Salmonella Enteritidis in Broiler Chickens, United States, 2000–2005

TOC summary line: Greater sampling and monitoring efforts are needed to reverse a significant increase in prevalence.

D uring the 1990s, Salmonella enterica serotype Enteritidis briefly surpassed S. Typhimurium as the predominant Salmonella serotype isolated from humans in the United States (1). Eggs were frequently implicated as the cause of outbreaks of human infection (2,3), and the outbreak strain was often detected in the implicated egg production flock (4). After egg producers implemented quality assurance programs in the late 1990s, human Salmonella Enteritidis infection rates decreased by ≈50% (1).
Recently, 2 US case-control studies in Foodborne Diseases Active Surveillance Network (FoodNet) sites identified eating chicken as a risk factor for sporadic human Salmonella Enteritidis infection (5,6), replicating findings of a case-control study performed in England in the late 1980s (7). While the overall incidence of human salmonellosis in FoodNet sites was lower in 2005 than in the mid-1990s, the incidence of Salmonella Enteritidis infections was ≈25% higher (8). We present US Department of Agriculture (USDA) Food Safety and Inspection Service (FSIS) Salmonella testing program data collected from 2000 to 2005 that suggest a need for interventions to prevent the emergence of this Salmonella serotype in broiler chickens in the United States.

FSIS Salmonella Testing Program
As of January 2000, an FSIS performance standard for Salmonella was set for all establishments that slaughter US broiler chickens (9). Establishments that slaughtered >20,000 chickens per year were eligible for FSIS regulatory Salmonella testing. These establishments accounted for >95% of raw poultry marketed in the United States.
The sampling frame for the present study included all eligible FSIS-inspected establishments. Each month, eligible facilities were randomly selected for Salmonella testing to begin in the following month. In each broiler slaughter setting that was tested, 1 broiler chicken carcass rinse (hereafter referred to as broiler rinse) was collected per day for 51 days of operation. The 51 broiler rinses constitute a "Salmonella set." Sets were scheduled approximately once a year. When a plant did not meet the Salmonella performance standard, a follow-up set was scheduled. To limit bias, this report does not include data from follow-up sets.
Carcasses were collected after they exited the chiller, downstream from the slaughter line. The chiller is designed to bring carcass temperatures down to the refrigeration range. The postchill collection site was selected as Salmonella Enteritidis in Broiler Chickens, United States, 2000-2005 the sampling site because interventions for pathogen reduction are generally located before this point.

Broiler Rinse Collection
Carcasses were collected after they exited the chiller and aseptically placed in a sterile bag. A 400-mL volume of buffered peptone water was added to the carcass in the bag. Half the volume was poured into the interior cavity and the other half over the skin. The carcass was rinsed with a rocking motion for 1 minute at a rate of ≈35 cycles per minute. After the carcass was removed from the bag, the rinse was poured into a sterile container and shipped on a freezer pack by overnight mail to 1 of 3 FSIS laboratories (Athens, GA; Alameda, CA; St. Louis, MO, USA) for analysis (10).

Microbiologic Testing
Testing of broiler rinses for Salmonella was performed by using standard FSIS isolation methods (11). Before October 2003, an immunoassay system (Assurance polyclonal enzyme immunoassay, BioControl Systems, Inc., Bellevue, WA, USA) was used to screen enrichment broths for Salmonella. Beginning in October 2003, Salmonella gene amplification (BAX System PCR Assay, DuPont Qualicon, Wilmington, DE, USA) was performed on lysed cells after overnight incubation in buffered peptone broth (35°C). Broiler rinses that tested positive on the screening test were cultured for Salmonella with standard methods (i.e., selective enrichment, plating, serologic and biochemical confirmation). Three presumptive Salmonella colonies with the predominant colony form were selected from each plate for biochemical and serologic confirmation. One confirmed Salmonella isolate was sent to the National Veterinary Services Laboratories (NVSL, USDA-APHIS-VS, Ames, IA, USA) for Salmonella serotyping (12).
Beginning in 2001, isolates of Salmonella Enteritidis were phage typed at NVSL (13). Because the predominant Salmonella Enteritidis phage types were clonal (6,14) and pulsed-field gel electrophoresis and antimicrobial susceptibility patterns were not available on all isolates during the study period, no further characterization of the isolates was performed for this report.

