HIV Risk and Prevention Estimates

The choice of estimate was prioritized based on the following criteria:

• Only evidence based on peer-reviewed published reports was considered when determining estimates. Unpublished data, including conference abstracts, were not considered to be as reliable as peer-reviewed publications because results may change as more data become available and data are re-analyzed or methods adjusted based on peer-review feedback. Additionally, the amount of information available for unpublished studies does not always allow us to adequately assess methods and quality of data and analysis. Non-peer reviewed publications may be used for supporting evidence.
• Only evidence regarding HIV transmission (e.g., HIV outcomes) was considered. Data for non-HIV outcomes (e.g., pregnancy prevention, STI prevention) were considered not to be good proxies for HIV transmission because modeling or other methods that require complex assumptions would be required to equate proxies with HIV transmission rates and introduce additional uncertainty.
• For the consensus estimates, a hierarchy was established for prioritizing the type of estimate to select.
• The greatest priority was given to estimates based on "optimal use" of the strategy or intervention; that is, if the strategy or intervention was used as intended to achieve maximum protection. In most cases, this means evidence from restrictive or subset analyses from larger studies was prioritized if those findings were restricted to those participants with optimal adherence to the strategy or intervention.
• "Optimal use" is typically determined by measuring level of adherence or exposure to the strategy or intervention (e.g., condom; PrEP); or it's determined by measuring its intended virologic response (e.g., viral suppression is a result of effective ART). Either way, priority was given to the most objective measure available for determining "optimal use" (e.g., maximum drug level in plasma).
• In some cases, the most objective measure for determining "optimal use" may not be ideal but may be the most relevant objective measure available (e.g. pharmacy data; directly observed therapy).
• If an objective measure for "optimal use" was not available, then we chose the best subjective measure available (e.g., self-report) based on assessing maximum adherence or "optimal use" (e.g., consistent use or always using) recognizing that self-report may overestimate actual use.
• Other findings reflecting effectiveness of the intervention, if used recently or at all, were not used for selecting the most appropriate effectiveness estimate for "optimal use". However, these other findings were summarized when available to provide a broader context of the evidence available for each strategy or intervention. In addition, findings from intent-to-treat, or modified intent-to-treat, (mITT/ITT) analyses from RCTs, comparing groups as assigned regardless of exposure or use were also summarized for broader context. These findings tend to help illustrate the extent to which the intervention effectiveness depends on adherence.
• An estimate from a published meta-analysis was used if available and relevant for the strategy/risk factor in question; otherwise, the most appropriate estimate or range of estimates from relevant studies were used.

