Association of Food Premises Inspection and Disclosure Program with retail-acquired foodborne illness and operator noncompliance in Toronto.

Introduction

Enteric illness is a common problem around the world (Lee & Middleton, 2003). Enteric diseases affect the gastrointestinal tract, and many are attributable to microbial contamination (bacteria, viruses, or parasites) or chemical contamination of food or water. Diseases that are acquired by consumption of contaminated food are referred to as foodborne illnesses. The effects can range from a relatively mild (self-limiting) gastroenteric infection to severe conditions requiring immediate medical attention and, in some cases, even death. While accurate numbers concerning the extent of foodborne illness in Canada are not available, Health Canada estimates that about 2 million cases occur each year (Health Canada, 2000; Health Canada, 2001a). The medical costs and productivity losses from foodborne illness or food poisoning are estimated at over $1 billion per annum (Health Canada, 2001b). In Ontario, foodborne diseases are a significant cause of morbidity, with an estimated 280 outbreaks (approximately 70 per year) having affected 3,057 individuals (765 per year) in the period from 1993 to 1996 (Public Health Research Education and Development, 2001). Between 1991 and 2004, over 60,000 cases of enteric diseases, most of which were caused by unsafe food or water, were reported to Toronto Public Health (TPH). It is generally felt that these numbers represent only a fraction of the problem. Underreporting is considered to be a major complication in provision of accurate estimates (Olsen, MacKinon, Goulding, Bean, & Slutsker, 2000). In fact, health authorities believe that for every case reported, there are at least 100 unreported cases (Farber & Todd, 2000). Recent U.S. data based on active surveillance rather than passive reporting estimate the annual incidence of foodborne diseases at 76 million illnesses, including 325,000 hospitalizations and 5,000 deaths (Mead et al., 1999). It's important to note that the majority of the foodborne illnesses (with known causes) reported in Ontario in 1993-1996 were sporadic cases occurring as a result of unsafe food practices in the home. Cases due to outbreaks (defined as two or more people developing foodborne illnesses within a short period of time after consuming the same foods) were associated with retail food premises (Lee & Middleton, 2003). Crucial infractions, such as improper temperature control and contamination from poor personal hygiene of food handlers, are the main causes of foodborne illness in restaurants, at catered events, and in health care institutions (Allwood, Lee, & Borden-Glass, 1999).

Local public health units in Ontario are required to implement food safety programs and services aimed at reducing or eliminating the incidence of foodborne illness (Ontario Ministry of Health [OMH], 1997a; OMH, 1997b). To ensure that food safety standards are upheld in all food premises, certified public health inspectors in each health unit are authorized to conduct regular inspections and enforce the legislative requirements. Health units can also undertake additional measures to achieve the ultimate goal of increasing the health of the public. The Food Premises Inspection and Disclosure Program is the first of its kind in Canada and was implemented on January 8, 2001, with the intention of enhancing the effectiveness and efficiency of the food safety program in Toronto (TPH, 2002). While not a grading system, the new disclosure aspects of the program are intended to provide the general public with easy access to detailed inspection results from all applicable food premises so that individuals may make informed decisions about which food premises they choose to visit. The information is made available by way of mandatory on-site postings of inspection notices or signs (i.e., Pass, Conditional Pass, or Closed) and mandatory provision of the inspection reports issued at all food premises. In addition, a comprehensive two-year history of the inspection results, including the violations, for all food premises can also be accessed on an Internet site (www.toronto.ca/fooddisclosure) as well as via a food safety hotline (416-338-FOOD). Although the general inspection requirements did not change after implementation, the new program uses standardized policies and procedures for inspection and enforcement activities that are applied consistently across Toronto by highly trained public health inspectors and monitored by internal quality assurance mechanisms. The program also utilizes the Toronto Healthy Environments Information System (THEIS) to ensure that all necessary data components are captured electronically to assist management in program planning, activity tracking, internal/external reporting, and performance evaluation.

