CONCISE COMMUNICATION
The impact of adherence to preexposure prophylaxis on the risk of HIV infection among people who inject drugs Michael Martina,b, Suphak Vanichsenic, Pravan Suntharasamaic, Udomsak Sangkumc, Philip A. Mocka,b, Manoj Leethochawalitd, Sithisat Chiamwongpaetd, Marcel E. Curlina,b, Supawadee Na-pompeta, Anchalee Warapronmongkholkula, Somyot Kittimunkonge, Roman J. Gvetadzeb, Janet M. McNichollb, Lynn A. Paxtonb, Kachit Choopanyac, for the Bangkok Tenofovir Study Group Objective: To describe participant adherence to daily oral tenofovir in an HIV preexposure prophylaxis (PrEP) trial, examine factors associated with adherence, and assess the impact of adherence on the risk of HIV infection. Design: The Bangkok Tenofovir Study was a randomized, double-blind, placebocontrolled trial conducted among people who inject drugs, 2005–2012. Methods: Participants chose daily visits or monthly visits. Study nurses observed participants swallow study drug and both initialed a diary. We assessed adherence using the diary. We examined adherence by age group and sex and used logistic regression to evaluate demographics and risk behaviors as predictors of adherence and Cox regression to assess the impact of adherence on the risk of HIV infection. Results: A total of 2413 people enrolled and contributed 9665 person-years of followup (mean 4.0 years, maximum 6.9 years). The risk of HIV infection decreased as adherence improved, from 48.9% overall to 83.5% for those with at least 97.5% adherence. In multivariable analysis, men were less adherent than women (P ¼ 0.006) and participants 20–29 years old (P < 0.001) and 30–39 years old (P ¼ 0.01) were less adherent than older participants. Other factors associated with poor adherence included incarceration (P ¼ 0.02) and injecting methamphetamine (P ¼ 0.04). Conclusion: In this HIV PrEP trial among people who inject drugs, improved adherence to daily tenofovir was associated with a lower risk of HIV infection. This is consistent with trials among MSM and HIV-discordant heterosexual couples and suggests that HIV PrEP can provide a high level of protection from HIV infection. Copyright ß 2015 Wolters Kluwer Health, Inc. All rights reserved.
AIDS 2015, 29:819–824 Keywords: adherence, HIV, people who inject drugs, preexposure prophylaxis, tenofovir a
Thailand MOPH – US CDC Collaboration, Nonthaburi, Thailand, bCenters for Disease Control and Prevention, Atlanta, USA, Bangkok Tenofovir Study Group, dBangkok Metropolitan Administration, Bangkok, and eThailand Ministry of Public Health, Nonthaburi, Thailand. Correspondence to Michael Martin, Chief, HIV Clinical Research, Thailand MOPH-US CDC Collaboration, DDC 7 Building, 4th Floor, Ministry of Public Health, Soi 4, Nonthaburi 11000, Thailand. E-mail: [email protected] Received: 11 December 2014; revised: 27 January 2015; accepted: 29 January 2015. c
DOI:10.1097/QAD.0000000000000613
ISSN 0269-9370 Copyright Q 2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
819
820
AIDS
2015, Vol 29 No 7
Introduction More than 2.1 million people were infected with HIV in 2013 [1]. Randomized trials have shown that daily use of tenofovir disoproxil fumarate (tenofovir)-emtricitabine as preexposure prophylaxis (PrEP) can reduce HIV incidence 44% among MSM [2], 62% among heterosexual men and women [3], and 75% among HIV-serodiscordant heterosexual couples [4] and that tenofovir alone can reduce HIV transmission 67% among HIV-serodiscordant heterosexual couples [4] and 49% among people who inject drugs (PWID) [5]. Based on these results, the US Public Health Service published guidelines for the use of PrEP to prevent HIV infection [6]. Adherence is required for PrEP to be effective [7–9]. During the Partners PrEP study among HIV-serodiscordant heterosexual couples, researchers provided additional counseling to a subgroup of 1147 participants if their adherence dropped below 80% [9]. In this subgroup, no incident HIV infections occurred in the treatment arms resulting in a PrEP effectiveness of 100% [95% confidence interval (CI), 84–100%] [4,9]. In contrast, in two PrEP trials, wherein drug level testing suggested poor adherence, effectiveness was not shown [10,11]. Unique among the PrEP trials, the Bangkok Tenofovir Study (BTS) offered participants daily follow-up and staff watched participants swallow study drug [5]. In this study, we describe participant adherence, examine demographic and risk factors associated with adherence, and estimate tenofovir effectiveness based on overall adherence.
