Nationwide Microbiological Baseline Data Collected by Sponge Sampling during 1997 and 1998 for Cattle, Swine, Turkeys, and Geese

Transcription

1 1848 Journal of Food Protection, Vol. 68, No. 9, 2005, Pages Nationwide Microbiological Baseline Data Collected by Sponge Sampling during 1997 and 1998 for,,, and D. R. EBLEN, P. LEVINE,* B. E. ROSE, P. SAINI, R. MAGEAU, AND W. E. HILL Microbiology Division, Office of Public Health Science, Food Safety and Inspection Service, U.S. Department of Agriculture, Washington, D.C , USA MS 05-14: Received 17 January 2005/Accepted 12 April 2005 ABSTRACT During 1997 and 1998, the U.S. Food Safety and Inspection Service completed nationwide microbiological baseline studies on four separate categories of livestock and poultry. Data were collected by sponge-sampling techniques. These studies were designed to provide nationwide estimates of the prevalence of Salmonella and prevalence and concentrations of Escherichia coli in cattle (n ), swine (n ), turkeys (n ), and geese (n ) in establishments under federal inspection. Salmonella prevalence ranged from 1.2% in cattle to 6.9% in swine, 13.7% in geese, and 19.6% in turkeys. The prevalence of E. coli was 16.6% in cattle (geometric mean 0.26 CFU/cm 2 ), 44.1% in swine (mean 0.78 CFU/cm 2 ), 92.7% in turkey (mean 2.46 CFU/cm 2 ), and 96.5% in geese (mean 1.97 CFU/cm 2 ). These values are similar to or somewhat lower than previous baseline values obtained for steers and heifers, cows and bulls, market hogs, and young turkeys. This study is the first in which nationwide microbiological baseline data have been compiled for geese. These data will be useful to individuals working with hazard analysis critical control point plans and risk assessment and to the research and academic communities. The inspection of food animals at the time of slaughter historically has been focused on the visual identification of symptoms of disease conditions that render the entire carcass or parts thereof unfit for use as human food. However, microflora (including foodborne pathogens) can asymptomatically reside in the digestive tract or on the hide, skin, or feathers of an animal. The U.S. Department of Agriculture Food Safety and Inspection Service (FSIS) has developed a number of microbiological carcass sampling programs that are used to verify that good manufacturing practices are adhered to during processing, thereby minimizing the numbers of foodborne pathogens to which the consumer will be exposed. These programs were initially developed using data gathered from 1992 to 1997, when individual year-long nationwide microbiological baseline studies were performed on steers and heifers, cows and bulls, market hogs, and young turkeys. In these studies, cattle and swine were sampled by the excision method, where a composite 60-cm 2 area of muscle tissue per carcass tested was removed for microbiological analysis (8). were sampled by the whole-bird-rinse method, which involved shipping a chilled carcass to the lab for testing, rinsing it in 600 ml of diluent, and testing a portion of the rinse fluid. Logistical problems were encountered during sample collection for these studies. Excision sampling of beef and * Author for correspondence. Tel: ; Fax: ; priscilla.levine@fsis.usda.gov. Mention of brand or firm names does not constitute an endorsement by the U.S. Department of Agriculture over other brands or firms of a similar nature that were not mentioned. Retired. swine carcasses can result in economic loss because the carcass surface constitutes a valuable part of these animals. The whole-bird-rinse method was extremely costly and occasionally resulted in operator injuries due to the weight of the birds. Mindful of these considerations, the FSIS used a sponge-sampling technique for collected under the pathogen reduction hazard analysis critical control points (PR/HACCP) rule. The PR/HACCP rule was published in 1996 (5) with the goal of reducing the prevalence and numbers of pathogenic microorganisms on meat and poultry products. The introduction of the PR/HACCP rule represented a fundamental shift from the regulatory philosophy of command and control, where certain meat processing requirements were mandated (many of which with tenuous links to food safety), to microbiological outcome or PR/HACCP based food safety process control programs, which allow industry more flexibility while still requiring robust verification. Under the PR/HACCP rule, federally inspected establishments are subject to routine Salmonella testing. These establishments must meet the relevant product-specific Salmonella performance standards (see 9 CFR (b) and (b)), which have been calculated to provide an establishment with an 80% probability of passing when it is operating at the national baseline prevalence of positive Salmonella results determined from previously conducted baseline studies., swine, and turkey slaughter establishments must also perform routine tests for Escherichia coli as an indicator of fecal contamination. The FSIS has completed nationwide microbiological baseline studies for cattle (including cows, bulls, steers,

