Low Disclosure of PrEP nonadherence and HIV risk behaviors associated with poor HIV PrEP adherence in the HPTN 067/ADAPT Study

Abstract

Objective.

We evaluated the relationship between two types of social relationships (i.e., 1) external support for use of HIV pre-exposure prophylaxis [PrEP] and related study supplies 2) participants’ disclosure of PrEP use and condom use and HIV PrEP adherence among daily-dosing regimen participants in HPTN 067, an open-label trial of oral tenofovir (TFV) disoproxil fumarate (TDF) 300 mg/emtricitabine (FTC) 200 mg.

Methods.

Using HPTN-067 survey data, we developed scales examining 1) Low Perceived External Support for PrEP: low perceived support by others for PrEP use or perceived negative reactions to the pill case (scoring ranges from 0 to 2) and 2) Participant-Staff Disclosure Challenges Scale which identifies challenges to sharing non-use of PrEP or condoms to study staff (scoring ranges from 0 to 4); these scales are the primary independent variables. Adherence, the dependent variable, was determined using log-transformed plasma TFV concentrations. GEE linear regression was used to assess the association between both scales and adherence.

Results.

Participants (n=161) included HIV-uninfected women in South Africa, and men who have sex with men and transgender women, in Thailand and the U.S. In multivariable analyses, higher scores in the Participant-Staff Disclosure Challenges Scale were significantly associated with lower PrEP adherence (exp(β) = 0.62, 95% CI (0.46- 0.84); p=0.002) as were increased days since the last PrEP dose (exp(β) = 0.73, 95% CI (0.65- 0.83); p≤0.001).

Conclusions.

Given the association with adherence, study staff-participant interactions and participants’ disclosure of PrEP challenges may be worthwhile intervention targets for improving PrEP adherence in confirmatory studies.

INTRODUCTION

Daily oral HIV pre-exposure prophylaxis (PrEP; tenofovir (TFV) disoproxil fumarate (TDF) 300 mg/emtricitabine (FTC) 200 mg formulation) is effective in preventing HIV infection in healthy, HIV-uninfected individuals ¹. However, recent studies have demonstrated that achieving high levels of adherence to daily PrEP use can be challenging, including in clinical trial settings ^2,3. Identifying barriers to PrEP adherence remains an important research topic: data may inform and support the implementation of PrEP and its uptake across diverse settings.

Low adherence to HIV antiretroviral medications, including PrEP, has been associated with individual, community, provider, or health-care system level variables as well as product characteristics ^2,4. In the context of biomedical, placebo-controlled, blinded research trials, non-adherence to study product was common (only <40% of women in the FEM-PrEP trial ⁵ and 30% of female participants in the VOICE ⁶ trial were adherent) and caused the early discontinuation of both studies. Concerns regarding the characteristics and side effects of the experimental drug and low support by others for PrEP use, including partner’s/family members’ concerns regarding the drug characteristics and HIV-stigma, were important factors in shaping PrEP use ^7–9 in several clinical trials.

Qualitative research can elucidate the context underlying participants’ PrEP use. A qualitative sub-study conducted with South African women who participated in the HIV Prevention Trials Network (HPTN) 067 Alternative Dosing to Augment Pre-Exposure Prophylaxis Pill Taking (ADAPT) study, examined barriers and facilitators to PrEP adherence². Data derived from qualitative interviews informed the development of the Mutuality Framework (MF); it posits that four domains explain individuals’ approaches to PrEP use (and non-use): (1) distrust of the intervention and/or places and people prescribing it — resulting in rejection and avoidance of PrEP; (2) uncertainty regarding the intervention and places/people providing it, thus producing intermittent PrEP use; (3) alignment with the idea and potential utility of PrEP, leading to acceptance of study/product and potential benefit to self/community and uptake and use of PrEP; and (4) mutuality — wherein the participant believes in PrEP and its utility for self and community (i.e., altruism) leading to consistent use and good adherence. The Mutuality Framework synthesizes many of the aforementioned challenges to PrEP adherence into a comprehensive framework that highlights the importance of relationships between those accessing PrEP and those providing it. Focusing on the relationship between the participant and study staff may also serve as a proxy for and enhance our understanding of the relationship between patients and providers in community settings.

