The Slaughter Module examines how handling practices and fabrication procedures influence E. coli O157:H7 contamination from the time when live cattle arrive at a slaughter plant to the time when pieces of trim are combined into boxes or bins destined for commercial ground-beef production. O157:H7 prevalence distributions developed in the Production Module serve as inputs to this module; its outputs are distributions of O157:H7 contamination in combo bins and trim boxes. The model separately factors in breeder and feedlot cattle and high-prevalence (June–September) and low-prevalence (October–May) seasons.
It is important to recognize that the slaughtering and fabrication processes that constitute the Slaughter Module may be the most crucial link, before cooking for consumption, in the farm-to-fork chain. The importance of the Slaughter Module is evident in the fact that through its steps sterile muscle tissue of healthy animals is converted into meat that may become contaminated with bacteria, including E. coli O157:H7. Interventions in the Production Module influence the extent (prevalence) and level (concentration of cells) of contamination introduced into the slaughter plant; interventions in the Preparation Module aim at eliminating the pathogen from the product before consumption. The Slaughter Module bridges the two, examining the factors that affect the amount of fecal and hide contamination that is introduced into and remains on carcasses and how this contamination is distributed on the meat cuts and trimmings that become ground beef. The extent and level of contamination in turn influence the efficacy of cooking in eliminating pathogens from the ready-to-eat product and thus human exposure. Excessive contamination levels
may lead to consumer illness due to pathogen survival even after recommended cooking procedures, and high prevalence of contamination at lower levels makes illness from undercooking or cross contamination more likely. Thus, the events evaluated in the Slaughter Module greatly influence the outcome of the whole model.
This chapter presents the committee’s review of the Slaughter Module. Five primary subjects are addressed: difficulties of data collection, sources of contamination and cross contamination during slaughter and fabrication, the levels (cell density) and extent of carcass and trim surface-area contamination, the effects of decontamination on pathogen prevalence (especially on pathogen load and surface area contaminated), and terminology. Some additional committee observations and comments are offered in Appendix B, and Appendix D is an independent review prepared by Edmund Crouch on the variables used in this module and their implementation in the spreadsheet environment.
LACK OF DATA AND DIFFICULTIES ASSOCIATED WITH DATA COLLECTION
To those unfamiliar with predictive modeling but with some experience in slaughter operations and microbiology, it may be difficult to understand how modeling could be used to predict contamination levels and the size of surfaces contaminated during slaughter and fabrication operations. The task would seem intractable because of the variability and potential unpredictability of events during those operations. Variability in contamination and cross contamination may originate in such factors as plant size, design, age, equipment, automation, speed of slaughter, and animal holding facilities; geographic location; season of the year; type, lot, and origin of animals; labor shift; and personnel training and turnover. As live animals arrive for slaughter, they may be free of E. coli O157:H7 contamination or be contaminated in their gastrointestinal systems or on their hides. Contamination may be localized or may have spread to larger or multiple locations of the hide; the concentration of cells in contaminated spots or niches may be variable. Hide contamination is unpredictable because it can be the result of fecal shedding by individual animal or of cross contamination on the farm, during transportation, or holding before slaughter, when animals enter the slaughter chain. Cross contamination can affect other animals or the plant environment, which in turn can contaminate animals, carcasses, or meat. Fed steers and heifers from one pen are shipped and slaughtered together; culled animals from different farm environments can be commingled and thereby contaminate one another.
Slaughter presents numerous opportunities for contamination and cross contamination that may vary among plants. As the hide is separated
for removal, contamination may be introduced onto the carcass surface. A single source (an animal or the plant environment and equipment) may contaminate carcasses not only during dehiding but also during later steps. Some operations are more likely than others to result in carcass contamination, and some carcass areas are more prone than others to exposure to potential contamination or cross contamination; but the prevalence (number of carcasses), extent or level (cell density), and size of the carcass surface area contaminated—especially the latter two—are difficult to estimate, or data are unavailable or impossible to collect for their estimation. During slaughter operations, there is always the opportunity for unpredictable cross contamination or spreading, removal, or inactivation of contamination.
Contamination of a carcass, especially with E. coli O157:H7, may be localized and of low or high density. The contamination is not always spread uniformly throughout a given area of the carcass surface. When a carcass sample is analyzed and reported to have a number of some kind of cells per square centimeter, those cells may have actually been present in the whole area being measured or in any fraction of it. The presence and extent of contamination and the size of surface area contaminated with the cells may change unpredictably during the slaughtering steps because of such events as carcasses touching each other, aerosols, worker activities, water splashing, knife trimming, steam vacuuming, spray washing, other decontamination interventions, and surface drying during chilling. Changes in contamination may include elimination, spreading to larger carcass surface areas, shifting to other carcass surface areas, direct cross contamination of other carcass areas through equipment or workers, reduction in numbers or in surface area contaminated, and reduction in prevalence. Those uncertainties do not include potential effects and variation due to differences in bacterial attachment and biofilm formation on surfaces, exposure to sublethal stresses, potential development of bacterial cell resistance, and cross-protection effects of such stresses or interventions as cooking (Samelis et al., 2001a,b, 2002a,b). There are no data for estimating or predicting any of those potential changes, and some of them may prove to be unpredictable. Available data exist only to support the suggestion that dehiding results in carcass contamination (in terms of prevalence) and that decontamination interventions can have a substantial effect in reducing prevalence and levels of contamination if applied properly. Considering the limitations of sampling and testing, those conclusions are well established. However, even for such well-established effects, there is variation among plants, lots of animals, slaughter times, and other factors. Thus, it is difficult to predict the extent (cell density) and the carcass surface area contaminated as a carcass leaves the slaughtering room and enters the chiller. The effect of the chiller on levels of contamination is also largely unknown because of the paucity
of data. It is feasible to determine its effect with well-designed studies— but not knowing the carcass area contaminated is still a limitation.
