Comparison of short messaging service self-reported adherence with other adherence measures in a demonstration project of HIV preexposure prophylaxis in Kenya and Uganda

    loading  Checking for direct PDF access through Ovid


Objective:Short messaging service (SMS) can collect adherence data on a frequent basis and is relatively anonymous, and therefore could potentially reduce recall and social desirability biases prevalent in other self-reported measures.Methods:We compared SMS self-reported adherence with three self-reported adherence questions (rating of ability to adhere, frequency of doses taken, percentage of doses taken) and two objective adherence measures [electronic adherence monitoring (EAM) and plasma tenofovir levels] using data from HIV-uninfected members of serodiscordant couples enrolled in a preexposure prophylaxis demonstration project in Kenya and Uganda.Results:Of 373 enrolled participants, 256 (69%) were male and median age at enrolment was 29 years (26, 35). Fifty-two percent were from Kenya and median education at enrolment was 10 years (7,12). Overall, median adherence was 90, 75, 85, 94 and 79%, respectively, for self-report by SMS, rating, frequency, percentage and EAM adherence. Spearman's correlation coefficient between SMS and interviewer-administered self-reported measures was 0.18 for rating and frequency, 0.22 for percentage and 0.14 for EAM (all P < 0.001). The estimated difference in average adherence between SMS and self-reported rating, frequency, percentage adherence and EAM was 8.1 (P < 0.001), 0.3 (P = 0.81), −5.2 (P < 0.001) and 9.5 (P < 0.001), respectively. Area under the receiver-operating curve assessing the ability of SMS self-report to discriminate between detectable and undetectable tenofovir was 0.51.Conclusion:Our study found low correlation between SMS self-report and other self-reported and objective adherence measures and did not discriminate between detectable and undetectable plasma tenofovir levels. Future use of SMS self-report should explore alternative means for reducing potential biases.

    loading  Loading Related Articles