A total of 180 lamb carcasses and 200 inert surfaces were sampled in two commercial abattoirs (plants A and B) from northwest Spain. A higher (P < 0.001) average microbial load (log CFU per square centimeter) on lamb carcasses was observed for total viable counts (TVC; 2.74 ± 1.15) than for Enterobacteriaceae (2.21 ± 1.16). Different microbial counts were found on carcasses from plants A and B, both for TVC (2.56 ± 0.96 versus 3.18 ± 1.47, respectively; P < 0.001) and Enterobacteriaceae (2.09 ± 0.97 versus 2.50 ± 1.61, respectively; P < 0.05). High correlations (P < 0.001) were observed for TVC and Enterobacteriaceae in both plants A (r = 0.708) and B (r = 0.912). The percentages of unsatisfactory daily mean log values for carcasses, according to European Union Regulation (EC) No 2073/2005, were 0.0 (TVC) and 30.8 (Enterobacteriaceae) in plant A and 10.0 (TVC) and 40.0 (Enterobacteriaceae) in plant B. Average counts for inert surfaces were all lower than 10 CFU/cm2 (TVC) or 1 CFU/cm2 (Enterobacteriaceae). The need to improve hygienic practices in order to adhere to the European Union microbiological performance criteria is emphasized. The detected different microbial counts between slaughterhouses could be attributed to differences in external hygiene of livestock and in the number of slaughterhouse workers. Microbiological analysis of carcasses and surfaces allows detection of hygienic concerns in the overall process.
It is of prime importance that consumers receive safe, wholesome products prepared under the best hygienic and sanitary conditions. Because microbiological hazards are undetectable by conventional veterinary meat inspection, European Union (EU) authorities are mandating the implementation of hazard analysis critical control point (HACCP) systems for abattoirs to ensure the microbiological quality of red meat and poultry carcasses (Regulation [EC] No 853/2004) (8).
Microbiological testing is desirable for the implementation and maintenance of effective HACCP procedures. The main use of microbiological testing is to verify that identified critical control points are successfully being controlled. Microbiological analysis must be used to investigate the effects of operations on general hygiene, to validate procedures adopted for controlling microbiological contamination, and to verify maintenance of control over the microbiological condition of products (3).
Commission Regulation (EC) No 2073/2005, on microbiological criteria for foodstuffs, requires that operators of meat establishments conduct regular checks on the general hygienic condition of production. The checks include microbiological testing of total viable counts (TVC) and Enterobacteriaceae on carcasses at abattoirs. This regulation also establishes microbiological performance criteria for the measurement of process hygiene (9).
Because there is a lack of information about the application of this rule in northwest Spain, the aims of this work were (i) to determine the microbiological profile of lamb carcasses and surfaces in two commercial abattoirs to assess whether the established criteria are being met and (ii) to contribute to the evaluation of the feasibility and diagnostic value of microbiological testing in routine abattoir HACCP systems monitoring.
MATERIALS AND METHODS
Samples were obtained from two abattoirs in northwest Spain that process 400 (plant A) and 100 (plant B) lamb carcasses per day. Both abattoirs undertake inverted dressing in a nonmechanized rail and do not have automatic pelt removal. The number of workers for slaughter and dressing ranged from four to five (plant A) and from one to two (generally one; plant B). Both plants comply with EU standards for construction and hygienic operation of abattoirs, and refrigeration capacity is adequate for the average daily slaughter volume.
Carcass and surface sampling was carried out throughout 26 (plant A) or 10 (plant B) weeks, according to Regulation (EC) No 2073/2005. The day of sampling was changed each week to ensure that each day of the week was covered. Five carcasses were tested on each sampling day, for a total of 180 carcasses (130 at plant A and 50 at plant B). Sampling was carried out after the completion of carcass dressing but before the start of chilling. Samples were taken halfway through the slaughter day. For each carcass, four separate tissue samples, each 5 cm2, were collected with a sterile scalpel and a template. The sample sites were flank, lateral thorax, brisket, and breast. Samples from each carcass were pooled and were placed in a separate stomacher bag, which was then transported in an ice chest (2 to 4°C) to the laboratory for testing, usually within 6 h, but never after 12 h, of being collected.
