruitment (100%) actually included in the study, the proportions of those eligible
for analysis c, and the proportion of the latter patients responding with patient diaries, as well as the median cluster sizes were not
significantly different between the study groups within each study period, nor between the study periods within each study group using
Generalised Estimating Equations and Kruskal Wallis Median Test respectively.
b: 1 GP responding in the Pre-Test did not have consultation data eligible for analysis c.
c: Analysis of the main outcome measures, differences in antibiotic prescribing rates
d: Only 27 GPs in the intervention and 29 GPs in the control group responded in both the Pre-Test and the Post-Test
e: In the intervention group 9 GPs responding in the Pre-Test did not respond in the Post-Test, compared to 6 GPs in the control
group. In the control group 3 GPs not responding in the Pre-Test did so in the Post-Test.

83

Chapter VI
The focus of this dialogue however was on dealing with barriers within the
individual prescriber, especially in dealing with diagnostic uncertainty. Using
a fishbone scheme we presented what was known about the accuracy of history
and clinical examination to differentiate between viral and bacterial respiratory
infections, upper and lower respiratory infections, and between bronchitis and
pneumonia,24 25 about the validity of a clinical prediction rule to assess
prognosis in case of community-acquired pneumonia,26 about the effectiveness
of antibiotics for acute cough,18 and about the effect of antibiotic consumption
on bacterial resistance in the community and the individual,27 28 to conclude
that after ruling out pneumonia patients with acute cough due to a respiratory
infection do not need antibiotics. After all possible benefits of antibiotics are
outweighed by their cost, and it is not possible to identify those patients who
will benefit from antibiotics. Nevertheless, the guideline also recommended
amoxicillin or doxycyclin as first choice antibiotics if for any reason the GP
decides to prescribe antibiotics. We addressed the effect of patient and
physician related non-medical reasons on the prescribing decision, especially in
case of diagnostic uncertainty, as well. We demonstrated the mismatch
between patients'expectations and GPs perceptions of these, stressing the latter
are described as important determinants favouring antibiotic prescribing,29-31
and we instructed the GPs on how to make patients'expectations regarding
antibiotic prescribing explicit and provided different strategies for different
patient expectations. To overcome an uncomfortable prescribing decision due
to GP related reasons, we stated that watchful waiting will prevent
complications more effectively than antibiotics, and will not jeopardize the
doctor-patient relationship. We thus tried to show that managing patients
according to the guideline might result in a win-win situation, more satisfied
GPs, more satisfied patients, and less antibiotic consumption. We thus tailored
the interventions to overcome identified barriers.
Before the post-test all intervention GPs also received one page with the key
messages of the guideline by mail as reminder (Box 1).
Our intervention was initiated in December 2000 and continued until January
2001. The study protocol was approved by the local research ethics committee.
Consent was obtained from GPs and patients.

84

Cluster randomised controlled trial of a tailored


professional intervention to optimise prescribing

Box 1 Key messages of the guideline for acute cough


This guideline concerns patients, aged 12 years or older, whose most prominent
complaint is acute cough with or without purulent sputum, not patients with
recurrent or chronic cough, chronic obstructive pulmonary disease or patients that
have been treated in the preceding week with antibiotics.
First, pneumonia, pulmonary embolism, left ventricular failure (pulmonary oedema),
pneumothorax, aspiration and irritation by toxic agents should be ruled out by
history and clinical examination. Although these are not frequent conditions, and
although acute cough may not be the most prominent complaint, these conditions
are treatable, and possibly life-threatening. They should not be missed.
If another cause than a respiratory infection is present (for example asthma, gastrooesophageal reflux disease, ACE-inhibitors) management needs to be adjusted
accordingly. Even though such conditions may not be obvious in a first encounter, it
is worthwhile to take them into account.
If finally a respiratory infection seems to be the most likely cause, it is not feasible to
distinguish between viral and bacterial infections. Nevertheless the decision
whether to prescribe antibiotics has to be made. Antibiotics are only needed for
patients with compromised immunity.
Besides the scientific arguments, we also recommend to integrate the GPs own
agenda as well as that of the patient in the final therapeutic decision.

85

Chapter VI

Data
GPs were asked to collect medical (demographics, medical history & risk,
circumstances of the consultation, symptoms, signs, test ordering and
prescriptions) as well as non medical data (e.g. GPs workload at the time of the
consultation) in 20 consecutive patients eligible for recruitment. If, due to time
constraints, this was not possible, one in two or one in three patients were to be
included. The GPs kept records of those patients eligible for recruitment but
not included. They collected the data themselves on pre-printed forms, with
clear instruction about how this should be done. To ensure patient
confidentiality, GPs completed the forms using patient identification numbers
only. GPs were also asked to deliver a package, containing a symptom diary
and clear instructions for its use, to all included patients.
Patients were asked to record their symptoms and medication consumption
starting the day of the consultation for 29 days or for as long as appropriate.
Each diary also contained an identification number and was to be returned to
the GP in a sealed envelope.
The GPs held a patient reference sheet with names of patients against those
numbers. This enabled them to assess and improve their patients'response.
They sent all completed data collection material to SC for analysis. The data
collection method had been previously piloted.
We compared data from control with data from the intervention group in each
study period, and data from the pre-test period, February-April 2000, with data
from the post-test period, February-April 2001 in each study group.

Outcomes
The primary outcome was the antibiotic prescribing rate by GPs for adult
patients with acute cough. We were also interested in the kind of antibiotics
prescribed, if any, and whether any change in antibiotic prescribing affected
symptom resolution. Finally, we measured the medication cost per patient from
the perspective of the National Sickness and Invalidity Insurance Institute
(NSIII). We expected that implementation of the guideline would reduce the
proportion of patients who were prescribed antibiotics for acute cough,
increase the relative proportion of first choice antibiotics, and that this would
not affect patients'symptom resolution.

86

Cluster randomised controlled trial of a tailored


professional intervention to optimise prescribing

Sample size
Sample size was calculated with antibiotic prescribing for acute cough as
primary outcome. Before the pre-test no data were available regarding
antibiotic prescribing rate for acute cough nor about the intracluster correlation
coefficient (ICC) needed to adjust the sample size because GPs rather than
patients were randomised. Therefore we calculated the sample size with a
method that takes into account the number of events, the expected effect, and
the power of the study, but not the ICC. We thus acted as if patients were
randomised, and assumed a minimum of 20 patients for each practice and a
worst case control group rate of 50%. Under these assumptions we anticipated
a power of 90% to detect a difference of 10% in rates between the two groups
at the 5% significance level with 30 practices in each study group. We
anticipated adjustment of the sample size for cluster randomisation and loss to
follow up. Therefore we planned to randomise 40 GPs in each group and to
adjust the number of patients to be included per practice for the post-test
according to the ICC found in the pre-test.

Statistical methods
We applied cluster specific methods taking into account the dependence among
patients of the same GP, known as the clustering effect: GPs rather than
patients were randomised, and variance in how patients were managed would
be partly explained by the GP.32
We used logistic regression to test for an effect of our intervention on antibiotic
prescribing (Box 2). To test for differences in medication cost a linear
regression model with random intercept was used to account for within-GP
correlation. To test for differences in time to symptom resolution we used
Cox'
s proportional hazard regression. Data were analysed assuming
independence and standard errors were then corrected for within-GP
correlation using a robust estimator. All models were estimated with SAS
v8.02.33 All other analyses were done with Statistica v6.0.34

87

Chapter VI

Box 2 Analysis of cluster data


To assess the effect of a tailored intervention on GPs'antibiotic prescribing for acute
cough, we first estimated a logistic model: logit p(X) = 0 + 1 G + 2 P + 3 G*P,
where p(X) is the probability of an antibiotic prescription, G is a dichotomous variable
for the study groups, which equals zero for the control and one for the intervention
group, P is a dichotomous variable for the study period, which equals zero for the pretest and one for the post-test.
The same kind of model was used to test for significant differences of the covariates.
All significant covariates in this analysis were included in the above model as possible
confounders. Then from this multivariable analysis non significant covariates were
removed, eliminating one by one the covariates with least significant type 3 score
statistics.
By testing the hypothesis H0: 1=0 we were able to test for differences between
control and intervention group during pre-test.
Second, we considered the model without the effect of the study group, forcing
antibiotic prescribing rates to be equal in both study groups during the pre-test, an
assumption
which
should
hold
with
randomisation.
Under this assumption we were tested for differences between control and
intervention group during the post-test by testing the hypothesis H0: 3=0.
By testing the hypothesis H0: 2+ 3=0 we were able to test for differences between
pre-test and post-test in the intervention group.
By testing the hypothesis H0: 2=0 we were able to test for differences between pretest and post-test in the control group.
We adjusted logistic regression estimates for clustering within our data (patients are
32
35
nested within GPs). We used generalized estimating equations (GEE). Using GEE
we can also provide an order of magnitude for the intra-cluster correlation coefficient
(ICC) . For significance testing of the ICC estimate we used alternating logistic
regression (ALR), a technique closely related to GEE. With ALR, estimating equations
are specified for marginal and association parameters. The association between pairs
of responses is measured by log odds ratios ( ), instead of correlations ( ) as with
36
GEE. The ICC in GEE is related to in ALR.

88

Cluster randomised controlled trial of a tailored


professional intervention to optimise prescribing
Results

GP flow & characteristics


The randomised GPs were similar to the other Flemish GPs (invited) with
respect to age and sex distribution.
They all received the material and instructions for data collection. Six GPs in
the intervention arm and seven GPs in the control arm did not respond in the
pre-test period (Figure 1). We did not get data eligible for the analysis of the
main outcome measures from one control GP. This left 36 of 42 GPs in the
intervention arm and 35 of 43 GPs in the control arm for the pre-test. These 36
GPs in the intervention arm received the entire intervention. Nine GPs in the
intervention arm and six GPs in the control arm did not respond in the posttest. Thus 27 GPs in the intervention arm and 29 GPs in the control arm
responded in both the pre-test and the post-test. Three GPs were recovered for
the post-test in the control arm. This left 27 of 42 GPs in the intervention arm
and 32 of 43 GPs in the control arm for the post-test.
No significant differences were found between the intervention and control
GPs (Table 1), nor between responding GPs and the non-responding group for
the same characteristics.

Patient flow & characteristics


Consultation data. The GPs collected data for 1978 patients eligible for
recruitment: in the pre-test 485 in the intervention group, 574 in the control
group, in the post-test 398 and 521 respectively. (Figure 1). They included
1800 patients in the study (445, 531, 356 and 468 respectively), of which 1503
patients were eligible for analysis of the primary outcome (365, 445, 292 and
401 respectively). Comparing between the four groups - we mean comparing
between both study groups within each study period and between both study
periods within each study group - the median cluster sizes were similar (Figure
1). Likewise similar proportions of patients eligible for recruitment were
actually included in the study, respectively eligible for analysis. The
proportions of male patients were not different whether patients eligible for
recruitment were included in the study or not. Only in the control group in the
post-test the proportion of male patients eligible for recruitment was greater in
those eligible for analysis than in those not eligible for analysis. Of the patients

89

Chapter VI
eligible for recruitment the patients included and those eligible for analysis
were younger than those not included, respectively not eligible for analysis.
Table 2 shows the characteristics of the patients with acute cough eligible for
analysis by study group for the pre-test and the post-test period. Except for risk
for thrombo-embolic disease, duration of cough, presence of sputum, of ronchi,
of loss of appetite, and of a referral, characteristics of the patients eligible for
analysis based on the data collected by the GPs were similar (Table 2).
Patients in the intervention group were less likely to be at risk for thromboembolic disease in the pre-test (odds ratio (OR) (95% CI)=0.17 (0.05-0.60) and
the post-test (OR=0.15 (0.03-0.79)). They were less likely to produce sputum
(OR=0.68 (0.47-0.98)), less likely to be referred (OR=0.23 (0.06-0.68)) and
coughing significantly less days before consulting in the pre-test only
(estimated difference (ED) (95%CI)= 0.96 (0.12-1.80). In the post-test patients
in the control group were coughing significantly less days before consulting
compared to the pre-test (ED=0.79 (0.12-1.46)) They were also less likely to
produce sputum (OR=0.68 (0.48-0.97)), have loss of appetite (0.60 (0.390.93)) or ronchi (0.58 (0.34-1.00)).
Patient diaries. Patient diaries of 1009 patients eligible for analysis were
available: in the pre-test 243 in the intervention group, 278 in the control
group, in the post test 208 and 280 respectively (Figure 1). Comparing between
the four groups the proportion of patients eligible for analysis responding with
patient diaries is similar. Except for age, duration of coughing, smoking, ACEInhibitors and percussion dullness, the characteristics of the patients eligible
for analysis were similar whether patients responded with patient diaries or not.
Patients responding with patient diaries were significantly older (estimated
difference (95%CI)= 4.00 years (2.57-5.42), and coughing not as long (0.57
days (0.04-1.11) They were more likely to be taking ACE-Inhibitors (OR=3.62
(95% CI=1.17-11.2) and less likely to be smoking (0.57 (0.44-0.73) or have a
clinical examination positive for percussion dullness (0.33 (0.15-0.93).
Comparing the presence of complaints on the day of the consultation (= day 1)
from these patients'diaries between the four group, only for complaints about
fever and headache differences were found between the study periods (Table
3). Patients in the control group were less likely to suffer from fever (0.62
(0.43-0.89)) and patients in the intervention group were more likely to suffer
from headache (1.64 (1.08-2.48) in the pre-test.

90

Cluster randomised controlled trial of a tailored


professional intervention to optimise prescribing

Table 1 Characteristics of study GPs in intervention and control group(1)


Characteristics

Intervention GPs
N=27

Figures are numbers (percentage) (2)


MEN
19 (70)
UA GRADUATES
12 (44)
PROFESSIONAL TRAINING
10 (40)
FEE FOR SERVICE
26 (96)
SINGLE-HANDED
13 (48)
(3)
GPPTs IN PRACTICE
9 (33)
PART TIME
5 (19)
PRACTICE IN ANTWERP
15 (56)
PEAKFLOW METER
25 (93)
SPIROMETER
7 (26)
TRAINING PRACTICE
10 (37)
ACADEMIC LINK
10 (37)
RECORDS OF HOME VISITS
21 (78)
COMPUTERISED RECORDS
17 (63)
COMPLEMENTARY MEDICINE
1
(4)
Figures are means (standard deviation)
AGE AT START OF THIS STUDY
43.6 (8.3)
PATIENT ENCOUNTERS PER
100 (43)
WEEK
HOME VISITS PER WEEK
34 (21)
MEDICAL REPRESENTATIVES
16 (11)
PER MONTH
(4)
ATC J COST RATIO
16.6 (9.0)
(4)
ATC J VOLUME RATIO
3.5 (1.6)

Control GPs
N=29
20
12
6
28
15
3
5
21
26
9
8
9
25
21
1

(69)
(41)
(22)
(97)
(52)
(10)
(17)
(72)
(90)
(31)
(28)
(31)
(86)
(72)
(3)

45.0 (8.1)
108 (43)
33
15

(18)
(11)

14 (6.1)
2.9 (1.2)

(1) There were no significant differences between the intervention and the control group using chisquare or Student'
s t test where appropriate when comparing the characteristics of the GPs
responding in the Pre-Test (n=72), of those responding in the Post-Test (n=59) or of those
responding in the Pre-Test and the Post-Test (n=56). The latter comparison is presented in this
table.
(2) Denominators vary due to missing values
(3) General Practitioner in Professional Training
(4) The ratios the gross amount for antimicrobials for systemic use (ATC J)/ the gross amount for all
pharmaceutical specialities and the volume (Daily Defined Dosage) DDD)) of ATC J/ the volume for
all pharmaceutical specialities are both expressed as percentages on individual prescribing
feedback from the National Sickness and Invalidity Insurance Institution (NSIII) to GPs. We asked
the participating GPs for these percentages and calculated their mean.

91

Chapter VI
Table 2 Characteristics of patients with acute cough: consultation data.
Figures are numbers (percentage) unless otherwise stated.
Characteristics
Demographics
MEN
History & Risk
ASTHMA
COPD (CARA)
HEARTFAILURE
ACE-INHIBITOR
CV-DISEASE
ASPIRATION RISK
LIVER DISEASE
RENAL DISEASE
NEOPLASTIC DISEASE
TROMBO-EMBOLIC RISK
SMOKING
Circumstances
HIGH WORKLOAD
SICK IMPRESSION
REQUEST AB
REQUEST MED
Symptoms
SPUTUM
FEVER
RUNNY NOSE
HEADACHE
MUSCLE EACHE
SOAR THROAT
WHEEZING
SHORT OF BREATH
CHESTPAIN
LOSS OF APPETITE
LIMITED ACTIVITY
Signs
ALTERED CONSCIOUSNESS
PULSE RATE > 125/'
RESPIRATORY RATE > 30/'
TEMPERATURE > 38C
SYSTOLIC BP<90mmHg
LESS VESICULAR
BREATHING
WHEEZING
RONCHI
CREPITATIONS
PERCUSSION DULNESS

92

PRE-TEST (2000)

POST-TEST (2001)

Intervention

Control

156 (43)

191 (43)

118

(40)

172

(43)

26 (7)
27 (7)
3 (1)
9 (2)
8 (2)
0 (0)
3 (1)
2 (1)
3 (1)
5 (1)
119 (33)

61 (14)
39 (9)
7 (2)
14 (3)
10 (2)
10 (2)
13 (3)
7 (2)
9 (2)
39 (9)
158 (36)

21
23
0
5
1
0
1
0
2
3
80

(7)
(8)
(0)
(2)
(0)
(0)
(0)
(0)
(1)
(1)
(27)

32
30
6
8
6
4
8
2
4
25
135

NS
(8)
NS
(7)
(1)
a DK
NS
(2)
NS
(1)
(1)
a DK
NS
(2)
(0)
a DK
NS
(1)
(6) b,c S
NS
(34)

196 (54)
158 (43)
37 (10)
175

(48)

242 (54)
197 (44)
88 (20)
216

(49)

153
112
27

(52)
(38)
(9)

243
171
66

(61)
(43)
(16)

198
94
221
177
119
227
61
98
108
99
164

(54)
(26)
(61)
(48)
(33)
(62)
(17)
(27)
(30)
(27)
(45)

285
149
287
232
168
242
89
134
159
130
221

(64)
(33)
(64)
(52)
(38)
(54)
(20)
(30)
(36)
(29)
(50)

159
64
179
146
86
173
52
79
92
62
105

(54)
(22)
(61)
(50)
(29)
(59)
(18)
(27)
(32)
(21)
(36)

222
114
235
198
122
214
81
110
138
82
167

(55) b,d S
NS
(28)
NS
(59)
NS
(49)
NS
(30)
NS
(53)
NS
(20)
NS
(27)
NS
(34)
(20)
dS
NS
(42)

3 (1)
4 (1)
3 (1)
39 (11)
16 (4)
49 (13)

