In a story last summer for Food Safety News, I looked to see if the U.S. Department of Agriculture’s Food Safety and Inspection Service (FSIS) recalls were biased toward any one day of the week. This inquiry followed what seemed to be several late Friday recall announcements, including a nationwide recall of contaminated ground turkey implicated in a multistate outbreak of Salmonella infection.
That Friday recall and others raised questions about whether the timing of recall notices could potentially leave consumers uninformed about unsafe food over a weekend.
That preliminary look did not show any significant differences in the daily publication of recall notices, but with the close of 2011, I decided to take a longer look.
This more comprehensive look included recalls going back to 2005 for USDA and Food and Drug Administration (FDA) monitored foods. The USDA data also detailed whether there was a specific pathogen found in the foods (or epidemiologically linked to an outbreak), if any illnesses were associated with the recalled item, and the class of the recall as designated by the USDA’s Food Safety and Inspection Service (FSIS). The FDA does not specify classes in its recall notices, so only the first two categories could be assessed.
The conclusions of this more detailed analysis show a different story from the earlier one.
On the USDA front; from 2005-2011 there were more recalls announced on Fridays than would be expected by random chance (p<.001). Indeed, 2006, 2007, 2009, 2010 and 2011 all had significantly higher numbers of recalls on Fridays (p<.05).
The total number of Class I recalls on Fridays were also higher from 2005-2011 (p=.003). Class I recalls, by definition, are for dangerous or defective products that could cause serious health problems or deaths.
While on a per year basis, only 2006 and 2011 have significantly higher number of Class I recalls on Fridays than expected, 2007 and 2009 are very nearly significantly higher (p=.055 and p=.062, respectively).
The red line indicates the expected number of recalls if equally distributed throughout the week. A proportion significantly higher than expected of p<.01 indicated by *.[/caption] In addition, if weekend recall dates are excluded, the pattern holds for total recalls with a greater number on Fridays than other weekdays (p=.002). For Class I recalls, there was no significant difference in the number of Friday recalls compared with the number expected for weekdays, but there were still more alerts published on that day than any other day of the week (p=.086). [caption id="attachment_59279" align="aligncenter" width="481"] The red line indicates the expected number of recalls if equally distributed throughout the week. A proportion significantly higher than expected of p<.01 indicated by *.[/caption] There was no trend in any day of the week having more pathogen or illness related recalls than any other day in the seven years analyzed. This was true when only weekdays were analyzed, as well. This indicates that recall notices related to outbreaks of foodborne illness and due to specific pathogens found in food have been evenly spaced throughout the week. While this is the case, when taken as a whole there were more Class I and Class II recalls on Fridays than would be expected. The significance of this difference for Class I recalls was eliminated when analyzing only weekdays, but held true for Class II notices, perhaps showing a propensity for putting out more urgent recalls on Fridays. One particularly interesting finding was an observable pattern in an increasing number of Class II and Class III recalls during the time period studied, while Class I recall numbers remained relatively static. There was a significant difference in the number of Class II recalls starting in 2008 the percentage that Class II recalls make of the total number of recalls has continued to increase (P<.05). Class III recalls remain rare and, as such, there is little statistical power in any observations made.
The increasing number of Class II and slightly Class III with the stagnant number of Class I led to an overall increase in the number of recalls per year by the USDA from 56 in 2005 to 108 in 2011.
The previous article on this topic did not evaluate FDA recalls. This oversight has been amended this time around, and the findings are similar to those for FSIS recalls.
A glance at the total recalls per day chart shows that between 2005 and 2011 a significantly higher number of recalls were issued on Fridays than would be expected (P<.01). [caption id="attachment_59282" align="aligncenter" width="481"] The red line indicates the expected number of recalls if equally distributed throughout the week. A proportion significantly higher than expected of p<.01 indicated by *.[/caption] This holds when excluding the weekends that had so few recalls. [caption id="attachment_59283" align="aligncenter" width="481"] The red line indicates the expected number of recalls if equally distributed throughout the week. A proportion significantly higher than expected of p<.01 indicated by *.[/caption] The FDA data again points to a bias toward publishing recall notices on Fridays over other days of the week and, if anything, the tendency is more pronounced with this agency than with the FSIS. Indeed, recall announcements related to foods testing positive for pathogens and those associated with illnesses were also more common on Fridays, compared to the whole week or just the work week, a trend not seen in FSIS recalls.
Fridays have consistently seen significantly more than an expected number of recall notices for both the whole week and weekdays only for the entire seven years examined. However, due to the low number of recalls related to illnesses in all the years, statistical tests cannot be performed on a per year basis. This is also true of pathogen-related recalls prior to 2009.
There was again a pattern of certain types of recalls increasing over the years, as well as the overall number of FDA recalls increasing. The total number of recalls in 2005 was 145 and this reached a max of more than 800 because of multiple peanut butter-related recalls in 2009. Excluding that outlier, there was still a steady increase in the number of FDA recalls from 2005 to the end of 2011.
This tendency toward releasing recall statements on Fridays, more than on other days of the week, is likely not intentional but could be problematic, because people who do not consume media over the weekend might be less likely to be informed in a timely manner and, as a result, could eat recalled food. On the other hand, Friday recall notices may be the result of a sense of urgency to get make the information public before the work week ends.
Indeed, processes within the regulatory agencies may be causing an inordinate number of recall notices to be issued on Fridays. Such factors could include laboratory procedures to find contamination in samples, communication delays between the lab, regulators and the companies voluntarily recalling their products, and the general timeline of work in the offices. These, along with other processes, including political pressures, could lead to significantly more recalls on Fridays.
Importantly, recalls are initiated by private companies once the potential for hazard is found. The company distributing or producing the food recalls the products in coordination with the regulatory agency, but the company is ultimately responsible for when the recall is initiated.
It could take days before information regarding potential hazards makes it from a lab or other area of initial discovery, through the regulatory agency to the producer, and then for a recall alert to be issued. Thus, the high number of Friday recalls could be seen as an attempt to get the information out before the weekend, while people are still likely to be tuned into the news.
While the causes leading to the high number of Friday recalls cannot be determined with this small look into the past few years, what can be seen is a potential for lack of communication between the consumer and the food regulators. Easy access to information regarding potentially dangerous foods should be an issue of primary importance for both the USDA and FDA, and these agencies can work to ensure that the information is not going out when people are less likely to get it.© Food Safety News