2 J. Food Prot., Vol. 68, No. 9 NATIONWIDE BASELINE DATA COLLECTED BY SPONGE SAMPLING 1849 heifers, and calves), swine (barrows, gilts, boars, and sows), young turkeys, and for the first time geese. A primary aim of these baseline studies was to obtain national microbiological data that would be useful to those updating and evaluating PR/HACCP plans and to those in the risk assessment, research, and academic communities. MATERIALS AND METHODS Selection of organisms. Two microorganisms were selected for analysis. Salmonella (non-typhi) was selected from a group of organisms most often associated with human illness in the United States, as determined by foodborne illness reports. E. coli was chosen as an indicator of general hygiene and process control. Establishments included in the sampling frames for the studies. For cattle, the sampling frame included all establishments (approximately 735) under federal inspection that slaughtered steers or heifers, cows or bulls, and/or calves during the sample selection period of June 1997 through May For swine, the sampling frame included all federally inspected establishments (approximately 680) that slaughtered barrows or gilts and/or boars or sows from June 1997 through May For turkeys, all establishments (approximately 65) under federal inspection that slaughtered young turkeys from July 1997 through June 1998 were included. Young turkey production accounts for approximately 99% of all turkey production, and older breeder turkeys and fryer-roaster turkeys account for the remainder. The goose baseline study was conducted from September 1997 through November 1997 because of the limited availability of geese at other times of the year. At the time of sampling frame development, there were eight federally inspected establishments slaughtering geese, and these establishments slaughtered a total of approximately 250,000 birds per year. However, of the eight establishments, only the two that slaughtered more than 5,000 geese per year were included in the sampling frame, representing approximately 97% of all geese slaughtered in the United States in Sample selection. For cattle and swine, a sample size of approximately 2,000 was chosen to ensure reasonable levels of precision for this nationwide survey. Approximately 3,000 were requested during the 52-week study to account for inevitable nonresponses, incorrectly collected or packaged, and delayed shipment. For the same reason, approximately 1,900 turkey were requested, with the goal of analyzing at least 1,200. Turkey sample collection was concentrated from May through November (39 birds per week), with fewer birds sampled from December through April (33 birds per week) to account for seasonal variation in production during a typical year. To achieve a random sample, sample selection was performed in two stages. For cattle, swine, and turkeys, the first stage was to randomly select an establishment from the sample frame. Establishments were selected with probabilities in proportion to the total number of animals slaughtered annually, by slaughter class if applicable. As a result, establishments with a large proportion of total animals slaughtered had a greater chance of being selected and could be sampled on multiple occasions. With this process, all carcasses had an approximately equal chance of being selected. Random of goose carcasses were requested from each of the two establishments in the sample frame, in proportion to their individual goose slaughter totals from July 1996 though July The second stage of sampling was to select the carcass from within the establishment. To successfully achieve the objectives of the program, it was necessary to derive data from an important point in the production process. Key factors in the microbial profile of cattle and swine are the slaughter and carcass dressing practices. Therefore, carcasses were sampled after chilling, which is the end