Evidence supporting effectiveness estimates

• Effectiveness estimates based on suppressive ART ("Optimal Use" of ART) as indicated by achieving and maintaining viral suppression:
  • Optimal use of ART is defined as taking ART daily as prescribed and achieving and maintaining a suppressed viral load (or viral suppression).
  • Four key studies provide evidence for the effectiveness of ART, when used optimally, on preventing the sexual transmission of HIV. These studies – HPTN052 (Cohen, 2016), PARTNER (Rodger, 2016), Opposites Attract (Bavinton, 2018), and PARTNER2 (Rodger, 2018) – observed zero linked sexual transmissions among HIV-discordant couples with viral suppression.
    • Each of these studies followed HIV-discordant couples while the partners with HIV were treated with ART with the intent of suppressing HIV replication. The follow-up assessments, at frequencies typical of what experts recommend for clinical care, included regular measurement of plasma HIV RNA concentrations and HIV testing of the partner without HIV. In each study, new HIV infections in the uninfected partners were assessed phylogenetically to determine whether they were genetically linked to their partner with HIV in the study.
    • The HPTN052 study (Cohen, 2016) followed 1,763 HIV-discordant couples (97% heterosexual; 3% MSM) for a median of 5.5 years. Zero genetically linked transmissions were observed while the partner with HIV was virally suppressed, defined as <400 copies/mL of plasma, resulting in a transmission rate estimate of 0.00 per 100 couple-years and an effectiveness estimate of 100%, if calculated (not reported in study). The confidence intervals for the effectiveness and transmission rate estimates were not reported and could not be calculated from data reported. The authors reported six partner infections that occurred during the study period where linkage could not be determined due to the inability to amplify HIV RNA; these infections were excluded from all analyses. Although linked infection could not be definitively ruled out, epidemiologic investigation strongly suggested most were not linked (Eshleman, 2017). Reported condom use was high (93%) among couples (Cohen, 2011) and likely contributed to the observed reduction in HIV transmission risk.
    • The PARTNER study (Rodger, 2016) followed 1,166 HIV-discordant couples (62% heterosexual; 38% MSM) for a median of 1.3 years while the partner with HIV was treated with ART and virally suppressed at baseline. During the 1,238 couple-years of follow-up time included in the analysis, where nearly 900 couples engaged in over 58,000 condomless sex acts, the partner without HIV did not use PrEP or PEP, and the partner with HIV was virally suppressed, defined as VL <200 copies/mL of plasma, zero genetically linked transmissions were observed. The resulting transmission rate estimate per 100 couple-years was 0.00, with a 95% confidence interval (CI) = (0.00, 0.30). The upper 95% confidence limit varied by risk group and sexual behavior due to the range of couple-years observed across the subgroups. For example, the estimate for the sexual transmission rate of HIV among discordant couples while the partner with HIV was virally suppressed was:
      • 0.00 (0.0 – 0.46) per 100 couple-years during any condomless sex among heterosexual men and women
      • 0.00 (0.0 – 0.89) per 100 couple-years during condomless anal sex among MSM
    • The Opposites Attract study (Bavinton, 2018) followed 343 HIV-discordant male-male couples for a median of 1.7 years while the partner with HIV was treated with ART, with most taking ART at baseline (80%). During the 232 couple-years of follow-up time included in the analysis, where the partner with HIV was virally suppressed (defined as <200 copies/mLof plasma) and couples reported over 12,000 episodes of any condomless anal sex acts and no PrEP use, there were zero genetically linked transmissions observed. This translates to a transmission rate estimate of:
      • 0.00 (0.00 – 1.59) per 100 couple-years during condomless anal sex among MSM
    • The PARTNER2 study (Rodger, 2019) was an extension of the PARTNER study that recruited more HIV-discordant male-male couples and extending the follow-up time for those enrolled in the PARTNER study, totaling 972 HIV-discordant male-male couples enrolled in PARTNER2. The final analysis included almost 800 couples followed for a median of 2.0 years. Over nearly 1,600 couple-years of follow-up while the partner with HIV was on ART and virally suppressed, defined as <200 copies/mL of plasma, and couples reported no PrEP use and over 76,000 episodes of condomless anal sex, zero genetically linked transmissions were observed. This translates to a transmission rate estimate of:
      • 0.00 (0.00 – 0.23) per 100 couple-years during condomless anal sex among MSM
  • Additional supporting evidence beyond the four individual studies includes:
    • Combining over 2,600 couple-years of follow-up and more than 125,000 episodes of sex without a condom or PrEP while the partner with HIV was virally suppressed, from the PARTNER, PARTNER2, and Opposites Attract studies, results in a combined HIV transmission risk estimate for condomless and PrEP-less sex among heterosexual or MSM couples of 0.00 (0.00 – 0.14) per 100 couple-years.
    • A recent review at the 2019 CROI conference combined the four studies above along with several previous observational studies, accumulating over 4,000 couple-years of follow-up, and reported a combined HIV transmission risk estimate while the person with HIV was virally suppressed, excluding unconfirmed viral loads, of 0.00 (0.00 – 0.07) per 100 couple-years (Vernazza, 2019).
    • No cases of linked HIV transmission to sexual partners when the person with HIV was virally suppressed have been documented.

• Earlier effectiveness estimates based on original RCT study:
  • Cohen (2011) was the first published RCT examining the protective benefits of ART for reducing HIV transmission. This paper reported the interim analysis of the HPTN 052 study, a randomized controlled trial (RCT) of providing early ART, compared with delayed ART, among 1,763 mostly heterosexual, serodiscordant couples followed for a median of 1.7 years. The effectiveness estimate for ART was 96%, based on the ITT results using verified linked cases of HIV.
    • Typically, findings from the primary analysis within an RCT include many participants assigned to the intervention strategy but not necessarily using the strategy. In this study, however, most participants in the "early ART" arm were taking ART consistently as evidenced by a high level of adherence to ART (79% had at least 95% adherence via pill count) and a high rate of viral suppression (89% were virally suppressed by 3 months). Given that this ITT analysis included time periods where the person with HIV was not taking ART or not virally suppressed, this effectiveness estimate for consistent use of ART is not an accurate estimate for optimal use of ART, where the person with HIV would be taking ART as prescribed and would have achieved viral suppression.
    • The 96% effectiveness of taking early ART, as well as a significant reduction in morbidity and mortality among participants with HIV, led to ending the RCT and offering all couples ART. Cohen and colleagues have continued to follow participants from this original study and offer ART to participants in both arms (thereby turning the study from an RCT to an observational design, although they continue also to analyze participants per their original random assignment) (Cohen, 2016). By the end of the study, 96% of persons with HIV in the "delayed ART" arm had started ART. The final HPTN 052 study ITT effectiveness estimate, including more than 5 years of follow-up, was 93% comparing "early ART" vs "delayed ART" (Cohen, 2016). Given that essentially all participants in both arms has started ART by the end of the study, this finding is not a better estimate of the effectiveness of taking ART (versus not taking ART) on reducing HIV transmission.
    • Based on the HPTN 052 RCT (Cohen, 2011), the best estimate for the overall effectiveness of taking ART consistently among heterosexuals is 96%. There are no comparable RCTs for MSM or PWID.

