Improving HIV clinical trials with mobile technology

HIV prevention and treatment strategies are routinely studied through enrolling at-risk participants in clinical trials. During these studies, self-reported data regarding sexual practices can be affected by recall or misclassification bias, as well as social desirability bias, where participants feel pressure to answer a certain way. Social desirability bias is strongest in face-to-face interviews with study coordinators, so researchers have implemented computer-assisted self-interviewing which has slightly increased the accuracy of self-reported data. Furthermore, public health and research initiatives have begun to introduce cell phone applications to disperse and collect information, however, the advantage of cell phone- versus computer-assisted interviewing in decreasing reporting bias has not been validated with biomarkers. Additionally, clinical trials currently do not have a reliable method to detect vaginal HIV exposure, a metric that would improve the accuracy in identification of transmission and enable screening of those at high risk for HIV infection. To address these problems in the HIV field, Dr. Maria Lemos, along with colleagues from the HIV Vaccine Trials Network and sought to test the effectiveness of mobile phone surveys in collecting self-reported sexual practice data as well as validate the use of vaginal swabs as a method of HIV virion detection, and published this recent work in JAIDS.

The authors enrolled in their study 50 HIV-negative South African women who were frequently sexually active or engaged in transactional and unprotected sex. These participants were asked to perform daily self-administered vaginal swabbing and to complete brief behavior questionnaires via mobile phone application, reporting data such as time and number of sex acts, whether condoms were used, and time of swabbing. The eluted swabs were then used for polymerase chain reaction (PCR) to detect the presence of HIV DNA. Additionally, PCR was used to detect Y-chromosome DNA (Yc-DNA), indicating that condoms were not used, and glycogen, a biomarker for vaginal secretions, which confirms whether vaginal swabbing was performed. These metrics were then used to validate the self-reported data collected via phone. 

At the conclusion of the study, PCR revealed that mobile applications accurately reflected swabbing adherence, as almost 94% percent of self-collected swabs contained glycogen, indicating that the swabs had actually reached the vaginal vault. This result suggested that not only were participants mostly honest about their swabbing when using a phone to report, but that they were able to accurately perform the swabbing without clinician oversight. Likewise, an absence of glycogen allowed the researchers to exclude the samples that had not actually entered the vagina, eliminating potential false-negative HIV exposure readings. Swabbing adherence was further confirmed with a measure of human DNA, which yielded similar results and corroborated glycogen PCR detection as a valid measure of adherence. Similarly, the authors found that condom use reported by patients via phone was highly accurate, as confirmed by the absence of Yc-DNA in 94.7% of swabs following reportedly protected sex. To determine exposure to HIV, the swabs that had been collected following unprotected sex were tested for the presence of HIV DNA, but only two swabs tested positive for HIV virus, even among samples where semen was present. HIV statistics for this region predict that the swabs would identify a higher frequency of HIV exposure, but the authors speculate that the low HIV detection could be due non-technical factors such as decreased HIV rates since the latest prevalence data or a reduction in HIV risk behavior during the study. However, they also suggest that HIV exposures that occurred more than four hours before swabbing could have been missed due to virus RNA instability.

Together, these findings demonstrate that “daily self-swabbing is feasible, well tolerated, and provides sufficient vaginal sample for assessment of condomless sex,” Lemos said. “This research also showed that mobile phone-based questionnaires are more accurate than those at the clinic, correlating best with biomarkers of condomless activity,” she added. The authors speculate that mobile phone reporting may outcompete other computer-assisted self-reporting methods simply because the participants can input their responses immediately after sexual activity and are not required to recall days at a time and are therefore better able to remember important details. Importantly, this research also demonstrated that vaginal swabs self-administered immediately after sex can be used to detect HIV exposures, although the limited window for HIV detection poses further challenges when considering swabbing as a tool for HIV prevention. These promising results raise further questions about how these techniques can be implemented, and the authors are further “extending these findings to men,” Lemos said. “In South Africa, Dr. Janan Dietrich is testing modifications to the mobile questionnaires to make them more user- friendly and applicable to men. In the US, I am studying (in partnership with Dr. Hong Van Tieu) whether mobile-based sexual reporting and rectal swabbing can be useful to measure biomarkers of condomless sex and HIV exposures in men who have sex with men in NYC.”

This work was supported by the National Institute of Allergy and Infectious Diseases, the South African National Research Foundation, and the HIV Initiatives Program.

UW/Fred Hutch Cancer Consortium members Julie McElrath and James Kublin contributed to this work.

Lemos MP, Lazarus E, Isaacs A, Dietrich J, Morgan C, Huang Y, Grove D, Andrasik M, Laher F, Hural J, Chung E, Dragavon J, Puren A, Gulati RK, Coombs R, McElrath MJ, Gray G, Kublin JG. 2019. Daily vaginal swabs and mobile phone report for assessing HIV virion exposure prospectively among a cohort of young sexually active women in South Africa (HVTN 915). JAIDS. DOI:  10.1097/QAI.0000000000002015.