The relationship between patients and study staff, is under-represented in research examining PrEP uptake and use. To examine the potential influence of how PrEP study participants interact with staff members associated with providing PrEP, we conducted a post-hoc analysis using questionnaire and biomedical data derived from the HPTN 067 study. Specifically, we examined participants’ perceptions of reactions by others to PrEP pills and containers and participants’ experiences or perceptions of disclosure of PrEP use and sex without a condom to study staff. The outcome of interest was adherence to PrEP, per TFV plasma concentrations. Consistent with the characterization of the Uncertainty and Rejection domains in the Mutuality Framework², we hypothesized that participants who reported greater challenges to disclosing non-adherence to PrEP or sex without a condom would be less adherent to the study-provided PrEP, as estimated through TFV plasma concentrations. Potentially, reducing disclosure barriers may be associated with greater trust and improved both self-reported and objectively measured adherence.

METHODS

Study Design

The HPTN 067 Alternative Dosing to Augment Pre-Exposure Prophylaxis Pill Taking (ADAPT) study methods have been described elsewhere ^10–12. Briefly, the HPTN 067/ADAPT study from which data for this analysis was derived was a phase II, randomized, open-label, multi-site pharmacokinetic study of daily versus non-daily oral FTC/TDF PrEP ¹⁰. It sought to compare non-daily vs. daily use of PrEP in diverse participants and adherence-related outcomes. The study was conducted between July, 2012 and May, 2014 and approved by ethics committees of the Thailand Ministry of Health, the Centers for Disease Control and Prevention, Columbia University Medical Center, and the Ethics Committee of the Health Science Faculty, University of Cape Town; and the Emavundleni Community Advisory Board and the Medicines Control Council of South Africa. The role of Gilead Sciences in the development of the protocol was restricted to the handling of study drugs

Participants

Participants were ≥18 years old and included cisgender HIV-uninfected women at a study site in Cape Town, South Africa, HIV-uninfected men who have sex with men (MSM), and transgender women (TGW) who have sex with men, recruited at clinical research sites in Bangkok, Thailand, and Harlem in New York City, USA. All participants exhibited normal renal function, were literate in the local language, and provided informed consent. Visits that occurred during a product hold or following seroconversion were removed from this analysis as plasma TFV concentrations were expected to be below detection limits at these visits. While the study included 540 participants, we restricted this analysis to participants randomized to the daily-dosing arm, since quantitative plasma TFV benchmarks have been established for this dosing regimen¹³.

Study Visits & Medication Dispensation

Study visits included a self-administered dosing period. Participants in the daily dosing arm were advised to take one tablet daily, regardless of sexual activity. Participants received 30 tablets of FTC/TDF every 4 weeks.

Laboratory Procedures

Plasma concentrations of TFV were determined via a validated liquid chromatographic-tandem mass spectrometric (LC-MS/MS) method¹³. The lower limit of quantification for TFV is 0.31 ng/mL. This analysis relies on plasma data that were obtained at weeks 18 and 30.

Survey Data

We analyzed Computer Assisted Self-Interview (CASI) data, and data obtained from research staff collected at baseline and weeks 18 and 30. Demographic factors assessed at baseline included age and education; the study sites recruited participants based on biological sex and sexual orientation and therefore these data are not reported as they are reflected by the study site variable. CASI data included: (1) Perceived Problematic Drug Use: whether others perceive the participant to have a drug use problem (measured at baseline; refers to past 3 months; 1: yes vs. 0: no); (2) Perceived Problematic Drinking¹⁴: Relative, friend or doctor or health care worker is concerned about the participant’s drinking and suggests they cut down (measured at baseline; 0: never or yes but not in past 6 months, 1: yes during past 6 months); and (3) participant’s perceived control over HIV infection (1: “whether or not I get HIV is out of my control” or “I have some control over whether or not I get HIV” vs. 0: “I can prevent myself from getting HIV”). The following biological factors that may influence TFV concentrations in plasma were considered: creatinine clearance (measured in mL/min at weeks 18 & 30 per the Cockcroft-Gault equation¹⁵) and participants’ weight (measured in kilograms at weeks 18 & 30). We considered participants’ report of ever experiencing study-related side effects (coded: yes/no) and time-since-last-dose (a continuous variable measured in days; dosing events occurring ≥8 days were collapsed to “≥8 days”), both measured at weeks 18 & 30.