8 (2)
8 (2)
11 (2)
63 (14)
7 (2)
70 (16)

0
0
0
22
6
32

(0)
(0)
(0)
(8)
(2)
(11)

2
2
5
39
13
36

(0)
(0)
(1)
(10)
(3)
(9)

68 (19)
95 (26)
18 (5)
1 (0)

78 (18)
118 (27)
32 (7)
9 (2)

49
71
13
2

(17)
(24)
(4)
(1)

63
70
26
5

(16)
(17)
(6)
(1)

n=365

n=445

Intervention

n=292

163

(56)

Control

n=401

180

(45)

GEE
NS

NS
NS
NS
NS

DK
DK
DK
NS
NS
NS
NS

d NS*
NS
NS

Cluster randomised controlled trial of a tailored


professional intervention to optimise prescribing
Table 2 Continued.
Characteristics
Investigations
RADIOGRAPH
SPUTUMANALYSIS
INFECTIONPARAMETERS
SEROLOGY
OTHER
Prescriptions
REFERRAL
SICK LEAVE
FOLLOW UP CONTACT
MEDICATION
PREDICTED MEAN AGE
(95% CI)
PREDICTED MEAN
DURATION
OF COUGH (95% CI)
FINE a
LUNG AUSCULTATION

PRE-TEST (2000)

Intervention

n=365
7
0
4
1
0

(2)
(0)
(1)
(0)
(0)

3 (1)
124 (34)
25 (7)
318 (87)

Control

n=445

23
12
12
7
26

5,6 (4,9-6,2)

n=292

Control

GEE

n=401

(5)
(3)
(3)
(2)
(6)

3
1
5
2
6

(1)
(0)
(2)
(1)
(2)

11
4
10
9
9

(3)
(1)
(2)
(2)
(2)

NS
a DK
NS
NS
a DK

18 (4)
193 (43)
57 (13)
388 (87)

1
107
29
285

(0)
(37)
(10)
(98)

12
165
33
377

(3)
(41)
(8)
(94)

bS
NS
NS
NS

41,9 (40,3-43,5) 40,9 (39,4-42,4)

142 (39)
139 (38)

POST-TEST (2001)

Intervention

6,5 (5,9-7,1)

175 (39)
165 (37)

40,2 (38,5-42,0)

5,2 (4,6-5,9)

98 (34)
102 (35)

41,7 (40,1-43,2)

5,7 (5,1-6,3)

147 (37)
119 (30)

NS

b,d S
NS

a No estimation possible. To assess differences for these characteristics we


26
constructed a new variable derived from Fine et al. , which equals one if the
patients age > 50 or if the patient has congestive heartfailure, cerebrovascular
disease, liver disease, kidney disease or neoplastic disease, or has altered
consciousness, pulse rate>125/'
, respiratory rate>30/'
, temperature>38C or
systolic blood pressure>90 mmHg, and which equals zero otherwise.
b,c,d,e Significant differences between intervention and control in pre-test (b), or in
post-test (c), between pre-test and post-test in control (d), or intervention group (e).
Lung auscultation is a constructed variable which equals one in the presence of less
vesicular breathing, wheezing, ronchi or crepitations, and which equals zero otherwise

93

Chapter VI
Comparing the number of complaints on the day of the consultation patients in
the control group suffer from less complaints in the pre-test compared to the
post-test (estimated difference (95%CI)= 0.50 (0.04-0.95), and compared to the
intervention group (0.52 (0.15-0.90)

Outcome
Outcome data were collected for 1503 consultations for acute cough. The ICC
for the primary outcomes was highly significant, indicating that cluster specific
analytical methods were appropriate.
Use of antibiotics. Table 4 shows the prescription rate of antibiotics and the
percentage difference in change of prescription rates for patients in the
intervention and the control group. In the pre-test period antibiotic prescribing
rates were not significantly different between the intervention and control
group (OR (95% CI) = 1.09 (0.68 - 1.76)); ORadj (95% CI) = 1.28 (0.76 - 2.16),
adjusted for duration of cough and presence of sputum). Using the model
forcing antibiotic prescribing rates to be equal in both study groups during the
pre-test, an assumption which should hold with randomisation, patients in the
intervention group were less likely to receive an antibiotic after our
intervention compared to controls (ORadj = 0.56 (0.36-0.87)). Also comparing
the antibiotic prescribing rate between the pre-test and the post-test, only
patients in the intervention group were less likely to receive antibiotics after the
intervention (ORadj = 0.56 (0.39-0.81), and not patients in the control group
(ORadj = 1.01 (0.76-1.33)).
Kind of antibiotics used. In the pre-test period prescribing rates of
recommended antibiotics were not significantly different between the
intervention and control group (OR = ORadj = 1.05 (0.52-2.12)). Under the
assumption of equal baseline rates, patients in the intervention group were
more likely to receive amoxicillin or doxycyclin than patients in the control
group (ORadj = 1.90 (0.96-3.75)) (Table 4). Also comparing the antibiotic
prescribing rate between the pre-test and the post-test, only patients in the
intervention group were more likely to receive the recommended antibiotics
after the intervention (ORadj = 1.98 (1.19-3.29)), and not patients in the control
group (ORadj = 1.03 (0.61-1.78)).

94

Cluster randomised controlled trial of a tailored


professional intervention to optimise prescribing
Table 3 Characteristics of patients with acute cough: data from patient diaries.
Figures are numbers (percentage).
PRE-TEST (2000)

Complaints on day 1
COUGHING
SPUTUM
FEVER
RUNNY NOSE
SOAR THROAT
HEADACHE
MUSCLE ACHE
LOSS OF APPETITE
SHORT OF BREATH
WHEEZING
CHESTPAIN

POST-TEST (2001)

Intervention

Control

Intervention

Control

n=243

n=278

n=208

n=280

239
149
62
150
139
117
80
89
84
49
87

(98)
(61)
(26)
(62)
(57)
(48)
(33)
(37)
(35)
(20)
(36)

275
177
98
175
160
151
114
110
113
72
108

(99)
(64)
(35)
(63)
(58)
(54)
(41)
(40)
(41)
(26)
(39)

205
112
56
128
111
126
65
74
75
45
65

(99)
(54)
(27)
(62)
(53)
(61)
(31)
(36)
(36)
(22)
(31)

273
170
71
160
148
159
106
97
94
62
95

(98)
(61)
(25)
(57)
(53)
(57)
(38)
(35)
(34)
(22)
(34)

a
b

a,b Significant differences between pre-test and post-test in control (a), or intervention group (b).

Table 4 Rate of use and percentage difference in change of use of (recommended)


antibiotics.

Use of antibiotics (AB):


Pre-Test
Post-Test
Percentage change
Percentage difference
ORadj (95%CI)*
Use of recommended AB:
Pre-Test
Post-Test
Percentage change
Percentage difference
ORadj (95%CI)*

Intervention

Control

43.0 (157/365)
27.4 ( 80/292)
-15.6

37.8 (168/445)
28.7 (115/401)
-9.1

-6.5
0.56 (0.36-0.87)

40.1 (63/157)
53.8 (43/ 80)
+13.6

37.5 (63/168)
37.4 (43/115)
-0.1

+13.7
1.90 (0.96-3.75)

* Odds ratios are based on the model assuming equal antibiotic prescribing rates in intervention and control group
in the Pre-Test period, and are adjusted for patient characteristics (see box 2)

95

Chapter VI
Cost of antibiotics. Looking at the medication cost from the perspective of the
NSIII means looking at the reimbursement cost of prescribed medication. Since
for many prescribed drugs for acute cough no reimbursements are made,
significance testing of the medication cost in all patients is hampered by
distributional problems. We tested for differences in reimbursement cost in the
subset of patients who were prescribed an antibiotic. Since antibiotics represent
the only reimbursed group of prescribed medication in this subset of patients,
we
actually
tested
for
differences
in
antibiotic
cost.
The antibiotic cost was lower in the intervention group after our intervention
compared to the control group (Mean Difference (MD)adj (95%CI) = -6.89 (11.77 - (-2.02) ), and compared to the pre-test (MDadj = -6.11 (-9.97 - (-2.24)
) (Table 5) .
Time to symptom resolution. Concerning the use of antibiotics and the kind of
antibiotics used the same conclusions can be drawn form the subset of patients
responding with patient diaries. Comparing the time to resolution of all
symptoms and to return to the activity and to the health status of before the
illness, no significant difference was found between the patients in the
intervention group and those in the control group after our intervention (Figure
2).
Other analyses. Large variations occurred across the included GPs in the
prescription of antibiotics and in the extent of change for this outcome measure
(Figure 3). The change in antibiotic prescription rates was not different in the
first month compared to the last two months of the post-test period.

Discussion
We were able to show that a tailored intervention to implement a guideline for
acute cough optimised GPs'antibiotic prescribing for adult patients with acute
cough. Compared to controls patients in the intervention group were prescribed
less antibiotics. If GPs in the intervention group prescribed antibiotics, these
were more in line with the guideline. No significant differences were found in
patients'symptom resolution.

96

Cluster randomised controlled trial of a tailored


professional intervention to optimise prescribing

Table 5. Mean and difference in change of medication cost in the subset of patients
with an antibiotic prescription from the perspective of the National Sickness and
Invalidity Insurance Institute (NSIII)
Medication cost
Pre-Test
Post-Test
Change
Difference
MDadj (95%CI)*

Intervention

Control

22.86
16.75
- 6.11

21.48
22.35
+0.87

- 6.97
- 6.89 (- 11.77 (- 2.02))

* Mean difference is based on the model assuming equal medication cost in intervention and control group in
the Pre-Test period, and is adjusted for patient characteristics (see box 2)

Figure 2 Symptom resolution of patients with acute cough: graph of times to symptom
resolution vs. cumulative proportion symptomatic patients (Kaplan-Meier).

1,0

Cumulative Proportion Symptomatic

0,9
0,8

Pre-Test Control
Post-Test Control
Pre-Test Intervention
Post-Test Intervention

0,7
0,6
0,5
0,4
0,3
0,2
0,1
0,0

p25

p50

p75

10

15

20

25

29

Days

97

Chapter VI

Figure 3 Rates of antibiotic use in consultations for acute cough before and after the
tailored interventions from all practices with more than 10 consultations in each period

Percentage of patients receiving antibiotics after intervention

100%
90%
80%
70%
60%
50%
40%
30%
20%

Intervention
Control

10%
0%
0%

10%

20%

30%

40%

50%

60%

70%

80%

Percentage of patients receiving antibiotics before intervention

98

90%

100%

Cluster randomised controlled trial of a tailored


professional intervention to optimise prescribing

These kind of trials should not only contribute to evidence of the effect, but
also to understanding of the mechanism.

The evidence of effect


Study limitations. The results may be biased due to the recruitment, nonresponse and response quality of the GPs. The recruited GPs did not differ
from the other 64 GPs approached for this study nor from the other 108 GPs
responding in the questionnaire study,17 as we reported elsewhere.37 Only more
male GPs agreed to participate (63/85 vs. 36/64: p=0.02, vs. 64/108:p=0.03
respectively). Furthermore, their age and gender distribution is similar to
national averages. The GPs responding in the Pre-Test (n=72), those
responding in the Post-Test (n=59) and those responding in one (n=74) or in
both study periods (n=56) did not significantly differ from the respective not
responding GPs. In all responding GPs subgroups GPs in the intervention
group
were
similar
to
those
in
the
control
group.
Response quality, assessed by means of testing for differences in the proportion
of patients eligible for recruitment actually included in the study, in the
proportion of the included patients eligible for analysis, and in the proportion
of the latter patients responding with patient diaries, as well as by testing for
differences in the median cluster size, was similar in both study groups and
study periods. It was not feasible to increase the sample size in the post-test
because GPs were unable to include more than 10 patients on average per study
period.
Patient characteristics also might influence the results of this study, and
because of the nature of the interventions, participating practices knew the
group to which they were assigned. The patients included and eligible for
analysis were younger than those eligible for recruitment. Their characteristics
however did not differ between the study groups nor between the study
periods, except for risk for thrombo-embolic disease, duration of cough,
presence of sputum, of ronchi, of loss of appetite, and of a referral.
The professional intervention. Adjusting the main outcome measure, the
antibiotic prescribing rate, for these differences, patients in the intervention
were less likely to receive an antibiotic prescription compared to controls after
our intervention.

99

Chapter VI
We might have underestimated the rates for antibiotic prescriptions because the
practitioners may not have registered this information correctly in some
instances on the pre-printed forms we provided. We preferred this data
collection method since we also aimed to collect information not available in
the many different electronic medical record systems used in Belgium or in
other sources of prescribing data linked to clinical information. This is unlikely
to have differed between the groups and is therefore unlikely to have affected
the results. It is possible that we underestimated the reduction in the
prescription of antibiotics for acute cough. We do not know how often patients
were told that antibiotics normally are not necessary but received an antibiotic
prescription for use "if needed." However, we know from the patient diaries
that patients in the intervention group did not purchase nor took the prescribed
antibiotics less often than patients in the control group. Figure 2 shows the
importance of using adequately sized cluster randomised controlled trials to
evaluate interventions to support the implementation of guidelines. Large
variation exists in practice and in the extent of change among practices.
The public campaign. A national campaign, which coincided with our
professional intervention, provided health education of the general public.
Though the study does not, and did not set out to, compare the effect of the
coincidence of a national public campaign and a professional intervention, the
design of our study also allowed to test for an effect of the national public
campaign on the antibiotic prescription rates for acute cough. We agree with
Flottorp et al. that uncontrolled or inadequately controlled before and after
evaluations in selected practices are likely to have spurious results that are, at
best, difficult to interpret.38 Nonetheless, the similarities of the effectiveness of
our intervention when assessing differences between the intervention and
control group after our intervention and when assessing pre-post differences in
the intervention group, together with the absence of pre-post differences in the
control group, when adjusting for differences in patient characteristics,
suggests the national campaign had no effect on antibiotic prescribing after our
intervention.
In contrast to the public campaign our professional intervention not only
resulted in a reduction of antibiotic consumption. It changed the kind of
antibiotics prescribed form less desirable to more desirable antibiotics as well.
Price to pay. This trial on the implementation of a guideline to optimise
antibiotic prescribing not only looked at prescribing as outcome, but took
patient outcomes into account as well. Though antibiotics are not needed for

100

Cluster randomised controlled trial of a tailored


professional intervention to optimise prescribing
most patients with acute cough, indiscriminately reducing antibiotic use will
withhold some patient subgroups from the benefits of antibiotics. Less and
other antibiotics prescribed by the GP did not affect the time to symptom
resolution of the patients in the intervention group compared to the controls
after our intervention.
Another notable aspect of our study was the short duration of the educational
intervention compared with some other studies that have used repeated
education over several weeks; for example, to improve adolescent health care
39
Despite the responding GPs were visited more than fifteen times by medical
representatives per month, a single outreach visit resulted in the desired
changes during the follow up.
We reached 36 out of 42 GPs with our intervention at a total cost of 8514.22,
or an intervention cost of 236.51 per GP. The intervention resulted in a
significant reduction of the reimbursement cost for antibiotics from the
perspective of the NSIII of nearly 7.

Understanding the mechanism


Recently, Gross and Pujat40 concluded that for implementing guidelines for
appropriate antimicrobial usage multifaceted implementation methods seem to
be the most successful. Although more complex interventions to implement
guidelines tend to be most effective, their effectiveness varies, they require
more resources, and it is difficult to know which interventions to use.
Identifying barriers to change and tailoring interventions to address these is a
logical approach to selecting appropriate interventions. Still the effectiveness
of tailored interventions remains uncertain.38 Two recent cRCTs reported the
effect of interventions addressing either general practices38 or primary
healthcare teams,41 not individual general practitioners. In contrast to the
tailored interventions which only had little effect on the antibiotic prescribing
rates for sore throat38 and in contrast to the educational outreach visits, which
also coincided with a national campaign, and which did not improve the
influenza vaccination rates,41 our intervention had a substantial effect on
antibiotic prescribing.
In stead of a one size fits all approach, we really tailored the interventions to
the needs of individual general practitioners. Tailoring the intervention at this
level might have greater effect. After all individual approaches seem to have a
greater impact on prescribing than group approaches.42 43 The individual

101

Chapter VI
approach of face to face meeting, academic detailing, to improve antibiotic
prescribing proved to be successful in 8 studies in primary care.40 Furthermore,
we actively supported the GPs with the outreach visits and we might have
identified important barriers to change. Our educational program mainly
addressed the barriers relating to the individual prescriber s barriers to change,
focussing on non-medical reasons for prescribing. We did not provide
individual feedback on prescribing or on decision criteria.44 Studies of scoring
rules for sore throat have failed to show that they lower the rates for antibiotic
prescription.45 46
We have not identified trials of the implementation of a guideline for acute
cough similar to ours. The key messages of the pre-final version of the
guideline used for this cRCT are the same as those of the final version.47 , also
available now for all general practitioners at http://www.wvvh.be. Although
our evidence base was rather poor, and uncertainty about the evidence may
affect doctor'
s behaviour, identifying, understanding and modifying tacit expert
knowledge and promoting the ownership of change amongst professionals
appeared to be more important in altering behaviour in accordance with the
guideline.48
If we also distinguish between an agenda for action and one for future research,
the evidence of effectiveness supports this implementation strategy of the
guideline to optimise antibiotic prescribing on a larger scale. Further research
efforts should be devoted to understand the interaction between public
campaigns and professional interventions and to cost-effectiveness studies.
Whereas the public campaign transiently reduced antibiotic consumption and
saved money,20 21 the involvement of the prescribers has the potential of
influencing the prescribing decision as well, e.g. the kind of antibiotics
prescribed.

Conclusions
The described strategy to support the implementation of the guideline, tailored
to address identified barriers to optimise antibiotic prescribing for acute cough,
achieved the goals of the public campaign: Antibiotics: Use them less often,
but better .

102

Cluster randomised controlled trial of a tailored


professional intervention to optimise prescribing
This trial is assigned the International Standard Randomised Controlled Trial
Number (ISRCTN) ISRCTN09811591 by Current Controlled Trials Ltd.

