Resumen de Prospective State and Trait Predictors of Daily Medication Adherence Behavior in HIV

Paul F. Cook, Sarah J. Schmiege, Whitney Starr, Jane M. Carrington, Lucy Bradley Springer

• Background Many persons living with HIV (PLWH) are nonadherent to medication. Trait level measures that ask about predictors of adherence in the abstract may not adequately capture state level daily variability that more directly impacts adherence.

  Objectives This preliminary study was designed to test six predictors of electronically monitored adherence at both the state and trait levels and to compare their relative effects.

  Methods Using a smartphone, 87 PLWH completed randomly cued daily surveys on thoughts, mood, stress, coping, social support, and treatment motivation. All participants also completed baseline surveys on each construct. These state and trait variables were tested as prospective predictors of next-day adherence in multilevel models, and their relative importance was quantified. The analysis sample consisted of 53 PLWH who stored their most frequent antiretroviral medication in a bottle that time-stamped openings to measure adherence.

  Results Higher state level motivation, OR = 1.55, 95% CI [1.07, 2.24], and negative mood, OR = 1.33, 95% CI [1.07, 1.63], predicted greater adherence the following day. Importantly, these effects were only found at the state level. Trait level control beliefs predicted greater adherence, OR = 1.65, 95% CI [1.17, 2.35], but contrary to prediction, validated trait level measures of mood, stress, coping, social support, and motivation did not.

  Discussion Trait and state level measures predicted adherence, but there were differences between them. Motivation for treatment and negative mood predicted adherence when measured the preceding day, but not as aggregate measures. At the trait level, only control beliefs predicted adherence. Researchers should consider state level variations in mood and motivation as possible explanations for nonadherence. Interventions could be developed to target state level variables.