Development and validation of a measure for intention to adhere to HIV treatment

Abstract

We present the development and validation of a theory-derived scale measuring patients' behavioral intention to adhere to HIV care. Adherence to HIV care includes attendance at appointments and adherence to highly active antiretroviral therapy (HAART) regimens. These two components have been independently associated with long-term HIV outcomes. Items were chosen to reflect behavioral intention as defined by the Health Action Process Approach to health-seeking behavior. Items reflecting self-reported HIV knowledge were also included after expert panel review. The study took place from October 2009 to April 2010 at two HIV clinics in Houston, Texas. Participants were 287 adults with HIV/AIDS (10.1% female, mean age 50.8); 56.5% were African-American and 17% were Hispanic. Of the total, 87.1% were on HAART at enrollment. Factor analysis of survey items resulted in the retention of two domains, knowledge and intention, based on scree plot analysis of eigenvalues. Questions with factor loadings >0.4 were retained, yielding 4 knowledge questions and 10 intention questions. The survey had good internal consistency for knowledge (Cronbach's α=0.83) and for intention (Cronbach's α=0.81). In multivariate analysis, intention was associated with HIV viral suppression, defined as HIV-1 viral load <400 RNA copies/mL, (odds ratio [OR]=1.75, 95% .confidence interval [CI]=1.00-3.07). Knowledge was also associated with HIV suppression (OR=1.55, 95%, CI=1.09-2.12). The resulting study describes the development and preliminary validation of an HIV treatment-seeking intention measure. Additional studies are needed to validate this instrument in other populations.

Figures

FIG. 1.

Application of the health action process approach to adherence to HIV care.

FIG. 2.

A scree plot of eigenvalues after exploratory factor analysis. Factors are shown in order of decreasing eigenvalues. Factors 1 and 2, with eigenvalues >1.0 and appearing in the vertical portion of the graph, were retained and interpreted, whereas factors with eigenvalues