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

The Restaurant Inspection as a Tool to Improve Food Safety A DISSERTATION SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOOL OF THE UNIVERSITY OF MINNESOTA BY Petrona Albertha Lee IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY Craig Hedberg, PhD, Adviser May 2012

UMI Number: 3513570 All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent on the quality of the copy submitted. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. UMI 3513570 Copyright 2012 by ProQuest LLC. All rights reserved. This edition of the work is protected against unauthorized copying under Title 17, United States Code. ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, MI 48106 - 1346

Copyright by Petrona Lee, PhD 2012 All Rights Reserved

i Acknowledgments So many times it has been said that listing the names of people who have played a vital role in one’s success is in itself a reckless task. Inevitably, there will be an omission of someone that should be acknowledged. However, at the risk of committing such a faux pas, I must acknowledge the following persons that without them, I would not have been successful in this endeavor. First, I would like to thank Dr. Bill Toscano, my advisor, who has been both inspirational and motivational in my getting the job done. Dr. Toscano has told me repeatedly, “you have to finish”. Dr. Toscano encouraged me to enroll in the PhD program, advised me in accepting an internship in Ecuador and to attend a global conference in India. Both of these experiences broadened my public health exposure and helped me to solidify my desire to continue my work in developing countries. Thanks Dr. Toscano for your wisdom and your generous support in my education. I also would like to thank Dr. Craig Hedberg for steering me into developing a coherent thesis research. Dr. Hedberg skillfully allowed me to take the leap from being the field environmental health specialist that applies public health policies to being the researcher that identifies gaps in the field and finds solutions to improve those deficiencies. Thanks also to the other dissertation committee members, Dr. Carlota Medus and Dr. Phil Peterson, for their advice, constructive critiques and their encouragements. Special Thanks to Shawn Tubbs from the University of Minnesota, for her assistance in collecting survey data; to Curt Fernandez, Tim Jenkins and Rebecca Sandell from the Minneapolis Environmental Health Department; Colleen Paulus, LeeAnn Austin and Peter

ii Lindell from the Minnesota Department of Health and Kris Kristoferson from Washington County Health Department for their assistance in providing Minnesota restaurant information for data analyses; also to personnel from Champaign, Christian, Fulton, LaSalle, Macon, Shelby, Sangamon and Winnebago for their gracious cooperation in providing me with data from the 2010 Salmonella restaurant outbreak restaurants in their jurisdictions. Thanks to Dr. Paul Allwood who provided me with the opportunity to be one of his teaching assistants. The experience was invaluable as it provided financial assistance as well as the opportunity to develop teaching tools that will be relevant to my future career. My family has been very supportive in every step of this adventure. One member that has always stood by me is my sister, Hyacinth Campbell Roberts. No one could possibly have a sister that has been more of a friend, consoler and a leader of my cheering section. She together with and her children Karif and Makela have made me realize that I belong to a great family and that rain or shine; they will be there for me. There are several other family members and friends that I wish to acknowledge at this time. These include Don Steele, Jerry Steele, Marguerita and Korry Franklin, Mary Jane Maser, Joan Dion, Patricia Lalley, Maggie Bovis, Doris Martinson, Geri Bares, Lorna Gerard, Lonna Bielke and Marsha Robbins. These are family and friends that have encouraged me to continue and have remained in my great circle of support. Finally, my greatest earthly support came from the values instilled by my aunt, Urseline Eugina Smith and my mother, Mercella Agatha Chambers. I was raised and nurtured by these two great women everyday of their lives until God called them home several years ago. My brother, Curtis Antonio Campbell (Bass), in his own way also inspired

iii me. Although he was mentally challenged, his smiles, his love of music and his unconditional love taught me that we do not have to be ‘Einsteins’ in order to be valued members of society. To God, I give all the praise and glory for putting these people in my life.

iv Abstract Routine restaurant inspections are important tools in food safety. Evidence of their importance is demonstrated in studies of the public’s expectation of frequent restaurant inspections and strict regulations that only allow restaurants to continue in operation after almost full compliance with food safety codes. Although foodborne outbreaks occur relatively infrequently, each event triggers further erosion of public confidence in foodsafety. Routine restaurant inspections is of immeasurable value, if regulators deliver uniform and consistent communication regarding restaurants’ food code compliance to the public and incorporate routine inspection results into a mechanism for predicting foodborne outbreaks. This research was conducted to examine three aspects of restaurant inspections where these objectives could be reached. These include analysis of the process to determine the effectiveness of restaurant inspections in capturing critical and minor violations; examination of methods of communicating results of the inspection to determine uniformity and consistency and analysis of the inspection to determine the predictability of restaurant inspections. There are very few studies that examine the restaurant inspection process or the consistency and uniformity of computing inspection scores; however, there are studies that fall on both sides of the debate of whether the inspection is predictive of an outbreak. This research revealed that the inspection process was valid in its effectiveness in capturing critical and major violations; that uniformity and consistency in reporting inspection results using letter grades occurred in approximately 56% of the reports, however, uniformity and

v consistency could be increased to approximately 84% by standardizing the reports and that the ability of routine restaurant inspections to predict outbreaks will likely depend on the nature of outbreaks. Keywords: routine restaurant inspection, standardized inspection reports, predict outbreak

vi Table of Contents Section 1 Acknowledgements………………………………………………………………… (i) Abstract…………………………………………………………………………… (iv) List of Tables………………………………………………………………………(ix) List of Figures………………………………………………………………………(x) Section 2 Chapter 1Introduction……………………………………………………………………1 Thesis Statement…………………………………………………………………………1 1.1 Food Safety in the United States…………………………………………………1 1.2 Number of meals served by restaurants………………………………………….1 1.3 Restaurant Inspections……………………………………………………………2 1.4 Frequency of Inspections………………………………………………………….3 1.5 The Restaurant Inspection Process………………………………..........................3 1.6 Communication of Restaurant Inspection Results………………………………...4

vii 1.7 Posting of Restaurant Inspection Scores…………………………………………..5 1.8 Extensiveness of Posting of Restaurant Inspection Scores………………………..5 1.9 Standardized Inspection versus Un-standardized Inspection Scores……………...5 1.10 Can restaurant inspections predict foodborne outbreaks? ....................................6 1.11 Hypothesis………………………………………………………………………..8 1.12 Specific aims to test my hypothesis include……………………………………..8 Chapter 2 Literature Review…………………………………………………………………..9 Chapter 3 Un-standardized Restaurant Inspection Reports Will Reduce Consistency and Uniformity in Scores and Letter Grades…………………………………………………….28 3.1 Introduction………………………………………………………………………28 3.2 Methods…………………………………………………………………………..30 3.3 Results……………………………………………………………………………32 3.4 Results ll…………………………………………………………………………32 3.5 Discussion and Conclusion………………………………………………………35 Chapter 4 The Impact of Standardization of Restaurant Inspection Reports on Posted Scores and Letter Grades…………………………………………………………………………….37

