Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Review
a r t i c l e i n f o a b s t r a c t
Article history: Community-acquired pneumonia (CAP) is a leading cause of morbidity and mortality. This review
Received 5 March 2015 compared two of the main treatment alternatives: quinolone or macrolide monotherapy versus their
Accepted 22 April 2015 combination with -lactams. A systematic review and meta-analysis of randomised controlled trials
(RCTs) including adult inpatients and outpatients with CAP that compared treatment with any res-
Keywords: piratory uoroquinolone or macrolide administered as single agent with combination therapy of a
Community-acquired pneumonia
-lactam plus either a uoroquinolone or a macrolide (four separate comparisons) were conducted.
Macrolides
The primary outcome was all-cause 30-day mortality. Secondary outcomes included clinical and micro-
Quinolones
Systematic
biological failure, treatment discontinuation and adverse events. A comprehensive search was conducted
Review with no date, language or publication status restrictions. Pooled risk ratios (RRs) with 95% condence
Meta-analysis intervals are reported. Sixteen RCTs randomising 4809 patients were included. All but one included
hospitalised patients. Mortality was low, and no differences between groups were observed in all com-
parisons. Quinolone monotherapy resulted in signicantly less clinical failures [RR = 0.72 (0.570.91)],
treatment discontinuations [RR = 0.65 (0.540.78)] and diarrhoea [RR = 0.13 (0.050.34)] compared with
-lactam/macrolide combinations (nine trials). Addition of a -lactam to quinolones did not improve
outcomes (three trials). In all comparisons, treatment discontinuation and diarrhoea were more frequent
in patients receiving combination therapy with a -lactam. Overall, there is no evidence for a benet of
-lactam/macrolide or -lactam/quinolone combination therapies over monotherapy with a respiratory
uoroquinolone. The ecological implications of selecting uoroquinolone or -lactam monotherapy as
the preferred regimen for hospitalised CAP among adults should be further investigated.
2015 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved.
1. Introduction for CAP should target the most effective and efcient antibiotic
regimens.
Community-acquired pneumonia (CAP) remains a common The American Thoracic Society/Infectious Diseases Society of
cause of morbidity throughout the world despite the availability America (ATS/IDSA) guidelines for the management of CAP in
of potent new antimicrobials and effective vaccines. The estimated adults recommend primarily a macrolide for healthy outpatients
yearly incidence of CAP is 12 cases per 1000 population. In the [1]. A respiratory uoroquinolone or a -lactam combined with a
USA, CAP is the sixth leading cause of death and the number one macrolide are recommended as alternatives for outpatients with
cause of death from infectious diseases. CAP is a common reason co-morbidities and for inpatients. The combination of a -lactam
for antibiotic prescription in the community and in hospitals, affect- and a respiratory uoroquinolone is recommended when Pseu-
ing resistance ecology in both settings. Thus, antibiotic treatment domonas aeruginosa is a consideration [1]. The British Thoracic
Society (BTS) guidelines recommend amoxicillin or clarithromycin
for outpatients or inpatients with CURB-65 of 01, amoxicillin plus
clarithromycin or a respiratory uoroquinolone for hospitalised
Corresponding author. Tel.: +972 4 777 1758; fax: +972 4 777 1620.
patients with CURB-65 of 2, and for severe CAP, a penicillin plus
E-mail addresses: drdudidu@gmail.com, d shasha@rambam.health.gov.il
a macrolide, a penicillin plus a quinolone or a cephalosporin plus a
(D. Shasha).
1
These two authors contributed equally to this work. macrolide [2].
http://dx.doi.org/10.1016/j.ijantimicag.2015.04.010
0924-8579/ 2015 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved.
Please cite this article in press as: Raz-Pasteur A, et al. Fluoroquinolones or macrolides alone versus combined with -
lactams for adults with community-acquired pneumonia: Systematic review and meta-analysis. Int J Antimicrob Agents (2015),
http://dx.doi.org/10.1016/j.ijantimicag.2015.04.010
G Model
ANTAGE-4589; No. of Pages 7 ARTICLE IN PRESS
2 A. Raz-Pasteur et al. / International Journal of Antimicrobial Agents xxx (2015) xxxxxx
2. Methods
RCTs of adults aged >18 years with CAP, treated in the hospital
[intensive care unit (ICU) and non-ICU] or in the community were
included. CAP was dened as a new inltrate on chest radiography
or computed tomography (CT) and at least one positive nding out
of the following: new onset of a cough; purulent tracheobronchial
secretions; fever; and focal respiratory abnormalities on physical
examination [2].