Independent Variables: Social Challenges to PrEP Adherence Scales

Using survey data, we used exploratory factor analysis to support the creation of two weighted scales. The first scale, Low Perceived External Support for PrEP, included the following items: I am worried about how people will react if ….. (a) they see me taking the study pills and (b) they see my study pill case. Responses were dichotomized: (1) agree and somewhat agree vs. (0) neither agree nor disagree, and somewhat disagree, and disagree. Items were summed to create a scale that ranged from 0 to 2; the scale’s Cronbach alpha was 0.9048.

The second scale, Participant-Staff Disclosure Challenges Scale relied on participants’ responses regarding the difficulty of reporting PrEP and condom non-use to research staff. Items included: (a) Reporting on the weekly calls that I did not take the study pills as recommended (prescribed). (If you have never missed a pill, please reply with how difficult or easy reporting a missed pill if it did happen would be); (b) Telling counselors about times when it really is hard for me to take the study pills as recommended (prescribed); (c) Reporting that I have not used condoms on the weekly study calls is/would be…; (d) Telling study counselors about times when it really is hard for me to use condoms is/would be… Responses to these four items were dichotomized: (1) difficult and somewhat difficult vs. (0) neither difficult nor easy and somewhat easy, easy. Responses for the four items were summed to create a scale that ranged from 0 to 4; the scale’s Cronbach alpha was 0.7884. A score of “0” indicates that the participant did not experience barriers related to this scale. This scale was not intended to represent whether participants liked/disliked study staff, rather it was used to quantify participants’ desire/avoidance of open discourse with study staff regarding pill or condom use. Missing components of either scale were interpolated as the mean of the observed components of the scale for that person at that visit.

Three additional variables pertaining to the participants’ perceptions about study staff behaviors were analyzed individually: (1) Staff Judgment: “In this study clinic, the staff sometimes judge me”. Responses were dichotomized as follows: (1) agree; somewhat agree vs. (0) neither agree nor disagree; somewhat disagree; disagree; (2) Staff Reactions: “Do you think study staff react to participants’ reports of pill taking?” Responses were dichotomized: (1) Yes, they seem to react negatively when I tell them I am taking the pills as recommended (prescribed) vs. (0) No, they seem to react neutral (not positive or negative) when I tell them I am taking the pills as recommended (prescribed) and Yes, they seem to react positively when I tell them I am taking the pills as recommended (prescribed). (3) Staff Support: “The staff at the clinic support me in taking the study pills”. Responses were dichotomized: (1) somewhat disagree and disagree vs. (0) agree and somewhat agree and neither agree nor disagree.

Dependent Variable: Adherence to PrEP

Adherence was assessed by plasma TFV concentration, a continuous variable, at weeks 18 and 30; this analysis included data from 302 plasma samples. For statistical analysis, plasma TFV concentrations below the limit of quantification (BLQ) were recoded to 0.16 ng/ml (half the lower limit of assay quantification)¹⁶. TFV values ranged from 0.16 to 743 ng/ml and were log-transformed for use in multivariable regression models. Results are presented as exponentiated coefficients which may be interpreted as the fold-change in plasma TFV concentration per one unit change in the covariate. To assess the magnitude of adherence impact on HIV protection, we used HPTN 066 plasma TFV benchmark concentrations for 7 and 4 doses per week, associated with complete protection in women and MSM and TGW, respectively ¹³.

