Modelling counterfactual incidence during the transition towards culture-independent diagnostic testing

Abstract

Background: Culture-independent diagnostic testing (CIDT) provides rapid results to clinicians and is quickly displacing traditional detection methods. Increased CIDT use and sensitivity likely result in higher case detection but might also obscure infection trends. Severe illness outcomes, such as hospitalization and death, are likely less affected by changes in testing practices and can be used as indicators of the expected case incidence trend had testing methods not changed.

Methods: Using US Foodborne Diseases Active Surveillance Network data during 1996-2019 and mixed effects quasi-Poisson regression, we estimated the expected yearly incidence for nine enteric pathogens.

Results: Removing the effect of CIDT use, CIDT panel testing and culture-confirmation of CIDT testing, the modelled incidence in all but three pathogens (Salmonella, Shigella, STEC O157) was significantly lower than the observed and the upward trend in Campylobacter was reversed from an observed 2.8% yearly increase to a modelled -2.8% yearly decrease (95% credible interval: -4.0, -1.4).

Conclusions: Severe outcomes may be useful indicators in evaluating trends in surveillance systems that have undergone a marked change.

Keywords: Campylobacter, Salmonella, Shigella; Bayesian; FoodNet; Shiga toxin-producing Escherichia coli, Vibrio; counterfactual modelling; culture-independent diagnostic test.