A variety of public disclosure systems are currently in place throughout North America, such as the posting of critical infractions on a Web site in New York City or the posting of letter grades in food premises in Los Angeles. No consensus exists, however, about the most effective methods of disclosing the information and the merits of such an intervention (Campbell et al., 1998; Fielding, Aguirre, & Palaiologos, 2001; Jones, Pavlin, LaFleur, Ingram, & Schaffner, 2004; Seiver & Hatfield, 2000; Simon et al., 2005; Thompson, de Burger, & Kadri, 2005; Wiant, 1999). An intervention of this type necessitates proper examination of its intended outcomes so that any health benefits (e.g., reduction of foodborne illness) can be assessed and promoted. A literature review revealed that there are very few published scientific evaluation studies of food safety inspection programs. While food premises are regularly inspected by health departments, few data are available on the effect that these inspections have on food safety (Jones et al., 2004). We undertook our study, therefore, to assess the association between the food safety intervention implemented by TPH and the local incidence of retail-acquired foodborne illness. The goal of the study was to conduct a systematic investigation of the merit, worth, or significance of the established program and determine whether it has made a difference and whether observed results are linked to specific program objectives. The primary objective was to assess the association of the Food Premises Inspection and Disclosure Program with the incidence of retail-acquired foodborne illness among Toronto residents post-implementation (i.e., 2001-2004) by comparison with the incidence pre-implementation (i.e., 1997-2000). A secondary objective was to assess the association of the program year with crucial inspection rates (i.e., event rates) among Toronto food premises operators post-implementation. The study also assessed whether crucial inspection rates correlate with retail-acquired foodborne-illness rates (i.e., episode rates) post-implementation.

Methods

Surveillance of foodborne illness in Toronto currently relies entirely on the mandatory reporting of disease episodes to TPH and subsequent entry into the province's Reportable Disease Information System (RDIS). We obtained episode data from this database in order to determine the association of the food safety program with the local incidence of foodborne illness. The term episode refers to a case of enteric illness (lab confirmed or not confirmed) that occurred in the period 1997 to 2004 and was reported to have been acquired from the ingestion of a pathogenic agent in food that was purchased in inspected food premises within the city of Toronto. Enteric illnesses reported through RDIS can be acquired through non-foodborne routes (e.g., contaminated water, person-to-person contact, and direct animal exposure). Therefore, episodes were included only if the client reported food as the "risk source" for the enteric illness. Toronto residents can acquire foodborne illness in a variety of public and private settings within the city as well as outside the jurisdictional boundaries of TPH. Therefore, the reported risk setting was restricted to food premises that are typically inspected by local health units (i.e., retail-acquired foodborne illness). This group included premises of the following types: "restaurant or food vendor," "daycare," "hospital," "correctional facility," "residential facility," "facility for the developmentally challenged," and "school." Episodes for which the risk source was given as "unknown" or for which the risk setting was something other that the settings listed were, however, also assessed individually and included on the basis of relevant information from other pertinent fields (e.g., "specific risk source" or "specific risk setting"). Total enteric disease was used as a denominator for calculation of standardized yearly and monthly episode rates. Food-poisoning episodes were analyzed separately from the general data.

The Food Premises Inspection and Disclosure Program relies on the Toronto Healthy Environments Information System (THEIS) to capture and maintain program information. THEIS is capable of performing many functions, including the documentation, reporting, and disclosure of food premises inspection activities. We accessed this database to obtain event data that would allow us to assess the effect of the food safety program on operator noncompliance. The term event is defined as a food premises compliance inspection conducted in the period 2001 to 2004 resulting in one or more crucial infractions. An infraction of this type has a direct impact on food and is most likely to result in a foodborne illness. Examples of crucial infractions include contamination or temperature abuse of food products as well as uninspected meats or ungraded eggs. Total inspection frequencies were used as denominators for calculation of standardized yearly and monthly event rates.

Statistical tests of the secondary data were conducted with SAS Version 9.1 software. The descriptive analyses for the study objectives include frequencies, means, proportions, test statistics, probabilities, relative risks, and confidence intervals. Pearson Chi-square tests were conducted to determine whether significant differences in retail foodborne illness rates occurred after the implementation of the new food safety program. We performed the Cochran-Armitage trend test to determine whether the proportional change in the event rate was statistically significant across the ordinal time intervals from 2001 to 2004. Finally, we measured rates of retail-acquired foodborne illness against the crucial noncompliance rates at monthly intervals post-implementation to determine the correlation between these two variables. Outliers were visually identified with a scatter plot and removed.