Statistical Methods

Participants who lacked data for TFV concentrations, the Low Perceived External Support for PrEP, or creatinine clearance were excluded from this analysis (n=9). Data from the remaining 161 daily-dosing arm participants across all study sites were analyzed. We employed Pearson Chi-Square tests for categorical data and t-tests for continuous data. GEE linear regression was used to assess the association between each social challenge scale (i.e., Low Perceived External Support for PrEP and Participant-Staff Disclosure Challenges Scale) and log-transformed plasma TFV concentrations. GEE linear regression (identity link, Gaussian distribution) with an exchangeable correlation matrix was used to account for multiple observations per participant. Variables that were statistically significant at p<0.10 in univariate analyses were included in the multivariable GEE linear regression models. The following variables are known to influence drug metabolism and plasma TFV concentration and thus were forced into regression models regardless of their significance level in univariate analyses: study-related side effects, participant weight, creatinine clearance, and self-reported days since last dose ^15,17. We also tested whether there was an interaction between site and both scales, however, the interaction term was non-significant after controlling for other covariates (i.e., study-related side effects, weight, creatinine clearance and self-reported days since last dose), thus, these results are not shown.

RESULTS

Participant Characteristics

The study included participants from Cape Town (n=50), Bangkok (n=53), and Harlem (n=47). At baseline, the mean age of all participants (n=161) was 30.9 years (standard deviation [SD]: 9.1 years; data not shown) and by site was: Cape Town: 28.5 years (SD: 8.8 years), Bangkok: 31.4 years (SD: 5.4 years), and Harlem: 33.2 years (SD: 12 years, Table 1). Regarding educational attainment, on average, 42% of all participants reported completing up to and including a secondary education (vs. technical training/college education: 58%) and variations by site were statistically significant (p<0.01). Only 1.9% of Bangkok participants reported completing through a secondary education (vs. technical/college education: 98.1%). Cape Town participants overwhelmingly reported a lower educational attainment (≤secondary education: 90.0%; technical/college education: 10.0%). In Harlem, 36.2% of participants completed through a secondary education (vs. technical/college education: 63.8%).

Table 1.

Demographic Characteristics of HPTN 067 daily dosing arm participants at baseline, July 2012-May, 2014.

Variable	Cape Town (n=50)	Bangkok (n=53)	Harlem (n=47)
Mean Age in years (SD)	28.5 (8.8)	31.4 (5.4)	33.2 (12.0)
Education* Secondary education or less	90.0%	1.9%	36.2%
Technical training or college education	10.0%	98.1%	63.8%

^*p<0.01

Average TFV and Social Challenges Scales

Table 2 summarizes averages for TFV concentrations in plasma, and the average score for both scales. The average TFV concentration across all samples was 78.1 ng/ml and ranged from BLQ to 743 ng/ml. Bangkok was identified to have significantly greater TFV concentrations (vs. Cape Town, p<0.01) while those of Harlem participants did not vary from Cape Town participants’ TFV levels (p=0.15). Specifically, 52.65% of records had ≥35.5ng/ml of TFV in plasma, reflecting adherence to daily TFV use (i.e., 7 doses/week, data not shown). Furthermore, Bangkok had the greatest proportion of records that reflected daily adherence (70.8% vs. 43.7% in Cape Town and 41.9% in Harlem; data not shown).

Table 2.

Plasma TFV concentration and social relationships scales stratified by HPTN 067 site, daily dosing arm participants, (Weeks 18 and 30 data combined; 161 participants, n=302 records), July 2012-May, 2014.