viii 4.1 Introduction………………………………………………………………………37 4.2 Materials and Methods…………………………………………………………...38 4.3 Results……………………………………………………………………………41 4.4 Discussion and Conclusions……………………………………………………..43 Chapter 5 The Ability of Restaurant Inspections to Predict Foodborne Outbreaks Likely Depends on the Nature of the Outbreak ……………………………………………………………….46 5.1 Introduction………………………………………………………………………46 5.2 Methods…………………………………………………………………………..48 5.3 Results……………………………………………………………………………50 5.4 Limitations……………………………………………………………………….54 5.5 Discussion and Conclusions……………………………………………………..55 Chapter 6 Thesis Defense……………………………………………………………………62 6.1 Limitations of the Study…………………………………………………………65 6.2 Future Researches……………………………………………………………….67 6.3 Conclusion………………………………………………………………………67 Bibliography…………………………………………………………………………………69 Appendix 1 Food Establishment Inspection Report ………………………………………..76

ix Appendix 2 Minnesota Food Establishment Form………………………………………….81

x List of Tables Table 3.1 Distribution of Violation Categories among Three Test Agencies………………..32 Table 3.2 Analytic Description of How Scores Match Using LA and NYC Grading Criteria……………………………………………………………………………………….34 Table 3.3 Summary of Grade Comparison between LA County and NYC Grading Criteria…................................................................................................................................35 Table 4.1 Summary of Status of Posting of Restaurant Inspection Reports in the U.S…………………………………………………………………………………………...41 Table 4.2 Summary of Grade Comparison between LA and NYC Grading Criteria……….42 Table 5.1 Distribution of Salmonella ser. Hvittingfoss cases………………………………..51 Table 5.2 Descriptive Statistics………………………………………………………………52 Table 5.3 (A) Restaurant Inspection Results during Year before Outbreak of Salmonellosis in a Restaurant Chain…………………………………………………………………………...53 Table 5.3 (B) Restaurant Inspection Results during Year before Outbreak of Salmonellosis in a Restaurant Chain…………………………………………………………………………..54 Table 5.4 Case Control studies of restaurant inspections to predict foodborne illnesses and their fit with the proposed models…………………………………………………………..61

xi List of Figures Figure 1 Model 1Foodborne Pathways Associated with Contaminated Raw Foods of Animal Origin………………………………………………………………………………………..57 Figure 2 Model 2 Foodborne Pathways Associated with Contaminated Fresh Fruits and Vegetables……………………………………………………………………………………58 Figure 3 Model 3 Foodborne Pathways Associated with Infected Food Workers………….59

1 Chapter 1 Introduction Thesis Statement: Restaurant Inspections are fundamental tools for foodsafety however there are areas that could be modified to improve its effectiveness. 1.1 Food Safety in the United States: The United States is said to have one of the best food safety systems in the world (USDA, 2009). However, food borne illnesses still occur. The Center for Disease Control and Prevention (CDC) recently re-calculated its estimates of foodborne illnesses in the United States as 48 million persons per year resulting in 128,000 cases that result in hospitalization and approximately 3,000 deaths. Food safety requires the cooperation of many agencies and regulations that ensure appropriate measures are in place to prevent foodborne illnesses and to provide the human and other resources necessary to investigate and contain outbreaks when they occur. The United States food safety system is based on science-driven principles that requires: (1) only safe and wholesome foods are marketed; (2) food safety regulations are science based; (3) the government is responsible for food safety regulations; (4) manufacturers, distributors, importers and others to comply with food safety regulations and are accountable if they do not; and (5) transparency of the regulatory system to the public (US Food and Drug Administration, 2003). 1.2 Number of meals served by restaurants: The very mobile and busy lifestyle of the American public relies on restaurants for a significant number of meals per week. Estimates vary depending on whether the study is directed at fast food meals or the age and gender included in the study. In one study using National Health Interview Survey (NHIS) data, the

2 estimates are that on average Americans eat approximately 2.8 meals per week in commercially prepared facilities (Kant & Graubard, 2004). In addition, persons 19-64 take an average of 1,875 trips that exceed 50 miles per year, while those that are over 65 take 220 such trips each year (Collia, Sharp, & Giesbrecht, 2003). This means that each year a significant number of individuals take their meals outside their political borders. 1.3 Restaurant Inspections: Keeping our food safe involves agencies at the federal, state and local level. At the federal level the US Food and Drug Administration (FDA) and the United States Department of Agriculture (USDA) share the primary responsibilities. One of the responsibilities of the FDA is the developing and updating of the food code. The Federal food code has a somewhat unusual role because it is only a recommendation. However, 49 of the 50 states and 52 of the 56 states and territories have adopted the food code that is legally enforceable (US Food and Drug Administration, 2011). There is some level of standardization; but there are modifications at the state or local levels to add food code or licensing issues that are unique to each agency. Restaurant inspection is one tool that has been used to ensure that commercially prepared foods are safe. States that have delegation agreements with the FDA are responsible for conducting restaurant inspections and enforcing the food code. There may also be delegation agreements between state and local agencies. In the event of foodborne outbreaks, there is usually collaboration between the state and the local agency, however, State Health Departments usually maintain primary responsibility for conducting foodborne investigations and for collecting and collating data for the entire state. Depending on the size of the local agency, e.g. at the counties or major cities, foodborne investigations may also be conducted at the local level with some

3 collaboration with the state. State, local and territorial health departments use a standard form to report results of foodborne investigations to the CDC. 1.4 Frequency of Restaurant Inspections: Frequency of inspections is usually set at the state level. However, the food code does recommend that frequency should be risk-based. Each state may set the frequency based on risk, but also on resources and priority setting of local city councils. One example of the variability in frequency of inspections is demonstrated by the difference in frequency of inspections between the state of Minnesota and the state of New York. Minnesota requires that High Risk Establishments are inspected at least once per year; Medium Risk Establishments are inspected at least once per 18 months and that Low Risk Establishments are inspected at least once per 24 months. New York City has a much more complicated system, where frequency of inspection is grade-dependent. An establishment that receives 0-13 points on an initial routine inspection is given an A and will be re-inspected in 12 months. One that receives ≥ 14 points during an initial inspection is not given a grade until a re-inspection is conducted. The re-inspection is conducted in a minimum of two weeks to allow for corrections. If the first re-inspection results in 14-27 points on its initial or re-inspection, it is inspected in 5 – 7 months after the re-inspection. One that receives ≥ 28 points on its initial or re-inspection receives re-inspection in 3-5 months after its inspection cycle ends which is defined as after the number of re-inspections needed for the restaurant to score ≤ 28 points. There are not many other agencies that have the financial and human resources to provide this level of inspection frequency. 1.5 The Restaurant Inspection Process: The Food Establishment Inspection Report Form 3-A developed by FDA, is the form that most states adopt either “as-is” or a modified version. The Form has 54 items that include such categories as Supervision; Employee

