Active Surveillance

FoodNet’s core activity is active surveillance for laboratory-confirmed bacterial infection caused by Campylobacter, Listeria, Salmonella, Shiga toxin–producing Escherichia coli (STEC) O157 and non-O157, Shigella, Vibrio, and Yersinia and parasitic infection caused by Cryptosporidium and Cyclospora. FoodNet does not track agents for which clinical laboratories do not routinely test (e.g., norovirus). As laboratory practices change, the surveillance system adapts. During the 2000s, culture-independent diagnostic tests (CIDTs) became commercially available to detect Shiga toxin. Recognizing that some laboratories might stop culturing for STEC O157, leading to a perceived decline in infections, FoodNet began surveillance for hemolytic uremic syndrome (HUS) as another way to track STEC O157 infections. Most cases of HUS are caused by STEC O157. As more clinical laboratories began using CIDTs to detect other pathogens, FoodNet responded by gathering data on laboratory practices and on pathogens detected by these tests and by encouraging reflex culturing of specimens that test positive (4).

FoodNet staff at each site receive reports of every identification of a pathogen under surveillance from clinical laboratories that conduct tests on patients’ specimens ordered by health care providers. They conduct periodic audits to ensure that all pathogens identified are reported (3). In 1999, to ensure the validity of data summarized across all sites, FoodNet developed and began tracking metrics related to reporting, a process unusual for CDC programs at that time. For each person with an infection, FoodNet staff collect demographic information and determine whether the person was hospitalized and whether he or she survived. In 2004, FoodNet began collecting data on whether the infection was part of an outbreak and whether the patient had traveled internationally.

FoodNet has the flexibility to conduct special surveillance projects in response to concerns about emerging infectious diseases, as when variant Creutzfeldt-Jakob disease emerged (3). In 2010, FoodNet conducted surveillance for Cronobacter sakazakii infections and found infection in all age groups, the highest rate of invasive infections in infants, and data suggesting that urine might be a more common site of infection than previously thought (5).

Tracking Incidence and Changes Over Time

FoodNet tracks the incidence of infections and HUS to assess the effectiveness of measures aimed at preventing infections and to monitor progress toward national health goals. To measure changes over time and minimize the spurious effect of annual fluctuations, FoodNet has used 2 baseline periods of 3 consecutive years each. The first, 1996–1998, is the initial 3 years of surveillance; the second, 2006–2008, was used to develop the US Department of Health and Human Services Healthy People 2020 goals (6). In 2008, FoodNet began also reporting changes from the average annual incidence for the 3 years preceding the year of the report. In 2012, FoodNet began reporting a measure of overall change in the incidence of bacterial foodborne illness. This measure combines data for infections caused by the 6 bacterial pathogens monitored by the network for which >50% of illnesses are estimated to be transmitted by food (7). To account for variations in the surveillance area, FoodNet uses a main-effects log-linear Poisson (negative binomial) regression model to assess changes in incidence rates (1).

Figure 2

Figure 2. Relative rates of culture-confirmed infections with Campylobacter, Escherichia coli O157, Listeria, Salmonella, Vibrio, and Yersinia compared with 1996–1998 rates, Foodborne Diseases Active Surveillance Network, United States, 1996–2014. The position of each...

Each spring, FoodNet summarizes preliminary data and changes in incidence for the preceding year (Figure 2) in CDC’s Morbidity and Mortality Weekly Report. Public health officials, regulatory agencies, industry, and consumer groups use these data to assess the effect of food safety interventions (3). FoodNet has documented significant decreases in the incidence of E. coli O157 infections since 1996–1998 and in HUS since 2001, supporting other data indicating that regulatory and industry actions have made ground beef safer (8). FoodNet also has documented lack of significant change in the overall incidence of Salmonella infections and marked changes in some specific serotypes, indicating that efforts targeting specific serotypes are needed to decrease Salmonella infections. In response to these findings and to recent outbreaks, FSIS created performance standards mandating the upper limit of allowable Salmonella contamination of chicken parts (8). Poultry is also a major source of Campylobacter infections (9). In 2011, in response to FoodNet data showing little progress in reducing these infections, FSIS issued the first performance standards that limited the allowable contamination of chicken and turkey with Campylobacter (10).