The study utilized secondary aggregate (group-level) data without the recruitment of individual subjects. While the issue of sample size did not apply directly, the methodology relied on the availability of sufficient data points for the primary variables of interest (i.e., episodes and events). In terms of ethical issues, the study adopted a principle-based approach that was in accordance with the Tri-Council Policy Statement on ethical conduct for research involving humans in Canada (Interagency Secretariat on Research Ethics, 2005). The study pursued scientifically valid and valuable research utilizing securely guarded secondary data in which results were reported at the aggregate level without the release of any confidential information.

Results

Proportions Test for Pre-program and Post-program Episode Rates (1997-2004)

The compiled RDIS data set covered the years 1997-2004 and comprised 3,141 episodes of the following foodborne diseases: amoebiasis, Campylobacter infection, cryptosporidiosis, cyclosporiasis, giardiasis, hepatitis A, listeriosis, paratyphoid fever, salmonellosis, shigellosis, typhoid fever, verotoxigenic E. coli (VTEC) infection, and yersiniosis. The most commonly reported disease was Campylobacter infection (38 percent), followed by salmonellosis (16 percent). The data also included episodes in the general category of food poisoning (32 percent), which consisted of reported enteric illnesses that had either not been confirmed or had resulted from agents that were not reportable. Retail-acquired foodborne illness accounted for 10.4 percent of all enteric illness reports (3,141 of 30,317). In terms of demographic information, 49.2 percent were males, 50.8 percent were females and the median age group was 30-34 years, which is representative of 2001 census data for Toronto. When age groups that are generally vulnerable to disease (i.e., young children & the elderly) are considered, the data show that 7 percent of all cases were younger than five years and 11 percent were older than 64 years. With respect to the distribution of the total episodes over time, the average number of reported cases from 1997 to 2000 was 328 per year (i.e., baseline data). From 2001 to 2004, the average number of episodes was 458 per year, which is an increase of 40 percent over the baseline data. When, however, the frequency of episodes of specific retail-acquired foodborne illnesses was considered, there was a decrease for many diseases in the post-implementation period. The proportions analysis showed a significant decrease in the episode rate for Campylobacter infection, VTEC, and hepatitis A. By contrast, the results showed a significant increase in the episode rate for amoebiasis and shigellosis (Table 1). There also was a significant increase, of over 400 percent, in the number of "food poisonings" that were reported after 2000.

Trend Test for Yearly Event Rates (2001-2004)

The THEIS data set comprised 7,373 events across Toronto from 2001 to 2004. Approximately 52 percent of all noncompliance inspections included one or more crucial infractions (i.e., events). The majority of events occurred in 2001, and the numbers then progressively decreased in each subsequent year. By comparison with the number of events in 2001, there was a 10 percent decrease in 2002, a 28 percent decrease in 2003, and a 48 percent decrease in 2004. When total yearly inspections are used as a denominator, the event rate for each consecutive year was 10 percent, 6.5 percent, 5.0 percent, and 3.3 percent, respectively The results of the Cochran-Armitage Trend Test show a highly significant decrease in the event rates over the four-year time period (Table 2).

Correlation Test for Monthly Event and Episode Rates (2001-2004)

A scatter plot of the data points shows a fairly linear distribution, with several outliers (Figure 1). The linearity of the data increases when the outliers are removed (Figure 2). Results of the Pearson correlation (r) test for the event and episode rates show a significant relationship between these two variables. The correlation coefficient increases from r = .51 [p = .00021 to r = .73 [p < .0001] when the identified outliers are removed.

[FIGURE 1 OMITTED]

Discussion

Our study was unique in its methodology because the set inclusion criteria for the RDIS episode data (i.e., risk source = food, risk setting = food premises) allows the direct measurement of foodborne illness attributable to retail food premises. While it has been postulated that the majority of foodborne illness is acquired in the home environment, little information has been available on quantifying the proportion of illness that may be attributed to food premises. Our study measured the incidence of retail foodborne illness at approximately 10 percent of all enteric illness. The demographics of the study sample showed that men and women were equally affected and that most episodes occurred among adults.

