The-Restaurant-Inspection-as-a-Tool-to-Improve-Food-Safety

37 Chapter 4 The Impact of Standardization of Restaurant Inspection Reports on Posted Scores and Letter Grades 4.1 Introduction Restaurant inspection reports have traditionally been a part of the public domain, (Almanza, Nelson, & Lee, 2003). While there are some even within the public health agencies themselves that argue against releasing these inspection reports, (Koeune, 2000), motivation to actively release them to the public may be due to increased media scrutiny, encouraging a higher level of compliance with the food code or a desire by agencies to be more transparent to the public. As releasing of inspection scores increase across the country, it would be most meaningful if there were some consistent and uniform method of communicating hygienic conditions of food establishments. If the intention is for providing the level of hygienic conditions in the restaurant, then customers should be able to decipher the meaning of the letter grade, score or report. In the United States, people 19-64 take on average of 1875 trips that exceed 50 miles per year, while people 65+ take on average of 220 such trips, (Collia et al., 2003). This means that in this very mobile society, a significant number of people need to be able to translate inspection codes, icons, scores, grades, etc. to some meaningful correlation with restaurant conditions, if not, the message is lost or miscommunicated at best. If the intention is to improve compliance with the food code, then it is imperative that any behavioral changes resulting from the posting of scores is accurately computed across all communities to understand the true impact on compliance as well as determining whether the

38 Los Angeles experience, of lower food-related hospitalizations, (Simon et al., 2005), can be replicated. In our study, we conducted a survey to determine the extensiveness of posting of scores, or letter grades. We needed to determine whether the level of this practice warranted a call for consistency and uniformity. We also researched the effects of a standardized set of inspection reports on uniformity and consistency of scores. In our previous study (Lee, P.A. & Hedberg, C.W., 2011; Lee, Petrona A. and Craig Hedberg, Submitted for publication), we found that there was a consistency of approximately 56% of grades that resulted from applying grading criteria to un-standardized restaurant inspection reports. In this study we repeated the analyses, using the grading criteria from Los Angeles (LA) and New York City (NYC) and applying them to approximately 600 standardized inspection reports. 4.2 Materials and Methods The first part of our study was done to determine the prevalence of the posting of grades in this country. We conducted a telephone survey to all 50 states. All states responded. Response was received either from environmental health or food safety program personnel. All interviewees were asked the following five questions: (i) Is your department responsible for restaurant inspections throughout the state? (ii) How many local agencies are there in your state? (iii) Do you or your local agencies post inspection scores? (iv) Do you or your local agencies use scores or letter grades? (v) If you do post inspections, do you post on the web, on site? Other?

39 The responses were tabulated to determine the number of states in which there is posting at the state or local level; number of agencies that post on websites; number of agencies that post on restaurant premises; number of states that post in newspapers; number of agencies using scores or letter grades. Many states were not able to readily provide us with an exact number of local agencies, so we were not able to determine this through the survey. Routine Health Department Inspection reports that were translated to FDA Food Code Form 3-A, and were from over 600 jurisdictions across the US were used to conduct part two of this study. The data was standardized and the following information was available: City, State, Region, Zip, Inspection, Date/Time, Food Code, Risk Factor, Code Number and Code. A full inspection report was generated by combining the data by City, State, Region, Zip and Inspection Date/Time to determine violations that resulted from the same inspection. Scores were assigned, first using the LA’s Retail Food Inspection Guide, (County of Los Angeles). Based on this guide, the scores were converted to letter grades of A, B, C or “Post Scores”. LA Retail Food Inspection Guide, converts percent to letter grades as follows: A = 90 – 100, B = 80 – 89, C = 70 – 79. Scores < 70% are posted as percent, (per LA’s grading criteria). We then repeated the review for each inspection and assigned scores based on NYC’s grading criteria. Violation points were added and converted to letter grades based on NYC’s grading criteria. NYC Food Establishment Sanitary Inspection Scoring Parameters----A Guide to conditions, (New York City Department of Health and Mental Hygiene, 2011), outlines scoring based on violation points as follows: A = 0 – 13, B = 14 -27 and C = >27 violation points.

40 The data was analyzed to determine the frequency with which the same letter grade resulted from each inspection. 4.3 Results Table 4.1: *Summary of Status of Posting of Restaurant Inspection Report in the U.S. Actions Number Percent (of all states) Percent (of states that Post Inspection Reports # of states contacted 50 100 -- -- 46 92 98 28 # of states that post inspection reports, (i.e. at 24 state or local level) 37 45 90 # of states that post inspection reports online, (i.e. at state or local level) 13 26 # of states that post inspection reports on restaurant premises, (i.e. state or local level) 11 22 # of states that post inspection reports in Newspapers, (i.e. at state or local level) 17 34 # of states that score/grade inspection reports, ( i.e. at state or local level) (Source: Telephone Survey of all 50 states and information from March 7, 2011 Marler Blog) Table 4.1 indicates that in 46, (92%), states restaurant inspection results are posted on websites, restaurant premises or in newspapers at either the state or local level. Ninety percent, overall

41 uses the web; twenty-six percent post the scores in windows or doors and twenty two percent release the scores or grades in newspapers. Over a third scores or grades the inspections. The four states that reported no posting at the state or local levels included: Hawaii, Louisiana, Maine and Minnesota. Table 4.2: Summary of Grade Comparison between LA and NYC Grading Criteria LA NYC Compared to grade in Number Percent Number Percent other City 1341 83.96 1341 83.96 Same 226 14.15 17 1.06 1-grade higher 17 1.06 226 14.15 1-grade lower 13 0.81 00 2-grades higher 00 13 0.81 2-grades lower 1597 ~100.0 1597 ~100.0 Total Table 4.2 indicates that in approximately 84% of the inspections the same letter grade resulted from applying the two respective grading criteria. In approximately 14% of the inspections, application of LA’s grading criteria resulted in a higher letter grade. Application of NYC’s grading criteria resulted in a higher letter grade approximately 1% of the inspections. In approximately 1% of the inspections, application of the grading criteria there was a 2 grade difference with the higher grade resulting from application of LA’s grading criteria.

