Contamination Revealed by Indicator Microorganism Levels during Veal Processing

Each year, approximately 100 million lb (ca. 450,000 kg) of veal are consumed in the United States. Veal is the meat of calves or young cattle in contrast to beef, which is the meat from older cattle. Most veal comes from male calves of dairy cattle breeds, but some beef calves enter the food chain directly as veal as well. There are generally two types of veal calves harvested in the United States. Bob veal calves are calves sold and slaughtered within the first few days of life and represent about 30% of the calf numbers slaughtered. A calf must be younger than 3 weeks of age and less than 150 lb to be regarded as bob veal (12). The other category of veal is formula-fed veal, which are calves usually separated from cows within 3 days after birth and raised on a milk replacer diet until about 20 weeks of age, weighing approximately 450 lb (12). This diet prevents the calf from developing a functioning rumen and maintains a lightly colored meat (7, 10). Since June 2012, the U.S. Department of Agriculture, Food Safety Inspection Service (FSIS) has identified a higher percentage of non-O157 Shiga toxin–producing E. coli compared with Escherichia coli O157:H7–positive samples collected from veal trimmings than from products produced from other cattle slaughter classes (14, 15). Thus, FSIS began separately reporting non-O157 Shiga toxin–producing E. coli verification results of veal samples from other raw beef samples (14, 15).

The reason behind the higher positive rate of non-O157 Shiga toxin–producing E. coli trim testing results in veal products relative to beef is unclear. Veal processing may be more challenging than beef processing due to the differences in the hides that must be removed during processing (15). Veal hides, especially those of bob veal, have much thinner fascia and are insubstantial compared with beef cattle hides. Their removal requires additional contact and handling that often results in cross-contamination (13). Indeed, hide removal is the primary contributor to carcass contamination in beef processing (2, 3, 9). Further, FSIS has released directives (11) and guidance documents (14) directly related to sanitary dressing deficiencies that were identified at veal slaughter establishments. In response to the recognition of increased contamination rates during veal harvest, we visited five veal processors that harvest formula-fed veal or bob veal or both and collected samples to assess general processing and contamination rates. After our initial sample collection, the veal processors implemented additional interventions and processing practices; therefore, follow-up samples were collected to appraise the impact of these changes.

Design. Five veal processors (designated A through E) that harvest formula-fed veal (n = 3), bob veal (n = 1), or both (n = 1) were visited for 1 day each in December of 2013 and represent the initial samples. The selected veal processors harvested approximately 300 to 500 veal calves in a day. Hides, preintervention carcasses, and final carcasses were sampled for bacterial indicators of contamination, which include aerobic plate count bacteria (APC), Enterobacteriaceae, E. coli, and coliforms (CF). In December of 2014, plants B, C, and D (two formula-fed veal and one bob veal) were revisited for the follow-up samples to assess the impact of newly implemented interventions and processing practices on contamination rates and bacterial transfer.

Sample collection. Hide samples were collected by using Speci-Sponges (Nasco, Fort Atkinson, WI) that were moistened with 20 ml of Difco buffered peptone water (BD, Sparks, MD) in Whirl-Pak bags (Nasco). Hide samples (n = 90 to 96 at plants A, B, C, D, and E, with an additional 48 collected at plant E) were collected before any hide-directed interventions were applied and generally after stunning and exsanguination at plants A, B, C, and E. However, at plant D, it was not physically possible to collect hide samples from stunned calves before a hide-directed wash was applied. Each hide sample was obtained by swabbing a 600-cm² area over the breast and plate region. A sample consisted of approximately five vertical and five horizontal passes (up and down or side to side was considered one pass), with the sponge flipped to the other side after the first five passes. After collection, each sponge was sealed in the Whirl-Pak bag, stored in an insulated cooler that contained ice packs, and then transported to the laboratory and processed the following day.

Preintervention carcass samples were collected as soon as safely accessible following complete removal of the hide and collected before any subsequent interventions or knife trimming occurred on the preintervention carcass. Preintervention carcass samples were not directly matched to hide samples but were collected during the same time frame as the hide samples. The carcass samples (n = 90 to 96 at plants A, B, C, D, and E, with an additional 48 collected at plant E) were collected from the brisket-plate and hock-round areas by using Speci-Sponges in Whirl-Pak bags (Nasco) wetted with 20 ml of buffered peptone water by using 10 bidirectional strokes of the sponge, which was turned over halfway through the process. For formula-fed veal preintervention carcasses, one sponge was used to collect the sample from an area of 6,000 ± 100 cm². Bob veal preintervention carcasses were smaller and not as uniform in size as formula-fed veal; therefore, samples were collected from an area of 2,000 ± 200 cm². After collection, each sponge was sealed in the Whirl-Pak bag, stored in an insulated cooler containing ice packs, and then transported to the laboratory and processed the following day.