4 Health; Good Hygienic Practices; Control of Hands as a Vehicle of Contamination; Approved Source; Protection from Contamination; Potentially Hazardous Food Time/Temperature; Consumer Advisory; Highly Susceptible Populations; Chemical; Conformance with Approved Procedures; Safe Food and Water; Food Temperature Control; Food identification; Prevention of Food Contamination; Proper Use of Utensils; Utensils, Equipment and Vending and Physical Facilities. The form is a model for conducting routine, follow-up and investigative inspections. It serves as a model for providing the compliance status following an inspection. Traditionally, the inspection was a detailed examination of all items in the code with much attention paid to the physical structures. More recently, environmental health specialists are focusing on items that have been identified as significant contributors to foodborne illnesses. The challenge for this re-direction of focus is to ensure that there is adequate attention paid to the physical facilities while focusing on the critical violations. 1.6 Communication of Restaurant Inspection Results: The result of the restaurant inspection can be in a narrative form, scores or letter grades. Some agencies use a combination where the Food Establishment Report Form includes carry code items with a check mark denoting whether an item is in or out of compliance, not observed or not applicable to that food establishment. Each item may also have a demerit point that is deducted if the item is out of compliance. At the end of the inspection the demerit points are added and listed as a total demerit score. In this format a zero would be a perfect score and the larger the score, the greater the level of non-compliance. Scores may also be translated to percentages by subtracting the total demerits from 100 and presenting the results as percent. Scores may also be converted to letter grades by applying a curve to the total demerit or

5 percent scores. Whichever method is chosen, the food code recommends that a clear and succinct summary of the compliance status of the food establishment is delivered to the permit holder or the person in charge at the conclusion of the restaurant inspection. 1.7 Posting of Restaurant Inspection Scores: Increasingly agencies have been releasing the results of restaurant inspections to the public. For some, it might be a response to media scrutiny, for others it might be a desire to increase transparency or as an attempt to re-gain the public’s confidence in food safety following several foodborne outbreaks. In Los Angeles, the posting of scores was documented as resulting in fewer cases of hospitalization due to food-related illnesses (Simon et al., 2005). In general, posting of restaurant inspection scores is a tool used to gain a higher level of compliance as well as to provide information that the public may use when choosing a safe place to eat. 1.8 Extensiveness of Posting of Restaurant Inspection Scores: There has been a sense that the posting of restaurant inspection scores has been increasing, but there was no one study where, that had been done. For this reason, my research has included a survey of the states to provide an overview of the extent to which this practice exists. 1.9 Standardized Inspection versus Un-standardized Inspection Scores: The FDA has a voluntary standardization program that allows agencies to have regulatory personnel with standardized training in i) in interpretation of code violations, ii) in conducting restaurant inspections, iii) conducting interviews with restaurant management, iv) checking for conformance with a hazard analysis of critical control points (HACCP) plan and v) differentiating when an item is out of compliance versus not observed during an inspection. The standardization program does not include scoring or grading of inspections or

6 communicating the results to the public. In this research, the impact of non-standardized scores was researched and compared with scores resulting from standardized scores. 1.10 Can restaurant inspections predict foodborne outbreaks? The public has viewed restaurant inspections as the key to keeping our food safe. In one Tennessee study, over half of the public surveyed indicated an expectation of having restaurant inspections conducted at least monthly basis and over a third indicated that a score of ≥ 90% was an acceptable inspection score (Jones & Grimm, 2008). Restaurant inspections are held in high regard by the public. Researchers and some in the public have asked whether the restaurant inspection can be the ultimate tool to be used in predicting foodborne outbreaks. There is documentation on both sides of the debate. Some studies found that the routine restaurant inspection can be used to predict foodborne outbreaks and there are those that concluded that it does not predict foodborne outbreaks (Cruz, Katz, & Suarez, 2001; Irwin, Ballard, Grendon, & Kobayashi, 1989). Routine restaurant inspection may be done in several styles. The more traditional manner is for the inspector to do a complete walk through of the establishment and check to find code items that are out of compliance. Depending on the food code specific to each agency, there are commonly 47 to 54 code items. Items that have been historically viewed as more significant to foodborne illnesses have been weighted heavier but in this traditional manner of inspecting, aesthetic items such as disrepair in the physical facilities might have received most of the attention because they were more visible. At the same time the latter are more capital intensive and frequently, have evolved as chronic items. This was particularly true for smaller, single-owned restaurants, where the ability to provide ongoing maintenance and replacement of large items are sometimes challenging.

7 A more proactive type of restaurant inspection is the hazard analysis of critical control points (HACCP). This form of inspection is more of an interview that may require longer and more frequent visits. It requires more time than the previously discussed routine inspection. It is a review of each process required to prepare the food from when it enters the establishment until it is served to the guests. It seeks to identify areas in the process that allow the introduction of a pathogen, preventative measures that must be taken, a setting up of control limits and corrective measures that should be taken if those limits are exceeded. While the HACCP is usually developed by third parties, it will also involve agency personnel and will require the type of inspection that would seek documentation to verify that the HACCP plan is being implemented as developed. Yet another type of inspection is the active managerial control inspection. It, like the HACCP inspection is risk based. This usually involves a review of the flow of foods from its raw state through to when it is cooked and served or its incorporation with other foods. It may also include a third party as the lead in the development of the plan. In many respects, it is similar to the HACCP where processes are reviewed and points where there the likelihood for foodborne contamination are identified and preventive measures are put in place. The more recent form of the restaurant inspection consists of a visit from an environmental health specialist where there is a walk-through inspection; however, the main focus is on the critical factors associated with foodborne outbreaks. Food code items that are minor contributors are not ignored, but usually a general statement that addresses them collectively rather than individually and by their location within the facility is typical. The food code recommends that all inspections be risk-based even with the limited time that the inspectors have.

8 1.11 Hypothesis: The restaurant inspection can be an effective tool in improving food safety. 1.12 Specific aims to test my hypothesis include: 1. Determine whether the inspection process is effective in identifying critical food code violations as well as aesthetic or minor contributors to foodborne illnesses 2. Determine whether standardizing inspection reports will increase uniformity and consistency in communicating the results of restaurant inspections to the public 3. Determine the ability of the routine restaurant inspection to predict foodborne illnesses

9 Chapter 2 Literature Review There are not many published studies available on the routine restaurant inspection process independent of a foodborne outbreak. However, much has been written about the HACCP program and the HACCP inspection. The following five inspections describe certain perspectives and application of the program. The Unnevehr (Unnevehr & Jensen, 1999) study addresses much of the economics of implementing HACCP and its applicability to specific systems. It is both a primer for HACCP as well as a resource for companies that wish to implement the program. It is detailed in its planning, training, start-up and maintenance of a HACCP program. Unnevehr and Jensen outlined specific benefits of HACCP that fits very well into a discussion of conducting a restaurant inspection and focusing on the critical points for limiting introduction or amplification of a pathogen. The paper points out that HACCP is not a “stand alone” program. Other elements of an inspection such as good hygienic practices and food protection, equipment and facilities in maintained and in good sanitary conditions also are necessary elements of a successful HACCP program. The inspector’s role in identifying these support systems will increase the potential for success. The Unnevehr and Jensen (1999) paper observed that a HACCP program eliminates the “command and control” type of intervention and moves regulation into a more science based format. HACCP, and the active managerial control, (not covered in this paper) are two programs in which there is a dialog between the inspector and the restaurant management. Procedures are reviewed and shortcomings addressed. There is a sense of working together to improve food safety rather than the traditional approach of trying to catch food workers doing wrong. One limitation is