References
1. Koninklijke Academie voor Geneeskunde van Belgi [Belgian Royal
Academy for Medicine]. Advies inzake het overgebruik van antibiotica
[Advise concerning the overuse of antibiotics]. Tijdschr Geneesk 1999;55:1734.
2. Wise R, Hart T, Cars O, Streulens M, Helmuth R, Huovinen P, Sprenger M.
Antimicrobial resistance. Is a major threat to public health [editorial]. BMJ
1998;317:609-10.
3. De Melker R. Efficacy of antibiotics in frequently occurring airway
infections in Family Practice. Ned Tijdschr Geneeskd 1998;142:452-6.
4. Cars O, Mlstad S, Melander S. Variation in antibiotic use in the European
Union. Lancet 2001;357:1851-3.
5. Seppala H, Klaukka T, Vuopio-Varkila J, Muotiala A, Helenius H, Lager K,
et al. The Effect of Changes in the Consumption of Macrolide Antibiotics on
Erythromycin Resistance in Group A Streptococci in Finland. NEJM
1997;337:441-6.
6. Huovinen P, Cars O. Control of antimicrobial resistance: time for action
[editorial]. BMJ 1998;317:613-4.
7. Gosden T, Torgerson D. The effect of fundholding on prescribing and
referral costs: a review of the evidence. Health Policy 1997;40:103-14.
8. O'
Connor P, Solberg L, Christianson J, Amundson G, Mosser G. Mechanism
of action and impact of a cystitis clinical practice guideline on outcomes and
costs of care in an HMO. Jt Comm J Qual Improv 1996;22:673-82.
9. O'
Connor P, Amundson G, Christianson J. Performance Failure of an
Evidence-Based Upper Respiratory Infection Clinical Guideline. J Fam Pract
1999;48:690-7.

103

Chapter VI
10. Saint S, Scholes D, Fihn S, Farrell R, Stamm W. The effectiveness of a
clinical practice guideline for the management of presumed uncomplicated
urinary tract infection in women. Am J Med 1999;106:636-41.
11. Gill P, Makela M, Vermeulen K, Freemantle N, Ryan G, Bond C, et al.
Changing doctor prescribing behaviour. Pharm World Sci 1999;21:158-67.
12. Perez-Cuevas R, Guiscafre H, Munoz O, Reyes H, Tome P, Libreros V, et
al. Improving physician prescribing patterns to treat rhinopharyngitis.
Intervention strategies in two health systems of Mexico. Soc Sci Med
1996:1185-94.
13. Coenen S, Kuyvenhoven M, Butler C, Van Royen P, Verheij T. Variation
in European antibiotic use [letter]. Lancet 2001;358:1272.
14. Deschepper R, Vander Stichele R, Haaijer-Ruskamp F. Cross-cultural
differences in lay attitudes and utilisation of antibiotics in a Belgian and a
Dutch city. Patient Educ Couns 2002;48:161-9.
15. Coenen S, Van Royen P, Denekens J. Diagnosis of acute bronchitis [letter;
see reply]. J Fam Pract 1999;48:471-2.
16. Coenen S, Van Royen P, Vermeire E, Hermann I, Denekens J. Antibiotics
for coughing in general practice: a qualitative decision analysis. Fam Pract
2000;17:380-5.
17. Coenen S, Michiels B, Van Royen P, Van der Auwera J-C, Denekens J.
Antibiotics for coughing in general practice: a questionnaire study to quantify
and condense the reasons for prescribing. BMC Fam Pract 2002;3:16 (10p).
18. Fahey T, Stocks N, Thomas T. Quantitative systematic review of
randomised controlled trials comparing antibiotic with placebo for acute cough
in adults. BMJ 1998;316:906-10.
19. Smucny J, Becker L, Glazier R, McIsaac W. Are Antibiotics Effective
Treatment for Acute Bronchitis? A Meta-Analysis. J Fam Pract 1998;47:45360.
20. A public campaign for a more rational use of antibiotics. 11th European
Congress of Clinical Microbiology and Infectious Diseases (ECCMID); 2001;
Istanbul, Turkey.
21. Watson R. Belgium cuts antibiotic use by 12%. BMJ 2001;323:710b-.

104

Cluster randomised controlled trial of a tailored


professional intervention to optimise prescribing
22. Soumerai S, Avorn J. Principles of educational outreach ('
academic
detailing'
) to improve clinical decision making. JAMA 1990;263:549-6.
23. The AGREE Collaboration. AGREE Instrument. Available at:
http://www.agreecollaboration.org. Accessibility verified February 7,2003.
24. Metlay J, Kapoor W, Fine M. Does This Patient Have CommunityAcquired Pneumonia? Diagnosing Pneumonia by History and Physical
Examination. JAMA 1997;278:1440-5.
25. Zaat J, Stalman W, Assendelft W. Hoort, wie klopt daar [Listen, who is
knocking]? Huisarts en Wetenschap 1998;41:461-9.
26. Fine M, Auble T, Yealy D, Hanusa B, Weisfeld L, Singer D, et al. A
prediction rule to identify low-risk patients with community-acquired
pneumonia. NEJM 1997;336:243-50.
27. Wang E, Kellner J, Arnold S. Antibiotic-resistant Streptococcus
pneumoniae. Implications for medical practice. Can Fam Physician
1998;44:1881-8.
28. Nava J, Bella F, Garau J, Lite J, Morera M, Marti C, et al. Predictive
factors for invasive disease due to penicillin-resistant Streptococcus
pneumoniae: a population-based study. Clin Infect Dis 1994;19:884-90.
29. Macfarlane J, Holmes W, Macfarlane R, Britten N. Influence of patients'
expectations on antibiotic management of acute lower respiratory tract illness
in general practice: questionnaire study. BMJ 1997;315:1211-4.
30. Britten N, Ukoumunne O. The influence of patients'hopes of receiving a
prescription on doctors' perceptions and the decision to prescribe: a
questionnaire survey [see comments]. BMJ 1997;315:1506-10.
31. Cockburn J, Pit S. Prescribing behaviour in clinical practice: Patients'
expectations and doctors'perceptions of patients'expectations - a questionnaire
study. BMJ 1997;315:520-3.
32. Wears R. Advanced statistics: Statistical methods for analyzing cluster and
cluster-randomized data. Acad Emerg Med 2002;9:330-41.
33. SAS System for Windows [program]. 8.02 version. Cary, NC: SAS
Institute Inc., 2001.

105

Chapter VI
34. STATISTICA for Windows [program]. 6.0 version. Tulsa, OK: StatSoft,
Inc., 2001.
35. Liang K, Zeger S. Longitudinal data analysis using generalized linear
models. Biometrika 1986;73:13-22.
36. Ridout M, Demtrio C, Firht D. Estimating intraclass correlation for binary
data. Biometrics 1999;55:137-48.
37. Coenen S, Michiels B, Renard D, Denekens J, Van Royen P. Antibiotics
for coughing in family practice: physicians'perception of patients'requests
determines prescription behavior. J Fam Pract 2002: Submitted.
38. Flottorp S, Oxman AD, Havelsrud K, Treweek S, Herrin J. Cluster
randomised controlled trial of tailored interventions to improve the
management of urinary tract infections in women and sore throat. BMJ
2002;325:367-70.
39. Sanci LA, Coffey CMM, Veit FCM, Carr-Gregg M, Patton GC, Day N, et
al. Evaluation of the effectiveness of an educational intervention for general
practitioners in adolescent health care: randomised controlled trial
Commentary: Applying the BMJ'
s guidelines on educational interventions.
BMJ 2000;320:224-30.
40. Gross P, Pujat D. Implementing practice guidelines for appropriate
antimicrobial usage: a systematic review. Med Care 2001;39:II55-69.
41. Siriwardena A, Rashid A, Johnson M, Dewey M. Cluster randomised
controlled trial of an educational outreach visit to improve influenza and
pneumococcal immunisation rates in primary care. Br J Gen Pract
2002;52:735-40.
42. van Eijk MEC, Avorn J, Porsius AJ, de Boer A. Reducing prescribing of
highly anticholinergic antidepressants for elderly people: randomised trial of
group versus individual academic. BMJ 2001;322:654-7.
43. Figueiras A, Sastre I, Tato F, Rodriguez C, Lado E, Caamano F, et al. Oneto-one versus group sessions to improve prescription in primary care: a
pragmatic randomized controlled trial. Med Care 2001;39:158-67.
44. Veninga C, Lagerlov P, Wahlstrom R, Muskova M, Denig P, Berkhof J, et
al. Evaluating an educational intervention to improve the treatment of asthma

106

Cluster randomised controlled trial of a tailored


professional intervention to optimise prescribing
in four European countries. Drug Education Project Group. Am J Respir Crit
Care Med 1999;160:1254-62.
45. Poses R, Cebul R, Wigton R. You can lead a horse to water--improving
physicians'knowledge of probabilities may not affect their decisions. Med
Decis Making 1995;15:65-75.
46. McIsaac W, Goel V. Effect of an Explicit Decision-support Tool on
Decisions to Prescribe Antibiotics for Sore Throat. Med Decis Making
1998;18:220-8.
47. Coenen S, Van Royen P, Michels J, al e. Aanbeveling voor goede medische
praktijkvoering: Acute hoest [Good Clinical Practice Guideline: Acute Cough].
Huisarts Nu 2002;31:391-411.
48. Kelley MA, Tucci JM. Bridging the quality chasm [editorial]. BMJ
2001;323:61-2.

107

108

VII
Patients views on respiratory symptoms and antibiotics

Introduction
In Europe there is a striking variation in outpatient antibiotic usage.1 In 1997,
in Belgium 27 Defined Daily Dosages per 1000 inhabitants per day were
consumed, compared to 18 for the UK and 9 for the Netherlands. And looking
at 1998 data from the Alexander Project antimicrobial resistance of
Streptococcus pneumoniae, especially macrolide resistance, is correlated with
these figures.2
Cough is the most common reason for encounter (for respiratory tract
infections) in general practice.3 4 Likewise patients frequently consult their GP
for other respiratory symptoms, such as earache and sore throat. And though
antibiotics have no or very limited effects,5-7 antibiotics are frequently
prescribed for these respiratory symptoms.8-10 Moreover, especially in primary
care and for respiratory tract infections antibiotics are overprescribed.11-13 This
overprescribing wastes money, exposes patients unnecessarily to the risk of
side effects, encourages reconsulting for similar problems and causes
antimicrobial resistant bacteria.8
Non-medical determinants, either patient- or physician-related, appear to have
a major influence on this overprescribing.14-17 Macfarlane showed non-medical
determinants influenced nearly half the prescribing decisions for acute lower
respiratory tract symptoms.18 Patient pressure was cited most frequently. Little

109

Chapter VII
showed non-medical factors influenced prescribing for upper respiratory tract
infections. 19 We found that GPs'perception of patients'request for an
antibiotic significantly influenced prescribing for acute cough.20 Furthermore,
patients seem to overestimate the effectiveness of antibiotics.18 In a US study,
79% of respondents believed antibiotics are effective for a discoloured nasal
discharge, and 31%- 61% believed antibiotics are effective against colds.21 22
Patients views on respiratory symptoms and antibiotics thus deserve
consideration as possible determinants of antibiotic use.1 23
And since antibiotic resistance is an international problem, an awareness of
possible similarities and differences in views between countries might be
helpful in designing international interventions to optimise antibiotic usage,
particularly in Europe. For a better understanding of patients'views on frequent
respiratory symptoms, cough, earache and sore throat, we performed a postal
questionnaire study with patients in Belgium, in the UK and in the Netherlands.
This study was a collaboration between the Department of General Practice,
Faculty of Medicine, of the University of Antwerp in Belgium, the Department
of General Practice, College of Medicine, of the University of Wales in the UK
and the Department of General Practice and Patient Oriented Research of the
University Medical Center Utrecht in The Netherlands, the initiator of the
study. For this thesis the emphasis will be laid on the results for Belgium.

Methods

Study sample
In Belgium, as well as in the UK and in the Netherlands, four general practices
were recruited purposefully to represent a sample with a range of social and
educational levels: two rural and two urban, each with one practice located in a
deprived and one in a non-deprived area. In each practice one hundred patients
(age between 18 and 65 years) were randomly selected. Those who were
unable to read the questionnaire (because of a language problem or mental
disorder) or were suffering from a serious disabling disease were excluded by
GP s screening the selection lists.

110

Patients views on respiratory symptoms and antibiotics

Data collection
All selected patients received a questionnaire. After ten days a first reminder
was sent. Two weeks after the first reminder a second reminder with a new
questionnaire was sent to patients from practices with a response rate lower
than 50%.

Questionnaire
The questionnaire was based on a questionnaire previously piloted in the
Netherlands24 and further developed in collaboration with the three centres.
The questionnaire contained the following domains: seriousness (need to
consult a general practitioner and perceived seriousness), self-limiting
character, effectiveness of antibiotics (to speed up recovery and to prevent
deterioration), adverse effects of antibiotics and aetiology. Patients were asked
to rate their agreement to statement on five-point scale with categories ranging
from 1 (strongly disagree) to 5 (strongly agree).

Data processing and statistical analysis


The data were entered with a 10% double-check. The analysis was performed
in the Dutch centre. Extremely skewed, bipolar items or those, which were not
completed by more than 20% of the responders were excluded from further
analysis. The items were grouped into 6 clusters of views relating to
respiratory tract symptoms: need to consult a general practitioner, perceived
seriousness, perceived self limiting character, perceived effectiveness of
antibiotics to speed up recovery, perceived effectiveness of antibiotics to
prevent deterioration, with each cluster containing items relating to cough, sore
throat and earache respectively, and side effects of antibiotics. This grouping of
items was controlled by means of principal component factor analysis with
Varimax rotation. The inter-correlation between the items in each cluster was
calculated by means of Cronbach s alpha. Results were expressed as means. In
addition, correlations between the views (Pearson correlation coefficient r)
were described, with correlations
.25 being concerned as relevant
(corresponding with R 0.05).
All analyses were performed with SPSS 10.0.

111

Chapter VII
Results

Respondents
In Belgium in total 243 evaluable questionnaires were returned (response rate
60.8%)(Table). Inter-practice variation was small. The respondents mean age
was 41.3 years, 60.2% were women and they were highly qualified. Reported
use of antibiotics during the past two years was very high (62.1%). This
applied to antibiotics prescribed by a medical doctor as well as use of
antibiotics not prescribed by a medical doctor for the illness for which they
were taken. Forty-five percent of the respondents reported experiencing
respiratory tract symptoms during the previous month.

Grouping of items
Principal component factor analysis with Varimax rotation endorsed the chosen
clusters. The clusters need to consult a general practitioner and perceived
seriousness both loaded on one factor, as did the clusters perceived
effectiveness of antibiotics to speed up recovery and perceived effectiveness of
antibiotics to prevent deterioration. However, for semantic reasons, they were
analysed apart. The inter-correlation between the items within the six clusters
of views was moderate to strong (Cronbach s ranged from .57 to .88), which
meant that the items reflect one concept underlying each cluster of items.

Views on respiratory tract symptoms and antibiotics


Belgian respondents reported a great need to consult a general practitioner with
respiratory tract symptoms (mean = 4.5) and considered these symptoms as
serious (mean = 4.2) and less as self-limiting (mean = 2.9). They reported
similar perceptions of the effectiveness of antibiotics to speed recovery (mean
= 3.1) and to prevent respiratory tract symptoms deteriorating (mean = 2.8).
Belgian patients often endorsed concerns about adverse effects from antibiotics
(mean = 4.3) and they also regarded a general practitioner as the best person to
consult with respiratory tract symptoms. More than 75% of the respondents
appeared to agree (strongly) with the statement that bacteria are an important
cause for respiratory tract symptoms.

112

Patients views on respiratory symptoms and antibiotics

Correlations between views


There was a moderate correlation between need to consult a general
practitioner and perceived seriousness (r:.41) and a strong correlation
between perceived effectiveness of antibiotics to speed up recovery and
perceived effectiveness of antibiotics to prevent deterioration (r:.61), as
could be expected from the factor analysis. The remaining correlations between
these four views were weak. The view that a general practitioner is the best
person to consult for respiratory tract symptoms was a core view being
correlated with all views except the views concerning perceived seriousness
and side effects. Perceiving of the self-limiting character of respiratory tract
symptoms was negatively correlated with the general practitioner as the best
person to consult and did not correlate with the remaining clusters (r:-.31).
Perceiving bacteria as an important cause of respiratory tract symptoms was
only correlated with the perceived effectiveness of antibiotics to prevent
deterioration (r:.27).

113

Chapter VII
Table Demographic characteristics (mean and SD) of respondents from the
Netherlands, UK and Belgium and their views on respiratory tract symptoms and
(1)
antibiotics (AB) (mean and SD; Cronbachs alpha)
Netherlands

UK

Belgium

All

(n=247)

(n=188)

(n=243)

(n=678)

61.7

37.6

60.2

52.2

40.1 (11.8)

44.9 (12.1)

41.3 (13.0)

41.9 (12.4)

64.0

55.6

65.8

62.3

-low

10.7

17.9

7.2

11.4

-medium

58.1

46.2

51.7

52.6

-high

31.2

35.8

41.1

36.1

31.8

54.0

62.1

48.8

10

12

27

Countries
Resonse rate (%)
Age
Gender (% female)
Highest level of education

Antibiotics past 2 years


-prescribed (%)
- not prescribed (abs)
In case of respiratory tract symptoms

Need to consult a general practitioner

3.8 (1.0) .75

3.6 (0.7) .66

4.5 (0.6) .57

4.0 (0.9) .74

Cough and raised temperature >2 days

3.5 (1.4)

3.4 (1.0)

4.4 (1.0)

3.8 (1.3)

Sore throat and raised temperature >2


days
A child with earache >2 two days

3.4 (1.4)

3.2 (1.0)

4.4 (0.9)

3.7 (1.3)

4.5 (0.8)

4.1 (0.7)

4.8 (0.5)

4.5 (0.7)

Perceived seriousness

3.6 (1.0) .80

3.4 (0.7) .74

4.2 (0.8) .79

3.8 (0.9) .82

Cough and raised temperature

3.5 (1.2)

3.3 (1.0)

4.1 (1.0)

3.7 (1.1)

Sore throat and raised temperature

3.5 (1.1)

3.3 (0.9)

4.2 (0.9)

3.7 (1.1)

A child with earache and raised


temperature

3.8 (1.0)

3.7 (0.8)

4.3 (0.8)

3.9 (1.0)

Perceived self-limiting character

3.6 (0.9) .67

3.4 (0.7) .67

2.9 (1.0) .67

3.3 (0.9) .70

Cough better without treatment < 2 weeks

3.7 (1.2)

3.5 (1.0)

2.9 (1.3)

3.4 (1.3)

Sore throat better without treatment < 1


week
Earache almost always gets better without
treatment within two days

4.0 (1.1)

3.7 (0.9)

3.2 (1.3)

3.6 (1.2)

3.1 (1.2)

3.0 (0.9)

2.7 (1.2)

2.9 (1.1)

114

Patients views on respiratory symptoms and antibiotics


Table continued.
In case of respiratory tract symptoms

Antibiotics speed up recovery

3.3 (1.1) .83

2.9 (0.7) ..69

3.1 (1..1) .84

3.1 (1.0) .82

Antibiotics speed recovery from coughs

3.1 (1.3)

2.7 (0.8)

2.9 (1.3)

2.9 (1.2)

Antibiotics speed recovery from sore


throats
Antibiotics speed recovery from earache

3.3 (1.2)

2.8 (1.0)

2.8 (1.3)

3.1 (1.2)

3.4 (1.2)

3.1 (1.3)

2.9 (1.3)

