Abstract
Background
Since 2014, multiple outbreaks of HIV among people who inject drugs (PWID) have occurred across the United States along with hepatitis C (HCV), skin and soft tissue infections (SSTI) and infective endocarditis (IE), creating a converging public health crisis.
Methods
We analyzed the temporal patterns of infectious disease and overdose (OD) using a hierarchical Bayesian distributed lag logistic regression model examining the probability that a given geographic area experienced at least one HIV case in a given month as a function of the counts/rates of OD, HCV, SSTI, IE and associated medical procedures at different lagged time periods.
Results
Current-month HIV is associated with increasing HCV cases; abscess incision and drainage, and SSTI cases, in distinct temporal patterns. For example, one additional HCV case occurring 5 and 7 months previously is associated with a 4% increase in the odds of observing at least one current-month HIV case in a given locale (odds ratios: 1.04, 1.04; 90% credible intervals: 1.01-1.10, 1.00-1.09). No such associations were observed for echocardiograms, IE, or OD.
Conclusion
Lagged associations in other infections preceding rises in current-month HIV counts cannot be described as predictive of HIV outbreaks but may point towards newly discovered epidemics of injection drug use and associated clinical sequalae, prompting clinicians to screen patients more carefully for substance use disorder and associated infections.
