My first go at consulting was over a decade ago, and a few of my clients were in the poultry industry. At the time, there were questions about whether the amount of Salmonella (quantification) was more important than straight prevalence (frequency of finding a positive, even if levels were low). I had the pleasure of working with top-notch industry experts to flesh out these ideas in a publication.¹ A decade later, people occasionally bring up this article and its relevance today; in August 2022, the U.S. Department of Agriculture's Food Safety and Inspection Service (USDA FSIS) began quantifying Salmonella in raw poultry rinses. I have to wonder: Why is change so slow?

A co-author on that paper, Dr. Angie Siemens with Cargill, reflects, "The publication in 2014 was presented after a series of Salmonella public health incidences and proposed a potential new pathway for controlling that portion of product with the greatest potential of creating illness. It was not revolutionary, as the approach had been used in other countries. It is a shame that the focus on current Salmonella Performance Standard metrics continued with little consideration of alternatives, until recently."

Salmonella, including in poultry, remained top of mind during my time representing the fresh produce industry. Poultry pellets and/or composted poultry manure are sometimes used to fertilize fields on which produce will grow. What progress has been made, and where do we still need more information? To what extent does poultry contribute to cases of salmonellosis, both directly and indirectly, and is there more the industry can do to protect public health? Clearly, targeting mitigations and assessing public health risk are reliant on appropriate sampling and testing.

Now that I am back in consulting, I have had the opportunity to re-engage with the poultry industry and, tangentially, their customers and industries that support poultry testing. Many of the topics are still the same: Where in or on the bird does Salmonella hide? What is the best way to find it and enumerate it? Does serotype matter? Which of these is most important to assess public health risk? Are virulence genes better indicators of public health risk? What do indicators indicate? What is the role of the farm? Despite extensive work, the answers are not clear. Progress toward achieving public health goals related to salmonellosis has stagnated. Yet, where and how to test fresh poultry throughout the supply chain in order to gauge public health risk is still the source of substantial discussion.

A session at the 2023 International Association for Food Protection (IAFP) meeting provided a good overview of the current regulatory and scientific considerations for poultry safety.² In the session, USDA FSIS shared compelling information about the need to try a different approach and gave an overview of the changes proposed by FSIS; the National Chicken Council identified many continued unknowns and potential unintended consequences of the proposal; and a USDA Agricultural Research Service (ARS) researcher shared results comparing the current rinse-based method to one using cloth-based sampling to illustrate how improved recovery, as well as quantification of Salmonella, can be used on turkeys. In autumn 2023, select members of the poultry industry, along with researchers in this area, expanded on the discussion and left me with an appreciation of the difficulty in finding and tracking Salmonella in a way that correlates with public health risk.

Reliable test results are a prerequisite for targeting interventions and improving processes. This article presents an outsider's view of poultry safety, with an eye toward stimulating discussion around new approaches while airing a hint of frustration at the glacial pace of progress.

What Has Not Changed?

The main thing that has not changed in decades is the incidence of salmonellosis. Although many foods serve as vehicles for the pathogen, almost 20 percent of outbreak-associated cases of salmonellosis are attributed to chicken, and just over 5 percent are attributed to turkey.³ On paper, fruits are responsible for almost 16 percent of salmonellosis cases, and seeded vegetables contribute another 11 percent. Taking a "One Health" perspective should prompt questions around how the organism entered the produce growing environment, and if poultry sources could have indirectly contributed to the presence of the pathogen, as was suspected in a 2020 outbreak associated with peaches.⁴ FDA also noted the relationship between Salmonella in turkey and cantaloupe: for example, Salmonella isolates from ground turkey and cantaloupe that were whole genome sequence (WGS) matches, the use of untreated turkey manure on fields later used to grow cantaloupes, and the proximity of turkey feeding operations to cantaloupe production.⁵

A recent Centers for Disease Control and Prevention (CDC) analysis shows that rates of salmonellosis in 2022 were unchanged compared to the 2016–2018 baseline, at 14.5 cases/100,000 population.⁶ Every decade, the U.S. Department of Health and Human Services establishes targets for various health measures, including rates of salmonellosis. The Healthy People 2030 target is 11.5 cases/100,000 population, and USDA FSIS has set a similar target to reduce Salmonella infections associated with FSIS-regulated products by 25 percent.⁷