After the chiller, carcass sides enter the fabrication process. Again, the only data that are available or could be obtained with reasonable certainty are related to the prevalence of a pathogen on carcass sides (with all the limitations of sampling); the extent of carcass area contaminated and levels of contamination are still unknown and highly unpredictable. The limitation in data becomes more pronounced during fabrication: carcasses are cut into parts; a major portion of the external carcass surface is removed and enters the rendering process; new and larger meat surfaces are exposed to potential contamination; meat comes into contact with table surfaces, equipment, and worker’s hands; and meat from different carcasses is mixed in combo bins. Contamination changes during fabrication are unknown and may be unpredictable.
The scientists who prepared the FSIS draft risk assessment have done a commendable job of developing their model, given the challenges they faced. However, the lack of publicly available data on crucial steps in the slaughter process, the variability of the operations modeled in the module, and the potential unpredictability of the effects of some activities on contamination during slaughter and fabrication complicate modeling and limit its ability to predict outcomes.
The committee recommends that the impact of data deficiencies and difficulties associated with data collection, which have been recognized in various parts of the FSIS draft risk assessment document, be more strongly emphasized in discussions of the outcomes calculated by the model. The data deficiencies identified by the risk assessment should serve as the foundation for a delineation of research priorities to be promoted or pursued so that the model (and E. coli O157:H7 policy decisions) can be improved in the future.
The committee also recommends that the authors add a discussion of the appropriate and inappropriate applications of the slaughter module in its present state of development—in particular, whether the module is ready to be used to draw conclusions about the factors most important in influencing the occurrence and extent of E. coli O157:H7 contamination in ground beef and the possible impact of interventions.
SOURCES OF CONTAMINATION AND CROSS CONTAMINATION DURING SLAUGHTER AND FABRICATION
Issues Regarding Fecal and Hide Contamination
The FSIS draft risk assessment states that “the number of E. coli O157:H7 organisms that initially contaminate a carcass depends on the
level of infected cattle.” However, carcass contamination (during dehiding and later steps) may originate not only from “infected” cattle but also from other sources, such as animal hides that cross-contaminate noninfected cattle during transportation, in holding areas, and during slaughter. In addition, carcasses and meat may become contaminated through O157:H7 niches that are established in plants and through cross contamination, not only during dehiding but also in handling at later stages of slaughter and fabrication (Samelis et al., 2002a).
As noted in the review of the Production Module, using fecal prevalence in slaughtered cattle as the sole measure of carcass contamination is a major weakness of the draft model. The FSIS draft risk assessment acknowledges that cross contamination may occur and notes that the hide is an additional source of contamination. The draft’s justifications for using fecal contamination as the only source of carcass contamination are in the discussion of the knock box and stunning operations: there are sparse data on hide prevalence, the contribution of hide contamination is implicit in the existing model, and research indicates that the fecal status of incoming cattle correlates most strongly with carcass contamination.
Not considering potential sources of carcass and meat contamination other than animals carrying the pathogen in their gastrointestinal systems simplifies modeling but has major drawbacks. Omitted factors include contamination from animal hides and the influence of variations in plant design, size, capacity, and operational procedures on extent of contamination.
The committee recommends that the final risk assessment emphasize these weaknesses of the model and state that the outcomes of the model need to be recalculated after additional data become available or state that the model cannot provide informative predictions at this stage of its development because of lack of data in key segments of the process.
Issues Regarding the Use of Limited Data to Determine Carcass Contamination
The research of Elder and colleagues (2000) is cited as the primary support for the notion that fecal E. coli O157:H7 prevalence data best predict the quantitative correlation between preharvest and postharvest contamination. However, a close reading of the paper reveals a more complex picture. The Elder et al. data indicated that prevalence in all fecal and hide samples was significantly correlated with prevalence of positive carcasses (p = 0.001) and that there was “no significant difference between the proportion of lots positive on fecal and hide samples and those positive on carcass samples (p = 0.2207).” The authors note that their data suggest a lack of association between hide prevalence and carcass con
tamination and that there appears to be a correlation between fecal prevalence and initial carcass contamination. However, a possible reason for these findings is the lower prevalence of E. coli O157:H7 on hides than in feces in the study. Those data do not preclude situations in which hide prevalence is higher than fecal prevalence, in which case carcass contamination may correlate better with hide contamination. In addition, existence of a higher correlation between fecal and carcass prevalence does not preclude hide contamination—irrespective of how low it is—as the source of direct contamination or cross contamination of carcasses or meat.
The observations summarized in Table 2 of Elder et al. (2000), outlined below, illustrate the complexity of the situation.
A group of cattle from a single source (a “lot”) with high fecal contamination (76.5%) and hide contamination (11.1%) yielded carcasses with 0% postprocessing contamination.
A lot with 0% fecal and hide contamination yielded carcasses with 75% contamination before evisceration and 0% after processing.
Fecal and hide prevalence of 12.5% and 6.3% were associated with 56.3% and 0% prevalences in carcasses before evisceration and after processing, respectively.
Fecal and hide prevalences of 11.1% and 77.8% were associated with 55.6% and 0% prevalences before and after evisceration, respectively.
Fecal and hide prevalences of 0% and 50% were associated with 30% and 0% prevalences before and after evisceration, respectively.
Some lots had 0% prevalence throughout the slaughtering process.
