Incremental value of non-pharmacokinetic adherence measures for dapivirine vaginal ring use in resource limited setting Authors: Andrew Max Abaasa, Sylvia Kusemererwa Published: 2026 ABSTRACT AbstractBackgroundMeasuring product use adherence remains a challenge in HIV prevention trials and routine care setting. We determined agreement among and between non-pharmacokinetic (non-PK) and pharmacokinetic (PK) adherence measures, and also assessed the incremental value of each measure among women using the dapivirine vaginal ring (DVR) in the Ring trial in southwestern Uganda.MethodsBetween 2013 and 2016, the International Partnership for Microbicides conducted the Ring trial in which women were randomised to dapivirine or placebo vaginal ring in a ratio of 2:1. A total of 197 women were randomised. Four adherence measures were used, including PK: (i) testing of the dapivirine residual levels in used rings and (ii) plasma drug concentrations; non-PK: (iii) self-reported and (iv) ring colour change, all measured every 4 weeks for 2 years. We defined product use adherence as having the expected dapivirine residual levels (≤ 23.5 mg of 25 mg) or plasma concentration (≥ 95 pg/mL) or having used the ring as instructed at ≥ 80% of the visits completed. We used dapivirine residual levels as a gold standard and measured agreement and incremental value of other measures in predicting residual adherence using Cohen’s kappa statistic, a discriminant C-statistic and Receiver Operating Characteristic (ROC) curve, stratified by age < 25 or ≥ 25 years.ResultsOf the 197 women randomised, 66% were aged ≥ 25 years and in both age groups, > 80% completed 2 years of follow-up. In women aged < 25 years, percent agreement between non-PK measures and that between non-PK and PK measures was moderate 57.8% to 66.7%. While that between PK measures was high 86.7%. In the women aged ≥ 25 years, agreements between combinations of measures were 80.5% to 86.2%. The incremental value of non-PK measures was poor in women aged < 25 years, AROC = 0.62, and good in those age 25 + years, area under receiver operating curve (AROC) = 0.77.ConclusionCombining non-PK measures to measure dapivirine residual level adherence in women aged ≥ 25 years could be useful in DVR use, in the context of resource-scarce settings. However, this is of limited value in young women aged < 25 years.Trial registrationReg No. NCT01539226. Registered on February 24, 2012.Keywords: Dapivirine ring, Adherence measures, Residual plasma, Colour Background HIV infection continues to be a global challenge with sub-Saharan Africa (SSA) bearing the highest burden [1]. The infections continue to occur despite the availability of effective prevention strategies like voluntary male surgical circumcision, pre-exposure prophylaxis (PrEP), antiretroviral therapy for HIV-positive partners to prevent transmission and post-exposure prophylaxis to prevent the establishment of infection in an exposed HIV negative individual among many others [2]. The dapivirine vaginal ring (DVR) was recently added to the HIV prevention tools that women can use to prevent HIV infection [2, 3]. UNAIDS estimates indicate that in 2023 alone, 0.64 million people in SSA acquired HIV, accounting for 50% of the global total of 1.3 million new infections [1]. Women, particularly adolescents, remain disproportionately affected [1]. One major driver of persistent infections in SSA is suboptimal adherence to proven HIV prevention interventions [4]. Unlike in the treatment settings where surrogate measures like viral loads or CD4 counts are available, measuring adherence to HIV prevention interventions poses unique challenges. In PrEP trials, measures of adherence like self-report and plasma among others have been employed either singularly or in combination with varying degrees of agreement [5]. Non-PK measures such as self-report were generally found to over-estimate adherence hence poor correlation with PK measures in these trials [5]. The agreement among and between non-PK and PK measures in DVR use adherence have not been established. PK measures, while considered more accurate, they are expensive and need established laboratory infrastructure which may be limited in the resource scarce settings. Assessment of how non-PK and PK measures correlate would come in handy to select the best feasible options for such settings. Additionally, discrepancies between measures could inform counselling strategies to improve adherence. Residual drug levels in DVR, the gold standard for measuring adherence, could help identify inaccuracies if women removed and re-inserted the ring prior to visiting a health facility. The ring will not have shaded the expected amount of dapivirine in the specified period. This level of accuracy in DVR use may not be picked by plasma concentration which is a short-term measure and likely to be influenced by the drug moisture. Another challenge is associated with