RCTs that compared any respiratory uoroquinolone or any
macrolide administered as single agent with combination ther-
apy consisting of a respiratory uoroquinolone or macrolide plus
a -lactam were included. Analyses were stratied by compari-
son. Trials in which the uoroquinolone or the macrolide in the
combination arm was optional (permitted according to physicians
discretion) were excluded.
The primary outcome was all-cause mortality at 30 days. If Fig. 1. Flow diagram of the selection of studies for inclusion in the meta-analysis.
not reported, all-cause mortality at the end-of-study follow-up
was collected and the length of follow-up was documented. Sec-
ondary outcomes included: clinical failure at end of treatment; were contacted for missing data regarding the primary outcome
the need for treatment discontinuation; microbiological failure; and risk of bias.
mean time to defervescence; duration of hospital stay for hos- Risk ratios (RRs) were calculated for individual trials, with 95%
pitalised patients; resistance development expressed as rates of condence intervals (CIs), and a meta-analysis was performed
any clinical superinfections and superinfections caused specically using the xed-effect model. A RR > 1 favours combination ther-
by bacteria resistant to the study antibiotics; and AEs address- apy. The CIs for mortality in one cluster-randomised trial were
ing specically diarrhoea [including Clostridium difcile-associated adjusted to clustering using a reported design effect of 1.18 [5], as
diarrhoea (CDAD)], nausea/vomiting and dermatological reactions. recommended in The Cochrane Handbook [4]. Heterogeneity was
The study denitions for clinical and microbiological failure were examined visually and statistically using the 2 test (P < 0.1) and
accepted. Data were extracted preferentially by intention to treat. the I2 test (I2 > 50% dening signicant inconsistency). To address
A comprehensive search with no date, language or publica- heterogeneity, we dened a priori subgroup analyses of pneumo-
tion status restrictions was conducted of PubMed, LILACS and coccal CAP, severe CAP (treated in ICU or CURB-65 > 2) and patients
The Cochrane Library until December 2014 as well as conference with bacteraemia. However, data allowed only the rst. Analyses
proceedings of the European Congress of Clinical Microbiology were performed using RevMan 5.3 (The Nordic Cochrane Centre,
and Infectious Diseases (ECCMID) and Interscience Conference of Copenhagen, Denmark).
Antimicrobial Agents and Chemotherapy (ICAAC) 20092014. In
addition, previous systematic reviews of CAP, all references of 3. Results
included trials and the National Institutes of Health (NIH) trial reg-
istry were searched. The following search strategy was adapted The search resulted in 563 publications, of which 530 were non-
for each database: [(community AND pneumonia) OR community- randomised. Twelve RCTs were excluded because the macrolide
acquired pneumonia OR (Community-Acquired Infections [MeSH] was optional, and one was excluded because the -lactam agent
AND pneumonia [MeSH])] AND (quinolone OR uoroquinolone OR was optional. After further excluding duplicate publications (Fig. 1),
macrolide OR -lactam OR individual antibiotic names). In PubMed, 16 RCTs fullling the inclusion criteria were included [520] Two
the Cochrane sensitivity- and precision-maximising lter for RCTs were published only as conference proceedings and contact with
was used [4]. the authors could not be established [6,13].