3.3 (1.1)

AB stop deteriorating symptoms

2.9 (1.2) .86

2.8 (0.8) .77

2.8 (1.2) .88

2.8 (1.1) .85

Antibiotics stop cough deteriorating

2.6 (1.4)

2.6 (0.9)

2.6 (1.3)

2.6 (1.2)

Antibiotics stop sore throats deteriorating

2.8 (1.3)

2.8 (1.0)

2.8 (1.3)

2.8 (1.2)

Antibiotics stop earache deteriorating

3.1 (1.3)

3.1 (1.3)

2.9 (1.3)

3.0 (1.2)

4.0 (1.0) .66

3.8 (0.9) .74

4.3 (1.0) .77

4.0 (1.0) .74

Frequent use can cause problems for your


health

4.1 (1.0)

3.9 (0.9)

4.5 (0.9)

4.2 (1.0)

Frequent use can cause problems for the


community

3.7 (1.3)

3.6 (1.1)

4.2 (1.2)

3.9 (1.2)

A GP is the best person to go

3.6 (1.1)

3.4 (1.0)

4.2 (0.9)

3.8 (1.1)

Bacteria are an important cause

4.2 (1.0)

3.8 (0.8)

4.1 (0.9)

4.0 (0.9)

Side effects of antibiotics

(2)

(1) The answers were ranged as follows: 1: totally disagree to 5: totally agree
(2) This view is based on two items, so the Pearson correlation coefficient was used in stead of
Cronbachs alpha

115

Chapter VII
Discussion
Because we sampled from only four practices, our results should be treated
cautiously.
In the table, the results for Belgium are presented together with those for the
Netherlands and the UK. Patient report of antibiotic use during the preceding
two years was highest in Belgium and lowest in the Netherlands with the UK in
between. Belgian respondents perceived a higher need to consult a general
practitioner with respiratory symptoms and viewed these as more serious and
less self-limiting compared to UK and Dutch respondents. This is congruent
with respondents higher reported use of antibiotics, as well as higher national
figures for antibiotic prescription in Belgium compared with the UK and the
Netherlands.1 This congruence suggests validity of our data. The
intercorrelation between the items with the six clusters was moderate to strong
(Cronbach s ranged from .70 to .85 for all respondents together).
There were smaller differences between the UK and Dutch respondents views
as might be expected, given the differences in national antibiotic use.
Countries health care delivery characteristics, such as having personal patients
lists, and the degree of participation of GP'
s in peer review groups addressing
prescribing behaviour, having national guidelines on management and patient
education, and physician availability also may contribute to the international
variance in views as well as in antibiotic use.23
Given the differences between the countries and the intercorrelation between
the clusters, patient-directed interventions might fruitfully highlight the benign
nature of the vast majority of respiratory tract symptoms, which makes
consulting the general practitioner generally unnecessary. Stressing the danger
of side effects of antibiotics might be less important. Further studies must be
concentrated on the responsiveness to changes in views of this scale and the
relevance of such changes in relation to reduction of antibiotic use.

116

Patients views on respiratory symptoms and antibiotics


References
1. Cars O, Mlstad S, Melander S. Variation in antibiotic use in the European
Union. Lancet 2001;357:1851-3.
2. Schito GC, Debbia EA, Marchese A. The evolving threat of antibiotic
resistance in Europe: new data from the Alexander Project. J Antimicrob
Chemother 2000;46:3-9.
3. De Maeseneer J. Huisartsgeneeskunde: een verkenning [General practice: an
exploration]. Proefschrift [Dissertation] Rijksuniversiteit Gent, 1989.
4. Okkes I, Oskam S, Lamberts H. Van klacht naar diagnose [From complaint
to diagnosis]. Bussum: Coutinho, 1998.
5. Smucny J, Fahey T, Becker L, Glazier R, McIsaac W. Antibiotics for acute
bronchitis (Cochrane Review). In: The Cochrane Library, Issue 4, 2002.
Oxford: Update Software.
6. Del Mar C, Glasziou P, Spinks A. Antibiotics for sore throat (Cochrane
Review). In: The Cochrane Library, Issue 4, 2002. Oxford: Update Software.
7. Glasziou P, Del Mar C, Sanders S, Hayem M. Antibiotics for acute otitis
media in children (Cochrane Review). In: The Cochrane Library, Issue 4,
2002. Oxford: Update Software.
8. Butler C, Rollnick S, Kinnersley P, Jones A, Stott N. Reducing antibiotics
for respiratory tract symptoms in primary care: consolidating '
why' and
considering '
how'
. Br J Gen Prac 1998;48:1865-70.
9. Coenen S, van Royen P, Denekens J. Reducing antibiotics for respiratory
tract symptoms in primary care: '
why'only sore throat, '
how'about coughing?
[letter]. Br J Gen Pract 1999;49:400-1.
10. Gonzales R, Bartlett J, Besser R, Cooper R, Hickner J, Hoffman J, et al.
Principles of appropriate antibiotic use for treatment of acute respiratory tract
infections in adults: background, specific aims, and methods. Ann Emerg Med
2001;37:690-7.

117

Chapter VII
11. Wise R, Hart T, Cars O, Streulens M, Helmuth R, Huovinen P, Sprenger
M. Antimicrobial resistance. Is a major threat to public health [editorial]. BMJ
1998;317:609-10.
12. Koninklijke Academie voor Geneeskunde van Belgi [Belgian Royal
Academy for Medicine]. Advies inzake het overgebruik van antibiotica
[Advice concerning the overuse of antibiotics]. Tijdschr Geneeskd
1999;55:173-4.
13. Kuyvenhoven M, Verheij T, de Melker R, van der Velden J. Antimicrobial
agents in lower respiratory tract infections in Dutch general practice. Br J Gen
Pract 2000;50:133-4.
14. Britten N, Ukoumunne O. The influence of patients'hopes of receiving a
prescription on doctors' perceptions and the decision to prescribe: a
questionnaire survey. BMJ 1997;315:1506-10.
15. Cockburn J, Pit S. Prescribing behaviour in clinical practice: Patients'
expectations and doctors'perceptions of patients'expectations - a questionnaire
study. BMJ 1997;315:520-3.
16. Coenen S, Van Royen P, Vermeire E, Hermann I, Denekens J. Antibiotics
for coughing in general practice: a qualitative decision analysis. Fam Pract
2000;17:380-5.
17. Coenen S, Michiels B, Van Royen P, Van der Auwera J-C, Denekens J.
Antibiotics for coughing in general practice: a questionnaire study to quantify
and condense the reasons for prescribing. BMC Fam Pract 2002;3:16 (10p).
18. Macfarlane J, Holmes W, Macfarlane R, Britten N. Influence of patients'
expectations on antibiotic management of acute lower respiratory tract illness
in general practice: questionnaire study. BMJ 1997;315:1211-4.
19. Little P, Williamson I, Warner G, Gould C, Gantley M, Kinmonth AL.
Open randomised trial of prescribing strategies in managing sore throat. BMJ
1997;314:722-7.
20. Coenen S, Michiels B, Renard D, Denekens J, Van Royen P. Antibiotics
for coughing in family practice: physicians'perception of patients'requests
determines prescription behavior. J Fam Pract: Submitted.

118

Patients views on respiratory symptoms and antibiotics


21. Braun B, Fowles J, Solberg L, Kind E, Healey M, Anderson R. Patient
beliefs about the characteristics, causes, and care of the common cold: an
update. J Fam Pract 2000;49:153-6.
22. Mainous Ar, Zoorob R, Oler M, Haynes D. Patient knowledge of upper
respiratory infections: implications for antibiotic expectations and unnecessary
utilization. J Fam Pract 1997;45:75-83.
23. Coenen S, Kuyvenhoven M, Butler C, Van Royen P, Verheij T. Variation
in European antibiotic use [letter]. Lancet 2001;358:1272.
24. van Duijn H, Kuyvenhoven M, Welschen I, den Ouden H, Slootweg A,
Verheij T. Patients'and doctors'views on respiratory tract symptoms. Scand J
Prim Health Care 2002;20:201-2.

119

120

VIII
Antibiotics for coughing in general practice:
General discussion

Summary of the results


In this dissertation we aimed to describe the management of complaints about
coughing in general practice, and to optimise this management, especially
regarding the antibiotic prescribing decision. Patients views on the topic were
addressed as well.
We described the management of patients with acute cough in Flemish general
practice by means of qualitative and quantitative research methodologies.
Focus group research enabled us to generate hypotheses on GPs decision
making regarding complaints about coughing and the determinants underlying
their decisions.1 GPs have to deal with diagnostic uncertainty when trying to
distinguish between infectious and non-infectious causes of coughing. In
suspected respiratory tract infections, GPs want to make a distinction between
clinical syndromes such as bronchitis and pneumonia, viral and bacterial
respiratory tract infections, and upper and lower respiratory tract infections.
This also cannot be achieved with certainty on the basis of medical history and
clinical examination. Dealing with diagnostic uncertainty, GPs decisions are
directed at whether or not to prescribe antibiotics. For this therapeutic decision,
patient- and doctor-related factors, such the patient s expectations and the GP s

121

Chapter VIII
perceptions of these, also play a role. These non-medical reasons give rise to a
shift in the action threshold in favour of antibiotics, a phenomenon explained
by the chagrin factor .
A questionnaire study with adequate response enabled us to quantify and
condense the focus group determinants and confirmed the focus group
findings.2 The participating GPs (mean age:42.8 years; 65.9% men) considered
all the items included in the questionnaire: always the items relating to the lung
auscultation, often the items determining whether there is something unusual
happening both medical reasons and to a lesser extent non-medical reasons.
Non-medical as well as medical reasons supported antibiotic treatment.
By means of multivariable analysis of medical and non-medical data registered
by GPs about adult patients consulting at their practice with acute cough as one
of the most prominent complaints, the findings of the previous studies were
validated.3 The GPs perception of the patients'request for an antibiotic, a nonmedical reason that was mentioned in the focus groups and scored high in the
questionnaire study as a determinant for antibiotic prescribing, was
significantly associated with antibiotic prescribing. Antibiotics were prescribed
more often when a patient s request for an antibiotic was perceived and the
lung auscultation was normal or revealed only one abnormal finding.
Abnormal auscultatory findings were also associated with more prescribing.
Antibiotics are thus prescribed for medical reasons if these are available and,
especially when GPs have to deal with diagnostic uncertainty, non-medical
reasons favour antibiotic prescribing as well. Hence, clinical practice
guidelines and interventions to optimise antibiotic prescribing have to take
non-medical reasons for antibiotic prescribing into account.
To change the described management of acute cough, especially regarding the
antibiotic prescribing decision, a clinical practice guideline was developed
according to a standardised methodology defined by the Scientific College of
Flemish General Practitioners and in line with the AGREE-criteria.4 A tailored
professional intervention, including this guideline, one educational outreach
visit and a written reminder, to implement the guideline, was successful in
optimising antibiotic prescribing for patients with acute cough. In addition to a
reduction in antibiotic prescriptions, prescribed antibiotics were more in line
with the guideline recommendations and less expensive from the perspective of
the National Sickness and Invalidity Insurance Institute (NSIII). The change in
antibiotic prescribing did not affect the patients symptom resolution.5

122

General discussion
For a better understanding of patients' views about frequent respiratory
symptoms, cough, earache and sore throat, and antibiotic treatment, we
performed a postal questionnaire study with patients in Belgium, in the UK and
in the Netherlands. Belgian respondents perceived a higher need to consult a
GP with respiratory symptoms and viewed these as more serious and less selflimiting compared to UK and Dutch respondents.

Limitations of the project


The internal and the external validity of the description of the management of
patients with acute cough in Flemish general practice is limited by the
measurement of the topic and the selection of the recruited and participating
GPs. To understand this complex decision making process, especially the
antibiotic prescribing decision, we triangulated three different research
methodologies, starting with a qualitative study. Therefore, stating that in the
prescribing decision non-medical reasons such as the patients request for
antibiotics also play a role, seems a valid general conclusion. If there is
diagnostic uncertainty this is almost unavoidable.6 In addition, we have shown
that this irrational prescription behaviour can be explained by the so-called
chagrin factor7: GPs consider it less appropriate not to prescribe antibiotics
when this may prove to be necessary, than to prescribe antibiotics when not
necessary. The latter caused less chagrin to GPs. Furthermore, these finding
are in line with Butler s qualitative research regarding the prescribing decision
for sore throat.8
We confirmed our general conclusion by studying larger samples of GPs. And
though these were not strictly representative samples, non-medical reasons
have been shown to affect prescribing behaviour of GPs for respiratory tract
infections in the work of Little9 and MacFarlane10, and it has been suggested
that GPs perception of patient expectations may be the strongest determinants
for antibiotic prescribing by others as well.11 12
The development of the clinical practice guideline for acute cough was
hampered by the lack of good evidence for the management of acute cough.
We chose to study the management of acute cough, and not the management of
a respiratory syndrome like acute bronchitis, since most patients consult a GP
with complaints about coughing,13 14 and the diagnosis of respiratory
syndromes is not valid, nor reliable in general practice.15 Pooling the limited

123

Chapter VIII
evidence concerning the effectiveness of antibiotics, the latter problem was
also acknowledged,15 resulting in the use of acute (productive) cough as a
synonym or in stead of acute bronchitis in the most recent meta-analysis.16
Currently there are no clinical criteria to identify subsets of patients who are
most likely to benefit form antibiotic treatment, and the overall benefit from
antibiotics for patients with acute (productive) cough is limited: antibiotics do
not influence the (duration of) productive cough, nor the (duration of) the
limitations in work or activities; and of every 10 patients with acute
(productive) cough more than 8 will be clinically improved after 7-11 days
regardless the use of antibiotics; less than one patient extra will be improved
due to antibiotics, but as many patients will experience the side effects of
treatment. The management recommended in the guideline not to treat
immuno-competent adult patient with antibiotics after ruling out possibly life
threatening conditions such as pneumonia (severity assessment), and
considering other possible causes for acute cough than an uncomplicated
respiratory infection is in line with the currently available evidence. The nonmedical reasons for antibiotic prescribing were taken into account in the
guideline as well. And probably a guideline for acute cough is fitting in better
with daily practice than a guideline for acute bronchitis.
Writing and publishing guidelines however is not sufficient to change
antibiotic prescribing practices in primary care. Our intervention to implement
the guideline included dissemination of the guideline for acute cough, a short
one-to-one conversations between a detailer and a practitioner, with the goal of
persuading the practitioner to change behaviour in concordance with the
guideline through tailored information and evidence, and a written reminder.
We thus preferred the individual approach of academic detailing as
implementation method, and actively supported the GPs with educational
outreach visits. Furthermore, we tailored the intervention to identified barriers
to change relating to the individual prescriber. To facilitate better
implementation of guidelines on appropriate antibiotic management of
respiratory infections(, excluding pneumonia,) multifaceted implementation
methods are most useful, according to a systematic review by Gross and
Pujat,17 but not all studies have shown a positive effect.18 Furthermore, it is
difficult to determine which part or parts of complex methods are critical to
successful implementation. Academic detailing however appears to be useful
as an individual implementation method in primary care. Up to the start of our
intervention all the evidence for the effectiveness of academic detailing to

124

General discussion
optimise antibiotic prescribing by general practitioners came from outside
Europe. The success of implementation methods used in before-and-after
studies19 20 however should not necessarily be viewed as definitive evidence for
the utility of academic detailing. Selection bias can also occur in controlled
studies if the control patients are not randomised.21-23 Randomised controlled
trials typically compensate for such bias. But only one RCT showing a
reduction in antibiotic prescription was performed,24 and two RCTs showed
academic detailing had a significant impact on recommended drug usage and
prescribing costs.25 26
We performed a cluster-randomised controlled before and after study in
Flanders, Belgium, to assess the effectiveness of our implementation strategy.
We measured the prescription rates of antibiotics and, if antibiotics were
prescribed, prescription rates of the recommended antibiotics. In addition to the
prescribing costs, we assessed the cost of the intervention as well. And we took
patient outcome into account. Our results are in line with the assessments of
other professional interventions in primary care. Yet the indicators we used to
assess success did not allow to distinguish the effectiveness of the guideline
from the effectiveness of the implementation methods.
Although the participating GPs, the proportion of patients eligible for
recruitment included in the study and in the analysis, and the characteristics of
the latter patients were similar for both study groups in both study periods, a
selection of GPs and patients might have influenced the results. Crosscontamination of the study groups was unlikely, as was a Hawthorne effect
because the control group knew that a study was being done to improve
antibiotic usage for acute cough. The concurrent control group controlled for
unknown factors, such as changes in the microbial patterns, in the medical
delivery system and in provider knowledge and practices that occurred over
time. Such changes between the two study periods limit the ability to draw
definitive conclusions on the merits of an intervention assessed in before-andafter studies.
Implementing evidence-based guidelines is one of the most known, and best
studied examples of professional interventions. Although more complex
interventions to implement guidelines tend to be most effective, their
effectiveness varies and should not be exaggerated they require more
resources, and it is difficult to know which intervention to use. Furthermore,
the issue seems to be one that takes time. But, although the participating

125

Chapter VIII
physicians were frequently visited by advocates for the use of antibiotics, a
single visit by advocates for improved communication with their patients
optimised their antibiotic prescribing significantly.
We did not provide individual feedback on their prescribing, nor did we use
patient education or media support. In the context of a national public
campaign, which coincided with our professional intervention patient
information leaflets and booklets were distributed among all GPs and
pharmacies. This campaign also included TV spots and radio messages
informing the public on over consumption and misuse of antibiotics, the
resulting resistance problem and the self-limiting character of most frequent
infections in the community. Although this campaign was no longer effective
at the time of our assessment, the effect of our professional intervention may
include a significant interaction of the public campaign with our professional
intervention. Still a patient information leaflet describing the uncertain value of
antibiotics reduced antibiotic use in patients presenting with lower respiratory
tract infections.27 However this approach is appealing, it is clearly not the final
answer. Half of the patients took the antibiotics that their physician felt
unnecessary
Each year the NSIII provides GPs with individual feedback on their prescribing
of reimbursed pharmaceutical specialities, e.g. the gross amount for and the
volume in Daily Defined Dosage (DDD) of antimicrobials for systemic use
(Anatomic Therapeutic Chemical class J (ATC J). There was a delay of about
two years between the prescribing and the prescribing feedback before the start
of our intervention study in 1999. Before the randomisation of the GPs willing
to participate received feedback on ATC J prescribing for 1997. Before our
intervention they received the data for 1998. On the other hand an intervention
of repeated mailings containing confidential profiles of the prescribing habits
of an individual provider compared with those of his or her local groups and
peers and accompanying educational materials however can be successful,
even without the use of direct personal contact by academic detailers, opinion
leaders, or nurse or pharmacist implementers.28
Computerised decision-support programs that make recommendations as
orders are typed into the computer appear to be an excellent, real-time method
of guideline implementation. But all studies are hospital based.17

126

General discussion
As pointed out by Schaffner and his colleagues two decades ago, the challenge
to the medical profession is whether to develop and implement effective
programs to correct excess in medical practice or whether to leave this
responsibility to others, such as the government and managed care
organisations.19 We have now provided more evidence for the effectiveness of
implementing guidelines on the appropriate use of antibiotics by means of
academic detailing.
Concerning the assessment of patients views on respiratory symptoms and
antibiotics a postal questionnaire study was performed with patients in
Belgium, in the UK and in the Netherlands. The questionnaire was based on a
questionnaire previously piloted in the Netherlands29 and further developed in
collaboration with the three centres. Though internal consistency is good, the
validity of the identified domains and the questionnaire items and the
questionnaire s responsiveness to change needs further study.

