Food poisoning provides one of the classic anecdotes of epidemiology—the potato salad that goes bad at the church picnic. Classic or not, foodborne illnesses (and infectious disease in general) had begun to lose the attention of epidemiologists by the 1970s. The illnesses inflicted by foodborne outbreaks were typically transient and affected small numbers of people. Our understanding of the causes of contamination and the management of outbreaks was thought to be complete. The challenge was gone, and intellectual excitement moved on to other areas.
Today, the potato salad story is a starkly outdated image for the emerging threats posed by foodborne illnesses. We find ourselves facing virulent strains of familiar pathogens, bizarrely novel pathogens such as the prion, with its long latency and deadly effects, and genetically modified food products. Furthermore, the geographic rules have changed. Contaminated food products from a single farm can be dispersed throughout a wide area with modern systems of transport, processing, and distribution. The epidemiology of foodborne illnesses encompasses a whole new set of challenges that are the topic of this commentary.
Perhaps the single most important development in foodborne illness has been the reemergence of infectious diseases. If epidemiologists in the 1970s had thought that the study of infectious disease was a thing of the past, we quickly discovered how wrong this was. By the mid-1980s, previously unrecognized pathogens were emerging, including the foodborne pathogens Campylobacter and Escherichia coli O157:H7. 1–3 The Shiga toxin of E. Coli 0157:H7 can produce hemolytic uremic syndrome, kidney damage, and death. This bacterium, found in cattle and their manure, passes easily to meat products as well as to farm produce that comes in contact with contaminated farm water or manure. As a further complication, this emergence of dangerous pathogens has taken place during a time when the number of immune-suppressed persons in the population has been increasing (through the spread of HIV, the rise of organ transplantation and its accompanying immunosuppressive therapy, and the extensive use of chemotherapy). As if this were not enough, there has been an increasing occurrence of antibiotic resistance among common (and previously innocuous) foodborne pathogens.
Other factors have contributed to the complexities of preventing foodborne illness. These include the globalization and centralization of food production, the vulnerability of the food supply to intentional contamination, and an increasing consumer dependence on prepared products and restaurant meals. Foods that in the past were grown, processed, and distributed locally are now being distributed throughout the country and abroad. Tracking food becomes a challenge. For the epidemiologist, it becomes even more daunting to identify outbreaks of illness that are scattered over time and across a large geographic area. Ironically, these trends are accompanied by rising expectations among consumers regarding the safety of their food supply.
The purpose of this commentary is two-fold. One purpose is to discuss the rapidly changing features of foodborne illnesses. The second is to suggest that the problems of foodborne illness are not unworthy of the skills of contemporary epidemiology. Food safety is a fundamental public health challenge that has become more, rather than less, challenging. To address this challenge, we need to draw on many streams of epidemiologic thinking, including both the practical aspects of traditional outbreak investigation and the sophisticated methods of infectious and chronic disease epidemiology, nutritional epidemiology, psychosocial epidemiology, and environmental epidemiology.
What Defines a Foodborne Illness?
One feature of foodborne illness is that it is defined conceptually by the route of transmission—a legacy of its roots in infectious disease. However, we cannot always operationally distinguish infections that are foodborne from those transmitted by water, pets, sexual activity, or other routes of exposure. Salmonellosis acquired by toddlers mouthing objects from the floor is not a foodborne illness, and campylobacteriosis acquired from drinking contaminated water is not foodborne, yet such cases would be included in counts of infections that are “commonly foodborne”.
Furthermore, foodborne illnesses are not restricted to acute illnesses. There may be long latency periods, such as there is with bovine spongiform encephalopathy (BSE). Finally, not all foodborne diseases are infectious. For example, there may be allergic reactions to genetically modified plant proteins, or poisoning from chemical contaminants in food.
In fact, the etiologic heterogeneity of “foodborne illness” makes it not entirely satisfactory as a disease category. This categorization does have value, however, from the point of view of intervention and prevention. For practical purposes, the logistics of maintaining a safe food supply require attention to all illnesses that may be acquired through consumption of food.