Evidence supporting effectiveness estimates

• Effectiveness estimates based on "Optimal or Consistent Use" of oral daily PrEP.
  • The effectiveness of oral daily PrEP is highly dependent on PrEP adherence (Riddell, 2018). The effectiveness estimate of PrEP, when taken daily or consistently, is presented here. The effectiveness estimates of PrEP as assigned within a trial or when used recently are presented below.
  • When taking oral PrEP daily or consistently, it is extremely effective in preventing HIV and HIV acquisition is extremely rare. Only three cases of seroconversion have been confirmed to date worldwide, while individuals without HIV were on PrEP with verified adherence.
  • The US Preventive Services Task Force (USPSTF) provides a Grade A recommendation for oral daily PrEP in preventing HIV acquisition in persons at high risk. The USPSTF also concludes with high certainty that the benefit of oral PrEP is substantial, but that adherence to PrEP is central to maximizing its benefit (USPSTF, 2019).
  • MSM: Several studies evaluated the effectiveness of PrEP use among MSM. These studies vary in study design methods (e.g. RCT, observational) as well as how PrEP adherence is measured; but all provide evidence for the effectiveness of PrEP when taken daily or consistently.
    • iPrEx OLE Study (Grant, 2014). This open-label extension (OLE) cohort study enrolled 1,603 MSM and transgender women previously enrolled in three PrEP trials (ATN 082; iPrEx; and US Safety Study) and followed participants for 72 weeks. All were offered free daily oral PrEP (TDF/FTC or Truvada), and 1,225 elected to take PrEP. PrEP adherence was measured by drug concentration of TFV-DP in dried blood spots. No new HIV infections were observed among MSM taking PrEP where drug levels indicated they had taken 4 or more doses per week.
      • Among those with the highest drug concentrations indicating daily PrEP use, as verified by drug level of TFV-DP in dried blood spots of >1250 fmol/punch (equivalent to ~7 pills/week), there were no new HIV infections. This resulted in a risk reduction estimate of 100% when compared to the previous placebo group from the iPrEx trial or the concurrent group of participants not on PrEP.
      • In addition, among those with drug concentration levels indicating at least 4 pills/week (>700 fmol/punch), there were no new HIV infections, which resulted in a risk reduction estimate of 100% when compared to either comparison group.
    • DEMO Project (Liu, 2015). This open-label observational study enrolled 557 MSM and transgender women in 2 STI clinics and a community health center in 3 U.S. cities and offered free daily oral PrEP (TDF/FTC) for 48 weeks. PrEP adherence was measured by drug concentration of TFV-DP in dried blood spots in a large sample of participants at all follow-up visits. At the end of follow-up, 527 had at least 1 follow-up visit, providing a total of 481 person-years of follow-up. Most of the participants (ranging from 80% to 86% of participants across the follow-up visits) of those assessed for PrEP adherence had drug levels considered protective (consistent with >4 pills/week). At the end of the study, 2 participants acquired HIV infection; however, both participants had drug levels indicative of < 2 doses/week or BLQ (below the limit of quantification) throughout the study. This means no new HIV infections were observed among those with protective levels of PrEP use.
    • PROUD Study (McCormack, 2015). The PROUD study was a randomized-control trial (RCT) evaluating immediate daily oral PrEP (TDF/FTC) vs delayed PrEP among MSM patients without HIV in 13 clinics in England from 2012-2014. A total of 554 MSM were randomized, 275 to immediate PrEP and 269 to the delayed group. After an interim analysis, the trial stopped early and all deferred patients were offered PrEP. More than 90% of the patients in each group were retained at the end of the study, providing ~500 person-years of follow up. The mITT results from the trial are reported below. Although there were 3 new HIV infections among those assigned to the immediate PrEP group, there were no HIV infections observed among those actually taking PrEP. All 3 new HIV infections in the immediate PrEP group, based on clinical indications, attendance, and prescription info, were not taking PrEP near the time of seroconversion – 2 never started taking PrEP and 1 infection was identified over 40 weeks after last clinic visit (where 90 PrEP pills were provided).