point for the slaughter and dressing procedures. Similarly, for turkeys and geese the slaughter and evisceration processes that take place under federal inspection are key determinants of the resultant microbial profile, so were selected from the drip line after the chill tank, the end point for slaughter and evisceration. By sampling at these points, the contribution of the slaughter and dressing-evisceration processes to the overall microbial profile of the carcass could be assessed without interference from further processing steps that could potentially introduce confounding variables. Sample collection. Samples for cattle, swine, and turkeys were aseptically collected by FSIS inspection program personnel following procedures described in FSIS directive 30.5 (7). Goose were collected following procedures described in FSIS Goose Microbiological Baseline Study: Procedures for Sponge Sample Collection and Shipment Final 9/9/97 (6). and swine were collected by hydrating a sterile sponge with 10 ml of buffered peptone water (BPW), which was used to swab three areas (10 by 10 cm, as defined by a sterile plastic template). were swabbed on the flank, brisket, and rump, and swine were swabbed on the belly, ham, and jowl. Sampling of cattle and swine carcasses resulted in a composite sample of 300 cm 2. For turkeys and geese, a 100-cm 2 skin-surface composite was collected, which included one midback site (50 cm 2 ) and one thigh site (50 cm 2 ) on each carcass sampled. Paired sponge were obtained for cattle and swine. One sponge sample was collected from one carcass half for Salmonella analysis. At the same time, a second sponge sample was collected from the other half of the carcass to test for E. coli. For turkeys and geese, one sponge sample was collected from one carcass for Salmonella analysis, and a separate sponge sample was collected from a second carcass for E. coli analysis at the same sampling time. Each sponge was individually bagged, placed in an insulated package with chilled gel ice packs capable of maintaining refrigeration temperatures, and shipped the same day as sample collection to the designated testing laboratory via an overnight delivery service. Only received at the laboratory the day after sample collection with a sample receipt temperature of0to10 C inclusive were analyzed. Samples received outside of those constraints were discarded. Only one analysis, for either Salmonella or E. coli, was performed on each individual sponge of the paired. Microbiological analysis. An additional 50 ml of BPW was added to the sponge designated for Salmonella analysis, bringing the total volume to 60 ml, and the sponge and BPW were stomached in the original bag for 2 min. The procedure used for Salmonella analysis was described in Appendix E of the PR/HACCP regulations (5) and was qualitative only. With this method, were preenriched in BPW at 36 C for 20 to 24 h, and selective enrichment was conducted in both tetrathionate broth and Rappaport-Vassiliadis broth incubated at 42 C for 22 to 24 h. Enrichment cultures were then streaked onto both double-modified lysine iron agar and brilliant green sulfa agar plates, which were incubated at 36 C for 22 to 24 h. Suspect colonies were selected and confirmed with biochemical testing procedures. This method of analysis, when performed as described, has been estimated to detect microorganisms at concentrations of at least 1 CFU/g in a 25-g sample. An additional 15 ml of BPW was added to the sponge designated for E. coli analysis, bringing the total volume to 25 ml.

3 1850 EBLEN ET AL. J. Food Prot., Vol. 68, No. 9 TABLE 1. Samples analyzed and mean nationwide prevalence of Salmonella and E. coli in the four commodities tested Commodity Salmonella Cows and bulls Steers and heifers Calves Barrows and gilts Boars and sows E. coli Cows and bulls Steers and heifers Calves Barrows and gilts Boars and sows analyzed 386 1, , , , positive , Prevalence (%) (95% confidence interval) 2.1 (0.5, 3.6) 0.6 (0.2, 1.0) 8.8 (1.3, 16.3) 1.2 (0.7, 1.7) 6.3 (5.3, 7.4) 19.0 (10.8, 27.2) 6.9 (5.8, 8.0) 19.6 (17.4, 21.7) 13.7 (6.6, 20.9) 20.2 (16.1, 24.3) 13.5 (11.8, 15.3) 60.3 (48.0, 72.7) (41.0, 45.4) 61.0 (50.9, 71.1) 44.1 (41.9, 46.2) 92.7 (91.3, 94.1) 97.1 (93.3, 100) The sponge and BPW were stomached in the original bag for 2 min and analyzed according to the procedure for quantitative E. coli analysis described in the FSIS Microbiology Laboratory Guidebook (9), using rehydratable E. coli coliform Petrifilm count plates (3M, St. Paul, Minn.). Following