Burden of Illness—Present State of Knowledge
Foodborne illness is closely identified with gastrointestinal illnesses, and the latter often serve as indicators of the burden of foodborne illness. Attempts have been made to estimate foodborne illness through the use of such indicator outcomes as “intestinal infectious diseases,” self-reported diarrhea, or culture-confirmed cases of specific pathogens. Each approach (whether based on symptom or pathogen) is limited in its description of the complete picture of foodborne illness. When a case is defined by symptoms or laboratory culture, the case group will include nonfoodborne cases—thus overestimating the burden of foodborne illness. On the other hand, many cases of foodborne illness are not reported; patients may decide not to seek care, doctors may choose not to collect specimens or order culture tests, and laboratories may not culture routinely for the pathogen in question. Data derived from clinical or surveillance sources may thus greatly underestimate the actual number of cases.
Surveillance systems are in place for acute infections, although these are limited. The FoodNet Surveillance System was established in 1996 to measure the number of laboratory-cultured cases for seven common foodborne pathogens (Salmonella, Campylobacter jejuni, E. Coli 0157, Shigella, Listeria, Yersinia, and Vibrio) and two parasites (Cryptosporidium and Cyclospora). 4,5 A case is counted if it is confirmed by culture and reported to the Public Health Laboratory System in any of the FoodNet catchment areas (the entire states of Connecticut, Maryland, Oregon, and Minnesota, plus some counties in California, New York, Georgia, Colorado, and Tennessee). The FoodNet cases are assumed to be predominantly foodborne, although some may have been transmitted through drinking water, sexual activity, exposure to pets, etc. The proportion of nonfoodborne cases in FoodNet has not been estimated. The FoodNet group has conducted surveys to estimate the proportion of cases not captured by laboratory surveillance.
The first comprehensive effort to adjust for underreporting and for nonfoodborne causes was undertaken by Mead and colleagues. 6 They estimated cases, hospitalizations, and deaths for 19 bacterial pathogens and strains, five parasites, and four viral pathogens; they then estimated the proportions thought to be foodborne. They estimated that in the United States there are 76 million cases of foodborne illnesses each year. Of these, 13.8 million have a known etiology (ie, the pathogen was identified). These include an estimated 553 deaths each year attributable to foodborne Salmonella (nontyphoidal), 499 to Listeria monocytogenes, and 375 to Toxoplasma gondii (Table). This is a first step in attempting to quantify the burden of foodborne infectious illness. Obtaining more precise estimates will present a formidable undertaking. 7
The problem of definition goes beyond laboratory screening for common infectious agents. Food safety officials also need to be able to recognize unexpected events, such as those attributable to chemical contamination of food, long-term sequelae of infections, and reactions to genetically modified foods. These health events are usually not considered in estimates of the burden of foodborne disease, but may nonetheless consume public health and other resources. They have major consequences for industry and trade, and receive a great deal of publicity. A good example is the task of keeping the meat supply free of bovine spongiform encephalopathy (“mad cow” disease, thought to cause variant Creutzfeldt-Jacob disease or vCJD). In the United States, there have been strenuous efforts to sample and test animals, restrict trade, regulate animal feed, and seize animals to prevent any occurrence of this foodborne illness. This public health program—so far successful—may be taken for granted because its success is inconspicuous. It consumes public health resources without a change in current health indicators.
A comprehensive approach to foodborne illness requires us to assess not only the illnesses spread by food, but also the threats posed by potential illnesses. Only in this way can we measure our successes and the impact of dollars allocated to food safety. Epidemiologists will be key in refining surveillance systems to (1) include all foodborne conditions, (2) adjust for factors that cause underestimates, (3) separate foodborne cases from nonfoodborne cases, and (4) detect rare or unexpected events.
Identifying the Sources of Pathogens and Other Contaminants
There are two general approaches to the study of foodborne illness. One is used for an acute and specific outbreak, when quick intervention may prevent further illness. This requires classic outbreak investigatory skills to identify the pathogen and the vehicle and source of contamination, and to control the specific outbreak. The second approach is to generalize beyond a specific outbreak to the universe of foodborne illness and thus, identify measures that will lead to an overall lower number of outbreaks and an overall lower burden of foodborne illness.
