Self-Reported Combination HIV Prevention Strategies Enacted by a Prospective Cohort of Midlife and Older Men Who Have Sex with Men in the United States: A Latent Class Analysis

Abstract

Insights into combination HIV prevention (CHP) strategies to reduce HIV incidence among midlife and older adult men who have sex with men (MSM) are limited. The current study is a secondary data analysis evaluating CHP in a sample of sexually active midlife and older adult MSM (N = 566) from the Multicenter AIDS Cohort Study Healthy Aging Substudy. Stratified by HIV serostatus, we used latent class analyses to identify CHP classes based on self-reported sociobehavioral and biobehavioral prevention strategies that participants and their male partners used in the prior 6 months. We identified three CHP classes among men living without HIV (MLWOH), including the following: high CHP overall (43.0%), high anal sex abstention (15.0%), and low prevention overall (42.0%). Among men living with HIV (MLWH), we identified four CHP classes, including the following: high CHP overall (20.9%), high CHP/low condom use (27.1%), high condom reliance (22.3%), and low prevention overall (29.7%). There were small differences by sociodemographic characteristics and sexual behavior practices between the classes; however, poppers use was often linked to being in high CHP groups. Our findings support that CHP is not one-size-fits-all for midlife and older adult MSM. There remains a need to scale up clinical providers' sexual health communication practices to assist midlife and older MSM incorporate prevention strategies, particularly biobehavioral prevention strategies that align with their patients' lived experiences.

Introduction

HIV prevention experts have widely supported combination HIV prevention (CHP) as a critical strategy for reducing incident cases of HIV among midlife and older (age 40+ years) adult cisgender men who have sex with men (MSM).^1,2 CHP describes the multitude of strategies (e.g., condom use and HIV testing) undertaken by persons to minimize their risk of HIV acquisition or transmission.^3,4 Community patterns and employment of CHPs have undergone surveillance over the last decade as the HIV prevention toolbox evolved, for example, biobehavioral strategies including pre- and postexposure prophylaxis (PrEP/PEP) and HIV treatment as prevention (TasP).⁵

Awareness and uptake of biobehavioral HIV prevention strategies continue to increase among MSM populations.^6,7 However, these strategies have not prioritized midlife and older adult MSM in the United States,⁸ despite that more than a third of new HIV diagnoses were reported in MSM 35 years and older in 2019.⁹ A recent study identified notable age disparities in PrEP uptake among men, with the lowest PrEP uptake among those aged 55–69 years irrespective of race.¹⁰ These disparities may be attributed to HIV prevention efforts having a prioritized focus on youth populations and low perceptions of HIV risk among those in midlife and older adulthood and their providers.¹¹

There remain limited studies evaluating nonoccupational PEP uptake among cisgender MSM; however, prior studies in high-income countries have provided mixed findings on age-related disparities in PEP awareness. Most studies have found statistically nonsignificant differences between age cohorts, while some have observed higher levels of PEP awareness within increasing age.^12–15

Developed by the Prevention Access Campaign, TasP Undetectable = Untransmittable (U = U) messaging has been implemented globally to educate communities that people living with HIV (PLHIV) who are taking HIV medications with an undetectable viral load have virtually no risk of transmitting the virus to a sexual partner.^16–18 Numerous recent studies have indicated high levels of awareness of U = U among MSM,^6,19–21 but there remain varying levels of trust and acceptability of TasP as an HIV prevention method.²² Further, studies to date have provided little insight into age-generational differences in the acceptability of TasP.