42 4.4 Discussion and Conclusions Our research reveals that posting of restaurant inspection results is being done widely throughout the country. Posting of scores is more common in the United States, although countries such as Canada and Singapore are also posting restaurant scores on a much smaller scale, (Boehnke & Graham, 2000). The United Kingdom has also started trial programs called posting Scores on Doors, (Worsfold & Worsfold, 2007). In forty six US states, restaurant inspection results are posted in windows or doors, in the newspapers or on websites. This includes posting that may be done by the state, county, city or other local agencies. Of those that post restaurant inspection reports, 17(%), use a score and 6(%), use a letter grade. It is apparent that posting of inspection scores is being used widely in the US and increasingly used, though slowly, in other countries. The posting of scores have been reported to be responsible for lowering hospitalizations due to Food foodborne illnesses, (Jin & Leslie, 2005) however other studies indicate that consumers are relying on other cues such as hygienic conditions of bathrooms, utensils, the physical conditions of the establishment, etc. as indicators for hygienic conditions of restaurants, (Henson et al., 2006). Usage of the information can be of greater public health benefit, if the grading is made relevant to some risk of avoiding foodborne illness and by standardizing the information released so that scores or letter grades communicate the same level of hygienic conditions regardless of political borders. While agencies are moving towards standardizing

43 environmental health specialists within their agencies, (Jones et al., 2004), we are not aware of any movement for a national standardization, though there have been recommendations calling for the standardization of a grading system, (Seiver & Hatfield, 2000). Standardization of restaurant inspection scores is important to management, especially large enterprises that try to standardize franchisees and company-owned facilities. They use inspection reports as a tool for training and consistency amongst franchised and company owned restaurants, (Hornsby, 2011) . Standardization in communicating to the public should not limit the information that agencies need to adequately address food protection, training, corrective actions or even enforcement, or other needs specific to individual establishments. More research is needed to develop a risk assessment and a tiered model that would deliver an accurate and concise message regarding the hygienic status to the public while allowing agencies to have detailed information for follow-up inspections, training, etc. In our study where we examined the results of standardized restaurant inspection reports, we found an increase from 56%, (Lee, P.A. & Hedberg, C.W., 2011) to 84% in consistency in letter grades. The increase in consistency was mainly due to the standardization of the data. Only the violation categories were included for grading. There were no inspector notes available, so there was not the ability to adjust the grades based the number of occurrences or whether they were chronic problems, etc. This level of standardization eliminated location, severity and other details that would give management, other restaurant workers and environmental health specialists the information to address critical violations in a comprehensive manner. While this level of standardization increased consistency and uniformity, it lacked in efficiency for the restaurant and the agency. It is

44 therefore important to define what is meant by standardization. Standardization, as defined for this paper refers only to how inspections are scored and reported to the public. There might be models would incorporate both levels needed, but further research is needed to address this adequately. A short list of some of the more comprehensive information that should be available would include reheating and cooling procedures, temperature records of potentially hazardous foods, thawing practices, HACCP procedures, hand washing issues, sanitizing practices, etc., so that there is adequate information for corrective actions and training specific to each establishment’s needs. Another item where it is crucial to gather routine information is the need to keep employee illness logs as these are critical in preventing and in the investigation of foodborne outbreaks. Research has shown that food workers play an important role in foodborne illnesses such as Salmonella, Norovirus, etc., (Todd, Greig, Bartleson, & Michaels, 2007). Other studies that support the role of the food worker in foodborne outbreaks include: (Roels et al., 1998), where food workers might have been implicated in cross-contaminating Tuna fish sandwich; (Mintz, Wragg, Mshar, Cartter, & Hadler, 1993), where employees at a corporate cafeteria, ate sliced raw tomatoes prepared by a food worker infected with Giardiasis; (Quiroz et al., 2000), in which a Cryptosporidiosis outbreak was linked to a food handler that prepared raw vegetables, (Gaynor et al., 2009) in which several persons internationally were infected by raw carrots prepared by a single caterer that served several international airlines. There are several other documented events where food handlers were either proved or suspected to be responsible for foodborne illnesses. Others on the list include: (Decker, Lavely, Hutcheson, & Schaffner, 1985), (Michaels et al., 2004),(Todd, Greig, Bartleson, & Michaels, 2009), among others.

45 In conclusion and for all the aforementioned reasons, detailed inspection reports are necessary for agencies to provide adequate resources for training, inspecting and enforcing where necessary. Detailed information is also important to establishment personnel so that they can address foodsafety issues specific to their restaurants. Also, any standardization should be structured to give some element of risk associated with any score or letter grade. Although this issue will not be addressed in this paper, our research group is conducting the studies that will allow us to fully address risks associated with scores or letter grades in future papers. The challenge will be to provide some uniform communication method that gives the consumer some clear, accurate and concise message that is a reflection of the foodsafety quality in the restaurant and yet retain enough information that meets he level of detailed information described earlier.

46 Chapter 5 The Ability of Restaurant Inspections to Predict Foodborne Outbreaks Likely Depends on the Nature of the Outbreak 5.1 Introduction Food safety requires vigilance that covers the entire journey of our food, from the “farm to the fork” i.e. everyone involved in growing, manufacturing, processing, transportation, storage and preparing and serving of each item of each meal every day. Everyday there are millions of meals prepared and served domestically and commercially. The average American eats approximately 2.8 meals commercially prepared every week, (Kant & Graubard, 2004). Although the incidence of foodborne outbreaks is relatively small, the most recent estimates by the CDC place the number of persons getting ill each year at 48 million. Stated another way, every year one in six people experiences a foodborne illness. This results in a hospitalization of 128, 000 persons and a death toll of 3,000 persons each year (US Food and Drug Administration, 4/11/2011). Restaurants are the most commonly identified source for foodborne outbreaks (Rangel, Sparling, Crowe, Griffin, & Swerdlow, 2005). There clearly is a challenge to find additional strategies to prevent contamination of our foods and for regulatory authorities to identify strategies that might predict foodborne incidences to further minimize the incidence and effects of food-related illnesses. An important part of keeping our food safe is the routine inspection of facilities that prepare food commercially. These include restaurants, schools, senior and daycare facilities, delicatessens, etc. The FDA food code recommends that agencies conduct risk-based

