When there is a foodborne outbreak, the ability to identify a common source is dependent on three things: recordkeeping, a narrow request for records, and limited product commingling.
The FDA final traceability rule addresses only the first component. My friend and colleague Dr. Bob Whitaker recently wrote a terrific white paper with retrospectives on 4 outbreaks, showing how the rule could have made a difference. But recordkeeping alone will not solve all traceability woes. At the same time, the traceback process can be substantially improved even if recordkeeping is imperfect, if the scope of records requests and/or product commingling is better managed. Regulators can increase the success of tracebacks by focusing their records requests. For products that are extensively commingled, even narrow records requests and good recordkeeping will not be able to tease out the implicated ingredient. Industry and regulators must look beyond recordkeeping if tracebacks are to be improved.
Better recordkeeping
Plenty has been written about the need for better records to aid in tracing the path of a product from the point of sale or purchase back toward its origin. The FDA final traceability rule requires retail and foodservice operators to have records of the lot numbers of products received. The “last mile” of the supply chain has long argued that in a recall situation, they take a conservative approach and generally remove all product, regardless of lot number. In a traceback situation, however, these “buffers” are not protective; instead they increase the number of possible sources of a contaminated product. The final rule alleges to address this gap. However, the benefit of having the lot number available at the retail establishment is quickly undone by broad requests for records by regulators.
Narrower records requests
Outbreak investigations start at the local level, generally the store at which someone purchased the food causing illness. By the time the outbreak investigation commences it could be weeks later, and people may not remember the exact day of purchase. The increase in online food purchases, whether of prepared meals or groceries, combined with the use of credit card and loyalty programs, should allow investigators to more frequently zero in on the purchase most likely associated with illness.
When possible, records requests should be focused on the single date most likely to be associated with illness. However, when someone has multiple exposures linked to an establishment (e.g., they eat lunch at the same restaurant every day), it is not possible to discern ‘the’ eating event that caused illness.
An industry led pilot in 2020 showed that when credit card or shopper card data were used to pinpoint a purchase, records indicating the possible suppliers and product lots could be readily provided.
Granted, a broader request for records can be helpful in revealing purchasing patterns and aberrations. However, as the industry complies with the recordkeeping requirements of the traceability rule, there should be limited need to look for patterns. If the establishment of interest can pinpoint the lot number(s) associated with product sold that day, as will be required, this should suffice. Asking for records that cover a longer timeframe is akin to retail food establishments being able to provide a range of possible suppliers and lot numbers, especially when the implicated product is one that moves through the system quickly. Regulators can undo the value of retailers maintaining lot numbers if records requests are large.
A third component of traceability that is not addressed through the traceability rule is commingling of product. The aforementioned leafy green pilots showed that, for processed products such as bagged salads, the processor could be quickly identified and the possible growers and corresponding lots of raw material could also be identified. However, once several raw material lots are combined into one bag of salad, it is not possible to determine which grower/ field was the origin of contamination. By comparing several suspect supply chains, hopefully as focused as possible, it could be possible to rule out some fields/ ranches and rule in some others.
Identifying the source of contamination is the goal of a traceback and requires as strong a signal with as little noise as possible. Records that are unable to definitively identify where a particular product came from contribute to noise, and the rule aims to address this. However, expansive record requests also create a lot of noise, as does product commingling. In order to benefit from the extensive efforts and investments made to comply with the rule, other practices need to change as well. On the regulatory side, this means tightening the timeframe for records requests, and on the industry side, this means thinking about opportunities to limit commingling.