Evaluation of 3 approaches for assessing adherence to vaginal gel application in clinical trials

Abstract

BACKGROUND: Accurate measurement of adherence to product use is an ongoing challenge in microbicide trials. METHODS: We compared adherence estimates using 2 applicator tests (a dye stain assay [DSA] and an ultraviolet light assay [UVA]), the Wisebag (an applicator container that electronically tracks container openings), and self-reported adherence (ability, frequency, and percent missed doses). Healthy, HIV-negative, nonpregnant US women aged 23 to 45 years received a Wisebag and 32 applicators filled with placebo gel were instructed to insert 1 applicator daily for 30 days, returned the Wisebag and all applicators, and completed an exit interview. Emptied applicators were tested by UVA and then DSA, and scored by 2 blinded readers. Positive and negative controls were randomly included in applicator batches. RESULTS: Among 42 women enrolled, 39 completed the study. Both DSA and UVA yielded similar sensitivity (97% and 95%) and specificity (79% and 79%). Two participants had fully inoperable Wisebags, and 9 had partially inoperable Wisebags. The proportion of participants considered to have high adherence (>/=80%) varied: 43% (Wisebag), 46% (UVA), 49% (DSA), and 62% to 82% (self-reports). For estimating high adherence, Wisebag had a sensitivity of 76% (95% confidence interval, 50%-93%) and a specificity of 85% (95% confidence interval, 62%-97%) compared with DSA. Although 28% of participants reported forgetting to open the Wisebag daily, 59% said that it helped them remember gel use. CONCLUSIONS: Dye stain assay and UVA performed similarly. Compared with these tests, self-reports overestimated and Wisebag underestimated adherence. Although Wisebag may encourage gel use, the applicator tests currently seem more useful for measuring use in clinical trials