Methods The BTS was a randomized, double-blind, placebocontrolled, endpoint-driven study conducted in 17 Bangkok Metropolitan Administration (BMA) drugtreatment clinics that offer clients an HIV-prevention package, social services, and medical care [5]. Participants chose daily visits or monthly visits and could change from daily to monthly visits or vice versa at monthly visits. At daily visits, study nurses observed participants swallow study drug, both initialed the adherence diary, and participants received 70 baht (about 1.8 US dollars) to compensate for time and travel. At monthly visits, all participants (daily and monthly) received individualized adherence counseling and 350 baht (about 9.2 US dollars) compensation and staff reviewed the diary, did a pill count, and worked with participants to resolve discrepancies. Staff gave participants appointment cards, phoned them before monthly visits, and contacted participants who missed monthly visits by phone or, if they could not contact them by phone, with a home visit. We defined adherence as the proportion of days, recorded in the diary, that the participant took study drug. We
examined adherence by age group and sex and used Wilcoxon or Kruskal–Wallis tests to assess group differences [12], adjusting for age with van Elteren’s test [13]. We used Spearman’s correlation to assess the association between days in daily follow-up and adherence [12], and Cox regression for the adherencebased effectiveness analysis [14]. We used logistic regression to evaluate demographics and risk behaviors during the 3 months before enrollment as predictors of less than 95% adherence [12]. We chose 95% because data from PrEP trials suggest a high level of adherence is associated with protection from HIV infection [2,4,5,9,15]. Variables associated with poor adherence in bivariable analysis (P 0.1) were evaluated in a multivariable model and retained in the final model if the P value was 0.05 or less. We included treatment group in all models and censored data at the last HIV test. We used SAS version 9.3 (SAS Institute, Cary, North Carolina, USA) for statistical analyses. The ethical review committees of the BMA, Thailand Ministry of Public Health, and US Centers for Disease Control and Prevention approved the protocol.
Results We described BTS results in previous publications [5,16,17]. Briefly, during 2005–2012, 2413 HIVnegative individuals who injected drugs the previous year enrolled and were randomly assigned to receive daily tenofovir or placebo. Exit visits were completed in June 2012 and participants contributed 9665 person-years (mean 4.0 years; maximum 6.9 years) of follow-up; 355 (14.7%) were lost to follow-up during the 7-year study. Fifty participants became HIV-infected: 17 in the tenofovir group (incidence, 0.35 per 100 person-years) and 33 in the placebo group (incidence, 0.68 per 100 person-years) indicating a 48.9% reduction in HIV incidence (95% CI, 9.6–72.2; P ¼ 0.01) among participants randomized to tenofovir. Participant’s median age at enrollment was 31 years (range 20–59), 1924 (79.7%) were men, 1507 (62.5%) reported injecting drugs during the 3 months before enrollment: 801 (53.2%) injected methamphetamine, 559 (37.1%) midazolam, and 527 (35.0%) heroin, and 1239 (51.4%) participants received methadone during the study. At enrollment, 2231 (92.5%) participants chose daily visits [i.e. 500 (95.4%) of those in the daily methadone programme at baseline and 1723 (91.6%) of those not in the programme (P ¼ 0.004); methadone use data were missing from eight participants]; 628 (26.0%) remained in daily follow-up, 1711 (70.9%) switched between daily and monthly visits, and 74 (3.1%) were in monthly follow-up throughout the study. Overall, participants were in daily follow-up an average of 86.9% of the time: 1534 (63.6%) were in daily follow-up at least 95% of the time, 1742 (72.3%) at least
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Adherence to HIV preexposure prophylaxis Martin et al.