47 inspections and that the level of inspection should be in proportion with those establishments that fall within the high risk category, (Code, 2009) Organizations such as the International Association for Food Protection (IAFP) and Conference for Food Protection (CFP) bring the stakeholders across the food safety spectrum together. The CFP has a membership that includes the federal, state and local agencies as well as the food industry and consumers. It forwards recommendations from its biennial meetings to the FDA where they may be used to modify the food code. The CFP was responsible for initiating the standardized Food Establishment Inspection Report form which has become Form 3-A (Appendix 1) in the food code and has been adopted by many states. However, the Food Establishment Report Form is frequently amended to meet unique local food safety and licensing requirements. Although the form is modified among agencies, there is still a level of standardization that allows for some data analyses. Between April 25 and June 30, 2010, there was an outbreak of Salmonella ser. Hvittingfoss in a chain of restaurants located in Illinois. Ninety-seven cases, (including 12 food handlers), were confirmed with S. ser. Hvittingfoss during this period. All cases either lived in or had travelled to Illinois. Three produce items, (lettuce, tomatoes or olives), distributed by the same distributor, appeared to be linked to the outbreak (Illinois Department of Public Health, 2011). A retrospective study was conducted on restaurant inspection reports from eight of the counties involved in the outbreak. The objective of the study was to determine whether there were indicators present in the restaurant inspections that might have predicted which of the chain A restaurants would be involved in the outbreak. Records were obtained from

48 outbreak and non-outbreak restaurants in the chain. The study period was from January 1, 2008 through May 27, 2010. 5.2 Methods Our study was conducted on restaurant inspection reports received from eight of the twenty-nine Illinois counties involved in the foodborne illness outbreak. For each of the eight counties, we obtained all the inspection reports from chain A restaurants from 2008 through 2010. The cut-off date for inclusion in the study was May 27, 2010 since that was the date from the Illinois Department of Health notified local health departments of the outbreak. Only routine inspections were included in the study. All pre-open, seasonal, complaint and follow-up inspections were excluded. Non-outbreak restaurants were compiled into a control group. We further subdivided the outbreak-restaurants into two groups, high and low. Those in the high group had more than one confirmed case of Salmonella ser. Hvittingfoss and with dates of exposure over a period of at least 3 days, suggesting possible amplification of salmonella within the restaurants. Those in the low group had only one confirmed case or multiple cases reporting exposure on the same or consecutive days, consistent with consumption of contaminated food items from the same source, such as a common box of produce. The average number of cases per establishment in the eight counties was determined by dividing the total number of cases (46) by the number of Chain A restaurants (16). The number of cases per restaurant was determined and compared to an expected number of cases based on the Poisson distribution with λ = the mean number of cases per restaurant. The sum of (0-E)²/E was calculated for case Frequencies per restaurant and the Chi square value for

49 the total calculated. A p-value <0.05 was used to indicate the presence of clustering of cases by establishment. Results of inspections were entered into Excel spreadsheets. The following information was extracted for analyses: county, address, identification number, type of inspection, date of inspection, violation category, county violation number, violation description and demerit points. The data was analyzed using Excel Descriptive Statistics and Pivot Tables. The final descriptive analyses were done using CDC Epi Info 7. Descriptive Statistics; ANOVA (Parametric Test for Inequality of Population Means), Chi square and the Mann-Whitney/Wilcoxon Two-Sample Test were calculated for each variable. The parameters studied included total demerit scores over the period of January 1, 2008 through May 27, 2010 in the outbreak and non-outbreak restaurants; the total demerit scores of the last inspections before the outbreak; hand washing and good hygienic practices violation scores; cross-contamination violation scores and all violation categories found per establishment per inspection. We also applied grading criteria from Los Angeles and New York City to evaluate the usefulness of these grading systems to predict whether a restaurant was likely to be involved in the outbreak. 5.3 Results Inspection results were collected from 106 Chain A establishments associated with the outbreak in the eight counties included in the study. Forty-six outbreak-associated cases were linked to 23 of the Chain A restaurants. There were no significant differences between the outbreak and non-outbreak restaurants for the number of demerit points attributed to hand washing or cross-contamination. Differences in the observed versus expected distribution of

50 cases by restaurant suggested some clustering of cases by establishment, consistent with either a non-uniform source of contamination or amplification of cases within some establishments (Table 1). Sixteen met the definition for “low-outbreak” (13 with single cases) and 7 were considered “high-outbreak” (3 with 5 or 6 cases). The data for the mean total demerit score for the overall period for the control restaurants and the outbreak restaurants were 13.5 and 10.9 respectively and the mean total demerit for the last inspections were 13.9 and 10.0 respectively (Table 2). There were no significant differences between controls and case restaurants. Table 5.1: Distribution of Salmonella ser. Hvittingfoss cases No. of cases/restaurant No. of restaurants Total no. of cases (46) N=1 12 12 N=2 5 10 N=3 2 6 N=4 0 0 N=5 0 0 N=6 3 18

51 Table 5.2: Descriptive Statistics Inspection/Demerit Non-outbreak Outbreak Point Category (# obs) mean (s.d) (# obs) mean (s.d.) Total 212 13.5 (11.4) 67 10.9 (10.4) Last Inspection 83 13.9 (11.5) 23 10.0 (9.3) Hand washing 210 1.0 (2.0) 67 1.0 (2.2) (total) 83 Hand washing 212 1.1 (2.1) 23 1.1 (2.1) (last) Cross- 83 1.9 (2.3) 67 1.6 (1.7) Contamination (total) 1.9 (2.2) 23 1.6 (1.6) Cross- Contamination (last)

