Concurrent, predictive validity and reliability of the algorithm in the stages of change for rapid HIV testing

Abstract

Objective. To evaluate the concurrent, predictive validity and reliability of an algorithm designed to determine the stage of change for HIV testing. Methods. A cross-sectional, correlational, two-stage study was conducted. The first, with 300 young university students, to evaluate the concurrent and predictive criterion validity with variables such as self-efficacy, knowledge about HIV testing and risk perception, measured with instruments that present Cronbach's a between ,73 and ,93. The second in 75 migrants, considered as at-risk population (n = 75), to assess reliability by means of a test-retest. Results. The predominant stages of change for HIV testing in both samples were precontemplation and contemplation. The stages of change correlated with two of the concurrent variables of self-efficacy and knowledge (p < ,01). The regression model was significant (F [3,296] = 53.68, p < ,001), predicting 35 % of the variance in stages of change. A significant intraclass correlation was found in the test-retest (ICC=,97, CI [,952-,981], p < ,01). Conclusions. The stage-of-change algorithm for HIV testing presents acceptable psychometric characteristics of concurrent and predictive validity, as well as high temporal stability of the measure, so its use in practice is recommended.