Variable	Cape Town (n=103)	Bangkok (n=106)	Harlem (n=93)	Overall (n=302)
Mean TFV concentration ng/ml, SD (min, max)	71.8, 108.2 (BLQ, 476)	81.7, 81.4 (BLQ, 419)*	81.0, 132.2 (BLQ, 743)	78.1, 107.9 (BLQ, 743)
Low Perceived External Support for PrEP Scale Proportion (95% Confidence Interval)
0 barriers/items reported	68.9% (59.2, 77.2)	80.2% (71.4, 86.8)	77.4% (67.6, 84.9)	75.5% (70.2, 80.0)
1 barrier/item reported	9.7% (5.2, 17.2)	3.8% (1.4, 9.7)	5.4% (2.2, 12.5)	6.3% (4.0, 9.7)
2 barriers/items reported	21.4% (14.4, 30.4)	16.0% (10.1, 24.4)	17.2% (10.7, 26.5)	18.2% (14.2, 23.0)
Participant-Staff Disclosure Challenges Scale* Proportion (95% Confidence Interval)
0 barriers/items reported	57.3% (47.4, 66.6)	73.6%* (64.2, 81.2)	83.9%* (74.8, 90.1)	71.2% (65.8, 76.0)
1 barrier/item reported	10.7% (5.9, 18.4)	11.3% (6.4, 19.0)	7.5% (3.6, 15.1)	9.9% (7.0, 13.9)
2 barriers/items reported	16.5% (10.4, 25.1)	12.3% (7.2, 20.1)	6.5% (2.9, 13.8)	11.9% (8.7, 16.1)
3 barriers/items reported	4.9% (2.0, 11.3)	1.9% (0.05, 7.3)	1.1% (0.1, 7.5)	2.7% (1.3, 5.2)
4 barriers/items reported	10.7% (6.0, 18.4)	0.9% (0.1, 6.6)	1.1% (0.1, 7.5)	4.3% (2.5, 7.3)

^*p<0.01 as compared to Cape Town; SD refers to the Standard Deviation; BLQ refers to “Below the Limit of Quantification”

With respect to the social challenges scales, it was observed that overall, three-quarters of responses did not endorse either of the two barriers within the Low Perceived External Support for PrEP scale and results did not vary by site (p=0.29, Table 2). Results for the Participant-Staff Disclosure Challenges Scale varied by site. Overall, about seven in 10 responses did not report any barriers related to staff disclosure. However, Cape Town reported the greatest proportion of barriers related to the Participant-Staff Disclosure Challenges Scale (p<0.01).

Unadjusted Relationship Between Variables and PrEP Adherence

Table 3 summarizes the unadjusted relationship between all variables and TFV concentrations. The following variables were positively and significantly associated with PrEP adherence at p ≤0.10: increasing age, self-report that staff judges patient, study related side effects, and participating at the Bangkok site (vs. reference group: Cape Town). Factors associated with lower adherence include less education, perceived problematic drug use, low perceived control over HIV infection, Low Perceived External Support for PrEP, Participant-Staff Disclosure Challenges Scale, staff have negative reactions to PrEP pill taking, weight, and days since last dose.

Table 3.

Bivariate regressions of independent variables with dependent plasma TFV concentrations, HPTN 067 daily dosing arm participants (All Sites; Weeks 18/30 Data Combined; 161 participants, n=302 records), July 2012-May, 2014.

Variable	exp(β) (95% Confidence Interval)	p-value
Age	1.04 (0.99- 1.08)	0.10
Less Than Secondary Education	0.29 (0.14- 0.60)	<0.01
Perceived Problematic Drug Use, Past 3 Months	0.17 (0.05- 0.64)	<0.01
Perceived Problematic Drinking, Past 6 Months	1.36 (−0.64- 3.37)	0.18
Perceived Low Control Over HIV Infection	0.55 (0.28- 1.07)	0.08
Low Perceived External Support for PrEP Scale (Range: 0-2)	0.57 (0.38- 0.85)	<0.01
Participant-Staff Disclosure Challenges Scale (Range 0-4)	0.58 (0.42- 0.78)	<0.01
Staff Judge Patient	2.15 (1.01- 4.40)	0.04
Staff Have Negative Reactions to PrEP Pill Taking	0.34 (0.10- 1.22)	0.10
Low Staff Support for PrEP Pill Taking	1.46 (0.54- 3.99)	0.46
Study-Related Side Effects	2.41 (1.05- 5.53)	0.04
Weight (At weeks 18 and 30; Kg)	0.98 (0.96- 0.99)	0.04
Creatinine Clearance (mL/min)	0.99 (0.97- 1.00)	0.15
Self-report Days Since Last Dose	0.70 (0.62- 0.79)	<0.01
Study Site (ref. Cape Town)
Bangkok	5.12 (2.22- 11.8)	<0.01
Harlem	0.53 (0.23- 1.26)	0.15

