Abstract
Objective
Among participants of a clinical trial to test the efficacy of tenofovir/emtricitabine in protecting heterosexual men and women living in Botswana from HIV infection, determine 1) if sexual risk behavior, specifically condomless sex acts and number of sex partners, changed over time, 2) factors associated with condomless sex acts and number of sex partners and 3) the effect of participant treatment arm perception on risk behavior to address the possibility of risk compensation.
Methods
A longitudinal modeling of rates of risk behaviors was used to determine if the rate of condomless sex acts (#acts/person) and rate of sex partners (#partners/person) changed over time and which factors were associated with behavior change.
Results
1200 participants were analyzed over 1 year. There was a 25% decrease in the rate of sex partners among participants sexually active in the last 30 days. The rate of reported condomless sex acts was greater for males (RR=1.34, CI=1.07–1.67) and participants whose sexual debut in years was ≤ 15 years of age (RR=1.65, CI=1.14–2.38) and 16–17 (RR=1.68, CI=1.22–2.31) compared to ≥20 years. Rate of reported sex partners was greater for males (RR=3.67, CI=2.86–4.71) and participants whose age at sexual debut in years was ≤15 (RR=2.92, CI=2.01–4.22) and 16–17 (RR=2.34, CI=1.69–3.24) compared to ≥20. There was no effect of participant treatment arm perception on risk behavior.
Conclusions
Our study of PrEP to prevent HIV infection found no evidence of risk compensation which may have been due to participants’ motivations to reduce their risk behaviors and risk-reduction counseling.
Keywords: clinical trial, tenofovir/emtricitabine, risk compensation, sexual risk behavior
Introduction
A clinical trial (TDF2 study) to test the efficacy of TDF/FTC as pre-exposure prophylaxis (PrEP) in protecting heterosexual men and women living in Botswana from HIV infection found a 63% reduced risk of infection [1]. Three other clinical trials also demonstrated the efficacy of tenofovir-based PrEP. The iPrEx trial found a 42% reduced risk of HIV infection among gay men and transgender women [2, 3]. The Partners PrEP trial, discontinued early after an interim data review, found that daily oral TDF/FTC (75%) and TDF (67%) reduced risk the of HIV acquisition for the HIV-negative partner in a discordant couple [4]. The Bangkok Tenofovir study found a 49% reduced risk of HIV among injection drug users [5]. In contrast, two trials, FEM-PrEP [6] and VOICE [7], both involving women, were stopped early because there was lack of evidence of effectiveness. Follow-up analysis of the VOICE and FEM-PREP [6] trials showed that the lack of effectiveness was due, at least in part to low adherence [8]. Because of the large role poor adherence plays in explaining variability of PrEP trial outcomes, an improved understanding of pharmacological variables [9] and psychological and social determinants of behavior [10, 11] will be needed. In 2012, TDF/FTC was approved for oral daily use as PrEP by the U.S. Food and Drug Administration (FDA).
Several factors will influence the effectiveness of PrEP in the real world including adherence and risk compensation [11]. Risk homeostasis theory postulates that persons attempt to produce an output or given level of risk to health that matches the target level of risk [12]. In the case of PrEP, if a person believes that it will prevent HIV, they would theoretically reduce restrictions on their risk behaviors [13]. Risk compensation behaviors and their frequency, likely to differ depending upon the prevention intervention and various social and individual level factors [13], have long been an area of concern when considering biomedical HIV prevention interventions [2, 14].
The purpose of the present analysis of the TDF-2 study data was to determine, among a general population of adult heterosexual men and women in Botswana participating in the study, 1) if sexual behavior, specifically condomless sex acts and number of sex partners, changed over time, 2) the factors associated with condomless sex acts and number of sex partners and 3) the effect of participant treatment arm perception on risk behavior to address the possibility of risk compensation. To supplement the self-report data, a secondary analysis was done to determine the frequency of pregnancy or sexually transmitted infection (STI) diagnosis during the study, and factors associated with each.
Methods
The TDF-2 study, a phase 3 randomized, double-blind, placebo-controlled clinical trial to determine if PrEP was effective in preventing HIV among heterosexual men and women, was carried out in the cities of Francistown and Gaborone, Botswana, between May 2007 and May 2010 (ClinicalTrials.gov Identifier: NCT00448669). The full description and results of the TDF2 study can be found in the primary report [1].
Self-reported behavioral data were collected from participants at baseline and monthly thereafter using face to-face interviews augmented by audio computer-assisted self-interviews at baseline and semi-annually thereafter. HIV prevention services were provided to all participants and included individualized counseling on risk reduction, free male and female condoms, and screening for STIs. If indicated, partner notification and treatment for STIs were carried out.
Measures
Dependent Variables
Our primary analysis consisted of determining factors associated with two longitudinal measures of sexual risk behavior as well as two cumulative proxy measures of condomless sex. The two longitudinal dependent variables that measured sexual behavior were:
Number of condomless vaginal sexual acts (“number of condomless sex acts”) with both casual and main partners among those who reported having had at least one sexual partner in the previous 30 days and
Number of casual and main vaginal sex partners in the last 30 days (“number of sex partners”) among all study participants.
The two cumulative proxy measures of condomless sex were pregnancy (among women) or STI diagnosis during the course of the study. The STIs included were laboratory-confirmed gonorrhea, syphilis, chlamydia, trichomoniasis or herpes simplex virus 2. See main study paper for specifics on laboratory assessments. [1]
Independent variables
Independent variables included time, treatment group (TDF/FTC, placebo) to which the subject was assigned at randomization, treatment arm perception (or the treatment group that the respondent believed that he/she had been assigned to [TDF/FTC/placebo/unsure]) measured at baseline, 6 months, and 12 months, demographic factors (age, gender, educational attainment, screening site (Gaborone or Francistown), marital status, occupational status), and age at sexual debut. Laboratory results for STIs were also considered as independent variables in the analysis of self-reported sexual behavior.