Opportunities for further research


To preserve the effectiveness of antibiotics for the management of respiratory
tract infections, the focus of future research activities should be in line with the
description of the topic: Management of respiratory tract infections, for the
first call for the Sixth Framework Programme (FP6) published December 2002
by the European Commission in the work programme of the thematic priority
area 1: Life Sciences, Genomics and Biotechnology for Health under the
headings Combating major diseases, Applications-orientated genomic
approaches to medical knowledge and technologies, and Combating
resistance to antibiotics and other drugs. Consequently, the focus should be to
address current fragmentation by integrating microbial and human genomics
with clinical research and cost-benefit/cost-effectiveness studies towards a
common understanding of an improved evidence-based management of
community acquired RTIs with the aim of reducing antibiotic resistance. The
activities should take into account validation and implementation of novel
treatment, prevention and diagnostic approaches for various bacterial and viral
respiratory pathogens.
Research activities on microbial genomics could comprise the development,
standardisation and validation of innovative molecular methods to establish the
aetiology of RTIs, to elucidate the relative importance of various pathogens
127

Chapter VIII
with regard to morbidity and to determine the presence of genetic elements
conferring antimicrobial resistance. Identification of new antibiotic resistance
determinants and description of new resistance profiles are necessary to predict
future trends in bacterial resistance and to ensure optimal therapy for patients
as well.
Activities on human DNA could identify genetic risk factors for RTIs with
worse prognosis, e.g. community-acquired pneumonia.
By using qualitative as well as quantitative research methods clinical research
could provide a deeper and setting specific understanding of antibiotic
prescribing and antibiotic use on a macro- and micro-level, both for primary
care and secondary care. In addition, these data could also be used to identify
domains and questions (item generation) for an evaluative instrument to
measure change in patients attitude, knowledge and beliefs about common
infections and their management. At the present time, no validated, responsive
instrument is available for measuring the effect of interventions aimed at
changing peoples attitudes, beliefs and knowledge about common infections,
despite the fact that high profile public campaigns and other interventions are
being undertaken with this aim.
Observational studies could contribute to gaining insight into the incidence,
aetiology, optimal diagnostic strategies, individual risks for severe outcome
and infection with resistant micro-organisms.
Furthermore, intervention studies are needed to solve the ongoing debate on the
overall benefit from antibiotics, to identify subgroups who will (not ) benefit
from antibiotic, and to improve the management of RTIs, especially antibiotic
prescribing. Evidence on the effectiveness of symptomatic treatment of RTIs
could also contribute to improve antibiotic prescribing.
For primary care these clinical research activities requires operational networks
of GPs, which for Belgium need to be established and/or supported.
This dissertation is line with the empirical evidence base for changing
behaviour, suggesting that academic detailing (outreach visits to practices)
supplemented by other interventions is likely to be the most effective way to
implement guidelines to change doctors prescribing behaviour. More evidence
on the effectiveness of computerised decision support programmes in primary
care is needed, as is research to determine which individual part of complex
methods is most successful.

128

General discussion
Finally economic evaluations of diagnostics, therapeutics, outreach visits and
other interventions to improve antibiotic prescribing could be performed in
cost-benefit/cost effectiveness studies.

References
1. Coenen S, Van Royen P, Vermeire E, Hermann I, Denekens J. Antibiotics
for coughing in general practice: a qualitative decision analysis. Fam Pract
2000;17:380-5.
2. Coenen S, Michiels B, Van Royen P, Van der Auwera J-C, Denekens J.
Antibiotics for coughing in general practice: a questionnaire study to quantify
and condense the reasons for prescribing. BMC Fam Pract 2002;3:16.
3. Coenen S, Michiels B, Renard D, Denekens J, Van Royen P. Antibiotics for
coughing in general practice: GPs'perception of patients'request determines
prescribing. Submitted.
4. Coenen S, Van Royen P, K VP, Michels J, Dieleman P, Lemoyne S, et al.
Aanbeveling voor goede medische praktijkvoering: Acute hoest [Good Clinical
Practice Guideline: Acute Cough]. Huisarts Nu 2002;31:391-411.
5. Coenen S, Van Royen P, Michiels B, Denekens J. Promotion of rational
antibiotic use in Flemish general practice: implementation of a guideline for
acute cough [abstract]. Prim Care Respir J 2002;11:56.
6. Fahey T. Antibiotics for respiratory tract symptoms in general practice.
British Journal of General Practice 1998;48:1815-6.
7. Feinstein A. The '
Chagrin Factor'and Qualitative Decision Analysis. Arch
Intern Med 1985;145:1257-9.
8. Butler CC, Rollnick S, Pill R, Maggs-Rapport F, Stott N. Understanding the
culture of prescribing: qualitative study of general practitioners'and patients'
perceptions of antibiotics for sore throats. BMJ 1998;317:637-42.
9. Little P, Williamson I, Warner G, Gould C, Gantley M, Kinmonth AL. Open
randomised trial of prescribing strategies in managing sore throat [see
comments]. BMJ 1997;314:722-7.

129

Chapter VIII
10. Macfarlane J, Holmes W, Macfarlane R, Britten N. Influence of patients'
expectations on antibiotic management of acute lower respiratory tract illness
in general practice: questionnaire study. BMJ 1997;315:1211-4.
11. Cockburn J, Pit S. Prescribing behaviour in clinical practice: Patients'
expectations and doctors'perceptions of patients'expectations - a questionnaire
study. BMJ 1997;315:520-523.
12. Britten N, Ukoumunne O. The influence of patients'hopes of receiving a
prescription on doctors'perceptions and the decision to prescribe: a
questionnaire survey [see comments]. BMJ 1997;315:1506-10.
13. Okkes I, Oskam S, Lamberts H. Van klacht naar diagnose.
Episodegegevens uit de huisartspraktijk. Bussum: Coutinho, 1998.
14. De Maeseneer J. Huisartsgeneeskunde: een verkenning [General practice:
an exploration]. Proefschrift Rijksuniversiteit Gent, 1989.
15. Arroll B, Kenealy T. Antibiotics for acute bronchitis. BMJ 2001;322:939940.
16. Becker L, Glazier R, McIsaac W, Smucny J. Antibiotics for Acute
Bronchitis (Cochrane Review). In: Software U, editor. The Cochrane Library,
Issue 4. Oxford, 2002.
17. Gross P, Pujat D. Implementing practice guidelines for appropriate
antimicrobial usage: a systematic review. Med Care 2001;39:II55-69.
18. O'
Connor P, Amundson G, Christianson J. Performance Failure of an
Evidence-Based Upper Respiratory Infection Clinical Guideline. J Fam Pract
1999;48:690-7.
19. Schaffner W, Ray W, Federspiel C, Miller W. Improving antibiotic
prescribing in office practice: a controlled trial of three educational methods.
JAMA 1983;250:1728-1732.
20. Stewart J, Pilla J, Dunn L. Pilot study for appropriate anti-infective
community therapy. Canadian Family Physician 2000;46:851-859.
21. Perez-Cuevas R, Guiscafre H, Munoz O, Reyes H, Tome P, Libreros V, et
al. Improving physician prescribing patterns to treat rhinopharyngitis.
Intervention strategies in two health systems of Mexico. Soc Sci Med
1996:1185-1194.

130

General discussion
22. Zwar N, Wolk J, Gordon J, Sanson-Fisher R, Kehoe L. Influencing
antibiotic prescribing in general practice: a trial of prescriber feedback and
management guidelines. Family Practice 1999;16:495-500.
23. Gonzales R, Steiner J, Lum A, Barrett PJ. Decreasing antibiotic use in
ambulatory practice: impact of a multidimensional intervention on the
treatment of uncomplicated acute bronchitis in adults. JAMA 1999;281:15121519.
24. DeSantis G, Harvey K, Howard D. Improving the quality of antibiotic
prescription patterns in general practice. The role of educational intervention.
Med J Aust 1994;160:502-505.
25. Avorn J, Soumerai S. Improving drug-therapy decisions through
educational outreach. A randomized controlled trial of academically based
"detailing". N Engl J Med 1983;308:1457-1463.
26. Ilett K, Johnson S, Greenhill G, Mullen L, Brockis J, Golledge C, et al.
Modification of general practitioner prescribing of antibiotics by use of a
therapeutics adviser (academic detailer). Br J Clin Pharmacol 2000;49:168173.
27. Macfarlane J, Holmes W, Macfarlane R. Reducing reconsultations for acute
lower respiratory tract illness with an information leaflet: a randomized
controlled study of patients in primary care. British Journal of General
Practice 1997;47:719-22.
28. Hux JE, Melady MP, DeBoer D. Confidential prescriber feedback and
education to improve antibiotic use in primary care: a controlled trial. CMAJ
1999;161:388-392.
29. van Duijn H, Kuyvenhoven M, Welschen I, den Ouden H, Slootweg A,
Verheij T. Patients'and doctors'views on respiratory tract symptoms. Scand J
Prim Health Care 2002;20:201-2.

131

132

Antibiotics for coughing in general practice:


Summary
Introduction
The discovery of antibiotics (penicillin) by Alexander Fleming in 1928
triggered enormous progress in the field of medicine. However, as early as
1944 Fleming observed that some bacteria were able to destroy penicillin, and
he warned that the misuse of antibiotics could lead to selection of resistant
bacteria. This warning was lost in the first flush of the discovery of increasing
numbers of antibiotics and the success of these medicines.
At the beginning of the 21st century, infectious diseases again cause more
deaths as antibiotics lose their effectiveness. Over the last decade, the causative
agent of the most frequent life-threatening bacterial infections, Streptococcus
pneumoniae - or pneumococcus for short -, has become less sensitive to
penicillin and other antibiotics. More than 30% of pneumococci isolated in
Belgium are resistant to erytromycine and tetracycline, whereas more than 5%
show full penicillin resistance.
This increase in bacterial resistance is associated with the increased use of
antibiotics, both in animals and in humans. In the case of humans, 80% of
antibiotics are prescribed by primary healthcare providers, that is outside the
hospital, especially by general practitioners (GPs). The best way of preserving

133

the effectiveness of antibiotics is to use them more appropriately, i.e. in cases


where patients will actually benefit. The alternative of continuing to develop
new antibiotics will only solve the problem of antibiotic resistance if the
principles of judicious use of antibiotics are implemented at the same time. The
growth in resistance is progressing faster than the development of newer
antibiotics.
This dissertation aims to develop a management model enabling GPs to reduce
the prescription of antibiotics without harming their patients. At the same time,
our research addresses the very core of primary care, the screening function of
general practice, with the missing of diagnoses as Scylla, and the excessive
treatment of everyday complaints as Charybdis (Chapter I).

The filter function of general practice


General practice has an important role to play in the organization of quality
health care. Its main characteristic is to screen various health problems in, on
the one hand, self-limiting conditions or conditions to be dealt with in primary
care, and, on the other, disorders that require a more specialist approach. For
many, even well-known, complaints this function still needs an evidence base.

Missing diagnoses
The fact of working at low cost and with few technological means in order to
deal with a wide array of health problems is an inherent feature in general
practice. As a result, there is a limited diagnostic certainty. This also applies to
the most common complaint, i.e. coughing. Respiratory tract infections (RTIs)
are not easy to distinguish from other conditions such as asthma since patients
complain about coughing in both cases. Furthermore, there is a low degree of
certainty when differentiating between acute bronchitis and pneumonia, and
between viral and bacterial infections. A bacterial pneumonia, however, can be
a life-threatening condition, requiring antibiotics, and even admission into
hospital.
As a result, the quest for evidence enabling GPs to exclude life-threatening
conditions with more certainty should not centre on diagnoses that are difficult
to make, such as acute bronchitis, but on symptoms like coughing, for which
patients seek help.

134

Summary

Excessive treatment of everyday complaints


For most conditions for which antibiotics are being used, there is no scientific
evidence to support their actual benefit. For most patients, the use of an
antibiotic in case of an acute cough has no benefit when compared with a
placebo. Nevertheless, especially for this condition, (too) many and ever more
expensive antibiotics are being prescribed. Apart form the high financial cost
and the medicalising effect, this overuse results in an increase in anti-microbial
resistance to the antibiotics available.
This dissertation aims to contribute to the development of effective strategies
for a more appropriate use of antibiotics. Since coughing is one of the most
common complaints in general practice, the appropriate use of antibiotics to
treat coughs is a key area of action in order to tackle the resistance problem.
Consequently, by describing, exploring and optimising the prescription of
antibiotics for coughing we can safeguard a major development in the field of
medicine, i.e. the use of antibiotics in the treatment of life-threatening
infections. Examining how GPs can identify patients with coughing complaints
who will (not) benefit from antibiotics, is another relevant element in this
regard.

Part 1: Exploration and description


In the first part of the dissertation we explored the way GPs currently manage
patients who consult them with complaints about coughing.

Qualitative part
In a qualitative study (Chapter II), we explored the diagnostic and therapeutic
decisions by Flemish general practitioners regarding adult patients who consult
them complaining about a cough as well as the determinants of their decisions
by means of focus groups. Twenty-four GPs participated in four semistructured group discussions centred on the following questions:

135

1.You are consulted by one of your adult patients who complains about
coughing. Which diagnoses come to mind?
2. How do you differentiate between the various possibilities in your patient?
3. You suspect an infection of the respiratory tract. Do you differentiate in any
way? Which distinctions do you make?
4. How do you differentiate between the various possibilities in your patient?
The recordings of these focus groups were transcribed and subsequently
analysed in accordance with the principles of qualitative content analysis . All
texts were coded according to the research questions. Interpretation of the
coded texts allowed a classification of the codes and the establishment of
relationships between the various codes or categories.
In the focus groups, GPs stated that they try to differentiate between infectious
and non-infectious causes of coughing, and between various types of RTIs. The
most important decision, however, is whether or not to prescribe antibiotics for
patients in case of suspected RTI. In terms of the latter decision, we made a
distinction between two kinds of determinants after analysis of all codes.
Medical determinants, such as signs and symptoms, determine the probability
of disease but offer little diagnostic certainty for the patients involved. Nonmedical determinants, such as defensive medicine by the GP (doctor-related)
and patient expectations (patient-related), help determine the threshold at
which to prescribe an antibiotic as well.
Following the analysis of the different codes, hypotheses were set up regarding
decisions made by GPs when faced with complaints about coughing and the
determinants underlying their decisions:

The first diagnosis to present itself to a GP is RTI. This diagnosis is


reached independently of the patient. Other hypotheses emerge only if they
are considered plausible as a result of knowledge of patient history.

GPs ask routine questions to confirm only the most likely diagnoses.
Explicitly ruling out other diagnoses is less often used in decision-making.

136

Summary

In suspected RTI, GPs want to make a distinction between clinical


syndromes such as bronchitis and pneumonia, viral and bacterial RTI and
upper and lower RTI. This cannot be achieved with certainty on the basis of
medical history and clinical examination. Dealing with diagnostic
uncertainty, GPs decisions are directed at whether or not to prescribe
antibiotics.

For this (therapeutic) decision, doctor- and patient-related factors also come
into play. These factors give rise to a shift in the action thresholds in favour
of antibiotics, a phenomenon explained by the Chagrin factor . GPS regret
less having unnecessarily prescribed antibiotics than not having prescribed
any if afterwards it appears that they were necessary. In this context,
necessary does not only mean necessary for curing the patient, but also,
for instance, to meet patient expectations and thus retain patient loyalty.
The decision to prescribe antibiotics is better explained by both types of
determinants than by the conventional diagnostic groups of RTIs.

Quantitative part
A questionnaire (Chapter III) was used to quantify and condense the
determinants generated in the focus group study. More specifically, we
assessed the extent to which GPs consider those determinants when making
decisions in the case of suspected RTI in a coughing patient and how strongly
the determinants act in favour of or against antibiotic treatment.
Of the 316 Flemish GPs who were sent the questionnaire, 200 replied, with 188
responses being eligible for analysis (59.5 % of overall response rate). Our
sample, which included 65.9 % men and an average age of about 43, was
typical of the Flemish GP population.
GPs seem to consider all the determinants included in the questionnaire. They
nearly always consider '
lung auscultation'
, but also whether the patient has a
fever, is coughing up (coloured) sputum, looks ill and whether s/he has a
medical history of COPD or smoking. Moreover, they often watch out for
anything out of the ordinary . GPs pay less attention to '
non-medical reasons'
,
whether they be patient- or doctor related.

137

GPs felt that the deterioration of the general condition of the patient favoured
treatment with antibiotics most . There were no items that argued strongly in
favour or against treatment with antibiotics. Non-medical reasons support the
prescription of antibiotic treatment, albeit to a lesser extent than medical
factors.

Validation of qualitative and quantitative studies


In order to validate the focus group and questionnaire findings we recorded GP
management of acute cough (Chapter IV).
Of the 85 GPs willing to participate in our intervention study (see Part II:
Optimisation), 72 included an average of 10 consecutive adult patients who
consulted them with acute cough between February and April of 2000 and
2001. They recorded medical as well as non-medical data, including the
prescription of antibiotics and the GP'
s perception of requests for antibiotic
treatment on the part of the patients.
These data also revealed that non-medical determinants may have a
considerable effect on the decision whether or not to prescribe an antibiotic.
After all, the fact that patients requested antibiotics proved to be an equally
important, statistically significant, independent predictor of an antibiotic
prescription, as were medical determinants, such as for example the presence
of sputum. Good clinical practice guidelines and interventions to optimise the
prescription of antibiotics have to take into account non-medical reasons such
as the patient'
s request for antibiotics. In order to implement the recommended
management approach, this has to fit in with the described, i.e. current,
common practice.

Part 2. Optimisation
The second part of this dissertation provides recommendations for changing
current practices and specifically for optimising the use of antibiotics for
coughing in general practice. We conducted a prospective, cluster-randomised,
controlled, before-and-after study. The intervention was based upon a clinical
practice guideline.