non-PK and PK measures not being aligned. Each measures adherence in different periods. Optimizing adherence measurement for DVR use would support its use post licensure in the routine care settings. We therefore assessed the Ring Study data at the MRC/UVRI and LSHTM Uganda research Unit site in Masaka district, Uganda to provide a comprehensive assessment of two of each non-PK and PK measures aligned over a month. We used both short-term (plasma drug concentration) and long-term (residual drug level) PK and non-PK (self-report and colour change) measures to assess concordance over a 28-day period and provide the incremental value of using a single or combination of measures to predict residual level (gold standard) adherence. Methods Recruitment and study procedures We retrospectively analysed data collected in the Ring Study (Reg No. NCT01539226) which evaluated safety and efficacy of dapivirine vaginal ring (DVR) against HIV infection in two countries, Uganda (one site) and South Africa (six sites) between 2013 and 2016 [6]. In this analysis, we used data collected at the Uganda site, i.e. the MRC/UVRI and LSHTM Uganda Research Unit in Masaka City, about 120KM Southwest of Kampala, the capital of Uganda. At this site, women at high risk of HIV infection residing in small townships along the trans-Africa highway (highway from North Africa through East Africa to Southern Africa) and on the shoreline of Lake Victoria were screened for eligibility and randomised to DVR or matching placebo ring arms. The details on recruitment [7] and follow-up [6] have been previously described. Briefly, women aged 18 to 45 years, HIV negative and at high risk of HIV infection defined as presence of any two of history of sexually transmitted infections, self-reported non-use of condoms when having sex with multiple sexual partners or a new partner and use of recreational drugs in the preceding 3 months were recruited. These women were largely identified from bars, small hotels/restaurants, hair salons and other social gathering spots. Women were randomised in a ratio of 2:1, i.e. DVR to placebo. At enrolment, data were collected on sociodemographic and behavioural characteristics, and women had physical examination completed. Randomised women were trained on how to self-insert a ring into a vagina and were provided with a self-inserted vaginal ring every 4 weeks for 104 weeks. They were tested for HIV infection monthly for 2 years. DVR adherence At each monthly clinic visit, a trial nurse took off a blood sample for plasma dapivirine drug concentration testing and completed an interviewer-administered questionnaire to gather information on ring removal and insertion and duration when the ring was removed. The returned ring was received and examined by a clinician for colour change and later for residual levels. DVR adherence in this trial was defined in four measures; (a) residual-level adherence—a used ring having ≤ 23.5 mg of 25 mg (if consistently in a vagina over 28-day period) dapivirine residual levels, (b) plasma adherence—plasma dapivirine drug concentration of ≥ 95 pg/mL, (c) ring colour change adherence—a used ring turning yellowish-brownish and self-reported adherence—a woman reporting inserting a ring and leaving it in the vagina as instructed. For the purposes of this analysis, we defined residual-level and plasma-ring adherence as having the expected residual, and plasma dapivirine concentration levels at 80% of the visits completed; self-reported adherence—a woman reporting a ring being consistently in the vagina as instructed at 80% of the visits completed; and ring colour change adherence—returning a yellowish-brownish ring at 80% of the visits completed. Statistical methods Statistical analyses were performed in Stata version 18.0 (Stata Corp, College Station, TX, USA). We limited our analysis to the DVR arm and summarised women baseline characteristics by counts and percentages stratified by age group (18–24 years and ≥ 25 years). We considered DVR residual level and plasma dapivirine drug concentration adherence measures as pharmacokinetics (PK) and DVR colour change and self-report as non-pharmacokinetic (non-PK). Adherence by each measure was defined as having expected dapivirine residual levels or plasma concentration or having used the ring as instructed at ≥ 80% of the visits completed. Further details are provided in the methods above. The cutoff of 80% was arbitrary. However, this has been associated with reduced risk of HIV infection in PrEP studies [8, 9]. For incremental value analysis, we considered DVR residual level adherence as the gold standard and determined how best other measures both PK and non-PK predicted this measure. We estimated categorical concordance when two measures were matched (i.e. adherent to adherent or non-adherent to non-adherent). We assessed the incremental value of multiple adherence measures when used in a combination by comparing