Growth in proportion and disparities of HIV PrEP use among key populations identified in the United States national goals: systematic review & meta-analysis of published surveys

Introduction

Pre-exposure prophylaxis (PrEP) has been shown to reduce HIV transmission in clinical trials as well as community-based (‘real-world’) studies.^1–4 The United States (US) Food and Drug Administration (FDA) licensed tenofovir disoproxil fumarate (TDF)/emtricitabine for PrEP in 2012 and emtricitabine/tenofovir alafenamide in 2019. The US Centers for Disease Control and Prevention (CDC) released the first PrEP guideline in 2014. Use of PrEP has been significantly associated with declines in new HIV diagnoses in the US.⁵

A list of updated national HIV goals from the US national HIV strategy identified seven key populations and geographic areas that are vulnerable to HIV infection in the US: gay, bisexual, and other men who have sex with men of all races and ethnicities (collectively referred to as MSM); black or African American women and men (hereafter referred to as blacks); Hispanic or Latino men and women (hereafter referred to as Hispanic/Latinos); persons who inject drugs (PWID); youth aged 13 to 24 years; people in the Southern US; and transgender women.^6,7 About 66% of US HIV infections diagnosed in 2017 were among MSM, 43% among blacks, 26% among Hispanic/Latinos, 7% among PWID, 21% among youth, and 52% among people in the Southern US.⁸ While surveillance data does not include information on transgender women (which may be included in MSM group), an estimated 1% of new HIV diagnoses were among this population.⁹ Since members of these key populations are at highest risk for HIV infection, they may be prioritized candidates for PrEP.

A recent assessment of US retail pharmacies estimated that 148,147 unique individuals have taken TDF/emtricitabine for PrEP as of September 2017.¹⁰ This number does not include clients in closed healthcare systems with their own pharmacies that are not included in retail pharmacy data. Also, due to the cost of antiretroviral medications (ARVs), lack of awareness that ARV can legally be prescribed for use by HIV-seronegative persons, or the hassle and time it takes to get a prescription, some HIV-seronegative persons have taken ARVs to prevent HIV without a prescription or supervision by healthcare providers.¹¹ A recent meta-analysis found that by 2016, about one in six people who responded to US surveys self-reported using PrEP (although this was not confirmed by medical record).¹² However, it is unclear how widely PrEP has been used within key populations identified in the US national HIV goals, and whether PrEP has reached populations with the highest need. The purpose of this meta-analysis was to: 1) assess trends in PrEP use by estimating the proportion of lifetime PrEP use across years and growth pre-/post- CDC clinical guideline released in 2014 and 2) assess PrEP use disparities among key populations and geographic areas identified in the US national HIV goals.

Methods

We implemented a two-step systematic literature search to identify PrEP-related citations in the CDC HIV/AIDS Prevention Research Synthesis (PRS) Project database. The process of creating a comprehensive systematic literature search strategy in MEDLINE, EMBASE, PsycINFO and CINAHL for the PRS database has been previously published (Appendix I, Step 1).¹³ The search for this review consisted of several queries of the PRS database (Appendix I, Step 2) and reference list checks of included citations with the last search taking place in April 2018. We also searched for any newly published literature in PubMed using HIV and PrEP terms (last searched May 2019). All identified citations were uploaded to DistillerSR (Evidence Partners, Ottawa, Canada) for screening, data abstraction, and quality assessment.

Screening, data abstraction, and quality assessment

Inclusion criteria for this review were: 1) primary studies with the number or proportion of PrEP users reported among the study participants, 2) implementation in the US, and 3) publication in English. A reviewer screened titles and abstracts to identify PrEP primary studies meeting criteria. Exclusions were validated by a second reviewer. Next, full texts of identified studies were screened again for eligibility. To abstract data from eligible studies (N), only data from baseline assessment were included for prospective and intervention studies. Studies that used the same survey/dataset were carefully screened and only data from unique samples and subgroups (k) were included in this review. If studies spanned multiple years, the midpoint of the time span was used to represent the year. For studies reporting the proportions for both lifetime and current/recent use, we used the lifetime proportion. Full text screening, data abstraction, and risk of bias assessment were conducted by two independent reviewers; conflicts were resolved through discussion. We contacted authors to obtain additional information for studies that did not report the data needed for this review.¹⁴