138

Summary

Recommended management
A group of GPs drafted a guideline for the diagnostic and therapeutic
management of acute cough. This text was based upon the available evidence,
our own descriptive research, and on a consensus within the author group if
evidence was lacking. The text was peer-reviewed by a multidisciplinary panel
of experts and subsequently revised.
The revised guideline for acute cough includes the following key points:

The guideline applies to patients, aged 12 years or older, whose most


prominent complaint is acute cough with or without purulent sputum, not
patients with recurrent or chronic cough, chronic obstructive pulmonary
disease or patients that received antibiotic treatment in the preceding week.

First, pneumonia, pulmonary embolism, left ventricular failure (pulmonary


oedema), pneumothorax, aspiration and irritation by toxic agents should be
ruled out by history and clinical examination. Although these are not
frequent conditions and acute cough may not be the most prominent
complaint, these potentially life-threatening conditions are treatable. They
should not be missed.

If a cause other than a respiratory infection is present (for example asthma,


gastro-oesophageal reflux disease, ACE-inhibitors) management needs to
be adjusted accordingly. Even though such conditions may not be obvious
in a first encounter, they should not be ruled out.

If eventually a respiratory infection seems to be the most likely cause, it is


not feasible to distinguish between viral and bacterial infections.
Nevertheless, the decision whether to prescribe antibiotics has to be made.
Antibiotics are only needed for patients whose immunity has been
compromised.

Besides the scientific arguments, we also recommend integrating the GP s


own agenda as well as that of the patient in the final therapeutic decision.

An educational package was developed in accordance with the guideline. In


addition, this text was further elaborated according to a standardized

139

methodology defined by the Scientific College of Flemish General


Practitioners (WVVH) to become the guideline for good clinical practice: acute
cough (Chapter V).

Implementation
Participants in the questionnaire study were asked whether they were willing to
join an intervention study including pre- and post-assessment of the diagnostic
and therapeutic management of coughing (Chapter VI). The pre-test of the
planned intervention study consisted of the previously mentioned registration
of the management of acute cough in the period February-April 2000. Before
the intervention, all 85 GPs who agreed to participate were divided at random
into two study groups.
Our intervention was preceded by a nation-wide public awareness campaign,
"Antibiotics: Use them less often, but better." The campaign included TV and
radio announcements, booklets and leaflets raising public awareness about the
overconsumption and misuse of antibiotics, the resulting resistance problem
and the self-limiting character of the most frequent infections.
In January 2001, all GPs in the intervention group received the guideline by
mail and were contacted by a facilitator to arrange an outreach visit at their
practice. They were asked to read through the guideline in advance of that visit.
During the 10-20 minute visit, the educational package was presented. This
presentation was adjusted to the needs or observations expressed by the GP.
Once all GPs had been visited, each received a written reminder of the key
recommendations by post.
Immediately after the intervention period, GPs started the post-test. This
consisted of recording data regarding consecutive patients with acute cough in
the period February-April 2001. After the first consultation at the GP'
s practice,
the patients involved each day recorded the presence of coughing, sputum,
fever, sore throat, headache, muscle ache, runny nose, loss of appetite,
shortness of breath, thoracic pain, as well as information about their health
status and level of activity. When assessing the intervention, we took into
account the amount and cost of the antibiotics prescribed and the time it took
for patient symptoms to disappear (symptom resolution).

140

Summary

Results
Of the 42 GPs in the intervention group, 36 received the entire intervention.
Fifty-six GPs, 27 in the intervention group, 29 in the control group,
participated in both pre- and post-tests. They included 1503 patients eligible for
analysis. Patient diaries of 1009 patients eligible for analysis were available.
Taking into account clustering of patients (Generalised Estimating Equations
analysis), we arrived at the following findings:
Use of antibiotics. The antibiotic prescription rates for acute cough in the
intervention group and the control group were 157/365 (43.0%) and 168/445
(37.8%) in the pre-test and 80/292 (27,4%) and 115/401 (28.7%) in the posttest, respectively. When antibiotics were prescribed, these were macrolides,
cephalosporines or combinations of penicillins and beta-lactamase inhibitors,
i.e. the non-recommended antibiotics, in 94/157 (59.9%) and 105/168 (62.5%),
and 37/80 (46.3%) and 72/115 (62.6%), respectively. Without adjustment for
the other registered variables, these prescription rates did not differ between the
intervention and the control groups, neither in the pre-test nor in the post-test.
Although by March 2001 the public awareness campaign no longer had an
effect on the use of antibiotics, there was a significant difference between preand post-test prescriptions (P=0.005 and P=0.03) in both groups. In cases
where antibiotics were prescribed, only the GPs in the intervention group
prescribed 14% less of the non-recommended antibiotics (P=0.06 vs. P=0.84).
Unlike the pre- and post-test comparison (cf. the public awareness campaign),
the comparison of the intervention and control groups allows a convincing
adjustment for the substantial differences in the incidence of acute respiratory
infections between 2000 and 2001. If, in addition, we adjust for differences in
the other registered variables between patients in both groups and assume the
antibiotic prescription rates to be equal in both study groups in the pre-test
phase, only GPs in the intervention group prescribed significantly fewer
antibiotics in the post-test than those in the control group (ORadj = 0.55 (0.360.85)), and compared with their own prescriptions in the pre-test period (ORadj
= 0.55 (0.38-0.80)).
The intervention not only influenced the number of antibiotic prescriptions, but
also resulted in a better choice of antibiotics for acute cough patients. Since

141

there was a significant pre- and post-test difference only in the intervention
group, the public awareness campaign no longer appears to have had an effect
on post-test antibiotic prescription.
Cost of antibiotics. During the pre-test, the mean medication cost from the
point of view of the National Sickness and Invalidity Insurance Institute
(NSIII) was 12 in the intervention group and 11 in the control group. In the
post-test this dropped to 8 in the intervention group, and 9 in the control
group.
By limiting the analysis to the subset of patients who were prescribed an
antibiotic, the mean medication cost increased to 22 in the intervention group
and 21 in the control group. This dropped significantly in the intervention
group in the post-test, i.e. 16, in comparison with the control group s 21
(Mean Difference (MD)adj (95%CI) = -6.89 (-11.77 - (-2.02) ) and with the
pre-test figure of 22 (MDadj = -6.11 (-9.97 - (-2.24) )
Time to symptom resolution. As far as the use and type of antibiotics are
concerned, the same conclusions can be drawn from the subset of patients who
completed diaries. Since we are especially interested in public health, the
outcome for the patient is of paramount importance. This is why we
investigated whether fewer and different antibiotics influenced the time to
symptom resolution. However, we found no significant difference between the
intervention and control groups in terms of the time to resolution of all
symptoms.

Part 3. The patients


In the decision to prescribe antibiotics, patient-related determinants also play a
role. International differences in outpatient antibiotic consumption might
correlate with differences in patient attitudes. Moreover, the problem of
antibiotic overuse is an international problem, which requires international
interventions and provides opportunities for international research.
By means of an international postal questionnaire study, in collaboration with
Utrecht (the Netherlands), Cardiff (UK), and Barcelona (Spain), 400 patients in

142

Summary
each country were asked about their views on respiratory complaints and
antibiotic use (Chapter VII). Belgian patients reported a higher need for
consulting a general practitioner when faced with respiratory symptoms and
considered these disorders to be more serious and less self-limiting than did
their Dutch and UK counterparts. These results might partially explain the
differences in antibiotic use between respondents. Patient counselling should
specifically highlight the benign and self-limiting nature of the vast majority of
respiratory tract complaints.

Conclusion
This dissertation clearly shows that the prescription of antibiotics for the most
frequent (RTI) complaint in general practice, i.e. coughing, requires
optimisation (Chapter VIII). This can be achieved by means of a guideline for
good clinical practice, provided the guideline fits in with current common
practice and is implemented by means of academic detailing.
In doing so, we can achieve the goals of the public awareness campaign:
Antibiotics: Use them less often, but better . GPs not only prescribed less, but
also better because of our intervention. Furthermore, this did not happen at the
expense of patient recovery.
There is significant interaction between doctor and patient in the decision to
prescribe antibiotics. So, in order to develop effective strategies aimed at a
more appropriate use of antibiotics, we need to focus on both the doctor and
the patient and, in particular, on doctor-patient communication. More
specifically, the discussion of patient expectations about antibiotic prescription
and the (in)appropriateness of it, supported by relevant and evidence-based
recommendations is the key to success for the GP. Likewise, interventions
should focus on both the prescribers and the consumers. Therefore we advise a
combination of interventions such as a national public awareness campaign and
an intervention like ours, which directly addressed GPs.
Finally, interventions to change behaviour - in this case the prescription
behaviour - cannot claim to have a lasting effect. They need to be repeated to
preserve the effectiveness of antibiotics in future health care.

143

144

Antibiotica voor hoesten in de huisartspraktijk:


Samenvatting

Inleiding
De ontdekking van antibiotica (penicilline) door Alexander Fleming in 1928
betekende een enorme vooruitgang op het gebied van de geneeskunde. Maar
reeds in 1944 noteerde Fleming dat sommige bacterin in staat waren
penicilline te vernietigen, en hij waarschuwde dat misbruik van antibiotica kon
leiden tot selectie van resistente bacterin. Deze waarschuwing ging verloren in
de roes van de ontdekking van steeds maar nieuwe soorten antibiotica en het
succes van deze geneesmiddelen.
Aan het begin van de 21e eeuw zijn we echter zover dat infectieziekten
opnieuw een hogere tol aan mensenlevens eisen omdat antibiotica hun
doeltreffendheid verliezen. Zo is de verwekker van de meest frequente
levensbedreigende bacterile infecties, de Streptococcus pneumoniae of
kortweg de pneumococ, gedurende het laatste decennium steeds minder
gevoelig geworden voor penicilline en andere antibiotica. Meer dan 30% van
de in Belgi gesoleerde pneumococcen zijn resistent tegen erythromycine en
tetracycline, meer dan 5% vertoont volledige penicilline-resistentie.
Deze toename van de bacterile resistentie hangt samen met het toegenomen
antibioticumgebruik, zowel bij dieren als bij mensen. Bij mensen worden 80%

145

van de gebruikte antibiotica voorgeschreven in de ambulante praktijk, d.w.z.


buiten het ziekenhuis, voornamelijk door huisartsen. De beste optie om de
doeltreffendheid van antibiotica te bewaren, is ze doelmatiger te gebruiken,
d.w.z. daar waar ze patinten voordelen bieden. Het alternatief, om steeds
nieuwere antibiotica te ontwikkelen, zal het probleem van antibioticaresistentie
ook enkel oplossen, als tegelijkertijd gewerkt wordt aan een doelmatiger
gebruik. Resistentie ontwikkelt zich namelijk sneller dan de ontwikkeling van
nieuwere antibiotica.
Dit proefschrift wil een beleidsmodel opstellen dat de huisarts toelaat het
voorschrijven van antibiotica te reduceren zonder dat dit ten koste gaat van zijn
of haar patinten. Tegelijkertijd raakt het aan het wezen van de
huisartsgeneeskunde, de filterfunctie van de huisartsgeneeskunde, met het
missen van diagnoses als Scylla en het overbehandelen van alledaagse klachten
als Charybdis (Hoofdstuk I).

De filterfunctie van de huisartsgeneeskunde


De huisartsgeneeskunde speelt een belangrijke rol bij de organisatie van een
kwaliteitsvolle gezondheidszorg. Haar voornaamste eigenschap is het filteren
van allerlei gezondheidsklachten in zelflimiterende of in eigen beheer op te
lossen problemen en in aandoeningen die een meer gespecialiseerde aanpak
vereisen. Voor veel, zelfs goed gekende aandoeningen dient deze functie echter
nog met wetenschappelijke evidentie onderbouwd te worden.

Het missen van diagnoses


Het is eigen aan de huisartsgeneeskunde dat veel klachten worden uitgewerkt
aan lage kostprijs en met weinig technologie. De zekerheid waarmee diagnosen
worden gesteld is derhalve beperkt. Ook voor de meest voorkomende klacht:
hoesten, is dit het geval. Het is niet eenvoudig luchtweginfecties te
onderscheiden van andere aandoeningen die zich zoals bijvoorbeeld astma ook
presenteren met hoestklachten. Bovendien is er onvoldoende zekerheid om te
zeggen dat het om een acute bronchitis gaat en niet om een pneumonie, laat
staan dat een virus en niet een bacterie de oorzaak is van de klachten. Een
bacterile pneumonie is nochtans levensbedreigend en vereist vooralsnog een
behandeling met een antibioticum, eventueel zelfs hospitalisatie.

146

Samenvatting
Zoeken naar wetenschappelijke evidentie die huisartsen toelaat met meer
zekerheid een levensbedreigende aandoening uit te sluiten, dient dan ook niet
te vertrekken vanuit voor huisartsen moeilijk te stellen diagnoses, zoals
bijvoorbeeld acute bronchitis, maar vanuit de klachten waarmee deze zich
presenteren, zoals bijvoorbeeld hoestklachten.

Het overbehandelen van alledaagse klachten


Wetenschappelijke onderbouwing van de voordelen van een antibioticum
ontbreekt bij het merendeel van de aandoeningen waarvoor antibiotica worden
gebruikt. Wat hoestklachten betreft, is er evidentie dat antibiotica meestal geen
voordelen bieden t.o.v. placebo. Toch worden vooral voor deze klachten (te)
veel en steeds duurdere antibiotica voorgeschreven. En 80 % van die
voorschriften levert de huisarts af. Naast de enorme kostprijs hiervan en de
medicalisering van hoestklachten heeft dit overgebruik vooral een toename van
de bacterile resistentie voor de beschikbare antibiotica tot gevolg.

Dit proefschrift draagt bij tot de ontwikkeling van doeltreffende strategien die
een doelmatiger gebruik van antibiotica beogen. Aangezien hoesten tot de
frequentste klachten in de huisartspraktijk behoort, is het nastreven van een
doelmatiger gebruik van antibiotica bij hoestklachten een belangrijk
aangrijpingspunt om het resistentie probleem het hoofd te bieden. Aldus kan
een belangrijke vooruitgang op geneeskundig gebied, met name de behandeling
van levensbedreigende infecties met antibiotica gevrijwaard worden.
Onderzoeken hoe huisartsen patinten met hoestklachten kunnen identificeren
die (geen) baat hebben bij een antibioticum speelt daarin eveneens een
belangrijke rol.

147

Deel 1: Beschrijven en exploreren


In het eerste deel van dit proefschrift exploreerden we het huidige beleid van
huisartsen bij patinten die met hoestklachten consulteren.

Kwalitatief deel
In een kwalitatief onderzoek met focusgroepen (Hoofdstuk II) hebben we de
diagnostische en therapeutische beslissingen van huisartsen bij patinten met
hoestklachten gexpliciteerd en de determinanten van deze beslissingen
nagegaan. Vierentwintig huisartsen namen deel aan vier groepsdiscussies,
gestructureerd rond de volgende sleutelvragen:
1. Voor u staat een volwassen patint uit uw praktijk met als
contactreden/klacht hoesten . Welke diagnoses komen in jullie op?
2. Hoe maakt u het onderscheid tussen de verschillende mogelijke diagnoses?
3. U vermoedt dat uw patint een luchtweginfectie heeft. Maakt u daarin een
onderscheid? Welk onderscheid?
4. Hoe maakt u het onderscheid tussen de verschillende mogelijkheden bij uw
patint?
De bandopnames van deze focusgroepen werden uitgeschreven en
geanalyseerd volgens de methode van qualitative content analysis . Alle
tekstfragmenten werden gecodeerd in functie van de onderzoeksvragen. De
interpretatie van de gecodeerde tekstfragmenten liet toe de codes te
categoriseren en verbanden te leggen tussen verschillende codes of categorien.
De huisartsen verwoordden in de focusgroepen dat ze beslissen of
hoestklachten al dan niet een infectieuze oorzaak hebben en dat ze trachten
verschillende luchtweginfecties te onderscheiden. De belangrijkste beslissing is
echter het al dan niet voorschrijven van antibiotica bij de patinten waarbij ze
een luchtweginfectie vermoeden. Voor deze laatste beslissing onderscheidden
we na de analyse van alle codes twee soorten determinanten. Medische
determinanten, zoals klinische tekens en symptomen, bepalen de kans op
ziekte, maar bieden bij deze patinten weinig diagnostische zekerheid. Nietmedische determinanten zoals defensief handelen door de huisarts
(artsgebonden) en de verwachtingen van patinten (patintgebonden) bepalen
mee de drempel om een antibioticum voor te schrijven, vaak ten voordele van
antibiotica.

148

Samenvatting
Na de analyse van de verschillende codes werden hypothesen gegenereerd over
de beslissingen die huisartsen nemen bij patinten met hoestklachten en de
determinanten die deze beslissingen bepalen:
1. Huisartsen denken in eerste instantie en onafhankelijk van de patint aan
een luchtweginfectie. Andere hypothesen komen slechts aan bod als die
aannemelijk zijn vanuit de voorkennis over de patint.
2. Huisartsen stellen routinevragen enkel om de meest waarschijnlijke
diagnosen aan te tonen. Het expliciet uitsluiten van andere diagnosen wordt
minder vaak gehanteerd in het besliskundig proces.
3. Bij vermoeden van een luchtweginfectie willen huisartsen een onderscheid
maken tussen klinische entiteiten zoals bronchitis en pneumonie, virale en
bacterile luchtweginfecties en infecties van de bovenste en onderste
luchtwegen. Met argumenten uit de anamnese en het klinisch onderzoek
kan dat niet met zekerheid. Huisartsen dienen om te gaan met diagnostische
onzekerheid en hun beslissing spitst zich dan ook toe op het al dan niet
voorschrijven van antibiotica.
4. Bij deze (therapeutische) beslissing spelen arts- en patintgebonden
factoren ook een rol. Deze factoren bepalen een verschuiving van de
actiedrempels ten voordele van antibiotica. De Chagrin factor verklaart
dit fenomeen. Huisartsen ervaren minder spijt als ze onnodig antibiotica
hebben voorgeschreven, dan wanneer ze geen antibiotica hebben
voorgeschreven, terwijl nadien zou kunnen blijken dat het nodig was.
Nodig betekent hier niet alleen nodig om de patint te genezen, maar ook
om bijvoorbeeld geen patinten te verliezen aan niet-ingeloste
verwachtingen. De beslissing om antibiotica voor te schrijven wordt beter
verklaard door beide soorten determinanten dan door de conventionele
diagnostische groepen luchtweginfecties.