52 Of 10 inspections conducted during May 2010 when cases were actively exposed, 8 were conducted among control restaurants and 2 were conducted in case-associated restaurants. One of these had six cases associated with it. An inspection on 5/4, and a repeat inspection on 5/18 noted violations for unclean food contact surfaces and inadequate sanitary solutions. Only one of the eight control restaurants had either of these violations cited. There were no differences observed from analyses for high and low outbreak restaurants separately or between all outbreak restaurants combined, compared to the control groups. There were no significant differences observed between outbreak and non-outbreak restaurant inspections with respect to grades imputed from either the LA County or the NY City restaurant inspection grading criteria (Tables 3A and 3B). Table 5.3 (A): Routine Inspection Results During Year Before Outbreak of Salmonellosis in a Restaurant Chain: Imputed Grade: LA Criteria Outcome AB ≤C 15 (28%) 7 (21)% 1 (9%) Outbreak Restaurants (n=23) Non-Outbreak 39 (72%) 27 (79%) 10 (91%) Restaurants (n = 76)

53 Table 5.3 (B): Routine Inspection Results During Year Before Outbreak of Salmonellosis in a Restaurant Chain: Imputed Grade: NYC Criteria Outcome AB ≤C Outbreak Restaurants 11 (29%) 8 (25%) 4 (14%) (n=23) 25 (86%) Non-Outbreak 27 (71%) 24 (75%) Restaurants (n = 76) 5.4 Limitations: We were not able to obtain inspection records for all of the chain A restaurants involved in this outbreak. However, our study includes restaurants accounting for almost half of the outbreak-associated illnesses. Given the lack of association observed between inspection results and illness, it is unlikely that doubling the size of the study would have materially changed the results. Also variation among agencies in the scoring of violations allowed for partial grading based on whether the violation was chronic or not. Because of this variation, scores for some critical violations were not always consistent within the same agency or among all agencies. In addition, categories of violation were not identical among agencies. An example of this was in the ‘Hygienic Practices’ where for some agencies ‘Hand washing’ was a stand-alone category, while in others it was included into ‘Hygienic Practices. Our solution was to group them all into ‘Hand washing and Hygienic Practices’.

54 5.5 Discussion and Conclusions Routine restaurant inspections are seen by the public as a way for ensuring that food safety is not compromised in restaurants (Jones & Grimm, 2008). Regulatory agencies also see restaurant inspections as effective tools for ensuring compliance with the food code and have associated a passing score with safer foods. Studies have also linked frequency of inspections (Allwood et al., 1999), total scores and types of violations ((Simon et al., 2005) to some measure of food compliance status and hence some measure of improved food safety (Buchholz et al., 2002; Buchholz, Run, Kool, Fielding, & Mascola, 2002; Irwin et al., 1989; Luby, Jones, & Horan, 1993; Sato, 2008; Shapiro et al., 1999). Other studies have shown that these measures by themselves are poor indicators of risk of foodborne illnesses for some outbreaks (12, 13, 18) and that other factors such as seating capacities, inspection history and the presence of certified kitchen managers, may influence the likelihood of foodborne outbreaks (Buchholz et al., 2002; Hedberg et al., 2006). In the studies where inspection scores predicted foodborne illnesses, total scores as well as specific violation categories were contributing factors. For example, improper holding or cooking temperatures contributed to foodborne pathogens such as Clostridium perfringens, Escherichia coli, etc. (Irwin et al., 1989; Irwin et al., 1989; Luby et al., 1993; Shapiro et al., 1999). As such, violations such as improper cooking and holding temperatures, are code items that are explicit in the code and are likely to be detected during an inspection. In some of these studies the low scores of the outbreak restaurants were accompanied by a number of other contributing factors, such as certified kitchen managers

55 (Hedberg et al., 2006), critical violations (Irwin et al., 1989) and presence of ill employees (White, Karen E¹., Michael T. Osterholm¹, Josep A. Mariotti¹, Jack A. Korlath¹, Don H. Lawrence², Terry L. Ristinen¹ & Harry B. Greenburg³, 1986). It could reasonably be interpreted that low scores were indicative of widespread system failures within these restaurants. Other preventative measures such as low water activity, pH, salinity and low sugar may not always be evident to all inspectors during a routine restaurant inspection and may require a HACCP or active managerial type inspections to identify some of these subtle critical control points. In our analysis of this outbreak, total demerit scores for the overall period or last inspections before the outbreak, routine inspections revealed no contributory factors leading to the salmonella outbreak. This included specific analyses for hand washing and cross- contamination as possible contributory factors. The Illinois Department of Health listed the contributory factor as item C7, (ingestion of contaminated raw product), and the suspected vehicle as produce items. There were 12 food handlers who tested positive for Salmonella ser. Hvittingfoss; however, they were ill at the same time as were the patrons. It appeared that by the time the outbreak was detected and stool cultures obtained, the outbreak was near its conclusion and although there were ill employees, there did not appear to be any further amplification that resulted in ill patrons (6) thus, the results of our analyses are consistent with this outbreak being caused by distribution of a contaminated fresh produce item without further amplification within individual restaurants. In order to explain for the results of this study as well as to understand the apparent conflicting results from other studies we examined the conditions under which a restaurant may be the source of a foodborne outbreak. Medus and colleagues described a model in which Salmonella is introduced via a contaminated source such as eggs (11). Undercooked foods

56 containing eggs provide a direct route for patrons (Figure 1). Failure to maintain good hygienic practices provide opportunities for contamination of food contact surfaces and ready- to-eat foods. Lack of proper holding, cooking, cooling and reheating will remove additional barriers for growth and multiplication. Infection of foodwokers and lack of other food protection practices will result in multiple exposure pathways. These failures in one or more systems for which the food code is built to detect are the type of violations that will be reflected in inspections. These are the type of violations that will be reflected in inspections. The ability for the routine restaurant inspection to capture factors that will lead to foodborne illnesses is optimal for outbreaks that fit this model (Model 1). Studies in which routine inspection results were associated with foodborne outbreaks have primarily involved this type of outbreak (Figure 1).