distinct categories of DVR residual level adherence with other measures. We used a discriminant (C-statistic) analysis for a single or combined measure(s) using logistic regression models and displayed by receiver operating characteristic (ROC) curve area. Results Screening and enrolment The trial screened 325 women and randomised 197 (60.6%). The primary reasons for screen failure included investigator decision (n = 37), genital infections (n = 25), and being HIV positive or having a discordant result (n = 14) among other reasons (Fig. 1). As shown in Fig. 1, over 80% of the women randomised in the DVR arm completed adherence assessments at each visit in the 2-year period. We excluded the DVR placebo arm because residual or plasma testing was not applicable to this arm. In the women aged 18–24 years; most, 57.8% had completed primary education, nearly half 53.3% were single and never married, majority had a main sexual partner 77.8%, were presently living with this partner 71.4% and had had unprotected sex in the last 3 months 91.1% (Table 1). Similarly, for women aged ≥ 25 years; half 50.6% were married or cohabiting, majority 85.1% had a main sexual partner, 55.4% were presently living with this partner and 73.6% had had unprotected sex in the previous 3 months. Trial profile of women screened and randomised into the Ring Study in Masaka Southwestern Uganda (Sept 2013–Dec 2014), N = 197 Table 1Baseline sociodemographic characteristics of 132 women randomised into the Ring Study in Southwestern Uganda (Sept 2013–Dec 2014)VariableAged (18–24) yearsAged (25+) yearsCount (n = 45)Percent (%)Count (n = 87)Percent (%)Highest education Incomplete primary1022.23843.7 Complete primary2657.83944.8 Secondary + 920.01011.5Marital status Single never married2453.33337.9 Single ever married*817.81011.5 Married/cohabiting1328.94450.6Has main partner Yes3577.87485.1 No1022.21314.9Presently lives with main partner Yes2571.44155.4 No1028.63344.6 Time (years) lived with main partner, median, IQR21–462–10Unprotected sex, last 3 months Yes4191.16473.6 No48.92326.4Drug/alcohol use, last 3 months Yes3884.46675.9 No715.62124.1IQR interquartile range*Divorced, widowed, separated Baseline sociodemographic characteristics of 132 women randomised into the Ring Study in Southwestern Uganda (Sept 2013–Dec 2014) IQR interquartile range *Divorced, widowed, separated Adherence by each measure In the women aged 18–24 years, the adherence over the 2-year period was self-report 60% (95% CI: 44–74%), colour change 78% (63–89%), plasma concentration 62% (47–76%), and residual 49% (34–64%). Similarly, in those aged ≥ 25 years, self-report 86% (77–93%), colour change 86% (77–93%), plasma 76% (65–84%) and residual 76% (65–84%). Concordance between adherence measures As shown in Table 2, in women aged 18–24 years, agreement between non-PK measures and that between non-PK and PK measures was moderate with a percent concordance of 57.8% to 66.7%. While that between PK measures was high 86.7%. Agreement between adherence measures in women aged ≥ 25 years was generally high with a percent concordance of 80.5% to 86.2% for both non-PK and PK measures. Table 2Concordance between PK and non-PK, ring adherence measures among women in the Ring Study in Uganda, (Sept 2013–Dec 2014), N = 132, age 25 + (bold, n = 87) and 18–24 (italics, n = 45)MethodAgreementResidual n (%)Plasma n (%)Colour-change n (%)Self-report n (%)ResidualConcordant1.039 (86.7)30 (66.7)26 (57.8)Discordant (high)na6 (13.3)4 (8.9)10 (22.2)Discordant (low)na0 (0.0)11 (24.4)9 (20.0)PlasmaConcordant75 (86.2)1.026 (57.8)26 (57.8)Discordant (high)6 (6.9)na3 (6.6)7 (15.5)Discordant (low)6 (6.9)na16 (35.6)12 (26.7)Colour-changeConcordant72 (82.7)70 (80.5)1.029 (64.4)Discordant (high)12 (13.8)13 (14.9)na12 (26.7)Discordant (low)3 (3.5)4 (4.6)na4 (8.9)Self-reportConcordant70 (80.5)66 (75.9)73 (83.9)1.0Discordant (high)13 (14.9)15 (17.2)7 (8.1)naDiscordant (low)4 (4.6)6 (6.9)7 (8.1)naPK pharmacokinetic, DVR dapivirine ring, na not applicable Concordance between PK and non-PK, ring adherence measures among women in the Ring Study in Uganda, (Sept 2013–Dec 2014), N = 132, age 25 + (bold, n = 87) and 18–24 (italics, n = 45) PK pharmacokinetic, DVR dapivirine ring, na not applicable Incremental value of two or more measures in combination vs residual level adherence Figures 2 and 3 show the ability of each adherence measure singularly or in combination of measures to discriminate among adherence or non-adherence for residual (gold standard) measure. In the women aged 18–24 years, the discrimination was poor for the non-PK measures individually (self-report, area under receiver operating curve (AROC) = 0.56, 95% CI (confidence interval) 0.40–0.71), colour change AROC = 0.60 (95% CI 0.47–0.74). While this was high for PK measure alone (plasma, AROC = 0.89, 95% CI (0.82–0.97)). When combining two non-PK measures, discrimination did not change much, AROC = 0.62. Also