Factors Independently Associated with PrEP Adherence

In multivariable GEE regression analyses, after controlling for statistically significant social and biological covariates from the univariate analysis, we found that higher scores in the Participant-Staff Disclosure Challenges Scale were significantly associated with lower PrEP adherence (exp(β) = 0.62, 95% CI (0.46- 0.84); p=0.002; Table 4). In contrast, the relationship between the Low Perceived External Support for PrEP scale and adherence was non-significant. Increasing self-reported days since the last dose were significantly associated with lower TFV concentrations (exp(β) = 0.73, 95% CI (0.65- 0.83); p≤0.001). Other model covariates were non-significant. To put the magnitude of the disclosure scale in context, if ciswomen from Cape Town endorsed any items from this scale, the associated plasma TFV concentration indicated less than daily adherence (≤4 doses/week); this dosing strategy is associated with less than 90% protection from HIV infection according to HPTN 066 benchmarks and randomized clinical trial outcomes ^13,18,19. For MSM and transgender women in Harlem and Cape Town, endorsement of ≥3 items was associated with plasma TFV concentrations below those associated with >90% HIV protection (2 or fewer doses per week).

Table 4.

Multivariable regressions of statistically significant univariate analysis variables on plasma TFV Concentration among HPTN 067 daily dosing arm participants (All Sites; Weeks 18/30 Data Combined; 161 participants, n=302 records) , July 2012-May, 2014.

	Exponentiated Adjusted Coefficient	95% Confidence Interval	p-value
Low Perceived External Support for PrEP Scale (Range: 0-2)	0.86	0.59- 1.25	p=0.42
Participant-Staff Disclosure Challenges Scale (Range 0-4)	0.62	0.46- 0.84	p=0.002
Self-report days since last dose	0.73	0.65- 0.83	p≤0.001

Model covariates are: age, education, problematic substance abuse, low perceived control over HIV infection, staff judge patient; staff have negative reactions to PrEP pill taking, study site, study-related side effects, weight, creatinine clearance.

DISCUSSION

This study advances our understanding of the connection between socially influenced factors and PrEP adherence in a culturally and sexually diverse clinical trial population. The study examines several relationships suggested within the Mutuality Framework and leverages TFV concentration data to objectively and quantitatively assess the role of these relationships and PrEP adherence. Research addressing challenges to PrEP adherence can benefit clinical trials by improving outcomes through influencing protocol design and execution; findings may also be pertinent to translational community-based studies.

The Participant-Staff Disclosure Challenges Scale may reflect limited communication around PrEP and condom use and thus, potentially, lower trust by the participant with the study staff to share this information. Increasing scores on the scale were significantly associated with reduced PrEP adherence and consequently, reduced protection from HIV infection. Among ciswomen, endorsement of disclosure scale items reduced adherence below daily dosing levels that are associated with the greatest level of protection from HIV. For MSM and transgender women, endorsement of ≥3 items was associated with adherence below the 4 doses per week level associated with the highest levels of protection.

Participant-staff dialogue has previously been identified as important for improving PrEP adherence in research studies. Qualitative interviews with PrEP-using participants in the FEM-PrEP trial revealed that adherence was positively influenced by adherence counseling, including staff availability to address participants’ unique challenges to PrEP use ^20,21. Similarly, community studies have reported gains in antiretroviral adherence when providers focus on patient’s specific concerns ²². Additionally, disclosure of barriers to daily medication use may also be tied to patients’ levels of trust (e.g., interpersonal trust developed through multiple interactions; social trust in institutions ^23–25). Findings associated with this scale suggest that assessing participant-study staff relationships, including open communication, is valuable. Interventions to promote communication between participants and study staff may improve PrEP adherence. Concepts of communication and trust should be studied using appropriate measures to identify their unique roles within PrEP adherence trials.