For surface monitoring, the following points were chosen: bleeding knives, flap doors, walls, hooks, knives used for skinning and evisceration, aprons, hand washers, external surface of sterilization devices for knives, trolleys, baskets, cold room door, and cold room walls. The internal surface of the knife tray was also tested in plant A. A total of 130 samples (10 for each tested surface) were taken at plant A, and 70 samples (five to six for each surface) were taken at plant B. Sampling was carried out in the morning, before the start of the work, on surfaces that had been cleaned and disinfected the previous day. Samples were taken throughout a total of 26 (plant A) or 10 (plant B) weeks (one sampling day per week). On each sampling day, five (plant A) or seven (plant B) determinations were carried out using the agar plate contact method (33.2-cm2 surface area; RODAC plates, BD Diagnostics, Baltimore, MD) on plate count agar for aerobic viable counts (Oxoid Ltd., Hampshire, England) or violet red bile glucose agar for Enterobacteriaceae (Oxoid). Plates were transported to the laboratory under chilled conditions (2 to 4°C).
Samples from each carcass were blended for 2 min in a Masticator (IUL, Barcelona, Spain) using 100 mL of maximum recovery dilution medium, composed of 0.1% peptone (Oxoid) and 0.85% sodium chloride solution (Panreac, Barcelona, Spain). Subsequent decimal dilutions were carried out using the same diluent. The pour plate technique was used to prepare duplicate plates, which were incubated under aerobic conditions at 30°C for 48 h for total aerobic counts (plate count agar, Oxoid) or at 37°C for 24 h for Enterobacteriaceae (violet red bile glucose agar, Oxoid). These conditions of incubation (time-temperature) were also used for RODAC plates.
Analysis of data
Laboratory data were recorded as CFU per square centimeter for each tested carcass (samples from each carcass were pooled) using this equation: CFU/cm2 = (A × V)/(S × D), where A = average CFU per plate; V = 100 (volume of original suspension); S = 20 cm2 (total surface area); and D = dilution factor. The limits of detection for both total viable and Enterobacteriaceae counts for carcasses and surfaces were 2.50 and 0.03 CFU/cm2, respectively.
Bacterial counts for each carcass were transformed to log values prior to calculation of mean log values. The daily mean of the log values was calculated as the arithmetic average of the counts (log CFU per square centimeter) for all carcasses sampled on a given day. Mean log values at plants A and B were compared for statistical significance using analysis of variance and Tukey's honest significant difference test. Regression analysis was used to examine the relationship between TVC and Enterobacteriaceae. Results for inert surfaces were recorded as CFU per square centimeter of surface area, and average counts referred to samples within a period of 2 weeks. Fisher's exact test was used to compare the prevalence of satisfactory results at each slaughterhouse. The tests were carried out using the Statistica 8.0 software package (Statsoft Ltd., Tulsa, OK).
RESULTS AND DISCUSSION
Microbial counts on lamb carcasses
Both TVC and Enterobacteriaceae were recovered from all carcass samples. Average microbial counts were 2.74 ± 1.15 log CFU/cm2 (TVC) and 2.21 ± 1.16 log CFU/cm2 (Enterobacteriaceae) (P < 0.001). The TVC found for lamb carcasses in the present study, in which the excision method was used, appear to be lower than those from other surveys (17, 22), although for most of those works, the swabbing technique was used for sampling. As previously reported (6, 11), excision gave higher counts than nondestructive sampling methods. However, the favorable comparison obtained may be partially due to the fact that the samples of this research were taken immediately after slaughter and dressing, whereas in most studies, carcasses are kept for several hours at refrigeration temperature before sampling. It has been observed that storage time has a significant influence on microbial loads on meat (5). Moreover, data from different authors should be compared with caution because the differences in procedures for enumerating bacteria at different laboratories would significantly affect the number of bacteria recovered, as previously suggested (1, 4).
