PMA: Ask Dr. Bob Whitaker

Julia Stewart of PMA passes on this link to PMA's "Ask Dr. Bob Whitaker" Web page. Whitaker is the chief scientific officer for the Produce Marketing Association. Here are a few of the questions and responses from the Web page, published yesterday:




How does a public health traceback work?
As a foodborne illness investigation moves forward, epidemiology and testing identify the causative microorganism and vehicle or food item that carry it. Then the activity shifts to a trackback investigation that is generally conducted by FDA in conjunction with various states and/or counties. One of the key questions the epidemiologists ask patients is where they contacted the food product – restaurant, retail store, family picnic, school, church social, friend’s house, etc. If they believe the patient is a credible source, they then visit the point of service, that is, the location where food was purchased (generally a restaurant or retail store).
From these visits, FDA and state investigators begin tracing the original source of the product. Stores and restaurants often have multiple sources of any one item, such as tomatoes, so the investigators need to determine what suppliers potentially provided the item under examination. They also need to assess the time periods and quantities of product involved, along with the various types or varieties of the product (e.g. for tomatoes, round red, cherry, Roma, grape, on-the-vine, etc.).
The efficiency of this process is largely based on access to purchase and receipt records, and their completeness. As the investigators visit multiple sites identified by credible patients, they can begin again to look for common threads, i.e. suppliers that are common to the sites. At this point the investigators contact the common suppliers and begin tracing the product back to the production site. They generally visit the distribution, production or supply facilities to interview the operators and view records.
The produce industry is a complex industry with many links in the distribution chain, so trackback can be challenging. Again, the efficiency of the traceback is dependent upon cooperation by those involved and availability of all records associated with the suspect product. From the data collected in this process, the investigators hope to identify a specific site or collection of ranches/farms where the initial contamination may have occurred. Of course, this is a vital outcome for the investigation and has very real meaning for our industry. It is at this point that the regulatory agencies can definitively say they can limit their alert or recall to product from a specific supplier, processor or ranch. This often allows the rest of the industry to supply and sell products without safety concerns.
Once the FDA and state investigators identify a supplier and/or sites, they begin the task of searching for a cause of the contamination. They often inspect the site and facilities and evaluate production records to verify that a food safety program was in place and was being followed as required by the plan. They may also choose to sample raw or finished products (especially if the same lots are still available), and may wish to take environmental samples from fields, equipment, agricultural inputs and water for microbial testing. They may interview employees and take personal health histories to see whether anyone who may have contacted the product had a similar infection. Their goal is to try to find the cause for contamination so it can be mitigated and the industry can learn from it.
Learning the process health authorities follow illustrates why it isn’t just a matter of going to a store or restaurant where a patient says he or she ate, and finding the supplier. The regulatory agencies’ goal is to protect public health. Therefore, they must follow the process to gather information systematically so they do not reach false conclusions, which could cause more injury to the public and ultimately to the industry.

Since the illness onset dates in this situation are so broad, what is the likelihood of finding the source?
Tomatoes can be a multi-harvest crop, i.e. tomatoes are harvested off the same vine or plant over a period of several weeks. Tomatoes also can remain in commerce for several weeks, as they can be distributed directly or sent to repackers who in turn hold them and ripen them as the market demands. So, it is not too surprising that the onset dates are broadly separated. As to the likelihood of finding the source, tomatoes represent a difficult challenge for investigators because of the many types of finished products and the diverse production locations. This is further complicated by the fact that the onset period straddled a seasonal transition point for Mexico, California and Florida. Given these issues, I think that as of the date of this writing (June 9) it is still likely that FDA will find a source, whether it is a specific grower, farm, packing shed, distribution center, etc., to which the records of commerce trace. What is more problematic is actually finding the S. saintpaul strain at the site. That has often proven hard to accomplish with any foodborne pathogen, as the bacteria that cause these outbreaks are present only in very low numbers and may not survive in soils or equipment surfaces or other touch points very long, especially if the soils are turned or routine clean-up procedures diminish their populations.


