Inroads for HIV prevention among men: findings from mixed methods research in the context of the DREAMS partnership in Southern Africa

Abstract

Objective: 

To assess trends in men's HIV risk factors and service use, and their experiences with prevention programming, during an intensive HIV response for adolescent girls and young women and their male partners.

Design: 

Independent cross-sectional surveys in 2016–2017 and 2018 with men in Eswatini (20–34 years-old, n = 1391) and Durban, South Africa (20–40 years-old; n = 1665), complemented by 74 in-depth interviews (IDIs) with men exposed to HIV services/prevention programming.

Methods: 

Survey recruitment was primarily at hot-spot venues. We assessed Round 1–2 trends in HIV risk factors and service use, overall and by HIV risk profiles. IDI respondents were identified via survey responses or program partners.

Results: 

HIV risk factors were prevalent in both countries at each survey round, although there were reductions over time among the highest risk profiles in South Africa. Most men were engaged in HIV services (e.g. nearly two-thirds tested for HIV in the last year at round 2, with large increases in Eswatini). Qualitative data suggest HIV service uptake was facilitated by increased convenience and supportive information/messaging about HIV treatment efficacy. Men described eagerly receiving the information and support offered in HIV prevention programming, and effects on HIV risk reduction and newly engaging in HIV services. However, less than 15% of survey respondents reported being reached by such programming.

Conclusion: 

Important inroads have been made to engage men in HIV services and prevention programming in the two countries, including among the high-risk profiles. Still, improving coverage of comprehensive HIV prevention programming is critical, particularly for men most at risk.

Introduction

Engaging men is a priority for the HIV response in sub-Saharan Africa ^[1], home to over half of the world's people living with HIV ^[2]. Men are more likely than women to die of AIDS-related illnesses, and less likely to test for HIV or initiate antiretroviral therapy (ART) ^[1,3,4], a pattern mirrored in men's lower life expectancies and poorer health-related practices worldwide ^[5,6]. Recent research also shows a pattern of transmission whereby heterosexual men contribute to disproportionately high HIV incidence among adolescent girls and young women (AGYW) ^[7,8] – who, for example, constitute 10% of the world's population but one-quarter of new infections ^[2]. Therefore, intensive efforts have been launched to reduce HIV risk among AGYW and men/male partners of AGYW in the region.

The DREAMS Partnership aims to prevent new HIV infections among AGYW in the highest HIV prevalence sub-national units. DREAMS applies a combination HIV prevention package for AGYW, including interventions that address individual, as well as structural sources of risk ^[9]. Engaging men/male partners is key to addressing AGYW's risk, and DREAMS has three main approaches to do so ^[9]. First and most prominent is reaching men/male partners of AGYW with HIV services – including HIV testing, voluntary medical male circumcision (VMMC), linkage to HIV care, and ART (funded largely via other PEPFAR programming) ^[9]. Second is reaching men/male partners with programming that increases information and skills to reduce HIV risk behaviors, with a particular focus on changing harmful gender norms and mitigating intimate partner violence (IPV). Third is male partner characterization, to identify characteristics of men at highest risk of transmitting HIV to AGYW – and reach them with services and prevention programming ^[10–12]. DREAMS implementation began in 2016 and is now institutionalized in PEPFAR's annual country planning process (COP); some governments and funding bodies have also rolled out complementary DREAMS-like programs.

In many respects, these approaches to engage men reflect trends emerging in recent research and program/policy approaches around engaging men in the HIV response. First, there is increasing emphasis on men as clients who, to access HIV services, require options that best meet their needs ^[13–15]. Community-based HIV services have proven an important alternative to facility-based services for reaching men, including the hardest-to-reach and potentially highest risk ^[16,17]. Second, there is now broad recognition that men's HIV risk and service use behaviors are fundamentally shaped by inequitable and restrictive gender norms, norms that also intersect with other inequalities, for example, based on class ^[18–20]. There is a critical need for primary prevention efforts that incorporate transforming gender norms to reduce HIV-related risk ^[21,22]. Third, there is increasing attention to the fact that men, like women, are not a homogenous group, including with respect to HIV risk ^[1,23]. Understanding heterogeneity in men's HIV risk factors, service use, and response to programming – between and within countries – is critical to effectively targeting and tailoring prevention efforts.

In this implementation science study, we used mixed methods to investigate men's engagement in and response to HIV programming in the context of DREAMS (including intensive efforts to engage men in HIV services) in Eswatini and South Africa. We sought to answer three questions, among samples recruited largely at community hot spot venues to reach men at high risk of acquiring or transmitting HIV. First, did HIV risk factors and service use change among men over the time-period when the intensive HIV response for AGYW and their male partners was rolled out? Second, to what extent was this the case for subgroups of men at highest risk of acquiring/transmitting HIV? Finally, what were men's experiences and perceptions of HIV prevention programming and services? To answer these questions, we conducted independent cross-sectional surveys with men at two timepoints in 2016–2017 and 2018 in Eswatini and Durban, South Africa, complemented by qualitative research with men who recently participated in HIV services and/or prevention programming at the second timepoint.

Methods

Study settings

Study sites were high-HIV burden sub-national units where DREAMS-related activities were implemented. In Eswatini, the research was conducted in 19 Tinkhundla (administrative subdivisions smaller than a district but larger than chiefdoms ^[24]), spanning the country's four regions. In South Africa, the study took place in two informal settlements in Ethekwini (Durban), a district in KwaZulu-Natal province.

Survey sampling, recruitment, and administration

We conducted two cross-sectional surveys with men ages 20–34 years in Eswatini and ages 20–40 years in South Africa. These age ranges reflect the DREAMS target age range for men/male partners of AGYW in each country. In Eswatini, round 1 (R1) surveys were conducted from December 2016 to February 2017 and round 2 (R2) from June to September 2018 (thus with a possible span of 17–22 months between R1 and R2). In South Africa, R1 surveys were conducted from May to September 2017 and R2 from June to August 2018 (possible span of 11–16 months).