Counts from plant A were lower than those from plant B, both for TVC (2.56 ± 0.95 versus 3.18 ± 1.47 log CFU/cm2, respectively; P < 0.001) and Enterobacteriaceae (2.09 ± 0.97 versus 2.50 ± 1.61 log CFU/cm2, respectively; P < 0.05). Large differences in the microbiological conditions of the carcasses produced at different abattoirs have also been reported by other authors (11, 22). However, significant differences among establishments should be interpreted with caution because differences in the number of bacteria on carcasses observed among visits to any one abattoir could be as great as, or even greater than, the differences among abattoirs (15).
Significant differences (P < 0.001) between microbial groups were found in both plants A and B. The mean Enterobacteriaceae counts were 33.9% of TVC in plant A and 20.9% in plant B. There was a strong correlation (P < 0.001) between TVC and Enterobacteriaceae in both plants A (r = 0.708) and B (r = 0.912), which suggests that the Enterobacteriaceae constituted a major proportion of TVC in most cases.
A wide range of carcass bacterial counts were obtained. Thus, contamination loads for individual carcasses ranged from 0.70 to 5.88 log CFU/cm2 (plant A) and from 0.70 to 6.19 log CFU/cm2 (plant B) for TVC. Data for Enterobacteriaceae were 0.70 to 5.10 log CFU/cm2 (plant A) and 0.70 to 5.54 log CFU/cm2 (plant B). Figures 1 and 2 show carcass distribution according to counts obtained. This heterogeneity in the contamination loads among lamb carcasses has been observed by other authors (19).
By EU Regulation criteria, contamination of carcasses with TVC for a sampling day is defined as unsatisfactory if the daily mean log is higher than M, as acceptable if the daily mean log is between m and M, and as satisfactory if the daily mean log is lower than m. The microbiological limits M and m are set at 5.0 and 3.5 log CFU/cm2 for TVC and 2.5 and 1.5 log CFU/cm2 for Enterobacteriaceae, respectively (9). Percentages of satisfactory, acceptable, and unsatisfactory log mean values for plants A and B are shown in Table 1. A lower percentage of satisfactory results was obtained from plant B for TVC (100 versus 40%; P < 0.001). On the other hand, similar (P > 0.05) percentages of satisfactory values were observed in plants A and B for Enterobacteriaceae.
Evolution of counts throughout the survey period is shown in Figures 3 (plant A) and 4 (plant B). The EU rules do not state the number of acceptable samples that are permitted to be defective before corrective action is required; however, in practice, corrective action is required after two consecutive acceptable results (14). In the present study, no corrective actions at plant A were required, taking into account criteria for TVC. By contrast, corrective actions were required based on TVC in plant B and, based on Enterobacteriaceae levels, at both plants A and B. Corrective actions could include evaluation of animal cleanliness, improving working procedure and instructions, retraining, review of cleaning and disinfection materials and maintenance and cleaning equipment, and improved supervision.
One possible use of microbiological data is to improve process hygiene. Thus, good practices during slaughter and dressing could be related to low numbers of bacteria, whereas increases would show deterioration in the abattoir procedures. As previously indicated, plant A had lower TVC and Enterobacteriaceae levels and a higher percentage of satisfactory values than did plant B, and it would be convenient to be able to identify reasons for these differences in the processing outcome. High counts of indicator organisms occur on carcasses when hygiene measures are so inadequate that this can already be discovered by mere observation of the slaughtering process (23). Although most processes and visual hygienic characteristics are similar for both abattoirs, before slaughter at plant B, animals are kept overnight in the slaughterhouse stables in poor hygienic conditions (on a concrete floor with a low amount of straw bedding). Thus, they are likely to carry large amounts of fecal material, and it might be expected that the livestock that entered plant B had higher contamination levels on their pelts than the livestock of plant A. This fact could explain, to a certain extent, the high contamination rates on carcasses from plant B, because external contamination of animals before slaughter influences microbial loads of carcasses (3). Thus, several studies have suggested that incision cuts through the skin and skinning operations are critical in determining contamination of carcasses; because these cuts serve as a vehicle for the horizontal transmission of bacteria to meat (7, 18, 20), a brief contact of slaughterhouse workers' hands or knives with fecal material from the pelt could cross-contaminate 10 or more successive carcasses (21). However, new microbiological analysis of the carcass at different points throughout slaughter and dressing should help clearly identify the microbiological effects of individual operations or processes.