10 that implementing a HACCP program can be costly therefore; it is more likely to be implemented by larger firms than by smaller or “mom and pop” type facilities. While HACCP is applicable to many systems, when mandated to an entire industry, it needs to be accompanied by risk assessments in order to determine its effectiveness on public health. A practical application of HACCP is described by Mortimore (Mortimore, 2001). It stresses the importance of management’s commitment to training and providing adequate resources. This top down approach sets the tone and allows all employees in the restaurant to realize that food the importance of food safety. Mortimore (Mortimore, 2001) supports the position that the need to assess other systems should be in place to supplement a HACCP program. Together these papers suggest that inspections could be thorough, but should be kept simple (Mortimore, 2001; Unnevehr & Jensen, 1999). This is important piece of advice to the restaurant industry. The educational level of many of the Foodworkers who will be carrying out the day to day aspects of the program are sometimes relatively low and many times they may not have English as their first language. Keeping it simple ensures that training and instructions are understood and implemented fully. The authors acknowledge the limitation of the need for financial and human resources and that periodic audits and a maintenance program need to be in place. Those responsible for HACCP should have a good understanding of the raw materials used and the food chemistry involved when these raw materials are put together (Mortimore, 2001). A critique that spotlights and elaborates on some of the limitations in the theory and practice of food inspection in restaurants was published by Sperber (26). The study pointed out that there is a trend toward opaqueness and that for some applications, inspections lack the appropriate preventative steps that are essential to food safety. Abattoirs are important

11 examples. Conventional steps of heat treatment are not appropriate at the slaughter house therefore protection starts much farther down the chain. Although slaughtering does not normally occur in restaurants, this is an important issue in the restaurant food safety chain. An adulterated meat source may start a chain of events that could lead to a foodborne outbreak. Sperber (Sperber, 2005) pointed out that setting an acceptable performance standard for Escherichia coli (E. coli) at five or fewer positive samples in a test batch of 53 could be problematic. Standards are set at some point where laboratory or epidemiologic studies demonstrate some acceptable margin of safety exists. It is good practice to allow some buffer. However, wherever the limit is set, on either side of the range, it will be regulated as unsafe. A sharp line is needed so inspectors will have an objective limit that can be equally be applied to everyone. In HACCP programs there can be a gradient in the corrective measures applied based on how far from the acceptable limits the test results are. The same argument can be applied to other areas in food protection especially around holding temperatures. If a product is found at one degree out of the temperature range, the product could be reheated or flash cooled to quickly bring it within the safe zone. Applying this to the E. coli performance standard could be that at 6 positive samples in 53, immediate retesting might be the response, if a second round has the same or greater results, pathogen reduction strategies might be implemented. Other limitations addressed in the paper include the setting of different pathogen levels among species and the protocol of allowing up to four additional rounds of repeat samplings, once the first is found unacceptable. In essence this allows continuation of the “unsafe conditions” for up to two years. This is an opportunity where environmental health specialists could provide some input during the inspection to review pathogen reduction strategies that are in place and as mentioned earlier, the success of

12 the program will depend on the education, training and skills of the people involved. There were several other limitations to HACCP that the Spencer paper highlighted, including i) science-based performance is sometimes used to discredit other Pathogen Reduction strategies, ii) retail establishments producing fresh fruit and vegetable juices are exempt from pasteurization, iii) rule permits a 5 log reduction after cleaning and culling of fruits and vegetables for multiple processes if they all occur within the same plant and iv) the rule allows for product sampling when control measures are not applied before the end product. This translates into the rule allowing a 20 ml sample to be a representative sample of 1,000 gallons of juice produced over one week. Similar indictment against the EU and its precautionary principle in issuing a ban effective in 2006 against antibiotic growth promoters due to the Bovine encephalopathy and dioxin in poultry feed incidences (Cervantes, ACPV, & Watkinsville, 2005; Vandeplas et al., 2008). All three papers concluded that HACCP is an effective tool where there are economical constraints; that there is a need for training and education in the practical application of the program and that there might be standards and protocols which need to be revisited with flexibility to accommodate a broader definition of pathogen reduction strategies. Allowing exemptions for retail juice production might be a soft spot in food safety, but one of the reasons might be the relatively smaller quantities produced and the fast turnover time. Time as a public health control has been introduced into the holding of potentially hazardous foods and might be the reasoning for this exemption. Further studies might be indicated to determine pathogen loads along some time and temperature curve. The issue of restaurant inspections was assessed by Soriano (25). The first three papers were broad view of HAACP that could be applied to various food industries including

13 larger chain restaurants, but perhaps best suited to food processors. Soriano (25) examined the impact of HACCP on two popular meals in a University cafeteria in Valencia, Spain. The two meals studied were Spanish potato omelet (containing potatoes, eggs, oil, onion and salt) and the pork loin. Background sampling established the levels of E. coli, Salmonella, Staphylococcus aureus (S. aureus) and Lysteria monocytogenes (L. monocytogenes). HACCP accompanied with training of the food handlers was implemented and the samples retested to determine pathogen levels. Background levels were established for E. coli and coagulase positive S. aureus at 21.0% and 17.9%, respectively in the Spanish potato omelet and 20.0% and 23.2% respectively in the pork loin. After implementation of HACCP plus training, the levels were found to be 1% E. coli in the Spanish potato omelet and 2% coagulase positive S. aureus in the pork loin. There were no Clostridium or Lysteria detected before or after implementation of HACCP. This study highlights the effectiveness of HACCP and that it can apply to relatively smaller setting. The presence of coagulase positive S. aureus suggested that there were improper hygienic practices among food handlers as S. aureus is a common micro flora associated with the mouth, nose or infected wounds. Implementation of HACCP would highlight the need for good hygienic practices such as prohibiting eating, drinking or smoking while handling food and proper hand washing after these practices. The application of HACCP to a relatively small restaurant setting is good indication that HACCP is not only for large chain restaurants but also applicable in smaller settings. A study examining the application and observation of HACCP in traditional and non- traditional restaurants was performed by Sun and Ockerman (7). The settings were international and included five different countries or ethnic cultures. HACCP was

14 implemented in Pakistan to address high levels of Salmonella on vegetables. Cross- contamination from knives, cutting boards and food workers hands were suspected as the vehicles. The control measure was the implementation of storage at 4-5°C. This is an instinctive requirement in food safety for us in this United States. However in developing countries where resources are limited and a culture where refrigeration is not always available, it required insight and intuition to recognize that HACCP could be modified to improve food safety. In another instance HACCP was implemented in 16 school cafeterias in Bahrain. Control measures included cooling since there were no cold holding during deliveries; good hygienic practices by food handlers and proper cooking and holding temperatures. These measures resulted in reduction of E. coli, S. aureus and Salmonella sp. The paper indicated that sharing the results with the workers helps in implementation of this program. Several of the papers reviewed on HACCP demonstrated the flexibility of the program although it may require some insight as to how the program can be modified to be used in non-traditional settings. The papers were all in agreement that it is essential that everyone involved in the program shared in the vision and that training and education had a direct correlation to success. One of the more innovative adaption of HACCP is the implementation of the program to food hawkers in Malaya where it was found that educational levels, culture and environment affected Foodworkers attitude to food safety. A study in Ohio also confirmed that when safety training was added, restaurants had significantly higher scores. What is missing from published studies is the lack of involvement by environmental health specialists, plan reviewers or other agency personnel. Agency personnel should be

