Background Although preventable through timely screening and treatment, congenital syphilis (CS) cases are rising in the United States. Individual CS risk factors are well-described, but county factors are not. We developed a predictive model that could identify county risk factors and use these to predict counties at highest risk for future CS.

Methods We included all 3,142 US counties. To identify county risk factors, we defined the outcome as a county having ≥1 CS case during 2012–2015. Case counts were taken from the National Notifiable Disease Surveillance system; county data were from a 2015 county health rankings analytic file. We used a stepwise logistic regression model to identify adjusted associations between CS and county factors. Retained predictor variables were each assigned a score based on the strength of their association with the outcome. Risk scores were calculated by summing predictor scores for each county. Counties with risk scores ≥24 were defined as high-risk for having ≥1 CS case. We cross validated the model using coefficients from the final 2012–2015 model to predict high-risk counties for 2016–2017 and compared predicted and actual counties by calculating the area-under-the-curve (AUC) value.

Results Our model identified 721 counties as high-risk for CS (sensitivity: 80.1%; specificity: 83.7%). County predictors included: 2015 Medicaid expansion status, presence of a metropolitan area, population size, income inequality, syphilis among women and men who have sex with men, violent crime rate, and the population proportions that were black, Hispanic, and uninsured. The final model based on 2012–2015 CS data was predictive of 2016–2017 CS counties (AUC: 88.1%).

Conclusion Given the damaging yet preventable nature of CS, enhancing prevention is a priority. The ability to predict counties at highest risk for CS based on county factors may help target CS resources where they are needed most.

Disclosure No significant relationships.