Investigating an Outbreak
In an acute outbreak, the main goal is to identify the source of exposure to prevent additional people from becoming ill. The reconstructed cohort method has been the primary approach used, particularly for outbreaks in which all exposed persons could be identified readily (such as the church picnic scenario). In recent years, investigators have also begun to use the case-control study. Fonseca and Armenian 8 reported an increase in case-control studies to study outbreak investigations, from 0.2% in 1960 to 17% in 1980, and others have also described the growing application of case-control studies. 9–13
If the contamination originated on farms or during processing, steps can be taken to recall contaminated products from retailers and to issue warnings to consumers. However, this requires identification of the food product by specific information (such as brand, lot, processing date, and sell-by date). A large part of the outbreak investigation may be dedicated to acquiring this detailed information. Investigators have painstakingly traced illnesses in disparate geographic locations to single-source products, to processing plants, or even to individual farms from which they originated. For example, an outbreak of Shigella sonnei in restaurants was traced to one lettuce-shredding plant. 14 In another outbreak, 202 cases of hepatitis A were associated with fresh produce contaminated before distribution to restaurants. 15S. flexneri has been traced to salad prepared at a central commissary of a restaurant chain. 16 An outbreak of Salmonella was traced to cilantro provided by one farm;17 an outbreak of E. Coli O157:H7 was traced back to farms where 1% of the cattle tested positive for E. Coli O157:H7. 18
Pathogens can be introduced by food handlers through a broad range of improper handling practices. An outbreak of Salmonella enteritis infection was attributed to a single employee. 19 A restaurant outbreak of E. Coli O157:H7 was associated with inadequate cooking of hamburgers, 20 and restaurant food storage and cooking practices have been associated with Salmonella enteritis infections. 21 Home preparation can also be a danger; an outbreak of Salmonella typhimurium was associated with improper reheating of roast pork in individual homes. 22
The United States has experienced one reported outbreak of a foodborne illness caused by intentional contamination of food. 23 In 1984, members of a religious commune deliberately contaminated salad bars in 10 restaurants in Oregon, resulting in 751 reported cases of Salmonella gastroenteritis. The possibility of intentional contamination has grown over the past decade, and epidemiologists (along with the rest of the public health community) have recently intensified their awareness of this issue.
In each of the above instances of food poisoning, the recognition of the pathway leading to contamination allows public health officials to control the outbreak and prevent further occurrence of disease. The ideal way to confirm a foodborne pathway would be to identify a sample of the food in the exact condition as when it was consumed. Sometimes this is possible, but not often. A food sample might be found in the refrigerator or obtained from the same lot. Even in such cases, there is often some uncertainty about whether the affected persons actually consumed the particular lot, brand, or food item. More often, the food has been completely consumed or, if some remains, it is no longer at the temperature it was when eaten, causing pathogen levels to change.
Verification that the ill person was truly exposed is surprisingly difficult to obtain. Dose information is even more difficult. In the example of BSE and CJD, it cannot be demonstrated that people with new variant CJD actually consumed meat with BSE prions—only that they had a likelihood of exposure because, decades after consumption, the meat is not available for testing. Similarly, when people reported allergic reactions to products possibly containing genetically modified corn, verification of exposure was not possible because the products had been consumed. With wide distribution of foods, prolonged food storage, long disease latencies, and relatively low incidence rates in the exposed population, it becomes increasingly difficult to identify the source of contamination and to verify the exposure of affected persons. This is similar to the problems of environmental epidemiology, in which exposure information is often ecologic or nonspecific.
Reducing the Overall Burden of Foodborne Illness
Acute outbreaks demand immediate attention, whereas, in contrast, reduction of the overall levels of foodborne illness requires a comprehensive understanding of the contribution of many different factors to the aggregate group of foodborne cases. Several approaches have been taken to summarize information and to generalize beyond individual outbreaks. These include: (1) summaries across outbreak investigations, somewhat like a meta-analysis; (2) population-based case-control studies to identify behaviors, food groups, and other factors contributing to risk of foodborne illness; and (3) studies across restaurants and food service sites to identify environmental antecedents of improper food-handling practices.
