Predicting distributions of blacklegged ticks, Lyme disease spirochetes and human Lyme disease cases in the eastern US, 2022, Burtis et al

Full title: Predicting distributions of blacklegged ticks (Ixodes scapularis), Lyme disease spirochetes (Borrelia burgdorferi sensu stricto) and human Lyme disease cases in the eastern United States

Abstract

Lyme disease is the most commonly reported vector-borne disease in the United States (US), with approximately 300,000 -to- 40,000 cases reported annually. The blacklegged tick, Ixodes scapularis, is the primary vector of the Lyme disease-causing spirochete, Borrelia burgdorferi sensu stricto, in high incidence regions in the upper midwestern and northeastern US. Using county-level records of the presence of I. scapularis or presence of B. burgdorferi s.s. infected host-seeking I. scapularis, we generated habitat suitability consensus maps based on an ensemble of statistical models for both acarological risk metrics. Overall accuracy of these suitability models was high (AUC = 0.76 for I. scapularis and 0.86 for B. burgdorferi s.s. infected-I. scapularis). We sought to compare which acarological risk metric best described the distribution of counties reporting high Lyme disease incidence (≥10 confirmed cases/100,000 population) by setting the models to a fixed omission rate (10%).

We compared the percent of high incidence counties correctly classified by the two models. The I. scapularis consensus map correctly classified 53% of high and low incidence counties, while the B. burgdorferi s.s. infected-I. scapularis consensus map classified 83% correctly. Counties classified as suitable by the B. burgdorferi s.s. map showed a 91% overlap with high Lyme disease incidence counties with over a 38-fold difference in Lyme disease incidence between high- and low-suitability counties. A total of 288 counties were classified as highly suitable for B. burgdorferi s.s., but lacked records of infected-I. scapularis and were not classified as high incidence. These counties were considered to represent a leading edge for B. burgdorferi s.s. infection in ticks and humans. They clustered in Illinois, Indiana, Michigan, and Ohio. This information can aid in targeting tick surveillance and prevention education efforts in counties where Lyme disease risk may increase in the future.

Paywall, https://www.sciencedirect.com/science/article/abs/pii/S1877959X22001030

 

CDC needs to be helping develop better diagnostics, not wasting its time playing with models.

High-predictive models that say Lyme cannot be in a given county or state, or are less likely to be there, are meaningless to the individuals who contract it in those counties and states. Historically, these kinds of models can discourage testing in those areas thought to be low incidence - testing which admittedly sucks. Which brings me back to my first sentence.