What’s the risk that FDA will implicate a region, rather than a farm? What are we doing to keep them disciplined?
The logical sequence of the investigation and traceback (see related question) is that FDA will identify a region first as they look at suppliers to points of service. Once investigators identify regions, they will begin to use the records and shipping documents to narrow the investigation to specific growers or packers. Once the FDA has a region, the additional traceback efforts to find a specific grower and farm take a few days depending on the quality of the shipping records. By its very nature, FDA is a very disciplined organization. Agents make decisions based on data and records. This can be frustrating for the industry in terms of the time it takes to perform a traceback, and FDA’s unwillingness to jump steps to identify a supplier. By the same token, this approach and adherence to process keeps the traceback focused and decisions fact-based.


What about Florida tomatoes, are they OK?
As of June 8, FDA has excluded Arkansas, California, Georgia, Hawaii, North Carolina, South Carolina, Tennessee, Texas, Belgium, Canada, Dominican Republic, Guatemala, Israel, the Netherlands and Puerto Rico. FDA understands the urgency the industry feels toward narrowing the sources down as quickly as possible so as not to penalize an entire industry and have excluded those regions. Other regions or states had not been excluded as of June 8. (Note: To view a current list of excluded production areas, visit www.pma.com/issues/tomatoesjun08.cfm.)


What about hot house, organic or hydroponic tomatoes?
Many types of tomatoes are sold across the country. Based upon interviews with people who became ill, as of June 8, the FDA and CDC pinpointed round red and plum/Roma tomatoes as linked to the illnesses in specific states. And, based on those interviews, FDA has excluded yellow tomatoes, cherry and grape tomatoes, tomatoes still on the vine and home-grown tomatoes. There is no distinction as to how the tomatoes were grown, e.g. field production, hot house, hydroponic, organic. Therefore, as of June 8, if the product is a round red, plum/Roma tomato, regardless of how it was produced, the alert applies to that product and to items made from that product unless they were grown in the excluded areas. (Note: To view a current list of excluded production areas, visit www.pma.com/issues/tomatoesjun08.cfm.)


Why did FDA go direct to consumers with a de facto recall?
The function of FDA is to protect public health. FDA’s consumer alerts have been driven by the epidemiology and the Salmonella fingerprint testing results identifying the genetic identity of S. saintpaul in people who became sick. Until CDC and FDA are sure the onset period is over, in other words that no one else could get sick, FDA has the obligation to alert consumers against the food item that might cause them to become ill. As of June 8, there had been no specific food recall, strictly speaking, as the source of the tomatoes in question was still unknown. Based on the alerts, retailers and foodservice operators and restaurants have made business decisions whether to provide red round, plum, or Roma tomatoes to consumers. Some have made the decision to remove these types of tomatoes from the market, as it would be unlikely that consumers would want them because FDA has advised them not to eat them.

I am using Romas to make fresh salsa for sale. How do we reassure our clients?
The best advice is to check the CDC and/or FDA Web sites daily to determine (note: the FDA and CDC sites can be accessed via www.pma.com/issues/tomatoesjun08.cfm) : (1) what production areas have been excluded; and (2) whether any S. saintpaul illnesses have been identified in your state and confirmed to have been caused by certain tomatoes.

Can we test products in warehouses now, and release the product if it tests negative?
While this sounds like a great idea, unfortunately this would not be advisable. The problem lies in the sampling. The general feeling in the scientific community is that these contaminations that lead to foodborne illness outbreaks are at low levels and not widely spread. This seems very logical and makes sense when you consider that if the contaminations were large and spread over the majority of the tomato fruit from a given field or farm, then many more than 100 people would be sick to date.
So, that puts us in a position of trying to understand how many tomatoes we would have to sample and test to be sure that we could find every tomato fruit that was contaminated, so we could keep them from going to consumers. Remember, even one sick consumer is one too many.
We could just test every tomato. It would be expensive, but let’s say we could do that. The problem is that the test essentially destroys the tomato. To perform the test required, the tomato is cut up and then juiced so that the microorganism can be isolated.
If you cannot test every tomato, then you need to rely on statistical sampling to have confidence that you can eliminate any tomato that is contaminated. The problem here is that scientists believe the contamination incidents are so infrequent, random and at low levels that there is not a satisfactory statistical solution to the problem.
Think of a field of tomatoes, and let’s say there are 1,000 plants in the block we want to harvest. Let’s also say that we are going to harvest 20 tomatoes from each plant over the next two weeks. That is 20,000 total tomatoes. Now let’s say that somehow during the harvest period, 200 fruits have become contaminated with a very low level of Salmonella. You have all your tomatoes in boxes; how do you sample to be sure you can eliminate those that are contaminated?
More visually, pretend you have 20,000 white marbles in a huge box and mixed in with them are 200 black ones randomly blended in. Since you cannot see bacteria on tomato fruits, our visual analogy would require us to select our black marbles blindly, so pretend that you are blindfolded and trying to pick out the black marbles from our huge box. How many times would you have to “sample” the box to get all 200 black marbles? It is a daunting question, and one we cannot answer now.