To capture a sample of likely high-risk men/male partners of AGYW, respondents were recruited at community hot spot venues identified by key informants as places where AGYW and men regularly meet and socialize, as we have described elsewhere ^[12,25]. These included informal and formal drinking establishments, kiosks/shops, and, in South Africa, taxi ranks and university surrounds. In South Africa, men were also recruited at community-based and facility-based HIV service sites (approximately one-third of the sample per round) to help capture experiences of men who used HIV services. Sampling/recruitment occurred at the same set of venues/service sites at R1 and R2. At each site, interviewers followed study protocols designed to select respondents in a random manner.

Survey data analysis

All analyses were performed in Stata v16 ^[26], and accounted for the clustered survey design. We limited the samples for analyses to men who reported having sex in the last year, given our interest in assessing current HIV risk. We assessed differences in sociodemographic characteristics, HIV risk factors and service use between R1 and R2, in bivariate and multivariate regression analyses. We also used Latent Class Analysis (LCA) to identify HIV risk profiles in Eswatini and South Africa, and to examine changes in HIV risk factors and service use by profile. LCA detects hidden classes or ‘profiles’ within survey data based on multiple observed variables. Models were chosen based on relative-fit statistics and interpretability (whether the latent classes made logical sense and were distinct from each other) ^[27]. Any variables that consistently had similar probabilities across all classes were dropped from models. Final profiles were labeled based on overall level of risk (per HIV risk factors) as well as age (younger/older) as the profiles were distinguishable based on these characteristics, and this seems an intuitive way for readers to understand the profiles.

We conducted multiple-group LCA, with survey round treated as the grouping variable. Following recommended procedures ^[27], we inspected latent class solutions at R1 and R2, noting the same number of interpretable classes with similar patterns of item response probabilities, and minimal differences (e.g. <5%) in sociodemographic characteristics between rounds. Therefore, we constrained all item response probabilities for sociodemographic characteristics to be the same across rounds. We also constrained probabilities of class membership as these too had minimal difference between rounds (<5%). We confirmed adequacy of fit statistics for final LCA models.

We then conducted postestimation analyses to assess HIV service use by profile at each round. Fit statistics confirmed low probabilities of latent class misclassification at both rounds.

Measures

All measures are described in Supplemental Table S1, https://links.lww.com/QAD/C475. Transactional relationships was based on reporting giving at least one item or service (the response categories) ‘mainly so you could start or stay in a sexual relationship’ with a partner, and categorized as less or more resource-intensive ^[28]. Hazardous drinking was assessed by the AUDIT-C ^[29], and endorsement of inequitable gender norms by the Gender Equitable Men scale ^[30,31]. In some cases, continuous/ordinal measures were dichotomized for clearer differentiation between classes.

Qualitative in-depth interview procedures

We conducted IDIs with men from March to December 2018, in the same study areas as surveys. Eligibility criteria included being at least 18 years old and having at least one AGYW partner. Other sampling/recruitment strategies were designed to identify men who had participated in HIV prevention programming/services (including oversampling of men known to be HIV-positive, to capture their experiences). We chose to interview program participants (vs. nonparticipants), to document their experiences with new service delivery and programming approaches. Although critical, a wealth of prior qualitative research (including our own research related to DREAMS ^[32,33]) has focused on barriers to men's HIV prevention behaviors/HIV service uptake. Recruitment was conducted via DREAMS implementing partners and based on survey responses (among participants providing consent to recontact, who were then re-consented). IDIs were conducted by experienced local interviewers (not affiliated with DREAMS) using semi-structured guides. Thematic data analyses followed a team-based, iterative process to arrive at final themes and compare them by country.

Ethical review

Study protocols were approved by the Population Council Institutional Review Board (IRB), the Swaziland National Health Research Review Board (in Eswatini), and the University of KwaZulu-Natal Biomedical Research Ethics Committee (in South Africa). Verbal informed consent was sought for the R1 survey in Eswatini; all other survey and IDI participants provided written informed consent.

Results

Survey findings

In Eswatini, 1391 men ages 20–34 years completed surveys (n = 650 at R1; 741 at R2) and were included in analyses. In Durban, 1665 men ages 20–40 years completed surveys (n = 876 in R1; 789 at R2) and were included in analyses. The refusal rate was less than 5% at each round in each country; this reflects individuals who refused to participate after moving to a more private/quiet location at the site with the interviewer, to begin the informed consent process. Sociodemographic characteristics (Table 1) were largely similar between rounds in each country; statistically significant differences were relatively small in magnitude. We accounted for these differences in sociodemographic characteristics when examining changes over time, as described further below.

Table 1 - Socio-emographic characteristics, HIV risk factors, and service use among men in sequential cross-sectional surveys in Eswatini and South Africa.