Participants took study drug an average of 83.8% of days; 84.4% while in daily follow-up and 88.9% while in monthly follow-up: 1132 (46.9%) took study drug at least 95% of the time, 1462 (60.6%) at least 90% of the time, 320 (13.3%) 80–89% of the time, 175 (7.3%) 70–79% of the time, and the number of participants ranged from 31–129 (1.3–5.4%) in each decile below 70%. Adherence did not differ by treatment group (P ¼ 0.16) or by time in study (P ¼ 0.22). There was a modest positive correlation between the number of days in daily followup and adherence of 0.32 (95% CI, 0.29–0.36). Adherence was higher among participants 40–59 years old [median 98.2%, interquartile range (IQR) 93.5– 99.5%] than participants 30–39 years old (median 94.2%, IQR 82.2–98.6%) and 20–29 years old (median 90.0%, IQR 68.9–97.5%) (P < 0.001). Controlling for age, women were more adherent (median 95.6%, IQR 81.1– 98.9%) than men (median 93.8%, IQR 78.8–98.7%) (P ¼ 0.04). Adherence, adjusted for age, did not differ significantly among participants reporting abdominal pain (P ¼ 0.15), nausea (P ¼ 0.32), diarrhea (P ¼ 0.14), or any adverse event (P ¼ 0.09) compared with participants who did not have these events. Among placebo recipients, adherence of the 33 participants who became HIV-infected (median 94.0%, IQR 65.0–98.8%) was similar to the 1174 who remained HIV-uninfected (median 94.3%, IQR 81.2–98.8%; P ¼ 0.45). However, among tenofovir recipients, median adherence of the 17 participants who became HIVinfected was 76.6% (IQR 65.3–94.1%), whereas median adherence of the 1187 who remained uninfected was 94.1% (IQR 77.8–98.6%; P ¼ 0.01). An adherence-based analysis, limited to participants coming daily who met adherence criteria (i.e., took study drug 71% of days with no more than 2 consecutive days off study drug) during a 2–3 month period before incident HIV infections occurred, showed a 55.9% (95% CI, 18.8 to 86.0; P ¼ 0.11) reduction in HIV risk associated with PrEP [5]. We repeated the analysis including all participants (i.e. those coming daily and monthly) and found a similar 55.9% (95% CI, 9.8 to 84.4; P ¼ 0.08) reduction in HIV risk, suggesting the visit schedule, daily versus monthly, did not alter the adherence-based estimate of PrEP effectiveness. Thus, using Cox regression to analyze adherence data from all participants, we found that as adherence improved, the effectiveness of PrEP increased; from 48.9% overall to 58.0% for participants with at least 75% adherence, and to 83.5% for those with at least 97.5% adherence (Fig. 1). In multivariable analysis, men were more likely to report less than 95% adherence than women (P ¼ 0.006) and
participants 20–29 years old (P < 0.001) and 30–39 years old (P ¼ 0.01) were more likely to report less than 95% adherence than older participants (Table 1). Other factors associated with less than 95% adherence included incarceration in prison (P ¼ 0.02), injecting methamphetamine (P ¼ 0.04), and having sex with a casual partner (P < 0.001) in the 3 months before enrollment. Participants in a methadone programme at enrollment were more likely to report at least 95% adherence (P ¼ 0.003).
Discussion In this HIV PrEP trial among PWID, we found that increasing levels of adherence were associated with a reduction in the risk of HIV infection; from a 48.9% reduction overall to 83.5% among participants with at least 97.5% adherence. This is consistent with trials among MSM [2] and HIV-discordant heterosexual couples [4] and suggests that PrEP can provide high levels of protection from HIV infection if adherence is maintained. We found better adherence among older participants (P < 0.001) and, controlling for age, among women (P ¼ 0.04). Interestingly, although the trial was not powered to assess effectiveness among subgroups, we found statistically significant effectiveness among women (78.6%; P ¼ 0.03) and participants at least 40 years old (88.9%; P ¼ 0.01) [5]. Participants who were incarcerated or injected methamphetamine before enrollment were more likely to report less than 95% adherence, suggesting poor adherence among some at-risk participants. In May 2014, the US Public Health Service issued guidelines advocating the use of PrEP for adults at substantial risk of HIV infection [6]; however, PrEP has stirred debate because of concerns about increased risk 100% 80%
Effectiveness
90% of the time, 225 (9.3%) 80–89% of the time, and less than 4.5% in each decile below 80%.
