JMIR Publications

ABSTRACT

Background:

Approximately 30% of people living with HIV (PLHIV) in Brazil have suboptimal adherence to antiretroviral therapy (ART). Theoretical models of health behaviors and objective evidence support the role of perceived barriers in hindering health behaviors.

Objective:

We sought to evaluate the psychometric properties of the perceived barriers to antiretroviral therapy adherence (PEDIA) scale among gay, bisexual, and other men who have sex with men (MSM) and transgender and non-binary persons (TGNB) in Brazil.

Methods:

Using a cross-sectional study design, we evaluated the factor structure, construct validity, convergent and discriminant factorial validity, and internal reliability of the PEDIA scale using two convenience samples of participants ≥18 years of age living in Brazil. Sample 1 was recruited online between Feb/Mar 2020 and sample 2 between May/2021 and Jan/2022. The study instrument included sociodemographic information, ART use and three measures of ART adherence. Using sample 1, we performed exploratory factor analysis (EFA) with parallel analysis to evaluate PEDIA’s factor structure. Based on the goodness of fit measures and theoretical relevance, we developed a reduced version of PEDIA (PEDIAr). Using sample 2, we performed a psychometric evaluation of PEDIAr, which included confirmatory factor analysis (CFA), examination of modification indexes and item-item and item-total correlations, and an assessment of its correlation with different measures of adherence to antiretroviral therapy (ART). Goodness of fit was evaluated based on multiple indices.

Results:

EFA conducted on sample 1 (n=1,746) revealed a two-factor structure with three factor loadings <0.4 (excluded). Using sample 2 (n=4,893), modification indices from the CFA and item-item and item-total correlations along with item relevance analysis suggested the exclusion of five additional items. CFA fit indices for PEDIAr were adequate (RMSEA=0.071, CFI=0.954, TLI=0.939, SRMR=0.052). Construct validity was supported by factor loadings above 0.6 and negative correlations between PEDIAr scores and three measures of adherence to ART. McDonald's omega was 0.795 and 0.859 for factors 1 and 2, respectively. The square-root of the average variance extracted (AVE) was 0.704 and 0.711 for factors 1 and 2, respectively, and the difference between AVE and the square of the factor correlations with other items was small (0.001 and 0.009, respectively) and not statistically significant for both factors (p = 0.941 and p = 0.546, respectively).

Conclusions:

PEDIAr, the 10-item reduced version of PEDIA, proved to be valid among gay, bisexual, and other MSM and TGNB in Brazil. This shorter instrument was able to capture two distinguished dimensions of the perceived barriers to adherence to ART (practical aspects and psychological aspects). By proactively identifying individuals struggling with adherence to ART, PEDIAr can facilitate timely interventions and improve personalized care.