Why is it taking so long to identify the source, is there an issue with a lack of industry capability to trace product?
The industry’s traceback capability is not an issue at this time. FDA and the involved states started their traceback efforts early the week of June 2, and were still working through their process at the time of this writing (June 9). The industry is working to provide FDA with information on tomato product movements and production areas to help them focus efforts in the trace.

Are tomatoes safe if they have been through a chlorine wash?
The use of sanitizers in fruit and vegetable wash systems is often misinterpreted. Many believe the sanitizer, e.g. chlorine, sanitizes the surface of the product being washed. In fact, sanitizers are used in wash systems to keep the wash water clean, not to sanitize the fruit or vegetable. The effectiveness of the sanitizer in keeping the water clean is dependent upon how clean the water is (i.e., the level of organic material, e.g. dirt), its pH, the sanitizer used, its concentration, etc. It is important to keep the water clean because, by the very fact that washing removes surface dirt and some portion of the bacteria and fungi on the surface, the wash system could easily become a reservoir for any dirt or bacteria that are washed off. In effect, the wash system could become a contamination point if it wasn’t treated with a sanitizer to kill microorganisms that wash off in the system.
Unfortunately, not all bacteria are washed off during washing. We know that most wash systems reduce the microbial populations on the surface of fruits and vegetables by about 2 logs. So what does that mean? If the product you are washing comes into the packinghouse with 1,000,000 organisms per square centimeter (not uncommon), then after washing in a well-maintained system it would likely have 10,000 organisms per square centimeter. So if raw product enters a packinghouse with a human pathogen on it, the pathogen most likely will not be reduced sufficiently during the wash to prevent illness.
Remember that surface microorganisms on fruits and vegetables are very common. Some are beneficial to the plant, some are beneficial to those consuming the product and some may be plant pathogens. In some ways these surface microorganisms actually protect the product as they compete with human pathogens like Salmonella and E. coli if they should contact the product and limit their growth or perhaps even eliminate them over time.

Why is FDA focusing on point of consumption? The tomatoes sent from a distribution center (DC) to stores in outbreak states are the same as tomatoes sent from the same DC to stores in other states. How is that protecting public health?
This is really several questions that cover many aspects of tomato distribution and FDA procedures. First, the FDA focuses on point of consumption simply because the agency has to start where the illnesses are actually occurring. It is from the people that are actually ill that health investigators can begin to gather important epidemiological data relating to what the person ate in the days leading up to the onset of illness and if that link can be made, where they might have consumed the contaminated tomatoes. From this point-of-consumption data, the FDA can then move to the point-of-service locations, and then on to determining where the tomatoes originally came from.
Distribution centers (DCs) are set up by retailers to serve defined geographical areas by acting as a storage center that receives products from the production locations and warehouses them until they are needed to replenish the stores in the region served by that DC. Often, retail chains will source products like tomatoes from several vendors. As an example, a retailer may use five or six vendors to supply red round tomatoes at any given time, and tomatoes from all or some of these vendors may be in any given DC. Retailers do this to ensure that their customers always have round red tomatoes available so that if one grower or supplier has a weather-related problem and can’t harvest, the other vendors can cover the volume. Price, logistics, quality, season and potential to supply other items also play into the concept of multiple vendors supplying individual DCs. Therefore, it is quite likely that a DC that sends tomatoes from a supplier who inadvertently supplied contaminated tomatoes can also ship tomatoes to stores from other suppliers whose tomatoes are safe to eat.
As part of the FDA’s traceback process, the investigators would use the shipping records into the DC to determine where and when the tomatoes from each vendor were received in the DC, and the outgoing DC shipping records to determine where and when they were shipped to the retail stores. The investigators then match this information with the epidemiologic data to link illness with stores, the DC and ultimately the source of the tomatoes that were received into the DC. It is an investigative process that takes time, but genuinely serves the public by identifying the source of contaminated product and ensuring it is removed from points of sale or consumption.