FoodNet data are used to guide the development of and monitor progress toward national goals and health objectives, such as the US Department of Health and Human Services’ high priority goal to reduce S. enterica serotype Enteritidis infections from eggs after implementation of the Egg Safety Rule that was passed in 2009 (6). FoodNet data also are used to monitor progress on 7 illnesses included in the Healthy People National Health Objectives.

Determining Sources and Outcomes of Infections

Understanding sources and settings of illnesses informs the development of recommendations, regulations, and interventions to reduce illnesses. FoodNet collects data to determine the relative importance of various routes of infection, including nonfood sources. Kendall et al. reported that, with wide variation by pathogen, 13% of persons infected with a pathogen monitored by FoodNet had recently traveled internationally; the authors identified regions to which travel carries the highest risk for illness (11). Hale et al. used data from FoodNet and other sources to estimate that 13% of illnesses caused by 7 enteric pathogens were attributable to contact with animals and their environments (12).

FoodNet has identified numerous risky food, environment, and animal exposures through case–control studies of sporadic Campylobacter, Cryptosporidium, Listeria, Salmonella, and STEC O157 infections, described in 19 journal articles (more information available at http://www.cdc.gov/FoodNet). It is currently conducting a case–control study of non-O157 STEC infections to assess risk factors and correlate virulence factors with symptoms. These studies have yielded rich data of long-term value. When the FDA needed data on sources of S. enterica serotype Enteritidis illnesses before the Egg Rule was implemented, Gu et al. reanalyzed data from an old FoodNet case–control study with a new method and determined that egg-related exposures had the highest attributable fraction (13). A case–control study of Listeria infections showed an unexpected association with eating melons (14). In response, CDC modified the Listeria Initiative questionnaire used to interview patients with Listeria infection. As a result, when Colorado detected a large Listeria outbreak, investigators already had information about cantaloupe consumption from many patients, and the melons were more quickly implicated and removed from the market (15). Case–control studies are resource-intensive and so are conducted infrequently. In 2014 FoodNet began routinely collecting exposure data from patients with some Salmonella infections and is exploring ways to use these data in models that attribute illnesses to sources.

Population and Laboratory Surveys

FoodNet conducted 5 population surveys beginning in 1996, with only a 2-year gap before the last survey ended in 2007. In addition to obtaining data for estimating illnesses (described in the next section), the surveys asked participants how recently they ate selected foods. These data have been used for many analyses. Shiferaw et al. found a higher proportion of men reported eating pink hamburger and runny eggs, whereas a higher proportion of women ate fruits and vegetables (16). The population surveys have been used frequently in outbreak investigations. Epidemiologists compare frequencies of specific exposures reported by outbreak patients with those of a comparable population in the survey to quickly generate, confirm, or refute hypotheses about sources of illness. The ready availability of these data saves time over traditional methods of finding controls (e.g., by random-digit telephone calls, followed by interviews of persons reached who agree to participate) (17). During a 2012–2013 multistate outbreak of S. enterica serotype Heidelberg infections, 79% of patients reported eating chicken in the week before illness began, significantly higher than the 65% reported in the 2006–2007 FoodNet population survey (18). That finding, with other epidemiologic, laboratory, and traceback findings, helped link the outbreak to chicken from 1 producer. The population surveys also have provided a platform for obtaining information quickly in a crisis. When bovine spongiform encephalopathy emerged as a public health concern during the mid-2000s, questions about hunting practices, eating venison, and travel to countries in which bovine spongiform encephalopathy had been reported in animals were added to the 2006–2007 survey (19).

FoodNet conducts surveys of clinical laboratories to determine practices. By analyzing data from surveys conducted in 1995, 1997, and 2000, Voetsch et al. determined that variations in laboratory practice by site might explain some of the observed differences in the incidence of STEC O157 infection (20). A survey in 2005 found that adherence to recommendations for isolation and identification of Campylobacter varied substantially among laboratories (21). A survey in 2007 showed that most laboratories complied with recommendations for testing STEC O157 but not with recommendations for non-O157 STEC (22). The laboratory surveys provided essential information for estimating the true number of enteric infections (23). In 2012, because of rapidly changing clinical laboratory practices, FoodNet began conducting a survey annually. The 2014 survey showed that CIDT methods were used most often to detect Campylobacter and STEC (4).