In addition to the high visible external contamination of animals, another hypothesis could explain the higher contamination level in carcasses from plant B. Plant A has several workers (four to five) who each perform a limited range of operations. By contrast, plant B has only one worker (rarely, two) who undertakes slaughter and dressing operations at a generally high working speed. Holding carcasses at different processing stages with the same hands and using the same knives for different slaughter and dressing operations (assessed in plant B) increases the risk of cross-contaminating carcasses (7). Moreover, it has been suggested (21) that the time spent in slaughter and dressing operations is related to the standard of performance. According to Roberts (21), high productivity is most often associated with poorer hygienic practices. Mackey and Roberts (16) indicated that increased throughput results in decreased hygiene standards due to operator carelessness and fatigue.
The grouping of slaughter and dressing processes as satisfactory, acceptable, or unsatisfactory does not discriminate between the hygienic performances at different points of the slaughter process. Thus, although the classification of a microbiological result as unsatisfactory precipitated investigative activities aimed at improving control over the process, it did not indicate the cause of failure, only the need for action to improve the process. The validity of a hypothesis will be established only when actions to improve the process, based on the extracted deductions (e.g., reduce the external contamination on the livestock entering the abattoir), are shown to have the expected effects. The microbiological analysis of carcasses and surfaces at different points of the production chain should be of great interest in verifying these hypotheses.
Microbial counts on inert surfaces
Although the microbiological criteria for inert surfaces (Decision 2001/471/EC) were repealed effective 1 January 2006, surface contamination can be compared with data from this rule to discuss the hygiene at a meat plant. According to the abovementioned Decision, the number of bacteria recovered from a surface was acceptable if it did not exceed 10 CFU/cm2 (TVC) or 1 CFU/cm2 (Enterobacteriaceae). Corrective actions were required for higher values. In this study, no bacteria were recovered from most inert surface samples. Only four (3.08%) individual samples in plant A and 12 (17.14%) individual samples in plant B yielded unacceptable numbers of bacteria. Most unacceptable values were due to TVC (three samples from plant A and 12 from plant B); only one sample from abattoir A exceeded regulated values for Enterobacteriaceae.
No unacceptable average counts (mean of data for 2 weeks) were observed. These findings agree with those from Gill et al. (12) and Gill and Landers (13) in meat plants in Canada; they found few or no bacteria on most cleaned surfaces. In a survey carried out in Spain, Benezet et al. (2) found values lower than 1 CFU/2.5 cm2, for both TVC and Enterobacteriaceae, on cleaned surfaces from different meat plants.
Finding unacceptable contamination on some clean surfaces was not unexpected because plant managers and regulatory authorities tend to leave the responsibility for cleaning surfaces, equipment, and utensils up to the workers. Such cleaning must inevitably be variable and, occasionally, inadequate. Workers' motivation has a large influence on the hygienic quality of meat and surfaces. Thus, Forte et al. (10) obtained substantial bacterial reductions in hands, work tools, and work places when workers received an economic incentive. The low number of slaughterhouse workers in plant B could also be responsible for the higher percentage of unacceptable counts on surfaces from this plant. Finally, no relationship was found between microbial levels on carcasses and surfaces on each sampling day.
From the findings of this work, we conclude that slaughter and dressing hygiene should be improved to meet EU microbiological criteria for lamb carcasses, especially for Enterobacteriaceae counts. The higher microbial load on carcasses and surfaces from plant B could be due to greater external contamination on livestock entering the slaughter process, as well as to the presence of only one operator who undertakes all slaughter and dressing operations. Finally, the primary objective of microbiological testing is to verify the control process at meat plants. Failures to meet microbiological standards require a secondary investigation to detect the worst stages and to suggest improvement. Results from this study should help meat industry managers and government personnel more effectively evaluate the microbiological quality of lamb carcasses and surfaces in abattoirs. Moreover, our research could contribute to obtaining baseline data to outline the initial levels of carcass hygiene, necessary to determine the performance of a HACCP system in Spanish meat plants.
The authors thank managers, workers, and official veterinarians from the abattoirs participating in this study for their excellent cooperation. This work was financially supported by the University of León, Spain.