Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
No additional articles relevant to the study major concerns and no or few minor con- methodological weaknesses for the 15
were found. In addition, the last 10 years of cerns that would seriously alter the results. strong or moderate quality studies.
Public Health and Epidemiology Report A designation of “moderate” indicated no
Ontario (PHERO) and Environmental major concerns, but at least one minor con- Study outcomes
Health Review were hand-searched since cern. A “weak” designation indicated one Inspection/Risk Assessment
these periodicals are not indexed on com- or more major concerns. One primary and Five studies8-12 evaluated the effective-
puter databases. one of four secondary reviewers assessed all ness of inspection of food service premises
Unpublished studies were identified studies independently. When the reviewers (primarily restaurants). Irwin et al. 10
using four strategies: (1) a computer search disagreed in their initial ratings, consensus observed that violation rates of inspection
of the Ontario Ministry of Health’s Public was reached on the final quality rating. criteria in restaurants were significantly
Health Research Database, (2) computer A data extraction tool was developed and associated with outbreaks, thereby validat-
search of student theses at Ryerson’s pretested to ensure that information was ing their use in inspections. Corber et al.9
School of Environmental Health, (3) direct extracted (from the strong or moderate found no difference in inspection scores
request of all Directors of Environmental studies) in a consistent manner by two between restaurants scheduled to receive 6,
Health in Ontario health units and all independent reviewers. Information was 9 or 12 inspections per year. Bader et al.8
provincial epidemiologists in Canada, and collected on study design, comparability of observed that premises inspected four
(4) key informants knowledgeable about the intervention and comparison groups, times per year produced better inspection
food safety research. description of the intervention, outcome scores than those inspected only upon pub-
measures and generalizability. Where dis- lic complaint (or once in two years).
Relevance and quality assessment crepancies occurred, discussion was held Kirshner11 found no significant difference
Studies were included in the review if and consensus reached. in inspection scores of restaurants receiving
they met all the following inclusion crite- 2, 4 or 6 inspections per year. Mathias et
ria: (1) the study described a food safety RESULTS al.12 determined that time intervals since
intervention or program, (2) the interven- the last inspection which were greater than
tion was within the scope of public health Study characteristics one year were significantly associated with
practice in Ontario, and (3) the study used A total of 168 potentially relevant stud- poorer inspection scores, however, time
outcome measures to assess the effective- ies were retrieved, which included 127 intervals of fewer than 3, 4-6, or 7-12
ness of the intervention. Two readers published and 41 unpublished studies. months since the last inspection were not
reviewed 60 articles and demonstrated a Application of the inclusion/exclusion cri- significantly different from each other.
high level of agreement regarding the deci- teria resulted in the retention of 34 rele- No studies were found that adequately
sion to include/exclude each study (kappa vant evaluation studies (23 published, 11 evaluated inspection programs based on
> 0.8). One reader screened the remaining unpublished) for quality assessment. Hazard Assessment Critical Control Point
108 articles for inclusion/exclusion. For the 34 evaluative studies assessed, 10 (HACCP) protocols, as implemented by
A quality assessment tool was developed, examined only inspections-based interven- staff of public health units. A few papers
pretested and then used to standardize the tions, 20 evaluated only food handler that described HACCP interventions,21-25
rating process among the five reviewers. training interventions, two evaluated both and two papers26,27 that discussed feasibility
The internal validity of each study was training and inspection, and two examined aspects, were retrieved.
assessed based on three major criteria: (1) community-based education interventions. Two studies examined risk assessment, a
comparability of the intervention and the Only 4 of the 34 studies assessed were component of an inspection strategy for
comparison groups [e.g., Was potential for randomized controlled trials (RCT). Six food premises which categorizes (after
bias due to selection, history, maturation, other studies were controlled trials (CT) inspection) each food premises according
resentful demoralization and compensatory without random allocation of study partic- to its potential for foodborne disease.