Analysis
Analysis was restricted to Salmonella sets performed in calendar years 2000-2005. A χ 2 test (2-sided) was used to test trends for annual percent of Salmonella Enteritidis isolates among Salmonella-positive broiler rinses and all analyzed broiler rinses, respectively. A χ 2 test for trend was also performed to assess the percent of establishments tested annually with Salmonella Enteritidis-positive broiler rinses, with subanalyses by establishment size. Approximately two thirds of establishments were large (>500 employees), one fourth were small (<500 but >10 employees), and 5% were very small (<10 employees). In addition, a χ 2 test for trend was performed on the number of isolates per Salmonella Enteritidis-positive establishment, by year (SAS version 9.1, SAS Institute, Inc., Cary, NC, USA).
The number of positive broiler rinses per state was plotted on a US map that showed the geographic density of broiler chicken production by county for the year 2002 (15). Results were plotted for 2 periods: calendar years 2000-2002 and 2003-2005. Phage types of isolates were tabulated by year.
The present study preceded a new FSIS policy to control Salmonella. The new policy emphasizes improvement in Salmonella control in product classes that have not reduced Salmonella prevalence in the past decade, such as broilers, and focuses on plants that test positive for common serotypes of human illness, such as Salmonella Enteriditis (16).

Results
During the 6-year study period, 280 (0.5%) Salmonella Enteritidis isolates were recovered from 51,327 broiler rinses ( increase in the number of positive establishments per year remained significant after stratification by large versus small establishment size. While most establishments with Salmonella Enteritidis-positive broiler rinses were large, in most years, a higher proportion of the small establishments that were tested had positive rinses. During the 6-year study period, the proportion of Salmonella Enteritidis-positive establishments with multiple positive broiler rinses per year also increased significantly (p<0.01, test for trend, Figure 1). In addition, the proportion of establishments with >4 positive broiler rinses per year (of 51 broiler rinse tests per set) increased, beginning in 2002.
From 2000 to 2002, Salmonella Enteritidis was isolated from broiler rinses in 14 states, compared with 24 states from 2003 to 2005 (Figure 2). Phage type (PT) 13 was predominant, accounting for half of all isolates, followed by Salmonella Enteritidis PT 8, which accounted for more than one third of isolates (Table 3). In 2005, the number of isolates that were PT 8 increased >3-fold compared with 2004.

Discussion
The principal finding of this study was a significant increase in the number of broiler chicken slaughter establishments with Salmonella Enteritidis-positive broiler rinses in the years from 2000 through 2005. The 90 slaughter establishments with positive rinses were dispersed across 24 states, reflecting the geographic distribution of the US broiler industry. During the study period, increases were seen in the proportion of both large and small establishments that had such positive broiler rinses.
Some caution is warranted when interpreting our findings. The purpose of the FSIS Salmonella program is to assess performance of individual establishments. The program is not designed to estimate national prevalence of poultry contamination because it does not fully account for production volume or regional or seasonal effects. Furthermore, samples are collected after slaughter processes that are intended to reduce carcass contamination. Nonetheless, the apparent emergence of Salmonella Enteritidis in broilers is noteworthy given the increase in human Salmonella Enteritidis infection rates in the United States (8) and recent findings that eating chicken is a new and important risk factor for sporadic infection (5,6). Additional epidemiologic studies are recommended to further elucidate the role of contaminated chicken in human Salmonella Enteritidis infections and estimate the extent of illness attributable to chicken. Retail food surveillance and laboratory subtyping studies (6) may also be valuable because they enable comparisons of human and poultry strains.
In this report, 2 Salmonella Enteritidis phage types, PT 8 and PT 13, accounted for most isolates from broiler rinses. In a recent FoodNet study, the association between Salmonella Enteritidis infection and eating chicken strengthened in analyses restricted to patients infected with these 2 phage types (6). The possible emergence of these phage types in broiler chickens suggests that industry should implement appropriate Salmonella Enteritidis controls for broiler chickens (17,18).
The present study preceded a new FSIS policy to control Salmonella in broilers that emphasizes common serotypes of human illness (16)  . Information from these meeting was used to prepare guidelines to help broiler plants control salmonellae (19). The agency is also monitoring progress of meat and poultry plants in controlling this organism. If, in July 2007, most plants (e.g., 90%) that manufacture a specific product (e.g., broiler carcasses) have not reduced the percentage of Salmonella tests that are positive to at least half the FSIS performance standard, the agency will consider actions to improve control of salmonellae. One option that FSIS is considering is to post Salmonella results on the web for product classes that have not made sufficient progress, listing data by plant name.
In the 1990s, successful voluntary quality assurance programs to control Salmonella Enteritidis were developed by the egg industry and state poultry health officials (20). Many of the interventions are adaptable to the control of this organism in broilers. For example, control points for the organism in broilers are likely to include monitoring and sanitation of breeding flocks, hatcheries, broiler flocks, and slaughter establishments. Serotype data that FSIS provides to plants on each isolate as part of its new Salmonella policy (16) may also assist plant officials to make informed SE risk management decisions.