the stomaching procedure, appropriate dilutions were made to obtain an appropriate final count, and 1 ml of each dilution was plated onto duplicate E. coli coliform Petrifilm count plates. The plates were incubated at 35 C for 48 h, and the counts were recorded as CFU per square centimeter. Data limitations. These studies were designed to provide nationwide estimates of Salmonella and E. coli prevalences and E. coli concentrations on cattle, swine, turkey, and goose carcasses. These studies were not designed to provide microbiological information for individual establishments. To obtain such information, a large volume of data would need to be collected from each establishment over an extended period of time. The Salmonella results obtained from these studies are independent of ongoing Salmonella regulatory testing; different criteria are used to select establishments for sampling in PR/HACCP verification programs. RESULTS AND DISCUSSION Table 1 includes the total numbers of analyzed for each of the four commodities tested (broken down by slaughter class for cattle and swine) and the mean nationwide prevalence of Salmonella and E. coli in these. TABLE 2. Mean nationwide concentration of E. coli in the four commodities tested Commodity analyzed positive , a Only data for positive are included. Geometric mean concentration (CFU/m 2 ) (95% confidence interval) a 0.26 (0.22, 0.32) 0.78 (0.65, 0.94) 2.46 (2.24, 2.70) 1.97 (1.47, 2.66) Table 2 lists mean nationwide concentrations of E. coli in the four commodities tested. Table 3 lists the distribution of E. coli from carcass sponge analyzed for each commodity tested. Two separate baseline studies in which the excision technique was used were previously performed for cattle: one with cows and bulls and one with steers and heifers (8). Direct comparison of these older excision baselines with the current sponge baseline for cattle, which is inclusive of both categories of cattle and also includes calves, revealed a similar Salmonella prevalence obtained with sponge sampling versus excision sampling: 2.1 versus 2.7% in cows and bulls and 0.6 versus 1.0% in steers and heifers, respectively. The E. coli prevalence (16.6% overall) in the present study is higher than that observed in the earlier cow and bull study (20.2 versus 15.8%) and steer and heifer study (13.5 versus 8.2%) (sponge sampling versus excision sampling, respectively). However, concentrations of recovered E. coli were 100-fold higher in the excision studies (geometric means of 33 and 35.3 CFU/cm 2 for cows plus bulls and steers plus heifers, respectively) than in the sponge study, where a geometric mean of 0.26 CFU/cm 2 was obtained. A small number of calves (n 68) were included in the sponge baseline study (Table 1). They represented 3.6% of all cattle tested and yielded 13.1% of the E. coli positive and 26.1% of the Salmonella positive. The 95% confidence intervals indicated that calves had a significantly higher proportion of Salmonella positive than did steers and heifers, but the proportion was not significantly different from that for cows and bulls. Calves also had significantly higher E. coli concentrations than did the other cattle categories. However, given the small number of calves sampled and the high variability of these data, the inclusion of calves in the data set did not have a significant impact on the overall rate of Salmonella or E. coli positive. In an earlier baseline study with market hogs, were collected by excision. The prevalence of Salmonella (6.9 versus 8.7% in the sponge and excision baseline studies, respectively) was similar in both studies. However, there were apparent differences in the prevalence (44.1 versus 31.0%) and concentration (0.78 versus 75.9 CFU/cm 2 ) of E. coli observed on raw pork carcasses tested in the sponge and excision baseline studies, respectively. A small number of boars and sows (n 100) were included in the sponge baseline study. These represented 4.7% of all swine tested and yielded 6.5% of the E. coli positive and