57 Model 2 is one in which an incoming fresh produce item may be contaminated at levels such that no further amplification within the restaurant is required to permit transmission to patrons. The restaurant merely serves as a ‘pass through’ to its customers (figure 2). For this scenario, routine inspections are not likely to predict foodborne illnesses. A third related model (Model 3) involves the introduction of foodborne pathogen by infected foodworkers. This is a major pathway for the occurrence of outbreaks of Norovirus, which is recognized as the most common cause of foodborne illness. Because Norovirus does not replicate on food or in the environment, no food handling violations beyond the initial contamination are required to cause the outbreak. Also, because food handler illnesses are sporadic events, routine inspections are not likely to detect them or prevent the outbreaks they cause. Studies that have failed to associate results of routine inspections with the occurrence of outbreaks have included a relatively high proportion of outbreaks following

58 models 2 and 3. The end result is that the restaurant inspection’s ability to predict an outbreak is based on whether the incidence was a result of the breakdown of the hygienic conditions, a ‘pass-through’ event, or a one-time or sporadic event resulting from the presence of ill employees. Because a high proportion of foodborne outbreaks represent these latter conditions, restaurant inspections have limited usefulness in communicating information about risk to the public. Thus, while the public may want to see a complex inspection represented as a simple letter grade, and the shame of a “bad” grade may motivate restaurant operators to improve, consumers should not interpret the grade as an indicator of the risk of acquiring a foodborne illness at the restaurant. Studies have shown that for some outbreaks where the restaurant inspection predicted the outbreak and corrective instructions were given, the outbreak was not prevented (Evans,

59 Tromans, Dexter, Ribeiro, & Gardner, 1996). This indicates that predictability does not necessarily translate to prevention. Additional measures are needed to ensure that restaurants maintain compliance. Food worker certification, additional training and education, more follow-up inspections, more severe penalties for chronic violations, health benefits/sick leave for food workers to encourage ill employees to stay home may be some of the strategies that should be further studied to determine their effectiveness on reducing foodborne illnesses that result from the presence of ill employees in food establishments. Ultimately this could benefit the restaurant in having a healthier work crew that may lead to improved hygienic conditions and reduced risk of illness. Translating these to quantifiable measures into inspection results is a considerable challenge. To illustrate the impact of these models on the association between inspection results and outbreaks, we have cataloged and applied these models to a series of studies that have been published (Table 5.4).

60 Table 5. 4: Case control studies of restaurant inspections to predict foodborne illnesses and their fit with the proposed models Author Year Agent Vehicle Routine Simple/ Explanation # Insp. Complex (Model 1, 2, Predicts 3) (Y/N) Irwin et al. 1989 Unknown Poultry Y Single 1 (Model 1) (most) (common) Bucholz et al. 2001 n/a n/a Y Insufficient 3 Insufficient Information Information) Jones et al. 2004 multiple multiple N Complex 2 (Model 3) Patel et al. 2008 Salmonella chicken Y Single 4 (Model 1) Montevideo Lee and 2012 Salmonella Produce N Single 6 (Model 2) Hedberg ser. Hvittingfoss There was a good fit for outbreaks where there was sufficient data from inspections performed before the outbreak and the agents and vehicles for transmission were known. Of four published studies comparing outbreak and non-outbreak restaurants, three fit our

61 models. The fourth did not provide enough data for us to apply the model scenarios. We also examined some case studies (15, 18, 20) and found that of the four studies presented all fitted our models. Additional studies are needed to provide a basis for updating the food code to adequately address the ‘pass-through’ and ‘one-time’ types of outbreaks. The FDA food code addresses the need to exclude employees with symptoms of foodborne illnesses. However, illness logs are not explicitly required or listed on Form 3-A Food Establishment Inspection Report. The employee illness logs should collect information regarding ill employees as well as any contact with persons that have symptoms of foodborne illnesses. The illness logs should be maintained and current and its review should be part of the inspection. Lastly, it should be recognized that prediction does not equate to prevention. More effort needs to be directed specifically at prevention once specific risk factors are identified by the inspection. Food protection should be vigilant over the entire length of the food chain. Prevention measures such as HACCP, (that is modified to be made relevant to specific systems), and other pathogen reduction measures need to be applied from the farm to the table. Other prevention strategies accompanied by appropriate levels of inspection must be in place to protect all food sources. The restaurant inspection is too close to the end of the chain for it to provide full protection and predictability for the entire food chain.

62 Chapter 6 Thesis Defense My hypothesis states that the routine restaurant inspection can be an effective tool to improve food safety. I attempted to prove the hypothesis by examining three major aspects of the inspection. These three areas included: 1. The Inspection Process 2. Uniformity and consistency in communicating of the results from the routine restaurant inspection 3. Application of the routine restaurant inspection in predicting foodborne illnesses. To respond to the first question, I conducted two researches. The first was an examination of routine inspections from three Minnesota agencies to determine whether there was a correlation between critical violations and those found in the physical facilities. I found that there was correlation and that there were approximately three violations found for every one found in the physical facilities. This indicated validity in the process and that major attention was being given to the critical factors that result in foodborne illnesses. As a valid food safety tool, it is important that all elements of the food code are captured during the inspection. To respond to the second question, I conducted three researches. The first was a study that converted the results from routine restaurant inspections from the three Minnesota agencies into graded scores using grading criteria from Los Angeles and New York City. This resulted in a consistency of approximately 56%. It was important to determine if grades and scores are being posted at a level where uniformity and consistency were of importance.

