Why Outbreaks Don’t Always Result in Recalls

On December 12, 2017, in response to a cluster of pathogenic E. coli illness reports in Eastern Canada, the Public Agency of Canada (PHAC) released the following: “The Public Health Agency of Canada is collaborating with provincial public health partners, the Canadian Food Inspection Agency (CFIA) and Health Canada to investigate an outbreak of Escherichia coli O157, commonly called E. coli. The outbreak involves three provinces and is linked to romaine lettuce. At this time, there are no product recalls associated with this outbreak. ….”

How could there be no recalls? The outbreak continued, with PHAC issuing a follow-up statement on December 21 urging Canadians in the affected provinces to eat lettuce other than romaine. A week later, the U.S. CDC issued a statement indicating that U.S. cases appeared linked to the Canadian outbreak, but noting that a food vehicle had not yet been identified. Many industry members and news reporters wondered how these announcements could be made without a corresponding recall.

The answer requires an understanding of how outbreak investigations are conducted. CDC has a great description of the process on its website. Essentially, the steps are:

1. People become ill and go to their health care provider.
2. The health care provider sends a sample to the laboratory for analysis.
3. Lab tests are reviewed by public health officials who confirm that illnesses are related (historically through pulsed field gel electrophoresis; now using whole genome sequencing [WGS]).
4. Public health officials interview case patients to ask what they ate in the days leading up to illness. This is the epidemiological investigation. It is one of the tricky points in the process — how accurate are people’s memories? At this point, a week or more has probably passed since illness began. Can people remember every topping on a sandwich, every seasoning they used, etc.? Shopper-card information and purchase history can help verify and supplement this information.
5. Public health officials match up the interviews from cases and compare the foods eaten to that of the regular population. If most or all cases report eating something that is less frequently consumed in the regular population, that is a signal that further investigation is warranted.
6. The traceback investigation follows on the heels of the epidemiological investigation. After a food(s) is suspected, regulators begin tracing it back to its source, starting with the point of purchase (restaurant, grocery store, etc.). This is done for as many cases as possible, to see if, for example, the path of a product someone ate in Michigan intersects with a product bought in California.

So, why not just test products? Sometimes, product testing can yield a smoking gun. More often, however, the contamination event was short lived, affecting a relatively small amount of product which is out of distribution by the time the outbreak is identified. This is particularly true when products are consumed quickly because they are perishable (e.g., fresh produce) or popular (e.g., bottled water).

There are several other challenges in the process that may render the traditional epidemiological investigation unsuccessful, such as the issue of specificity. For example, in the case of romaine, did people eat bagged salad, romaine hearts, leaves, etc.? There also may be situations in which a common source would be very difficult to tease out of a questionnaire, such as an ingredient — as in the salami outbreak ultimately linked to black pepper.

There are also challenges in the traceback investigation. Despite ongoing efforts, recordkeeping practices and supply-chain information transfer can be insufficient, resulting in dead ends that hinder or prevent a successful traceback. Even perfect records don’t always reveal a single production or manufacturing location. If, for example, there was a short-term issue with an irrigation water system, farms of different companies growing different crops may have used the water. So finding a common food and source would be impossible, even though illness was related to one specific issue.

These scenarios represent situations in which an outbreak is ongoing without a recall being issued. In too many instances, the root of the problem is never solved. However, advances in science and technology provide hope that improvements are on the horizon.

WGS is the best tool to infer relationships between microorganisms. Not only can it help signal an outbreak when people become ill, but public health agencies can look in an ever-growing database to determine if the WGS pattern has been seen before. Alhough the industry has concerns about the assertion of causality, a sequence match could provide a piece to the puzzle — but it should not be the only piece.

More information is collected today than ever before, though we may not always have the capability to use it. “Big data” promises to make more sense of existing information, and may be able to find patterns that the human eye could never discern.

The bottom line is that there are many reasons an outbreak investigation may not be able to find a cause, in which case, there is nothing to recall. However, those days are numbered as society becomes more equipped to collect and handle data. Prevention will always be the best defense against a recall. Do all you can to produce safe food so an outbreak isn’t associated with your company.