15 part of the process as management develops the plan and before implementation and for developing the appropriate inspection format. While HACCP programs and the resulting HACCP inspections are an effective form of restaurant inspections, HACCP is not the major type of restaurant inspections. As implied earlier, many small businesses do not implement a HACCP program. The active managerial control is another form of interacting comprehensively between food inspectors and the restaurant management. In this study (Reske, Jenkins, Fernandez, VanAmber, & Hedberg, 2007) conducted in Minneapolis, Minnesota the Minneapolis Environmental Health Division together with the University of Minnesota collaborated with food operators to better engage in foodborne illness prevention strategies. This two-year study examined the result of having announced inspections followed by announced inspections compared to unannounced inspections. For the announced inspections, the interviews were menu and process directed. The environmental health specialists focused on hazards specific to each establishment. Critical points were identified and preventative steps were reviewed and a routine walk through inspection done with management and the health inspector making the observations. The key is that it was interactive with managers working with the environmental health specialists in identifying key issues in food safety. After the announced inspections, an unannounced inspection was done within a year. The study found key areas of improvement in the demonstration of knowledge of foodborne disease prevention by the person in charge. These findings supported previous studies which found that the presence of a certified kitchen manager was a key element in the reduction of risk of foodborne illnesses (Hedberg et al., 2006). There was improvement in the prevention of cross- contamination an in the cleanliness of the floors walls and ceilings; reduction of violations on

16 non-food contact surfaces, food equipment and food protection during storage. These improvements were significant differences from the unannounced inspections. In addition, operators reported a better relationship with the environmental health specialists. Active managerial inspection is adaptable to both large and smaller establishments but it does require more time and resources from agencies. At the local and state levels, frequency of inspections usually is determined at administrative levels. Implementing programs such as the active managerial control may not allow for the agencies to meet these pre-determined goals. The active managerial approach allows the restaurant management and the health inspector to work together in approving food safety much as in the HACCP program, but it shifts much of the increase in resources to the agency. It deserves additional attention. Kovats (Kovats et al., 2004) reported on data from national surveillance centers of ten European countries. A correlation between ambient temperatures and the incidence of salmonella infections was observed. It appeared that the rise in salmonellosis was due to the increase of outdoor cooking and grilling of more meat and poultry which may have presented more opportunities for cross-contamination. Providing outdoor seating has been a feature that many restaurant goers appreciate as the weather warms up in the colder parts of the United States. Some restaurants have tried to expand on the idea by bringing some cooking outside as well. Prevention strategies usually begin at the time the restaurant seeks a permit for outdoor cooking. Some strategies for prevention could include HACCP or active managerial control as discussed earlier. Cross- contamination is one violation for which the environmental health specialist checks, however, because of the short time spent during an inspection, it may not be observed until it becomes a widespread system failure. A HACCP or active managerial inspection would provide a preventative approach.

17 Another item that is explicitly included in the food code and for which most inspectors are trained to observe is hand washing practices in the restaurant. Hand washing as a food safety strategy is applicable for most foodborne pathogens therefore the study conducted by the Environmental Health Specialist Network (EHS-Net) is relevant to the role of the inspection in detecting critical violations. This study (Green et al., 2006) included all six of the catchment states of Colorado, Connecticut, Georgia, Minnesota, Oregon and Tennessee. Observations were made of the number of times that food workers washed their hands after performing activities that necessitated hand washing. Credit was given to attempts at hand washing as well as to hand washing that were performed adequately. The study revealed that there were activities after which there should have been hand washing at a rate of 8.3 times per hour. Further, the study showed that workers performed hand washing in some form 32% of the time. This study was in contrast with a previous study, (Clayton & Griffith, 2004) that found the number of times that required hand washing was 17 times per hour. The difference between the two studies is likely due to the difference in the data collection methods. In the Clayton and Griffith (Clayton & Griffith, 2004) the study counting was extracted from a taped recording of the activities. A routine inspection of a large facility will take approximately 1.5 - 2 h. In the EHS-Net study, an hour was spent observing for only one activity. The potential for the inspection to reflect only a short moment in time is probable by methodology, because the inspector can only observe a limited number of activities. The number of observations is confined by the physical layout of the restaurant and the types of activities that are being performed during the inspection. The benefit of this study is that it presented a real time view of the inspection. The study highlighted the need for prevention to start at the plan reviewing phase because a better flow

18 control might reduce the level for hand washing generated in the food preparation. Highlighting the need for more appropriate and adequate hand washing will assist management in determining their training needs. Other studies have indicated that hand washing quality is improved when the person in charge has been trained or if there is a certified kitchen manager. In the Allwood 2004 (Allwood, Jenkins, Paulus, Johnson, & Hedberg, 2004) studies, environmental health specialists interviewed the persons in charge to determine their knowledge regarding hand washing. Food workers were selected to demonstrate how to wash their hands. In this study, there was a positive correlation between the level of knowledge of the person in charge and how well and appropriately foodworkers washed their hands. The study is yet another that reinforces the importance of having a certified kitchen manager or person in charge to raising the level of hygienic conditions in the restaurant. The health inspection of the restaurant is perceived as an important strategy in improving food safety. Minimizing the critical violations is crucial in preventing foodborne illnesses. HACCP and active managerial approaches are developed to prevent or minimize critical violations and therefore improve food safety. Although the focus is appropriately placed on the items that contribute to foodborne illnesses, cleanliness and good repair in the physical facilities cannot be overlooked. There is discussion among some restaurant owners and managers that these are aesthetic issues that should be left to the restaurant management. The argument is partly correct, but minor items may contribute indirectly to a food-related illness outbreak. An example of this may be a hole in the roof could allow insects, birds and rodents to enter the restaurant and resulting in the cross-contamination of foods. A major reason for environmental health specialists to pay some attention to the physical facilities is

19 that the public has no other evidence of how well the inspection is conducted except what is seen in the dining rooms, lobbies and bathrooms. In the Wetterling study (Wetterling, 2009), research was conducted in retail facilities in Germany and France where the impact of cleanliness of the physical facilities on customers were observed. The study found that customers may differ on what was aesthetically appealing in the display of products but they were unanimous in their distaste for any lack in cleanliness. When customers are so displeased by physical evidence that management is not meeting the mark, it is quite likely that they may assume that food safety is being short changed as well. The limitation of this study is that its setting was in a retail facility where there are many more items and areas that might enter into the picture than if it were in the dining room of a restaurant where conditions are more easily controlled. On the other hand, in the restaurant setting, a customer might have a higher expectation for cleanliness and an aesthetically appealing environment. These reviews represent an overview of several issues that are covered in restaurant inspections as well as the different types of food safety inspection that may be warranted based on the programs implemented in the restaurant. Some of the critical items such as hand washing and cross contamination are highlighted because they are significant contributors to a foodborne outbreak. The food code recommends that the compliance status of the restaurant be presented to the permit holder or person in charge in a concise and succinct manner soon after the restaurant inspection (Dundes, 2001). Many agencies use a score accompanied by a letter grade as a clear and concise means of communicating the level of compliance with the food code. It may be reasonable to assume that everyone understands that an A through D is equivalent to a passing score. However, while this may be understood in the college