Of the 29 lots sampled, 21 were positive for feces and 11 for hides; of the 30 at the carcass stage, 26 were positive before evisceration, 17 after evisceration, and 5 after processing.
Those observations confirm that contamination originates in feces and hides as well as from cross contamination (even if animal testing yields zero prevalence); hide removal is the most important operation that results in exposure of carcasses to contamination; slaughter-plant operations after dehiding, including decontamination processes, reduce contamination greatly, in terms of both lot and sample prevalence; and carcass contamination before evisceration, after evisceration, and especially after processing does not correlate well with fecal or hide contamination. Prevalence varies with animal lot and slaughter plant irrespective of extent of animal contamination, and decontamination greatly reduces pathogen prevalence.
The lack of higher correlation between hide and carcass prevalence in the Elder et al. study may alternatively or additionally be due to sampling limitations. The authors found the relatively low hide prevalence to be
surprising and noted that “preliminary studies had indicated good concordance in isolation rates between fecal and hide samples on individual cattle.” They state:
One explanation for this apparent discrepancy is choice of sampling site [and it] is possible that other sites on the hide have higher levels of contamination, and are, therefore, greater risks for generating direct or airborne carcass contamination. It is also possible that survival rates of EHEC O1571 differ by site on the hide. It is clear that hides do contribute to the total bacterial load, which may contribute to carcass contamination. Further studies are required to address the relative importance of hides as a source of carcass contamination by EHEC O157.
A more recent study by Ransom et al. (2002) supports that observation, finding that E. coli O157:H7 hide prevalence varied from 13.3% to 23.3%, depending on the method of sampling.
In summary, a limitation of the model—admittedly due to lack of data—is overreliance on a single study in the Slaughter Module—Elder et al. (2000). The committee suggests that consideration be given to using available data on other pathogenic or indicator organisms to estimate proportional transfer of contamination (prevalence and levels) from live animals and the plant environment to carcasses during dehiding and possibly during later steps of the process. The committee also suggests that future studies be promoted to provide improved data for this part of the model.
The Effect of Slaughter-Plant Methods on Cross Contamination of Carcasses
The FSIS draft risk assessment states that cross contamination of hides may occur in the knock box as noncontaminated cattle fall to the floor or come into contact with sides of the chute after contaminated cattle have passed through. It also cites literature that finds that additional contamination can occur if cattle emit feces or rumen contents at the knock box or if dirty knives are used. Thus, cross contamination is expected to occur, but it is difficult to predict. Data from studies by Elder et al. (2000), Bacon et al. (1999, 2000), and Ransom et al. (2002) indicate that the extent of contamination of live animals and carcasses immediately after hide removal and after carcass washing or chilling varies greatly among plants. That suggests that there is no obvious correlation between animal con
tamination and carcass contamination, especially after processing and in individual animal lots. Depending on the plant, lot, or time of slaughter, heavily contaminated cattle may be linked with high or low numbers of contaminated carcasses or with carcasses showing no detectable E. coli O157:H7 contamination. Slaughter-plant characteristics and operations clearly have an important influence on carcass contamination (irrespective of the extent of live-animal contamination), extent of cross contamination, and the effectiveness of decontamination. Thus, slaughter plants may influence contamination levels in ways that are not captured by the cow-bull versus steer-heifer modeling performed in the draft risk assessment.
Cross Contamination of the Carcass During Evisceration
The FSIS draft risk assessment identifies evisceration as a step in the slaughter process in which contamination may be introduced through unintentional perforation of the gastrointestinal tract. Although the draft asserts that “studies indicate that evisceration is usually carried out with minimal contamination,” opportunities for additional contamination, cross-contamination, and spreading of contamination exist during evisceration and may vary with plant operations.
Although contamination and cross contamination originating in leakage of intestinal contents during evisceration are expected, available data on their extent are sparse or inadequate, and the cross-contaminating effect may be unpredictable. The draft cites a personal communication within the US Department of Agriculture (USDA) as the sole support for the assumption that gastrointestinal tract perforation potentially occurs in 1% of carcasses. Even if that rate of perforation, or leakage, is an underestimate, setting its probability between 0% and 2%, as the draft does, is almost certainly appropriate. The committee notes that the cross contamination and redistribution of contamination that may occur at this stage (and other stages of the slaughter process) may at times be substantial and suggests that the final risk assessment explicitly acknowledge this.
Factoring in Cross Contamination During Processing
The Elder et al. (2000) paper clearly indicates that cross contamination occurs during processing. It notes that “the overall prevalence of carcass contamination with EHEC O157 was significantly greater than that of fecal and hide prevalence” and that carcass samples in the same lot were positive even when no animals were fecal- or hide-positive. And the FSIS draft risk assessment recognizes the importance of self-contamination and
cross contamination in several places, acknowledging that “the exterior surface of the hide and the environment in the dehiding area are recognized sources of pathogens (Grau, 1987)” and that “cross-contamination can occur via workers’ gloves, knives, clothing, or during the changing of the hide-puller from one carcass to the next (Gill et al., 1999).”2
It can be argued that by estimating the frequency of contaminated carcasses at 120% and 160% of the prevalence of incoming contaminated cattle during the low- and high-prevalence seasons, respectively, the draft model is in effect factoring in cross contamination. However, the distributions around the values need to be based on more data. The calculation of the low-prevalence season figure is particularly problematic. It is derived by creating a “mixture” of the high-prevalence season transformation ratio and a uniform distribution ranging from near 0 to a maximum of the high-season ratio. The committee suggests that the risk assessment emphasize the need for additional data so that the frequency of cross contamination can be estimated with more confidence in later refinements of the model.