The risk of bias was examined using the individual domain The trials included a total of 4809 patients with a mean or
approach, for all domains recommended in The Cochrane Hand- median age between 37 and 72 years (Table 1). A single trial
book [4]. Assessment was based on the methods described in the included outpatients that were recruited in the emergency room
publication, trial registry and correspondence with authors if nec- [7], and the remainder included inpatients. Hospitalised patients
essary. We planned to examine the effect of risk of bias on results had moderate-to-severe pneumonia; one study included only
through sensitivity analyses, but the paucity of trials under each patients with mild pneumonia [19], and in four studies pneumo-
comparison limited our ability to do this. Three reviewers indepen- nia severity was not described [6,13,15,20]. Low risk of bias was
dently applied inclusion/exclusion criteria and extracted all data described for allocation generation and concealment in four tri-
from all trials. Differences were resolved by discussion. Authors als [5,11,14,18], and four further trials only addressed allocation
Please cite this article in press as: Raz-Pasteur A, et al. Fluoroquinolones or macrolides alone versus combined with -
lactams for adults with community-acquired pneumonia: Systematic review and meta-analysis. Int J Antimicrob Agents (2015),
http://dx.doi.org/10.1016/j.ijantimicag.2015.04.010
ANTAGE-4589; No. of Pages 7
G Model
http://dx.doi.org/10.1016/j.ijantimicag.2015.04.010
lactams for adults with community-acquired pneumonia: Systematic review and meta-analysis. Int J Antimicrob Agents (2015),
Please cite this article in press as: Raz-Pasteur A, et al. Fluoroquinolones or macrolides alone versus combined with -
Table 1
Study characteristics.
Study Monotherapy Combination therapy Blinding Allocation No. randomised Age (years) (mean S.D.) Location Study year
generation/ [monother- [monother- end
concealmenta apy/combination] apy/combination]
Etemadi et al. (2011) [6] i.v. azithromycin i.v. ceftriaxone 2 g b.i.d. + p.o. Probably B/B 32/31 NS Iran NS
500 mg OD clarithromycin 500 mg b.i.d. open
Fogarty et al. (2004) [18] i.v. levooxacin i.v./i.m. ceftriaxone 12 g Open A/A 134/135 61.6 16.59/59.8 18.1 USA 2000
500 mg OD OD + i.v. erythromycin
5001000 mg QID
Frank et al. (2002) [7] i.v./p.o. i.v. ceftriaxone 1 g OD + i.v. Open A/B 115/121 67.8 13.11/67.3 13.17 USA 1999
ARTICLE IN PRESS
Leroy et al. (2005) [9] i.v. levooxacin i.v. cefotaxime 1 g t.i.d. + i.v. Open B/B 196/202 59.8 17.4/59.5 16.2 France, Tunisia, 2002
500 mg b.i.d. ooxacin 200 mg b.i.d. South Africa
Lin et al. (2007) [10] i.v. levooxacin i.v. amoxicillin/clavulanic acid Open A/B 26/24 65.3 13.2/71.0 11.4 Taiwan 2006
750 mg OD 500/1000 mg OD + p.o.
clarithromycin 500 mg b.i.d.
Portier et al., 1996 [12] p.o. sparoxacin p.o. amoxicillin 1 g DB B/B 110/103 60.0 2.0 France 1993
400 mg OD t.i.d. + ooxacin 200 mg b.i.d.
Portier et al. (2005) [11] p.o. moxioxacin p.o. amoxicillin/clavulanic acid Open A/A 174/175 59.3 17.9/62.4 18.0 France 2002
400 mg OD 1000/125 mg t.i.d. + p.o.
roxithromycin 150 mg b.i.d.
Postma et al. (2015) [5] i.v. moxioxacin i.v. penicillin, amoxicillin, Open A/A 888/739 Median (IQR): 71 The 2013
400 mg OD or amoxicillin/clavulanic acid or a (5979)/70 (5980) Netherlands
levooxacin third-generation
500 mg OD cephalosporin + p.o.
azithromycin, i.v. erythromycin
or p.o. clarithromycin
Ramirez and File (2003) [13] i.v. levooxacin i.v. ceftriaxone + p.o. Probably B/B 37/36 NS NS NS
clarithromycin DBb
Rovira et al. (1999) [19] p.o. clarithromycin p.o. cefuroxime 500 mg Open B/B 45/45 37.0 14.0/38.0 15.0 Spain 1997
500 mg OD b.i.d. + p.o. clarithromycin
500 mg OD
Torres et al. (2008) [14] i.v. moxioxacin i.v. ceftriaxone 2 g OD + i.v. DB A/A 271/367 66.0 16.2/64.8 16.7 Europe, Latin 2005
400 mg OD levooxacin 500 mg OD America, South
Africa
Vetter et al. (1997) [20] i.v. clarithromycin i.v. cefuroxime 1.5 g t.i.d. + i.v. Open B/B 118/117 60.0 2.0 Europe, Canada NS
500 mg OD erythromycin 1 g t.i.d.