Analysis
To measure the potential impact of selection bias at both enrollment and randomization, we assessed demographic characteristics and risk behaviors at screening by enrollment status (screened and enrolled versus screened and not enrolled) as well as among those enrolled by treatment group (TDF/FTC and placebo). For time-varying outcomes (i.e., number of condomless sex acts and number of sex partners), we used negative binomial hurdle (NBH) random-effects models [15, 16] to estimate the occurrence and rate of condomless sexual acts (#acts/person) and sexual partners (#partners/person) over time and determine factors associated with sexual behavior. To assess factors associated with a pregnancy or STI diagnosis occurring at any time during the study, we estimated the risk ratios associated with each predictor using log binomial regression.
Our NBH models analyze the data for each outcome measure independently using two components: logistic regression to estimate the odds of reporting zero condomless sex acts (or zero sex partners) and negative binomial regression to estimate the rate ratio (RR) of condomless sex acts (or sex partners) over time and across subgroups. Thus, in the analysis of number of condomless sex acts, our model estimates the odds of reporting no condomless sex acts, and for respondents reporting at least one condomless sex act we estimate the rate ratio of condomless sex acts (i.e., number of condomless sex acts per person) at different time points or across different subgroups during the previous 30 days. This allowed us to determine if the number of participants who reported always using condoms changed over time or was significantly associated with selected predictors. Also, for those who had at least one condomless sex act in a time period, this modeling strategy allowed us to determine if the rate of reported condomless sex acts changed over time or was significantly associated with selected predictors. The analysis of number of sex partners was done similarly and allowed us to determine if having no sex partners changed over time (or differed with regard to selected predictors) and, for those who had at least one sex partner, if the rate of reported sex partners changed over time (or differed with regard to selected predictors).
We accounted for correlated responses (multiple observations within subject) by treating the participant as a random effect in both components of the NBH model (i.e., logistic regression and negative binomial components). Additionally, we identified predictors associated with the outcome. Factors that were significant at the p≤0.20 level in base models were considered for inclusion in the final multivariable model. However, time and treatment perception were used regardless of statistical significance. We performed a backwards elimination variable selection process to arrive at our final multivariable models. The final model for each outcome measure consisted of predictors that were significant in each model component independently at the 0.05 level.
Results
Of the 2533 volunteers who were screened for the study, 52.2% were eligible for enrollment. The enrolled and not-enrolled participants were significantly different on several variables. Compared to not-enrolled participants, a greater proportion of enrolled participants were 30–39 years of age, had post-secondary school education, were from Gaborone, and were employed. The enrolled versus not enrolled participants also differed on variables related to eligibility criteria (Supplemental Table 1). The final number of persons who were enrolled and randomized in the TDF2 study was 1219. Of the 1219, 19 participants were excluded from analysis (3 were HIV infected at enrollment and 16 never started study medication), leaving 1200 study participants.
Subjects were randomized to TDF or placebo groups and were followed for 1563 person-years (median, 1.1 years; maximum, 3.7 years). There were no significant differences in demographic characteristics, presence of STIs in the last year, or sexual behaviors in the last 30 days between the treatment and placebo groups at enrollment (Table 1). Participant perception of their study arm assignment was recorded at baseline (TDF2 n=228 (19.4%), placebo n=72 (6.1%), not sure n=878 (74.5%)), 6 months (TDF2 n=164 (20.8%), placebo n=79 (10.0%), not sure n=546 (69.2%)), and 12 months (TDF2 n=81 (18.2%), placebo n=50 (11.3%), not sure n=313 (70.5%)).
Table 1.
Demographic and clinical characteristics and risk behaviors at enrollment by treatment group in the TDF2 clinical trial. Botswana 2007–2010.
| Demographic Characteristic or Risk Behavior |
Combined (N=1219) |
TDF/FTC (N=611) |
Placebo (N=608) |
p-value |
|---|---|---|---|---|
| Gender | ||||
| Male | 662 (54.3) | 331 (54.2) | 331 (54.4) | 0.9254 |
| Female | 557 (45.7) | 280 (45.8) | 277 (45.6) | |
| Education | ||||
| Primary or less | 40 (3.3) | 20 (3.3) | 20 (3.3) | 0.9962 |
| Secondary (Jr/Sr) | 891 (73.1) | 446 (73.0) | 445 (73.2) | |
| Postsecondary | 288 (23.6) | 145 (23.7) | 143 (23.5) | |
| Marital Status | ||||
| Married or Cohabitating | 70 (5.7) | 32 (5.2) | 38 (6.3) | 0.4465 |