Risk of bias was assessed using a Newcastle-Ottawa Scale.¹⁵ This scale adapted for cross-sectional studies assesses qualities in five domains: selection of participants, sample size, comparability of respondents, ascertainment of PrEP uptake, and quality of descriptive statistics reporting. We further adapted this scale for this review. A strength of this scale is that validity and reliability has been refined and establised over time by several expertswhile simple risk of bias scales may not assess other biases.^16–19 Total scores were calculated by counting the number of ‘Yes’ responses. The total possible value was five points (zero to five), with three or more points considered “low risk of bias.”^16–19

Data synthesis

The review describes characteristics of included studies using narrative synthesis. First, studies were categorized as focusing on one or more of seven key populations. Studies combining MSM and transgender women, and studies presumed to focus on MSM (e.g., participants surveyed at gay pride events) were considered MSM studies. Next, because the majority of studies focused on MSM, we separated out non-MSM and MSM studies. For non-MSM studies, we created subgroups for each key population (e.g., black non-MSM). We also created two time groups based on CDC PrEP clinical guidelines: pre-(in/before 2014) and post-(after 2014).

We conducted a series of analyses. First, we estimated pooled proportions of participants reporting PrEP use and included all studies across years by using a fixed-effects as well as random-effects models in the meta-analyses since the proportions varied across studies. Random effects models are conducted by adding a variance component tau-squared (estimating the variance among the true effect sizes) to each study-specific variance.²⁰ Second, to assess the trends of PrEP use, we estimated the pooled proportion of PrEP use for each year of study. Third, we repeated this analysis for each key population and non-MSM subgroup in recent years (2015–2017). For assessing differences in PrEP use, we compared 95% confidence interval (CI) for odds ratios (OR). Fourth, we run multivariable logistic regression models to estimate adjusted odds ratio (aOR) for the MSM and non-MSM overall groups. Finally, mixed effects logistic regression models estimated crude OR for growth rates of PrEP use pre-/post-CDC PrEP clinical guideline era for overall, key populations, and non-MSM subgroup key populations.

We assessed heterogeneity with the I² index that quantifies the degree of true heterogeneity (i.e., between-studies variability) rather than variability due to sampling error within studies; heterogeneity greater than or equal to 75% was considered high as this indicates that ¾ of total variability among effect sizes is caused by variation between studies and this could lead to biased interpretation of results.^20,21 To reduce conceptual heterogeneity caused by population and outcome variations, we grouped studies into three categories based on the number of studies as well as considering FDA approval in 2012 and the clinical guideline release in 2014²²: 2004–2012, 2013–2014, and 2015–2017, then sorted by key populations and non-MSM subgroups.

We conducted sensitivity analyses to assess bias. For studies reporting no PrEP use among study participants, we applied continuity correction and added 0.5 (half of an individual) to event and non-event values to compute logit event rates.^23,24 To assess the bias caused by this strategy, we compared the estimated overall proportion for all identified studies to the proportion without studies with zero proportions. To assess the stability of the results, we used fixed-effects models in order to present simple percentages in each category. We conducted multivariable logistic regression on the comparison of proportions between the overall MSM and non-MSM subgroups in recent years excluding surveys reporting use in past 12 months (i.e., including ever, current, and past 6 months use). To assess bias due to overlap between key populations (e.g., black MSM), we also conducted sensitivity analyses comparing the estimated population for each key population strata to the proportions for each subgroup strata (e.g., blacks strata vs. black MSM substrata).

All meta-analyses were conducted using Comprehensive Meta-Analysis Software Versions 2 and 3 (Biostat, Englewood, NJ). P-values less than 0.05 and non-overlap of 95% CI for ORs were considered statistically significant.

Results

We identified 2,019 citations in the PRS database and an additional 30 newly published citations through hand searches in PubMed (Figure 1). We excluded 1,448 citations at the title and abstract level and 506 studies at the full text level. The remaining 95 studies were included in this review (Supplemental Table 1, Appendix II and Appendix III). Data were collected between 2004 and 2017. Sample sizes of the included studies varied from 12 to 19,587 respondents; this review included a total of 95,854 responses.