20 environment, the same level of understanding may not be easily transferred to restaurant inspections. In one report (Dundes, 2001), undergraduate students and a group of environmental health professionals were asked to interpret a score of ‘86’ and a “C” as results of restaurant inspections. The college students were more likely to find a score of “86” as not acceptable and were less likely to eat at one graded as a “C”. There was some confusion among college students in their understanding of restaurant grades even though many had worked in restaurants. College students were more comfortable in eating at restaurant graded with a “C” than one receiving a score of “86”. Their take on grades for restaurants is that they should either be opened or closed. A majority (66%) indicated the signs meant nothing to them and would prefer that health inspectors make the decision as to whether the restaurant met acceptable levels. Environmental health professionals were more likely to eat at restaurants receiving both scores and appeared to better understand the grades. The confusion as to the choosing of letter grade of “C” could be due to undergraduates past experience in what is “passing”. Depending on the class, an “86” may range from an excellent to a poor score. The study underlines the confusion and the miscommunication of results of restaurant inspections when communicated in the format of scores and grades. The college students’ skepticism regarding scores and grades may be justified because it may be important to know the jurisdiction or location of the restaurant and how that agency scores or grades its inspections. Scores are released by agencies as a means to achieve and maintain food code compliance. The Dundes paper (8) indicates that some members of the public have difficulty deciphering the meaning of scores and letter grades. However agencies have found that posting of scores is a tool to boost food code compliance in their jurisdictions. Jin and Leslie

21 conducted research to determine the impact of posting of scores in Los Angeles County for the period of 1996 through 1998 (Jin & Leslie, 2003). The researchers tracked all inspection reports, revenue for all restaurants and foodborne admissions to hospitals. They determined that foodborne related hospitalizations were reduced by 5%; that hygiene conditions of the restaurants were increased and consumers became more sensitive to hygienic conditions in restaurants. The researchers determined that the posting of scores led to improvement in hygienic conditions and that a boost experienced in revenue could not be attributed solely to a shift in choosing restaurants with better scores. This research indicated that consumers can be receptive to the use of letter grades to choose a safer place to eat. What may not be clear is whether consumers were making restaurant choices based on the grades posted or whether the changes were due to changes made by management as a response to this new method of communicating to the public. Since there was also an indication of higher scores the researchers determined that the posting of scores did modify behavior among food handlers. It is unclear whether the behavior changes have been modified or sustained. Having an effective inspection program that communicates effectively to the public is essential to food safety. Although the incidence of foodborne illness outbreaks are relatively rare, when they do occur they take a devastating toll on the public confidence in food safety in restaurants. Public confidence is essential in public health in general but particularly so for food safety. If the routine restaurant inspection could be determined to be a tool that effectively predict food-related outbreaks more outbreaks could be averted. The following are studies conducted to determine the predictability of foodborne outbreaks by the routine restaurant inspection.

22 The Seattle-King County study (Irwin et al., 1989) is well-known for changing the focus of the discussion concerning the role that restaurant inspections play in food safety. Other studies (Allwood, Lee, & Borden-Glass, 1999) have linked variables such as frequency of inspections to better scores. The assumption has been that better scores indicated lower risk for acquiring foodborne illness at a restaurant with a passing score. However, studies (Cruz et al., 2001; Jones, Pavlin, LaFleur, Ingram, & Schaffner, 2004; Kalluri et al., 2003; Penman, Webb, Woernle, & Currier, 1996) have demonstrated that routine inspections that resulted in satisfactory inspection scores have been involved in foodborne outbreaks. In the Irwin retrospective study (13), 28 case restaurants that met their definition of permanent facilities with an outbreak between January 1, 1986 and March 31, 1987 were studied. Each case was matched with two non-outbreak restaurants based on location and routine inspection dates (±30 days). The study was conducted on 25 cases and 48 controls. The outbreaks had a range of one to six persons and a mean of 2.9 persons and the pathogens were unknown for most cases. The researchers determined that the routine inspection was predictive of the foodborne outbreak. The vehicle most commonly implicated was poultry and the contributing factors were improper food temperature controls for cooking, holding reheating, cooling and storage. Analysis of routine inspections indicated a significant lower total score for case restaurants compared to non-outbreak restaurants. Other contributing factors included the presence of any critical violations; the presence of any three individual violations and the duration of the inspection. These findings are in line with our proposed explanation of the types of outbreaks that the routine inspection will identify and fits well into the first type of scenario that may result in a foodborne outbreak. The food code was developed to prevent foodborne illnesses resulting from pathogens such as Clostridium

23 perfringens. As such violations such as temperature control and cross contamination are explicitly identifiable during the routine inspection. Outbreaks that result from a breakdown of hygienic conditions such as temperature, cross-contamination and other food protection and hygienic practices are likely to be predicted by the routine restaurant inspection. A study that examined two consecutive outbreaks in a bakery which occurred in Cardiff, Wales was prepared by Evans et al. (9). A second outbreak occurred although corrective steps were issued after inspections revealed cross-contamination in the use of custard mix after the first outbreak. The second outbreak involved inadequate cleaning of the decorating nozzles in the making of cream cakes. Cases were matched with two other persons matched by age and location and were either nominated by the cases or randomly selected from the local telephone directory. There were 32 cases and 59 controls in the first outbreak and 22 cases in addition to 22 controls for the second outbreak. In the first outbreak it was determined that although the custard mix was made from a mix and cultures from the leftover mix did not indicate any salmonella contamination, food handlers had prepared the mix using the same bowl that had contained shelled eggs and therefore cross contamination might have occurred. For the second outbreak, there was no laboratory evidence, but it was suspected that there might have been salmonella contamination from the first outbreak to cause another cross-contamination episode. Similar to the Irwin study (8), these outbreaks fit our model in that there were breakdown of systemic hygienic conditions that provided the introduction and amplification of pathogen and the resulting outbreak. Not all studies on the ability to predict foodborne illness indicate that the inspection predicted the outbreak. In a recent study it was determined that routine inspections did not predict foodborne outbreaks (Jones et al., 2004). This study involved 49 outbreaks that occurred in

24 Tennessee from 1999 to 2002. Mean inspection scores of the last inspections done before the outbreaks was 81.2 and the next previous was 81.6. These scores were not significantly different from the mean of all inspections in the study period. The researchers cited improper storage of toxic chemicals and improper hand washing techniques as the two most commonly cited violation in a small sample of outbreak restaurants. An examination of this study indicated that most of the inspections fit a Norovirus-like category or are ones that appear to be contaminated sources that fit the ‘pass-through’ model that we propose. For Norovirus outbreaks, food may act as a vehicle as well as through person to person contacts. An ill employee can individually deliver an infective dose of the virus without the presence of overall poor hygienic conditions in the restaurant. It is important to keep ill food workers away from food handling, food equipment or food contact surfaces. Although the food code prohibits employees with foodborne illness symptoms from having food contact, neither the Form 3-A, the model Food Establishment Inspection Report form nor the food establishment inspection report forms used in many agencies includes illness as an item for which the inspector checks. Including the illness log in the inspection would improve its implementation and further foodsafety in restaurants. In the Jones paper (Jones et al., 2004), several of the outbreaks were due to Salmonella sp. as the agent. The presence of Salmonella sp. can be in response to poor hygienic conditions as well as a onetime pass through that resulted from adulterated sources. In cases where the organism is amplified by systemic conditions in the restaurant, the routine inspection will be predictive of the outbreak. In cases where the adulterated sources have levels of the organism high enough to cause illness, the routine restaurant inspection will not