| Single | 1145 (93.9) | 578 (94.6) | 567 (93.3) | |
| Separated or Divorced | 4 (0.3) | 1 (0.2) | 3 (0.5) | |
| Screening Site | ||||
| Gaborone | 651 (53.4) | 326 (53.4) | 325 (53.5) | 0.9724 |
| Francistown | 568 (46.6) | 285 (46.6) | 283 (46.6) | |
| Occupational Status Employed Unemployed |
480 (39.4) 739 (60.6) |
245 (40.1) 366 (59.9) |
235 (38.7) 373 (61.4) |
0.6052 |
| HIV | ||||
| Positive | 3 (0.3) | 1 (0.2) | 2 (0.3) | 0.5603 |
| Negative | 1216 (99.8) | 610 (99.8) | 606 (99.7) | |
| Syphilis | ||||
| Positive | 14 (1.2) | 5 (0.8) | 9 (1.5) | 0.5462 |
| Negative | 1182 (97.0) | 594 (97.2) | 588 (96.7) | |
| Missing | 23 (1.9) | 12 (2.0) | 11 (1.8) | |
| Gonorrhea | ||||
| Positive | 24 (2.0) | 12 (2.0) | 12 (2.0) | 0.4820 |
| Negative | 1119 (91.8) | 566 (92.6) | 553 (91.0) | |
| Missing | 76 (6.2) | 33 (5.4) | 43 (7.1) | |
| Trichomonas* | ||||
| Positive | 33 (5.9) | 19 (6.8) | 14 (5.1) | 0.5400 |
| Negative | 471 (84.6) | 237 (84.6) | 234 (84.5) | |
| Missing | 53 (9.5) | 24 (8.6) | 29 (10.5) | |
| Chlamydia | ||||
| Positive | 97 (8.0) | 43 (7.0) | 54 (8.9) | 0.2435 |
| Negative | 1047 (85.9) | 535 (87.6) | 512 (84.2) | |
| Missing | 75 (6.2) | 33 (5.4) | 42 (6.9) | |
| HSV-2 | ||||
| Positive | 428 (35.1) | 208 (34.0) | 220 (36.2) | 0.2082 |
| Negative | 766 (62.8) | 392 (64.2) | 374 (61.5) | |
| Indeterminate | 7 (0.6) | 1 (0.2) | 6 (1.0) | |
| Missing | 18 (1.5) | 10 (1.6) | 8 (1.3) | |
| Number of Sex Partners Last Month |
||||
| 0 | 151 (12.4) | 73 (12.0) | 78 (12.8) | |
| 1 | 815 (66.9) | 410 (67.1) | 405 (66.6) | |
| 2 | 172 (14.1 | 86 (14.1) | 86 (14.1) | |
| 3+ | 60 (4.9) | 32 (5.2) | 28 (4.6) | |
| Missing | 21 (1.7) | 10 (1.6) | 11 (1.8) | 0.9732 |
| Total Number of All Sex Acts with Any Partner Last Month |
||||
| 0 | 175 (14.4) | 87 (14.2) | 8 (14.5) | 0.9815 |
| 1 | 94 (7.7) | 45 (7.4) | 849 (8.1) | |
| 2 | 143 (11.7) | 74 (12.1) | 69 (11.4) | |
| 3+ | 786 (64.5) | 395 (64.7) | 391 (64.3) | |
| Missing | 21 (1.7) | 10 (1.6) | 11 (1.8) | |
| Number of Condomless Sex Acts (Any) Last Month |
||||
| 0 | 805 (66.0) | 402 (65.8) | 403 (66.3) | 0.1976 |
| 1 | 67 (5.5) | 31 (5.1) | 36 (5.9) | |
| 2 | 38 (3.1) | 27 (4.4) | 11 (1.8) | |
| 3+ | 137 (11.2) | 68 (11.1) | 69 (11.4) | |
| Did Not Have Sex | 151 (12.4) | 73 (12.0) | 78 (12.8) | |
| Missing | 21 (1.7) | 10 (1.6) | 11 (1.8) |
Note: There were 1219 subjects enrolled in the study. However, 18 of these subjects were carried over from the TDF1 study and were never screened for the TDF2 analysis. So, the “Screened and Enrolled” population for the TDF2 study consists of the original 1219 minus the 18 carryover subjects (n=1201). The analytic dataset for this study was constructed using a different set of inclusion criteria. We included the carryover participants. Our analysis, however, excludes 3 subjects because they were diagnosed HIV+ by blood test at enrollment. Also, 16 subjects were excluded because they never received their medication. So there were 1200 subjects included in the actual analysis (the original 1219 minus the 19 who were deemed ineligible).
Condomless Sex Acts
Twenty-four participants (2.0%) reported no sex acts during the study period and were, therefore, not included in this analysis. Of the remaining 1176 subjects, 15 (1.3%) always reported no condomless sex acts (i.e., consistent condom use). Six hundred thirty-three (53.8%) reported both condomless sex acts as well as sex with a condom (i.e., inconsistent condom users) and 528 (44.9%) never reported using a condom (i.e., non-condom users). The number of reported condomless sex acts in a given 30-day period ranged from 0–168 acts.
The percent or rate of reported sex acts with 100% condom use in the last 30 days showed no apparent change over time (Fig. 1a), however the odds of reporting no condomless sex acts in the past 30 days increased approximately 16% per year (p=0.0042) (Supplemental Table 2). The percent or rate of reported condomless sex acts among those with 1 or more reported condomless sex acts in the past 30 days showed no significant change over time (p=0.3683)(Supplemental Table 2) (Fig. 1b).
Figure 1.
Change in the occurrence and frequency of sexual risk behavior over time. TDF2 clinical trial, Botswana 2007–2010.
Number of sex partners
Of the 1200 subjects, 20 (1.7%) reported no sex partners throughout the entire study. Seven hundred and two participants (58.5%) reported having 1 or more sex partners during each 30-day window of observation. In addition, 478 participants (39.8%) reported having 1 or more sex partners during at least one but not all 30-day periods. The number of reported sex partners in a given 30-day period ranged from 0–17 partners (mean=1.0; SD=0.76 median=1.0; IQR=1.0–1.0).
The percent of participants who reported no sex partners in the last 30 days increased (Fig. 1c). In addition, the odds of reporting no sex partners in the past 30 days increased approximately 40% per year (p≤0.0001) (Supplemental Table 3). For those who reported at least 1 sex partner during a time period, the rate of reported sex partners decreased approximately 25% per year (p≤0.0001) (Supplemental Table 3) (Fig. 1d).