Final carcass samples were collected from chilled or chilling carcasses after they entered the final cooler. Final carcasses were not directly matched to preintervention carcasses but were from carcasses mostly processed the same day as the preintervention carcasses and generally during the same time frame. In some cases, the final carcass samples included carcasses that were from the previous day. Samples for final carcasses (n = 90 to 96 at plants A, B, C, D, and E, with an additional 48 collected at plant E) were collected in the same fashion and from the same surface areas as the preintervention carcasses, as described previously. The interventions applied to carcasses varied by plant and consisted of differing combinations of lactic acid spay, hot water wash, and knife trimming. The lactic acid spray was 2 or 4%, depending on the plant, and usually applied by hand by using a wand attached to a pump sprayer. Hot water (75 to 85°C) washes were mostly applied by hand spraying from the hose with nozzle. Knife trimming focused on the removal of visible contamination.

Sample processing. All sponge samples were hand massaged thoroughly in the bags, and an aliquot (1.0 ml from hide samples and 4.0 ml from preintervention and final carcasses samples) was removed from each sample bag and transferred to a test tube for enumeration of the indicator organisms.

Measurement of indicator organisms. All samples were serially diluted to an acceptable range and plated to Petrifilm (3M Microbiology, St. Paul, MN) for enumeration of APC, Enterobacteriaceae, CF, and E. coli, following the manufacturer's instructions. Indicator counts were determined by using a Petrifilm Plate Reader (3M Microbiology). Values were log transformed to CFU/100 cm². As used in this study, the level of detection (LOD) for indicator organisms on hides was 3.6, 0.7, 2.5, and 2.5 log CFU/100 cm² for APC, Enterobacteriaceae, CF, and E. coli, respectively. The LOD on preintervention carcasses was 1.8 log CFU/100 cm² for APC and −0.4 log CFU/100 cm² for Enterobacteriaceae, CF, and E. coli. The LOD for all organisms on final carcasses was −0.4 log CFU/100 cm². The LOD was equivalent to one colony in the most concentrated dilution of sample tested. Hide-to-carcass transfer (HTCT) ratios were calculated by dividing preintervention carcass APC log CFU by hide APC log CFU. Carcass intervention efficacy (CEI) coefficients were calculated by dividing final carcass APC log CFU and by preintervention carcass APC log CFU.

Statistical analysis. Analysis of variance of the mean levels of indicator organisms (log transformed) for each sample type by plant and veal type (formula fed and bob) was performed by using GraphPad Prism version 5.03 (GraphPad Software, La Jolla, CA). Specifically, the nonparametric data were analyzed using a one-way analysis of variance, with a Bonferroni multiple comparison posttest and significance level of 0.05. Samples with indicator bacteria levels below the LOD were not included in the analysis.

Rates of contamination measured during initial sample collection. Increased attention and scrutiny of veal processing and veal products began in 2012, as FSIS reported a higher percentage of non-O157 Shiga toxin–producing E. coli compared with E. coli O157:H7–positive samples collected from veal products than from beef products. To investigate this observation, we visited five veal processors and examined indicator organisms on hides, preevisceration carcasses, and final carcasses to assess levels of sanitary dressing procedure in veal processing plants.

Because the hide is the general source of contamination in beef processing, we examined veal hides but did not find any differences in the concentrations of indicator organisms from previous reports of cattle hides. The APCs from veal hides ranged from 6.50 to 7.48 log CFU/100 cm² (Table 1). A previous study of veal calf hides at harvest found APC of 7 to 8 log CFU/100 cm² before any hide interventions were applied (15). Comparatively, cull cows have been observed to have hide APC concentrations ranging from 6.17 log CFU/100 cm² in the winter, similar to the samples described herein, to 8.19 log CFU/100 cm² in the summer (6). Significant differences in APCs were observed between the veal plants and is a common observation (4, 6). Cattle hide APC can be highly variable, depending on the location and group of cattle examined. Levels as low as 4.5 log CFU/100 cm² of hide (5) to as high as 9.0 log CFU/100 cm² of hide (1) have been reported. The Enterobacteriaceae, CF, and E. coli concentrations on hides were likewise not different from other previous reports of hide levels of contamination. From this, we conclude that the level of contamination on veal calf hides is not introducing an unexpected or overwhelming level of APC contamination that cannot be controlled by proper sanitary dressing procedures or by commonly used processing interventions.