Few studies have evaluated the integration of biobehavioral HIV prevention methods alongside other commonly used sociobehavioral strategies at the individual level (i.e., condom use or alternative methods to anal penetrative intercourse).²³ In a US study, a higher proportion of Black and Latino MSM preferred to use both condoms and PrEP than one method versus the other. However, when comparing preferences for using both methods against using condoms only, men who reported any recent condomless sex were less likely to prefer both methods than men who reported no recent condomless sex.²⁴

In a recent study of CHP conducted with a mixed-HIV serostatus sample of gay and bisexual men aged 16 years and older (mean = 38 years, standard deviation = 14) in Montreal, Canada, Doyle et al observed four classes of prevention behaviors [i.e., low use of prevention, condoms, seroadaptive behavior (selecting partners based on HIV serostatus or sexual positioning), and biomedical] among men living without HIV (MLWOH) and three classes of prevention behaviors [e.g., mainly antiretroviral treatment (ART) with viral suppression, ART with viral suppression and condoms, and ART with viral suppression and seroadaptive behavior] among men living with HIV (MLWH).²³

To our knowledge, no prior study has focused evaluating CHP among midlife and older adult MSM in the United States. The need to evaluate CHP in this population is threefold. First, the Centers for Disease Control and Prevention have highlighted that MSM 55 years and older comprise the only age group where HIV diagnoses increased between 2015 and 2019.²⁵ Second, prior studies have observed age-related disparities indicating lower adoption rates of biobehavioral prevention strategies such as PrEP among MSM in older age cohorts compared with MSM in younger cohorts.¹⁰ Lastly, as the proportion of older MLWH continues to increase, HIV health care services are needed to maximize accessibility of sexual health resources that align with their lived experiences.^26,27

To address this need, we aimed to identify and characterize group-level patterns of CHP among midlife and older MSM based on self-reported prevention behaviors stratified by HIV serostatus, specifically evaluating the integration of biobehavioral prevention methods with other prevention strategies. Second, we evaluated differences in group-level patterns by sociodemographic characteristics, sexual practices, and drug-use behaviors (e.g., condomless sex and substance use).

Methods

Study design and data collection

The data for this study are derived from the healthy aging substudy of the Multicenter AIDS Cohort Study (MACS), a prospective multi-center longitudinal cohort study initiated in 1983 to evaluate the natural and treatment trajectories of the HIV epidemic among MSM in the United States.²⁸ MACS participants were recruited at study sites in the Baltimore, Maryland-Washington, DC metropolitan corridor, Chicago, Illinois, Pittsburgh, Pennsylvania, and Los Angeles, California. Between 2018 and 2019, 1901 MLWH and MLWOH completed a study site visit.²⁹ As an ongoing cohort, participants complete biannual physiological and behavioral health assessments. The healthy aging substudy, Understanding Patterns of Healthy Aging Among Men Who Have Sex with Men [“Healthy Aging Study,” NIMHD, R01M010680, Principal Investigator (PI): Stall, Friedman, Plankey], is an observational cohort study to evaluate HIV- and age-related risk and resiliency factors.^30,31

To be eligible for the substudy, prospective participants had to be: (1) an active MACS participant; (2) at least 40 years old by April 1, 2016; and (3) report any sex with another man since enrollment in the MACS. Substudy participation included completing an online or in-person survey every 6 months (six surveys total) between April 2016 and April 2019. All study procedures were approved by the IRB at the coordinating center at the University of Pittsburgh and at each individual MACS site. Informed consent was obtained before each survey.

Measures

CHP strategies

Participants were offered a list of HIV prevention, including talking about HIV serostatus, talking about HIV viral load, using condoms, taking PrEP, taking PEP, pulling out before cumming (ejaculating), and choosing not to have anal sex. Participants were instructed to check off all the strategies that they or their partners used at least once during sex since their last survey completion visit. Capturing self- and partner-specific strategies provides insight into how safer-sex practices are negotiated between partners. Correlations between each prevention strategy stratified by serostatus are depicted in Supplementary Table S1 (MLWOH) and Supplementary Table S2 (MLWH). Although imperfect, talking about viral load was treated as a proxy for incorporating or considering TasP.

Correlates: HIV serostatus

Participants' HIV serostatuses, including viral load for MLWH, were measured prospectively as part of the core study. Among participants with HIV, viral load was measured at each semiannual visit by polymerase chain reaction using the Roche COBAS TaqMan assay. Viral load was dichotomized as detectable (≥40 copies/mL) and undetectable (<40 copies/mL). Participants were classified as follows (0 = MLWOH, 1 = MLWH undetectable, and 2 = MLWH detectable).