Figure 1:

Flow diagram of included studies

Since this review focused on self-reported surveys and none of the included studies validated the participants’ PrEP use with medical records, no studies achieved the highest possible score (5 points) for study quality (Supplemental Table 1, Appendix II). Thirty-three studies scored four points while nearly half of the studies (n=43) scored three points. Seventeen studies attained two points and two study garnered only one point.

Of the 95 included studies, 89 exclusively focused on one or more key populations. Eighty (84.2%) studies reported the proportion of MSM while 18 (18.9%) studies reported the proportion among non-MSM. This review also included 39 (41.1%) studies which presented data on blacks, 26 (27.4%) studies on Hispanic/Latinos, 19 (20.0%) studies on youth, 19 (20.0%) studies on people in the Southern US, nine (9.5%) on transgender women, and six (6.3%) studies on PWID. Only four studies (4.2%) focused on female members of one of the groups women identified in national HIV goals (n=2, 2.1%) or on women in general (n=2, 2.1%).

Proportions of self-reported PrEP use overall and by year

The unstratified overall average proportion of reported PrEP use across all years and populations was 5.4% (95%CI:4.6–6.3, k=95, I²=97%). Because the tau was calculated each year separately, the stratified overall proportion from a random-effects model was 2.9% (95%CI:1.7–5.0, k=95, I²=97%) (Figure 2). Overall proportions reporting PrEP use each year in 2004–2017 were: 3.0%; 0.3%; 2.3%; 2.0%; 1.0%; 2.3%; 0.7%; 0.9%; 3.0%; 2.7%; 3.2%; 9.5%; 11.6%; and 15.8%, respectively; note some estimates are from single studies.

Figure 2:

Forest plot of pooled proportions of self-reported pre-exposure prophylaxis (PrEP) use stratified by year (N=95)

Proportions of self-reported PrEP use in recent years (2015–2017)

In terms of PrEP use in recent years (2015–2017), the overall proportion was 11.3% (95%CI:9.9–12.9, k=46, I²=96%) (Table 1). Within key populations, MSM reported the highest proportion of using PrEP (13.9%, 95%CI:8.8–21.1, k=39, I²=96%) followed by Hispanic/Latinos (11.5%, 95%CI:7.1–18.1, k=12, I²=95%), transgender women (11.2%, 95%CI:5.8–20.6, k=5, I²=73%), blacks (9.9%, 95%CI:8.3–11.8, k=18, I²=96%), people in the Southern US (9.9%, 95%CI:4.7–19.7, k=8, I²=94%), youth (7.3%, 95%CI:4.7–11.2, k=8, I²=92%), and PWID (3.7%, 95%CI:0.8–16.1, k=3, I²=73%).

Table 1:

Pooled proportion of self-reported PrEP ever use by year and strata among US study participants (N=95)

