In 1996, the U.S. Department of Agriculture's Food Safety and Inspection Service (USDA FSIS) required that all poultry slaughter facilities must be evaluated for Salmonella on an intermittent basis.[1] In general, a USDA Inspector-in-Charge (IIC) receives notification that he or she should begin testing and the following will occur:

1. One carcass per day is selected and rinsed. The rinsate is mailed to USDA FSIS and tested for the presence of Salmonella.

2. Carcasses are selected and tested for approximately 51 processing days, or until 51 carcasses have been evaluated.

3. Thirteen or more carcasses out of 51 samples (>23.5 %) that test positive for Salmonella results in a failure.

4. Once the first failure occurs, the plant is given 30 days to make corrections, and the testing series (51 samples) begins again.

5. After a second failure, the company must write an action plan detailing corrective actions that will be taken to prevent the problem from recurring.

6. Testing resumes 30 days after the second testing series has been completed.

7. Once the third failure has occurred, inspection will be withdrawn, which effectively closes the processing plant.

This action by the USDA would result in layoffs (900 to 1,400 employees), loss of reputation and lost business ($500,000 per day minimum). In addition, five million birds at different stages of growth in preparation for processing must be sold to another company or, if possible, processed by other plants within the company in question.

This situation is very serious and at the heart of the regulation is an important scientific issue. Of what value is determining the prevalence of Salmonella positive chickens with regard to food safety? For example, Amy Waldroup, Ph.D., wrote a review article published in 1994 comparing results from studies to determine the number of Salmonella on chicken carcasses that were found to be positive.[2] Waldroup found that, on average, Salmonella positive carcasses usually have less than 30 Salmonella cells on the entire skin surface. Most of the time, the number of Salmonella cells on positive carcasses was in the 4 to 5 cell range.

Therefore, according to the way the USDA FSIS protocol is structured, it is more dangerous to have 23 carcasses out of the sample window of 51 (performance failure) with one Salmonella cell on them than to have 15 of 51 carcasses with millions of Salmonella cells on each carcass (passing score). Intuitively, it is unlikely that a few cells of Salmonella on a carcass will result in an infection. In contrast, carcasses with millions of Salmonella on them would be a hazard and many of these bacteria will come off of the carcass during processing and spread to others during immersion scalding and chilling.

So why do we test carcasses for prevalence as opposed to number of Salmonella? If you have ever attempted to run a Salmonella most probable number (MPN) assay on a chicken carcass, the answer is clear: It is extremely tedious, cumbersome and expensive (Figure 1). The purpose of this article is to discuss the development of methods to rapidly enumerate specific types of bacteria.

Because the Salmonella MPN procedure is so difficult and tedious, few people ever run these types of assays and they are almost never conducted in an industrial setting. Could a better way to enumerate Salmonella be developed? If so, could it be used as a viable regulatory method as opposed to Salmonella prevalence? The following is an example of the development of a rapid method for enumerating spoilage bacteria as a means of predicting shelf life.

Rapid Enumeration of Pseudomonas fluorescens
In 1977, Mead and Adams stated that "the rapid isolation of psychrophilic spoilage bacteria is of interest in relation to various foods and numerous attempts have been made previously to develop selective media which would permit more rapid growth of the organisms at higher temperatures whilst inhibiting the growth of mesophiles." This statement makes an important point. The traditional procedure for enumerating psychrotrophic bacteria requires incubating plates at 7C for 10 days. By the time microbiological results are achieved, the product has already been eaten or has spoiled. Thus, the method is of no practical use for predicting shelf life.

By reviewing this statement along with the optimal growth temperatures of pseudomonads (the predominant spoilage bacteria on fresh meats), it is apparent that these bacteria multiply rapidly at temperatures of 25C and can be enumerated in hours as opposed to 10 days. The only reason that the plates are incubated at such low temperatures is that if the plates were incubated at 25C, the mesophilic bacteria in the sample would easily outgrow the psychrotrophs and ruin the plate count reading. Thus, incubation temperature is the only criterion used for selection of the psychrotrophic bacteria using the traditional method.

What if a selective medium could be developed to allow only psychrotrophs to multiply, while suppressing meso-philic growth at an incubation temperature of 25C? This would provide a possible solution to the question posed by Mead and Adams. We chose the commercially available Pseudomonas Isolation Agar (PIA) and examined its selective ingredients. It was found that only Irgasan is used as the selective component in this medium. We rinsed chicken carcasses (which may contain up to 50 different genera of bacteria) and plated it on PIA. We found that pseudomonads grew well, but five other species of bacteria also grew, including Aeromonas salmonicida salmonicida, Aeromonas sobria, Aeromonas hydrophila caviae, Serratia liquefaciens and Vibrio alginolyticus (Figure 2).