Vergis et al. (2000) [15] i.v. azithromycin i.v. cefuroxime 750 mg Open A/B 83/86 NS USA 1996
500 mg OD t.i.d. + i.v./p.o. erythromycin
5001000 mg q.i.d.
Xu et al. (2006) [16] i.v. moxioxacin i.v. cefoperazone 2 g OD + p.o. Open B/B 20/20 NS China NS
400 mg OD azithromycin 500 mg OD
Zevros et al. (2004) [17] i.v. levooxacin i.v. ceftriaxone 1 g OD + p.o. Open B/B 107/112 72.84/70.73
USA, Europe, 2002
500 mg OD azithromycin 500 mg OD Canada
S.D., standard deviation; i.v., intravenous; OD, once daily; b.i.d., twice daily; p.o., oral; NS, not specied; t.i.d., three times daily; DB, double-blind; IQR, interquartile range; q.i.d., four times daily.
a
A = adequate; B = not specied.
b
Placebo mentioned with no further description of blinding methods.
3
G Model
ANTAGE-4589; No. of Pages 7 ARTICLE IN PRESS
4 A. Raz-Pasteur et al. / International Journal of Antimicrobial Agents xxx (2015) xxxxxx
Table 2
Meta-analysis results.
concealment [7,8,10,15]. Three were double-blind [1214]. One 3.2. Quinolone versus -lactam/quinolone
was a cluster-randomised, crossover trial comparing treatment
strategies assigned to hospitals in dened study periods as the unit No signicant differences were observed with regard to all out-
of randomisation [5]. Trial methods were not described in the other comes, based on a small number of patients. Two trials reported on
trials. mortality (RR = 1.00, 95% 0.691.45) [9,14], with an overall mortal-
The trials compared quinolone monotherapy versus -lactam/ ity in these studies of 8.7% (97/1116 patients), and all three reported
macrolide combinations (nine trials) [5,7,8,10,11,13,1618], on clinical failure (RR = 1.11, 95% CI 0.891.38; 1252 patients;
quinolone monotherapy versus -lactam/quinolone combina- Fig. 2) [9,12,14]. There was a signicantly higher risk of diarrhoea
tions (three trials) [9,12,14] and macrolide monotherapy versus with the addition of the -lactam in one trial (RR = 2.05, 95% CI
-lactam/macrolide combinations (four trials) [6,15,19,20]. The 1.133.73) [14], with no signicant difference in overall discontin-
specic antibiotic comparisons are detailed in Table 1. There were uations due to AEs (RR = 1.37, 95% CI 0.872.16). No discontinuation
no trials comparing macrolide monotherapy versus -lactam/ was reported.
quinolones. Antibiotic treatment was given for 714 days and
follow-up ranged between 3 and 6 weeks. The results are provided 3.3. Macrolide versus -lactam/macrolide
in Table 2 and are summarised below.
Mortality was low in these trials, one of which included out-
3.1. Quinolone versus -lactam/macrolide patients (overall 18/467 deaths; 3.9%) [19], and there was no
difference between arms with large CIs (RR = 1.00, 95% CI 0.402.46;
No difference in mortality was observed, without hetero- three trials) [6,15,20]. No signicant differences were observed
geneity and large CIs (RR = 0.99, 95% CI 0.701.40; ve trials) in clinical failure reported in all trials (Fig. 2) or other efcacy
[5,7,11,17,18], with an overall mortality in these trials of 5.1% outcomes. In one trial comparing clarithromycin with cefuroxime
(137/2683 patients). Clinical failure (RR = 0.72, 95% CI 0.570.91; and erythromycin [20], diarrhoea, nausea/vomiting, and discon-
nine trials, 2441 patients; Fig. 2) [5,7,8,10,11,13,1618] and treat- tinuations due to AEs were signicantly more common in the
ment discontinuation (RR = 0.65, 95% CI 0.540.78; six trials, 2179 combination arm.
patients) [5,7,8,10,11,18] were signicantly less common with In all trials, data on resistance development were not provided.