	Eswatini				South Africa
	Round 1 (n = 650)	Round 2 (n = 741)	OR/beta	aOR/adj beta	Round 1 (n = 876)	Round 2 (n = 789)	OR/beta	aOR/adj beta
Sociodemographic characteristics
Age – mean (SD, range)	26.5 (4.3, 20–34)	26.4 (4.2, 20–34)	−0.17 (−0.62 to 0.28)	–	27.6 (5.5, 20–40)	28.7 (5.7, 20–40)	1.12∗∗∗ (0.58, 1.65)	–
Currently in school	15.4%	13.5%	0.86 (0.64, 1.16)	–	15.3%	12.9%	0.82 (0.62, 1.09)	–
Education (highest completed)
Some secondary or less	–	–	–	–	21.8%	25.2%	0.72∗∗∗ (0.59, 0.86)	–
Secondary	–	–	–	–	55.5%	60.1%		–
Technical college/University	–	–	–	–	22.7%	14.7%		–
Married/cohabiting	31.4%	24.8%	0.72∗∗ (0.57, 0.91)	–	15.2%	14.7%	0.96 (0.74, 1.26)	–
Employed	54.3%	58.3%	1.18 (0.95, 1.45)	–	59.8%	54.3%	0.80∗ (0.66, 0.97)	–
Urban (vs. rural)	47.5%	49.0%	1.06 (0.86, 1.31)	–	n/a	n/a	n/a	n/a
HIV risk factors
Number of sexual partners in last year
1	51.4%	46.8%	1.05 (0.86, 1.28)	1.00 (0.77, 1.32)	25.9%	39.7%	0.50∗∗∗ (0.42, 0.61)	0.53∗∗∗ (0.42, 0.67)
2–4	33.5%	43.3%			49.9%	47.7%
5 or more	15.1%	9.9%			24.2%	12.7%
Age disparity: % age disparate relationships, or mean (SD) years younger, with last three nonmarital partners	34.7% (n = 473)	34.1% (n = 622)	0.97 (0.78, 1.25)	1.02 (0.73, 1.42)	3.4 years (3.6) (n = 844)	3.4 years (3.6) (n = 704)	−0.02 (−0.38 to 0.34)	−0.30 (−0.53 to −0.06)
Consistent condom use with last 3 nonmarital partners, % (n/N)	37.4% (n = 473)	41.8% (n = 622)	1.20 (0.94, 1.53)	1.19 (0.91, 1.55)	20.9% (n = 845)	25.2% (n = 705)	1.27∗ (1.01, 1.62)	1.29 (0.98, 1.70)
Hazardous drinking	53.2%	46.3%	0.76∗ (0.61, 0.94)	0.76 (0.53, 1.08)	52.4% (n = 870)	43.1%	0.69∗∗∗ (0.57, 0.83)	0.70∗∗ (0.55, 0.88)
Transactional relationships
None	–	57.6%	–	–	43.5%	52.9%	0.78∗∗ (0.65, 0.94)	0.81 (0.63, 1.04)
Less resource-intensive	–	30.2%	–	–	44.3%	32.6%
More resource-intensive	–	12.2%	–	–	12.2%	14.6%
Endorsement of inequitable gender norms	22.6%	20.9%	0.91 (0.70, 1.17)	0.91 (0.64, 1.29)	24.5% (n = 869)	12.6%	0.44∗∗∗ (0.34, 0.57)	0.47∗∗∗ (0.35, 0.62)
IPV perpetration	14.6%	14.7%	1.01 (0.75, 1.36)	1.04 (0.68, 1.57)	20.0% (n = 875)	15.9% (n = 785)	0.76∗ (0.59, 0.98)	0.76∗ (0.59, 0.97)
HIV service use/prevention program exposure
Tested for HIV (in last 12 months)a	41.5%	61.8%	2.28∗∗∗ (1.84, 2.82)	2.26∗∗∗ (1.74, 2.93)	65.1% (n = 547)	66.0% (n = 470)	1.04 (0.80, 1.35)	1.04 (0.81, 1.33)
Is circumcised	37.4%	42.9%	1.26∗ (1.02, 1.56)	1.28 (0.94, 1.74)	62.8% (n = 871)	58.6%	0.84 (0.69, 1.02)	0.94 (0.68, 1.30)
Currently taking antiretroviral therapy (ART) (among men reporting HIV-positive status)	74.4% (n = 23)	87.5% (n = 24)	Low power	Low power	91.3% (n = 69)	99.2% (n = 133)	12.57∗ (1.48, 106.65)	9.80∗∗ (2.41, 39.83)
Aware that HIV treatment can prevent onward transmissionb	–	52.8%	–	–	49.2%	66.6%	2.06∗∗∗ (1.69, 2.51)	2.16∗∗∗ (1.60, 2.92)
Attended HIV prevention meetingsc	13.9% (in last year)	9.3% (in last 6m)	0.64∗∗ (0.46, 0.89)	0.67 (0.37, 1.21)	2.7% (in last year, n = 865)	3.5% (in last year, n = 601)	1.33 (0.73, 2.42)	1.32 (0.69, 2.57)
Number of HIV prevention meetings attended – mean (range)	3.6 (1–25) (n = 89)	3.1 (1–20) (n = 65)	Low power	Low power	1.6 (1–10) (n = 22)	2.0 (1–10) (n = 21)	Low power	Low power

Dashes (−) indicate unmeasured factors at R1. Multivariate analyses controlled for sociodemographic characteristics (age, schooling, marital, and employment status) with Inkhundla in Eswatini and site type in South Africa. n = 537 in Eswatini and 98 in South Africa were dropped from the sample as they reported no partners in the last year (≥1 partner was an eligibility criterion for this analysis). Eighty-three percent of men in Eswatini and 73% in South Africa reported having at least one partner ages 15–24 years in the last year, less than 1% in each country reported having sex with a man in the last year.

^aIn South Africa, assessed among venue-based sample only (n = 582 at R1, 478 at R2) as service-based sample included many coming for HIV testing.

^bResponded ‘yes’ to Can taking ART reduce the risk of transmitting the HIV/AIDS virus to another person?’.

^cDefined as answering ‘yes’ to having attended any ‘community, group, and/or one-on-one meetings about HIV’.

^∗P less than 0.05.

^∗∗P less than 0.01.

^∗∗∗P less than 0.001.

Overall change in HIV risk factors and service use over time

In Eswatini, all HIV risk factors were prevalent at R1 (Table 1), and while most shifts over time were towards lower risk, no changes reached statistical significance (including after adjusting for sociodemographic characteristics). The proportion reporting HIV testing in the last year rose substantially from R1 to R2, from 42 to 62% (P < 0.001). Circumcision (status) also increased, from 37 to 43% (not significant in multivariate analyses). Among men self-reporting being HIV-positive, current ART use rose from 74 to 88% (small sample size precluded testing statistical significance of the difference). Regarding awareness of treatment as prevention (TasP) at R2 (not assessed at R1), 53% responded ‘yes’ when asked ‘Can taking ART reduce the risk of transmitting the HIV/AIDS virus to another person?’. Recent participation in HIV prevention meetings was less than 15% at each round, with participants attending three to four meetings on average (range 1–25).