Summarizing Across Outbreaks
Summarizing across outbreaks is an intuitively appealing approach to understanding the causes and consequences of foodborne outbreaks. Since 1974, the Centers for Disease Control and Prevention (CDC) has maintained a database describing reported foodborne outbreaks. 24 They define a foodborne outbreak as “two or more cases of a similar illness resulting from the ingestion of a common food”; 2751 such outbreaks were documented between 1993 and 1997. The database also provides information on the number of cases and deaths, month of occurrence, foods implicated, and place where the food was eaten. Importantly, it also contains information on outbreaks caused by contamination with chemicals, such as heavy metals and monosodium glutamate.
Summarization of such data across outbreak reports is analogous to meta-analysis of observational studies. However, data on outbreaks are not easily found. Outbreaks are infrequently published in peer-reviewed journals, perhaps because they are not always considered a scientific contribution. Other factors, such as the time demands of the investigator or an institution's aversion to negative publicity, may also discourage publication. States now use the Electronic Foodborne Outbreak Reporting System (EFORS) to report outbreaks to the CDC surveillance system, but information may be less complete than would be desired for a meta-analysis. Information about individual outbreaks is not generally available as a text document, and review or validation is minimal. Further, not all outbreaks are reported to authorities. State and local health departments vary in their response to outbreak reports and in the resources available for outbreak investigations, and this variation may contribute to differences in reporting. For example, between 1993 and 1997, Washington state reported twice as many outbreaks of foodborne illness as California, 24 despite the fact that Washington has one-sixth the population. The biases introduced by reporting differences have not been explored or documented. The field would clearly benefit from rigorous methodologic review of inconsistencies in surveillance and reporting.
Population-Based Case-Control Studies
Investigators are increasingly using population-based case-control studies to identify factors associated with the probability of becoming infected. In the population-based case-control study, individuals may have acquired their infections through various pathways, vehicles, and routes of transmission. This is in contrast to the use of case-control studies within an outbreak investigation; in this latter situation, one assumes a common source, pathway, and exposure. Examples of population-based case-control studies include studies of salmonellosis in Spanish children, E. Coli O157:H7 in the United States and in Belgium, campylobacteriosis in England, cholera in Ecuador, listeriosis in the United States, and toxoplasmosis in pregnant women in Europe and Scandinavia. 25–34
These studies compare cases with controls with respect to eating habits, recalled dietary intake, and other factors such as owning pets, visiting farms, or traveling abroad. This is a fascinating example of a concept taken from chronic disease epidemiology and applied to infectious diseases. Ironically, the approach of studying multiple risk factors in chronic disease epidemiology was originally viewed as inconsistent with the concepts of causality that had been developed in infectious disease epidemiology. The adaptation of the population-based case-control study to the study of heterogenous cases of foodborne illness may require more thought and attention to methodologic issues. Sophisticated statistical analysis is required to assess the interrelation among variables such as education and social class, potential risk factors such as food-handling practices or eating in restaurants, and accompanying risk factors for nonfoodborne acquisition of “commonly foodborne” pathogens (such as water supply or pet ownership). In particular, precise estimates of attributable risk associated with specific food groups or handling practices may require further work.
Environmental and Food Service Studies
The role of food handling introduces an additional layer of complexity in maintaining food safety. Proper food handling can eliminate risks associated with pathogens in some foods, whereas improper food handling can introduce pathogens not previously present, or promote growth of pathogens already present. The association of code violations or failed health inspections with disease outbreaks has not been simple to quantify. Outbreaks have been reported at frequently penalized restaurants, 35 and a case-control study in King County, Washington found a correlation between mean inspection score and reporting an outbreak. 36 A second case-control study in Miami-Dade County, Florida did not find routine restaurant inspection violations to be predictive of foodborne outbreaks except for one critical violation, evidence of vermin. 37 Epidemiologic research should be able to help identify practices that are most strongly predictive of food safety and thus provide a systematic basis for regulation and enforcement.
Another approach to the problem of poor food handling has been taken by the Environmental Health Specialists Network (EHS-Net), an initiative of the CDC. This is an effort to describe the events and conditions that encourage improper food handling practices (and outbreaks) at retail establishments. This reintroduces epidemiology to its old cousin, sanitarianism, to identify areas for intervention and prevention.