quinolone monotherapy, without heterogeneity. In the subgroup Hospitalisation duration was reported in a single trial of inpatients
of patients with pneumococcal pneumonia, the opposite direc- [17], with another two reporting on ICU stay or length of stay
tion was observed, with failure rates higher in the quinolone related to CAP [9,13]. The time to defervescence was not reported.
monotherapy arm without statistical signicance (RR = 2.03, 95% CDAD was reported in two trials (964 patients), with two events
CI 0.944.38; seven trials, 145 patients) [7,8,10,11,1618]. Treat- reported both in the -lactam arm [7,14].
ment discontinuations were in part related to AEs, as these were
less common with quinolones (mainly diarrhoea, RR = 0.13, 95% 4. Discussion
CI 0.050.34, reported in three trials using roxithromycin [11] or
azithromycin [7,8] as the macrolide). The microbiological failure In the current review, the question was addressed of whether
rate was similar. the use of a respiratory uoroquinolone alone or a macrolide
Please cite this article in press as: Raz-Pasteur A, et al. Fluoroquinolones or macrolides alone versus combined with -
lactams for adults with community-acquired pneumonia: Systematic review and meta-analysis. Int J Antimicrob Agents (2015),
http://dx.doi.org/10.1016/j.ijantimicag.2015.04.010
G Model
ANTAGE-4589; No. of Pages 7 ARTICLE IN PRESS
A. Raz-Pasteur et al. / International Journal of Antimicrobial Agents xxx (2015) xxxxxx 5
Fig. 2. Clinical failure for quinolone or macrolide monotherapy versus their combination with a -lactam (BL). A xed-effect MantelHaenszel (MH) meta-analysis is shown
with results presented as risk ratios with 95% condence intervals (CIs).
alone, both providing adequate antimicrobial coverage against similarly documenting an advantage to quinolones with regard
the pathogens causing CAP, are as effective as double-covering to clinical failure and gastrointestinal AEs, but no difference in
regimens consisting of a -lactam combined with each. Few mortality [22]. In a systematic review including both RCTs and
trials addressed each comparison (none comparing macrolide observational studies (analysis dominated by the observational
monotherapy versus -lactams/quinolones), and nearly all focused data), there was no difference in mortality between respira-
on hospitalised patients with CAP. Mortality was low in the tory uoroquinolone monotherapy versus -lactam/macrolide
trials and no differences were observed for all comparisons. combinations (RR = 1.17, 95% CI 0.911.50, with RR > 1 favour-
Quinolone monotherapy resulted in signicantly less treatment ing the quinolone arm; 16 studies, 12,624 patients) [23]. The
failure (RR = 0.72, 95% CI 0.570.91) than -lactam/macrolide com- results are similar to ours showing similar 30-day mortality with
bination regimens. The corresponding numbers needed to treat quinolones versus -lactam/macrolide combinations (RR = 0.99,
(NNT) with a quinolone to prevent one treatment failure was 28 95% CI 0.701.40; ve trials, 2683 patients) [5,7,11,17,18]. These
patients (95% CI 1889), with a control event rate (CER) of 12.4%. analyses point at a potential advantage of respiratory uoro-
Treatment failure was dened most commonly as a composite of quinolones with regard to tolerability and the pace of pneumonia
no cure at end of treatment, need for treatment modication due to resolution, but ultimately no difference in mortality.
clinical failure, or death related to infection. Quinolone monother- The main limitation of this analysis and previous analyses
apy also resulted in less need for any treatment discontinuation is the paucity of randomised data overall, considering that the
(RR = 0.65, 95% CI 0.540.78; NNT = 9, 95% CI 1320; CER = 21.8%) questions addressed are basic in the management of CAP. No
and less diarrhoea. Addition of a -lactam to either a quinolone or RCTs conducted among adults with CAP in the community were
a macrolide resulted in signicantly higher rates of diarrhoea with identied. The main comparisons of interest were addressed in
no difference in clinical outcomes, with large CIs due to the paucity three (quinolones versus -lactam/quinolones), four (macrolides
of trials. versus -lactam/macrolides) and nine (quinolones versus -
Several previous systematic reviews and meta-analyses com- lactam/macrolides) RCTs. The paucity of trials and patients did not
pared treatment regimens for CAP. Vardakas et al. investigated allow an analysis of subgroups of interest, mainly patients with
whether respiratory quinolone monotherapy was superior to severe pneumonia. Neither individual trials nor their compilation
-lactams, macrolides or -lactam/macrolide combinations, the have ever shown an effect of antibiotic choice on mortality in CAP.