Correlates: health behaviors

At each core study visit, participants are asked to self-report the number of male sexual partners since their last visit, including partners with whom they had condomless intercourse, stratified by insertive and receptive roles. These responses were dichotomized to any versus none for both sexual positioning roles. Participants were asked about binge drinking behaviors based on the frequency with which they drank 6 or more drinks on one occasion since their last visit. Alcohol consumption was recoded to 0 = None, 1 = Any, No Binge Consumption, 2 = Any Binge Consumption. Participants were asked whether they had used the following substances since their last study visit: marijuana, poppers, crack or any form of cocaine, uppers (including crystal, methamphetamine, speed, and ice), ecstasy/MDMA, heroin/other opiates, gamma hydroxybutyrate (GHB), and any downers. Three dichotomous variables were created to indicate any versus none for marijuana use, poppers, and all other recreational drugs.

Correlates: sociodemographic characteristics

Participants self-reported their age, race/ethnicity, and education level. Similar to prior MACS analyses, race/ethnicity and educational attainment were recoded to account for small cell sizes (race/ethnicity: 0 = White, non-Hispanic; 1 = Person of color; education: 0 = Less than college degree; 1 = College degree). To account for potential MACS cohort effects, participants were distinguished between those who enrolled in the MACS before 1988 and those enrolled after 2000.

Statistical analysis

Figure 1 depicts the construction of the final MACS Healthy Aging analytic sample. Although there were six substudy data collection waves, HIV prevention data were only collected in waves 3 (Visit 67; April 2017 to September 2017), 5 (Visit 69; April 2018 to September 2018), and 6 (Visit 70; October 2018 to March 2019). We completed a cross-sectional analysis with data from participants' most recent wave participation. The number of unique participants with survey responses for at least one of these data collection waves was N = 1241 midlife and older adult MSM. Given the small proportion of missingness (8.5%) of variables of interest, we removed missing cases using listwise deletion yielding an analytic sample of n = 1135. Participants with missing responses were more likely to have been enrolled after the year 2000 (χ²₁ = 4.53, p = 0.033).

FIG. 1.

MACS healthy aging subsample included as final analytic sample. MACS, Multicenter AIDS Cohort Study.

To evaluate the aims of the study, participants were included in the final sample if they were currently sexually active with a man, which was determined by self-report of having sex with a man in the past 6 months. Our final analytic sample was n = 566 (45.6%; n = 293 MLWOH, n = 273 MLWH).

We generated descriptive statistics and conducted t and chi-square tests to evaluate differences in sociodemographic characteristics, health behaviors, and prevention strategies by serostatus. To identify group-level patterns in comprehensive HIV prevention (Aim 1), we estimated HIV serostatus-stratified latent class models using Mplus version 7.³² Latent class analysis (LCA) is a statistical technique that assumes the existence of underlying (latent) groups or subpopulations within a sample that may emerge when evaluating response patterns to observed variables.³³ Identifying groups based on behavioral patterns may maximize the utility for tailoring or prioritizing specific strategies when developing community-based interventions. LCA offers an opportunity to evaluate how midlife and older adult MSM have incorporated multiple methods into their sexual behaviors.

Due to sample size limitations, we evaluated latent class models up to five classes based on standard criteria, including the Bayesian information criterion (BIC), adjusted BIC, Bootstrapped Lo-Mendell-Rubin (BLMR) Likelihood Ratio Test, and entropy values (≥0.8).³⁴ Using recommended practices, we selected final class solutions based on these criteria alongside considerations of interpretability. To complete Aim 2, we evaluated bivariate differences in CHP classes by sociodemographic characteristics, condomless sex behaviors, and substance use. Because latent class relies on conditional probabilities, we evaluated condomless sex as a potential predictor with the assumption that prevalent condomless anal intercourse would appear across all CHP groups. We evaluated substance-use behaviors based on robust relationships with condomless sex and HIV seroconversion in prior studies.^35,36

