CFE 2019: Start Registration
View Submission - CMStatistics
Title: Regular HIV testing: Why it matters and how to measure it Authors:  John Rice - University of Colorado, Denver (United States) [presenting]
Brent Johnson - University of Rochester (United States)
Robert Strawderman - University of Rochester (United States)
Abstract: Screening for infectious diseases such as HIV is an important public health priority, but traditionally only the rate of screening has received attention. We argue that it is equally important to examine the regularity of screening. Modeling testing events as a renewal process, we show that the mean delay in diagnosis increases quadratically with the coefficient of variation (CV) of the intertest times, a quantity that is inversely proportional to regularity. This result implies that greater regularity in screening is crucial from a public health perspective because delays in diagnosis may lead to worse prognoses and greater likelihood of unaware patients infecting others. In order to estimate this delay in a population of at-risk individuals, we fit regression models for the mean and CV of the intertest time distribution to data in which study participants were asked to recall the time of their most recent HIV test within pre-specified intervals. Finally, we present power analysis of a likelihood ratio test comparing the mean and CV between two groups, concluding that while efficiency is lost with discretization of backward recurrence times, an optimal set of intervals may be chosen such that this loss is minimized.