For the other indicator organism levels (Enterobacteriaceae, CF, and E. coli) on hides, those at plant D were significantly lower than the levels at other plants. This was most likely due to the presence of a hide-directed intervention that did not physically allow samples to be collected from stunned veal calves before its application. Therefore, the levels of Enterobacteriaceae, CF and E. coli on hides at plant D need to be interpreted as such. However, this still represents the level of bacteria on hides at this plant before hide removal and still offers the opportunity to evaluate cross-contamination during hide removal at this plant. We did not focus on the effects of hide interventions at these plants, as that has been reported previously (15).

Preevisceration carcass contamination is mostly due to cross-contamination during hide removal (9). The levels of indicator organisms found during the initial sample collection supports this, as there was generally a linear relationship between the hide levels and preintervention carcass levels of APC, Enterobacteriaceae, CF, and E. coli. The levels of APC on preintervention carcasses ranged from 3.08 to 5.22 log CFU/100 cm², while Enterobacteriaceae, ranged from 1.16 to 3.47 log CFU/100 cm². CF and E. coli, subpopulations within the Enterobacteriaceae, were similar to the Enterobacteriaceae, ranging from 0.21 to 3.47 log CFU/100 cm² and −0.07 to 3.10 log CFU/100 cm² for CF and E. coli, respectively. There were significant differences among plants, wherein indicator organisms on preintervention carcasses at each plant were different (P < 0.05) from most of the other plants. A number of reports have described beef preevisceration carcasses, before interventions were applied, to have APC in the ranges of 3.0 to 4.7, 6.1 to 6.4, 4.5 to 5.3, and 6.1 to 9.1 log CFU/100 cm² differing by the day, group of cattle, and processing plant sampled (1, 2, 4, 5). Preevisceration carcasses of cull cows at harvest from four different processors were found to have mean APC ranging from 4.18 to 5.39 log CFU/100 cm², and vary from 4.24 log CFU/100 cm² in the winter to 6.47 log CFU/100 cm² in the summer (6, 8). The mean APC of veal preintervention carcasses is not remarkably different from reports of beef carcasses following hide removal. A similar conclusion can be drawn for Enterobacteriaceae, CF, and E. coli of preintervention veal carcasses as well.

HTCT ratios are indirect measurements of the efficacy of dressing procedures and indicate how well a processor prevents cross-contamination from the hide to the carcass during the hide removal process (6). Tracking APC levels from hides to preintervention carcasses at plants A and C showed that these two plants with the lowest preintervention carcass APC, had the lowest HTCT ratios as well (Table 2). In contrast, plant D had hide APC levels similar to plant C, but preintervention carcass APC levels 1.6 log higher than that at plant C, suggesting a less optimal hide removal process. As mentioned previously, hides at plant D had significantly lower levels of Enterobacteriaceae, CF, and E. coli compared with the other plants, yet plant D demonstrated the greatest level of transfer of these organisms to the carcass during hide removal (HTCT ratio 0.72).

Levels of indicators on final carcasses also were variable. Final veal carcasses in the cooler had mean APC ranging from 0.36 to 3.04 log CFU/100 cm². In comparison, APC on final chilled beef carcasses have been reported to range from 0.9 to 2.0 log CFU/100 cm², depending on the plant and day samples were collected (1). Other studies have reported APC on beef carcasses after chilling, at different beef processing plants ranged from 2.3 to 5.3 log CFU/100 cm² (2). More recent results of Canadian beef and cull cow processing have reported final chilled carcasses to have 4.38 and 2.89 log CFU/100 cm², respectively (3, 8). Likewise Enterobacteriaceae of chilled veal carcasses ranged from −0.21 to 1.78 log CFU/100 cm². Final carcasses at plants B and E had the highest Enterobacteriaceae, CF, and E. coli levels measured. The Enterobacteriaceae of these veal carcasses were significantly higher than Enterobacteriaceae counts of final carcasses at the other veal plants and higher than Enterobacteriaceae levels reported for chilled final beef carcasses, which ranged from 0.2 to 0.7 log CFU/100 cm² (1). When considering studies of beef and cull cow final carcasses that reported CF and E. coli, the ranges found on veal are not out of line and are somewhat lower. Final beef carcasses have been reported to have CF of 0.9 to 1.3 log CFU/100 cm² and 1.63 log CFU/100 cm² (2, 3), while cull cow final carcasses had CF of 2.02 log CFU/100 cm² (8). E. coli counts on beef were found to range from 0.9 to 1.45 log CFU/100 cm² (2, 3), while counts on cull cow carcasses were found to be 2.01 log CFU/100 cm² (8).