latter pooled together (23 trials) [21]. Clinical success rates The outcome of clinical success or failure, for which there are differ-
were signicantly higher and AEs signicantly fewer with uoro- ences between antibiotic regimens, is ill dened in most trials. It is a
quinolone monotherapy, with no signicant difference in mortality. composite of different non-validated endpoints and its signicance
We previously compared macrolides versus quinolones, both for the individual patient is unclear. Trials do not report consistently
given either as monotherapy or in combination therapy regimens, on the time to clinical stability, duration of hospital stay, time to
Please cite this article in press as: Raz-Pasteur A, et al. Fluoroquinolones or macrolides alone versus combined with -
lactams for adults with community-acquired pneumonia: Systematic review and meta-analysis. Int J Antimicrob Agents (2015),
http://dx.doi.org/10.1016/j.ijantimicag.2015.04.010
G Model
ANTAGE-4589; No. of Pages 7 ARTICLE IN PRESS
6 A. Raz-Pasteur et al. / International Journal of Antimicrobial Agents xxx (2015) xxxxxx
Please cite this article in press as: Raz-Pasteur A, et al. Fluoroquinolones or macrolides alone versus combined with -
lactams for adults with community-acquired pneumonia: Systematic review and meta-analysis. Int J Antimicrob Agents (2015),
http://dx.doi.org/10.1016/j.ijantimicag.2015.04.010
G Model
ANTAGE-4589; No. of Pages 7 ARTICLE IN PRESS
A. Raz-Pasteur et al. / International Journal of Antimicrobial Agents xxx (2015) xxxxxx 7
[22] Skalsky K, Yahav D, Lador A, Eliakim-Raz N, Leibovici L, Paul M. Macrolides vs. [26] Donnelly JP, Baddley JW, Wang HE. Antibiotic utilization for acute respiratory
quinolones for community-acquired pneumonia: meta-analysis of randomized tract infections in U.S. emergency departments. Antimicrob Agents Chemother
controlled trials. Clin Microbiol Infect 2013;19:3708. 2014;58:14517.
[23] Asadi L, Sligl WI, Eurich DT, Colmers IN, Tjosvold L, Marrie TJ, et al. [27] Lafaurie M, Porcher R, Donay JL, Touratier S, Molina JM. Reduction
Macrolide-based regimens and mortality in hospitalized patients with of uoroquinolone use is associated with a decrease in methicillin-
community-acquired pneumonia: a systematic review and meta-analysis. Clin resistant Staphylococcus aureus and uoroquinolone-resistant Pseudomonas
Infect Dis 2012;55:37180. aeruginosa isolation rates: a 10 year study. J Antimicrob Chemother 2012;67:
[24] Garin N, Genne D, Carballo S, Chuard C, Eich G, Hugli O, et al. -Lactam 10105.
monotherapy vs -lactam-macrolide combination treatment in moderately [28] Patel SN, McGeer A, Melano R, Tyrrell GJ, Green K, Pillai DR, et al. Susceptibility
severe community-acquired pneumonia: a randomized noninferiority trial. of Streptococcus pneumoniae to uoroquinolones in Canada. Antimicrob Agents
JAMA Intern Med 2014;174:1894901. Chemother 2011;55:37038.
[25] Eliakim-Raz N, Robenshtok E, Shefet D, Gafter-Gvili A, Vidal L, Paul M, [29] Bennett J, Dolin R, Blaser M. Mandell, Douglas, and Bennetts princi-
et al. Empiric antibiotic coverage of atypical pathogens for community- ples and practice of infectious diseases. 8th ed. Philadelphia, PA: Elsevier;
acquired pneumonia in hospitalized adults. Cochrane Database Syst Rev 2015.
2012;9:CD004418.
Please cite this article in press as: Raz-Pasteur A, et al. Fluoroquinolones or macrolides alone versus combined with -
lactams for adults with community-acquired pneumonia: Systematic review and meta-analysis. Int J Antimicrob Agents (2015),
http://dx.doi.org/10.1016/j.ijantimicag.2015.04.010