Modeling the Declining Positivity Rates for Human Immunodeficiency Virus Testing in New York State

摘要

Objective: New York health care providers have experienced declining percentages of positive human immunodeficiency virus (HIV) tests among patients. Furthermore, observed positivity rates are lower than expected on the basis of the national estimate that one-fifth of HIV-infected residents are unaware of their infection. We used mathematical modeling to evaluate whether this decline could be a result of declining numbers of HIV-infected persons who are unaware of their infection, a measure that is impossible to measure directly. Design and Setting: A stock-and-flow mathematical model of HIV incidence, testing, and diagnosis was developed. The model includes stocks for uninfected, infected and unaware (in 4 disease stages), and diagnosed individuals. Inputs came from published literature and time series (2006-2009) for estimated new infections, newly diagnosed HIV cases, living diagnosed cases, mortality, and diagnosis rates in New York. Main Outcome Measures: Primary model outcomes were the percentage of HIV-infected persons unaware of their infection and the percentage of HIV tests with a positive result (HIV positivity rate). Results: In the base case, the estimated percentage of unaware HIV-infected persons declined from 14.2% in 2006 (range, 11.9%-16.5%) to 11.8% in 2010 (range, 9.9%-13.1%). The HIV positivity rate, assuming testing occurred independent of risk, was 0.12% in 2006 (range, 0.11%-0.15%) and 0.11% in 2010 (range, 0.10%-0.13%). The observed HIV positivity rate was more than 4 times the expected positivity rate based on the model. Conclusions: HIV test positivity is a readily available indicator, but it cannot distinguish causes of underlying changes. Findings suggest that the percentage of unaware HIV-infected New Yorkers is lower than the national estimate and that the observed HIV test positivity rate is greater than expected if infected and uninfected individuals tested at the same rate, indicating that testing efforts are appropriately targeting undiagnosed cases.