    • Kaiser Permanente Observational Study (Volk, 2015; Marcus, 2017). This observational study followed 1,045 Kaiser Permanente (KP) patients, mostly MSM (98-99%), who were referred to a specialized PrEP program in KP San Francisco during 2012-2015, and then later extended through February 2017. PrEP use was measured based on pharmacy refill data. Among the 2,107 patients never starting PrEP, there were 22 new HIV infections. Among the 4,991 who started PrEP, although we don't know how many were always taking PrEP daily, there were no new HIV infections while PrEP prescriptions were filled (over 12.4 months; 5,104 person-years on PrEP). Of the 1,303 patients who stopped PrEP (prescription not re-filled), 11 new HIV infections were later observed after stopping PrEP, by the end of the follow-up.
    • In summary, the effectiveness of PrEP among MSM when used daily or consistently is estimated to be 100% in studies. However, a few cases of new HIV infections have been reported with PrEP verified adherence, indicating that the risk has not been completely eliminated and that the effectiveness of PrEP cannot be exactly 100%. Given the number of persons on PrEP worldwide (prepwatch.org), the risk reduction (or effectiveness of PrEP) would likely need to be very high and close to100% to observe only three confirmed cases of PrEP failure (new HIV infection despite taking PrEP daily or consistently) to date. To represent the protective value of PrEP while also acknowledging the small number of failures, we indicate the effectiveness of PrEP is about 99%.
  • Transgender women: The iPrEx OLE cohort study (Grant, 2014) enrolled mostly MSM, but included 175 transgender women previously enrolled in three PrEP trials (ATN 082; iPrEx; and US Safety Study) and offered free daily oral PrEP (TDF/FTC or Truvada) for 72 weeks. PrEP adherence was measured by drug concentration of TFV-DP in dried blood spots. One transgender woman seroconverted while receiving PrEP and one seroconversion occurred in a woman who elected not to use PrEP. No new HIV infections were observed among transgender women who were taking PrEP where drug levels indicated they had taken 4 or more doses per week. However, the iPrEx trial results described below show no benefit of PrEP among transgender women, likely due to low PrEP adherence (Deutsch, 2015).
  • Heterosexual men and women: There is no effectiveness estimate of PrEP when taken daily or consistently among heterosexuals. There is evidence for the effectiveness of PrEP when used recently, which is estimated to be 88 – 90%, as described below. These estimates come from subset analyses among heterosexual men or women with evidence of taking PrEP recently (based on detecting TFV in plasma). These subset analyses likely include people who vary in PrEP adherence, including those who used PrEP recently but not consistently, used PrEP consistently but not daily (e.g. ~4 times/week), or used PrEP daily. Given that the effectiveness of PrEP is highly dependent on PrEP adherence, the effectiveness of PrEP when taking PrEP daily or consistently is likely to be greater than when taking PrEP recently; therefore, likely to be greater than 90% and similar to what is observed for MSM. Data show that it takes longer (~13 days longer) to reach a maximum drug level of PrEP in vaginal tissue as compared to rectal tissue (CDC, 2018), but once maximum drug levels are reached, the effectiveness of PrEP in preventing acquisition during sex should be similar for vaginal or anal sex, and for men or women.
  • PWID: The Bangkok Tenofovir Study (BTS) (Choopanya, 2013) was an RCT evaluating oral daily PrEP use (TDF alone) against placebo among persons without HIV who inject drugs.
    • When taking PrEP (TDF) nearly daily, as verified by TFV detected in plasma and directly observed therapy (DOT) (with at least 70% of days were DOT, with no gaps of >2 days without DOT; equivalent to ~5 days/week), the risk of HIV acquisition was reduced by 74% among HIV-uninfected injecting drug users (subset analysis; BTS; Choopanya, 2013).
    • When taking PrEP (TDF) nearly daily, when defined as 97.5% adherence, based on daily diary (most often confirmed daily by DOT staff) and monthly pill count, the risk of HIV acquisition was reduced by about 84% (subset analysis; BTS; Martin, 2015). This study also showed a dose-response between adherence and protection from PrEP, with greater adherence resulting in a greater effectiveness estimate for PrEP.
    • This BTS study evaluated TDF (Tenofovir) rather than the combination drug TDF/FTC (Truvada). The effectiveness of two-drug oral therapy has not been assessed among PWID but may be higher than TDF alone. TDF alone had been shown to have a slightly lower efficacy than TDF/FTC, although not statistically different, among heterosexual HIV-discordant couples in the Partners PrEP study (Baeten, 2012; Baeten, 2014). In addition, since the measures used in the BTS study for assessing PrEP adherence included those taking PrEP nearly daily but not daily, the effectiveness of daily PrEP use may in fact be greater.
    • Note that TDF (Tenofovir) is recommended in the U.S. as an alternative to TDF/FTC (Truvada) among PWID.