In South Africa, all HIV risk factors were also prevalent at R1. From R1 to R2, there were declines in numbers of sexual partners in the last year from R1 to R2 (P < 0.001) – including a decline from 24 to 13% in the proportion with 5+ partners, inequitable views towards gender norms (from 25% to 13% from R1 to R2; P < 0.001), hazardous drinking (52–43% from R1 to R2; P < 0.001), and transactional sexual relationships (44–53% with none, P < 0.01). Additionally, consistent condom use increased from 21 to 25% (P < 0.05). HIV testing in the last year remained at about 65% from R1 to R2, comparable with Eswatini at R2. Circumcision status also remained at about 60%, higher than in Eswatini at R2. The proportion of HIV-positive men currently on ART rose from 91 to 99% (P < 0.01). Awareness of TasP was low at R1, at 49%, but increased significantly to 67% at R2 (P < 0.001). Finally, participation in HIV-prevention meetings was at 3–4% in R1 and R2, with participants joining two meetings on average (range 1–10).

In both countries, low levels of participation in HIV prevention meetings precluded assessing associations between such participation and HIV risk factors.

Changes over time by HIV risk profile in Eswatini

In each country, we found four distinct HIV risk profiles, constituting a younger and older profile at markedly high risk. Fit statistics for the final LCA models can be found in Supplemental Table S2, https://links.lww.com/QAD/C475. In Eswatini (Table 2), the younger and older high-risk profiles both had markedly higher probabilities than their lower risk counterparts of having two to four and more than five sexual partners, age-disparate partners, hazardous drinking, and inequitable views towards gender norms, as well as lower probabilities of consistent condom use. The younger high-risk profile (constituting about 26% of the sample) was 24 years old on average, unmarried, with a lower probability of being employed compared with most other profiles. The older high-risk profile (constituting 21% of the sample) was 30 years old on average and likely unmarried (unlike the older low risk of whom nearly 70% were married) and employed. In addition to other prevalent HIV risk factors, over three-quarters of this profile had age-disparate relationships. Notably, profiles were largely similar across urban and rural areas (leading us to drop this variable from the LCA model).

Table 2 - Eswatini: changes by HIV risk profile from round 1 (n = 650) to round 2 (n = 741) among men ages 20–34 years.

	Younger moderate risk			Younger high risk			Older low risk			Older high risk
Probability of class membership	28.5%			25.7%			24.7%			21.1%
Sociodemographic characteristics
Age (mean)	22.5 years			24.1 years			30.2 years			30.2 years
Married/cohabiting	9.5%			17.3%			68.7%			17.9%
Employed	28.9%			49.6%			85.3%			68.1%

HIV risk factors	R1	R2	Difference	R1	R2	Difference	R1	R2	Difference	R1	R2	Difference
Number of sexual partners in last year
1	61.0%	56.4%	−4.6%	27.3%	32.6%	5.3%	75.8%	66.2%	−9.6%	36.5%	30.1%	−6.4%
2–4	32.5%	34.6%	2.1%	42.2%	56.2%	14.0%	16.8%	27.3%	10.5%	45.3%	57.0%	11.7%
5 or more	6.4%	9.0%	2.6%	30.5%	11.3%	−19.2%∗	7.4%	6.6%	−0.8%	18.2%	12.9%	−5.2%
Age disparate relationships (≥5 years age difference) with last three nonmarital partners	11.6%	16.7%	5.1%	11.6%	7.8%	−3.9%	53.3%	68.0%	14.7%	78.5%	69.6%	−8.8%
Consistent condom use	59.2%	66.4%	7.2%	8.9%	16.4%	7.5%	57.7%	71.0%	13.3%	25.1%	21.5%	−3.6%
Hazardous drinking	29.0%	15.5%	−13.5%	74.1%	69.6%	−4.5%	57.4%	34.1%	−23.3%∗	56.3%	72.4%	16.1%
Endorsement of inequitable gender norms	22%	19%	−3%	26%	24%	−2%	22%	20%	−2%	21%	21%	1%

HIV service use/prevention program exposure (postestimation analyses)	R1 (n = 185)	R2 (n = 217)	aOR	R1 (n = 156)	R2 (n = 199)	aOR	R1 (n = 175)	R2 (n = 184)	aOR	R1 (n = 134)	R2 (n = 141)	aOR
Tested for HIV (in last 12 months)	43.8%	61.1%	2.20∗∗∗ (1.38, 3.51)	43.0%	57.8%	1.82∗ (1.15, 2.89)	42.3%	65.2%	2.56∗∗∗ (1.62, 4.02)	35.8%	61.0%	2.80∗∗ (1.63, 4.02)
Is circumcised	38.4%	47.9%	1.48 (0.89, 2.44)	39.7%	42.7%	1.13 (0.72, 1.77)	41.7%	42.4%	1.03 (0.65, 1.62)	27.6%	36.2%	1.49 (0.80, 2.75)
Aware that HIV treatment can prevent onward transmissiona	–	49.3%	–	–	56.7%	–	–	59.2%	–	–	56.7%	–
Attended HIV prevention meetingsb	15.1%	9.2%	0.57 (0.25, 1.32)	9.6%	6.0%	0.60 (0.28, 1.28)	16.0%	12.5%	0.75 (0.32, 1.73)	14.2%	9.9%	0.67 (0.24, 1.86)

All adjusted analyses controlled for Tinkhundla using SVY function. Dashes (−) indicate unmeasured factors at R1. Wald tests were used to assess significance of differences in parameters for HIV risk factors between rounds. For number of sexual partners, the asterisk indicates that the parameter representing ‘1’ and ‘2–4’ vs. ‘5 or more’ significantly differed between rounds. Current use of ART is not reported given low sample size of men reporting an HIV-positive status. Fit statistics for this model: AIC = 19 602, BIC = 19 943, entropy = 0.61 (more detailed information is available in Supplemental Table S2, https://links.lww.com/QAD/C475).