when combining the two non-PK with plasma, discrimination was not any different from that offered by plasma alone, AROC = 0.9. On the other hand, in the women aged ≥ 25 years, the discrimination was moderate for non-PK measures, individually (self-report 0.66 and colour change 0.69), while this was high for plasma alone 0.81. When combining two or more measures; (a) non-PK with non-PK, (b) non-PK with PK and (c) all measures, the discrimination improved with AROC of ≥ 0.77 for all combinations. Discrimination of ring residual adherence measure based on a single alternate measure or combinations of measures among 18–24-year-olds in the Ring Study, Uganda Discrimination of ring residual adherence measure based on a single alternate measure or combinations of measures among 25+-year-olds in the Ring Study, Uganda Discussion In the Ring Study of women using monthly vaginal DVR, we determined the incremental value of non-pharmacokinetic (non-PK) measures of adherence to predict a pharmacokinetic (PK) adherence measure in a resource scarce setting. The results showed evidence of high adherence for each PK and non-PK measure in the women aged at least 25 years and moderate to poor adherence in those below 25 years. This is consistent with other studies on oral PrEP, and other HIV prevention tools [10, 11]. Young women have challenges associated with poor understanding of product use, information vacuum, negative stigma and stereotypes reinforced by cultural norms portrayed on social media, television and film [12]. The results of concordance analysis were moderate when the comparison was among the non-PK or between non-PK and PK measures in the women aged < 25 years. On the contrary, either combination provided high concordance in the women aged at least 25 years. Taking residual-level adherence as a gold standard, the concordance in adherence was very high in the women aged at least 25 years, i.e. about four in every five assessments for both non-PK and PK measures. This was high for PK and reduced to about three in every five assessments for non-PK measures among women aged below 25 years. Social desirability among young people has been cited by multiple prior studies as a key contributor to inaccuracies in self-reported adherence [13, 14]. This may not have been possible for PK measures. On the contrary, older women are likely to follow instructions as provided. A receiver operating characteristic (ROC) analysis was used to estimate the incremental value of multiple adherence measures used either singularly or in combinations. This was aimed at discriminating among good (≥ 80%) and poor (< 80%) residual testing adherence. Under this analysis, self-report singularly performed poorly (AROC = 0.56) in women aged < 25 years and fairly (AROC = 0.66) in those aged at least 25 years. Slightly better discrimination was observed for colour change measure in both age groups. This is similar to the Partners PrEP study in which self-report discrimination of 0.52 was observed. The results show that monitoring DVR in plasma provided the best discriminant ability (AROC > 0.8) for both age categories. This finding suggests that plasma concentration could best predict ring adherence over a month period as captured by residual testing for all ages. When measures were used in combinations, for women aged < 25 years, non-PK measures alone did not improve discrimination ability, though this improved to AROC = 0.77 in the women aged ≥ 25 years. Hence, in the older age category, the combination of non-PK measures explained nearly as much variability as a combination of non-PK and a PK measure. In resource scarce settings such as ours, these non-PK measures could provide a relatively comprehensive image of monthly ring adherence in older women. The strengths of this analysis include having simple adherence measures such as ring colour change that could be applied even in the most remote setting. This was coupled with self-report and PK measures for easy comparison. There was good retention in the trial and multiple adherence data collection time points allowed for comprehensive analysis. Even when the trial was not primarily powered to do this kind of comparison, the data from this trial exhibit a high level of ability of non-PK measures predicting ring adherence in older-age women. Our analysis is not without limitations. The colour change observation was done by a single study clinician, therefore, this was dependent on the observer. There could have been variations in the conclusions about observations for the same ring colour had two or more trial clinicians assessed the same ring for colour change. Lastly but not least, the ring adherence in the placebo-controlled setting could be different from the routine care setting. Therefore, the results of this analysis need to be interpreted with caution when applied to real-world setting since women will be using a vaginal ring already known to prevent