Multivariable Analysis: Factors associated with reporting no condomless sex acts, and for no sex partners in past 30 days
The adjusted odds of reporting no condomless sex acts significantly increased by approximately 23% per year (adjusted odds ratio (aOR)=1.23, 1.10–1.37). Odds of no condomless sex were significantly lower for participants whose sexual debut in years was ≤15 years of age (aOR=0.37, CI=0.23–0.61) and 16–17 (aOR=0.49, CI=0.32–0.72) compared to ≥20 and for participants who were diagnosed with an STI during the study (aOR=0.43, CI=0.27–0.69). There was no significant effect of participant perception that they were receiving TDF-2 compared to participant perception that they were receiving a placebo (aOR=1.05, CI=0.77, 1.44) (Table 2).
Table 2.
Multivariable risk factors for condomless sex acts in the past 30 days. TDF2 clinical trial, Botswana 2007–2010.
| Demographic Characteristic or Risk Behavior |
No condomless sex acts | ≥1 condomless sex acts | ||||
|---|---|---|---|---|---|---|
| Odds Ratio |
95% CI | p-value | RateRatio | 95% CI | p-value | |
| Time (per year) | 1.23 | 1.10, 1.37 | .0002 | 0.98 | 0.89, 1.11 | .6910 |
| Treatment Arm Perception |
||||||
| TDF | 1.05 | 0.77, 1.44 | .7422 | 0.95 | 0.73, 1.24 | .7135 |
| Not Sure | 1.06 | 0.80, 1.38 | .6967 | 0.89 | 0.70, 1.11 | .2954 |
| Placebo | REF | REF | ||||
| Sexual Debut | ||||||
| 15 or less | 0.37 | 0.23, 0.61 | <.0001 | 1.65 | 1.14, 2.38 | .0081 |
| 16–17 | 0.49 | 0.32, 0.72 | .0005 | 1.68 | 1.22, 2.31 | .0016 |
| 18–19 | 0.72 | 0.50, 1.04 | .0857 | 1.34 | 1.00, 1.81 | .0551 |
| 20+ | REF | REF | ||||
| Gender | ||||||
| Female | -- | -- | -- | REF | ||
| Male | -- | -- | -- | 1.34 | 1.07, 1.67 | .0100 |
| STI | ||||||
| Yes | 0.43 | 0.27, 0.69 | .0003 | -- | -- | -- |
| No | REF | -- | -- | -- | ||
| HSV-2 | ||||||
| Yes | -- | -- | -- | 1.32 | 1.06, 1.66 | .0147 |
| No | -- | -- | -- | REF | ||
Note: “—“ indicates the variable was not included in that component of the final multivariable model.
The adjusted odds of reporting no sex partners significantly increased approximately 2% per year (aOR=1.02, CI=1.01–1.03). Odds of reporting no sex partners were significantly lower for participants who were married (aOR=0.33, CI=0.17–0.62), employed (OR=0.76, CI=0.58–1.00), and participants who tested HSV-2 positive during the study (aOR=0.58, CI=0.44–0.76). There was no significant effect of participant perception that they were receiving TDF-2 compared to participant perception that they were receiving a placebo (aOR=0.92, CI=0.68, 1.22) (Table 3).
Table 3.
Multivariable risk factors for number of sex partners in the past 30 days. TDF2 clinical trial, Botswana 2007–2010.
| Characteristic | No sex partners | 1 or more sex partners | ||||
|---|---|---|---|---|---|---|
| Odds Ratio |
95% CI | p-value | Rate Ratio |
95% CI | p-value | |
| Time (per year) | 1.02 | 1.01, 1.03 | <.0001 | .97 | .97, .98 | <.0001 |
| Treatment Arm Perception |
||||||
| TDF | 0.92 | 0.68, 1.22 | .5476 | 1.04 | 0.84, 1.29 | .7328 |
| Not Sure | 0.94 | 0.73, 1.20 | .6008 | 1.02 | 0.85, 1.23 | .8357 |
| Placebo | REF | REF | ||||
| Marital Status | ||||||
| Single | REF | -- | -- | -- | ||
| Married | 0.33 | 0.17, 0.62 | .0006 | -- | -- | -- |
| Age | ||||||
| 18–20 | -- | -- | -- | REF | ||
| 21–29 | -- | -- | -- | 3.12 | 1.18, 8.24 | .0224 |
| 30+ | -- | -- | -- | 2.14 | 0.75, 6.12 | .1568 |
| Occupational Status | ||||||
| Employed | 0.76 | 0.58, 1.00 | .0467 | -- | -- | -- |
| Unemployed | REF | -- | -- | -- | ||
| Sexual Debut | ||||||
| 15 or less | -- | -- | -- | 2.92 | 2.01, 4.22 | <.0001 |
| 16–17 | -- | -- | -- | 2.34 | 1.69, 3.24 | <.0001 |
| 18–19 | -- | -- | -- | 1.23 | 0.91, 1.67 | .1834 |
| 20+ | -- | -- | -- | REF | ||
| Gender | ||||||
| Female | -- | -- | -- | REF | ||
| Male | -- | -- | -- | 3.67 | 2.86, 4.71 | <.0001 |
| STI | ||||||
| Yes | -- | -- | -- | 1.79 | 1.34, 2.41 | .0001 |
| No | -- | -- | -- | REF | ||
| HSV-2 | ||||||
| Yes | 0.58 | 0.44, 0.76 | .0001 | -- | -- | -- |
| No | REF | -- | -- | -- | ||
Note: “--“ indicates the variable was not included in that component of the final multivariable model.
Multivariable Analysis: Factors associated with reporting ≥1 condomless sex acts, and for ≥1 sex partners in past 30 days
For those who reported ≥1 condomless sex act during a time period, the rate of reported condomless sex acts was significantly greater for participants who tested positive for HSV-2 (aRR=1.32, CI=1.06–1.66), males (aRR=1.34, CI=1.07–1.67), and for participants whose age at sexual debut in years was ≤15 (aRR=1.65, CI=1.14–2.38) and 16–17 (aRR=1.68, CI=1.22–2.31) compared to ≥20 of age. There was no significant effect of participant perception that they were receiving TDF2 compared to participant perception that they were receiving a placebo (aRR=0.95, CI=0.73, 1.24) (Table 2).
