Identification of United States Counties at Elevated Risk for Congenital Syphilis Using Predictive Modeling and a Risk Scoring System, 2018

Abstract

Background: The persistence of congenital syphilis (CS) remains an important concern in the United States. We use the 2018 data to refine a previous predictive model that identifies US counties at elevated risk for CS in 2018.

Methods: Using county-level socioeconomic and health-related data from various sources, we developed a logistic regression predictive model to identify county-level factors associated with a county having had 1 or more CS case reported to the National Notifiable Diseases Surveillance System in 2018. We developed a risk scoring algorithm, identified the optimal risk score cutpoint to identify counties at elevated risk, and calculated the live birth to CS case ratio for counties by predicted risk level to compare counties at elevated risk with counties not at elevated risk.

Results: We identified several county-level factors associated with a county having 1 or more CS case in 2018 (area under the curve, 88.6%; Bayesian information criterion, 1551.1). Using a risk score cutoff of 8 or higher (sensitivity, 83.2%; specificity, 79.4%), this model captured 94.7% (n = 1,253) of CS cases born in 2018 and identified 850 (27%) counties as being at elevated risk for CS. The live birth to CS case ratio was lower in counties identified as at elevated risk (2,482) compared with counties categorized as not at elevated risk (10,621).

Conclusions: Identifying which counties are at highest risk for CS can help target prevention efforts and interventions. The relatively low live birth to CS case ratio in elevated risk counties suggests that implementing routine 28-week screening among pregnant women in these counties may be an efficient way to target CS prevention efforts.