The present study leveraged biological monitoring data, which is important since prior studies have documented discrepancies between self-reported PrEP use and biological assessments of TFV concentrations ^5,25,26. Consequently, researchers have proposed using biomedical adherence monitoring (e.g., drug concentration in hair, dried blood spots, and plasma) as one objective method for supporting staff-participant adherence discussions ^25,27,28. Plasma samples shed light on recent PrEP use (≤7 days) (data collected via other methods were not systematically available in HPTN 067). Adherence monitoring through objective methods with feedback to participants was used in the iPrEx Open Label Extension and other implementation projects ²⁹. A sub-sample of iPrEX OLE participants found this strategy to be acceptable, though these data were not universally perceived as useful. A minority of participants reported improving their PrEP use following receipt of the adherence data. Importantly, participants desired specific guidance regarding the relationship between PrEP dosing and HIV protection ²⁷, and providing timely feedback is also an important component of problem-solving around PrEP adherence ^25,30,31. Incorporating biomedical adherence monitoring can enhance discussions regarding challenges to PrEP adherence and support the development of participant/patient-specific solutions over time; these conditions may be critical for improved adherence. Furthermore, ensuring that cost-effective and suitable time-relevant specimens are routinely collected can help researchers characterize PrEP adherence and subsequently, dialogue with study participants regarding barriers to PrEP use.

The Low Perceived External Support for PrEP scale was not found to be significantly associated with PrEP adherence in this analysis; this finding is counter to those of qualitative studies of PrEP, including HPTN 067 data, which have consistently demonstrated that social relationships and perceived reactions of sexual partners and family/friends have an important bearing on PrEP use^2,11. Participants’ receipt of encouragement for taking PrEP by important persons in their lives may facilitate habit formation ²⁰, which is a key factor in daily medication use. Alternatively, “HIV Stigma” has been reported by healthy individuals who are assumed to be seropositive due to their PrEP use, or labeled as sexually promiscuous by partners of PrEP users; it may contribute to reduced PrEP uptake ¹¹. Importantly, studies in other settings have found that when participants’ contacts are supportive of PrEP use, individuals’ adherence behaviors have been favorable²⁰. Developing questionnaire items that reflect community resilience and encouragement may shed light on external influences and their potential role as facilitators of PrEP use. Improving PrEP use may require ongoing multi-level social interventions that target individuals, families and communities; such approaches may help overcome social barriers to PrEP delivery and use ^32–34.

Limitations

Although we used the Mutuality Framework to guide our study (i.e., Distrust and Uncertainty domains), survey data were not available to test other domains. Findings may not be generalizable to other dosing regimens, communities, or non-trial settings and merit further examination in those contexts. The small number of participants and plasma samples may have affected our ability to detect statistically significant relationships between the scales and adherence and these concepts should be tested with larger samples. We lacked staff members’ perspectives regarding the staff-participant relationship, such data can inform our understanding of how staff-related factors can support PrEP adherence.

Conclusions

Our findings suggest that interventions to promote positive interactions between participants and study staff, especially open dialogue, disclosure, and trust, should be explored to reduce PrEP non-adherence. These factors can be independent from “liking” study staff and may reflect social, political, economic and historical factors that contextualize how members of communities view medical or HIV-specific services. A spectrum of comprehensive strategies to facilitate daily PrEP uptake are likely needed and can aid communities in reaching their PrEP implementation goals. Future studies should consider including survey items that assess dimensions of the participant-study staff relationship and community-level barriers and facilitators to PrEP uptake. Fielding these instruments within larger, sexually and geographically diverse samples will complement qualitative studies on these topics. Findings may shed light on targeted interventions that can be implemented in clinical trials and community settings.

Meetings at which parts of data were presented:

• Ojeda VD. Amico KR, Hughes J, Wilson E, Holtz TH, Chitwarakorn A, Grant RM, Dye BJ, Bekker LG, Mannheimer S, Marzinke M, Hendrix CW. Associations Between Research Staff-Participant Interactions and HIV PrEP Adherence in HPTN 067. 12th International Conference on HIV Treatment and Prevention Adherence, June 4-6, 2017.

• Ojeda VD. Amico KR, Hughes J, Wilson E, Holtz TH, Chitwarakorn A, Grant RM, Dye BJ, Bekker LG, Mannheimer S, Marzinke M, Hendrix CW. Associations Between Research Staff-Participant Interactions and HIV PrEP Adherence in HPTN 067. National Hispanic Science Network Annual Conference. Phoenix, AZ. October 5, 2017.