• Effectiveness estimates based on "Recent Use" of oral daily PrEP.
  • Recent use of oral PrEP is measured based on drug detected, typically detecting FTC or TFV, in plasma. All effectiveness estimates presented here come from subset analyses within larger RCTs restricting to participants with drug detected in plasma indicating recent use of PrEP. These estimates do not reflect optimal or consistent use of PrEP, which resulted in greater effectiveness estimates among MSM and PWID as described above.
  • MSM: The iPrEx Trial (Grant, 2010) was an RCT evaluating oral daily PrEP use (TDF/FTC) against placebo among MSM. The findings from a case/control sub-analysis show that effectiveness of PrEP, when recently used, was estimated to be 92%. This measure of recent use of PrEP was based on detecting FTC or TFV in plasma or detecting FTC-TP or TFV-DP in PBMC.
  • Heterosexual men and women: The Partners PrEP Study (Baeten, 2012) was an RCT with three arms, evaluating oral daily PrEP use as TDF/FTC and as TDF alone against a placebo arm, among HIV-discordant heterosexual men and women.
    • The effectiveness of PrEP (TDF/FTC), when used recently, was estimated to be 88% – 90%, which comes from two separate sub-analyses from the Partners PrEP Study.
    • A case/control sub-analysis reported the effectiveness of PrEP, when used recently (based on detecting TFV in plasma), was estimated to be 90% among HIV-uninfected heterosexual men and women (Baeten, 2012).
    • Another restricted analysis of the same study was based on TFV drug levels in plasma. When taking PrEP (TDF/FTC) recently, as defined by >40 ng/ml of TFV in plasma (unknown equivalent pills/week), the risk of HIV acquisition was reduced by 88% among HIV-uninfected heterosexual men and women (Donnell, 2014). Given these levels of TFV in plasma do not translate to a known level of PrEP adherence or known number of pills/week, this finding more accurately corresponds to those taking PrEP recently rather than daily or consistently.
  • PWID: The Bangkok Tenofovir Study (BTS) (Choopanya, 2013) was an RCT evaluating oral daily PrEP use (TDF alone) against placebo among persons without HIV who inject drugs.
    • A case/control sub-analysis reported the effectiveness of PrEP (TDF), when used recently (based on detecting TFV in plasma), was estimated to be 70% among PWID.
    • This BTS study evaluated TDF (Tenofovir) rather than the combination drug TDF/FTC (Truvada). The effectiveness of two-drug oral therapy has not been assessed among PWID but may be higher than TDF alone. TDF alone has been shown to have a slightly lower efficacy than TDF/FTC when compared to placebo, although not statistically different, among heterosexual HIV-discordant couples in the Partners PrEP study (Baeten, 2012; Baeten, 2014).
    • Note that TDF (Tenofovir) is recommended in the U.S. as an alternative to TDF/FTC (Truvada) among PWID.