60% 40% 20% 0% ≥60
≥65
≥70
≥75
≥80
≥85
≥90
≥95
≥97.5
1462
1132
849
Percent adherence Number 2086
2033
1957
1890
Effectivenessa
1782
1664
95% confidence interval
Fig. 1. Estimated reduction in the risk of HIV infection (effectiveness) among 2413 participants in the Bangkok Tenofovir Study by level of adherence, 2005–2012. aEffectiveness was estimated using Cox regression.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
821
822
AIDS
2015, Vol 29 No 7
Table 1. Results of bivariable and multivariable analysis using logistic regression to evaluate baseline demographic characteristics and risk factors as predictors of poor (
The impact of adherence to preexposure prophylaxis on the risk of HIV infection among people who inject drugs Michael Martina,b, Suphak Vanichsenic, Pravan Suntharasamaic, Udomsak Sangkumc, Philip A. Mocka,b, Manoj Leethochawalitd, Sithisat Chiamwongpaetd, Marcel E. Curlina,b, Supawadee Na-pompeta, Anchalee Warapronmongkholkula, Somyot Kittimunkonge, Roman J. Gvetadzeb, Janet M. McNichollb, Lynn A. Paxtonb, Kachit Choopanyac, for the Bangkok Tenofovir Study Group Objective: To describe participant adherence to daily oral tenofovir in an HIV preexposure prophylaxis (PrEP) trial, examine factors associated with adherence, and assess the impact of adherence on the risk of HIV infection. Design: The Bangkok Tenofovir Study was a randomized, double-blind, placebocontrolled trial conducted among people who inject drugs, 2005–2012. Methods: Participants chose daily visits or monthly visits. Study nurses observed participants swallow study drug and both initialed a diary. We assessed adherence using the diary. We examined adherence by age group and sex and used logistic regression to evaluate demographics and risk behaviors as predictors of adherence and Cox regression to assess the impact of adherence on the risk of HIV infection. Results: A total of 2413 people enrolled and contributed 9665 person-years of followup (mean 4.0 years, maximum 6.9 years). The risk of HIV infection decreased as adherence improved, from 48.9% overall to 83.5% for those with at least 97.5% adherence. In multivariable analysis, men were less adherent than women (P ¼ 0.006) and participants 20–29 years old (P < 0.001) and 30–39 years old (P ¼ 0.01) were less adherent than older participants. Other factors associated with poor adherence included incarceration (P ¼ 0.02) and injecting methamphetamine (P ¼ 0.04). Conclusion: In this HIV PrEP trial among people who inject drugs, improved adherence to daily tenofovir was associated with a lower risk of HIV infection. This is consistent with trials among MSM and HIV-discordant heterosexual couples and suggests that HIV PrEP can provide a high level of protection from HIV infection. Copyright ß 2015 Wolters Kluwer Health, Inc. All rights reserved.
AIDS 2015, 29:819–824 Keywords: adherence, HIV, people who inject drugs, preexposure prophylaxis, tenofovir a
Thailand MOPH – US CDC Collaboration, Nonthaburi, Thailand, bCenters for Disease Control and Prevention, Atlanta, USA, Bangkok Tenofovir Study Group, dBangkok Metropolitan Administration, Bangkok, and eThailand Ministry of Public Health, Nonthaburi, Thailand. Correspondence to Michael Martin, Chief, HIV Clinical Research, Thailand MOPH-US CDC Collaboration, DDC 7 Building, 4th Floor, Ministry of Public Health, Soi 4, Nonthaburi 11000, Thailand. E-mail: [email protected] Received: 11 December 2014; revised: 27 January 2015; accepted: 29 January 2015. c
DOI:10.1097/QAD.0000000000000613
ISSN 0269-9370 Copyright Q 2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
819
820
AIDS
2015, Vol 29 No 7
Introduction More than 2.1 million people were infected with HIV in 2013 [1]. Randomized trials have shown that daily use of tenofovir disoproxil fumarate (tenofovir)-emtricitabine as preexposure prophylaxis (PrEP) can reduce HIV incidence 44% among MSM [2], 62% among heterosexual men and women [3], and 75% among HIV-serodiscordant heterosexual couples [4] and that tenofovir alone can reduce HIV transmission 67% among HIV-serodiscordant heterosexual couples [4] and 49% among people who inject drugs (PWID) [5]. Based on these results, the US Public Health Service published guidelines for the use of PrEP to prevent HIV infection [6]. Adherence is required for PrEP to be effective [7–9]. During the Partners PrEP study among HIV-serodiscordant heterosexual couples, researchers provided additional counseling to a subgroup of 1147 participants if their adherence dropped below 80% [9]. In this subgroup, no incident HIV infections occurred in the treatment arms resulting in a PrEP effectiveness of 100% [95% confidence interval (CI), 84–100%] [4,9]. In contrast, in two PrEP trials, wherein drug level testing suggested poor adherence, effectiveness was not shown [10,11]. Unique among the PrEP trials, the Bangkok Tenofovir Study (BTS) offered participants daily follow-up and staff watched participants swallow study drug [5]. In this study, we describe participant adherence, examine demographic and risk factors associated with adherence, and estimate tenofovir effectiveness based on overall adherence.