Year		Key populations proportion (95%CI), I2[k]
	Overall	MSM	Hispanic/Latinos	TGW	Blacks	So US	Youth	PWID
2004	3.0(1.2–7.3), 91.2[2]	3.0(1.2–7.3), 91.2[2)	0.2(0–5.5), 0[1]*	-	0.6(0–10.1), 0[1]*	6.8(1.7–23.1), 0[1]	0.3(0–5.9), 0[1]*	-
2005	0.3(0–2.2), [1]	0.3(0–2.2), 0[1]	-	-	-	-	-	-
2006	2.3(0.8–6.6), 0[1]	2.3(0.8–6.6), 0[1]	-	-	-	-	-	-
2007	2.0(0.6–6.1), 0[1]	2.0(0.6–6.1), 0[1]	-	-	-	-	-	-
2008	1.0(0.3–3.6), 66.0[2]	1.0(0.3–3.6), 66.0[2]	0.5(0–12.0), 0[1]*	-	1.8(0.1–24.2), 0[1]*	-	0.8(0–13.3), 0[1)*	-
2009	2.3(0.6–8.1), 0[1]	2.3(0.6–8.1), 0[1]	-	-	-	-	-	-
2010	0.7(0.3–2.0), 0[2]	0.7(0.3–2.0), 0[2]	-	-	-	-	-	-
2011	0.9(0.5–1.7), 82.0[7]	0.9(0.5–1.7), 82.0[7)	1.1(0.2–5.6), 0[4]	-	1.8(0.6–5.5), 18.5[4]	0.1(0–0.9), 0[2]	0.5(0.1–4.2), 0[2]*	3.8(0.0–40.3), 0[1]*
2012	3.0(1.6–5.6), 94.3[5]	3.5(1.9–6.5), 95.3[4] 0.2(0–3.2), 0[1]	- -	- -	0.2(0–1.4), 0[2]* 0.2(0–2.6), 0[1]*	- -	0.7(0.2–2.3), 0[1] -	0.2(0–2.6), 0[1] 0.2(0–2.6), 0[1]*
2013	2.7(2.0–3.8), 67.5[13)	2.6(1.8–3.7), 67.6[11] 4.9(1.6–13.9), 0[3]	4.0(0.4–28.0), 0[1]	5.1(1.4–17.4), 0[2] -	2.5(1.7–3.8), 56.3[5] -	1.7(0.5–5.8), 0[1] 3.8(0–100), 0[1]*	0.5(0–8.6), 0[1]* -	- -
2014	3.2(2.3–4.4), 90.2[14]	4.2(3.0–5.9), 91.2[9] 0.6(0.2–1.5), 0[5]	3.5(1.5–7.7), 20.6[7] 4.4(0.6–25.7), 0[2]*	1.5(0.2–9.4), 0[2] -	4.4(3.3–6.0), 74.7[9] 6.4(0.9–34.4), 0[2]*	4.4(2.5–7.6), 75.8[6] 0.4(0–100), 0[1]	2.7(1.4–5.0), 0[5] 1.5(0.2–9.8), 0[1]	- -
2015	9.5(7.5–11.8), 93.5[20]	10.0(8.0–12.5), 93.9[18] 5.2(2.3–11.5), 60.4[3]	9.5(4.3–19.8), 78.58[5] 5.0(0.7–28.2), 0[1]	11.4(3.2–33.6), 0[2] -	8.3(6.3–10.9), 39.0[8] 5.8(3.2–10.2), 0[1]	8.6(4.7–15.2), 0[4] -	7.8(4.9–12.1), 89.7[5] 8.3(4.5–14.7), 0[1]	0.8(0.1–5.8), 0[1] 0.8(0.1–5.8), 0[1]
2016	11.6(9.2–14.5), 94.7 [17]	11.8(9.3–14.9), 95.4[15] 6.2(3.5–10.8), 85.4[5]	11.3(5.8–20.9), 7.6[6] 15.5(11.1–21.4), 0[1]	11.2(4.9–23.6), 84.6[3] -	9.5(7.3–12.3), 73.8[7] 12.4(8.3–18.2), 0[1]	16.3(8.0–30.5), 96.0[3] -	7.2(3.3–4.5), 89.5[2]	9.7(6.9–13.4), 0[2] 0[1]**
2017	15.8(11.2–21.8), 97.8[9]	23.5(16.4–32.4), 98.5[6] 3.4(1.4–8.1), 0[3]	28.6(5.9–71.8), 0[1] -	- -	18.1(11.7–27.1), 0[3] -	4.8(1.2–16.6), 0[1] 4.8(0–100), 0[1]	4.8(1.5–14.4), 0[1] 4.8(2.4–9.2), 0[1]	3.0(0.4–18.6), 0[1] 3.0(0.4–18.6), 0[1]
Overall	2.9 (1.7–5.0), 97.1[95]	3.1(1.7–5.4), 97.2[80] 2.7(1.0–6.6), 79.8[20]	4.5(1.8–11.1), 93.1[26] 9.5(3.2–25.0), 9.1[4]	6.7(2.6–16.1), 63.2[9] -	4.4(2.4–7.8), 88.1[40] 5.4(1.9–14.6), 69.9[5]	4.1(1.7–9.5), 93.3[19] 2.0(0–100), 61.1[3]	2.3(1.0–5.1), 90.8[19] 5.3(2.4–11.0), 44.1[3]	2.1(0.4–10.7), 0[6] 0.9(0.2–5.1), 31.3[4]
‡Recent years	11.3 (9.9–12.9) 95.8 [46]	13.9(8.8–21.1), 96.4[39] 5.3(3.7–7.5), 73.1[11]	11.5(7.1–18.1), 95.7[12] 12.9(5.5–27.3), 31.3[2]	11.2(5.8–20.6), 72.99[5] -	9.9(8.3–11.8), 96.6[18] 8.8(4.1–17.9), 78.8[2]	9.9(4.7–19.7), 94.3[8] 4.8(2.4–9.2),0[1]	7.3(4.7–11.2), 92.2[8] 6.4(3.7–10.8), 30.7[2]	3.7(0.8–16.1), 72.9 [4] 1.6(0.4–6.2), 0[2]