against HIV acquisition. Conclusion In conclusion, our results indicate that PK measures of DVR adherence are more accurate in predicting residual level testing adherence over a month and for all age categories. The differences in the performance of the non-PK measures in the different age categories (< 25 or ≥ 25 years), must be considered carefully when using these measures in future trials and/or routine settings. In resource scarce settings where residual level and/or plasma drug concentration testing are not feasible, non-PK measures examined in this trial could provide a good estimate of DVR use over a month for women aged at least 25 years. Acknowledgements The International Partnership for Microbicides’ (IPM)—a not-for-profit product-development partnership work is made possible by generous support from many donors including: Irish Aid, the Ministry of Foreign Affairs of Denmark, the Ministry of Foreign Affairs of the Netherlands, the Norwegian Agency for Development Cooperation, the United Kingdom Department for International Development, the Bill and Melinda Gates Foundation, the American people through the United States Agency for International Development, and the President’s Emergency Plan for AIDS Relief. This research is jointly funded by the UK Medical Research Council (MRC) and the Foreign Commonwealth and Development Office (FCDO) under the MRC/FCDO Concordat agreement. We wish to thank the study team who worked tirelessly to plan, recruit and collect data for this trial; Vincent Basajja, Henry Ssemaganda, Victoria Mugwaneza, Sarah Nakato, Beatrice Kimono Washi, Sylvia Masawi, Penelope Akankunda, Aeron Namirembe, Wilson Kakeeto, Paul Taire, Joseph Kitumba, and Elizabeth Mbabazi Atuhura. Abbreviations Acquired immunodeficiency syndrome Area under receiver operating curve Dapivirine ring Confidence interval London School of Hygiene and Tropical Medicine Medical Research Council Pharmacokinetic Pre-exposure prophylaxis Texas United States of America Joint United Nations Programmes on HIV/AIDS Uganda Virus Research Institute Authors’ contributions AA, SK designed the analysis and AA completed the analysis and wrote the initial manuscript version. SK directed the work. All authors contributed to the interpretation of the results and critically commented and provided revisions to the manuscript. All authors approved the final version of the manuscript. Funding The trial was both funded and sponsored by IPM (www.ipmglobal.org) (IND # 110,659). The funder participated in the design of the main trial and reviewed this manuscript before submission but had no role in either data analysis or interpretation of the results. Data availability The MRC/UVRI and LSHTM Uganda Research Unit encourages data sharing and data-sharing policy and procedures are available https://apps.mrcuganda.org/mrcdatavisibility/Home/ The policy summarises the conditions under which data collected by the Unit can be made available to other bona fide researchers, the way in which such researchers can apply to have access to the data and how data will be made available if an application for data-sharing is approved. Should any of the other researchers need to have access to the data from which this manuscript was generated, the processes to access the data are well laid out in the policy. The corresponding author can be contacted for any clarifications and/or support to access the data. Declarations Ethics approval and consent to participate This trial involved human participants and the approval for its conduct was obtained from regulatory bodies of each participating country, Uganda: the Uganda Virus Research Institute Research Ethics Committee (Ref#-GC/127/13/03/33), the Uganda National Council of Science and Technology (Ref#- HS1362) and the National Drug Authority (Ref #-166/ESR/NDA/DID-07/2013). Before any trial procedures, written informed consent was obtained from each woman. Women that became HIV positive during the trial were referred to an HIV/AIDS care provider of their choice in the community for treatment and care. Consent for publication Not applicable. Competing interests All authors declare that they have no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The MRC/UVRI and LSHTM Uganda Research Unit encourages data sharing and data-sharing policy and procedures are available https://apps.mrcuganda.org/mrcdatavisibility/Home/ The policy summarises the conditions under which data collected by the Unit can be made available to other bona fide researchers, the way in which such researchers can apply to have access to the data and how data will be made available if an application for data-sharing is approved. Should any of the other researchers need to have access to the data from which this manuscript was generated, the processes to access the data are well laid out in the policy. The corresponding author can be contacted for any clarifications and/or support to access the data.