Issues Regarding Cross Contamination During Fabrication
There is a need for data to attempt to estimate the frequency and extent of cross contamination during fabrication, although this would be a difficult, costly, and time-consuming undertaking and could yield results of great uncertainty. As stated by Newton et al. (1978), “structural and work surfaces may be as important as the hide as sources of bacterial contamination of meat.” To deal with the lack of data, the FSIS draft model relies on output data from the grinder segment of the Preparation Module, and it appears that inputs were adjusted to fit expectations. The committee suggests that consideration be given to whether it could be more appropriate to adjust inputs at previous stages of the model (for example, with data on prevalence in carcasses in the chiller) to predict contamination better.
Issues Regarding the Accuracy with Which Data Were Copied from Sources
A transformation ratio (TR) is used in the FSIS draft risk assessment to relate the frequency of contaminated carcasses to the frequency of cattle in a lot carrying the pathogen. The fraction of carcasses contaminated
during dehiding is based on the results of the study of Elder et al. (2000) that reports on cattle and carcass prevalences in four slaughter plants during July–August 1999. Concerns associated with the calculation of those variables include the use of a single study for such an important part of the model (it is recognized that there is little research on the topic) and the observation that there are discrepancies in the numbers cited in the draft risk assessment and Elder et al. (2000). Specifically, the draft risk assessment states: “In lots showing evidence of E. coli O157:H7 in cattle or on carcasses, 91 of 307 [327 according to Elder et al. (2000)] cattle (30% [28%]) and 148 of 312 [341 according to Elder et al.] carcasses at dehiding (47% [43%]) were E. coli O157:H7 positive.” The independent review by Edmund Crouch in Appendix D notes additional concerns with the calculation of the TR for the low prevalence season.
The committee recommends that the numbers found in the FSIS draft risk assessment and attendant model be cross-checked for accuracy with the data presented in the published study of Elder et al. (2000).
LEVELS AND EXTENT OF SURFACE-AREA CONTAMINATION
The FSIS draft risk assessment, citing Galland (1997), notes that the number of E. coli O157:H7 organisms that initially contaminate a carcass and the level of infected cattle are affected by the average concentration of organisms per unit contaminated area and the total area of a carcass that is contaminated. However, it acknowledges that no published information is available on those factors. Collecting data to develop estimates (especially for the total surface area of the carcass and trim that is contaminated) may be difficult, but an informative model needs to account for the factors that greatly affect the extent of contamination.
Issues Regarding Contamination of the Carcass During Dehiding (Step 2)
To estimate the number of E. coli O157:H7 organisms on a contaminated carcass, it is necessary to know the number of organisms per square centimeter and the total contaminated surface area of the carcass. As the draft risk assessment notes, however, there is a lack of data on both those variables. The lack of information was handled in the draft by using data from a study conducted by the USDA Food Safety and Inspection Service (FSIS) in 1994. In that study, carcass surface tissue was excised after carcass chilling in plants throughout the United States and shipped to laboratories for analysis of various microorganisms, including E. coli O157:H7. Samples of 60 cm2 were taken from carcasses originating in feedlots. Of 2,081 samples tested, four were positive for E. coli O157:H7: two at less
than 0.03 CFU/cm2 and two at 0.301–3.0 CFU/cm2. That information was used to estimate the number of E. coli O157:H7 organisms on a contaminated carcass at dehiding. The data were combined with sampling information from the Elder et al. (2000) study—in which 6 of 330 carcass samples taken at the post-processing stage tested positive for O157:H7— to form a ratio that was used to adjust the FSIS figures for contamination below the detection limit of that study.
The committee notes several weaknesses associated with that approach: the number of samples found positive was extremely small; the positive samples were analyzed after carcass chilling, at which point the contamination density may have been reduced by carcass washing and chilling, compared with the density at dehiding; the surface area analyzed was sampled with two methods—tissue excision in the 1994 FSIS study and swabbing in the Elder et al. (2000) study; the studies used different analytic methods; and, the amount of carcass surface area sampled and analyzed was much smaller in the FSIS study (60 cm2) than in the Elder et al. study (450 cm2). It also observes that available information on the confidence interval around the percentage of positive samples in the Elder et al. study (listed in Table 1 of the paper) was not used.
On the basis of those observations, the committee believes that the weaknesses render the calculation of the adjustment ratio problematic and raise questions about the reliability of the estimates derived from the analysis. The committee recommends that these weaknesses be highlighted and the need for additional data be emphasized in the risk assessment.
Issues Regarding the Difficulties of Determining Surface Contamination
Lack of information on the surface area contaminated is also of concern. The FSIS draft authors manage it by subjectively setting the minimum area of contamination at 30 cm2 and the maximum area at 3,000 cm2 because “initial model runs showed that contaminated surface areas greater than 3,000 cm2 produced results that were infeasible in comparison with FSIS ground beef sampling data.” The range of contamination density predicted with those assumptions is 1 to 9,000 cells.
The committee acknowledges that such data are difficult to obtain or predict and that contamination of the carcass surface is expected to be localized, nonuniform, random, and nonhomogeneous, and notes that the range of contamination density predicted may well be wide enough to account for all the unknowns in the calculation. However, it points out that the derivation of the values in the draft is arbitrary and unscientific.