Methods The BTS was a randomized, double-blind, placebocontrolled, endpoint-driven study conducted in 17 Bangkok Metropolitan Administration (BMA) drugtreatment clinics that offer clients an HIV-prevention package, social services, and medical care [5]. Participants chose daily visits or monthly visits and could change from daily to monthly visits or vice versa at monthly visits. At daily visits, study nurses observed participants swallow study drug, both initialed the adherence diary, and participants received 70 baht (about 1.8 US dollars) to compensate for time and travel. At monthly visits, all participants (daily and monthly) received individualized adherence counseling and 350 baht (about 9.2 US dollars) compensation and staff reviewed the diary, did a pill count, and worked with participants to resolve discrepancies. Staff gave participants appointment cards, phoned them before monthly visits, and contacted participants who missed monthly visits by phone or, if they could not contact them by phone, with a home visit. We defined adherence as the proportion of days, recorded in the diary, that the participant took study drug. We
examined adherence by age group and sex and used Wilcoxon or Kruskal–Wallis tests to assess group differences [12], adjusting for age with van Elteren’s test [13]. We used Spearman’s correlation to assess the association between days in daily follow-up and adherence [12], and Cox regression for the adherencebased effectiveness analysis [14]. We used logistic regression to evaluate demographics and risk behaviors during the 3 months before enrollment as predictors of less than 95% adherence [12]. We chose 95% because data from PrEP trials suggest a high level of adherence is associated with protection from HIV infection [2,4,5,9,15]. Variables associated with poor adherence in bivariable analysis (P 0.1) were evaluated in a multivariable model and retained in the final model if the P value was 0.05 or less. We included treatment group in all models and censored data at the last HIV test. We used SAS version 9.3 (SAS Institute, Cary, North Carolina, USA) for statistical analyses. The ethical review committees of the BMA, Thailand Ministry of Public Health, and US Centers for Disease Control and Prevention approved the protocol.
Results We described BTS results in previous publications [5,16,17]. Briefly, during 2005–2012, 2413 HIVnegative individuals who injected drugs the previous year enrolled and were randomly assigned to receive daily tenofovir or placebo. Exit visits were completed in June 2012 and participants contributed 9665 person-years (mean 4.0 years; maximum 6.9 years) of follow-up; 355 (14.7%) were lost to follow-up during the 7-year study. Fifty participants became HIV-infected: 17 in the tenofovir group (incidence, 0.35 per 100 person-years) and 33 in the placebo group (incidence, 0.68 per 100 person-years) indicating a 48.9% reduction in HIV incidence (95% CI, 9.6–72.2; P ¼ 0.01) among participants randomized to tenofovir. Participant’s median age at enrollment was 31 years (range 20–59), 1924 (79.7%) were men, 1507 (62.5%) reported injecting drugs during the 3 months before enrollment: 801 (53.2%) injected methamphetamine, 559 (37.1%) midazolam, and 527 (35.0%) heroin, and 1239 (51.4%) participants received methadone during the study. At enrollment, 2231 (92.5%) participants chose daily visits [i.e. 500 (95.4%) of those in the daily methadone programme at baseline and 1723 (91.6%) of those not in the programme (P ¼ 0.004); methadone use data were missing from eight participants]; 628 (26.0%) remained in daily follow-up, 1711 (70.9%) switched between daily and monthly visits, and 74 (3.1%) were in monthly follow-up throughout the study. Overall, participants were in daily follow-up an average of 86.9% of the time: 1534 (63.6%) were in daily follow-up at least 95% of the time, 1742 (72.3%) at least
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Adherence to HIV preexposure prophylaxis Martin et al.