I’m a greenhouse grower. Can I set up pathogen testing, for example to test the water and the product for pathogens?
The simple answer to this question is that you can never test your way to food safety. Testing is merely a tool, and an imperfect one at that. Without knowing the grower in question, I would first advise that they have a documented food safety program in place to govern their greenhouse operations. This food safety plan needs to be driven by a risk assessment that outlines every operation at the facility and identifies where potential contaminations could take place, e.g. the water quality used for irrigation, potential for contamination entering the facility via the air from surrounding areas, hygiene practices by workers handling the fruit, sanitation of harvest baskets, purity of the potting soil, etc. The absolute best way to avoid an illness associated with their products is to prevent the contamination from ever occurring.
I suspect this greenhouse grower has a food safety program if they are asking about testing, but strongly encourage them to go back through it and see if it is adequate. Many people in our industry mistakenly think a food safety program is the documentation you use to pass food safety audits. In fact, a food safety program is a risk assessment/risk management program, and the documentation is simply the verification that you are indeed managing the risks to the best of your ability.
Microbial testing can have a beneficial role in validating the effectiveness of risk management procedures. Water testing is one of these areas. The greenhouse operator should test the incoming water for the operation on at least a monthly basis. Before testing, they should confirm their water source; i.e. what is the source (e.g. municipal, well, canal, etc.), and what risks does it pose? Obviously a municipal water source or a deep well poses much lower risks than an open water source like a canal or river. In any event, the operator needs to put management procedures in place to ensure the water does not become contaminated.
Once that is done, microbial testing is a way to monitor the effectiveness of the risk management procedures. Often, testing for specific pathogens is not necessary, but perhaps a simple test for generic E. coli as an indicator for fecal contamination would suffice and be more cost-effective.
In the end, there is no perfect organism to test for, and the greenhouse operator will have to make his or her own decision on whether to test for pathogens directly or not. If the operator feels through a risk assessment that there is a potential for Salmonella or E. coli O157:H7 contamination because the water source is in proximity to a potential source, e.g. dairy, poultry, bird nesting areas, etc., then pathogen testing may be part of managing that risk and would be advisable. Microbial testing can also be an excellent way to test for the efficacy of equipment sanitation methods. Simple swab testing after cleaning and sanitation of equipment, harvest totes or any food contact surfaces is a great way to verify that the cleaning chemicals and methods used for sanitation are adequate. Often a quick test for total plate count (referred to in shorthand as TPC) is sufficient for this purpose. There are also several “do it yourself”, ATP-based (Adenosine triphosphate) test kits available that have the advantage of lower cost and time savings that would fit this bill.
Product testing has become a hot topic in our industry. A greenhouse operator can certainly develop a testing program for finished tomatoes, but it is important to understand what the testing data means and more important to know what it does not mean.
First it is important to find a certified laboratory that uses FDA and/or AOAC-approved methodologies for measuring pathogens. It is important that your grower understands the test procedures, handling practices by the laboratory and how the data will be reported and to whom. Once the testing itself is understood, then you move to the more difficult issue of sampling.
This is where perception needs to come to grips with reality. If this greenhouse operator had a large, wide-spread contamination, e.g. the water used for rinsing the harvest totes was contaminated, a sampling of tomatoes post-harvest may indeed detect the contamination and the tomatoes could be destroyed. But absent a catastrophic event, the general feeling in the scientific community is that these contaminations that lead to foodborne illness outbreaks are at low levels and not widely spread, e.g. a single ill or recovering worker that did not wash his/her hands properly after use of a restroom and handled perhaps a few hundred fruit.
So that puts us in a position of trying to understand how many tomatoes we would have to sample and test to be sure that we could find every tomato fruit that was contaminated so we could eliminate them from going to consumers. (See related question.) Your grower cannot test every tomato, as it is too expensive and it results in the destruction of the fruit. So while the idea of product testing for pathogens sounds great, in reality the resources required are better spent in prevention of contamination in the first place.