After reviewing the literature, it was discovered that the following chemicals or antibiotics could be used to inhibit competing bacteria while allowing Pseudomonas to multiply: ampicillin, carbenicillin, cephalothin, chloramphenicol, Irgasan, nitrofurantoin, and trimethylamine N-oxide (TMAO). By adding each of these chemicals to a basal media in various combinations and at various concentrations, it was found that brain-heart infusion broth with carbenicillin, Irgasan and nitrofurantoin was excellent for selecting for the growth of Pseudomonas fluorescens.[3-7] This medium was patented as a means of counting Pseudomonas fluorescens, using it with agar as a plate count assay (24 hours at 25C) or in the Bactometer Microbiological Monitoring System M128 (bioMérieux) and monitoring capacitance.[7] The capacitance assay can be conducted in less than 12 hours and usually requires 6-8 hours for most samples. Thus, an assay for predicting spoilage by enumerating the primary spoilage bacteria on meat and poultry at day of processing can be completed in less than 12 hours, solving the problem posed by Mead and Adams. The premise behind the assay is that selective agents were used, as opposed to incubation temperature, to inhibit the multiplication of mesophilic bacteria while enhancing the multiplication of psychrotrophic pseudomonads. This can be done at the pseudomonads' optimum growth temperature, allowing for rapid enumeration.

Possible Rapid Method for Enumerating Salmonella
Returning to the Salmonella problem posed at the outset, we now can ask whether it is possible to incorporate the steps of concentration, preincubation, selection, detection and confirmation into one step and combine it with a system that allows enumeration. We used an optical detection method vial (BioSys) and layered the following media: preincubatory medium (Universal Preenrich-ment Medium), Rappaport Vassiliadis (RV) Medium, selenite cystine medium (SC), or tetrathionate medium with Hajna (TT), and an xylose lysine desoxycholate XLT agar plug (Figures 3-7).

Using this approach, Salmonella was able to grow in the preincubatory medium, move through the selective media and produce metabolites that were detected in the detection window in a very short period of time. The time required for this to occur should be directly proportional to the number of Salmonella in the sample, enabling enumeration. The next question is, what about samples such as carcass rinses (400 mL of liquid) that have only a few Salmonella in them? How do you get those few Salmonella into the small vial? Matrix Microscience has developed an immunomagnetic separation procedure that is AOAC International-approved and has been independently evaluated to be more effective than the current USDA method for recovering Salmonella from meat and poultry samples. The immunomagnetic separation procedure can be completed in just three hours, which means that Salmonella in large volumes of liquid or homogenate may be captured using immunomagnetic beads. The beads could then be placed onto the surface of the layered vial, and Salmonella should be able to be enumerated within one processing day. If this method is evaluated and found to be appropriate for enumerating Salmonella, it may be a more viable means of enumerating the pathogen and thus, a better way of determining the safety of foods, as opposed to conducting Salmonella prevalence tests.

Another option for confirmation of Salmonella, would be to place a specific antibody for Salmonella in the agar plug. If you make the agar plug semi-solid, then the Salmonella will move into the agar plug, react with the antibody, form an opaque zone (similar to the BioControl 1-2 test) and the instrument will detect it. Theoretically, one would have an all-in-one detection, enumeration and confirmation assay that could be done in hours.

In summary, rapid methods for enumerating bacteria from foods are within reach. An enormous amount of research has been conducted with regard to detecting pathogenic bacteria; however, in many cases, pathogenic bacterial counts may be more appropriate. This is especially true with pathogenic bacteria such as Campylobacter. Conducting prevalence assays for Campylobacter on poultry products would be meaningless; whereas, a rapid procedure for counting Campylobacter may allow the USDA to implement applicable regulatory procedures.

Scott M. Russell, Ph.D., is associate professor in the Department of Poultry Science at the University of Georgia, Athens, GA., where he has conducted research and provided extension services to the poultry industry for the past nine years. A Food Safety Magazine editorial advisory board member, his main areas of interest are developing rapid and automated methods for identifying and enumerating pathogenic and spoilage bacteria from foods of animal origin, and identifying methods for eliminating pathogenic and spoilage organisms from poultry during rearing and processing.

References
1. USDA. Federal Register. 9 CFR Part 304, et al. Pathogen Reduction; Hazard Analysis and Critical Control Point (HACCP) Systems; Final Rule. Vol. 61, No. 144, pp. 38846-38848. 1996.
2. Waldroup, A.L. Pathogens on raw poultry—a review. Broiler Industry Magazine, pp. 18, 20, 22, 24, 26, 27, and 28. May 1994.
3. Russell, S.M. A rapid method for determination of shelf-life of broiler chicken carcasses. Journal of Food Protection 60(2):148-152. 1997a.
4. Russell, S.M. A rapid method for enumeration of Pseudomonas fluorescens from broiler chicken carcasses. Journal of Food Protection 60(4):385-390. 1997b.
5. Russell, S.M. Rapid prediction of the potential shelf-life of broiler chicken carcasses under commercial conditions. Journal of Applied Poultry Research 6(2):163-168. 1997c.
6. Russell, S.M. A rapid method for predicting the potential shelf-life of fresh fish. J. Food Prot. 61(7):844-848. 1998a.
7. Russell, S.M. A selective additive for enumerating Pseudomonas fluorescens and for predicting the potential shelf-life of fresh poultry, beef, ground beef, fish, and milk using standard and electrical microbiological methods. U.S. Patent Number 5,741,663. 1998b.