The committee suggests that one potential approach to deal with
concerns about determining surface contamination may be to use available data on density and extent of carcass contamination with indicator organisms, such as E. coli biotype I. The draft risk assessment notes that Bell (1997), reporting on New Zealand operations, measured densities of generic E. coli on carcasses. In addition to those results, data from several studies of North American plants may be useful in the estimation (for example, Bacon et al., 2000; Gill et al., 1996a,b; Graves Delmore et al., 1997, 1998; Reagan et al., 1996; Sofos et al., 1999a,b,c,d; Van Donkersgoed et al. 1997). Some of those studies and others report data at several steps in the slaughtering process that might be useful in estimating proportional transfer and cell density on carcasses. Future studies could be proposed to provide better data for the improvement of this part of the model.
Issues Regarding the Effects of the First (Step 3) and Second (Step 5) Decontamination on Prevalence, Levels, and Extent of Contamination
Decontamination is modeled at two points in the draft Slaughter Module: a first decontamination after dehiding (Step 3) and a second after carcass splitting (Step 5). The committee has several observations and suggestions concerning those steps.
The committee notes that not all plants apply a first decontamination; this should be explicitly recognized in the risk assessment. The FSIS draft notes that a variety of organic acids are used for decontamination, but it does not specify which, if any, are in common use. The committee observes that lactic and, to a lesser extent, acetic acid are used. It also observes that the use of hot water cited in the draft may be limited before evisceration because of potential condensate formation.
Gill (1999) and Dorsa et al. (1997) are cited to justify the range of most-likely log reductions due to first decontamination—0.3 and 0.7, respectively. However, the draft fails to justify the choice of the range of maximal values used—0.8–1.2; they might simply be 0.5 log additions to the “most-likely” values. The committee recommends that the risk assessment delineate how the maximal magnitudes of contamination reduction were determined.
The draft states that “while visible signs of foreign matter can be readily identified and removed, bacterial colonies themselves are not directly observable.” It should be pointed out that bacteria on carcasses shortly after dehiding are probably not in the form of colonies, but instead in the form of cells or clusters of cells that are not macroscopically
visible. The term colony typically refers to a mass of cells that is formed by multiplication of single cells or clumps of cells on a nutrient agar plate or a stored product and that is macroscopically visible.
The draft report needs to clarify whether the “total outside surface area” of a carcass includes the area of the body cavity, which is also exposed to the environment and may be contaminated during evisceration or after carcass splitting.
The committee notes two factors that are not reflected in the draft’s first-decontamination model:
The extent of decontamination may also be affected by level (density) of initial contamination, especially when “hot” (highly contaminated) spots exist on a carcass (Graves Delmore et al., 1998; Dorsa, 1997; Smulders and Greer, 1998; Sofos and Smith, 1998).
Initial contamination by knives, gloves, and other equipment is noted in the text as a potential source. However, the draft does not allow for the possibility of an increase in contamination at this step.
The committee recommends that these omissions be at least acknowledged as weaknesses in the model.
The committee believes that several details regarding the second decontamination ought to be better reflected in the text:
The second decontamination step is applied after “zero tolerance” carcass inspection3 and any associated trimming or steam vacuuming that may be necessary to meet “zero tolerance” inspection requirements.
The procedures used for the second decontamination depend not only on the size of the plant, as stated in the draft risk assessment, but also on such factors as equipment availability, costs, plant design, space available, and steam availability.
All plants use knife trimming and some type of water rinsing or spraying, but steam pasteurization is not used universally (although it is quite common).
The use of steam pasteurization, hot-water rinsing (thermal pasteurization), and organic-acid rinsing varies among plants, depending on the factors mentioned above.
A number of references in addition to Bell (1997) have reported on the results of decontamination interventions. These include Bacon et al. (2000); Bolton et al. (2001); Graves Delmore et al. (1997, 1998); Nutsch et al. (1997, 1998); Reagan et al. (1996). Such references should be consulted to determine whether they could better inform the modeling of this step.
The efficacy of decontamination procedures depends on such factors as pressure, temperature, cabinet or chamber design, nozzle configuration and operation, and length of application.
The assumption that “large plants typically use a steam pasteurization process” is not entirely correct, in that some large plants use hot-water or organic-acid rinses after carcass washing.
The draft states that “Phebus et al. (1997) found 3.53 ± 0.49 log CFU/cm2 reduction in E. coli O157:H7 on inoculated carcasses.” It should be clarified that those authors evaluated carcass samples inoculated with over 5 logs and that the reduction was achieved after 15 seconds of steam pasteurization, which is about twice as long as practical applications. It should be noted that findings with inocula in experimental circumstances can be very different from those with natural flora in commercial circumstances.
Two Nutsch et al. papers (1997, 1998) are cited to support the statement that “other studies have shown reductions in prevalence of E. coli O157:H7-contaminated carcasses from steam pasteurization.” The cited studies evaluated carcasses in plants but not E. coli O157:H7; they involved nonpathogenic contaminants.
The committee suggests that these observations be considered in revising the FSIS risk assessment to provide a more complete depiction of plant operations. However, it notes that the wide variations in practices for second decontamination may not necessarily result in substantially different reductions in contamination.
The FSIS draft states that “given standard industry behavior and available evidence, variability in steam pasteurization efficacy… was modeled using triangular distribution with a minimum value of 0 logs, an uncertain most-likely value of 0.5 to 1.5 logs, and an uncertain maximum value of 1.51 to 2.5 logs.” Those values appear reasonable, but the committee recommends that the risk assessment explicitly state the reasoning underlying them and note, if appropriate, that they apply not only to steam pasteurization but also to other methods applied individually or in combination in carcass decontamination.