The number of carcass samples that were enumerable in our study illustrates that many of the lowest mean values for the indicators were based on a fraction of the samples. The LOD of Enterobacteriaceae, CF, and E. coli on preintervention and final carcasses was −0.4 log CFU/100 cm² due to the initial volume (20 ml), area (2,000 and 6,000 cm²) of the sample, and the dilution prepared prior to plating. The number of samples that were enumerable with cell counts above the LOD are shown in Table 1. At plant A, 79, 66, and 27% of preintervention carcass samples had enumerable Enterobacteriaceae, CF, and E. coli, respectively. Plants C and D also had significantly fewer enumerable final carcass samples than the other plants (P < 0.05), with only 5 and 6% of samples having enumerable E. coli present.

Similar to HTCTs, CIE coefficients can be calculated to indicate how effectively residual contamination has been removed by carcass directed antimicrobial interventions. CIE coefficients can be calculated as the ratio of final carcass APC log CFU to preintervention carcass APC log CFU. These values reflect the efficacy of the processing interventions in place at each plant in our study (Table 2). Most plants demonstrated a significant reduction in APC, Enterobacteriaceae, CF, and E. coli from preintervention to final carcass. Four of the five plants showed APC reductions that ranged from 1.84 to 3.21 log CFU/100 cm², Enterobacteriaceae from 1.13 to 1.88 log CFU/100 cm², while CF and E. coli reductions ranged from 0.35 to 1.98 log CFU/100 cm² with CIE coefficients ranging from 0.62 to 0.12. However, reductions between preintervention and final carcasses at plant A were all less than 1 log CFU/100 cm², and the CIE coefficient at this plant was 0.81. This suggests that although plant A appeared to have one of the best hide removal techniques (low HTCT) that prevented carcass contamination, efficacy of interventions applied to the carcasses postintervention had little impact (high CIE). Thus, the microbiological quality of plant A final carcasses ended up third best among the five plants. During FSIS on-site reviews of veal processing establishments, they noted a common deficiency in which antimicrobial interventions failed to achieve proper carcass coverage owing to the practice of suspending veal carcasses from the rail system with both hind limbs on a single hook (13). During our sample collection, this practice was not observed at any of the five plants visited. However, it was observed that most of the veal processors applied antimicrobial interventions by hand using a pump sprayer and wand. Applying antimicrobials in this fashion can result in incomplete coverage of a carcass or inconsistent coverage from carcass to carcass.

From the initial sample collections, we concluded that although veal calves at harvest are not substantially different in the levels of indicator bacteria on their hides, greater levels of cross-contamination of the carcass occurred during hide removal, and final carcasses at some plants had greater levels of contamination than has been reported for fed beef carcasses.

Rates of contamination measured during follow-up sample collection. After our initial sample collection, the involved veal processors implemented a number of changes in their harvest processes that included different combinations of new interventions, additional training of personnel, and modifications of current interventions. The changes followed the best practices described by FSIS (14) and included increased attention to sanitary dressing procedures and the introduction of additional applications of 4% lactic acid or 400 ppm of peroxyacetic acid to carcasses after hide removal. In some cases, the new interventions were applied by hand using wands attached to pump sprayers, and in other cases, automated spray bars were installed. The additional intervention sprays were applied to carcasses as surfaces were exposed during hide removal, before or after evisceration or both, and when entering the chiller. Because these changes were substantial, we decided a follow-up and sample collection was warranted to determine the current status of veal processing and to provide feedback to the processors on the efficacy of the changes made. We were able to revisit three of the original five plants, plants B, C, and D.

The levels of indicator organisms on hides for the follow-up samples collected 12 months later (Table 3) showed that incoming load on hides was variable. For plant B, nearly a full log less APC was observed, while plants C and D hides had a 1.3 and 1.5 log higher load APC, respectively. Despite the increased incoming load of APC at plants C and D, the APC levels on preintervention carcasses at these two plants, as well as plant B, were lower compared with the initial sample collection. HTCT ratios showed each plant had improved. The follow-up samples, however, did not identify any significant changes in the efficacy of interventions used on carcasses between preintervention and the final cooler. No significant difference between APC, Enterobacteriaceae, CF, and E. coli was observed on final carcasses between the two sample collection trips. CIE coefficients were unchanged or slightly increased at each plant. Note, however, that lower microbial load on preintervention carcasses results in lower loads on final carcasses, so the improved hide removal had an impact on final carcass levels.

