Salmonella Serotypes Associated with Illnesses after Thanksgiving Holiday, United States, 1998–2018

We sought to determine which Salmonella serotypes cause illness related to the Thanksgiving holiday in the United States and to foods disproportionately eaten then (e.g., turkey). Using routine surveillance for 1998–2018 and a case-crossover design, we found serotype Reading to be most strongly associated with Thanksgiving.


The Study
The Laboratory-based Enteric Disease Surveillance (LEDS) system captures enteric infections with Salmonella species in the United States through passive surveillance of laboratory-confi rmed isolates. State and territorial public health laboratories serotype Salmonella isolates; any unusual serotypes are sent to the Centers for Disease Control and Prevention's National Salmonella Reference Laboratory (National Center for Emerging and Zoonotic Infectious Diseases, Division of Foodborne, Waterborne, and Environmental Diseases, Enteric Diseases Laboratory Branch) for further characterization. This surveillance system obtains demographic information; specimen source; collection date; test result; serotype; and, if available, outbreak association. We included infections that had fully serotyped Salmonella isolates that occurred from 1998 through 2018 and excluded isolates confi rmed to be outbreak-associated.
Using a case-crossover design, we determined a case window for each year using the date of Thanksgiving, a minimum incubation period, and a window length. To account for seasonal variation in infections, we created nonexposure case windows before and after Thanksgiving by using the same case window length with a washout period separating the exposure and nonexposure windows (Appendix Figure, https://wwwnc.cdc.gov/EID/article/28/1/21-1986-App1.pdf). Our primary analysis used a 7-day case window after a minimum 2-day incubation period after Thanksgiving, with a 5-day washout. Control windows were also 7 days.
We conducted sensitivity analyses evaluating different minimum incubations, case windows, and washouts and stratifi ed all analyses by serotype. We sought to determine which Salmonella serotypes cause illness related to the Thanksgiving holiday in the United States and to foods disproportionately eaten then (e.g., turkey). Using routine surveillance for 1998-2018 and a case-crossover design, we found serotype Reading to be most strongly associated with Thanksgiving.
during holidays and nonholidays, we used all Salmonella cases associated with serotypes other than the serotype under consideration as the comparison group. We calculated odds ratios (ORs) for the entire study period and compared ORs for 1998-2007 with those for 2008-2018. We conducted descriptive demographic and clinical analyses with SAS versions 9.4 (https://www.sas.com) and calculated ORs and unadjusted p values with R version 3.6.1 (R Foundation for Statistical Computing, https:// www.r-project.org) using unconditional maximumlikelihood estimation with 95% CIs from the normal approximation.

Conclusions
Salmonella Reading was the serotype most strongly associated with illness during the Thanksgiving holiday. Given the dramatic increase in turkey consumption around Thanksgiving, one might expect that serotypes we identified are primarily associated with turkey consumption, and indeed, Reading caused a multistate outbreak with a raw turkey source during 2017-2019 (8), and a new clone of this serotype has emerged since 2014 in commercial turkey production (9). Other serotypes significantly associated with Thanksgiving in our study (i.e., Hadar, Schwarzengrund, and Heidelberg) have also been associated with turkey (6,10).
Other significantly associated serotypes are not among those most commonly identified in turkey (e.g., Heidelberg and 4, [5],12:i-are more commonly identified in chicken; Derby, Brandenburg,  . However, all these serotypes have been found in turkeys and in retail samples of turkey or have been associated with outbreaks attributed to turkey (11)(12)(13)(14)(15). Some of the serotypes significantly increased after Thanksgiving, such as, Baildon and Ohio, were rare, causing <200 illnesses annually, and were not reported among food animals, retail products, or outbreaks during 2015-2020 (7). Although our study may have identified serotypes associated with other foods eaten during the Thanksgiving holiday, particular attention probably should be paid to evidence of these serotypes emerging in turkey production.
The first limitation of our study is that LEDS is a passive surveillance system and does not capture mild or asymptomatic infections for which ill persons do not seek healthcare or submit a specimen. Although we removed cases reported as outbreakassociated, those data are not reported by all states, and some outbreak-associated cases most likely are included. Missing data in LEDS varies, but serotype and date are largely complete. Although sensitivity analyses demonstrate consistency across time windows, misalignment of windows with causative exposures could have resulted in biases, possibly differential, from, for example, differences in healthcare seeking because of the holiday itself. Our study may be subject to ecologic bias because individual food exposures are unknown. We did not adjust for multiple testing because this analysis is intended to be hypothesis-generating rather than confirmatory.
Our case-crossover approach could be helpful for other pathogens and their subtypes that are likely to cause illnesses from a certain food disproportionately eaten during a brief period, such as turkey during the Thanksgiving holiday. Our technique provides unique insights into the causes of sporadic illness throughout the year and their changes over multiyear periods using no more than routine surveillance data and may provide valuable information to industry, regulators, and public health officials that could help guide monitoring and interventions to prevent illnesses and their associated morbidity and mortality. Consumers can also help protect themselves from Salmonella by following the 4 steps to food safety (https://www.cdc.gov/ foodsafety/keep-food-safe.html). Appendix Figure. Example of case windows used in primary analysis before and after Thanksgiving Day, United States, 2017. *The 2-day incubation period represents the minimum incubation period; the 7-day case window includes the maximum incubation and reporting period.