MSM: men who have sex with men, PrEP: pre-exposure prophylaxis, PWID: persons who inject drugs, So US: people in the Southern United States, TGW: Transgender women

Bold and italic: non-MSM (all non-MSM under the MSM column; non-MSM who were also members of that subgroups under all the other columns)

No study or unable to estimate odds ratio due to limited number of studies (-)

I²: Heterogeneity, k: number of surveys

^*Zero proportion (no PrEP use among study participants)

^†Only one PWID in the study who did not report PrEP use

^‡From 2015 to 2017

Among subgroups of non-MSM (overall proportion: 5.3%, 95%CI:3.7–7.5, k=11, I²=73%), the highest proportion of PrEP use was observed among Hispanic/Latino non-MSM (12.9%, 95%CI:5.5–27.3, k=2, I²=31%), followed by blacks (8.8%, 95%CI:4.1–17.9, k=2, I²=78%), youth (6.4%, 95%CI:3.7–10.8, k=2, I²=31%), non-MSM in the Southern US (4.8%, 95% CI: 2.4–9.2, k=1, I²=n/a), and PWID (1.6%, 95% CI: 0.4–6.2, k=2, I²=0%).

Non-overlapping 95% CI were indication of some significant differences between strata. These included lower prevalence of PrEP use among non-MSM compared to black MSM; non-MSM living in the South compared to MSM; and non-MSM PWID compared to MSM, Hispanic/Latino MSM or black MSM.

Moreover, the odds of reporting ever PrEP use were twice (aOR=2.07, 95% CI: 1.27–3.38) as high among MSM compared to non-MSM after adjusting for the year between 2015–2017. Hispanic/Latinos was the only strata where non-MSM population reported a higher proportion of PrEP use than the combined group including both MSM and non-MSM.

Growth rates of PrEP use

The overall growth rate significantly increased both within the pre-guideline era (in/before 2014, OR=1.11/year, 95%CI:1.01–1.21) and the post-guideline era (after 2014OR=1.34/year, 95%CI:1.09–1.64); the growth rate in post-guideline era was larger, but not statistically significantly (Figure 3).

Figure 3:

Growth rate of proportion of self-reported PrEP use within both pre-CDC PrEP clinical guideline era (2004–2014) and post-guideline era (2015–2017) time periods among US study participant

The proportion of PrEP use grew significantly during the pre-guideline era among all key populations except among transgender women (OR=0.29/year, 95%CI:0.05–1.67) and people in the Southern US (OR=0.99/year, 95%CI:0.85–1.15). Due to the limited number of studies, we were not able to calculate a growth rate among PWID nor among any of the non-MSM subgroups.

In the post-guideline era, the largest point estimate for growth rate was among Hispanic/Latinos (OR=1.59/year, 95%CI:0.62–4.08), however this increase was not significant. Among significant increases during the post-guideline era, the largest point estimate was among MSM (OR=1.53/year, 95%CI:1.21–1.93), followed by blacks (OR=1.44/year, 95%CI:1.13–1.83). The proportion was stable among transgender women (OR=1.02/year, 95 %CI: 0.16–6.37) while the proportion did not demonstrate growth among people in the Southern US (OR=0.94/year, 95%CI:0.29–3.18). Youth significantly improved PrEP use over time in pre-guideline era (in/before 2014) (OR=1.48/year, 95%CI:1.21–1.80), but in post-guideline era (after 2014)-proportion of use (OR=0.82/year, 95%CI:0.43–1.55) did not demonstrate growth.