Prevention and Control
The food safety community has put forward the concept of the “farm-to-table continuum,” emphasizing the importance of all stages of food production, transportation, and consumption in assuring food safety. We know from experience with individual outbreaks that contamination or improper handling at any step along the way can result in illness or death. The epidemiology of foodborne illness, which has historically been involved in well contained individual outbreaks, now needs to integrate its work across many levels of food production, processing, and distribution.
Sound policies on prevention of foodborne illness depend increasingly on risk assessment. This is true for policies at all levels of government and at the World Health Organization. Although risk assessment for environmental exposures has been undertaken for almost 2 decades, risk assessment for food safety policy is more recent, and the focus is somewhat different. 38–40 A major part of environmental risk assessment is hazard identification or hazard characterization—an assessment of the causal relation between the exposure and diseases in question. In microbial food safety risk assessment, the causal role is known. The pathogen in question is by definition a cause of disease. The unknown is the dose-response relation, or a minimum “safe” dose. Human dose data are scarce. There is a wide range of potential doses that can be delivered with various foods, methods of preparation, cooking and storing temperatures, etc. Food safety risk assessment is often used to evaluate the impact of potential interventions, changes in production and processing, and regulations regarding food production and handling. Such models use the quantitative relation between dose and infection to estimate the number of cases potentially prevented by the proposed change in production practice or regulation.
Environmental risk assessment has produced an extensive literature over the past 2 decades regarding literature search strategies, meta-analyses, and quantitative tools for developing estimates from multiple data sources. 41–45 Much of this can be applied to the problems of microbial food safety risk assessment, with certain adaptations to address the specific issues of food safety risk. Epidemiologists can contribute greatly to this area.
A new area of concentration in epidemiology is emerging—a new epidemiology of foodborne illness. This is broader and more comprehensive than the outbreak investigations of the past. This new approach has its roots in the study of infectious disease but also draws on other branches of epidemiology. The extensive expertise of nutritional epidemiology in measuring dietary intake has been applied to some degree in the epidemiology of foodborne illness, but perhaps not as fully as possible. As in pharmacoepidemiology, the epidemiology of foodborne illness must sometimes include monitoring rare events and adverse reactions, as well as estimating the risk of such events with extremely limited data. The dividing line between chronic and infectious disease epidemiology has been eroding in all areas of epidemiology. Foodborne illness is one more example of the blurring of this distinction, as newly discovered agents carried in food are implicated in diseases of long latency.
One of the most informative areas of epidemiology for the study of foodborne illness may be environmental epidemiology. Food, like water and air, is essential and ubiquitous. The causal agents carried by food (as with food and water) can be invisible, low level, in flux over time, and difficult to reconstruct in retrospect. Exposures leading to disease may come from more than one source, and exposure to multiple contaminants can occur simultaneously. Unlike air and water, however, individuals play a sizable role in choosing and preparing their foods and in determining their exposures. Food itself carries considerable social and cultural meaning, which further complicates efforts to alter consumption patterns for the benefit of the consumer. Preventive programs need to incorporate information from psychosocial epidemiology in understanding and encouraging behavioral change. This may include modifying the public's taste for rare or undercooked meats, helping ethnic groups develop safer preparations for traditional recipes, and effectively training food workers in hygiene and food handling.
The resources presently devoted to the epidemiology of foodborne illness are scant by comparison with other areas of study. This may help explain the limited activities of epidemiologists in this area. However, it is also probably true that most epidemiologists do not fully appreciate the importance of food safety or the extent of the intellectual challenges it encompasses. The opportunities for epidemiology to contribute to this field are still unfolding.
1. Altekruse SF, Stern NJ, Fields PI, Swerdlow DL. Campylobacter jejuni–an emerging foodborne pathogen. Emerg Infect Dis 1999; 5: 28–35.
2. Foster EM. Historical overview of key issues in food safety. Emerg Infect Dis 1997; 3: 481–482.
3. Tauxe RV. Emerging foodborne diseases: an evolving public health challenge. Emerg Infect Dis 1997; 3: 425–434.
4. Centers for Disease Control and Prevention. Foodborne diseases active surveillance network. MMWR Morb Mortal Wkly Rep CDC
5. Wallace DJ, Van Gilder T, Shallow S, et al
. Incidence of foodborne illnesses reported by the foodborne diseases active surveillance network (FoodNet)-1997. FoodNet Working Group. J Food Prot 2000; 63: 807–809.