4 J. Food Prot., Vol. 68, No. 9 NATIONWIDE BASELINE DATA COLLECTED BY SPONGE SAMPLING 1851 TABLE 3. E. coli concentration from carcass sponge analyzed for each of the four commodities tested (%) Concentration (CFU/cm 2 ) 0 a ,000 1,569 (83.4) 230 (12.2) 62 (3.3) 20 (1.1) 1, ,000 10, ,000 (100) 1,190 (55.9) 598 (28.1) 158 (7.4) 101 (4.7) 59 (2.8) 20 (0.9) 1 (0.0) (100) (7.3) 480 (34.4) 582 (41.7) 191 (13.7) 28 (2.0) 4 (3.5) 37 (37.5) 43 (43.4) 16 (15.6) 12 (0.9) 1 (0.1) (100) (100) a Below detection limit, which was CFU/cm 2 for cattle and swine baselines and CFU/cm 2 for turkey and geese baselines. 13% of the Salmonella positive. The 95% confidence intervals indicated that boars and sows had a significantly higher proportion of Salmonella positive and E. coli positive than did barrows and gilts. However, given the small number of boars and sows sampled and the high variability of these data, the inclusion of this slaughter class in the data set did not have a significant impact on the overall rates of Salmonella or E. coli positive. Comparing the results of the sponge baseline studies with those of previous baseline studies performed with the same animals may seem to be an obvious step. The passage of time between the excision studies and the sponge studies, particularly considering the introduction of the PR/HACCP rule in the interim, may account for some of the differences observed. The sampling methods used may also be responsible, and although the results presented here do indicate some similarity between the results of these baseline studies, there has been much debate on the appropriateness of such direct comparisons. In one unpublished FSIS study, the increased sensitivity of the three-site sponge method was attributed to the greater surface area sampled (300 cm 2 compared with 60 cm 2 with the excision method). The researchers also found that although sponging generally gave average microbial counts lower than those of the excision baseline method, numerous negative excision results and negative sponge results presented difficulties that prevented the calculation of an appropriate conversion factor. Although for calculation of prevalences recovery of organisms with the sponging technique is at least as good as if not better than excision, many researchers have found that more organisms can be recovered from carcasses with the excision technique. Phillips et al. (3) found that the prevalence of E. coli on beef carcasses was higher based on excised than on sponged. Ware et al. (11) found that higher bacterial numbers were recovered from beef carcasses by excision than by the sponging technique. However, these authors also stated that either sponging or excision would be appropriate to use in carcass sampling for process control verification provided that results were compared based on specific criteria and the same sampling procedure. Conversely, when beef carcasses were tested for E. coli and Salmonella by both the sponge and the excision techniques, Ransom et al. (4) reported no difference in bacterial recovery between the two sampling methods. Gill and Badoni (1) reported that sampling of poultry carcasses by excision and by rinsing led to recovery of similar numbers of bacteria. Gill and Jones (2) compared recovery of E. coli on beef and pork carcasses by excision- and sponge-based methods and concluded that the numbers of bacteria recovered by excision or swabbing with sponge were comparable. Many factors unrelated to sampling methodology, such as process variability, can affect the numbers of bacteria recovered from meat. Therefore, any assumed relationship between populations recovered by sponging and by excision would have to be verified on a per-product basis to determine the relationship between the recovery of bacteria by either method (3). Any interpretation of the present comparison must consider differences in methodologies between baseline studies. Careful examination of this issue and, if appropriate, design of a conversion factor that would allow direct comparison of results from the sponge baselines with those of earlier baseline studies would be useful. Federally inspected establishments presently are required to conduct routine testing of carcasses for E. coli in slaughter operations as an indicator of process control of fecal contamination. Comparison of the results from the present studies with performance criteria derived from sponge-sampling data would be appropriate. However, these results suggest that contrast of sponge-sampling data with excision-sampling baselines may be falsely encouraging because of the generally lower counts obtained with the sponge method when compared with excision sampling. Comparison of results for turkeys between the earlier rinse baseline and the sponge baseline revealed that the E. coli populations were similar, with means of 2.45 and 3.72 CFU/cm 2 for the sponge and rinse baselines, respectively. Presently, turkey carcasses are not subject to a Salmonella performance standard because when the initial rule was written no baseline study of this product had been completed. Comparison of the prevalences reported in the two baselines, 19.6 and 18.6% for sponge and rinse studies, respectively, suggests that there is a baseline level of Salmonella in turkey carcasses of around 19% (data from 1997 and 1998), similar to that in broiler chickens tested during 1994 and 1995 (8). Prior to this study, the FSIS had not performed any