^aResponded ‘yes’ to Can taking ART reduce the risk of transmitting the HIV/AIDS virus to another person?’

^bDefined as answering ‘yes’ to having attended any ‘community, group, and/or one-on-one meetings about HIV’

^∗P less than 0.05.

^∗∗P less than 0.01.

^∗∗∗P less than 0.001.

When assessing changes in HIV risk factors between R1 and R2 (after constraining sociodemographic characteristics to be equal across rounds), few changes reached statistical significance. Among the younger high-risk profile, the proportion having more than five partners decreased from 31 to 11% (P < 0.05). Most other shifts, while nonsignificant, were towards lower HIV risk across the profiles.

In postestimation analyses, HIV testing in the last year, at 38–51% across profiles at R1, increased over time by about 15–25% in each profile (all statistically significant). Being circumcised (status) also increased in nearly all profiles, albeit not significantly. Participation in HIV prevention meetings in the last year was less than 20% for each profile.

Changes over time by HIV risk profile in South Africa

In Durban (Table 3), the younger and older high-risk profiles were characterized by very high prevalence of HIV risk factors, including probabilities of having two to four and more than five sexual partners, transactional relationships, hazardous drinking, and inequitable views towards gender norms, and, among the older high-risk, high age difference with partners. Of note, condom use was similarly low between profiles (and was thus dropped from the LCA models) at about 20–25% (data not shown). The younger high-risk profile (constituting 21% of the sample) was 27 years old on average and had a high probability of employment (unlike in Eswatini). The older high-risk profile (also 21% of the sample) was 36 years old on average and also likely employed, and, unlike in Eswatini, was most likely of the profiles to be married (although only at 38%).

Table 3 - South Africa: changes by HIV risk profile from round 1 (n = 876) to round 2 (n = 789) among men ages 20–40 years.

	Younger moderate risk	Younger high risk	Older low risk	Older high risk
Probability of class membership	35.4%	21.4%	21.9%	21.4%
Sociodemographic characteristics
Age (mean)	22.7 years	27.4 years	29.7 years	36.2 years
Education (highest completed)
Some secondary or less	17.1%	17.1%	26.8%	36.8%
Secondary	54.3%	67.4%	60.5%	50.5%
Technical college/University	28.6%	15.5%	12.7%	12.6%
Employed	23.4%	73.9%	75.4%	77.8%
Married/cohabiting	2.4%	5.9%	21.3%	38.2%

HIV risk factors	R1	R2	Difference	R1	R2	Difference	R1	R2	Difference	R1	R2	Difference
Number of sexual partners in last year
1	22.8%	40.4%	18%∗∗∗	2.6%	3.4%	0.8%	50.2%	55.1%	4.8%	32.8%	54.1%	21%∗∗
2–4	52.5%	46.9%	−6%	55.8%	75.0%	19%	40.4%	36.8%	−3.6%	48.3%	36.4%	−12%
5 or more	24.8%	12.7%	−12%∗∗	41.5%	21.7%	−20%∗	9.3%	8.1%	−1.2%	18.9%	9.5%	−9%∗
Age difference with last three nonmarital partners (mean years younger)	1.1	1.4	0.3	3.6	3.6	0.0	3.6	3.4	−0.2	8.1	6.7	−1.4∗∗
Transactional relationships
None	47.3%	56.9%	10%	5.1%	22.3%	17%	72.5%	68.9%	−3.6%	49.5%	57.6%	8%
Less resource-intensive	47.0%	36.5%	−11%∗	77.0%	56.8%	−20%∗	17.5%	15.6%	−1.9%	29.6%	23.1%	−7%
More resource-intensive	5.7%	6.6%	1%	17.9%	20.9%	3%	10.0%	15.5%	5.5%	20.9%	19.3%	−2%
Hazardous drinking	43.8%	37.9%	−6%	72.0%	64.2%	−8%	42.1%	34.9%	−7.2%	57.8%	40.4%	−17%∗∗
Endorsement of inequitable gender norms	24.6%	17.0%	−8%∗	39.6%	3.5%	−36%∗∗∗	6.9%	13.2%	6.3%	26.2%	13.3%	−13%∗∗

HIV service use/prevention program exposure (postestimation analyses)	R1 (n = 325)	R2 (n = 267)	aOR	R1 (n = 210)	R2 (n = 156)	aOR	R1 (n = 177)	R2 (n = 173)	aOR	R1 (n = 164)	R2 (n = 193)	aOR
Tested for HIV (in last 12 months)a	67.4% (n = 206)	65.7% (n = 175)	0.92 (0.54, 1.59)	66.4% (n = 140)	66.0% (n = 106)	0.98 (0.59, 1.65)	60.0% (n = 105)	71.2% (n = 94)	1.65 (0.88, 3.12)	63.5% (n = 96)	61.1% (n = 95)	0.90 (0.49, 1.66)
Is circumcised	75.3% (n = 324)	70.4%	0.78 (0.46, 1.33)	58.5% (n = 207)	60.9%	1.11 (0.75, 1.64)	60.5% (n = 177)	49.7%	0.65∗ (0.43, 0.97)	46.0% (n = 163)	48.2%	1.09 (0.59, 2.01)
Currently taking ART (among men reporting HIV-positive status)	87.5% (n = 8)	100% (n = 20)	Low power	88.2% (n = 17)	100% (n = 15)	Low power	95.0% (n = 20)	100% (n = 41)	Low power	91.7% (n = 24)	98.2% (n = 57)	Low power
Aware that HIV treatment can prevent onward transmissionb	47.2%	65.9%	2.16∗∗ (1.44, 3.24)	46.4%	69.9%	2.68∗∗ (1.40, 5.13)	54.8%	60.5%	1.26 (0.70, 2.28)	50.6%	70.5%	2.33∗ (1.20, 4.54)
Attended HIV prevention meetingsc	1.5% (n = 323)	3.6% (n = 195)	2.37∗ (1.15, 4.89)	3.4% (n = 206)	7.8% (n = 115)	2.41∗ (1.12, 5.21)	2.8% (n = 176)	2.9% (n = 139)	1.01 (0.21, 4.98)	3.8% (n = 160)	0.7% (n = 152)	0.17 (0.02, 1.46)

All adjusted analyses controlled for site type using SVY function. Wald tests were used to assess if HIV risk factor parameters significantly differed between rounds. Fit statistics for this model: AIC = 34 644, BIC = 35 094, entropy = 0.73 (more detailed information is available in Supplemental Table S2, https://links.lww.com/QAD/C475).