rivalry adequately addressed?], (2) assess- ipants. One study used a cohort design and Briley and Klaus6 observed that restaurants
ment of the interventions [e.g., Was the another was a case-control study. Thirteen designated as high risk (inspected every 1-2
intervention applied consistently and as studies were pretest/post-test studies (with- months) changed to a lower risk rating,
intended to everyone? Was the comparison out a control group). Seven studies were whereas medium and low risk premises
group blinded and protected from contami- cross-sectional, one was ecological and one (inspected approximately every 4 or 12
nation?], and (3) assessment of the out- was a time series design. months, respectively) did not. One impor-
comes [e.g., Was there potential for bias in The quality assessment revealed a limit- tant criteria for risk assessment in the
the outcome measurement due to testing, ed number of studies with good evidence Briley and Klaus study was previous
instrumentation, regression to the mean, or of effectiveness. Of the 34 evaluative stud- inspection score. In a Canadian study 7
validity and reliability concerns?]. Criteria ies assessed, 1 was rated as strong, 14 were with a different method of assessing risk
were rated as “no concerns”, “minor con- moderate, and 19 were weak. Table I pro- categories, about 10% of low risk food
cerns” and “major concerns”. Studies were vides a summary of the designs, partici- premises had changed to medium risk
designated as “strong” if there were no pants, interventions, outcomes and upon reassessment one year later.
TABLE I
Summary of Effectiveness Studies on Food Safety Interventions
Sandford & Amorim, Cross-sectional Inspection of previously designated “low 9% of previously designated “low Generalizability limit-
19967 (moderatea,b,c ) risk” food service premises to assess cur- risk” premises changed from low ed by non-random
(Canada) rent risk rating. to medium risk category. sample selection.
Inspection (only)
Bader et al., 19788 RCT Inspections 4 times per year or only after Inspection scores for premises “Problem” establish-
(U.S.A.) (moderatea,b,c ) public complaint inspected 4 times/year better ments excluded from
than those inspected only after study.
Intervenor: Public health inspectors complaint.
Setting: Food service establishments
(schools, restaurants & caterers) (n = 392)
Target: Food service establishment own-
ers & operators
Corber et al., 19849 RCT Inspections 6, 9 or 12 times per year No significant difference in Generalizability limit-
(Canada) (strongb ) inspection scores based on fre- ed by insufficient
Intervenor: Public health inspectors quency of inspections. information on char-
Setting: Food service establishments (n = acteristics of premises
392) included in study.
Target: Food premise supervisors & food
handlers
Irwin et al., 198910 Case control Comparison of violations of inspection Significant association between Provides evidence of
(U.S.A.) (moderatea,b,c ) criteria in restaurants with outbreaks to violations of many inspection cri- ability of inspection
those without. teria and outbreaks (OR 15.8, criteria to predict out-
95% CI 2.0 - 124.1). breaks, thus validat-
Intervenor: Public health inspectors ing their use in
Setting: Restaurants with outbreaks (n = 28); inspections.
matched controls without outbreaks (n = 56)
Target: Restaurant owners and operators
Mathias et al., 199512 Cross-sectional Time intervals since last inspection were Inspection scores significantly Good generalizability
(Canada) (moderatea,b ) 0-3, 4-6, 7-12 and > 12 months. Past poorer in restaurants last inspect- due to random selec-
food safety training. ed more than a year before. tion of restaurants
within 21 different
Intervenor: Public health inspectors Restaurants in which staff com- health units across 3
Setting: Non-franchise restaurants (n = 630) pleted food handler training provinces.
Target: Restaurant supervisors & food courses had better inspection
handlers scores than those without.
Training (only)
Cook & Casey, 197913 Non-random CT Food safety training: three 5-hour ses- No significant difference in Generalizability limit-
(U.S.A.) (moderatea,b,c ) sions; manual; 6 homework assignments; inspection scores of the interven- ed due to non-
examination. tion vs. control group. random sample selec-
tion and lack of infor-
Intervenor: Food service educator from mation on establish-
restaurant association ment characteristics.
Setting: Food service establishments (n =
20 with training; n = 35 control)
Target: Food premise managers
TABLE I — continued
Dalrymple, 199414 Pretest/post-test0 Food safety lecture1 Significant increase0 in knowl- Generalizability0 lim-
(Trinidad) (moderatea,b,c ) edge scores after food safety ited due to non-
Intervenor: Public health inspector training (p < 0.0001). random sample selec-
Setting: Education institution tion and lack of con-
Target: Food handlers (n = 75) trol group.
Kneller & Bierma, Time series Food safety training: 15 contact hours Inspection scores significantly Used multiple pre- (n
199015 (moderatea,b,c ) classroom training; examination. better (p < 0.01) after food safety = 483) and post-
(U.S.A.) training and certification. certification inspec-
Intervenor: Public health inspectors tion scores (n = 539)
Setting: Food service establishments (n = to compensate for
60) normal variation in
Target: Food service managers inspection scores.