For those reporting ≥1 sex partners during a time period, the rate of reported sex partners significantly decreased by approximately 3% per year (aRR=0.97, CI=0.97–0.98). There was a significantly increased rate of reported sex partners for those whose age at sexual debut in years was ≤15 (aRR=2.92, CI=2.01–4.22) or 16–17 (aRR=2.34, CI=1.69–3.24) compared to ≥20 and among males (aRR=3.67, CI=2.86–4.71). The rate of reported sex partners was also significantly greater for participants 21–29 years of age (aRR=3.12, CI=1.18–8.24) compared to 18–20 years of age and for participants diagnosed with an STI during the course of the study (aRR=1.79 CI=1.34–2.41). There was no significant effect of participant perception that they were receiving TDF-2 compared to participant perception that they were receiving a placebo (aRR=1.04, CI=0.84, 1.29) (Table 3).
STI diagnosis and Pregnancy
Factors associated with being diagnosed with an STI or becoming pregnant during the course of the study
During the course of the study, of participants enrolled 155 of 1200, 12.9%, were diagnosed with an STI, while 96 of the 544 women, 17.6%, became pregnant. Although nearly all participants reported condomless sex at some point during the study, of the 155 persons diagnosed with an STI, 109 (70.3%) reported no condomless sex acts during the month immediately preceding the STI diagnosis. Also, of the 96 women who became pregnant during the study, only 1 (1.0%) reported consistent condom use. As noted previously, only 15 participants (8 females and 7 males) reported consistently using condoms throughout the entire study. One of these 15 participants was diagnosed with an STI and 1 of the female participants became pregnant during the study. No variables, including perception of treatment arm assignment, were significantly associated with pregnancy or STI diagnosis during the study.
Discussion
Using our measures, risk compensation was not observed in our study of PrEP to prevent HIV infection among heterosexual participants living in Botswana. While some studies have provided data suggesting that risk compensation is possible [17, 18], our findings concur with the findings of other HIV prevention interventions such as male circumcision [19], vaccines [20], and microbicides [21] as well as other PrEP trials [2, 3, 14, 22–24], including the open label extension of iPrEX [25]. Not only was there no evidence of risk compensation in our study, there was, in fact, a trend toward less reported sexual risk over time. For those reporting no condomless sex in the last 30 days, the absence of condomless sex continued over the trial and the odds increased 16% per year. Although the odds increased 16% it equates to approximately a 3% increase in the 100% condom use. In addition, for those reporting no sex partners in the past 30 days, the odds of continued reporting of no sex partners increased 40% per year. There are several potential reasons for the reduction in risk behavior. First, participants in controlled trials may join the trial to obtain help in reducing their risk behaviors because they have already made a decision to change their behavior to avoid HIV infection. For instance, in the VAX004 HIV vaccine trial, 56% of participants reported joining the trial to reduce their risk behavior [26]. Second, it may be that the risk-reduction counselling and testing, as well as provision of condoms, a key part of our study, was effective in motivating participants to reduce their high-risk sexual behavior. Third, it may be that the participants did not know to which treatment arm they were assigned nor even if the drug would protect them. In fact, the message given to the participants from study staff was that assignment to treatment arm and efficacy of the drug were unknown. There is also the possibility of social desirability bias [27]. Finally, there may be regression to the mean where extreme behaviors even out due to natural fluctuations [28]. In the context of this randomized controlled clinical trial, it is not possible to exclude any or assess the relative contributions of these factors.
Multivariable analyses showed that younger age at sexual debut, male gender, and other sexually transmitted infections were associated with more HIV risk. Young age at sexual debut is an important risk factor for HIV [29, 30]. One way to mitigate early sexual debut may be to reduce school drop-out [31–33]. Education is postulated to have an effect in reducing HIV infection not only through exposure to HIV prevention and health information, but more importantly, through the knowledge and skills obtained through a basic education [34, 35]. These skills help individuals to find and use information, translate knowledge into behavior change, and plan for the future. In addition, it keeps boys and girls in school, exposed to same age mates, and occupied with studying [36]. Male gender was associated with greater HIV risk in our study. This finding seems contradictory given that HIV prevalence in Botswana is 3 times as high among young females as among young males (15.3 percent of young women and 5.1 percent of young men living with HIV in 2007). However, the qualitative portion of a PrEP study carried out in Ghana revealed multiple reasons why women reduced their high-risk sexual behaviors in that study and may have relevance to our study results. Two of the reasons were: 1) because the study gave them access to condoms and 2) knowing that they were negative reduced their number of sex partners [37]. STIs, including HSV-2, have long been associated with persons engaging in high-risk sexual behaviors and with increased risk for HIV infection [38].
Our study results must be interpreted in the context of several potential limitations. First, there may have been attrition bias as a result of higher than expected participant withdrawal from the study due to relocation or other obligations [1], though because we do not believe that this attrition occurred differentially among participants who were at different levels of treatment perception and degrees of change in risk behavior over time, this did not likely impact our findings. Second, our results indicating no risk compensation may not be predictive of a person’s risk perception and their risk behaviors in the real world outside the study environment. It has been suggested that future studies of risk compensation should not only assess perception of treatment condition (which we did assess) but also perception of efficacy of whatever pill the participant was taking (regardless of whether the participant thought it was the placebo or active drug), which we did not assess [39]. Third, while there was a qualitative component to our study, a question as to specific reasons participants modified their sexual behavior was not asked. Finally, increasing safer sexual behaviors over the course of the study could be due to dropout of higher-risk individuals [1] or social desirability bias in reporting [27]. It is of note that the large difference in the 40% unadjusted and the 2% adjusted increase in odds of no sex partners in last 30 days over time indicates the strong influence of other factors such as being married, where the spouse could be living in a separate location, and being HSV-2 positive, where the infected person may withdraw from having sex partners.