Although the total number of reported episodes, not including food poisonings, decreased in the post-intervention time period (2001-2004) as compared with the pre-intervention period (1997-2000), there was no significant difference in the standardized episode rate. There was, however, a highly significant decrease in the incidence of Campylobacter infections, VTEC infections, and hepatitis A in Toronto. This result is considerably beneficial with respect to frequency and severity of morbidity because Campylobacter infection is the most frequently occurring enteric disease in Toronto, and both VTEC and hepatitis A have very serious and long-lasting health impacts, especially among vulnerable population groups. Although there may have been decreases in the general, or standard, incidence of these diseases in other jurisdictions during the noted time period, the result is still meaningful since these episode rates are specific to food-premises risk settings, as opposed to all settings (i.e., different numerator), and also take into account the total number of enteric disease in each of the four years, as opposed to the total population at risk (i.e., different denominator).

The significant increase in the episode rate for shigellosis is a direct result of a provincewide outbreak in 2002 that was caused by contaminated food prepared outside of the Toronto area and sold in local supermarkets. Over 90 percent of post-intervention episodes caused by Shigella were a result of this outbreak. The significant increase in the episode rate for amoebiasis coincides with a general increase of such cases in the population during the post-intervention time period. Since transmission commonly occurs person-to-person via the fecal-oral route, the increase in episodes could be due to a misattribution of food as the source of infection or of food premises as the perceived risk setting in individual cases. There was a highly significant increase in the number of "food-poisoning" episodes reported post-intervention. This increase was probably due to an increase in food safety awareness among the general public in Toronto as a direct result of the disclosure program and related media coverage. Food-poisoning episodes are most often reported by the client and usually involve unconfirmed or no laboratory results. Although these self-reports are generally more open to biases, they can be used as a proxy indicator of disease incidence. Because the validity and reliability of the reports are questionable, however, the food-poisoning episodes were analyzed separately to prevent any potential confounding or effect modification of the general results.

With respect to the event data, descriptive analysis showed that over half of all noncompliance inspection results include infractions that are directly related to foodborne illness. There are socio-economic incentives for operators of food premises to comply with food safety requirements (Fielding et al., 2001; Fielding, Aguirre, Spear, & Frias, 1999). The existence of these incentives may be evidenced by the finding that there has been a highly significant decrease in the number of inspections resulting in crucial infractions since the implementation of the new food safety program in Toronto. While it can be argued that the increased number of inspections alone could account for the decrease in violations, this association would be valid only if the increased occurred within the same food premises. This was not the case, for the most part, as evidenced, for example, by the fact that the total number of premises varied each year because of new operations commencing and existing businesses shutting down. Ultimately, the important thing to note is that Toronto's ability to meet the minimum provincial inspection frequencies for all the different food premises in each risk category (low, medium, and high) increased by 17 percent in the first three years of implementing the new food safety program (Thompson et al., 2005). This result can likely be attributed to improved efficiency, effectiveness in program delivery, and the socio-behavioral impact(s) on operators as well as their patrons. The public disclosure of infractions may be perceived by some as having negative financial repercussions for individual food premises because patrons may not choose to visit such establishments until the operators have made the necessary corrections. Even after the operators have done so, the reputation of some food premises may be damaged as a result of noncompliance with food safety requirements. Thus, one of the benefits of a food safety disclosure program is that it encourages operators to properly maintain their premises and undertake the necessary food-handling procedures. This effect ultimately helps eliminate the occurrence of a health hazard and reduce the incidence of retail-acquired foodborne illness in the population. To this effect, the supporting data analysis showed a significant correlation between a decrease in crucial infractions and a decrease in the incidence of retail foodborne illness in Toronto.

The data sets prepared in our study were carefully selected and reviewed to ensure good analytical quality The RDIS database has specific limitations, however, which need to be taken into account when the results are interpreted. One of the biggest challenges is that the RDIS database is focused on laboratory-confirmed illnesses, and most foodborne illnesses are neither laboratory confirmed nor reported to health departments. Therefore, the true incidence of foodborne disease in any given year is difficult to quantify. The RDIS database is also deficient to a certain extent because of missing information in required fields (e.g., risk setting or source of infection). This problem may be due to the retrospective nature of the collection process, or there may be issues pertaining to proper form completion principles and data entry practices. This concern is important because, depending on the extent of missing information for a particular data field, the result could be summaries that may not be representative of the incidence of retail foodborne illness in Toronto. Another limitation is that RDIS data reflect only city of residence for each case at the time of diagnosis, which is not necessarily where the disease was actually acquired. In effect, a client may live in Toronto, but may have been exposed to contaminated food from premises in another health unit and vice-versa. Finally, the retrospective nature of the data collection process in foodborne-illness investigations means that these investigations rely heavily on the memory of the person who got sick to recall the correct information. Depending on the situation and the time lapse, some clients may be more apt to recall certain details of information (e.g., risk source or setting) than are others.