63 I conducted a study to determine the prevalence of the practice and found that in 46 of the 50 states there was posting of scores at the state or local level. With the public taking a significant number of their meals in restaurants and also travelling outside of their political borders, I determined that uniformity and consistency is an important issue that will allow the public to make informed and accurate decisions based on the results from the routine restaurant inspection. I further conducted research using a standardized dataset to determine its impact on uniformity and consistency in the resulting letter grades. I found that consistency in the letter grades improved from 56% using un-standardized reports to 84% using standardized restaurant inspection reports. The limitation is that standardizing the reports also filters out important details that both restaurant managements and environmental health specialists will need for correcting hygienic failures and for training and planning for additional resources. In order to address the third question, I conducted a research using a foodborne outbreak in an Illinois restaurant chain to determine whether routine inspections predicted the outbreak. I found that the routine inspections done before the outbreak were not predictive of the outbreak and that by applying the grading criteria from LA and NYC did not change the outcome. In order to understand why some studies indicate that the routine restaurant inspection will predict a foodborne outbreak and others indicate that the inspection is not predictive of a foodborne outbreak my project advisor and I examined the conditions that are representative for foodborne outbreaks and determined that the outbreaks usually occur under one of three conditions, namely:

64 1. Poor hygienic conditions in the restaurant which serve to amplify the pathogen and the contributory factors will be identified in the routine restaurant inspection. 2. Contaminated sources that require no further amplification and a breakdown of hygienic conditions in the restaurant. The restaurant merely serves as a pass-through vehicle and the routine restaurant inspection will not be predictive of these types of outbreaks. 3. A one-time or sporadic failure of good hygienic practices by a single individual that is affected with or has contact with a someone that is affected with a foodborne pathogen. Although there may be food code items that are detectable in the inspection, timing of the inspection or because it is not widespread, it may not be detected in the inspection, therefore the routine restaurant inspection will not predict this type of outbreak. Examination of well-known studies as to their fit in our models presented good fits with our models. 6.1 Limitations of the Study: In applying the grading criteria from Los Angeles or New York City, some of the interpretation of the criteria involved some subjective decisions. I contacted department personnel for some of the grading decisions, but it is likely that grading by an un-standardized third party may not always agree with those of the environmental health specialists that have been standardized with the agency. In reviewing data from eight separate Illinois counties, there was inconsistency in how violations were grouped and scoring techniques between agencies. My response was to include violations into the broadest sub-category to capture all similar violations.

65 In conclusion, the research has shown that the restaurant inspection is an effective tool in achieving food code compliance. The inspection process identifies critical items as well as minor items responsible for foodborne illnesses. The use of scores and letter grades is effective in providing a clear and concise status of code compliance to restaurant management and owners. However there are several areas that could be addressed to improve its effectiveness as a foodsafety tool. The areas where additional improvements were identified are: 1. Communication to the public. There needs to be uniformity and consistency in the scoring and reporting of grades. A percent or letter grade should mean the same throughout the country. This will allow the public to better understand hygienic conditions as patrons travel between jurisdictional borders. 2. Standardization of scores and grades. To aid in the uniformity of scores and grades, it will be necessary to standardize inspection reports so that scores and letter grades represent the same hygienic conditions in all jurisdictions. Our study indicated an improvement from 56% uniformity and consistency to 84% by using standardized report data. Standardization should not eliminate essential details of the inspection that is necessary for training, education, data collection and enforcement. The study indicated that in 92% of the states inspection results are posted at the state, county or local level to the public, therefore the practice is widespread enough so that the communication should be uniform. 3. Predictability of the routine restaurant inspection. The routine restaurant inspection is a useful tool in predicting outbreaks resulting from a breakdown of systemic hygienic conditions within the restaurant. It is not useful in identifying outbreaks that do not

66 require amplification within the restaurant. There are mainly two types of outbreaks that fit this scenario. The first is the ‘pass through’ outbreak. In instances where sources are adulterated to the level of an infective dose, the restaurant serves as a vehicle to pass through the pathogen to the customer. Typically, the agent is a low infective dose agent similar to salmonella. Likely vehicles are fruits and vegetables that are eaten or incorporated into other foods without a kill step. For this type of outbreak, good source control is the key. As it exists today, the restaurant inspection is not likely to predict this outbreak. Management must ensure that all sources are wholesome. Food protection logs may be necessary to track the journey of the food to the restaurant. In the second scenario for which the routine inspection is not likely to detect the outbreak, it involves the presence of an ill employee or one that has had contact with persons having foodborne illness. Norovirus foodborne outbreaks are typical agents for this pathway. Failure of good hygienic practices by the individual may result in a one-time outbreak event. With other barriers in place, the outbreak will be contained. Although the contributory factors may include hand washing and cross-contamination, the routine inspection may not identify the violations because they are momentary or sporadic and are not system-wide. For these outbreaks, the food code needs to be expanded with explicit requirement for illness logs and other prevention measures to be included as checklist items during the inspection. 6.2 Future Researches: It is important to note that prediction of the inspection does not always translate to prevention. Additional research is needed to identify the role of follow- up and other measures that should be in place as prevention strategies. Research is also needed to determine the most effective means of communicating the results of the routine

67 restaurant inspection to the public and how to include the risk of avoiding foodborne illnesses in the communication. 6.3 Conclusion: Finally, it is evident that the routine restaurant inspection does what it was constructed to do, that is, it identifies pathogens such as Clostridium perfringens and others that follow a similar pathway. However, for pathogens that follow the two alternate pathways, the routine inspection is not likely to predict the resulting foodborne outbreaks. To add to the confusion, the posting of scores is implying some risk of illness. However, scores and letter grades are not uniform among agencies. Additional studies are needed to identify how the food code should address these shortcomings.