6. Mead PS, Slutsker L, Dietz V, et al
. Food-related illness and death in the United States. Emerg Infect Dis 1999; 5: 607–625.
7. Powell M, Ebel E, Schlosser W. Considering uncertainty in comparing the burden of illness due to foodborne microbial pathogens. Int J Food Microbiol 2001; 69: 209–215.
8. Fonseca M, Armenian H. Use of the case-control method in outbreak investigations. Am J Epidemiol 1991; 133: 748–752.
9. Brownson RC. Outbreak and cluster investigations. In: Brownson RC, Petitti DB, eds. Applied Epidemiology: Theory to Practice. New York: Oxford University Press, 1998; 71–103.
10. Dwyer D, Strickler H, Goodman R, Armenian H. Use of case-control studies in outbreak investigations. Epidemiol Rev 1994; 16: 109–123.
11. Gregg M. Conducting a field investigation. In: Gregg M, Dicker R, Goodman R, eds. Field Epidemiology. New York: Oxford University Press, 1996; 44–59.
12. Hubert B, Rufat P. [Case control studies in investigation of food-borne infection outbreaks. Study of their utilization in France]. Rev Epidemiol Sante Publique 1992; 40: 156–163.
13. Petersen KE, James WO. Agents, vehicles, and causal inference in bacterial foodborne disease outbreaks: 82 reports (1986–1995). J Am Vet Med Assoc 1998; 212: 1874–1881.
14. Davis H, Taylor JP, Perdue JN, et al
. A shigellosis outbreak traced to commercially distributed shredded lettuce. Am J Epidemiol 1988; 128: 1312–1321.
15. Rosenblum LS, Mirkin IR, Allen DT, Safford S, Hadler SC. A multifocal outbreak of hepatitis A traced to commercially distributed lettuce. Am J Public Health 1990; 80: 1075–1079.
16. Dunn RA, Hall WN, Altamirano JV, Dietrich SE, Robinson-Dunn B, Johnson DR. Outbreak of Shigella flexneri linked to salad prepared at a central commissary in Michigan. Public Health Rep 1995; 110: 580–586.
17. Campbell JV, Mohle-Boetani J, Reporter R, et al
. An outbreak of Salmonella serotype Thompson associated with fresh cilantro. J Infect Dis 2001; 183: 984–987.
18. Ostroff SM, Griffin PM, Tauxe RV, et al
. A statewide outbreak of Escherichia coli 0157: H7 infections in Washington State. Am J Epidemiol 1990; 132: 239–247.
19. Hedberg CW, White KE, Johnson JA, et al
. An outbreak of Salmonella enteritidis infection at a fast-food restaurant: implications for foodhandler-associated transmission. J Infect Dis 1991; 164: 1135–1140.
20. Bell BP, Goldoft M, Griffin PM, et al
. A multistate outbreak of Escherichia coli O157: H7-associated bloody diarrhea and hemolytic uremic syndrome from hamburgers. The Washington experience. JAMA 1994; 272: 1349–1353.
21. McNeil MM, Sweat LB, Carter SL Jr, et al
. A Mexican restaurant-associated outbreak of Salmonella Enteritidis type 34 infections traced to a contaminated egg farm. Epidemiol Infect 1999; 122: 209–215.
22. Gessner BD, Beller M. Protective effect of conventional cooking versus use of microwave ovens in an outbreak of salmonellosis. Am J Epidemiol 1994; 139: 903–909.
23. Torok TJ, Tauxe RV, Wise RP, et al
. A large community outbreak of salmonellosis caused by intentional contamination of restaurant salad bars. JAMA 1997; 278: 389–395.
24. Centers for Disease Control and Prevention. CDC Surveillance Summaries. MMWR Mor Mortal Wkly Rep CDC Surveill Summ
, 2000:49(No. SS-1).
25. Mead PS, Finelli L, Lambert-Fair MA, et al. Risk factors for sporadic infection with Escherichia coli O157: H7. Arch Intern Med 1997; 157: 204–208.