[FIGURE 2 OMITTED]

Conclusion

The public health and socioeconomic implications of retail-acquired foodborne illness are ongoing challenges and concerns across health agencies at all levels of government. The implementation of an enhanced food safety program in local health jurisdictions may be an effective way of decreasing retail-acquired foodborne illness in each area. The results of our study show that the Food Premises Inspection and Disclosure Program in Toronto was associated with a highly significant decrease in the incidence of certain key foodborne diseases that can be attributed to retail food premises. A highly significant decrease in operator noncompliance with the requirements that are most closely related to food safety (i.e., crucial infractions) was also demonstrated over the four-year period since the implementation of the new food safety program. In addition, the rate of inspections with crucial infractions was found to be positively correlated with the incidence rate of retail foodborne illness. Further research, such as an ecological study, should be conducted to investigate the impact of the disclosure program in Toronto by comparison with conditions in other, similar health units in Ontario that have not implemented any such program.

Acknowledgements: The authors wish to thank Olayemi Kadri, epidemiologist with Toronto Public Health, for his contributions to the completion of this study.

Corresponding Author: Tino Serapiglia, Environmental Health Officer, Professional Practice Leader, Toronto Public Health, 44 Victoria Street, 18th Floor, Toronto, Ontario M5C 1Y2. E-mail: [e-mail].

REFERENCES

Allwood, P.B., Lee, P., & Borden-Glass, P. (1999). The vital role of restaurant health inspections. Journal of Environmental Health, 61(9), 25-29.

Campbell, M., Gardener, C., Dwyer, J., Isaacs, S., Kruger, P., & Ying, J. (1998). Effectiveness of public health intervention in food safety: A systematic review. Canadian Journal of Public Health, 89(3), 197-202.

Farber, J.M., & Todd, E.C.D. (2000). Safe handling of foods. New York: Marcel Dekker, Inc.

Fielding, J.E., Aguirre, A., & Palaiologos, E. (2001). Effectiveness of altered incentives in a food safety inspection program. American Journal of Preventive Medicine, 32(3), 239-244.

Fielding, J.E., Aguirre, A., Spear, M.C., & Frias, L.E. (1999). Making the grade: Changing the incentives in retail food establishment inspection. American Journal of Preventive Medicine, 17(3), 243-247.

Health Canada. (2000). Policy development for raw foods of animal origin. Retrieved March 16, 2007, from http://www.hc-sc.gc.ca/fn-an/legislation/pol/rfao-aoca_e.html.

Health Canada. (2001a). Assessment policy: Health Canada policy: Food safety assessment program: Preamble. Retrieved March 16, 2007, from http://www.hc-sc.gc.ca/fn-an/securit/eval/pol/index_e.html.

Health Canada. (2001b). Policy development framework for the management of risks associated with the consumption of raw food of animal origin (RFAO). Retrieved March 16, 2007, from http://www.hc-sc.gc.ca/fn-an/legislation/pol/policy_development_frame-work-cadre_d_elaboration_politiques_e.html.

Interagency Secretariat on Research Ethics. (2005). Tri-Council Policy Statement: Ethical conduct for research involving humans. Ottawa, Ontario: Public Works and Government Canada.

Jones, T.F., Pavlin, B.I., LaFleur, B.J., Ingram, L.A., & Schaffner, W. (2004). Restaurant inspection scores and foodborne disease. Emerging Infectious Diseases, 10(4), 688-692.

Lee, M.B., & Middleton, D. (2003). Enteric illness in Ontario, Canada, from 1997 to 2001. Journal of Food Protection, 66(6), 953-961.

Mead, R.S., Slutsker, L., Dietz, V., McCaig, L.F., Breesee, J.S., Shapiro, C., Tauxe, R.V, & Griffin, R.M. (1999). Food-related illness and death in the United States. Emerging Infectious Diseases, 5(5), 607-625.