^aAmong venue-based sample only, as service-based sample included many coming for HIV testing.

^bResponded ‘yes’ to ‘Can taking ART reduce the risk of transmitting the HIV/AIDS virus to another person?’

^cDefined as answering ‘yes’ to having attended any ‘community, group, and/or one-on-one meetings about HIV’.

^∗P less than 0.05.

^∗∗P less than 0.01.

^∗∗∗P less than 0.001.

From R1 to R2, there were significant declines in several risk factors among the higher risk profiles. Each had decreased probabilities of having multiple partners – including more than 5 partners (<0.05) – and endorsing inequitable views towards gender norms (P < 0.001 to P < 0.01). The younger high-risk profile was also less likely at R2 to have less-resource-intensive transactional relationships (P < 0.05). Finally, the older high-risk profile had a decreased mean age difference with partners as well as lower probability of hazardous drinking (both P < 0.01).

HIV testing in the last year started at about 60–70% in each profile at R1 and did not increase by R2. ART use among men reporting living with HIV, and TasP awareness, increased most in the higher risk profiles. Participation in HIV prevention meetings in the last year was less than 10% in any profile at R1 and R2.

Qualitative findings: men's perceptions of HIV prevention programming and services

We conducted IDIs with 26 men in Eswatini and 48 in South Africa. Per sampling/eligibility criteria, all had recently participated in HIV services and/or HIV prevention programming. Mean age was 27 years in Eswatini and 26 in South Africa (range 20–40 across the countries). Less than 15% of men in either country was married/cohabiting. Many described a history of substantial HIV risk behaviors, including having multiple sex partners and perpetrating IPV. About one-quarter of men in each country were HIV-positive (as noted above, men living with HIV were over-sampled).

Key themes that emerged from these data are described below. Illustrative quotes for each theme are included in Table 4. Findings were markedly similar between the countries; any findings unique to a country are noted.

Table 4 - Illustrative quotes from in-depth interviews with men in Eswatini and South Africa.

Themes	Quotes
HIV prevention programming was eagerly received by men and supported positive behavior change.	We talked about various topics…[like] about ways to keep relationships alive and happy. Here we were told that we need to stay away from this ‘men must control everything’ attitude because it brings gender-based violence in our communities…These meetings are like information hubs…[like] I’ve learned that one must always have their [HIV] status checked because it is better to know. (South Africa, age 24)…we are always out at night…[and] you find yourself having sex outside the bar…you see I stay at [___], so those things are happening around here. [Separate segment] [Stepping Stones] was for both men and women…I like the fact that we were all communicating. We are from the same community and that class helped because some people we did not know that much …but we were able to get to know each other and understand each other… It influenced me a lot…They helped us that we should know our status, so that you can work and in your relationship. There is nothing you cannot mend. (South Africa, age 27)They teach us a lot about HIV prevention, abuse, respect and a lot more…It made me learn a lot of things from the other young men…it also helps in changing one's behavior from bad to good. (Eswatini, age 22)
Men's-only meetings played particular role: coping with life challenges.	The ‘Men to Men’ program is there to listen, share and provide coping skills for these issues men are facing. So, we learn a lot, we share our daily struggles with our peers. The more we hear each other struggles, the more we realize that we are not alone…We [also] discuss many topics like discipline, love, relationship and health issues, all of which are aimed at men. (South Africa, age 24)
	[We] share experiences, defeats and triumphs everyone has encountered. I think this is done to give each other tips on how to get through life. Also, we will be chatting to each other about diseases young people, suffer with and how these can be cured or treated. (South Africa, age 20)…we discussed mostly about life, how to avoid the diseases as young people and how to look after ourselves as men…they looked at the future and helped you according to your needs …you get to know how other people live their lives. (Eswatini, age 25)…we were talking as men alone…I felt good and comfortable because it was only us guys and we spoke about anything without any reservation, we felt free. We talked about everything ranging from relationships to HIV and how we can help ourselves regarding these things. (Eswatini, age 24)
Comprehensive prevention programming supported men to link to HIV services.	People who have stressed the importance of testing for HIV are your ‘Stepping Stones’ crew…We discussed many topics around disease and health awareness…[and] I like the fact that these meetings teach us about way we can live in diverse community harmony…It has opened my mind up to many things, I now appreciate and know the significance of these services, like VMMC… (South Africa, age 24)[They encourage testing by] giving a person hope when we meet as a community. In these meetings there is usually someone who gives out the information to take care and support others. (Eswatini, age 25)
For some, men's meetings were an important entry-point for engaging in HIV services.	[Men to Men] was created to take care of men's needs specifically…even those who do not want to test at clinics will be able to go and test. Sometimes we need those men's groups where everyone will be free to talk about our problems. (South Africa, age 33)We were around 18 members, it was only men. I liked that as guys we were able to share our own experiences…I learned that one must always have their status checked because it's better to know…Also, being infected with HIV doesn’t mean you should stop having goals because life continues with or without the virus. [And] it was Men to Men that referred me to [get circumcised]. (South Africa, age 23)
Convenient options facilitated routine HIV service use.	Nowadays they provide door to door visits, they provide mobile units talking and calling people to test and know their HIV statuses…it really helps us as men. There is a huge change. (Eswatini, age 24)Now you don’t have to wait to go to the health centers for testing. There are more, more, and more of the mobile testing available around. (Eswatini, age 29)I got the [self-test] from the college…a lady standing at the main gate…I thought I should do it because it had been a while since I last did HIV testing. [Testing negative] was a relief, so from then I had to make sure that I protect myself. (South Africa, age 20)Male circumcision is brought to the communities, unlike before when we had to go elsewhere. (Eswatini, age 29)
Supportive messaging about the effectiveness of early HIV diagnosis and treatment was also critical.	The counsellor said that there is no difference between a person living with HIV and a negative one, it is all in the mind of the individual. (South Africa, age 31, HIV-positive)The way they test you and the way they talk with you…[they] give you certain comfort plans and tell you that it is not the end of the world and you can live for more than forty years if you follow certain rules, you will live your life to the fullest. (South Africa, age 21)…what was important was the fact that they offered counselling once one decided to get tested…They told us that the results were not a death sentence if you tested positive, that you can take treatment. (Eswatini, age 24, HIV-positive)I only know that [ARVs] keep your immune system strong, I don’t know about the transmission part. I think that when you know your HIV positive status you cannot be having sex without a condom. (Eswatini, age 28, HIV-positive)