The FSIS draft factors in the effect of decontamination on cell density
(specifically, the log reduction in cell density). The committee suggests that consideration be given to the effects of decontamination on pathogen prevalence and the contaminated surface area. In general, although data available on potential changes in contamination loads (cell densities) and contaminated surface area are sparse or nonexistent, data on the prevalence of E. coli O157:H7 on carcasses after decontamination and chilling are available (Elder et al., 2000 and others). The committee suggests that the importance of these variables in model development be acknowledged, reconsidered, and explained.
As noted elsewhere in this review, the FSIS draft risk assessment sometimes defines or uses terms in nonstandard ways. The committee found a few circumstances in the Slaughter Module in which that might confuse readers.
The draft risk assessment defines trim as a byproduct of processing carcasses to create cuts of meat (such as steaks and roasts) when the carcasses originate as feedlot cattle and as the primary product that results from deboning carcasses that originate as breeding cattle. Trim is not necessarily a byproduct, considering its volume and value. The meat industry considers such items as intestines, tongues, livers, and stomachs to be byproducts. Although a major proportion of cows and bulls becomes trim, a substantial amount is also used for less expensive steaks or in roast-beef production.
Later, in the discussion of fabrication, the expression leftover trim is used. It is confusing because trim is by definition what is left after primal cuts are removed. The text should also make it clear that only a small amount of such trim is typically vacuum packaged.
Lot is defined in the draft as the number of cattle necessary to fill one combo bin with trim; and a single lot may take one or more truckloads of cattle. To avoid potential confusion, it should be explained that in slaughter operations a lot is typically defined as a group of animals for slaughter that have a common source (a ranch or feedlot, for example) and are slaughtered together. The latter definition is used, for example, in the Elder et al. (2000) paper. During fabrication, the industry may consider product that moves from one cleanup operation to the next as a lot.
Plants are modeled as those which slaughter culled cows and bulls, called “breeding cattle”, and those which slaughter cattle fed in feedlots, called “feedlot cattle”. These terms, used throughout the exposure assessment, might not be the most appropriate choices, because they are somewhat misleading. For example, dairy animals, which are in the “breed
ing” category, are not breeding. Potential alternative terms are fed or feeder steers or heifers versus non-fed, cull, mature, or market cows or bulls.
In the discussion of the transportation segment, the term susceptible cattle is used. It is not defined, and it is not clear what the authors are referring to.
The committee suggests that the authors either adopt new terminology that clearly states concepts and definitions or align their definitions with those conventionally used in risk assessments to minimize confusion and misunderstanding. The review of the draft’s risk characterization (in Chapter 6) also addresses this point.
The FSIS draft risk assessment models cows and bulls separately from steers and heifers because most operations slaughter one of the two groups of animals. However, as indicated in the draft, a small number of plants (perhaps two to four) slaughter both types of animals. If those plants are excluded from the model, that should be explicitly acknowledged and the decision explicated.
The draft states (p. 59) that the probability and extent of E. coli O157:H7 contamination or decontamination during slaughter are modeled as dependent on
the status of the incoming animal (separate variables for the numbers of infected breeding and feedlot cattle arriving at slaughter in a truck and extent of contamination),
the type of slaughter plant (modeled indirectly—breeding and feedlot cattle are presumed to be sent to separate plants),
the type of equipment and procedures used (implicitly modeled— different types of decontamination are assumed for large versus small plants—hot water versus steam—and different efficacies are associated with them,
the efficacy of decontamination procedures (separately modeled for large versus small plants), and
the sanitation processes (which may be implicitly modeled via the efficacy of decontamination techniques, although this is not explicitly stated).
The text does not mention whether those variables are included as factors in the model or simply considered to be included through the assumptions made in developing the model. It is true that there are few data available for determining the roles of most of, if not all, these variables. However, such factors as plant design and capacity, specific combina
tions of decontamination interventions used, and efficacy of sanitation procedures may be highly influential in carcass contamination, and the committee recommends that the text make their role in the model more clear.
More generally, the draft considers seven steps of the slaughter process. It indicates that the slaughter process contains other steps but that they are not explicitly modeled. Although the role of the steps omitted may not be easy to predict or model, depending on the plant, the steps may be important in cross contamination, so their roles should be better acknowledged in the text.
In the “Arrival” step (1), the proportions of weight that amounts to trim from cow carcasses (53%) and bull carcasses (90%) may have been overestimated, considering that sizable amounts of these carcasses may be used for lower-priced steak and roast-beef items. Furthermore, although the FSIS draft risk assessment states that those values represent midpoints of uncertainty distributions that “can range ±20%” (p. 61), the draft’s Appendix C states that the variable representing the proportion (ρ) is deterministic rather than stochastic (p. 174). The risk assessment must clarify whether the variability identified in the text is actually reflected in the model.
The Arrival step discussion also indicates that the number of combo bins to which a steer or heifer carcass contributes trim depends on the number of trim “sortings,” which is based on fat content. That needs to be acknowledged in the risk assessment; if it is not modeled, a reason should be provided.
The discussion of carcass fabrication (Step 7) indicates that “Scanga et al. (2000) found no difference in the concentration of E. coli O157:H7 across fat content” in different types of beef trimmings. It should be clarified that those authors did not analyze for E. coli O157:H7 (which was considered an adulterant). However, they found higher bacterial counts and prevalence of other pathogens in trimmings as the fat content increased.
Although the draft discussion of contamination from a single carcass states that the amount of O157:H7 contamination in a combo bin depends on the number of contaminated carcasses and the amount of contamination that each carcass contributes, a major factor in the contamination of individual combo bins may be redistribution of contamination and cross contamination during fabrication. That unmodeled factor may in turn be affected by plant size, plant procedures and design, animal type, and the like.
