Estimation of new HIV diagnosis rates among high-risk, PrEP-eligible individuals using HIV surveillance data at the Metropolitan Statistical Area level in the United States.

INTRODUCTION

New HIV diagnoses have fallen in the past decade due to increased HIV testing, earlier diagnosis, earlier antiretroviral treatment, improved linkage to care and engagement in care, and the recent increased uptake of pre-exposure prophylaxis (PrEP). We propose a novel method to compute the rate of new HIV diagnoses at the Metropolitan Statistical Area (MSA) level in the US to support the evaluation of comprehensive treatment and prevention efforts over time.

METHODS

The number of new HIV diagnoses, number of individuals with a PrEP indication and aggregated person-time exposed to PrEP during the years 2012 to 2017 were used to compute a new HIV diagnosis rate for people at risk of HIV excluding those already on PrEP for the 105 MSAs in the US with published HIV surveillance data. In our calculation of person-time with a PrEP indication, time-at-risk excluded time on PrEP and time after an HIV diagnosis. We used a multivariate Poisson regression model to estimate HIV diagnosis rates by year and location.

RESULTS

From 2012 to 2017, the aggregate HIV diagnoses rate among high-risk individuals with an indication for PrEP in the 105 MSAs decreased from 4.14 per 100 person-years (PY) (95% CI 4.10 to 4.19) to 3.26 per 100 PY (95% CI 3.22 to 3.30). For the 25 US MSAs that overlapped with an ongoing large randomized clinical trial of PrEP in men who have sex with men (MSM), the HIV diagnosis rate from 2012 to 2017 decreased from 4.86 per 100 PY (95% CI 4.80 to 4.93) to 3.61 per 100 PY (95% CI 3.56 to 3.66), a decline that was more rapid than in non-study MSAs (IRR for trial site 1.19, 95% CI 1.18 to 1.20).

CONCLUSIONS

We propose a model to estimate the background HIV diagnosis rate in people at risk for HIV and with a PrEP indication in US MSAs (excluding those on PrEP) using publically available surveillance data which can evaluate trends over time. Data generated using this methodology could be used by policy makers and local HIV prevention specialists to evaluate and monitor their prevention efforts for the population at risk in their communities.