Effectiveness of Clinic-Based Patient-Led Human Papillomavirus DNA Self-Sampling among HIV-Infected Women in Uganda

Abstract

In Uganda, the uptake of cervical cancer (CC) screening services is low, at 46.7%, among HIV-infected women, and only 9% of these women adhere to annual CC screening. Some studies have evaluated the possibility of community or home-based human papillomavirus (HPV) self-collected vaginal swabs, but not clinic-based HPV self-collected vaginal swabs. Therefore, we propose a study to determine the efficacy of clinic-based versus home-based HPV DNA self-sampling among HIV-infected women attending a rural HIV clinic in Uganda. We believe that a randomized, single-blinded trial would achieve this objective, and so we have chosen it to guide the study. Including a total of 382 participants from a rural HIV clinic, randomized into a ratio of 1:1 for clinic- and home-based HPV self-sampling, would allow us to appropriately ascertain the difference in the uptake of HPV self-sampling between the two arms. The Integrated Biorepository of H3 Africa Uganda Laboratory would be used as a reference laboratory for the HPV DNA extraction, typing, and sequencing. At baseline, modified Poisson regression models would be used to measure factors associated with the prevalence of HPV and uptake in both arms at baseline. Visual inspection under acetic acid (VIA), as a gold-standard test for CC to grade for CIN, would be performed at 0 and 6 months among a random sample of 75 women with a self-collected HPV sample. The difference in uptake could be determined using the intention-to-treat analysis. The difference in the groups by each variable would be summarized as the standardized mean difference (i.e., the mean difference divided by the pooled standard deviation). The predictors of the time for which participants would continue with HPV self-sampling in both arms, recovery, and Cox proportional hazards regression would be used. At the bivariate level, the associations between each independent variable and time, with the time of continuing HPV self-sampling, would be computed. Crude hazard ratios and their 95% confidence interval would be used in the presentation of the results, with p-values < 0.05 considered significant at the bivariate level. Incremental cost-effectiveness analysis (CEA) using a Markov model would be used to determine the cost of clinic-based HPV self-sampling. We believe that screening approaches to disease stratification could provide an insight into the merits and limitations of current approaches to the diagnosis of cervical cancer, and how these could eventually be implemented into HIV clinics in Uganda and other developing African countries. It is anticipated that the findings would guide the development of step-by-step guidelines for the HPV self-sampling approach.