The committee is aware of research on the impact of seasonal variation and other factors on the incidence of E. coli O157 and Salmonella in slaughter facilities being sponsored by the National Cattlemen’s Beef Association. These results, when available, may help to fill some of the data gaps identified here.
In summary, the FSIS draft risk assessment correctly acknowledges that published data on E. coli O157:H7 prevalence and levels (cell density) during slaughter and fabrication are scarce. In addition, data on the surface area contaminated and the extent of cross contamination are lacking. One reason for the paucity of published data may be the status of E. coli O157:H7 as an adulterant in ground beef and other nonintact beef products; this may have discouraged studies of the detection of E. coli O157:H7 in fresh beef. The draft recognizes the importance of data by stating that the occurrence and extent of carcass contamination, effectiveness of decontamination procedures, and effect of carcass chilling are among the factors that most influence the occurrence and extent of E. coli O157:H7 contamination in ground beef. It also recognizes the scarcity of available data and the need for more research to obtain additional information on the contribution of the hide to carcass contamination; on the prevalence, extent, and density of E. coli O157:H7 contamination on carcasses after dehiding; on the contribution of cross contamination to product contamination; on the effect of carcass decontamination and chilling on increases or decreases in E. coli O157:H7 organisms; and on the influence of fabrication activities on redistribution of contamination in meat cuts and trimmings. The correlation analysis presented as part of the sensitivity analysis indicates that the E. coli O157:H7-contaminated carcass surface area and the effects of carcass chilling have the greatest influence on the occurrence of the pathogen in combo bins and grinder loads. However, the lack of publicly available data on crucial steps in the slaughter process, the variability of the operations modeled in the Slaughter Module, and the potential unpredictability of the effects of some activities on contamination during slaughter and carcass fabrication complicate modeling and limit the module’s predictive capacity.
The lack of data has made development of the model difficult and has created the necessity to rely heavily on the results of one study (Elder et al., 2000) to make major assumptions, and to adjust some inputs to fit the expected outcomes of the model. Although assumptions are often necessary in model development, the committee recommends that those difficulties and deficiencies be more strongly emphasized in discussions of the outcomes calculated by the model and that the need for more data for model improvement be highlighted throughout the report, including in the Conclusions, the Executive Summary, and the Interpretive Summary. Furthermore, the committee recommends that the report stress the potential influence of some plant activities on cross contamination, on the level of contamination, and on the extent of carcass or trim surface area contaminated. It must be made clear that the impact of those activities, al
though important, may be difficult to characterize empirically. The committee thus recommends that the authors add a discussion of the appropriate and inappropriate applications of the model in its present state of development—in particular, whether the Slaughter Module is ready to be used to draw conclusions about the factors most important in the occurrence and extent of E. coli O157:H7 contamination in ground beef and the possible impact of interventions. The committee thus suggests that the report explicitly recognize that the myriad factors that influence E. coli O157:H7 presence, spread, growth, and elimination throughout the slaughter process are at present very difficult to characterize empirically. This means that model limitations must be acknowledged, that alternative approaches such as using data on other pathogenic or indicator organisms be considered for use in the short term, and that research priorities identified in the risk assessment be promoted or pursued so that the model can be improved in the future.
Bacon RT, Belk KE, Sofos JN, Smith GC. 1999. Incidence of Escherichia coli O157:H7 on hide, carcass and beef trimmings samples collected from United States packing plants. Proceedings of the 53rd Annual Reciprocal Meat Conference. pp. 106–108.
Bacon RT, Belk KE, Sofos JN, Clayton RP, Reagan JO, Smith GC. 2000. Microbial populations on animal hides and beef carcasses at different stages of slaughter in plants employing multiple-sequential interventions for decontamination. Journal of Food Protection 63:1080–1086.
Bell RG. 1997. Distribution and sources of microbial contamination on beef carcasses. Journal of Applied Microbiology 82:292–300.
Bolton DJ, Doherty AM, Sheridan JJ. 2001. Beef HACCP: Intervention and non-intervention systems. International Journal of Food Microbiology 66(1-2):119–129.
Dorsa WJ. 1997. New and established carcass decontamination procedures commonly used in the beef-processing industry. Journal of Food Protection 60(9):1146–1151.
Dorsa WJ, Cutter CN, Siragusa GR. 1997. Effects of acetic acid, lactic acid and trisodium phosphate on the microflora of refrigerated beef carcass surface tissue inoculated with Escherichia coli O157:H7, Listeria innocua, and Clostridium sporogenes. Journal of Food Protection 60(6):619–624.
Elder RO, Keen JE, Siragusa GR, Barkocy-Gallagher GA, Koohmaraie M, Laegreid WW. 2000. Correlation of enterohemorrhagic Escherichia coli O157 prevalence in feces, hides, and carcasses of beef cattle during processing. Proceedings of the National Academy of Sciences 97(7):2999–3003.
FSIS (Food Safety and Inspection Service). 1994. Nationwide Beef Microbiological Baseline Data Collection Program: Steers and Heifers—October 1992–September 1993.
Galland JC. 1997. Risks and prevention of contamination of beef carcasses during the slaughter process in the United States of America. Revue scientifique et technique (International Office of Epizootics) 16(2):395–404.
Gill CO. 1999. HACCP by guesswork or by the numbers? Food Quality (January/February). pp. 28–32.
Gill CO, Badoni M, Jones T. 1996a. Hygienic effects of trimming and washing operations in a beef-carcass-dressing process. Journal of Food Protection 59(6):666–669.