68 Bibliography Allwood, P. B., Jenkins, T., Paulus, C., Johnson, L., & Hedberg, C. W. (2004). Hand washing compliance among retail food establishment workers in minnesota. Journal of Food Protection&# 174;, 67(12), 2825-2828. Allwood, P. B., Lee, P., & Borden-Glass, P. (1999). The vital role of restaurant health inspections. Journal of Environmental Health, 61(9), 25-29. Almanza, B. A., Ismail, J., & Mills, J. E. (2002). The impact of publishing foodservice inspection scores. Journal of Foodservice Business Research, 5(4), 45-62. Almanza, B. A., Nelson, D. C., & Lee, M. L. (2003). Food service health inspectors' opinions on the reporting of inspections in the media. Journal of Environmental Health, 65(10), 9-14. Boehnke, R. H., & Graham, C. (2000). International survey on public posting of restaurant inspection reports, and/or grade card posting schemes based upon health inspections. Region of Ottawa- Carleton Health Department, Bryan, F. L. (1990). Hazard analysis critical control point (HACCP) systems for retail food and restaurant operations. J.Food Prot, 53(11), 978-983. Buchholz, U., Run, G., Kool, J., Fielding, J., & Mascola, L. (2002). A risk-based restaurant inspection system in los angeles county. Journal of Food Protection&# 174;, 65(2), 367-372. Cervantes, H., ACPV, D., & Watkinsville, G. (2005). ASSESSING THE RESULTS OF THE EUROPEAN UNION BAN ON ANTIBIOTIC FEED ADDITIVES. Paper presented at the POULTRY NUTRITION CONFERENCE, pp. 38.

69 Choi, J., Nelson, D. C., & Almanza, B. (2010). The impact of inspection reports on consumer behavior: A pilot study. Food Control, Clayton, D. A., & Griffith, C. J. (2004). Observation of food safety practices in catering using notational analysis. British Food Journal, 106(3), 211-227. Code, F. D. A. F. (2009). FDA: Food Code, Collia, D. V., Sharp, J., & Giesbrecht, L. (2003). The 2001 national household travel survey: A look into the travel patterns of older americans. Journal of Safety Research, 34(4), 461-470. County of Los Angeles. Retrieved 4/24, 2010, from http://www.lapublichealth.org/eh/RFIG_Old_org/RET Cruz, M. A., Katz, D. J., & Suarez, J. A. (2001). An assessment of the ability of routine restaurant inspections to predict food-borne outbreaks in miami-dade county, florida. American Journal of Public Health, 91(5), 821. Decker, M. D., Lavely, G. B., Hutcheson, R. H., & Schaffner, W. (1985). Food-borne streptococcal pharyngitis in a hospital pediatrics clinic. JAMA: The Journal of the American Medical Association, 253(5), 679. Dundes, L. (2001). Scores and grades: A sampling of how college students and food safety professionals interpret restaurant inspection results. Journal of Environmental Health, 64(5), 14-22. Evans, M., Tromans, J., Dexter, E., Ribeiro, C., & Gardner, D. (1996). Consecutive salmonella outbreaks traced to the same bakery. Epidemiology and Infection, 116(02), 161-167.

70 Gaynor, K., Park, S., Kanenaka, R., Colindres, R., Mintz, E., Ram, P., et al. (2009). International foodborne outbreak of< i> shigella sonnei</i> infection in airline passengers. Epidemiology and Infection, 137(03), 335-341. Green, L. R., Selman, C. A., Radke, V., Ripley, D., Mack, J. C., Reimann, D. W., et al. (2006). Food worker hand washing practices: An observation study. Journal of Food Protection&# 174;, 69(10), 2417-2423. Hedberg, C. W., Smith, S. J., Kirkland, E., Radke, V., Jones, T. F., & Selman, C. A. (2006). Systematic environmental evaluations to identify food safety differences between outbreak and nonoutbreak restaurants. Journal of Food Protection&# 174;, 69(11), 2697-2702. Henson, S., Majowicz, S., Masakure, O., Sockett, P., Jones, A., Hart, R., et al. (2006). Consumer assessment of the safety of restaurants: The role of inspection notices and other information cues. Journal of Food Safety, 26(4), 275-301. Hornsby, G. D. (2011). Consistency between franchised and company-owned locations: A study of sonic drive-in food safety scores in the state of oklahoma. Illinois Department of Public Health. (2011). IDPH 2010, subway foodborne outbreak investigation [report]. Illinois: IDPH. Irwin, K., Ballard, J., Grendon, J., & Kobayashi, J. (1989). Results of routine restaurant inspections can predict outbreaks of foodborne illness: The seattle-king county experience. American Journal of Public Health, 79(5), 586.

71 Jin, G. Z., & Leslie, P. (2003). The effect of information on product quality: Evidence from restaurant hygiene grade cards*. Quarterly Journal of Economics, 118(2), 409-451. Jin, G. Z., & Leslie, P. (2005). The case in support of restaurant hygiene grade cards. Jones, T. F., & Grimm, K. (2008). Public knowledge and attitudes regarding public health inspections of restaurants. American Journal of Preventive Medicine, 34(6), 510-513. Jones, T. F., Pavlin, B. I., LaFleur, B. J., Ingram, L. A., & Schaffner, W. (2004). Restaurant inspection scores and foodborne disease. Emerg Infect Dis, 10(4), 688-692. Kalluri, P., Crowe, C., Reller, M., Gaul, L., Hayslett, J., Barth, S., et al. (2003). An outbreak of foodborne botulism associated with food sold at a salvage store in texas. Clinical Infectious Diseases, 37(11), 1490. Kant, A. K., & Graubard, B. I. (2004). Eating out in america, 1987-2000: Trends and nutritional correlates. Preventive Medicine, 38(2), 243-249. Koeune, T. J. (2000). Should restaurant inspection reports be published? Journal of Environmental Health, 62(8) Kovats, R., Edwards, S., Hajat, S., Armstrong, B., Ebi, K., Menne, B., et al. (2004). The effect of temperature on food poisoning: A time-series analysis of salmonellosis in ten european countries. Epidemiology and Infection, 132(3), 443-453. Lee, P.A. & Hedberg, C.W. (2011). Restaurant inspection as a tool for food safety. Unpublished manuscript.