• Effectiveness estimates based on modified intent-to-treat (mITT) analyses in trials, regardless of level of PrEP use:
  • MSM:
    • The iPrEx Trial (Grant, 2010) was an RCT designed to evaluate the efficacy of oral daily PrEP (TDF/FTC) versus placebo in preventing HIV acquisition among 2,499 HIV-uninfected MSM and transgender women. After a median of 1.2 years of follow-up, the risk of HIV acquisition was reduced by 44% among HIV-uninfected MSM assigned to daily PrEP (TDF/FTC) (mITT analysis). This estimate includes all participants assigned to take daily PrEP, regardless of actual use.
    • The PROUD Study (McCormack, 2015) was an RCT evaluating immediate daily oral PrEP (TDF/FTC) versus delayed PrEP among patients without HIV in 13 clinics in England from 2012-2014. A total of 554 MSM were randomized, 275 to immediate PrEP and 269 to the delayed group. After an interim analysis, the trial stopped early and all deferred patients were offered PrEP. More than 90% of the patients in each group were retained at the end of the study, providing ~500 person-years of follow-up.
      • RCT results (mITT analysis) – At the end of interim analysis, 3 new HIV infections were observed in the immediate PrEP group and 20 in delayed group, resulting in a risk reduction estimate of 86%.
      • There were no HIV infections observed among those taking PrEP. All 3 new HIV infections in immediate PrEP group, based on clinical indications, attendance, and prescription information, were not taking PrEP near the time of seroconversion – 2 never started taking PrEP and 1 infection was identified over 40 weeks after last clinic visit (where 90 PrEP pills were provided).
    • The IPERGAY Trial (Molina, 2015) was an RCT evaluating the efficacy of "on-demand" PrEP (TDF/FTC) regimen (defined as taking 2 pills 2-24 hours before sex, 1 pill 24 hours later, and a 4^th pill 24 hours after the 3^rd) versus placebo among 400 MSM. At the interim analysis of the trial, after 1 year of follow-up, the efficacy of "on-demand" PrEP was estimated to be 86% in the mITT analysis and 82% in the ITT analysis. By measured plasma drug levels in a subset of those randomized to TDF/FTC, 86% had TDF levels consistent with having taken the drug during the previous week.
      • The IPERGAY OLE (Molina, 2017) study. Following the interim analysis where the efficacy of "on-demand" PrEP was determined, the placebo group was discontinued, all study participants were offered TDF/FTC in an OLE phase of the study, and 361 enrolled. Although not part of the trial, the IPERGAY OLE study reported the risk of HIV acquisition was reduced by 97% when comparing the MSM taking PrEP as part of the OLE cohort to the placebo arm of the IPERGAY trial (Molina, 2017). Seventy-one percent of those in the OLE cohort had TDF levels consistent with having taken the drug during the previous week.
      • Two participants in the "on-demand" PrEP arm of the RCT seroconverted after enrollment and 1 participant in the OLE cohort seroconverted during follow-up. In all three cases, study records showed that the participants were not taking PrEP at the time of the diagnosis (no drug detected in plasma and all had returned all or most of their PrEP pills at the most recent visit). No new HIV infections were observed among participants taking PrEP.
      • A small sub-study of the IPERGAY trial reported high effectiveness of on-demand PrEP among those MSM participants with less frequent sexual intercourse (Antoni, 2017). This subset analysis reported an estimated 100% reduction in HIV incidence among a subset of participants reporting less frequent sexual intercourse (median of 5 sex acts/month) when reportedly taking on-demand PrEP, about 9.5 pills/month (or ~2-3 pills/week), compared to placebo.
      • Daily dosing is the only Food and Drug Administration (FDA)-approved schedule for taking PrEP to prevent HIV. However, the International Antiviral Society-USA supports the "off-label" but evidence-based use of on-demand PrEP, as an alternative to daily PrEP, for gay, bisexual and other men who have sex with men with infrequent sexual exposures (Saag, 2018). Given limited data on the effectiveness of on-demand PrEP for heterosexual men and women, PWID, and transgender persons, IAS-USA does not currently recommend on-demand PrEP for these populations. Several health departments have developed guidance on off label use of on demand PrEP for MSM, including the New York City Department of Health.
  • Transgender women: A follow-up sub-analysis of the iPrEx Trial evaluated the effectiveness of PrEP (TDF/FTC) versus placebo among 339 transgender women (Deutsch, 2015). No benefit of PrEP was identified (HR=1.1, 95% CI: 0.5 – 2.7); however the transgender women appeared to have lower PrEP adherence than MSM within iPrEx.
  • Heterosexual men and women:
    • The Partners PrEP study was an RCT among 4747 HIV-discordant heterosexual couples assessing the efficacy of oral daily PrEP by comparing three treatment arms – TDF/FTC (Truvada), TDF alone, and placebo. The risk of HIV acquisition was reduced by 75% among HIV-uninfected heterosexual men and women assigned to TDF/FTC (Truvada) compared to placebo (mITT analysis; Baeten, 2012). This estimate included all participants assigned to take daily PrEP, regardless of actual use.
    • The TDF2 study was an RCT among 1219 heterosexual men and women without HIV comparing TDF/FTC (Truvada) to placebo and found the risk of HIV acquisition was reduced by 62% (mITT analysis; Thigpen, 2012). This estimate included all participants assigned to take daily PrEP, regardless of actual use. An as-treated analysis, restricting to those participants taking PrEP recently based on self-reported PrEP use in last 30 days, found the risk of HIV acquisition was reduced by 78%. This, however, was based on self-report and not an objective measure of recent use.
    • There are additional PrEP trials among women reported in the literature not summarized here. Riddell (2018) and the USPSTF (2019) reviewed the trial findings for PrEP and described additional trials among women showing no significant effects of PrEP, primarily due to extremely low adherence among women in the studies.
  • PWID: The Bangkok Tenofovir Study (BTS) was an RCT evaluating oral daily PrEP use (TDF alone) against placebo among persons without HIV who inject drugs. This trial showed the risk of HIV acquisition was reduced by 49% among HIV-uninfected injecting drug users assigned to oral daily PrEP (TDF) (mITT analysis; Choopanya, 2013). This estimate included all participants assigned to take daily PrEP, regardless of actual use.