In conclusion, our study of PrEP to prevent HIV infection showed no significant increase in sexual risk behavior among heterosexual participants in Botswana. In fact, the proportion of participants who always reported using a condom and having no sex partners increased over time. Risk-reduction counselling, regular HIV testing and provision of condoms are the standard of care in HIV prevention and HIV prevention research, and likely contributed to the increase in reported safe sexual practices in the study. This increase is consistent with results from other studies of PrEP [2, 4, 6, 14, 22, 23]. In this regard, it is important to note that certain groups may respond differently to counseling [37]. Although there was no risk compensation in any of the PrEP trials to date or the iPrEx open label extension, it does not mean that perception of HIV risk could not change when there is a more widespread use of PrEP [40], similar to the perception by MSM outside a study environment that greater risks could be taken in the age of HAART [41]. PrEP will work best when it is part of a toolkit that includes behavioral interventions supported by structural interventions [42] that address poverty [43, 44], including food insecurity [45, 46] and general education [47–51], as well as the integration of health services (e.g. drug use, mental health, and primary care) [52].
Supplementary Material
Acknowledgments
We are grateful to all of the participants of the TDF2 study and the study staff. This study was supported by the Division of HIV/AIDS Prevention, Centers for Disease Control and Prevention, and the Division of AIDS, National Institutes of Health.
Footnotes
Disclaimer: The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention. Use of trade names is for identification purposes only and does not constitute endorsement by the Centers for Disease Control and Prevention or the Department of Health and Human Services.
References
- 1.Thigpen MC, Kebaabetswe PM, Paxton LA, Smith DK, Rose CE, Segolodi TM, et al. Antiretroviral preexposure prophylaxis for heterosexual HIV transmission in Botswana. N Engl J Med. 2012;367:423–434. doi: 10.1056/NEJMoa1110711. [DOI] [PubMed] [Google Scholar]
- 2.Grant RM, Lama JR, Anderson PL, McMahan V, Liu AY, Vargas L, et al. Preexposure chemoprophylaxis for HIV prevention in men who have sex with men. N Engl J Med. 2010;363:2587–2599. doi: 10.1056/NEJMoa1011205. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Anderson PL, Glidden DV, Liu A, Buchbinder S, Lama JR, Guanira JV, et al. Emtricitabine-tenofovir concentrations and pre-exposure prophylaxis efficacy in men who have sex with men. Sci Transl Med. 2012;4:151ra125. doi: 10.1126/scitranslmed.3004006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Baeten JM, Donnell D, Ndase P, Mugo NR, Campbell JD, Wangisi J, et al. Antiretroviral prophylaxis for HIV prevention in heterosexual men and women. N Engl J Med. 2012;367:399–410. doi: 10.1056/NEJMoa1108524. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Choopanya K, Martin M, Suntharasamai P, Sangkum U, Mock PA, Leethochawalit M, et al. Antiretroviral prophylaxis for HIV infection in injecting drug users in Bangkok, Thailand (the Bangkok Tenofovir Study): a randomised, double-blind, placebo-controlled phase 3 trial. Lancet. 2013;381:2083–2090. doi: 10.1016/S0140-6736(13)61127-7. [DOI] [PubMed] [Google Scholar]
- 6.Van Damme L, Corneli A, Ahmed K, Agot K, Lombaard J, Kapiga S, et al. Preexposure prophylaxis for HIV infection among African women. N Engl J Med. 2012;367:411–422. doi: 10.1056/NEJMoa1202614. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Marrazzo J, Ramjee G, Nair G, Palanee T, Mkhize B, Nakabiito Taljaard M, Piper J, Gomez Feliciano K, Chirenje M. Pre-exposure prophylaxis for HIV in women: daily oral tenofovir, oral tenofovir/emtricitabine or vaginal tenofovir gel in the VOICE study (MTN 003). 20th Conference on Retroviruses and Opportunistic Infections; March 3–6, 2013; Atlanta, GA. 2013. Abstract 26LB. [Google Scholar]
- 8.Marrazzo JM, Ramjee G, Richardson BA, Gomez K, Mgodi N, Nair G, et al. Tenofovir-based preexposure prophylaxis for HIV infection among African women. N Engl J Med. 2015;372:509–518. doi: 10.1056/NEJMoa1402269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Hendrix CW. Exploring concentration response in HIV pre-exposure prophylaxis to optimize clinical care and trial design. Cell. 2013;155:515–518. doi: 10.1016/j.cell.2013.09.030. [DOI] [PubMed] [Google Scholar]
- 10.Amico KR. Adherence to preexposure chemoprophylaxis: the behavioral bridge from efficacy to effectiveness. Curr Opin HIV AIDS. 2012;7:542–548. doi: 10.1097/COH.0b013e3283582d4a. [DOI] [PubMed] [Google Scholar]
- 11.Amico KR. Behavioral threats to PrEP success. Presented at the TasP PrEP evidence summit 2012 (International Association of Physicians in AIDS Care and British HIV Association); June 11–12, 2012; London, UK. 2012. [Google Scholar]
- 12.Wilde G. Target Risk: Dealing with the danger of death, disease, and damage in everyday decisions. Ontario, Canada: PDE Publications; 1994. [Google Scholar]