Lawrie 198816 Cross-sectional Food safety lectures provided in 2nd Food safety knowledge scores Generalizability limit-
(Canada) (moderatea,b,c ) year of 4-year hospitality program. highest in 2nd and 3rd year, after ed due to cross-
training, compared with knowl- sectional design;
Intervenor: Instructors at educational edge in 1st year prior to food sample receiving
institution safety training. intervention different
Setting: Hospitality and Tourism Program from sample without
at post-secondary institution food safety training.
Target: Students (n = 155)
Rinke et al., 197517 Pretest/post-test Two food safety training methods: live Both groups significantly (p = Generalizability limit-
(U.S.A.) (moderatea,b,c ) instruction; taped instruction. 0.01) improved their knowledge ed by small sample
score over pretest. Groups did size and application
Intervenor: University instructors not differ on post-test. to single facility.
Setting: University residence
Target: University residence food han-
dlers (n = 31 “live”; 29 “taped” instruc-
tion)
Soneff et al., 199418 RCT Food handler training consisted of either A combined educational Limited generalizabil-
(Canada) (moderatea,c ) a workshop and manual, or manual approach (workshop presentation ity due to insufficient
only. and manual) demonstrated signif- information about
icant improvements in food safe- characteristics of
Intervenor: Dietitians (auditors), regis- ty knowledge scores. premises.
tered dietitian nutritionist (workshop pre-
senter) Negligible benefits of providing a
Setting: Adult care facilities manual alone.
Target: Staff of adult care facilities
Community-based
Education
Pivarnik et al., 199419 Pretest/post-test Food safety awareness training: four 60- Significant increase in knowl- Good generalizability
(U.S.A.) (moderatea,b,c ) minute lessons; interactive; audiovisual edge scores after intervention in that program deliv-
aids. (p < 0.05). ered at multiple
schools and in ethni-
Intervenor: Paraprofessionals from Food cally and racially
& Nutrition Program diverse areas.
Setting: Elementary schools (n = 11)
Target: Children (n = 561)
Radostits et al., 199320 Pretest/post-test Food awareness training: 2-hour guided Significant increase in knowl- Generalizability limit-
(Canada) (moderatea,b,c ) supermarket tour; in-store demonstra- edge after intervention. ed due to non-
tions. Improvements in self-reported random sample selec-
behaviour. tion and implementa-
Intervenor: Nutritionist & health inspec- tion at single loca-
tor tion.
Setting: Food store
Target: Public (shoppers; n = 57)
Note
Food Handler Training some cases, improving actual food han- in post-test measures (compared with
Evaluation studies of food handler train- dling operations on the food service pretest measures) after participation in a
ing provide evidence for the effectiveness premises (based on improved inspection training intervention. In a study by Soneff
of training programs in increasing knowl- scores). Of the eight evaluative studies et al.,18 improvements between pre- and
edge of proper food safety practices, and in assessed,11-18 six showed a positive change post-intervention inspection scores were
noted in the group that participated in a The inspection studies reviewed did not diverse strategies (e.g., school curricula
workshop, but not in the group that assess the effectiveness of HACCP inspec- development, workshops, dissemination of
received only a food safety manual. Two tions as conducted by public health staff in resources, peer education), and many dif-
studies11,13 did not show any improvements Ontario health units. It is possible that ferent ethnocultural groups (each with
in inspection scores after training. All eight HACCP inspections may be more effective their own food preparation practices).
evaluative studies were of moderate quali- than routine inspections since they specifi- Although the literature contains a growing
ty. cally address critical control points or risk body of studies that describe innovative,
factors epidemiologically implicated in community-based education strategies, the
Community-based Education foodborne illness.21,29-32 While the litera- effectiveness of most of these strategies has
Community-based education interven- ture supports at least one routine inspec- not been formally evaluated.
tions include promotion of public aware- tion per food service premises per year, this In summary, there is evidence for the
ness, public disclosure of violations, does not preclude additional inspections, effectiveness of multiple public health
awards, development of school curricula based on what is deemed necessary to interventions to ensure food safety. It is
and peer education. Of the 12 potentially bring a premises with poor safety condi- important to emphasize the need for ongo-
relevant studies retrieved, 3 were evalua- tions and practices into compliance with ing high-quality research to inform deci-
tion studies, of which 2 were assessed as acceptable food safety standards. sion making. Many of the studies identi-
moderate and 1 as weak. Based on studies30,31 that have linked the fied in this literature review process were of
One moderate study,20 which evaluated occurrence of foodborne illness with com- poor quality and therefore not useful in
the effectiveness of a supermarket food mon risk factors, there is evidence to sup- establishing practice guidelines.