25 be predictive of the outbreak. This is especially true for pathogens such as Salmonella sp. where the infective dose is very low. A case control study by Cruz et al. (Cruz et al., 2001) was conducted in Miami Dade County and included outbreak restaurants from 1995. Each case restaurant was matched by random selections with two non-outbreak restaurants based o year and month of inspections. Only the last inspections before the outbreak were included in the analyses. There were 51 outbreak or case restaurants and 76 non-outbreak or control restaurants. There were no significant differences between the overall ratings of the case restaurants and the control restaurants although the case restaurants had a non-significant trend towards requiring earlier follow-up inspections than the control restaurants. Case restaurants were more likely to require longer duration of inspection time; had larger seating capacity and were 3 times as likely to be cited for the presence of insects and rodents. This study is very similar to the Irwin Seattle-King County study except that they are approximately ten years apart and although the scores from the Seattle-King County Study predicted the outbreak, the Miami Dade County did not predict an outbreak. The Miami Dade County study was relatively smaller than the Seattle King County study and there are not enough data available on the pathogens or vehicles for the outbreak for detailed examination. It is possible, however, to identify contributing factors in the outbreak such as length of the inspection and seating capacity, which are not items that are addressed in the food code and therefore would not be identified in routine inspections. The presence of insects and rodents is a food code item that should be identified in a routine restaurant inspection. One explanation for why the inspection totals may not be reflective of rodents may lie in how the item is weighted. Some agencies e.g. Los Angeles County, splits that item such that an active infestation is a critical

26 violation and weighted accordingly (6 demerit points) but an inactive infestation is a non- critical violation and is weighted the same as other non-critical items such as dirty floors or walls and ceilings. This scoring may allow for total scores that are not severe enough to predict an outbreak. It is difficult to interpret whether this format is applicable to the study since that information was not available. In a case study by Luby et al. (Luby & Jones, 1993), one restaurant had had a number of failing inspections. The foodborne outbreak that resulted from a number of food protection failures, such as lack of temperature control during storage and transportation and serving of unwholesome foods led to salmonellosis in over 800 persons. In this study the outbreak was predicted, though not prevented by routine restaurant inspections. This outbreak contained all the contributory factors for which the food code was developed to identify so the inspection identified and predicted the failures. Predicting does not equate to prevention. Although the environmental health specialists issued corrective orders, the restaurant chronically slipped back to its previous states of non-compliance. Additional measures are indicated for restaurants that chronically fail to maintain a compliance status. Requiring the restaurant to be closed until re-training of all food workers, kitchen manager certification, implementing active managerial or HACCP are possible approaches that may be necessary to avoid an outbreak in these cases. Penman et al. (Penman et al., 1996) conducted a case study that involved two outbreaks. In the first outbreak, although the restaurant had received good routine inspection scores an outbreak occurred. The study indicated 37 of 68 persons met the case definition of being ill. The pathogen was not identified and the suspected vehicle was potato salad. No employees were identified as being ill, however, the cook had contact with ill persons from a

27 nursing home in which she worked and ill children in her home. This scenario fits well into our model of foodborne illness where the overall condition of the restaurant does not present overall breakdown of hygienic conditions however, there may exist a brief breakdown of good hygienic practices by a single individual. The restaurant inspection will have difficulty in identifying these sporadic lapses. This supports the need for illness logs and that such logs should be expanded to include contacts of foodworkers with individuals showing symptoms of food-related illnesses. In the second outbreak reported in this study (21), the last inspection before the outbreak resulted in a failed score. The failure included poor general hygiene, lack of food protection during storage, and lack of thermometers. The pathogen was Bacillus cereus which resulted in 79 of 142 persons contacted meeting the case definition. As discussed in the Luby et al. (Luby et al., 1993) study, outbreaks that require amplification due to poor hygienic conditions within the restaurant are likely to be predicted by the routine restaurant inspection thus fitting our scenario for predictability. Shapiro et al. (Shapiro et al., 1999)conducted a case study where an outbreak occurred in three separate luncheons catered by the same restaurant and the vehicle was roast beef in all three incidences. The contributory factors included improper refrigerator temperature, inadequate internal temperature of potentially hazardous foods and cross- contamination. For the reasons discussed above, this outbreak also was predicted by the routine inspection.

28 Chapter 3 Un-standardized Restaurant Inspection Reports Will Reduce Consistency and Uniformity in Scores and Letter Grades. 3.1 Introduction Foodsafety is an integral part of public health. Every year approximately, one in six Americans develops a foodborne illness (Almanza, Ismail, & Mills, 2002), as well as management’s ability to designate adequate financial and human resources to provide maintenance in a timely manner. A more recent practice encourages the environmental health specialists to focus their attention on the critical food handling violations that have been identified as being significant risk factors for outbreaks of foodborne illnesses (Medeiros, Hillers, Kendall, & Mason, 2001). This allows health inspectors and kitchen managers to have meaningful discussions on how food must be handled to prevent food- related illnesses instead of reviewing a list of violations found. These risk-based inspections allow inspectors more time to do hazard analyses and assess critical control points, i.e. specific points where food safety protection steps, (e.g. temperature control) (Bryan, 1990), need to be incorporated in order to enhance food protection in the facility. This reinforces food safety knowledge among kitchen managers and encourages them to take an active role in ensuring the effectiveness of the controls (Unnevehr & Jensen, 1999). Results of how well restaurants are doing are frequently communicated through scores or grades to the management (Allwood, Jenkins, Paulus, Johnson, & Hedberg, 2004). Scores also serve as an indicator to environmental health specialists (Simon et al., 2005), to determine the level of follow-up actions needed by an establishment, i.e. an establishment

29 with lower scores may require more frequent re-inspections than an establishment with higher scores. Besides communicating to owners and management, inspection scores are increasingly being used as a tool by customers to help choose a place to eat. Theoretically, a good score, (i.e. fewer violations), will attract more customers and poor scores, (i.e. more violations), (Choi, Nelson, & Almanza, 2010), will attract fewer patrons at the respective establishments. The potential financial loss is thought to be a motivating factor for owners and managers to improve and maintain hygienic conditions and overall food safety, (Simon et al., 2005). This concept was supported a 13% reduction in foodborne-related hospitalizations in Los Angeles after they started posting restaurant inspection scores, (Simon et al., 2005). Also, an Indiana study showed improvement in restaurant scores, and decrease in consumer complaints after release of inspection scores in the media, (Almanza et al., 2002) Several other studies demonstrated the effect on restaurant performance by the releasing of scores to the public, e.g. (Almanza et al., 2002), (Boehnke & Graham, 2000), (NYC, 2010). Despite the evidence, the posting of scores to improve compliance and/foodsafety remains controversial. Some have argued that the frequency of inspections, (Allwood, Lee, & Borden-Glass, 1999) were more important than how the results are disseminated. Others argue that restaurant scores alone can be misleading if there is not public education and an inspection history to accompany the published scores or grades. Such a position is supported by the study done by Dundes, (Dundes, 2001) where college students and food-safety workers, (i.e.,