Heterogeneity

Overall heterogeneity and heterogeneity among strata were as high as 97%. We were able to reduce some heterogeneity during various subgroup analyses, but the heterogeneity remained high for some strata - particularly for 2015–2017 and people in the Southern US. (Supplemental Table 2, Appendix II).

Sensitivity analysis

The proportion of PrEP use after excluding studies reporting zero proportion (3.3%, 95%CI:2.0–5.5, k=83, I²=97%) did not much differ from a model that included all studies (2.9%, 95%CI:1.7–5.0, k=95, I²=97%). Our estimated proportions were stable with the continuity correction.

Differences among proportions reporting ever, current, and past 6 months PrEP use were similar for overall, MSM, and non-MSM (Supplemental Table 3, Appendix II). Moreover, logistic regression excluding past 12-month use showed that the difference of proportions between MSM and non-MSM in recent years after accounting for years of study remained significant (aOR=2.34, 95%CI:1.63–3.50). These analyses suggest our findings were reliable even though the review included proportions of lifetime use as well as other recall periods.

Finally, we assessed biases caused by overlap between key populations (e.g., black MSM, Hispanic/Latinos in the Southern US). The proportions of PrEP use among MSM subgroup strata (4.7% for black MSM, 4.4% for Hispanic/Latinos MSM, 2.0% for MSM youth, and 4.0% for MSM in the Southern US) did not much differ from proportions that included both MSM and non-MSM studies (4.4%, 4.5%, 2.3%, 4.1%, respectively) except PWID (6.2% for MSM and 2.1% for MSM and non-MSM). Therefore, our strategy to separate out non-MSM and report proportions for non-MSM subgroups (but not for MSM subgroups) appears acceptable. Due to the limited number of studies, we were not able to compare other key population subgroups (e.g., black vs black in the Southern US) and assess those biases.

Discussion

Our review pooled data from 95,854 responses to highlight trends and disparities in PrEP use. Because this is a systematic review of published surveys, our estimated proportions cannot be generalized beyond the people who took the surveys; however, our findings may be important to consider for future research and clinical practice.

CDC’s national HIV behavioral surveillance system reported that 3.5% of MSM participants had taken ARVs before sex in the past 12 months in 2014.²⁵ Our sample reported a higher rate (4.2%) than the behavioral surveillance estimates. The difference in estimates may be due to our review reporting lifetime use while the CDC’s behavioral surveillance study focused on PrEP use in the past 12 months. Moreover, our samples may not be representative of the key populations identified in the US national goals.

In this review, MSM and Hispanic/Latinos reported higher PrEP use and faster growth rates than other key populations. Proportions of PrEP use among blacks and people in the Southern US in recent years were similar, yet PrEP use among blacks improved significantly in the post-guideline era while people in the Southern US were not much impacted by the guideline release. Moreover, PrEP use significantly increased among youth in the pre-guideline era; however, after the guideline release, growth stopped. Our review also found a 34% increase per year in the odds of lifetime PrEP use in the post-guideline era, and the odds of reporting PrEP use among MSM were twice as great than among non-MSM.

We found very few studies for PWID. With the limited number of studies, PWID, especially non-MSM PWID, reported the lowest PrEP use compared to other key populations or non-MSM subgroups. Even though clean injecting equipment can reduce the risk of HIV transmission, PWID are 22 times more likely to acquire HIV compared with the general population due to high levels of HIV risk behaviors (e.g., sharing injection equipment with people with HIV) suggesting the importance of offering PrEP to this key population.^26,27 Further studies on identifying barriers and facilitators of PrEP prescription and use in PWID may help us understand this population’s low PrEP use.