Olsen, S.J., MacKinon, L.C., Goulding, J.S., Bean, N.H., & Slutsker, L. (2000). Surveillance for foodborne disease outbreaks--United States, 1993-1997. Morbidity and Mortality Weekly Report, 49(SS-1), 1-72.

Ontario Ministry of Health. (1997a). Mandatory health programs and services guidelines for local boards of health. Toronto, Ontario: Author.

Ontario Ministry of Health. (1997b). Food safety: Program descriptions and a systematic review of the effectiveness of documented interventions. Toronto, Ontario: Author.

Public Health Research Education and Development Program. (2001). Effective public health practice project--The effectiveness of food safety interventions. Toronto, Ontario: Author.

Seiver, O.H., & Hatfield, T.H. (2000). Grading systems for retail food facilities: A risk-based analysis. Journal of Environmental Health, 63(3), 22-33.

Simon, P.A., Leslie, P., Run, G., Jin, G.Z., Reporter, R., Aguirre, A., & Fielding, J.E. (2005). Impact of restaurant hygiene grade cards on foodborne-disease hospitalizations in Los Angeles County. Journal of Environmental Health, 67(7), 32-36.

Thompson, S., de Burger, R., & Kadri, O. (2005). The Toronto food inspection and disclosure system: A case study. British Food Journal, 107(3), 140-149.

Toronto Public Health. (2002). Food premises inspection and disclosure program: Evaluation report. Toronto, Ontario: Author.

Wiant, C.J. (1999). Scores, grades and communicating about food safety Journal of Environmental Health, 61(9), 37-39.

Although most of the information presented in the Journal refers to situations within the United States, environmental health and protection know no boundaries. The Journal periodically runs International Perspectives to ensure that issues relevant to our international constituency, representing over 60 countries worldwide, are addressed. Our goal it to raise diverse issues of interest to all our readers, irrespective of origin.

Tino Serapiglia, M.H.Sc., C.P.H.I.(C)

Erin Kennedy, M.H.Sc.

Sylvanus Thompson, M.Sc., C.P.H.I.(C)

Ron de Burger, C.P.H., C.P.H.I.(C)

TABLE 1 Proportions Analysis for Episodes: Pre-program and Post-
program--Toronto, 1997-2004*

                    Episodes   Chi-               Relative
Illness            Pre   Post  Square    p-Value  Risk      95% CL**

Amoebiasis           10   22     6.4157   .0113   0.3940    0.1866-
                                                              0.8317
Campylobacter       676  503     7.0240   .0080   1.1649    1.0404-
  enteritis                                                   1.3043
Cryptosporidiosis     1    4           Insufficient numbers
Cyclosporiasis       23    2           Insufficient numbers
Giardiasis           30   39     2.8313   .0924   0.6667    0.4145-
                                                              1.0726
Hepatitis A          29    7    10.5547   .0012   3.5909    1.5736-
                                                              8.1944
Listeriosis           4    7           Insufficient numbers
Paratyphoid fever     1    1           Insufficient numbers
Salmonellosis       287  225     1.2959   .2550   1.1056    0.9300-
                                                              1.3144
Shigellosis          19  130   100.3309  <.0001   0.1267    0.0783-
                                                              0.2049
Typhoid fever         1    1           Insufficient numbers
VTEC infection       41   25     6.4991   .0108   1.8878    1.1487-
                                                              3.1025
Yersiniosis          25   32     2.1667   .1410   0.6770    0.4014-
                                                              1.1418
Food poisoning      164  832   495.4164  <.0001   0.1884    0.1596-
                                                              0.2224
All                1147  998     0.0085   .9267   0.9962    0.9182-
                                                              1.0808

* Source: Reportable Disease Surveillance System, Toronto, 2005.
** Confidence limits (lower-upper).

TABLE 2 Event Rates and Cochran-Armitage Trend Test*--Toronto, 2001-
2004**

Year  Total Inspections  Events  Event Rate (%)

2001  23,107             2337    10.0
2002  33,078             2134     6.5
2003  33,622             1680     5.0
2004  36,773             1222     3.3

* Trend test = test statistic: Z = 32.4419, p < .0001.
** Source: Toronto Healthy Environments Information System, 2005.