- 13.Eaton LA, Kalichman S. Risk compensation in HIV prevention: implications for vaccines, microbicides, and other biomedical HIV prevention technologies. Curr HIV/AIDS Rep. 2007;4:165–172. doi: 10.1007/s11904-007-0024-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Peterson L, Taylor D, Roddy R, Belai G, Phillips P, Nanda K, et al. Tenofovir disoproxil fumarate for prevention of HIV infection in women: a phase 2, double-blind, randomized, placebo-controlled trial. PLoS Clin Trials. 2007;2:e27. doi: 10.1371/journal.pctr.0020027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hu MC, Pavlicova M, Nunes EV. Zero-inflated and hurdle models of count data with extra zeros: examples from an HIV-risk reduction intervention trial. Am J Drug Alcohol Abuse. 2011;37:367–375. doi: 10.3109/00952990.2011.597280. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Saberi P, Johnson MO, McCulloch CE, Vittinghoff E, Neilands TB. Medication adherence: tailoring the analysis to the data. AIDS Behav. 2011;15:1447–1453. doi: 10.1007/s10461-011-9951-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Chesney MA, Chambers DB, Kahn JO. Risk behavior for HIV infection in participants in preventive HIV vaccine trials: a cautionary note. J Acquir Immune Defic Syndr Hum Retrovirol. 1997;16:266–271. doi: 10.1097/00042560-199712010-00007. [DOI] [PubMed] [Google Scholar]
- 18.Grund JM, Hennink MM. A qualitative study of sexual behavior change and risk compensation following adult male circumcision in urban Swaziland. AIDS Care. 2012;24:245–251. doi: 10.1080/09540121.2011.596516. [DOI] [PubMed] [Google Scholar]
- 19.Mattson CL, Campbell RT, Bailey RC, Agot K, Ndinya-Achola JO, Moses S. Risk compensation is not associated with male circumcision in Kisumu, Kenya: a multi-faceted assessment of men enrolled in a randomized controlled trial. PLoS One. 2008;3:e2443. doi: 10.1371/journal.pone.0002443. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Bartholow BN, Buchbinder S, Celum C, Goli V, Koblin B, Para M, et al. HIV sexual risk behavior over 36 months of follow-up in the world's first HIV vaccine efficacy trial. J Acquir Immune Defic Syndr. 2005;39:90–101. doi: 10.1097/01.qai.0000143600.41363.78. [DOI] [PubMed] [Google Scholar]
- 21.Foss AM, Vickerman PT, Heise L, Watts CH. Shifts in condom use following microbicide introduction: should we be concerned? AIDS. 2003;17:1227–1237. doi: 10.1097/00002030-200305230-00015. [DOI] [PubMed] [Google Scholar]
- 22.Marcus JL, Glidden DV, Mayer KH, Liu AY, Buchbinder SP, Amico KR, et al. No Evidence of Sexual Risk Compensation in the iPrEx Trial of Daily Oral HIV Preexposure Prophylaxis. PLoS One. 2013;8:e81997. doi: 10.1371/journal.pone.0081997. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Liu AY, Vittinghoff E, Chillag K, Mayer K, Thompson M, Grohskopf L, et al. Sexual risk behavior among HIV-uninfected men who have sex with men participating in a tenofovir preexposure prophylaxis randomized trial in the United States. J Acquir Immune Defic Syndr. 2013;64:87–94. doi: 10.1097/QAI.0b013e31828f097a. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Campo R, DeJesus E, Bredeek UF, Henry K, Khanlou H, Logue K, et al. SWIFT: prospective 48-week study to evaluate efficacy and safety of switching to emtricitabine/tenofovir from lamivudine/abacavir in virologically suppressed HIV-1 infected patients on a boosted protease inhibitor containing antiretroviral regimen. Clin Infect Dis. 2013;56:1637–1645. doi: 10.1093/cid/cis1203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Grant R, Guanira J, Buchbinder S, Mosquera C. No excess in HIV incidence after stopping oral emtricitabine/tenofovir pre-exposure prophylaxis: The iPrEx Trial. 20th Conference on Retroviruses and Opportunistic Infections; March 3–6, 2013; Atlanta. 2013. Abstract #27. [Google Scholar]
- 26.Colfax G, Buchbinder S, Vamshidar G, Celum C, McKirnan D, Neidig J, et al. Motivations for participating in an HIV vaccine efficacy trial. J Acquir Immune Defic Syndr. 2005;39:359–364. doi: 10.1097/01.qai.0000152039.88422.ec. [DOI] [PubMed] [Google Scholar]
- 27.Paulhus D. Two-component models of socially desirable responding. Journal of Personality and Social Psychology. 1984;46:598–609. [Google Scholar]
- 28.Barnett AG, van der Pols JC, Dobson AJ. Regression to the mean: what it is and how to deal with it. International Journal of Epidemiology. 2005;34:215–220. doi: 10.1093/ije/dyh299. [DOI] [PubMed] [Google Scholar]
- 29.Asiimwe-Okiror G, Opio AA, Musinguzi J, Madraa E, Tembo G, Carael M. Change in sexual behaviour and decline in HIV infection among young pregnant women in urban Uganda. AIDS. 1997;11:1757–1763. doi: 10.1097/00002030-199714000-00013. [DOI] [PubMed] [Google Scholar]
- 30.Pettifor AE, van der Straten A, Dunbar MS, Shiboski SC, Padian NS. Early age of first sex: a risk factor for HIV infection among women in Zimbabwe. AIDS. 2004;18:1435–1442. doi: 10.1097/01.aids.0000131338.61042.b8. [DOI] [PubMed] [Google Scholar]
- 31.Stockl H, Kalra N, Jacobi J, Watts C. Is early sexual debut a risk factor for HIV infection among women in sub-Saharan Africa? A systematic review. Am J Reprod Immunol. 2013;69(Suppl 1):27–40. doi: 10.1111/aji.12043. [DOI] [PubMed] [Google Scholar]