149

Kwantitatief deel
Met een enqute (Hoofdstuk III) hebben we vervolgens de bevindingen van
het focusgroepen onderzoek gekwantificeerd en met behulp van factoranalyse
gecondenseerd. We gingen meer bepaald na in welke mate huisartsen bij
patinten met hoestklachten en het vermoeden van een luchtweginfectie letten
op een selectie van de determinanten uit het focusgroepen onderzoek. Daar
naast werd aan huisartsen gevraagd hoe sterk deze determinanten volgens hen
pleiten voor of tegen het behandelen met een antibioticum. Van de 316
aangeschreven Vlaamse huisartsen hebben er 200 geantwoord, en van 188
(59.5 %) was de respons bruikbaar. Met bijna twee derde mannen (65.9 %) en
een gemiddelde leeftijd van bijna 43 jaar was onze steekproef vergelijkbaar
met de Vlaamse huisartsenpopulatie.
De huisartsen letten op alle determinanten gencludeerd in de vragenlijst. Bijna
altijd hielden ze rekening met het resultaat van de longauscultatie - maar ook
of er (gekleurd) sputum wordt opgehoest, of er koorts is, de patint bekend is
met COPD of rookt, er ziek uitziet of kortademig is. Ze letten vaak op
gegevens om het onderscheid te maken tussen pluis en niet-pluis situaties ,
beiden zijn medische determinanten, en in mindere mate op niet-medische
determinanten , hetzij arts- hetzij patintgebonden.
De achteruitgang van de algemene toestand van de patint pleitte volgens de
huisartsen sterk voor een behandeling met een antibioticum. Voor het overige
vonden ze dat niets sterk voor of tegen een behandeling met een antibioticum
pleitte. Zoals medische determinanten pleitten ook niet-medische
determinanten voor het behandelen met een antibioticum, maar in mindere
mate.

Validatie kwalitatief en kwantitatief onderzoek


Met een registratie van praktijkgegevens (Hoofdstuk IV) hebben we
vervolgens gevalideerd wat de huisartsen in het focusgroepen onderzoek
zegden en wat ze antwoordden in het vragenlijsten onderzoek. Van de 85
huisartsen bereid tot deelname aan het interventie onderzoek (zie Deel 2.
Optimaliseren) hebben 72 huisartsen gemiddeld 10 opeenvolgende volwassen
patinten gencludeerd die hen consulteerden met acute hoestklachten in de
periode februari-april 2000. Zij registreerden zowel medische als niet-medische
gegevens, waaronder respectievelijk het voorschrijven van antibiotica en de

150

Samenvatting
inschatting door de huisarts van de vraag van de patint om een behandeling
met een antibioticum.
Ook uit deze gegevens bleek dat niet-medische determinanten van belang
kunnen zijn bij de beslissing al dan niet een antibioticum voor te schrijven. Zo
was de vraag van de patint om antibiotica een even belangrijke, statistisch
significante en onafhankelijke voorspeller van een antibioticumvoorschrift, als
medische determinanten zoals bijvoorbeeld de aanwezigheid van sputum.
Aanbevelingen en interventies om het voorschrijven van antibiotica te
optimaliseren dienen ook rekening te houden met niet-medische factoren, zoals
de vraag van de patint. Om het aanbevolen beleid te implementeren sluit dit
immers best aan bij de beschreven, c.q. gangbare praktijk.

Deel 2. Optimaliseren
In een tweede deel van dit proefschrift wilden we het beschreven beleid
trachten te benvloeden, meer bepaald het gebruik van antibiotica bij de klacht
hoesten in de huisartspraktijk optimaliseren. Hier was het opzet een
gecontroleerd cluster gerandomiseerd interventie onderzoek met voor- en
nameting. De interventie was gebaseerd op een aanbeveling.

Het aanbevolen beleid


Voor de diagnostische en therapeutische aanpak van acute hoestklachten werd
een eerste aanbevelingstekst ontwikkeld door een auteursgroep van huisartsen.
Deze aanbevelingstekst was gebaseerd op de beschikbare onderzoeksliteratuur,
eigen onderzoek en op consensus binnen de auteursgroep indien
onderzoeksbewijs ontbrak. De aanbevelingstekst werd ter beoordeling
voorgelegd aan experten huisartsen, pneumologen en microbiologen en
vervolgens aangepast.

151

In de aangepaste aanbevelingstekst acute hoest werden volgende


sleutelboodschappen geformuleerd:
Deze aanbeveling betreft patinten van 12 jaar of ouder met als
voornaamste klacht acute hoest al dan niet met purulent sputum; patinten
met chronisch obstructief longlijder, patinten met recidiverende of
chronische hoestklachten of patinten die in de voorafgaande week met
antibiotica zijn behandeld, worden in deze aanbeveling buiten beschouwing
gelaten.
In een eerste stap dienen met anamnese en klinisch onderzoek behandelbare
aandoeningen, waarbij onmiddellijk levensgevaar bestaat, uitgesloten te
worden, ook al zijn ze weinig waarschijnlijk: longembolie, congestief
hartfalen (longoedeem), pneumothorax, aspiratie en pneumonie.
Is in een tweede stap een niet levensbedreigende en niet-infectieuze
oorzaak duidelijk, dient het beleid hieraan aangepast. Meestal echter zijn
deze diagnosen niet duidelijk bij een eerste contact. Ze dienen niet expliciet
aangetoond of uitgesloten in dat eerste consult.
Is tenslotte een luchtweginfectie de meest waarschijnlijke diagnose, dan is
het niet haalbaar virale van bacterile luchtweginfecties te onderscheiden.
Er dient wel beslist of antibiotica nodig zijn. Deze zijn enkel nodig bij
patinten met gecomprommiteerde immuniteit: bv. oncologische patinten,
patinten met diabetes mellitus
In de uiteindelijke therapeutische beslissing dienen bovendien de ideen
van de patint en die van de huisarts gentegreerd te worden.
Op basis van deze aanbevelingstekst werd dan een deskundigheidsbevorderingspakket (pakket DKB) ontwikkeld. Bovendien is deze aanbevelingstekst
volgens een door de Wetenschappelijke Vereniging van Vlaamse Huisartsen
(WVVH) vastgelegde procedure verder uitgewerkt tot de aanbeveling voor
goede medische praktijkvoering: Acute Hoest (Hoofdstuk V).

152

Samenvatting

De implementatie
De deelnemers aan het vragenlijsten onderzoek werd gevraagd of ze bereid
waren deel te nemen aan een interventie onderzoek met voor- en nameting
i.v.m. het diagnostisch en therapeutisch beleid bij hoestklachten (Hoofdstuk
VI). De voormeting van het geplande interventie onderzoek bestond uit het
hoger vermeld registratie onderzoek in de periode februari-april 2000. De 85
huisartsen die deelname toezegden werden vr de interventie at random
verdeeld in twee groepen.
Juist voor onze interventie, in december 2000, startte de federale overheid een
nationale mediacampagne Antibiotica, minder vaak en beter . Met spots op
radio en televisie, affiches en folders richtte deze campagne zich tot de
bevolking met informatie over het overmatig en ondoelmatig gebruik van
antibiotica, de gevolgen hiervan voor de ontwikkeling van resistentie, en het
zelf limiterend karakter van de meest frequente infecties.
Alle huisartsen in de interventiegroep ontvingen in januari 2001 de
aanbevelingstekst acute hoest per post en werden door een artsenbezoeker
gecontacteerd om een praktijkbezoek te plannen. De artsen werd gevraagd de
aanbevelingstekst vooraf door te nemen. Tijdens het bezoek presenteerde de
daartoe getrainde artsenbezoeker gedurende 10 tot 20 minuten het pakket
DKB, aangepast aan de noden en opmerkingen welke de arts te kennen gaf
(academic detailing). Wanneer alle huisartsen bezocht waren, ontvingen ze een
herinneringsbrief met de sleutelboodschappen per post.
Onmiddellijk na de interventieperiode startten de huisartsen de nameting. Deze
bestond uit het registreren van gegevens bij opeenvolgende patinten met acute
hoestklachten in de periode februari-april 2001. De gencludeerde patinten
registreerden elke dag vanaf het eerste consult bij de huisarts gegevens over
hoesten, slijmen, koorts, keelpijn, hoofdpijn, spierpijn, neusloop, verminderde
eetlust, kortademigheid, pijn op de borst, alsook over hun algemene
gezondheidstoestand en graad van activiteit. Voor het beoordelen van de
interventie hanteerden we het volume en de kost van de voorgeschreven
antibiotica, alsook de duur tot het verdwijnen van de symptomen bij patinten
(symptoomresolutie) als uitkomstmaten.

153

De resultaten
Van de 42 huisartsen in de interventiegroep ontvingen er 36 de volledige
interventie. 56 huisartsen, 27 in de interventie en 29 in de controle groep,
namen deel aan de voor- en de nameting. De huisartsen includeerden 1503
patinten geschikt voor analyse. Van 1009 patinten (67%) beschikken we over
hun registratieformulieren. Deze patinten verschilden niet van de andere
patinten, zonder patinten-registratieformulier, wat betreft de door de
huisartsen geregistreerde gegevens. Rekening houdend met de clustering van
de patinten (Generalised Estimating Equations analyse) kwamen we tot
volgende uitkomsten:
Volume antibioticavoorschriften. De proporties antibioticavoorschriften bij
acute hoest patinten in de interventie en controle groep tijdens de voor-,
respectievelijk tijdens de nameting zijn 157/365 (43.0%) en 168/445 (37.8%),
respectievelijk 80/292 (27,4%) en 115/401 (28.7%). Als er antibiotica werden
voorgeschreven, gingen het in respectievelijk 94 (59.9%), 105 (62.5%), 37
(46.3%) en 72 (62.6%) om macroliden, cephalosporinen of combinaties van
amoxicilline en clavulaanzuur, c.q. de niet aanbevolen antibiotica. Zonder
correctie voor de andere geregistreerde variabelen zijn er geen verschillen
tussen de interventie en de controle groep, noch tijdens de voor-, noch tijdens
de nameting.
Ondanks het feit dat ook de nationale mediacampagne in Belgi in maart 2001
geen effect meer had op het antibioticagebruik is er in beide groepen wel een
significant verschil tussen de voor- en de nameting wat het aantal voorschriften
betreft (P=0.005 vs. P=0.03). Als er antibiotica werden voorgeschreven,
schreven enkel de huisartsen van de interventie groep 14% minder de niet
aanbevolen antibiotica (P=0.06 vs. P=0.84).
In tegenstelling tot de vergelijking tussen voor- en nameting, cf. de nationale
campagne, laat de vergelijking met een controle groep toe overtuigend te
corrigeren voor de aanzienlijke verschillen in incidentie van acute
luchtweginfecties tussen 2000 en 2001. Corrigeren we bovendien voor
verschillen tussen de patinten in de respectievelijke groepen betreffende de
andere geregistreerde variabelen en gaan we uit van gelijke proporties
antibioticavoorschriften en gelijke antibiotica keuze tussen interventie en
controle groep tijdens de voormeting, dan schrijven enkel de huisartsen in de
interventiegroep significant minder antibiotica voor tijdens de nameting

154

Samenvatting
vergeleken met de controle groep (OR (95% BI): 0.55 (0.36-0.85)), en
vergeleken met hun voormeting (0.55 (0.38-0.80)). En als ze antibiotica
voorschrijven zijn dit ook meer de in de aanbeveling voorgestelde antibiotica
vergeleken met de controle groep (1.87 (0.933.76), en significant meer
vergeleken met hun voormeting (1.99 (1.16-3.42)).
De interventie benvloedde niet alleen het aantal antibioticum voorschriften,
maar resulteerde even goed in een betere keuze van het antibioticum voor acute
hoest patinten. Aangezien er enkel voor de interventie groep een significant
verschil is tussen voor- en nameting en niet in de controle groep, lijkt de
nationale campagne ook geen effect meer gehad te hebben op het voorschrijven
van antibiotica voor acute hoest tijdens de nameting.
Kost antibioticavoorschriften. Tijdens de voormeting bedroeg de gemiddelde
medicatiekost vanuit het perspectief van het Rijksinstituut voor Ziekte- en
Invaliditeitsverzekering (RIZIV) 12 in de interventie groep en 11 in de
controle groep. Tijdens de nameting was dit minder, 8 in de interventiegroep
en 9 in de controle groep. Beperken we de analyse tot de patinten die
antibiotica voorgeschreven kregen, dan bedroeg tijdens de voormeting de
gemiddelde medicatiekost 22 in de interventiegroep en 21 in de controle
groep. Tijdens de nameting was dit significant minder in de interventie groep,
16 , vergeleken met de controle groep ( 21) en met de voormeting (22).
Duur tot symptoomresolutie. Aangezien ons vooral de volksgezondheid
interesseert zijn de uitkomsten van de patinten ook een belangrijke, zoniet de
belangrijkste, uitkomstmaat. We gingen daarom na of minder en andere
antibioticavoorschriften bij acute hoest patinten de duur tot
symptoomresolutie benvloedden, maar vonden geen significant verschil tussen
controle- en interventie groep wat betreft de resolutie van alle symptomen
samen.

155

Deel 3. De patinten.
Bij de beslissing antibiotica voor te schrijven spelen ook patintgebonden
determinanten een rol. Zo houden internationale verschillen in antibioticagebruik mogelijk verband met verschillen in opvattingen van patinten.
Bovendien is het overgebruik van antibiotica een internationaal probleem, dat
internationale interventies noodzakelijk maakt en zich leent tot internationaal
onderzoek.
Met een internationale post-enqute, in samenwerking met Utrecht
(Nederland), Cardiff (Verenigd Koninkrijk) en Barcelona (Spanje), zijn per
land 400 patinten gevraagd naar hun opvattingen over luchtwegklachten en
antibioticagebruik (Hoofdstuk VII). Belgische patinten ervaren LWIs als
ernstiger, eerder als een reden om een arts te raadplegen en minder als
zelflimiterend dan Nederlandse patinten en patinten uit het Verenigd
Koninkrijk. Deze resultaten verklaren de verschillen in antibioticagebruik
tussen de respondenten ten dele. Voorlichting van patinten dient zich
voornamelijk te richten op het over het algemeen banale en zelflimiterende
karakter van luchtwegklachten.

Tot besluit
Dit proefschrift heeft aangetoond dat het voorschrijven van antibiotica bij de
meeste frequente klacht (voor luchtweginfecties) in de huisartspraktijk,
hoesten, geoptimaliseerd dient te worden (Hoofdstuk VIII). Dit kan op basis
van een aanbeveling voor goede medische praktijkvoering als deze aansluit bij
de gangbare praktijk en wordt gemplementeerd door middel van individuele
artsenbezoeken (academic detailing).
Doelstellingen zoals geformuleerd voor de nationale campagne Antibiotica,
minder vaak en beter werden op die manier bereikt. Huisartsen schreven niet
alleen minder, maar ook beter voor ten gevolge van onze interventie.
Bovendien ging dit niet ten koste van de genezing van de patinten.
Bij de beslissing al dan niet antibiotica voor te schrijven is er een belangrijke
interactie tussen arts en patint. Om doeltreffende strategien te ontwikkelen

156

Samenvatting
die een doelmatiger gebruik van antibiotica beogen, dienen we dus zowel
aandacht te hebben voor de arts als voor de patint, en in het bijzonder voor de
arts-patint communicatie. Meer bepaald het bespreken van de verwachtingen
van de patint inzake een antibioticum voorschrift en het (on)nut hiervan,
ondersteund door relevante en onderbouwde aanbevelingen, is voor de huisarts
de sleutel tot succes. Ook de interventies zelf richten zich wellicht bij voorkeur
op de voorschrijvers en op de gebruikers. Het lijkt ons daarom erg raadzaam
interventies zoals de nationale campagne naar het publiek en onze interventie
naar de huisartsen te combineren.
Tenslotte kunnen interventies om gedrag, hier het voorschrijfgedrag, te
veranderen geen blijvend effect claimen. Ze dienen dus ook regelmatig
herhaald te worden, om de doeltreffendheid van antibiotica voor de
toekomstige gezondheidszorg te behouden.

157

158

Curriculum Vitae
Samuel Jules Adeline Coenen was born in Antwerp, Belgium, May 21 1972. In
secondary school he studied Latin and Greek (St. Xaverius College 1990: cum
Laude). He graduated as a Candidate in Medical Sciences from the University
of Antwerp (RUCA 1993: Summa cum Laude; a 3-year study programme),
and as Medical Doctor (UIA 1997: Summa cum Laude; a 4-year study
programme). His Master s thesis was on the diagnostic value of history and
clinical examination for the diagnosis of respiratory tract infections. In stead of
continuing his professional training as a general practitioner, he choose to
focus on research in general practice for which he was granted a Fellowship as
Research-assistant of the Fund for Scientific Research-Flanders (1997-1999,
and renewed 1999-2002). His research focused on the exploration, description
and optimisation of antibiotic prescribing for acute cough in the context of
increasing antimicrobial resistance. In the mean time he supervised thesis
students and occasionally lectured at the University of Antwerp. In the
Doctoral Study Programme for Ph.D. students in Medical Science he made
himself familiar with subjects ranging from statistics over evidence based
medicine and (qualitative) research in general practice to the philosophy of
science, from scientific reporting in English and French over Powerpoint to
webauthoring, and, even more universalist, from the monetary union over the
financial world crisis to investments, emotional intelligence and the Socratic
conversation
To date, he (co-)authored over 20 contributions in
(inter)national peer-reviewed journals (a detailed list is included in this book)
and gave about 20 oral presentations at (inter)national conferences. He
received the Specia-prize for Excellence during Medical Studies (1997), the
ADVISA-prize for Young Researcher in General Practice (2000), the
Pharmacia Award for Flemish Research in General Practice (2001), and was
nominated for the Special Equip Quality Improvement Prize (2002) and the
Pharmacia Corporation-prize UA Antwerp (2002). He was granted funding by
the Small Project Fund of the Research Council of the University of Antwerp
(1998), the Scientific College for Flemish General Practitioners (WVVH)
(2000) and the Special Projects Fund of the European Society of General
Practice/Family Medicine (2001). He is currently working at the Department of
General Practice as a research assistant of the University of Antwerp, coauthoring the Five-Year Plan of the Belgian Antibiotic Policy Coordination
Committee (BAPCOC), member of the steering committee guideline
development of the WVVH, and member of the editorial staff of the journal of

159

the WVVH, Huisarts Nu. He occasionally reviews manuscripts for


international peer-reviewed journals and is (co-)author of WVVH guideline for
acute cough and the BAPCOC guideline for lower respiratory tract infections.
He is married to Sylvie Van Bylen and the father of Seppe (1998) and
Lieselotte (1999).