safety educational tour, noted improve- port the use of risk assessment to classify
ments in food safety knowledge and self- food service premises into differential risk ACKNOWLEDGEMENTS
reported practices after the intervention. categories, based on: food property risks
The other moderate study,19 which evalu- (e.g., meat products more risky than We gratefully acknowledge funding sup-
ated a food safety curriculum for elemen- bread), population risk (e.g., type and port from the Ontario Ministry of Health.
tary students, demonstrated an increase in size), establishment history (e.g., past The authors also acknowledge the valuable
knowledge scores after the curriculum was infractions), and food service operational input from members of the Food Safety
introduced. risks (e.g., complexity of cooking, cooling, Literature Review Committee: Bob Hart,
handling, storage steps).6 There is some Chuck LeBer, Winston Miller, Dhan
DISCUSSION evidence to support that risk assessment Persaud, Pat Robinson, Fred Ruf and Janet
should be conducted annually for each Leung.
This systematic review provides evidence premises.7
that at least one routine inspection per This literature review provides evidence REFERENCES
food service premises per year is likely to for the effectiveness of food handler train- 1. Todd ECD. Preliminary estimates of costs of
reduce the risk of foodborne illness, as ing programs, in conjunction with certifi- foodborne disease in Canada and costs to reduce
determined through improved inspection cation, to improve the knowledge and salmonellosis. J Food Protection 1989;52:586-94.
2. Archer DL, Kvenberg JE. Incidence and cost of
scores. This result is consistent with the practices of food handlers. In establishing a foodborne diarrhea disease in the United States.
critical appraisal of the literature by Riben training program, consideration should be J Food Protection 1985;48(10):887-94.
3. Todd ECD. Foodborne and Waterborne Disease
et al. 28 While the individual studies on given to targeting both managers and food in Canada - Annual Summaries. Ottawa: Supply
which this evidence is based are of moder- handlers, and to making the educational and Services Canada, 1980-82; 1983-84; 1985-
ate or strong quality, there is concern experience active (such as with a work- 86; 1988-89; and 1990-91.
4. Nickol R. Food safety: Hazard analysis critical
about the generalizability of the findings in shop) rather than passive (such as with dis- control points (HACCP) - a system and protocol
these studies to the current situation in semination of resource materials only). for Ontario. PHERO 1991;2(16):246-48.
5. North York Public Health Department. Food
Ontario. The studies included in this Based on burden of illness data,1,33 com- Safety: Program Descriptions and a Systematic
review differed from each other in their munity-based food preparation (whether in Review of the Effectiveness of Documented
inspection protocols (different protocols the home or associated with community Interventions. Prepared for the Ontario Ministry
of Health. North York, 1997.
were likely used in different jurisdictions; events such as church gatherings) contin- 6. Briley RT, Klaus EF. Using risk assessment as a
protocols likely differed in the 1980s ver- ues to be an important setting for food- method of determining inspection frequencies.
Dairy and Food Sanitation 1985;5(12):468-74.
sus 1990s, and differed in Canada versus borne illness that adversely affects a large 7. Sandford J, Amorim E. Low Risk Premises
the United States). Furthermore, most number of people each year. The develop- Project. North York: City of North York Public
studies did not differentiate between high-, ment of community-based food safety Health Department, 1996.
8. Bader M, Blonder E, Henriksen J, et al. A study
medium- and low-risk premises. interventions is a complex undertaking of food service establishment sanitation inspec-
Consequently, it is not possible to make because it involves diverse target audiences tion frequency. Am J Public Health
1978;68(4):408-10.
evidence-based recommendations for the (e.g., homemakers, extended families, vol- 9. Corber S, Barton P, Nair RC, Dulberg C.
optimal inspection frequency for premises unteer cooks), many different settings (e.g., Evaluation of the effect of frequency of inspec-
tion on the sanitary conditions of eating estab-
in different risk categories. schools, community centres, churches),
lishments. Can J Public Health 1984;75(6):434- 18. Soneff R, McGeachy F, Davison K, et al. 27. Ying J. A feasibility study of the Ontario
38. Effectiveness of two training methods to improve Ministry’s HACCP protocol in high risk non-
10. Irwin K, Ballard J, Grendon J, et al. Results of the quality of foodservice in small facilities for institutional foodservice operations. Environ
routine restaurant inspections can predict out- adult care. J Am Diet Assoc 1994;94(8):869-73. Health Rev 1995;Spring:16-19.
breaks of foodborne illness: the Seattle-King 19. Pivarnik LF, Patnoad MS, Giddings M. A food 28. Riben PD, Mathias RG, Campbell E, Wiens M.