Gill CO, McGinniss J, Badoni M. 1996b. Assessment of the hygienic characteristics of a beef carcass dressing process. Journal of Food Protection 59(2):136–140.
Gill CO, Badoni M, McGinnis JC. 1999. Assessment of the adequacy of cleaning of equipment used for breaking beef carcasses. International Journal of Food Microbiology 46(1):1–8.
Gill CO, McGinnis JC, Bryant J. 2001. Contamination of beef chucks with Escherichia coli during carcass breaking. Journal of Food Protection 64(11):1824–1827.
Grau FH. 1987. Prevention of microbial contamination in the export beef abattoir. In: Smulders FJM, ed. Elimination of Pathogenic Organisms from Meat and Poultry. Amsterdam: Elsevier Press, pp. 221–223.
Graves Delmore LR, Sofos JN, Reagan JO, Smith GC. 1997. Hot-water rinsing and trimming/washing of beef carcasses to reduce physical and microbiological contamination. Journal of Food Science 62:373–376.
Graves Delmore LR, Sofos JN, Schmidt GR, Smith GC. 1998. Decontamination of inoculated beef with sequential spraying treatments. Journal of Food Science 63:890–893.
Newton KG, Harrison JCL, Wauters AM. 1978. Sources of psychrotrophic bacteria on meat at the abattoir. Journal of Applied Bacteriology 45:75–82.
Nutsch AL, Phebus RK, Riemann MJ, Schafer DE, Boyer JE Jr, Wilson RC, Leising JD, Kastner CL. 1997. Evaluation of a steam pasteurization process in a commercial beef processing facility. Journal of Food Protection 60:485–492.
Nutsch AL, Phebus RK, Riemann MJ, Kotrola JS, Wilson RC, Boyer JE Jr, Brown TL. 1998. Steam pasteurization of commercially slaughtered beef carcasses: evaluation of bacterial populations at five anatomical locations. Journal of Food Protection 61:571–577.
Phebus RD, Nutsch AL, Schafer DE, Wilson RC, Riemann MJ, Leising JD, Kastner CL, Wolf JR, Prasai RK. 1997. Comparison of steam pasteurization and other methods for reduction of pathogens on surfaces of freshly slaughtered beef. Journal of Food Protection 60:476–484.
Ransom JR, Belk KE, Bacon RT, Sofos JN, Scanga JA, Smith GC. 2002. Comparison of sampling methods for microbiological testing of beef animal rectal/colonal feces, hides and carcasses. Journal of Food Protection 65(4):621–626.
Reagan JO, Acuff GR, Buege DR, Buyck MJ, Dickson JS, Kastner CL, Marsden JL, Morgan JB, Nickelson R II, Smith GC, Sofos JN. 1996. Trimming and washing of beef carcasses as a method of improving the microbiological quality of meat. Journal of Food Protection 59:751–756.
Samelis J, Sofos JN, Kendal PA, Smith GC. 2001a. Influence of the natural microbial flora on acid tolerance response of Listeria monocytogenes in a model system of fresh meat decontamination fluids. Applied and Environmental Microbiology 67:2410–2420.
Samelis J, Sofos JN, Kendall PA, Smith GC. 2001b. Fate of Escherichia coli O157:H7, Salmonella Typhimurium DT104 and Listeria monocytogenes in fresh meat decontamination fluids at 4 and 10°C. Journal of Food Protection 64:950–957.
Samelis J, Sofos JN, Kendall PA, Smith GC. 2002a. Effect of acid adaptation on survival of Escherichia coli O157:H7 in meat decontamination washing fluids, and potential effects of organic acid interventions on the microbial ecology of the meat plant environment. Journal of Food Protection 65(1):33–40.
Samelis J, Sofos JN, Ikeda JS, Kendall PA, Smith GC. 2002b. Exposure to water meat decontamination washing fluids sensitizes Escherichia coli O157:H7 to organic acids. Letters in Applied Microbiology 34:7–12.
Scanga JA, Grona AD, Belk KE, Sofos JN, Bellinger GR, Smith GC. 2000. Microbiological contamination of raw beef trimmings and ground beef. Meat Science 56:145–152.
Smulders FJM, Greer GG. 1998. Integrating microbial decontamination with organic acids in HACCP programmes for muscle foods: Prospects and controversies. International Journal of Food Microbiology 44:149–169.
Sofos JN, Smith GC. 1998. Nonacid meat decontamination technologies: Model studies and commercial applications. International Journal of Food Microbiology 44:171–188.
Sofos JN, Belk KE, Smith GC. 1999a. Processes to reduce contamination with pathogenic microorganisms in meat. Proceedings of the 45th International Congress of Meat Science and Technology (Japan). Vol. II, pp. 596–605.
Sofos JN, Kochevar SL, Bellinger GR, Buege DR, Hancock DD, Ingham SC, Morgan JB, Reagan JO, Smith GC. 1999b. Sources and extent of microbiological contamination of beef carcasses in seven United States slaughtering plants. Journal of Food Protection 62:140–145.
Sofos JN, Kochevar SL, Reagan JO, Smith GC. 1999c. Extent of beef carcass contamination with Escherichia coli and probabilities of passing U.S. regulatory criteria. Journal of Food Protection 62:234–238.
Sofos JN, Kochevar SL, Reagan JO, Smith GC. 1999d. Incidence of Salmonella on beef carcasses relating to the U.S. meat and poultry inspection regulations. Journal of Food Protection 62:467–473.
Van Donkersgoed J, Jericho KWF, Grogan H, Thorlakson B. 1997. Preslaughter hide status of cattle and the microbiology of carcasses. Journal of Food Protection 60:1502–1508.