Another thing that has not changed is the need for more data. With the continued focus on absence/presence of Salmonella, limited resources have been put into looking at alternatives such as enumeration, serotypes, or pathogenicity factors throughout the supply chain. There are still questions around where in, or on, the bird Salmonella reside. This influences how best to sample and test for the pathogen, and how to interpret results of serotypes and the levels/ quantification. The results of a risk assessment, as has been contracted by USDA FSIS, will need to be viewed with these limitations in mind.

What Has Changed in the Past Decade?

The most obvious change has been on the regulatory side. The poultry industry has made strides in meeting the current performance standards. Many establishments are able to achieve "Category 1" status, meaning that their percent positive for the organism is less than half of the USDA FSIS performance standard. However, performance standards may be more useful to gauge process control than food safety.

USDA FSIS has proposed a three-component framework aimed at reducing salmonellosis associated with poultry products.⁷ The first component focuses on the live birds, with the premise that if Salmonella can be better controlled coming into the processing facility, then public health will be improved. The second component considers how monitoring indicator organisms throughout various parts of the process may also give insight into risk. The third component proposes the establishment of enforceable finished product standards for raw poultry that harness today's science and satisfy the joint FSIS and industry responsibility to ensure that products are safe for consumers. All of these goals are contingent on gaining meaningful and relevant measurements of Salmonella spp., high-risk serotypes, and/or indicators.

According to Michael Taylor, the former FSIS administrator who established in 1996 the system of Salmonella performance standards that is still in force today, "The most critically needed poultry safety reform is to establish enforceable finished product standards that provide regulatory incentive and accountability for processors to innovate and prevent contamination with the levels and types of Salmonella that are making people sick. Testing to verify that prevention systems are working to meet such finished product standards is key to protecting consumers."

A deeper discussion of assessing Salmonella in poultry houses, issues around indicators, and figuring out where best to test for Salmonella follows.

Another change that has impacted the breadth of the food and beverage industry has been WGS. This technology has been a game-changer across the food industry by enabling public health officials to more rapidly identify an outbreak and search for clues about potential sources, based on sequences from product and other environmental isolates submitted to a database. Scanning a phylogenetic tree can give hints that a certain sequence may be commonly associated with a specific host (e.g., avian) or be found in a certain part of the country. The insights we can gain from deeper genetic analyses have yet to be fully exploited. Not only can we compare strings of A, C, G, and T between isolates to determine evolutionary similarity, we can also begin to assess the relevance of genetic differences. Analysis of antibiotic resistance genes is already underway, and it is likely that we will begin to better understand from a genetic standpoint why certain organisms are more likely to cause illness, persist in the environment, or display other characteristics.

A deeper understanding of microorganisms also shows that not all Salmonella serotypes are equal. Although all are considered pathogenic, it has become clear that the serotype most commonly found in poultry (particularly at preharvest), Salmonella Kentucky, does not cause human illness at the rates one would expect based on prevalence.⁸ In other words, the public health risk differs based on serotype, and the historical performance standard approach did not take this into account. Our understanding of virulence factors and genetic attributes that impact fitness, including the ability to evade processing mitigations, may trigger additional evolution in how Salmonella are managed in the poultry industry.

The On-Farm vs. In-Plant Dilemma

The first component of the USDA FSIS framework proposes that the incoming flock be tested for Salmonella before receipt at the establishment. The possible sources of Salmonella in the pre-harvest stage were discussed by Thippareddi and Singh.⁹ Opinions diverge on the correlation between on-farm and in-plant data. The first component of the FSIS framework suggests that there is a correlation. While this seems reasonable, it assumes that Salmonella recovered from the live bird eventually contributes to public health risk. Are the serotypes found in poultry houses or in/on live birds the same as those causing human illness? And if they are, is illness attributed to poultry or to a secondary vehicle?

A key question is what to test: boot swabs and water samples each represent a composite of sorts, and can give an indication of Salmonella that may be in the gut. If Salmonella are internalized in other parts of the animal, how relevant is testing at the poultry house? Members of industry have observed that "hot" houses do not always correspond with higher Salmonella levels as poultry is processed.

Industry experts have also shared variation in levels recovered by boot swabs, even when variables were seemingly well managed. Is this a consequence of the actual variation in different birds, suggesting that some excrete higher levels of the organism than others? Or is this a methodological artifact? To what extent does the sampler impact the results (e.g., based on weight/pressure applied)? Different materials have different recovery efficacy, and this is an area that can be further explored, optimized, and standardized.

I listened with interest to the industry discussion around analyses in poultry houses, given that, as previously mentioned, poultry pellets can be used to fertilize produce fields. Regardless of whether testing poultry houses is a useful indication of public health risk associated with poultry products, it may still provide value to the broader food community using poultry byproducts. There have been questions related to the role that improperly treated poultry pellets may have played in produce-related outbreaks. Is there an opportunity for the produce industry—and especially those treating poultry pellets—to gain insight into how the microbial load, and specifically the levels of Salmonella might change, impacting the efficacy of the microbial reduction treatment? Using boot swabs in this way requires a reliable means to quantify Salmonella levels, as well as a sampling strategy that can identify when high loads are present.

Dr. Laura Strawn, an Associate Professor and Extension Specialist in Produce Safety at Virginia Tech who was able to join me in the discussion with the poultry industry as part of her professional deployment, feels that, "Gaining insights into the specific strains and levels of Salmonella within poultry houses is key to understanding the potential impact on the broader environmental ecology. This knowledge can enable us to establish meaningful connections, aligning with the systematic One Health perspective."

Post-Chill Carcass Rinsing

The second component of the USDA FSIS framework seeks to establish requirements for demonstrating process control during processing. This includes considering new or additional locations within the establishment where sampling would occur. Today, the utility of carcass rinsing post-chill (which is destructive), as an indicator of risk, is the subject of debate. While it proves useful in gauging the effectiveness of a process in reducing levels of indicator organisms, its reliability in detecting pathogens raises questions. In most cases, carcass rinses post-chill yield negative results for Salmonella. However, when positive, the quantity is often extremely low. One hypothesis is that Salmonella become entrapped during chilling. If this is true, rinsing does not effectively release them, but they are still present, leading to downstream positivity.

Due to their large size, carcass rinsing is not used for whole turkeys. Still, the cellulose sponge sampling method shows a fraction of a percent positive for Salmonella post-chill, while sampling later in the process shows significantly higher rates (almost 15 percent positive for comminuted turkey and 38 percent positive when mechanically separated).¹⁰ Preliminary research results from USDA ARS shared at the 2023 IAFP annual meeting show that both the point in the process, as well as the sampling method, influence results. Using different materials (a polyurethane swab; the proprietary, non-woven polyolefin cloth swab used to sample beef trim; and a mitt sampler made of the same proprietary cloth material) to sample post-chill yielded similar results (close to zero); however, when testing was done at rehang, the mitt sampler significantly outperformed the cellulose and polyurethane swabs (28.7 percent positivity vs. 13.9–14.8 percent).¹¹

Tracing Salmonella: The Elusive Culprit

As previously discussed, when the rinse method is used to assess the presence or absence of Salmonella in chickens, the results would lead one to conclude that Salmonella is not a huge issue. If higher rates are detected in later stages of processing, as USDA results show, where is the pathogen coming from?¹⁰

Today's approaches to sampling and testing—using carcass rinsing or cellulose swabs post-chill—give misleading results. Identifying the source of Salmonella in birds is imperative for targeted interventions. Whether it resides in the skin (entrapped, as discussed above), deep muscle tissue, or bone is a critical question. Historical industry data on deep muscle tissue and pneumatic bone rarely show positives. Combined, this prompts a deeper exploration into the elusive location of Salmonella within poultry.

Clearly, there is no spontaneous generation of Salmonella. The organism must be somewhere. Better sampling techniques are needed to physically recover the organism, and perhaps better test methods are needed to resuscitate the organism so it can be detected. Improvements in sampling and testing are the linchpin to solving the Salmonella conundrum.

Continuous Testing for 'Hot' Birds

The premise of our 2014 paper¹ was that the number of Salmonella matters. Positivity does not tell the whole story. While the idea that some birds or some flocks may have high levels of Salmonella does not explain the re-hang vs. post-chill positivity rates, it could explain the higher positivity rates of comminuted and mechanically separated product. Salmonella from one "hot" bird could become distributed to additional product, whether through mixing or through cross-contamination. Traditional testing methods, such as individual carcasses or bird part testing, may prove insufficient due to statistical limitations (one cannot test enough to achieve the desired level of confidence). The exploration of continuous testing of product contact surfaces emerges as a potential solution, providing more representative data.

Aggregate sampling provides a more representative view of changes in Salmonella levels over time. Analyzing levels of Salmonella—for example, using a continuous sampling device—in the context of other variables (season, house, and likely other metadata) could be a useful application of "big data." Even when results could not be received in time to affect the disposition of the processed product, these data can provide system/process feedback, influencing future practices rather than immediate product release. If data are properly collected on a continuous basis, then the opportunities to employ predictive analytics could be powerful. Eric Wilhelmsen, Principal Consultant at the Alliance of Technical Professionals, notes, "The dynamics of aggregate sampling provide a better view for a lot than individual grabs, where the view will be slanted toward rare detections of a 'hot' bird."

The third component of the USDA FSIS framework is an enforceable final product standard. FSIS states, "FSIS regulations should prevent product with high levels of contamination and/or specific serotypes from entering commerce," and goes on to say that criteria for considering Salmonella an adulterant would include serotypes, infectious dose, severity of illness, and consumer cooking practices (e.g., dependent on the specific product).⁷ Salmonella may not be evenly distributed between different parts of a bird, and may not be uniform from bird to bird, or flock to flock, especially if considering serotypes and levels. The likelihood of a final product standard detecting "hot" birds is questionable, and additional consideration should be given to sampling plans and approaches that can be practically implemented to verify compliance with a finished product standard and make the greatest difference in public health protection.

Key Poultry Industry Needs

The needs of the poultry industry seem straightforward: trace the Salmonella causing illness back from the finished product to the earlier point(s) in the process where mitigations can be implemented. This necessitates identifying (serotyping or virulence gene detection) and quantifying Salmonella, both of which require reliable sampling and testing:

• Non-destructive sampling that does not require taking out and discarding poultry products is ideal. The beef industry has transitioned to using a non-woven polyolefin cloth-based sampling method, as opposed to an excision method.^12,13 Although this approach has not been validated for the poultry industry, early research shows promising results.
• Methods that allow for reproducible quantification and serotyping are needed. Industry experts shared that internal research showed high variation in results, even when they tried to control variables (e.g., operator/sampler). Are actual levels of Salmonella truly that variable, or is this a consequence of the sampling or testing approach?
• Standardized approaches and processes within the industry that facilitate comparing research study findings are needed.
• The relationship between on-farm positivity, load, and serotype and in-plant positivity, load, and serotype is not yet well established:
  • Determining where and what to test on-farm and in the plant that will make the greatest positive impact on public health remains unresolved.
  • The potential benefit of understanding Salmonella excreted from live birds within the context of the practices of those who process and use poultry byproducts (e.g., manure) deserves additional attention.

• Continuous sampling of second process surfaces (or parts) may be able to reveal flocks or birds with higher-than-expected levels of Salmonella:

  • Traceback may reveal opportunities to improve the process, either in the plant or before (on-farm).

Although there are still legitimate questions to be answered, research to be done, and advances in methodological and sampling approaches to be developed, consumers, regulators, and industry must work together to ensure that the next decade will see more meaningful public health gains with respect to poultry-associated salmonellosis, compared to the last decade.

Acknowledgement

The author thanks Garth Hoffmann, Angie Siemens, Laura Strawn, Michael Taylor, and Eric Wilhelmsen for their thought-provoking conversations and critical reviews of this article.

References

1. 1 McEntire, J., D. Acheson, A. Siemens, S. Eilert, and M. Robach. "The Public Health Value of Reducing Salmonella Levels in Raw Meat and Poultry." Food Protection Trends 34, no. 6 (2014): 386–391.
2. International Association for Food Protection. "Poultry Sampling Symposium: The Path to Improved Poultry Safety through Salmonella Assessments." IAFP Annual Meeting. Toronto, Canada, 2023. https://foodprotection.org/upl/downloads/meeting/archive/64e8d700dbbd79453a8ec.pdf.
3. International Food Safety Analytics Collaboration (IFSAC). "Foodborne Illness Source Attribution Estimates for 2021 for Salmonella, Escherichia coli O157:H7, and Listeria monocytogenes Using Multi-Year Outbreak Surveillance Data." U.S. Department of Health and Human Services, Centers for Disease Control and Prevention (CDC), U.S. Food and Drug Administration (FDA), U.S. Department of Agriculture's Food Safety and Inspection Service (USDA FSIS). 2023. https://www.cdc.gov/foodsafety/ifsac/pdf/P19-2021-report-TriAgency-508.pdf.
4. FDA. "Investigation Report: Factors Potentially Contributing to the Contamination of Peaches Implicated in the Summer 2020 Outbreak of Salmonella Enteritidis." 2021. https://www.fda.gov/media/149804/download?attachment.
5. FDA. "Investigation Report: Factors Potentially Contributing to the Contamination of Cantaloupe Implicated in the Outbreak of Salmonella Typhimurium During the Summer of 2022." 2023. https://www.fda.gov/media/167640/download?attachment.
6. CDC. "Preliminary Incidence and Trends of Infections Caused by Pathogens Transmitted Commonly Through Food—Foodborne Diseases Active Surveillance Network, 10 U.S. Sites, 2022." Morbidity and Mortality Weekly Report. June 29, 2023. https://www.cdc.gov/foodnet/reports/preliminary-data.html.
7. USDA Food Safety and Inspection Service. "Proposed Regulatory Framework to Reduce Salmonella Illnesses Attributable to Poultry." October 14, 2022. https://www.fsis.usda.gov/inspection/inspection-programs/inspection-poultry-products/reducing-salmonella-poultry/proposed.
8. Obe, T., A.T. Siceloff, M.G. Crowe, H.M. Scott, and N.W. Shariat. "Combined Quantification and Deep Serotyping for Salmonella Risk Profiling in Broiler Flocks." Applied and Environmental Microbiology 89, no. 4(2023): e02035-22. https://doi.org/10.1128%2Faem.02035-22.
9. Thippareddi, H. and M. Singh. "A Critical Look at Reducing the Risk of Salmonella from Poultry—Part 1." Food Safety Magazine August/September 2022. https://www.food-safety.com/articles/7939-a-critical-look-at-reducing-the-risk-of-salmonella-from-poultrypart-1.
10. USDA FSIS. "Annual Sampling Summary Report: Fiscal Year 2022." 2023. https://www.fsis.usda.gov/sites/default/files/media_file/documents/FY2022-Sampling-Summary-Report.pdf.
11. Arthur, T.M. "Poultry Sampling Symposium: The Path to Improved Poultry Safety through Salmonella Assessments." IAFP Annual Meeting, Toronto, Canada, 2023.
12. Wheeler, T.L. and T.M. Arthur. "Novel Continuous and Manual Sampling Methods for Beef Trim Microbiological Testing." Journal of Food Protection 81, no. 10 (2018): 1605–1613. https://doi.org/10.4315/0362-028X.JFP-18-197.
13. USDA FSIS. "LMG 5C.03. Detection, Isolation, and Identification of Top Seven Shiga Toxin-Producing Escherichia coli (STEC) from Meat Products, Carcass, and Environmental Sponges." February 1, 2023. https://www.fsis.usda.gov/sites/default/files/media_file/documents/MLG-5C.03.pdf.