County experience. Am J Public Health safety curriculum for second- and third-grade ele- The evaluation of the effectiveness of routine
1989;79(5):586-90. mentary students. J Am Diet Assoc restaurant inspections and education of food han-
11. Kirshner B. An Effectiveness and Equivalence 1994;94(8):865-68. dlers: Critical appraisal of the literature. Can J
Trial Evaluating the Impact of Education and 20. Radostits J, Clemens R, McMillan D, et al. Public Health 1994;85(Suppl 1):S56-S60.
Frequency of Inspection to Regulation 243/84 Supermarket Smarts Program - An Evaluation 29. Bryan FL. Procedures for local health agencies to
under the Ontario Public Health Act [thesis]. Report. London: Middlesex-London Health institute a hazard analysis critical control point
Toronto, Ontario: University of Toronto, 1990. Unit, 1993. program for food safety in foodservice operations.
12. Mathias RG, Sizto R, Hazlewood A, Cocksedge 21. Bryan FL. Hazard analysis critical control point: J Environ Health 1985;47(5):241-45.
W. The effects of inspection frequency and food What the system is and what it is not. J Environ 30. Bryan FL. Impact of foodborne diseases and
handler education on restaurant inspection viola- Health 1988;50(7):400-1. methods of evaluating control programs.
tions. Can J Public Health 1995;86(1):46-50. 22. Dovell R. Evaluation of Hazard Analysis Critical J Environ Health 1978;40(6):315-23.
13. Cook CC, Casey R. Assessment of a foodservice Control Point Approach in Medium Risk Food 31. Bryan FL. Risks of practices, procedures and
management sanitation course. J Environ Health Premises [thesis]. Toronto, Ontario: Ryerson processes that lead to outbreaks of foodborne dis-
1979;March/April:281-84. Polytechnic University, 1991. eases. J Food Protection 1988;51:663-73.
14. Dalrymple RD. Evaluation of Food Handlers’ 23. Guzewich JJ. Statewide implementation of a 32. Bryan FL. Hazard analysis critical control point
Training Programme at St. George Central, HACCP food service regulatory program. (HACCP) systems for retail food and restaurant
Trinidad [thesis]. Toronto, Ontario: Ryerson J Environ Health 1986;49(3):148-52. operations. J Food Protection 1990;53:978-83.
Polytechnic University, 1994. 24. Mak A, Castillo M, Plunkie B. Implementation 33. LeBer C, Styliadias S, Brodsky M. Reported
15. Kneller P, Bierma T. Food service certification - of HACCP in a hospital’s cook-freeze foodservice Food-Borne Disease Outbreaks in Ontario from
measuring the effectiveness of a state program. system. Environ Health Rev 1991;Fall:60-65. 1982 to 1989. Toronto: Ontario Ministry of
J Environ Health 1990;52(5):292-94. 25. Rockbrune RA. Evaluation of Hazard Analysis Health, 1989.
16. Lawrie C. An Assessment of Food Handler Critical Control Point Approach in Medium
Education at the Post-secondary School Level Risk Food Premises [thesis]. Toronto, Ontario: Received: August 22, 1997
[thesis]. Toronto, Ontario: Ryerson Polytechnic Ryerson Polytechnic University, 1993. Accepted: November 20, 1997
University, 1988. 26. Stronach DA. Hazard analysis critical control
17. Rinke WJ, Brown NE, McKinley MM. Two point (HACCP) - an evaluation in Halton
methods for instructing personnel about foodser- Region. Environ Health Rev 1989;Summer:38-
vice sanitation. J Am Diet Assoc 1975;67:364-67. 41.
Available from
CPHA Health Resources Centre, 400-1565 Carling Avenue, Ottawa, ON, K1Z 8R1
613-725-3769 Fax: 613-725-9826 E-mail: hrc/cds@cpha.ca