72 Lee, Petrona A. and Craig Hedberg. (Submitted for publication). Restaurant inspections as a tool for foodsafety. Unpublished manuscript. Luby, S., & Jones, J. (1993). A large salmonellosis outbreak associated with a frequently penalized restaurant. Epidemiology and Infection.London, New York NY, 110(1), 31-39. Luby, S., Jones, J., & Horan, J. (1993). A large salmonellosis outbreak associated with a frequently penalized restaurant. Epidemiology and Infection.London, New York NY, 110(1), 31-39. Medeiros, L. C., Hillers, V. N., Kendall, P. A., & Mason, A. (2001). Food safety education: What should we be teaching to consumers? Journal of Nutrition Education, 33(2), 108-113. Michaels, B., Keller, C., Blevins, M., Paoli, G., Ruthman, T., Todd, E., et al. (2004). Prevention of food worker transmission of foodborne pathogens: Risk assessment and evaluation of effective hygiene intervention strategies. Food Service Technology, 4(1), 31-49. Mintz, E. D., Wragg, M. H., Mshar, P., Cartter, M. L., & Hadler, J. L. (1993). Foodborne giardiasis in a corporate office setting. Journal of Infectious Diseases, 167(1), 250. Mortimore, S. (2001). How to make HACCP really work in practice. Food Control, 12(4), 209-215. NYC. (2010). Retrieved 4/24, 2010, from (http://home2.nyc.gov/html/doh/downloads/pdf/notice- 81-amend-1209.pdf New York City Department of Health and Mental Hygiene. (2011).

73 Penman, A. D., Webb, R. M., Woernle, C. H., & Currier, M. M. (1996). Failure of routine restaurant inspections: Restaurant-related foodborne outbreaks in alabama, 1992, and mississippi, 1993. Journal of Environmental Health, 58(8), 23-23. Quiroz, E. S., Bern, C., MacArthur, J. R., Xiao, L., Fletcher, M., Arrowood, M. J., et al. (2000). An outbreak of cryptosporidiosis linked to a foodhandler. Journal of Infectious Diseases, 181(2), 695. Rangel, J. M., Sparling, C., Crowe, P. H., Griffin, P. M., & Swerdlow, D. L. (2005). Epidemiology of escherichia coli O157:H7 outbreaks, united states, 1982–2002. EID, 11, 602-609. Reske, K. A., Jenkins, T., Fernandez, C., VanAmber, D., & Hedberg, C. W. (2007). Beneficial effects of implementing an announced restaurant inspection program. Journal of Environmental Health, 69(9), 27-80. Roels, T., Wickus, B., Bostrom, H., Kazmierczak, J., Nicholson, M., Kurzynski, T., et al. (1998). A foodborne outbreak of campylobacter jejuni (O: 33) infection associated with tuna salad: A rare strain in an unusual vehicle. Epidemiology and Infection, 121(2), 281-287. Sato, H. D. (2008). Risk assessment model to predict foodborne illness outbreaks at restaurants. Risk Assessment Model to Predict Foodborne Illness Outbreaks at Restaurants, Pro Quest Dissertations & Theses, 207. Seiver, O. H., & Hatfield, T. H. (2000). Grading systems for retail food facilities: A risk-based analysis. Journal of Environmental Health, 63(3)

74 Shapiro, R., Ackers, M. L., Lance, S., Rabbani, M., Schaefer, L., Daugherty, J., et al. (1999). Salmonella thompson associated with improper handling of roast beef at a restaurant in sioux falls, south dakota. Journal of Food Protection&# 174;, 62(2), 118-122. Simon, P. A., Leslie, P., Run, G., Jin, G. Z., Reporter, R., Aguirre, A., et al. (2005). Impact of restaurant hygiene grade cards on foodborne-disease hospitalizations in los angeles county. Journal of Environmental Health, 67(7), 32-37. Sperber, W. H. (2005). HACCP and transparency. Food Control, 16(6), 505-509. Todd, E. C. D., Greig, J. D., Bartleson, C. A., & Michaels, B. S. (2007). Outbreaks where food workers have been implicated in the spread of foodborne disease. part 3. factors contributing to outbreaks and description of outbreak categories. Journal of Food Protection&# 174;, 70(9), 2199-2217. Todd, E. C. D., Greig, J. D., Bartleson, C. A., & Michaels, B. S. (2009). Outbreaks where food workers have been implicated in the spread of foodborne disease. part 6. transmission and survival of pathogens in the food processing and preparation environment. Journal of Food Protection&# 174;, 72(1), 202-219. Unnevehr, L. J., & Jensen, H. H. (1999). The economic implications of using HACCP as a food safety regulatory standard. Food Policy, 24(6), 625-635. US Food and Drug Administration. (2003). US Food and Drug Administration. (2011). , November 29, 2011.

75 US Food and Drug Administration. (4/11/2011). What you need to know about foodborne illness- causing organisms in the U.S.. Retrieved December 12, 2011, from http://www.fda.gov/Food/ResourcesForYou/Consumers/ucm103263.htm USDA. (2009). Presidents food safety working group. Retrieved December 20, 2011, from http://www.foodsafetyworkinggroup.gov/ContentLearn/HomeLearn.htm Vandeplas, S., Marcq, C., Dubois Dauphin, R., Beckers, Y., Thonart, P., & Théwis, A. (2008). Contamination of poultry flocks by the human pathogen campylobacter spp. and strategies to reduce its prevalence at the farm level. Biotechnologie, Agronomie, Société Et Environnement [= BASE], 12(3) Wetterling, A. (Ed.). (2009). Cleanliness in retail (1st ed.). Zurich: rheingold Institute for Qualitative Market and Media Research. White, Karen E¹., Michael T. Osterholm¹, Josep A. Mariotti¹, Jack A. Korlath¹, Don H. Lawrence², Terry L. Ristinen¹ & Harry B. Greenburg³. (1986). A FOODBORNE OUTBREAK OF NORWALK VIRUS GASTROENTERITIS EVIDENCE FOR POST-RECOVERY TRANSMISSION. American Journal of Epidemiology, 124(1), 120-126. Worsfold, D., & Worsfold, P. M. (2007). Evaluating food hygiene inspection schemes:'scores on doors' in the UK. International Journal of Consumer Studies, 31(6), 582-588.

76 Appendix 1

77

78

79

80 Appendix 2 Minnesota Food Establishment Form