On February 5, 2018, staff at the State Hygienic Laboratory (SHL) at the University of Iowa (Coralville, Iowa) observed a notable increase in the number of stool samples submitted for Salmonella testing. Whole-genome sequencing and serotyping revealed patterns of genetic similarity between isolates obtained from these samples. By completing epidemiologic interviews of affected persons and performing food sample testing on suspected food products, the Foodborne Rapid Response team of the Iowa Department of Public Health (IDPH) was able to identify the source of the outbreak as prepackaged chicken salad sold by a Midwest grocery store chain. By February 9, the grocery store chain voluntarily removed the product from all of its Iowa stores; on February 13, IDPH and the Iowa Department of Inspections and Appeals (DIA) issued a joint consumer advisory notification warning customers that the product was implicated in multiple cases of Salmonella illness.

Using the PulseNet national molecular subtyping network for foodborne illness surveillance (a national laboratory network that compares the DNA of bacteria from patient samples to find clusters of disease that might represent unrecognized outbreaks), we identified a total of 265 persons from 8 states with Salmonella Typhimurium illness as part of this outbreak. Of those, 240 were from Iowa. Ninety-four hospitalizations were reported (35.5% of cases), including 1 person from Iowa who died. Decisive, cooperative actions undertaken between the Iowa SHL and IDPH, together with the Food and Consumer Safety Bureau within the Iowa DIA, resulted in the removal of contaminated product within 3 days from the initial identification of the genetically related samples, averting what could have been a larger outbreak.

Using a method reported by Scharff et al. (8), we estimated the number of cases of Salmonella Typhimurium averted by responding rapidly to this outbreak (Table 1). As a result of the alert raised by Iowa public health staff, the US Department of Agriculture (USDA) issued a recall for ≈20,630 pounds of potentially contaminated chicken salad from grocery stores in 8 states (9). According to USDA records, the manufacturer reported that 5,397 pounds (26.2%) of the product were recovered (9). Using estimates for product loss at the consumer level (10,11), we calculated that 13,481 pounds of the product were consumed, implying ≈20 confirmed cases of Salmonella Typhimurium per 1,000 pounds of product consumed. Assuming this rate of disease transmission also applied to the quantity of product that was successfully recalled, we conclude that ≈94 cases of Salmonella Typhimurium were averted through the expedient recall of the chicken salad (i.e., cases that would have been reported had the recalled product been consumed). This estimate also assumes that all reported cases of Salmonella infection in the outbreak resulted from consumption of the contaminated product (Appendix 1)

Our estimate does not account for potential underdiagnosis resulting from variations in medical care seeking, specimen submission, and laboratory testing. Scallan et al. (7) estimated that, for every reported case of nontyphoidal Salmonella, 29.3 (90% credible interval 21.8–38.4) cases are likely not reported; therefore, the number of cases averted due to the product recall, including underdiagnosed cases, is estimated as 2,751 (range 2,047–3,605). Those cases would have occurred had the recalled product been consumed; of those, 94 would have been reported and the rest underdiagnosed (Appendix 1 Table 4).

Using cost of illness estimates for nontyphoidal Salmonella generated by USDA/ERS (12), we estimated the economic costs to society averted by responding rapidly to this outbreak (Table 2). All costs are inflation-adjusted to 2018 US dollars. For our estimate of 94 cases averted, we calculated averted economic costs to society of $601,563 in direct medical costs and $31,618 in productivity losses resulting from missed working days in nonfatal cases. The total estimate of averted costs rises to $844,000 to $1 million when accounting for underdiagnosis. These numbers likely constitute an underestimate because we were conservative in selecting input parameters in cases where uncertainty or feasible ranges exist (Appendix 1). Furthermore, our analysis does not consider secondary effects that could provide additional benefits, such as prevention of future potential outbreaks through providing industry with information by which to improve their processes.

Quantifying and communicating effects such as the amount of illness and economic costs prevented by response and prevention efforts to policymakers and other appropriate audiences using a clear and systematic approach helps to show the value in investing in a robust, responsive, and collaborative public health infrastructure. Although data from outbreak events may lack some of the information desired for a direct calculation of the effect of interventions on population health, methods do exist that aid in making conservative estimates. Routinely calculating and communicating these estimates using direct and relatable outcome indicators for a variety of public health actions helps demonstrate the importance of investing in the ability to respond to outbreaks when they occur and of sustained investment in measures that prevent these outbreaks from occurring. Future analyses could expand upon our approach by examining the sequence of public health actions in relationship to the rise and fall of daily case counts, which may provide additional useful insights into the value of timely information. Incorporating information about the distribution of product sales and recovery could yield specific knowledge for future studies.