30 two populations that have experiences with grades, scores and food safety), gave varying interpretations of what is meant by a letter grade of “C” and a score of 86%. (Dundes, 2001) Regardless of which position one takes, the posting of scores is being used by many agencies to encourage and maintain compliance to the food code. For foodsafety to be maximized, several changes need to occur. Among others, the scores or grades being posted need to communicate a uniform and consistent message, i.e. the same score or letter grade should mean the same regardless of the location of the restaurant. For this to occur, agencies need to have a uniform code and a uniform and consistent manner in which to score and grade inspections. In this study we examined the impact of different grading criteria on the results of inspections from a group of restaurants conducted by three Minnesota agencies. We consecutively applied the grading criteria from Los Angeles and New York City. We then compared the results of the two sets of grades for consistency. 3.2 Methods We obtained restaurant inspection records from the Minnesota Department of Health and two local environmental health agencies. The target population included High Risk Food Establishments. Each establishment must have had inspections in 2008 and 2009. For this study, High Risk Food Establishments were defined as any food facility that stores, prepares, serves or holds food and through these activities and/or because of the volume have the potential to cause food-related illnesses. We excluded schools, daycares, senior cares and fast food establishments that had limited food preparation. We selected five percent of the eligible establishments from each agency. Establishments were selected using the online QuickCalc Random Number Generator. These inspections were not scored or graded and in each case, in addition to the Food

31 Establishment Report Form, the inspectors’ comments were reviewed to determine the score, based on the respective agency’s grading criteria. Inspections were taken from both years. For the first part of the study, each inspection violation was coded according to the following categories: 1. Risk Factor, (rf), which includes all the high risk violations on the Minnesota Food Establishment Report Form, nos.1-27. (See Appendix 2). 2. Good Retail Practices, (GRP, which includes all other violations on the Minnesota Food Establishment Report Form, nos. 28-54. (See appendix 2). 3. Physical Facilities, (PF), which is a subset of the GRP, nos 48-54. (See Appendix 2). We also recorded the total number of repeated violations in the respective categories. The data was done on Excel Spreadsheets and modified to be made ready for analyses, using the Data Analyses, (STATA), software program. For the second part of the study, based on LA’s, and NYC’s, grading criteria each inspection report was reviewed, and based on the summary page and comments by the inspectors, a violation score was assigned. LA County scores were assigned, using LA’s Retail Food Inspection Guide, (http://www.lapublichealth.org/eh/RFIG_Old_org/RET). 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.

32 We 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, (http://home2.nyc.gov/html/doh/downloads/pdf/notice-81-amend-1209.pdf ), outlines scoring based on violation points as follows: A = 0 –13, B = 14-27 and C = >27 violation points. 3.3 Results Among the 167 inspections, a total of 1056 violations were recorded. These included 523 (50%) Food Borne Risk Factors, 312 (30%), Good Retail Practices and 221 (21%) Physical Facilities violations (Table 3.1). There was significant correlation between Risk Factor violations and the Physical Facilities violation. For approximately every 2.4 Risk Factor Violations reported, there was one Physical Facilities Violation found (p <0.05). Table 3.1: Distribution of Violation Categories among Three Test Agencies Agency 1 Agency 2 Agency 3 No. (%) N0. (%) No. (%) Total RF 105 (50) 319 (54) 99 (39) 523 (50) GRP 62 (29) 156 (27) 94 (37) 312 (30) PF 44 (21) 113 (19) 64 (25) 221 (21) Total 211 (100) 588 (100) 257 (100) 1056 (100)

33 3.4 Results (ll) Overall, approximately 56% of the scores were the same using both grading criteria. However, this ranged from approximately 35% of the scores Agency 1, to 66% of the scores from Agency 3, ( Table 3.2) For scores that differed by grading criteria, LA County criteria consistently produced higher grades. Overall 42% of the combined scores that differed by at least one grade difference had a higher grade using the LA grading criteria. For 2% of inspections, LA grades were 2 grades higher and in only 1% of inspections were a higher grade assigned by NYC criteria.

34 Table 3.2: Analytic Description of How Scores Match Using LA and NYC Grading Criteria Agency Number LA Grade Inspection NYC Grade Total (X = RF) Higher (%) Scores the Higher (%) same 17 (100) Agency 1 (6.2) 11 (65) using both LA 0 102 (100) Agency 2 (3.1) 45 (40) and NYC 38 (100) Agency 3(2.6) 13 (34) Grading 167 (100) Total 69 (42) Criteria (%) 6 (35) 65 (58) 2 (2) 25 (66) 0 96 (56) 2 (2) There were 167 inspections scored and graded using the LA grading criteria. Of the 167, 99 received an A. When the NYC grading criteria was applied, 59 received an A, 37 were downgraded to a B, and 3 to C’s. Of the 167, 49 received a B. When the NYC grading criteria was applied, one was upgraded to an A, 19 remained as B’s and 29 were downgraded to C’s. Of the 167, 15 received a C. When the NYC grading was applied, none was upgraded to an A, one was upgraded to a B, and 14 remained as C. Overall, there were 60 A’s, 57 B’s and 50 C’s. There were four applications that resulted in a non-graded score following the LA grading criteria. These would have all received a grade C following the NYC’s grading criteria. None were upgraded when the NYC grading criteria was applied (Table 3.3). Tables 3.1 indicated that Agencies 1 and 2 are similar in the number of risk factor practices found while t Agency 3 had less of a focus on risk factors. However, Table 3.2 indicated that

35 Agencies 2 and 3 were similar in the average number of Risk Factor violations per inspection while Agency 1 had twice as many average Risk Factor violations per inspection. 3.5 Discussion and Conclusions Consistency and uniformity in posted scores will add to food safety since restaurants are responsible for most of the commercially prepared meals and since the average American now travels between 316 to 370 land miles each year, (Collia, Sharp, & Giesbrecht, 2003). A result of 2.4 Risk Factor violations compared to one Physical Facilities violation suggested that health inspectors are correctly prioritizing their inspection time on items which are the most likely contributory factors to foodborne illness outbreaks. Although agencies commonly use scores and letter grades to communicate the compliance status of restaurants to the public, we found that for un-standardized inspection

36 reports, there was only a 56% consistency in the imputed LA County and NYC scores. In 42% of the applications, LA County grading criteria resulted in higher grades while NYC grading criteria resulted in higher grades in only 2% of the applications. The NYC grading criteria appeared to be more sensitive to the number of critical violations found than were the LA County grading criteria, however, both inspection criteria will only identify items that are explicitly stated in the food code. Both grading criteria were effective at identifying risk factors as well as physical factors. The final scores, (and hence consistency and uniformity), were more influenced by the number of violations found than they were by variations in weighting of the violations. In conclusion, we found that the restaurant inspection process allows health inspectors to prioritize their attention on critical food handling practices. Overall, only 56% of inspections were assigned the same grade, however, this varied by a range of 34% - 65% discordance across the three test agencies. Although each agency’s grading criteria varied in how violations were weighted, the study indicated that scores/grades were most influenced by the number of violations found during the inspection and the amount of information provided. The study also highlighted that a determination of risk requires more information than a letter grade or score