As noted previously the proportion of PrEP use in youth demonstrated significant growth in the pre-guideline era while growth was no longer significant post-guideline era despite low PrEP use in this population. One reason for the low rate may be that TDF/emtricitabine was not approved for individuals under aged 18 until May 2018. With the recent FDA approval in addition to CDC support document released in November 2018 for individuals aged 13 years and older, PrEP use in youth aged 13 to 24 years may significantly increase in the next few years.^28,29 A follow-up study assessing PrEP use in youth after FDA approval may help us learn the impact of these FDA approval and new guidance from the CDC.

Both blacks and people in the Southern US reported similar proportions of PrEP use in recent years; however, the proportion did not demonstrate growth among people in the Southern US, while PrEP use among blacks improved significantly in the post-guideline era. This may show geographical disparities in PrEP access and availability. People in the Southern US account for more than half of HIV diagnoses in the US.⁸ To reduce the healthcare disparity due to geographical location, more studies are needed to identify barriers for PrEP use in this geographical area to understand and reduce barriers.

We identified only four studies that focused on women. While three of these studies reported less than 1% of PrEP use among participating women, one recent study reported 2.4% of PrEP use in 2017,³⁰ the overall proportion of 0.8% was the lowest proportion among any of the populations described in this review. Moreover, black women comprised 59% of women diagnosed with HIV in 2017 while representing 13.4% of women in the US.^31,32; however our review found no studies that focused on PrEP use among black women. Black women are particularly vulnerable to HIV and in need of PrEP compared to other non-MSM populations, but are neglected from PrEP studies. Further PrEP intervention studies focusing on or including women may be necessary to increase PrEP uptake in this population.

Heterogeneity remained high for some subgroups (e.g., study years 2015–2017, people in the Southern US, Supplemental Table 2). This might be due to the significant improvement of PrEP use over three years of the time period. Also, heterogeneity tended to be high among people in the Southern US, which may be due to the large geographic area of 17 states, with potential differences in other variables in each study location. Finally, this review included 19 studies with high risk of bias. These studies could potentially improve the scores (i.e., reduce risk of bias) by having recruited more participants from multiple locations.

This review has several other limitations. Participants may have been counted more than once in our review as they may have been included in more than one study sample, although we excluded studies with duplicated datasets. We cannot ignore that there are a limited number of studies (if any) in certain years for some key populations. Most included studies had high study quality with low risk of bias, but there may be other biases that were not assessed. Finally, a high heterogeneity of included studies is noteworthy; subgroup analyses reduced only some heterogeneity. Despite these limitations, this is the first review we know of to estimate proportions of PrEP use in key populations and geographic areas identified in the US national HIV prevention goals by synthesizing identified published self-reported surveys.

Conclusions

This study found increased proportions of self-reported PrEP use as well as PrEP disparities among key populations identified by national HIV goals as vulnerable to HIV infection in the US. MSM and Hispanic/Latinos were more likely to report having ever used PrEP than other key populations (i.e., blacks, PWID, youth, people in the Southern US, transgender women), and the proportions had increased since 2015. On the other hand, people in the Southern US and youth had lower levels of PrEP use and did not demonstrate PrEP use growth in recent years. Furthermore, few studies focused on PWID and non-MSM subgroups, such as black women and non-MSM PWID. One of the goals introduced in the most recent CDC’s Division of HIV/AIDS Prevention’s (DHAP) Strategic Plan 2017–2020 is to reduce HIV-related disparities and health inequity.³³ PrEP may help to reduce disparities in HIV incidence rates if accessible to populations with the highest disparities. This study indicates that disparities in PrEP access/availability may exist, which could contribute to or exacerbate existing disparities in HIV acquisition. Culturally-tailored approaches targeting these most vulnerable populations in addition to approaches targeting providers serving these populations are needed to increase PrEP access and use and thereby reduce disparities in HIV acquisition.

Supplementary Material

Supplementary Material

Additional file 1: Appendix I: PrEP search strategy

Additional file 2: Appendix II: Supplemental Tables and Figures

Supplemental Table 1: Included studies’ populations, proportions, and quality scores by study years (N=95)

Supplemental Table 2: Heterogeneity by study year and key populations groups in US study participants.

Supplemental Table 3: Fixed effects pooled proportion of PrEP use by recall period in US study participants overall and in MSM and non-MSM participants

Additional file 3: Appendix III: List of Included studies (N=95)