Evidence supporting effectiveness estimates

• Effectiveness Estimates based on "Optimal Use" of Condoms.
  • Optimal use of condoms is defined here as both consistent and correct use during every sex act.
  • Laboratory studies show that (latex-based, polyurethane, or other synthetic material-based) condoms provide an impermeable barrier to passage of HIV. Even during optimal use, however, condoms may not offer complete protection all the time due to the rare chance of product failure.
  • Measures are in place to ensure high quality control on product development. Condoms are regulated as class II medical devices by the U.S. Food and Drug Administration (FDA). FDA requires every condom to be tested electronically for holes and weak spots before it is packaged and released for sale. In addition, samples of condoms undergo a series of additional laboratory tests for leakage, strength, and other factors. Condom samples must be at least 99.6% effective in laboratory "water leak" tests, which means that at least 996 out of every 1000 condoms sampled must pass the test. (Warner, 2018; FDA)
  • Other laboratory testing has estimated that the worst-case product failure would lead to less than 0.01% of volume leakage during sex. In other words, the worst-case scenario would still eliminate about 99.99% of volume exposure during sex, in the event of product failure. (Carey, 1992)
• Effectiveness Estimates based on "Consistent Use" of Condoms.
  • Although rare, and not easily measured, condoms may break, slip, or leak during use, even if used correctly. In addition, not using condoms correctly (user failure) increases the risk of breakage, slippage, leakage, or incomplete coverage which can increase exposure to HIV and, thus, may decrease condom effectiveness. Because male condoms are applied by the user during sex, user error or failure is an ongoing risk during each sexual episode. User error is difficult to eliminate; however, over time, as the user becomes more experienced, it is minimized. In addition, not using condoms consistently, meaning during every sex act, may further increase potential exposure to HIV and decrease effectiveness even more. Below are effectiveness estimates for consistently using condoms in practice as measured in observational studies.
  • Heterosexual Men and Women: The Weller 2002 Cochrane review of 13 longitudinal cohort studies among HIV discordant heterosexual couples reported results comparing those reporting "Always" vs "Never" using condoms during vaginal sex from 5 of the 13 studies with data available at the longest follow-up. Vaginal versus anal and insertive versus receptive sex were not distinguished in these analyses. Always using condoms, based on self-report, during sex with a partner who have HIV reduces the risk of HIV acquisition per person-year of follow-up by an estimated 80% among heterosexual men and women. This measure does not account for the possibility of different numbers of sex acts over time between condom users and non-users.
  • MSM: Two recent studies have estimated the effectiveness of consistent condom use on HIV risk among MSM without HIV having sex with men with HIV.
    • The Smith 2015 study combined data from two longitudinal studies among MSM (EXPLORE & Vax004) and compared MSM without HIV who reported "Always" vs "Never" using condoms during receptive anal sex, during insertive anal sex, and during any anal sex, with partners who have HIV.
      • MSM, Receptive Anal Sex — Always using condoms, based on self-report, during receptive anal sex with partners who have HIV reduced the risk of HIV acquisition per person-year by an estimated 72% among MSM.
      • MSM, Insertive Anal Sex — Always using condoms, based on self-report, during insertive anal sex with partners who have HIV reduced the risk of HIV acquisition per person-year by an estimated 63% among MSM. This analysis does not take into account whether MSM without HIV also engaged in receptive anal sex, with or without condoms, which could affect this estimate.
      • MSM, Any Anal Sex — Always using condoms, based on self-report, during any (insertive or receptive) anal sex with partners who have HIV reduced the risk of HIV acquisition per person-year by an estimated 70% among MSM.
      • These measures do not account for the possibility of different numbers of sex acts over time between condom users and non-users.
    • The Johnson 2018 study examined condom effectiveness per partner in four cohorts of MSM (EXPLORE, Vaxx004, JumpStart, and Vaccine Preparedness Study) by comparing those "Always" using condoms versus "Not always" using condoms, based on self-report, throughout the sexual partnerships. Among HIV-uninfected MSM engaging in receptive anal sex with their partner who have HIV, always using condoms during receptive anal sex throughout the partnership reduced the risk of HIV acquisition per partner by an estimated 91%. This measure does not account for the possibility of different numbers of sex acts per partner between condom users and non-users.
  • The estimates provided here likely underestimate the effectiveness of condoms when used consistently and correctly in practice due to measurement error regarding both aspects of condom use – consistent use and correct use.
    • These estimates for "consistent use" are based on observational cohort studies because no RCTs exist, due to ethical and feasibility concerns with assigning a no condom use arm. In addition, only subjective measures of condom use (self-report) are available in studies with HIV as an outcome, which may overestimate actual condom use, resulting in underestimating condom effectiveness. Therefore, the effectiveness of consistent condom use is likely greater.
    • These studies also did not measure whether condoms were used correctly. If used incorrectly, condoms may break, slip, leak, or not provide complete coverage, which may increase exposure to HIV. The studies among MSM, however, did ask MSM to count "breakage" and "slippage" as "not using a condom" in an attempt to account for user failure – but this relies on knowledge of failure and self-report and likely underestimates true failure. If these analyses included any data where condoms were used incorrectly but misclassified as consistent and correct use, then these estimates are likely underestimating condom effectiveness when used correctly, and the effectiveness of correct condom use is likely greater.