HIV prevention programming was eagerly received by men and supported positive behavior change

Men consistently described welcoming the opportunity to participate in HIV prevention programming, which ranged from a few brief sessions to multiple more intensive sessions held over many weeks. Commonly reported effects of prevention programs on men's HIV risk factors were reductions in numbers of sexual partners, improved communication and conflict resolution in relationships, and reduced violence in relationships. These changes were often described as arising out of the foundation of understanding the harm of control and power in relationships and importance of respect for others in one's community, particularly for more intensive, and/or mixed gender, programs.

Men-only meetings played particular role: coping with life challenges

Participants of men's meetings consistently described these meetings as providing opportunities for men to discuss the challenges they face and learn positive coping strategies, in addition to encouraging HIV risk reduction and linkage to services. Men who participated in men's meetings in each country rarely also participated in programming that included women.

Comprehensive prevention programming supported men to link to HIV services, including those previously reluctant to do so

Prevention programming that provided HIV information (reducing number of sexual partners; condom use) and referrals to HIV services, as well as critical reflection around dynamics and power/violence in intimate relationships and the importance of respect for others in the community, seemed to provide a foundation of information and support that facilitated engagement in HIV services, particularly among those formerly reluctant to do so.

For some, men's meetings were an important entry-point for engaging in HIV services

By building trust because of recognition of men's need for mutual support around life challenges, men's meetings seemed to serve as an entry-point to learn about and link to HIV services. This was particularly the case for men at higher risk and/or those reluctant to join sessions they anticipated would be dominated by women.

Convenient options facilitated routine HIV service use

Most men described a habit of routinely testing for HIV, sometimes contrasting this with their first-ever HIV test as a difficult and sometimes delayed decision. In addition, those testing positive described being on ART. This was often attributed to the convenient and varied options now available. For example, along with facility-based testing, there were mobile, door-to-door, workplace, and self-testing options. This expansion of convenient options was often described as a very recent change in Eswatini, where our survey also showed a large uptick in testing, while in South Africa it was described as the way things are now.

Supportive messaging about the effectiveness of early HIV diagnosis and treatment was also critical

This information/messaging was often communicated to men during pretest counseling, and was consistently described as easing men's fears around testing and subsequently linking to care if HIV-positive. However, when we asked, few men living with HIV in Eswatini and South Africa knew about the effectiveness of early HIV diagnosis and treatment for preventing transmitting HIV to their partners (i.e., treatment as prevention).

Discussion

Findings from this mixed methods study suggest that important inroads have been made to engage men in HIV services and prevention programming, including among those at highest risk of acquiring HIV or transmitting HIV to their partners. A majority of men were engaged in HIV testing, VMMC and care and treatment (including substantial recent improvements in Eswatini), which qualitative data suggested is facilitated by convenience of services and supportive information and messaging about the effectiveness and benefits of early HIV diagnosis and treatment. However, it is also clear from analyses by HIV risk profile that men at highest risk are not being adequately engaged in these services. Qualitative findings suggested that comprehensive HIV prevention programming (i.e. that includes HIV-related information and links to services as well as addressing gender norms and stigma) could serve as a foundation from which men feel informed about and supported to access services. Extensive evidence ^[9,21,34,35] suggests that such prevention programming will also be critical to reduce the very high levels of multiple HIV risk factors evident from both rounds of surveys (even after reductions in South Africa). In qualitative interviews, men described eagerly receiving the information and support offered and effects on HIV risk reduction and averting violence in their relationships.

The trends towards increasing HIV testing and circumcision status in Eswatini, and already relatively high levels of testing and circumcision status in Durban, are also reflected in population-based assessments ^[36–38] and suggest effectiveness of differentiated service delivery and outreach efforts in reaching most men. Qualitative findings suggested that for men who had already overcome the hurdle of their first HIV test, convenient community-based testing options were important to encourage routine testing. Men in Eswatini described the recent rapid expansion of such services, perhaps explaining the uptake in testing found in the surveys. Still, 50–60% is well below the UNAIDS goal of at least 90–95% status awareness, and is particularly concerning for the high-risk profiles. Such a ‘testing gap’ among the highest risk men is likely fueling continued high community HIV incidence despite overall increases along the HIV care cascade, as emerging evidence from recent Test-and-Treat trials suggests ^[39].