5 1852 EBLEN ET AL. J. Food Prot., Vol. 68, No. 9 microbiological baseline studies on goose carcasses, and no performance standard existed for this product. The results for Salmonella and E. coli prevalence (13.7 and 96.5%, respectively) and E. coli concentration (1.97 CFU/cm 2 ) fall within the ranges observed for other poultry (i.e., turkeys and chickens (8)). Performance standards in force are based on data from FSIS baseline studies, such as those described here. The 1996 PR/HACCP rule states that performance standards for Salmonella require all slaughter establishments to reduce the incidence of Salmonella contamination of finished meat and poultry carcasses below the national baseline prevalence as established by the most recent FSIS nationwide microbiological baseline data for each major species. The baseline studies described here constitute the most recent studies in these product categories, and guidance for processors at federally inspected establishments on the use of these data was recently published in the Federal Register (10). Although comparison of the studies described here might suggest that revision and reconsideration of existing performance standards is warranted, as already discussed, results obtained by excision and sponging methods are not directly comparable. The results from these studies suggest that it is unlikely that the prevalence of Salmonella would be underestimated by the sponge method. However, there is evidence to suggest that E. coli concentrations detected by the sponge method may be considerably lower than those that would be detected by the excision method. ACKNOWLEDGMENTS These studies were designed through consultation with various groups in the FSIS and advice from scientists and organizations outside the FSIS. The Microbiology Division in conjunction with the Data Analysis and Statistical Support Staff coordinated the conduct of these programs and provided data analysis. Thanks are given to Celine Nadon (Microbiology Division, Office of Public Health Science, FSIS) for additional statistical analysis. The microbiological analyses were conducted by the FSIS Field Service Laboratories (Athens, Ga.; St. Louis, Mo.; and Alameda, Calif.). Sample collection was the responsibility of the FSIS inspectors-in-charge, without whom these studies could not have been performed. REFERENCES 1. Gill, C. O., and M. Badoni Recovery of bacteria from poultry carcasses by rinsing, swabbing or excision of skin. Food Microbiol. 22: Gill, C. O., and T. Jones Microbiological sampling of carcasses by excision or swabbing. J. Food Prot. 63: Phillips, D., J. Sumner, J. F. Alexander, and K. M. Dutton Microbiological quality of Australian beef. J. Food Prot. 64: Ransom, J. R., K. E. Belk, R. T. Bacon, J. A. Scanga, and G. C. Smith Comparison of sampling methods for microbiological testing of beef animal rectal/colonal feces, hides, and carcasses. J. Food Prot. 65: U.S. Food Safety and Inspection Service Pathogen reduction; hazard analysis and critical control point (HACCP) systems; final rule. Fed. Regist. 61: U.S. Food Safety and Inspection Service Goose microbiological baseline study: procedures for sponge sample collection and shipment final 9/9/1997. U.S. Food Safety and Inspection Service, Washington, D.C. 7. U.S. Food Safety and Inspection Service Self-instruction guide for collecting raw meat and poultry product for Salmonella analysis. FSIS directive U.S. Food Safety and Inspection Service, Washington, D.C. 8. U.S. Food Safety and Inspection Service The baseline data collection program. Available at: Baseline Data/index.asp. Accessed 6 June U.S. Food Safety and Inspection Service Microbiology Laboratory Guidebook, 3rd. ed., sect. 3.5c. U.S. Food Safety and Inspection Service, Washington, D.C. 10. U.S. Food Safety and Inspection Service Generic E. coli and Salmonella baseline results. Fed. Regist. 70: Ware, L. M., M. L. Kain, J. N. Sofos, K. E. Belk, and G. C. Smith Comparison of sponging and excising as sampling procedures for microbiological analysis of fresh beef-carcass tissue. J. Food Prot. 62: