I am approaching this topic with some trepidation. Although I have decades of experience in food safety, all of my experience has been for or at the behest of industry. I have not walked one mile in the shoes of the regulators. I am also very aware of the glass in my own home; therefore, I am reluctant to throw stones. Industry is involved in outbreak investigations and will have roles to play in any improved system. With these caveats in mind, the discussion that follows is more about asking questions than providing recommendations.  

The topic of outbreak investigation improvement is large, and there are many details that can be considered. The following discussion is biased toward produce, particularly fresh-cut produce, where my personal efforts have been largely focused in recent years. The discussion also has a U.S./North American bias. However, I suspect that, at the high levels of this discussion, similar analysis will find similar problems in other markets and other industry sectors.  

To aid the reader and provide focus, the following discussion is divided into seven sections that address the following questions: 

• Is improvement in the outbreak investigation system needed?  
• Do competing objectives hinder the investigation process?  
• Is the airline industry a model for improvement?  
• Can risks be separated from potential hazards?  
• Is timeliness a problem?  
• What are the limitations of whole genome sequencing (WGS) in this context?  
• Where and how should society spend its food safety dollars?  

Is Improvement in the Outbreak Investigation System Needed?  

It is clear that outbreaks and illnesses related to food are occurring. For better or worse, society as a whole seems to believe improvement is needed. These outbreaks and illnesses are not rare events, but they are mostly small compared to the volumes of servings sold and consumed. My news feed is filled with reports of concerns about the safety of various products. Many of these relate to non-bacterial problems, but there are still many related to pathogens.  

Some reports can also be misguided. A recent example is the Consumer Reports article¹ condemning bagged salads in preference to naked lettuce. This seems counter-intuitive when one realizes that the same lettuce feeds into both channels of distribution. Additionally, branded products are subject to greater scrutiny. The number of illnesses associated with produce is very small compared to the billions of servings consumed each year, but society remains focused on the impossible task of "zero risk." The outbreak investigation system does not deliver zero risk. However, does this really mean improvement is needed?  

I would opine, as a counterpoint, that the North American food supply currently presents less risk to consumers than at any other time in history. Granted, some people are suffering from food insecurity and some people are eating too much (and too much of the wrong foods); but these are not the risks that our outbreak investigation is charged with addressing. Is society wasting resources with the current investigation system? Is society chasing hazards that present little risk when there are other, more important risks that need to be addressed? 

The prevalence of inconclusive investigations that result in reports pointing at the usual suspects—e.g., land adjacencies, agricultural water, soil amendments, and environmental monitoring issues—may indicate inefficiency in the system. The prevalence of repeat outbreaks, such as the Salmonella issue with cantaloupe,² the annual Cyclospora event attributed to produce and leafy greens in particular, and the E. coli O157 contamination associated with romaine lettuce may also indicate inefficiency in the investigation system. Despite the recent focus on traceability, it seems reasonable to ask if the current system is up to the challenges related to the global economy and nationwide distribution channels. This challenge is especially problematic for highly perishable and short shelf-life products with national distribution.  

Some will argue that the system is not broken, or that it is working. The generalities discussed here support the idea of room for improvement, but allow that care must be taken to avoid "throwing out the baby with the bath water."  

Do Competing Objectives Hinder the Investigation Process? 

It is ludicrous to think that anyone would want unsafe food. Still, the concept of what safe food means varies with the stakeholder: 

• Consumers expect that food will not cause them illness. As discussed above, there is an assumed zero risk. When an individual becomes ill, the perception is that the system failed and someone is responsible.  
• Producers want to produce products that consumers will continue to buy, which adds an economic perspective to the concept of food safety. Spend too much on safety, and no one will buy your product. Make people ill, and you will lose market share or the entire market.  
• Regulatory enforcement seeks to minimize or prevent consumer exposure to risks or potential risks. It is politically important to be seen as proactive. Action is often taken based on partial information. It seems important to assign blame or at least potential responsibility. Sometimes, activities seem to be punitive because they occur after the hazard has disappeared.  
• Regulatory rulemaking and legislation seek to prevent illness by establishing practices that, if followed by everyone, would prevent unsafe product from being produced or sold. These efforts are hampered by the need to accommodate the real and perceived limitations of all producers. Exemptions and delays for small businesses and small farms are the archetypes of this limitation.  
• Plaintiff attorneys seek to assign liability for food-related illness to compensate their clients. They also have a profit motive that cannot be ignored. This tort process can increase the public perception of real and potential risks, even when the magnitude of the risk is small compared with other risks associated with life. The prevalence and success of this activity make these liability costs part of the cost of doing business.  

Each of these positions makes sense, based on the needs of the stakeholder. They are consistent with the concept that food-related illness is bad. Can we bring these disparate objectives into alignment? So far, I would opine that we have not. The engendered friction results in much of the inefficiency in the system. What changes can be made? Do these competing goals limit the sharing of information? Do these goals limit progress? Is the system more adversarial than it needs to be? Can we separate preventing future occurrences from assigning blame and liability? 

Is the Airline Industry a Model for Improvement?  

On more than one occasion, it has been suggested that the Federal Aviation Administration and the airline industry are a model that should be emulated by the food industry to improve food safety. The proponents argue for systems of checklists, investigation of all safety events and near-miss safety events, and consistent pursuit of root cause as the keys to the amazing safety record of air travel. When viewed from the outside, the safety record of air travel is probably unparalleled.  

However well-earned this safety reputation is, it has been tarnished by recent events: 

• A door plug bursting out mid-flight, and others with loose screws 
• Fraudulent parts being used 
• Failed computer systems causing massive delays 
• A dropped tire 
• Lost panels.   

This point and counterpoint serve to highlight that there are good and bad things to be learned from the airline industry. One is right to question how much of this "reputation tarnishing" is due to hyper media attention. Mistakes draw eyeballs to advertisers. Air travel will never be 100 percent safe, just as food cannot be made 100 percent safe.  

The airline industry's collaborative model of investigation and root cause analysis, followed by efforts to implement true corrective action, is to be emulated. I suggest that the airline industry has done better than the food industry at melding the needs of all stakeholders to achieve this safety performance. It is also clear that the airline industry's work is not finished. It is reasonable to expect that undesirable events will continue to occur, requiring further self-examination.  

I have not seen the final reports for the few events cited above, so I am speculating on the root causes. I think it is reasonable to expect that cost cutting, complacency, and information loss have played roles, irrespective of what the reports assert. There is constant pressure to do more with less. As time passes without issue, vigilance naturally fades. It is hard to refresh the knowledge about what is important over time, especially with the changing of personnel. Unfortunately, I see these same types of shortcomings in the food industry. They are probably found everywhere.  

These observations lead to another series of questions. How do we get everyone to consistently do what should be best practices in various segments of the food industry? How do we prevent irresponsible cost-reduction efforts? How do we pay for more of the right things to improve safety?  

Can Risks be Separated From Potential Hazards?  

Before we examine this question, I want to point out that I have no desire to see anyone suffer a food-related illness. Every illness represents a situation where something bad has happened to an individual. Every day I go to work, I strive to help the industry prevent these bad things from happening. 

One of humanity's great skills is the ability to identify patterns and extrapolate these patterns to new situations. This skill creates bias in our perception of risk. The closer we come to a bad outcome or the greater the impact it has on our thoughts, the greater the perceived risk. For better or worse, media is an important amplifier. This perception of risk is generally not quantitative. If you are in a serious car accident, then driving is perceived as a more serious risk, especially if the accident was the other driver's fault. If a trusted information source presents a problem with a product, then the perceived risk increases, even if the actual risk is small. Is the risk associated with driving to and from the grocery store greater than the risk associated with consuming a serving of grocery store sushi or a bagged, ready-to-eat salad? Data and mathematics can compare these risks, but will this change people's perceptions?  

The greater food industry, including academics and government, has been successful at identifying potential hazards. Some of these hazardous situations occur with enough frequency that they pose substantial food safety risks. When these risks become great enough, strong steps like regulations are implemented to try to prevent the hazardous situation from occurring. Botulism poisoning is a well-characterized hazard and is subject to many regulations to mitigate the risk of occurrence. Nevertheless, botulism poisoning has not been eradicated. How do we place risks on this spectrum from where something might happen rarely to where something must be prevented as much as possible?  

It is tempting to decide that data and mathematics should make the decisions; this would be the science-led process. This is not our current reality. Hundreds of illnesses associated with billions of servings of produce are perceived as an important risk, while people are still allowed to smoke tobacco and drive cars. I speculate that the cancer risk from pumping gasoline into our cars is greater than the risk associated with eating a pre-washed, bagged salad. The data exists, but even when someone conducts a thorough analysis, it often does not change people's perceptions.  

Once a hazard is confirmed as possible, the response is to devise mitigations. I opine that this process has led to tremendously long audit lists filled with minutia. So much of food safety is in the minutia, and it is often about consistently doing everything right. In this environment, how do we ensure that the most important things always get done, and done well? This is the process that separates hazards from risks. Sadly, it is not always being done well.  

On this front, I am encouraged by developments in risk modeling and residual risk analysis. These models provide a means of comparing the importance of various mitigations and process controls. I anticipate that these models will support my personal observations in fresh-cut produce processing that high-quality good agricultural practices (GAPs) and chemistry control (pH and chlorine, for the most part) are much more important to fresh-cut leafy greens safety than testing programs. Prevention is much more effective than any after-the-fact mitigation. I can only wonder if this modeling will change the perception of the marketplace that wants and demands finished product testing, which has minimal impact on safety unless very serious problems are present. Can the greater industry come together to make adjustments based on science? Will there be a focus on getting everyone to equally execute what is known and shown to be best practice?  

Is Timeliness a Problem?  

There are four kinds of timeliness problems: 

• Desire to immediately prevent illness or potential illness 
• Lag between detection and when the causative event occurred  
• Span of time over which the problem is deemed to have occurred 
• Lag between knowing and doing when a change is needed.  

The challenge of preventing illness or protecting consumers from potential illness is almost a thankless task. These efforts generally start with the identification of a situation, such as two people being made ill with the same organism or a product testing as positive for a pathogen. The challenge is then to identify the risk, because zero risk is impossible. The product associated with the hazard may be out of the marketplace, but has the hazard actually run its course? Is the hazard restricted to one source or lot? Is there any action that will reduce risk or potential risk?  

Related to the first problem is finding cause. Retrospective investigations rarely succeed because the causative situation is usually transitory. Consistent situation would lead to much larger incidents. Even the produce outbreak that shut down sales of romaine lettuce in 2018 resulted in only approximately 200 reported illnesses over millions of servings. Communication and detecting the results are getting faster, but incubation times and distribution times present hurdles that will not be overcome. I see no way for detection to be fast enough to identify a transitory problem in progress. In the face of this reality, would energy be better spent on preventive measures and execution of known best practices?  

Moving to the third timeliness problem, what is a reasonable duration for an outbreak event? How does a transitory event span years? It is reasonable that a similar situation has occurred again, but it does not seem reasonable that a few illnesses that occurred more than several years previous are related to the same event. This leads to the topic of the next question. There may be solid technical reasons for the aggregation that is presently occurring, but I remain unconvinced. Should there be a "statute of limitations" for the duration of an outbreak? What is the minimum frequency for an event to occur to be classified as the same issue?  

Finally, I ask the question of why it takes so long for scientific finds to become best practices or regulatory standards? My experience indicates that for most products, there is a technical leader. Some portion of this standing relates to food safety. Why does it take so long for the demonstrated best practices to be adopted? I am well aware of the problems of capital spending. As they say, "Rome was not built in a day." However, schedules for adoption need to be put in place. Workarounds to minimize risks in the interim need to be put in place. How can a processor be expected to add a food safety cost when competitors continue to use a less expensive and less safe practice? How do we avoid a race to the bottom with unwise cost reduction measures? Incremental progress occurs only when change occurs. How do we foster change?  

What are the Limitations of Whole Genome Sequencing in This Context?  

Whole genome sequencing (WGS) is an amazing tour de force and a powerful tool. It has exonerated and convicted criminals years after the events transpired. Powerful libraries and search tools permit analytics that were unimaginable even a few decades ago. Meta-genomic analysis is a close kin and is still developing applications. CRISPR (clustered regularly interspaced short palindromic repeats) is the new kid on the block, and we are just starting to see its impacts. Still, I ask the question in sincerity: Are these technologies helping food safety?  

On the positive side, WGS has provided or confirmed the nexus between illnesses and organisms in products and/or facilities. These determinations probably provide good focus for root cause analysis (RCA) and investigation, but may prompt preconceived notion bias in the investigation and wrongly attribute the entire problem to a single source.  

WGS is presented as a tool to create a nexus between illnesses scattered across several states. It is also presented as a tool to create a nexus across time for illnesses that have occurred months and even years apart. I am even more concerned about preconceived notion bias in these types of investigations. WGS provides only positive information. It cannot be applied to the negative information.  

Examining a specific type of case at a very high level, we can assume that a cluster of illnesses is identified across three states and that a product class shows up in the interviews. The investigation identifies only one producer present in all three markets. The positive information seems solid. However, if the processor shipped this product to eight other states that have no reported illnesses, and the processor has over 50 percent penetration in the markets where the illnesses were reported, I opine that the information is much less conclusive. WGS provides simple answers, when the genuine answer may be more complicated. WGS information is often assumed to prove tight links.  

WGS data and WGS libraries can create what I will call "the gift that keeps giving." The WGS-identified Listeria problem in cheese that clustered a small number of illnesses over a decade is an example. This has led to more testing and more product recalls over the potential hazard. In a zero-tolerance world, there is no other choice. However, as an outsider looking in, products lots were destroyed that had little or no proven risk. This actual level of risk had been acceptable for years. This does not mean that RCA and change are not needed; but at some point, some level of risk is inherent in life.  

WGS is a blunt instrument. Its use needs to be nuanced in some way to ensure focus on the real risks and not just on the hazard of the day. Lumping illness over years based on libraries will not foster a collaborative environment. It will provide grist for the plaintiff attorneys and yield settlements for the injured; but again, I do not think it will reduce risk for most consumers. Can research and science provide more insight into what matching WGS data means? Would the concept of a "statute of limitations" make a difference? 

Where and How Should Society Spend its Food Safety Dollars?  

In many respects, the answer to this question is the sum of the entire previous discussion. This discussion already comes perilously close to making recommendations. Hopefully, the reader who has gotten this far will have collected a few questions to ponder.  

Let me conclude with some final “big” questions that will hopefully generate even more discussion and change: 

• Should enforcement action be separated from investigation?  
• Can triage focus investigation resources on the highest-risk hazards? 
• Can the greater industry come together to accept the costs associated with changes where science indicates that risk will be reduced?  
• Can a more collaborative environment be created that produces more of the positive results that have been seen in the airline industry?  
• Can we figure out a way to mitigate information loss and avoid the pain of relearning?  
• Lastly, would energy be better spent on preventive measures and execution of known best practices?  

References

1. Gill, Lisa L. "10 Risky Recalled Foods You Should Know About." Consumer Reports. March 30, 2023. https://www.consumerreports.org/health/food-recalls/risky-recalled-foods-you-should-know-about-a4109713872/. 
2.  Wilhelmsen, Eric. "What Was Not Learned from the FDA Investigation of Salmonella on Cantaloupe in 2022?" Food Safety Magazine December 2023/January 2024. https://www.food-safety.com/articles/9105-what-was-not-learned-from-the-fda-investigation-of-salmonella-on-cantaloupe-in-2022. 

Eric Wilhelmsen, Ph.D., CFS, is a recognized expert in food safety and sampling, serving for over 30 years in both academic and industrial positions. In these roles, he has provided technical leadership and innovation for industrial collaborations. His technical contributions and practical innovations have been fundamental in developing proprietary technology, as well as establishing new revenue streams and profitable businesses in juices, dietary supplements and botanicals, agricultural commodities, byproducts, and beverages. He can be reached at the Alliance of Technical Professionals: eric.wilhelmsen@atpconsultants.com.