160

List of Publications
In national peer reviewed journals
- Coenen S. Antibiotica voor acute hoest bij volwassenen? Bespreking van
Fahey T et al. Quantitative systematic review of randomised controlled trials
comparing antibiotic with placebo for acute cough in adults. BMJ
1998;316:906-10, in Huisarts Nu (MINERVA) 1999;2:174-6 (Rechtzetting
Huisarts Nu (MINERVA) 1999;2:220).
- Coenen S. Steroden voor nachtelijke hoest bij kinderen. Bespreking van
Davies MJ et al. Persistent nocturnal cough: randomised controlled trial of
high dose inhaled corticosteroid. Arch Child Dis 1999;81:38-44, in Huisarts
Nu (MINERVA) 2000;8:376-7.
- Coenen S, Van Royen P, Vermeire E, Hermann I, Denekens J. Antibiotica bij
hoestklachten in de huisartsenpraktijk: een kwalitatief besliskundig
onderzoek.Huisarts nu 2001;30:390-7.
- Coenen S. Antibiotica voor acute hoest: tijd voor actie [Editoriaal]? Huisarts
Nu 2002;31:388-9.
- Coenen S, Van Royen P, Van Poeck K, Michels J, Dieleman P, Lemoyne S,
Denekens J. Aanbeveling voor goede medische praktijkvoering: Acute Hoest.
Huisarts Nu 2002;31:391-411.
- Coenen S, Michiels B, Van Royen P, Van der Auwera JC, Denekens J.
Antibiotica voor hoestklachten in de huisartspraktijk: determinanten van het
voorschrijfgedrag. Huisarts Nu 2003;32: 180-9
- Coenen S. Azithromycine en acute bronchitis. Bespreking van Evans A et al.
Azithromycin for acute bronchitis: a randomised, double-blind, controlled
trial. Lancet 2002;359:1648-54, in Minerva 2003;2:45-6.
In international peer reviewed journals

As first author
- Coenen S, Avonts D, Van Royen P, Denekens J. Chronic obstructive
pulmonary disease: don t forget the gatekeeper [Letter]. The Lancet
1998;352:649.
- Coenen S, Van Puymbroeck H, Debaene L, Denekens J, Van Royen P.
Irrational prescribing because of shifting therapeutic thresholds for sore
throats and for coughing [eLetter]. eBMJ.
161

- Coenen S, Van Royen P, Denekens J. Reducing antibiotics for respiratory


tract symptoms in primary care: '
why'only sore throat, '
how'about
coughing?[Letter]. Br J Gen Pract 1999;49:400-1.
- Coenen S, Van Royen P, Denekens J. Diagnosis of Acute Bronchitis [Letter].
J Fam Pract 1999;48:741-2.
- Coenen S, Van Royen P, Vermeire E, Hermann I, Denekens J.
Correspondence [Letter]. Fam Pract 2000;17:209.
- Coenen S, Van Royen P, Vermeire E, Hermann I, Denekens J. Antibiotics for
coughing in general practice: a qualitative decision analysis. Fam Pract
2000;17:380-5.
- Coenen S, Van Royen P, Avonts D, Denekens J. The GP forgotten again
[eLetter]. EBMJ.
- Coenen S, Kuyvenhoven MM, Butler CC, Van Royen P, Verheij TJM.
Variation in European antibiotic use [Letter]. Lancet 2001;358:1272.
- Coenen S, Michiels B, Van Royen P, Van der Auwera J-C, Denekens J.
Antibiotics for coughing in general practice: a questionnaire study to quantify
and condense the reasons for prescribing. BMC Family Practice 2002, 3:16.
(10 pages; URL: http://www.biomedcentral.com/1471-2296/3/16)
- Coenen S, Van Royen P, Michiels B, Denekens J. Promotion of rational
antibiotic use in Flemish general practice: implementation of a guideline for
acute cough [Abstract]. Prim Care Respir J 2002;11:56.
- Coenen S, Van Royen P, Michiels B, Van der Auwera J-C, Denekens J. A
mytical diagnosis or a practicable symptom? [eLetter]. eBMJ.

As co-author
- Vermeire E, Van Royen P, Coenen S, Denekens J. The compliance of type 2
diabetes patients [Letter]. Aust Fam Phys 1999;28:720.
- Remmen R, Van Royen P, Coenen S, Denekens J. Diagnosis and general
practice [Letter]. Br J Gen Pract 2001;51:232.
- B. Michiels, S. Coenen, D. Avonts, P. Van Royen, J. Denekens. Who benefits
from an influenza vaccination of GPs [Letter]? Vaccine 2001;20:1-2.
- Vermeire E, Van Royen P, Griffiths Fr, Coenen S, Peremans L, Hendrickx K.
The critical appraisal of focus group research articles. Eur J Gen Pract 2002;
8 104-108
- Vermeire E, Van Royen P, Coenen S, Wens J, Denekens J. The adherence of
type 2 diabetes patients to their therapeutic regimens: patients'perspective; a
qualitative study. Practical Diabetes International. Accepted for publication.

162

List of publications
Abstracts

National Conferences
- Coenen S, Van Royen P, Vermeire E, Hermann I, Denekens J. Met welke
argumenten voert de huisarts zijn diagnostisch beleid bij de contactreden
"hoesten". VHI 16de referatendag 1998, Gent.
- Vermeire E, Van Royen P, Coenen S, Denekens J. Opvattingen van
apothekers over compliance van hun clinten in het algemeen en type 2diabetespatinten in het bijzonder [Poster]. VHI 16de referatendag 1998,
Gent.
- Van Heerde M, Stuer H, Coenen S, Denekens J, Van Royen P. Risico s van
hypolipemirende behandelingen [Poster]. 16e VHI Referatendag 1998,
Gent.
- Vermeire E, Van Royen P, Coenen S, Patteet R, Denekens J. De compliance
van type 2 diabetes patinten met hun behandeling: focusgroepen onderzoek.
17e VHI Referatendag 1999, Antwerpen.
- Sepers G, Coenen S, Denekens J, Van Royen P. De niet-bacteriologische
determinanten voor het gebruik van antibiotica bij luchtweginfecties in de
huisartsgeneeskunde [Poster]. 17e VHI Referatendag 1999, Antwerpen.
- Coenen S. De implementatie van Evidence-Based Medicine in de dagelijkse
praktijk. ADVISA Workshop Evidence-Based Medicine 2000, Antwerp.
- Coenen S, Van Royen P, Van der Auwera J, Denekens J. Antibiotica voor
hoestklachten: hoe beslissen huisartsen? Eerste Eerstelijnssymposium 2000,
Brussel.
- Coenen S. Gecontroleerd interventieonderzoek naar het effect van een
deskundigheidsbevorderingspakket op het gebruik van antibiotica bij de
klacht hoesten in de huisartspraktijk. Belgian Drug Utilization Research
Group 2000, Brussel.
- Coenen S, Van Royen P, Michiels B, Van der Auwera JC, Denekens J.
Doelmatig antibioticagebruik in de huisartspraktijk: implementatie van een
aanbeveling voor acute hoest. Tweede Eerstelijnssymposium 2001, Leuven.
- Meewe M, Coenen S, De Backer W, Van Royen P, Denekens J. De
behandeling van acute hoest: Zijn betamimetica nuttiger dan antibiotica ter
behandeling van luchtweginfecties met acute hoest [Poster]. Tweede
Eerstelijnssymposium 2001, Leuven.
- Coenen S. Antibiotica voor hoesten: hoe beslissen huisartsen. Research Club
UZA 2001, Antwerpen.
- Coenen S, van Duijn HJ, Van Royen P, Kuyvenhoven MM, Tudor Jones R,
Butler CC. Opvattingen van patinten over luchtwegklachten en antibiotica:

163

een vergelijking tussen Belgi, Groot-Brittanni en Nederland. Derde


Eerstelijnssymposium 2002, Antwerpen.
- Coenen S, Michiels B, Renard D, Denekens J, Van Royen P. Het effect van
de inschatting door de huisarts van de vraag van de patint op het
voorschrijven van antibiotica voor acute hoest. Derde Eerstelijnssymposium
2002, Antwerpen.

International Conferences
- Coenen S, Van Royen P, Vermeire E, Hermann I and Denekens J. What
determines medical decision-making in patients with coughing as the reason
for encounter? Focus group research with general practitioners. First
European Network Organisations Open Conference-WONCA 1999, Palma
de Mallorca
- Vermeire E, Van Royen P, Coenen S, Denekens J. Compliance of type 2
diabetes patients with their therapeutic regimen [Poster]. First European
Network Organisations Open Conference-WONCA 1999, Palma de
Mallorca.
- Vermeire E, Van Royen P, Coenen S, Wens J, Denekens J. Compliance of
type 2 diabetes patients with their therapeutic regimen: Focus groups.
European General Practitioners Research Workshop 2000, Maastricht.
- Vermeire E, Van Royen P, Peremans L, Hendrickx K, Coenen S, Griffiths F.
Critical appraisal of focus group research articles [Poster]. Cochrane
Colloqium 2000, Cape Town.
- Coenen S. A Flemish Recommendation for the Management of Acute Cough
in General Practice. Second General Practice Respiratory Infection Network
Symposium 2000, Gent
- Van Royen P, Coenen S, Denekens J, Dieleman P, Michels J. From practice
guidelines to implementation of good clinical practice. A Flemish guideline
for acute cough and rational antibiotic use in general practice. WONCA
2001, Tampere.
- Coenen S. Antibiotica voor hoestklachten: hoe beslissen huisartsen [Poster]?
NHG-Wetenschapsdag, Amsterdam 2001, Nederland.
- Coenen S, Van Royen P, Van der Auwera JC, Denekens J. Promotion of
rational antibiotic use in Flemish general practice: implementation of a
guideline for acute cough. EURODURG 2001, Praag.
- Coenen S, Van Royen P, Michiels B, Van der Auwera JC, Denekens J.
Promotion of rational antibiotic use in Flemish general practice:
implementation of a guideline for acute cough. Third General Practice
Respiratory Infection Network Symposium 2001, Helsinki.

164

List of publications
- Coenen S. Promotion of rational antibiotic use in Flemish general practice:
implementation of a guideline for acute cough. European Conference on
Antibiotic Use in Europe 2001, Brussel.
- Coenen S. Indications for antibiotic treatment of lower respiratory tract
infections. World IPCRG Conference, Amsterdam 2002.
- Coenen S, Van Royen P, Michiels B, Van der Auwera JC, Denekens J.
Promotion of rational antibiotic use in Flemish general practice:
implementation of a guideline for acute cough. World IPCRG Conference
2002, Amsterdam (Abstract: Prim Care Respir J 2002;11(2):56).
- Coenen S, Van Royen P, Michiels B, Van der Auwera JC, Denekens J.
Promotion of rational antibiotic use in Flemish general practice:
implementation of a guideline for acute cough. WONCA Europe 2002,
London.
- Coenen S, Van Royen P, Michiels B, Van der Auwera JC, Denekens J.
Optimaliseren van het voorschrijven van antibiotica in de Vlaamse
huisartspraktijk: de implementatie van een aanbeveling voor acute hoest.
NHG-Wetenschapsdag 2002, Nijmegen.
- Coenen S, Michiels B, Renard D, Denekens J, Van Royen P. Antibiotics for
coughing in general practice: GPs perception of patients demand
determines prescribing. Fourth General Practice Respiratory Infections
Network Symposium 2002, Winchester.
- Coenen S, Michiels B, Denekens J, Van Royen P. Optimising antibiotic
prescribing for acute cough: a quality improvement report from Belgium.
EquiP 2002, Lissabon.
Masters Thesis

As first author
- Coenen S, Wens J, Denekens J, Van Royen P. Diagnostiek van
luchtweginfecties: een literatuurstudie naar de kracht van medisch
besliskundige argumenten binnen het bereik van de huisarts. 1997.

As supervisor
- De Wever V: Welke kans is er op de diagnose van psych(iatr)ische
aandoeningen bij patinten die zich in de huisartspraktijk presenteren met de
klacht hoesten, en met welke argumenten kan de huisarts deze diagnose
aantonen dan wel uitsluiten, 1998.
- Martens V: De plaats van hoestremmers bij luchtweginfecties, 1998.

165

- De Leeck A: Allergie voor huisdieren bij astmatische kinderen, 1999.


- De Smedt P: Electrocardiografie en risico op het ontwikkelen van uitgebreid
myocardinfarct en/of plotse dood, 1999.
- Devos M: Argumenten om bij de klacht hoesten antibiotica voor te schrijven,
1999.
- Sepers G: De niet-bacteriologische determinanten voor het gebruik van
antibiotica bij luchtweginfecties in de huisartsgeneeskunde, 1999.
- Van Den Branden A: Aantonen van een bovenste luchtweginfectie vanuit de
klacht hoest, 1999.
- Van Dessel E: Behandeling van Pelvic Inflammatory Disease, 2000.
- Moret J: Beleid bij prikaccidenten. Effecten en toxiciteit van
chemoprofylaxis, 2000.
- Palit Y: Alcohol en vaatlijden, 2000.
- Jacobs S: Depressie bij ouderlinge, 2000.
- Sahbaz H: De gezondheidstoestand van de bejaarde migranten, 2000.
- Kraak J: Acute bronchitis: het juiste antibioticum, 2001.
- Marichal P: COPD: anticholinergica en/of betamimetica, 2001.
- Meewe M: De behandeling van acute hoest, 2001.
- Janssen E: Langwerkende betamimetica bij chronisch astma, 2001.
Other publications
- Coenen S, Denekens J, Van Royen P. Patint en arts hebben baat bij minder
antibiotica. De Standaard 1999 19 februari:10.
- Coenen S, Michiels B, Denekens J, Van Royen P. Optimising Antibiotic
Prescribing for Acute Cough: a Quality Improvement Report from Belgium,
submitted for the Equip Quality Improvement Prize 2002.
- Coenen S. Antibiotica voor hoestklachten: beslissen in de
huisartsgeneeskunde. Onderzoeksverslag voor de Wetenschappelijke Prijs
Pharmacia Corporation 2002.
- Coenen S. Antibiotica voor hoestklachten in de huisartspraktijk: exploreren,
beschrijven en optimaliseren van het voorschrijven. Onderzoeksverslag voor
de Wetenschappelijk Prijs 2003 McKinsey&Co voor Doctoraatswerken
2003.

166

Dankwoord
Bedankt , dat is een dankwoord. Al wat daarna komt is knap lastig als je zo
dankbaar bent voor de zovele dingen die zoveel mensen in de voorbije jaren
voor je gedaan hebben en je al deze mensen daarvoor willen danken. Ik zal niet
nalaten te proberen iedereen te bedanken voor wat hij of zij betekent heeft voor
dit proefschrift, maar vooral wil ik het volgende zeggen. Het heeft me steeds
ontzettend deugd gedaan dat de hieronder vermelde mensen, maar ook zo vele
anderen, mij zo welgezind zijn geweest.
In min of meer chronologische volgorde bedankt ik mijn ouders, zij gaven me
alle kansen, ook die om te studeren en arts te worden; Johan Wens, omdat hij,
Annick en hun kinderen me hebben laten zien wat huisartsgeneeskunde in de
praktijk betekent; met hem allen die meewerkten aan het onderzoeksprotocol
voor de aanvraag van een Aspirant-mandaat van het Fonds voor
Wetenschappelijk Onderzoek Vlaanderen (FWO); Jan Heyrman, die als
referent optrad; Frank Buntinx, voor zijn bijdrage aan de herziening van het
protocol ter gelegenheid van de hernieuwing van mijn FWO-mandaat; het
FWO, omdat ze vertrouwen hadden in dit huisartsgeneeskunde onderzoek, de
Universiteit Antwerpen, om haar gastvrijheid en financile steun; Eric
Mathieu, die zijn hulp en een ontzettend boeiend programma aanbood in de
doctoraatsopleiding; alle medewerkers van de Vakgroep Huisartsgeneeskunde:
Lydie Van Laerhoven, Louise Gentils en Veerle Jordant, voor het
secretariaatswerk; Ingeborg Hermann, Ria Patteet en, in het bijzonder, Etienne
Vermeire, voor de assistentie bij de kwalitatieve start van dit project; Barbara
Michiels, voor haar aanzienlijk aandeel in het kwantitatieve vervolg, JeanClaude Vander Auwera, voor zijn statistisch advies en andere inzichten; Jo
Goedhuys, voor zijn feedback betreffende de factor analyse; Didier Renard en
Geert Molenbergs, die me toelieten uiteindelijk zelf de geclusterde data te
analyseren; Lieven Annemans, Bob Vander Stichele, Monique Elsevier en Jan
Van Campen, omdat ze me introduceerden in de gezondheidseconomie en het
onderzoek van geneesmiddelen gebruik; An de Sutter en Jan Matthys, voor hun
boeiend gezelschap tijdens de Opleiding Huisart-Onderzoek; Rogier
Hopstaken, Jean Muris en Willy Graffelman, omdat ze hun ideen en expertise
deelden; de Wetenschappelijke Vereniging voor Vlaamse Huisartsen (WVVH),
voor de financile steun bij de ontwikkeling van de WVVH-aanbeveling voor
acute hoest; auteurs en experts betrokken bij deze aanbeveling, o.a. Peter
Dieleman, Sabine Lemoyne, Jan Michels, respectievelijk Jef Boecks, Veerle

167

De Bock en Hugo Van Bever voor hun bijdrage; Isabelle Kloeck, Goedele
Truyen, Karel Van Poeck, Isabelle Janssens, Bart Blyweert en Guillaume van
Melckebeke en alle anderen die van ver of nabij hebben geholpen bij het
interventie onderzoek; ADVISA vzw in de persoon van Jean Colin, voor de
financile steun; Ineke Welschen, Marijke Kuyvenhoven, Huug van Duijn en
Theo Verheij, voor het plezier van samenwerking; Herman Goossens en Marc
De Meyere, voor de unieke kansen om mijn werk te presenteren; de ander
leden van het General Practice Respiratory Infection Network (GRIN), onder
andere Ian Williamson en Alastair Hay voor hun feedback; Jo Verhoeven en
Daniel Newman, die hebben geholpen met het Engels, en Veronique
Verhoeven, Luc Debaene en de andere medewerkers die voor een leuke
werksfeer zorgden.
Uiteraard dank ik ook al de huisartsen, en hun patinten, die bereid waren om
deel te nemen aan de verschillende onderzoeken, zonder hen was dit
proefschrift er gewoonweg niet gekomen; Marc Debroe, Wilfried De Backer en
Andr Meheus, voor hun vakkundige adviezen tijdens de jaarlijkse evaluatie
van dit doctoraat; Geert-Jan Dinant and Paul Little, for your very valuable
review of this dissertation.
Maar, ik ben vooral erg tevreden dat ik met velen onder jullie waarschijnlijk
nog vaak zal kunnen samenwerken. Daarvoor ben ik nu reeds naast de eerder
genomineerden Mark Haggard, Erwin Offeciers en Chris Butler dankbaar.
Ik hoop dat ik jullie allemaal als vrienden mag beschouwen, en dat doe ik zeker
voor twee bijzondere mensen: Joke Denekens en Paul Van Royen, wiens
bijdrage aan deze universiteit, zowel wat betreft de onderwijskundige
onderbouw als de wetenschappelijke invulling van de basisopleiding, zeker
voor mij niet ongemerkt is gebleven.
Joke bedankt voor het vertrouwen dat je me geeft.
Paul bedankt voor je betrokkenheid en immer accurate reflectie bij alle mijn
ondernemingen.
En tot slot - over mijn waardevolste onderneming Sylvie Van Bylen, Seppe
en Lieselotte, het doet me goed te weten dat jullie trots op me zijn. Dit werk
draag ik aan jullie op.

168