The substantial reductions in multiple HIV risk factors in South Africa among the higher risk profiles are harder to explain (including by our qualitative data), given low exposure to HIV prevention programming that explicitly targets such changes. However, recent research found similar ‘secular trends’ in men's HIV risk factors (inequitable gender norms and IPV) in a community-based trial similarly employing cross-sectional surveys in South Africa from 2014 to 2018 (in rural Mpumalanga province, just north of KZN) ^[40]. Complementary qualitative research suggested this was because of rapidly escalating access to media, especially because of satellite TV and smartphones, which increased exposure to serial dramas modeling more equitable relationships and negatively portraying violence ^[40]. This may also have time-use effects on reducing numbers of partners and hazardous drinking. We are unable to say whether this is the case here as we did not assess media exposure at baseline (although when assessed at endline over 75% of men in each country owned a smartphone and used social media (data not shown)). Broad changes in the social context, perhaps especially in media access, warrant greater attention within the HIV response and could serve as opportunity to for programs to solidify norm changes and gain traction towards behavior change ^[40].

Taken together, our findings have several implications for the HIV responses in each country, and perhaps the region. First, differentiated, client-centered services should continue to be provided as a complement to facility-based services, and should continue to prioritize sharing information and supportive messaging around HIV treatment efficacy ^[41]. Such efforts are already well underway, the most prominent example being the PEPFAR-supported MenStar Coalition ^[15]. However, the importance of reaching the highest risk subgroups with HIV services remains. Our qualitative findings suggest supportive prevention programming that invites and meaningfully engages men (and male partners of AGYW), including in opportunities to discuss life challenges with other men, is promising for supporting uptake of services for previously reluctant men. Considerations about informing and engaging the highest risk men will become all-the-more important as preexposure prophylaxis (PrEP) access and options (including long-acting injectable PrEP) become more widely available in the region. It is also clear from our findings that TasP/Undetectable = Untransmittable (U = U) awareness has a long way to go in certain countries. Such awareness is critically important as part of promoting men's informed decision-making about their own health and preventing transmission to their partners, as we have argued previously ^[42,43].

However, given extremely high levels of multiple HIV risk factors, scaling up comprehensive, evidence-based HIV prevention programming is imperative alongside biomedical prevention approaches ^[9,21,34,35]. DREAMS purposefully built into its design community mobilization with men/male partners of AGYW. Our findings regarding the low participation in meetings related to HIV suggests that, as of 2018, few men may have been participating in such efforts. It is unclear the extent to which this is because of low volume of sessions being offered (information that is not readily accessible), and/or to substantial numbers of men refusing to participate. However, DREAMS implementing partner organizations and donors have not reported largescale refusal of men to participate when such programming was offered to them. Critical questions remain about the optimal content and structure of prevention programming. Reviews of interventions suggest that alongside health education, effective programs include critical reflection about gender roles and risk, last beyond a few sessions, and engage both men and women (together and/or in separate groups) ^[22,44]. Our qualitative findings not only support the importance of direct engagement of men alongside women to improve communication and mutual support but also the opportunity for male-only groups as safe spaces to express worries, share personal stories, and seek advice about life challenges. However, an important consideration is the potential consequences (e.g., of inadvertently reinforcing certain inequitable gender norms) if men only participate in men's only meetings, without also engaging in dialogues that include women ^[44–46].

This study had several limitations. First, responses were based on self-report, potentially introducing social desirability bias. Second, the sequential cross-sectional design resulted in certain differences in sociodemographic characteristics between rounds. In addition, the time between survey rounds was relatively brief (∼1.5 years in Eswatini; 1 year in South Africa). However, these timepoints also fell early and late in the initial 2-year DREAMS implementation period and intensive programming to get high-risk men into HIV services. The specific number of months between rounds may also have differed somewhat by venue/service site. Third, limiting our in-depth interview sample to men who had participated in HIV services/programming may engender biases in findings as those who participated may have been more likely to positively perceive these services/programming. Nonetheless, the male partners we interviewed likely represent the ‘right’ men to reach in future efforts to reduce HIV incidence, including among AGYW as many reported a history of HIV risk behaviors and/or were living with HIV. Finally, results may neither be fully representative of men attending hot spots/service sites in the study areas, nor generalizable to men beyond those sites or to other areas within each country. However, the marked similarities between HIV risk profiles in Eswatini and South Africa also suggest similar patterns of HIV risk across contexts, an important area for future research.

In conclusion, findings from this study highlight critical inroads as well as lessons learned with regard to engaging men in the HIV response. They also underline the need for our field to continue to grapple with difficult questions around how we can build a more holistic and inclusive HIV response that meets the information, healthcare, and psychosocial support needs of the most vulnerable subgroups, and positively influences and capitalizes on existing shifts towards healthier and more gender equitable social contexts. We hope this study has contributed actionable evidence to further strengthen the HIV response moving forward.

Acknowledgements

We thank all study participants for sharing their views with us, and data collectors for their work. We also appreciate the contributions of the following individuals during data collection and analysis: Zahra Reynolds, Elsa Marshall, Dominique O’Donnell, Bheki Mamba, Feziwe Makhubu, Vimbai Tsododo, Nrupa Jani, Pamela Keilig, John Mark Wiginton, and Sangram Patel. Finally, we thank stakeholders from the following organizations/entities: in Eswatini, the National Emergency Response Council on HIV and AIDS and Swaziland National AIDS Programme; in South Africa, the Government of South Africa Department of Health and South Africa National AIDS Council; the US President's Plan for AIDS Relief (PEPFAR), the U.S. Agency for International Development (USAID), and DREAMS implementing partner organizations in Eswatini and South Africa.

This study was supported by the Bill & Melinda Gates Foundation, Grant Numbers OPP1150068 and OPP1136778.

Author contributions: study conception: A.G., J.P., S.M.; study design/protocol development: A.G., J.P., S.M., J.O.; data collection: P.S., B.L., C.C., D.K., G.S., A.K., L.A., J.O.; data analysis and article preparation: C.H., A.G., J.P., G.S., A.K., L.A., J.O.; article review and final version approval: all authors.

Conflicts of interest

There are no conflicts of interest.