26. Slutsker L, Ries AA, Maloney K, Wells JG, Greene KD, Griffin PM. A nationwide case-control study of Escherichia coli O157: H7 infection in the United States. J Infect Dis 1998; 177: 962–966.
27. Weber JT, Mintz ED, Canizares R, et al
. Epidemic cholera in Ecuador: multidrug-resistance and transmission by water and seafood. Epidemiol Infect 1994; 112: 1–11.
28. Baril L, Ancelle T, Goulet V, Thulliez P, Tirard-Fleury V, Carme B. Risk factors for Toxoplasma infection in pregnancy: a case-control study in France. Scand J Infect Dis 1999; 31: 305–309.
29. Bellido Blasco JB, Gonzalez Cano JM, Galiano JV, Bernat S, Arnedo A, Gonzalez Moran F. [Factors associated with sporadic cases of salmonellosis in 1- to 7-year-old children. Study of cases and controls]. Gac Sanit 1998; 12: 118–125.
30. Cook AJ, Gilbert RE, Buffolano W, et al
. Sources of toxoplasma infection in pregnant women: European multicentre case-control study. European Research Network on Congenital Toxoplasmosis. BMJ 2000; 321: 142–147.
31. Pierard D, Crowcroft N, De Bock S, et al
. A case-control study of sporadic infection with O157 and non-O157 verocytotoxin-producing Escherichia coli. Epidemiol Infect 1999; 122: 359–365.
32. Rodrigues LC, Cowden JM, Wheeler JG, et al
. The study of infectious intestinal disease in England: risk factors for cases of infectious intestinal disease with Campylobacter jejuni infection. Epidemiol Infect 2001; 127: 185–193.
33. Schuchat A, Deaver KA, Wenger JD, et al
. Role of foods in sporadic listeriosis. I. Case-control study of dietary risk factors. The Listeria Study Group. JAMA 1992; 267: 2041–2045.
34. Kapperud G, Jenum PA, Stray-Pedersen B, Melby KK, Eskild A, Eng J. Risk factors for Toxoplasma gondii infection in pregnancy. Results of a prospective case-control study in Norway. Am J Epidemiol 1996; 144: 405–412.
35. Kassa H. An outbreak of Norwalk-like viral gastroenteritis in a frequently penalized food service operation: a case for mandatory training of food handlers in safety and hygiene. J Environ Health
2001;64:9–12, 33, quiz 37–38.
36. Irwin K, Ballard J, Grendon J, Kobayashi J. Results of routine restaurant inspections can predict outbreaks of foodborne illness: the Seattle-King County experience. Am J Public Health 1989; 79: 586–590.
37. Cruz MA, Katz DJ, Suarez JA. An assessment of the ability of routine restaurant inspections to predict food-borne outbreaks in Miami-Dade County, Florida. Am J Public Health 2001; 91: 821–823.
38. Lammerding AM, Paoli GM. Quantitative risk assessment. An emerging tool for emerging foodborne pathogens. Emerg Infect Dis 1997; 3: 483–487.
39. Samet JM, Schnatter R, Gibb H. Epidemiology and Risk Assessment (invited commentary). Am J Epidemiol 1998; 148: 929–936.
40. Samet JM, Burke TA. Epidemiology and risk assessment. In: Brownson RC, Petitti DB, eds. Applied Epidemiology: Theory to Practice. New York: Oxford University Press, 1998.
41. Evaluation and use of epidemiological evidence for environmental health risk assessment. WHO guideline document. Environ Health Perspect 2000; 108: 997–1002.
42. Berlin JA, Colditz GA. The role of meta-analysis in the regulatory process for foods, drugs and devices. JAMA 1999; 281: 830–834.
43. Federal Focus. Epidemiology in Hazard and Risk Assessment. Washington DC: Federal Focus, Inc., 1999.
44. Graham JD, ed. The Role of Epidemiology in Regulatory Risk Assessment. Amsterdam: Elsevier Science, 1995.
45. Stroup DF, Berlin JA, Morton SC, et al
. Meta-analysis of observational studies in epidemiology. JAMA 2000; 283: 2008–2012.