Strengths and limitations of effectiveness estimates

• Most of the evidence is based on observational studies and circumcision status is primarily based on self-report; only some studies are based on medical exam (objective measure of exposure).
• MSM Insertive Anal Sex – A Cochrane review of 7 observational studies among MSM reporting mainly or only "insertive" sex reports a significant protective effect of circumcision on acquiring HIV through insertive anal sex, 73% risk reduction (Wiysonge 2011). Exposure (circumcision) was primarily measured via self-report (subjective measure), although genital exams occurred in some studies. Two more recently published observational studies show non-significant effects of circumcision on HIV acquisition during insertive anal sex (Sanchez, 2011; Doerner, 2013). With conflicting results, the evidence is inconclusive and an updated meta-analysis is needed.
• MSM Receptive Anal Sex – A Cochrane review of 3 observational studies among MSM reporting primarily "receptive" sex reports a non-significant effect estimate for circumcision (of the insertive partner) on HIV acquisition during receptive anal sex, with exposure measured by self-report (Wiysonge 2011). A more recently published observational study reports a significant effect of circumcision (based on self-report) on HIV acquisition during receptive anal sex among MSM (Schneider, 2012). With conflicting results, the evidence is inconclusive, and an updated meta-analysis is needed.
• Heterosexual Men – A Cochrane review of 3 RCTs synthesizes ITT results on the effects of circumcision on risk of HIV acquisition during sex among heterosexual men without HIV (Siegfried, 2009).
• Heterosexual Women – A meta-analysis (including one RCT and several observational studies) reports that there is insufficient evidence to conclude that male circumcision reduces the risk of HIV acquisition during sex among heterosexual women without HIV (Weiss, 2009). Two more recent reports, 1 RCT and 1 observational study, also show non-significant effects of male circumcision (confirmed by medical exam) on HIV acquisition in women among HIV-discordant heterosexual couples (Baeten, 2010; Wawer, 2009). The evidence is inconclusive, and an updated meta-analysis is needed.

Strengths and limitations of risk estimates

• Hughes, 2012 is a longitudinal cohort study of 3,297 discordant heterosexual African couples, with 86 confirmed linked HIV transmissions. This risk estimate for ulcerative STIs is specific to genital ulcer disease among partners without HIV, where genital ulcer disease includes genital herpes (due to HSV-1 or HSV-2), syphilis, and chancroid.
• The evidence supports an increased risk of HIV acquisition when the uninfected partner has an STI. It is unclear, however, what the most appropriate estimates are for each population, behavior, and type of STI given the observational nature of these kinds of studies. Several studies have found that ulcerative STIs may confer higher risk (e.g., risk ratio ranges from 2.2 to 11.3) than non-ulcerative, inflammatory STIs (e.g., risk ratio ranges from 3 to 4) (Fleming, 1999; Galvin, 2004; Berman, 2006). However, an updated systematic review of the published literature is needed for more accurate risk estimates by population, behavior, and type of STI.
• Although no direct empirical evidence has been identified for MSM at this time, it is biologically and epidemiologically plausible for STIs to also increase the risk of HIV acquisition among MSM without HIV. The estimate for heterosexual men and women is the best proxy estimate for MSM until more direct evidence is available.

Strengths and limitations of risk estimates

• Gray, 2001 is a longitudinal cohort study of 174 monogamous discordant heterosexual couples, with 38 HIV transmissions to uninfected partners, from a larger Rakai (Uganda) community-randomized trial of STD control for AIDS prevention. This risk estimate for ulcerative STDs is specific to the presence of genital ulceration.
• The evidence suggests an increased risk of HIV transmission due to the partner with HIV having an STD. It is unclear, however, what the most appropriate estimates are for each population, behavior, and type of STD given the observational nature of these kinds of studies. Indirect evidence of STDs increasing the infectiousness of HIV exist. Ulcerative STDs generally increase HIV shedding in the genital tract and inflammatory STDs increase the concentration of HIV in the urethra, semen, and cervical fluid (Galvin, 2004). An updated systematic review of the published literature is needed for more accurate risk estimates by population, behavior, and type of STD.
• Although no direct empirical evidence has been identified for MSM at this time, it is biologically and epidemiologically plausible for STDs to also increase the risk of HIV transmission among MSM with HIV. The estimate for heterosexual men and women is the best proxy estimate for MSM until more direct evidence is available.

Strengths and limitations of risk estimates

• Wawer, 2015 estimates the increased risk of HIV transmission due to acute HIV infection from a retrospective sub-sample of 235 monogamous, HIV-discordant heterosexual couples with follow up time from a larger Rakai (Uganda) community-randomized trial of STD control for AIDS prevention. This study looked at early-stage infection (defined as up to 5 months after seroconversion, a 2.5-month midpoint), established infection or "middle" stage of infection (>6 months after seroconversion), and late-stage infection (6 to 25 months before death).
• No other published study of empirical data on increased risk of HIV transmission during acute infection exists.
• Acute HIV infection is clinically defined as the time between viral infection and development of detectable antibodies against HIV-1. During this time, concentrations of HIV in blood and semen are highest and transmission risk is therefore greatest; this period typically last a few weeks (Cohen, 2005; Pilcher, 2004). Pilcher, 2007 estimated that viral load reaches its peak at 17 days after seroconversion in blood and around 4 weeks after seroconversion in semen. A modeling paper (Pilcher, 2004) estimated the probability of heterosexual transmission of HIV during acute infection is ~8 to10 times the probability during later stage of infection, where the viral load peak in semen was modeled at about day 20 after infection.
• Although no direct empirical evidence has been identified for MSM at this time, acute infection is likely to also be associated with increased HIV transmission risk among MSM. The estimate for heterosexual men and women is the best proxy estimate for MSM until more direct evidence is available.