Estimating Actual Foodborne and Acute Gastrointestinal Illnesses, Hospitalizations, and Deaths

FoodNet data are central to estimating the numbers of US foodborne illnesses, hospitalizations, and deaths (23,24). Regulatory agencies and lawmakers use these estimates to help decide how to allocate resources for prevention. Mead et al. published the first estimates in 1999 in Emerging Infectious Diseases (24) using early active surveillance data from FoodNet and data from other sources. By 2010, this article was the most frequently cited article published in this journal. After these estimates were published, FoodNet began addressing data gaps and developing improved methods, resulting in revised comprehensive estimates published by Scallan et al. in 2011 (23). Major improvements resulted from the availability of data from >5 times more respondents to population surveys and more detailed information about illnesses reported in those surveys. For both the 1999 and the 2011 estimates, these surveys provided key data on the severity of illnesses, medical care–seeking behavior, and specimen submission. These data and data from FoodNet surveys of laboratories were used to estimate the total number of illnesses for every reported laboratory-confirmed illness of each pathogen. The surveys also provided essential information about the rate of acute gastroenteritis illnesses, which was used to estimate illnesses caused by viral pathogens and by unknown agents (25). An important advancement in the 2011 article was separate estimation of the numbers of illnesses acquired domestically and during international travel, enabled by FoodNet’s collection of information about recent international travel. These new foodborne illness estimates formed the basis for a ground-breaking analysis estimating the number of illnesses attributed to specific food categories (26).

Other Contributions

Linking FoodNet to the National Antimicrobial Resistance Monitoring System (NARMS) has expanded the impact of both surveillance systems. A FoodNet case–control study linked fluoroquinolone-resistant Campylobacter infections with eating poultry at a commercial establishment and with international travel (27). Data from this study contributed to the body of evidence that led FDA to withdraw approval for the use of fluoroquinolones in poultry (28). Krueger et al. conducted a joint FoodNet–NARMS study that showed bloodstream infection was more common among patients infected with resistant than susceptible Salmonella strains (29). By linking FoodNet and NARMS data, Shiferaw et al. found that Shigella isolates from Hispanics and recent international travelers were more likely than other isolates to be resistant to trimethoprim/sulfamethoxazole (30). In another study linking these 2 databases, O’Donnell et al. found that two thirds of persons with S. enterica serotype Enteritidis infections resistant to nalidixic acid had recently traveled internationally (31).

The ability to geocode FoodNet surveillance data and link it to census data has increased FoodNet’s ability to examine health disparities. By geocoding Campylobacter cases and linking to census tract socioeconomic status (SES) measures, Bemis et al. found the incidence of campylobacteriosis in Connecticut increased as neighborhood SES increased except among children <10 years old, for whom incidence increased as SES decreased (32).

The benefits of FoodNet for public health are far-reaching. As experts in surveillance, FoodNet site epidemiologists are often leaders in conducting multistate outbreak investigations, many of which result in industry or regulatory changes that make food safer. Examples include an outbreak of Salmonella infections linked to pot pies (33), an outbreak of E. coli O157 infections linked to spinach (34), and the outbreak of Listeria infections linked to cantaloupe (15). FoodNet site personnel also train local public health nurses, epidemiologists, sanitarians, and laboratorians about foodborne disease surveillance, outbreak detection, investigation, and response. Training helps local public health professionals recognize outbreaks and maintain skills and knowledge needed to respond appropriately.

FoodNet’s influence reaches beyond the United States. Australia’s OZFoodNet and FoodNet-Canada have modeled some of their activities on FoodNet surveillance, and FoodNet staff have collaborated with scientists from other countries to compare the prevalence of diarrheal illness (35). FoodNet scientists have been active in the World Health Organization Global Foodborne Infections Network, which works to enhance the capacity of countries to detect, respond to, and prevent foodborne and other enteric infections.