Recently, FSIS released sanitary dressing guidance documents specifically targeted to veal processors. This information and additional actions by veal processors appears to have resulted in improved hide removal and sanitary dressing techniques, as measured by indicator bacteria levels. However, our data suggested more attention needs to be applied to carcass interventions. Veal carcasses are smaller, and in the cases of bob veal, often quite variable in size compared with beef carcasses. Spray cabinet designs and the application of antimicrobials with hand wands and sprayers need to be optimized for veal harvest to ensure uniform and consistent delivery. During our observations, we noted that in automated wash cabinets, veal processors must apply spray washes at a lower pressure than is used for beef, because the smaller veal carcasses easily swing back and forth under these pressure washes and can fall from the rail system.

Differences in contamination between bob veal and formula-fed veal processing. As mentioned previously, bob veal and formula-fed veal calves differ considerably at harvest in size and age, as well as the production origin environment. During our initial sample collection, we visited plants that harvested bob veal and that harvested formula-fed veal. One plant harvested both types of veal and offered the opportunity to collect samples from both types of veal in one location to address potential plant-to-plant variation in the comparison of veal types. Our follow-up sample collection included one bob veal and two of the formula-fed veal plants. To analyze differences that may be present between these two types of veal processing, we aggregated all the results from bob veal and all the results from formula-fed veal plants at both collection times (Table 4). With the exception of hide APC, all indicator counts were higher on bob veal hides, preintervention carcasses and final carcasses than those on formula-fed veal. Only hide Enterobacteriaceae were not significantly different (P > 0.05) between the two groups of veal. Indicator bacteria levels on bob veal preintervention carcasses were generally about 2 log CFU/100 cm² higher than those on formula-fed veal carcasses. This is a function of a less effective hide removal process in bob veal that had an HTCT of 0.66, while formula-fed veal had a HTCT of 0.51. Although bob veal hides had greater levels of contamination, the difference was still within half a log of formula-fed veal hides. Hide removal from bob veal carcasses likely poses challenges for processors, and an effective hide-directed intervention should be a priority for bob veal processors (15). Final carcasses of formula-fed veal had indicator bacteria levels about 1 to 1.5 log lower than bob veal final carcasses. However, the CIE coefficients for bob veal and formula-fed veal were nearly identical, 0.44 and 0.45, respectively. Thus, the difference between the two groups final carcasses was owing to the lower starting load on the preintervention carcasses.

Examining the bob and formula-fed veal calves harvested at the same plant on the same day also indicated that bob veal was more challenging to process with minimal contamination than formula-fed veal (Table 5). At this plant, there were significant differences in APC and Enterobacteriaceae, CF, and E. coli on hides of the two groups. Formula-fed veal had higher (P < 0.05) APC, while bob veal had higher (P < 0.05) Enterobacteriaceae, CF, and E. coli on their hides. While the transfer of APC during hide removal was about the same (HTCTs of 0.69 and 0.71), the transfer of Enterobacteriaceae, CF, and E. coli was about 70% greater on bob veal preintervention carcasses than formula-fed veal carcasses following hide removal. The subsequent interventions at this plant reduced the final carcass contamination by about 2 log CFU/100 cm² compared with the preintervention carcass levels. Because formula-fed veal carcasses started with less contamination than bob veal carcasses at preintervention, the formula-fed veal final carcasses were often below the LOD of Enterobacteriaceae, CF, and E. coli Clearly, greater efforts are needed during the removal of bob veal hides and effective bob veal hide–directed interventions.

In conclusion, we visited five veal processors to assess contamination during veal harvest. We found that, in general, veal calves at harvest are not remarkably different than fed beef and cull cows at harvest in regard to levels of indicator organisms. A follow-up visit to three of the plants found that hide removal dressing practices had improved with training of personnel. However, our data showed that veal processors need to continue addressing carcass-applied interventions, in particular, using techniques that ensure uniform and consistent delivery. Further, bob veal was found to have greater levels of indicator organisms than formula-fed veal. Because veal was singled out by FSIS, the control of contamination during processing has started to improve, but the nature and processing of this group of animals still presents challenges to processors.

The authors thank the cooperating veal processing plants for access to sample collection, the American Veal Association and its members for their contributions to this project, Greg Smith, Lawnie Luedtke, Bruce Jasch, and Frank Reno for technical support, and Jody Gallagher for secretarial assistance. Trade names are necessary to accurately report experimental results. The use of trade names by a U.S. Department of Agriculture author implies no approval of the product to the exclusion of others that may also be suitable; further, the USDA neither guarantees nor warrants the standard of any products mentioned.