- 32.Jukes M, Simmons S, Bundy D. Education and vulnerability: the role of schools in protecting young women and girls from HIV in southern Africa. AIDS. 2008;22(Suppl 4):S41–S56. doi: 10.1097/01.aids.0000341776.71253.04. [DOI] [PubMed] [Google Scholar]
- 33.Ghebremichael M, Larsen U, Paintsil E. Association of age at first sex with HIV-1, HSV-2, and other sexual transmitted infections among women in northern Tanzania. Sex Transm Dis. 2009;36:570–576. doi: 10.1097/OLQ.0b013e3181a866b8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Vandermoortele JDE. The "Education Vaccine" against HIV. Current Issues in Comparative Education. 2000;3:6–13. [Google Scholar]
- 35.Coombe CKM. Education as a vehicle for combating HIV/AIDS. Prospects. 2001;XXXI:435–445. [Google Scholar]
- 36.Global_Campaign_for_Education. Learning to survive: How education for all would save millions of young people from HIV/AIDS. [Accessed 12 March 2014];2004 http://www.campaignforeducation.org/docs/reports/arch/learn.pdf. [Google Scholar]
- 37.Guest G, Shattuck D, Johnson L, Akumatey B, Clarke EE, Chen PL, et al. Changes in sexual risk behavior among participants in a PrEP HIV prevention trial. Sex Transm Dis. 2008;35:1002–1008. [PubMed] [Google Scholar]
- 38.Hayes R, Watson-Jones D, Celum C, van de Wijgert J, Wasserheit J. Treatment of sexually transmitted infections for HIV prevention: end of the road or new beginning? AIDS. 2010;24(Suppl 4):S15–S26. doi: 10.1097/01.aids.0000390704.35642.47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Underhill K. Study designs for identifying risk compensation behavior among users of biomedical HIV prevention technologies: balancing methodological rigor and research ethics. Soc Sci Med. 2013;94:115–123. doi: 10.1016/j.socscimed.2013.03.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.De Man J, Colebunders R, Florence E, Laga M, Kenyon C. What is the place of pre-exposure prophylaxis in HIV prevention? AIDS Rev. 2013;15:102–111. [PubMed] [Google Scholar]
- 41.Crepaz N, Hart TA, Marks G. Highly active antiretroviral therapy and sexual risk behavior: a meta-analytic review. JAMA. 2004;292:224–236. doi: 10.1001/jama.292.2.224. [DOI] [PubMed] [Google Scholar]
- 42.Roberts ET, Matthews DD. HIV and chemoprophylaxis, the importance of considering social structures alongside biomedical and behavioral intervention. Soc Sci Med. 2012;75:1555–1561. doi: 10.1016/j.socscimed.2012.02.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Handa S, Halpern CT, Pettifor A, Thirumurthy H. The Government of Kenya's Cash Transfer Program Reduces the Risk of Sexual Debut among Young People Age 15–25. PLoS One. 2014;9:e85473. doi: 10.1371/journal.pone.0085473. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Pettifor A, MacPhail C, Nguyen N, Rosenberg M. Can money prevent the spread of HIV? A review of cash payments for HIV prevention. AIDS Behav. 2012;16:1729–1738. doi: 10.1007/s10461-012-0240-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Weiser SD, Young SL, Cohen CR, Kushel MB, Tsai AC, Tien PC, et al. Conceptual framework for understanding the bidirectional links between food insecurity and HIV/AIDS. Am J Clin Nutr. 2011;94:1729S–1739S. doi: 10.3945/ajcn.111.012070. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Miller CL, Bangsberg DR, Tuller DM, Senkungu J, Kawuma A, Frongillo EA, et al. Food insecurity and sexual risk in an HIV endemic community in Uganda. AIDS Behav. 2011;15:1512–1519. doi: 10.1007/s10461-010-9693-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Hargreaves JR, Glynn JR. Educational attainment and HIV-1 infection in developing countries: a systematic review. Trop Med Int Health. 2002;7:489–498. doi: 10.1046/j.1365-3156.2002.00889.x. [DOI] [PubMed] [Google Scholar]
- 48.Hargreaves JR, Bonell CP, Boler T, Boccia D, Birdthistle I, Fletcher A, et al. Systematic review exploring time trends in the association between educational attainment and risk of HIV infection in sub-Saharan Africa. AIDS. 2008;22:403–414. doi: 10.1097/QAD.0b013e3282f2aac3. [DOI] [PubMed] [Google Scholar]
- 49.Baird SJ, Garfein RS, McIntosh CT, Ozler B. Effect of a cash transfer programme for schooling on prevalence of HIV and herpes simplex type 2 in Malawi: a cluster randomised trial. Lancet. 2012;379:1320–1329. doi: 10.1016/S0140-6736(11)61709-1. [DOI] [PubMed] [Google Scholar]
- 50.Cho H, Hallfors DD, Mbai II, Itindi J, Milimo BW, Halpern CT, et al. Keeping adolescent orphans in school to prevent human immunodeficiency virus infection: evidence from a randomized controlled trial in Kenya. J Adolesc Health. 2011;48:523–526. doi: 10.1016/j.jadohealth.2010.08.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Adimora AA, Schoenbach VJ, Martinson FE, Coyne-Beasley T, Doherty I, Stancil TR, et al. Heterosexually transmitted HIV infection among African Americans in North Carolina. J Acquir Immune Defic Syndr. 2006;41:616–623. doi: 10.1097/01.qai.0000191382.62070.a5. [DOI] [PubMed] [Google Scholar]
- 52.Rotheram-Borus MJ, Swendeman D, Chovnick G. The past, present, and future of HIV prevention: integrating behavioral, biomedical, and structural intervention strategies for the next generation of HIV prevention. Annu Rev Clin Psychol. 2009;5:143–167. doi: 10.1146/annurev.clinpsy.032408.153530. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.