Participants took study drug an average of 83.8% of days; 84.4% while in daily follow-up and 88.9% while in monthly follow-up: 1132 (46.9%) took study drug at least 95% of the time, 1462 (60.6%) at least 90% of the time, 320 (13.3%) 80–89% of the time, 175 (7.3%) 70–79% of the time, and the number of participants ranged from 31–129 (1.3–5.4%) in each decile below 70%. Adherence did not differ by treatment group (P ¼ 0.16) or by time in study (P ¼ 0.22). There was a modest positive correlation between the number of days in daily followup and adherence of 0.32 (95% CI, 0.29–0.36). Adherence was higher among participants 40–59 years old [median 98.2%, interquartile range (IQR) 93.5– 99.5%] than participants 30–39 years old (median 94.2%, IQR 82.2–98.6%) and 20–29 years old (median 90.0%, IQR 68.9–97.5%) (P < 0.001). Controlling for age, women were more adherent (median 95.6%, IQR 81.1– 98.9%) than men (median 93.8%, IQR 78.8–98.7%) (P ¼ 0.04). Adherence, adjusted for age, did not differ significantly among participants reporting abdominal pain (P ¼ 0.15), nausea (P ¼ 0.32), diarrhea (P ¼ 0.14), or any adverse event (P ¼ 0.09) compared with participants who did not have these events. Among placebo recipients, adherence of the 33 participants who became HIV-infected (median 94.0%, IQR 65.0–98.8%) was similar to the 1174 who remained HIV-uninfected (median 94.3%, IQR 81.2–98.8%; P ¼ 0.45). However, among tenofovir recipients, median adherence of the 17 participants who became HIVinfected was 76.6% (IQR 65.3–94.1%), whereas median adherence of the 1187 who remained uninfected was 94.1% (IQR 77.8–98.6%; P ¼ 0.01). An adherence-based analysis, limited to participants coming daily who met adherence criteria (i.e., took study drug 71% of days with no more than 2 consecutive days off study drug) during a 2–3 month period before incident HIV infections occurred, showed a 55.9% (95% CI, 18.8 to 86.0; P ¼ 0.11) reduction in HIV risk associated with PrEP [5]. We repeated the analysis including all participants (i.e. those coming daily and monthly) and found a similar 55.9% (95% CI, 9.8 to 84.4; P ¼ 0.08) reduction in HIV risk, suggesting the visit schedule, daily versus monthly, did not alter the adherence-based estimate of PrEP effectiveness. Thus, using Cox regression to analyze adherence data from all participants, we found that as adherence improved, the effectiveness of PrEP increased; from 48.9% overall to 58.0% for participants with at least 75% adherence, and to 83.5% for those with at least 97.5% adherence (Fig. 1). In multivariable analysis, men were more likely to report less than 95% adherence than women (P ¼ 0.006) and
participants 20–29 years old (P < 0.001) and 30–39 years old (P ¼ 0.01) were more likely to report less than 95% adherence than older participants (Table 1). Other factors associated with less than 95% adherence included incarceration in prison (P ¼ 0.02), injecting methamphetamine (P ¼ 0.04), and having sex with a casual partner (P < 0.001) in the 3 months before enrollment. Participants in a methadone programme at enrollment were more likely to report at least 95% adherence (P ¼ 0.003).
Discussion In this HIV PrEP trial among PWID, we found that increasing levels of adherence were associated with a reduction in the risk of HIV infection; from a 48.9% reduction overall to 83.5% among participants with at least 97.5% adherence. This is consistent with trials among MSM [2] and HIV-discordant heterosexual couples [4] and suggests that PrEP can provide high levels of protection from HIV infection if adherence is maintained. We found better adherence among older participants (P < 0.001) and, controlling for age, among women (P ¼ 0.04). Interestingly, although the trial was not powered to assess effectiveness among subgroups, we found statistically significant effectiveness among women (78.6%; P ¼ 0.03) and participants at least 40 years old (88.9%; P ¼ 0.01) [5]. Participants who were incarcerated or injected methamphetamine before enrollment were more likely to report less than 95% adherence, suggesting poor adherence among some at-risk participants. In May 2014, the US Public Health Service issued guidelines advocating the use of PrEP for adults at substantial risk of HIV infection [6]; however, PrEP has stirred debate because of concerns about increased risk 100% 80%
Effectiveness
90% of the time, 225 (9.3%) 80–89% of the time, and less than 4.5% in each decile below 80%.
60% 40% 20% 0% ≥60
≥65
≥70
≥75
≥80
≥85
≥90
≥95
≥97.5
1462
1132
849
Percent adherence Number 2086
2033
1957
1890
Effectivenessa
1782
1664
95% confidence interval
Fig. 1. Estimated reduction in the risk of HIV infection (effectiveness) among 2413 participants in the Bangkok Tenofovir Study by level of adherence, 2005–2012. aEffectiveness was estimated using Cox regression.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
821
822
AIDS
2015, Vol 29 No 7
Table 1. Results of bivariable and multivariable analysis using logistic regression to evaluate baseline demographic characteristics and risk factors as predictors of poor (