Statistically significant associations at p < 0.05 were incorporated into multinomial logistic regression models to characterize the CHP classes. Multivariable models were controlled for race/ethnicity given the ongoing racial disparities in HIV infection in the United States.³⁷

Results

Participant characteristics

The mean age of participants was 61.1 (±8.2; range: 42–84) years. As seen in Table 1, participants were predominantly non-Hispanic White (71.6%), had attained at least a college degree (76.3%), and were enrolled in the MACS before 1988. The proportion of men reporting recent condomless sex was 44% and 42% for insertive and receptive anal intercourse, respectively. More than 80% of participants reported alcohol consumption, including 26.1% who reported binge drinking. Among MLWH, 23.1% had a detectable viral load. More than one-third of participants reported any marijuana (37.5%) and poppers use (33.6%), respectively, and 11.1% reported recreational drug use. Bivariate tests indicated that compared with MLWOH, MLWH were younger, more likely to be a person of color, less educated, report higher rates of condomless receptive anal intercourse, lower rates of alcohol consumption, and higher rates of marijuana, poppers, and other recreational drug use.

Table 1.

Sample Characteristics

Variable	Full sample, N = 566		MLWOH, n = 293 (51.8%)		MLWH, n = 273 (48.2%)		t or χ2	p
	M (SD)	n (%)	M (SD)	n (%)	M (SD)	n (%)
Age, range 42–84	61.1 (8.2)	—	63.1 (8.0)	—	58.9 (7.9)	—	6.21	<0.001
Race/ethnicity							36.37	<0.001
White, non-Hispanic	—	405 (71.6)	—	242 (82.6)	—	163 (59.7)
Person of color	—	161 (28.4)	—	51 (17.4)	—	110 (40.3)
Educational attainment							17.88	<0.001
No college degree	—	134 (23.7)	—	48 (16.4)	—	86 (31.5)
College degree	—	432 (76.3)	—	245 (83.6)	—	187 (68.5)
Viral load								—
Undetectable	—	—	—	—	—	210 (76.9)	—
Detectable	—	—	—	—	—	63 (23.1)	—
MACS enrollment							41.44	<0.001
Before 1988	—	368 (65.0)	—	227 (77.5)	—	141 (51.6)
After 2000	—	198 (35.0)	—	66 (22.5)	—	132 (48.4)
Condomless receptive anal intercourse							19.94	<0.001
None	—	238 (58.0)	—	196 (66.9)	—	132 (48.4)
Any	—	238 (42.0)	—	97 (33.1)	—	141 (51.6)
Condomless insertive anal intercourse							2.27	0.132
None	—	317 (56.0)	—	173 (59.0)	—	144 (52.7)
Any	—	249 (44.0)	—	120 (41.0)	—	129 (47.3)
Alcohol consumption							10.15	0.006
No consumption	—	90 (15.9)	—	37 (12.6)	—	53 (19.4)
Any—no binge	—	328 (58.0)	—	188 (64.2)	—	140 (51.3)
Any—binge	—	148 (26.1)	—	68 (23.2)	—	80 (29.3)
Marijuana use							14.28	<0.001
None	—	354 (62.5)	—	205 (70.0)	—	149 (54.6)
Any	—	212 (37.5)	—	88 (30.0)	—	124 (45.4)
Taken poppers							5.66	0.017
None	—	376 (66.4)	—	208 (71.0)	—	168 (61.5)
Any	—	190 (33.6)	—	85 (29.0)	—	105 (38.5)
Recreational drug use							15.28	<0.001
None	—	503 (88.9)	—	275 (93.9)	—	228 (83.5)
Any	—	63 (11.1)	—	18 (6.1)	—	45 (16.5)

MACS, Multicenter AIDS Cohort Study; MLWH, men living with HIV; MLWOH, men living without HIV; SD, standard deviation.

Individual HIV prevention strategies

Figure 2 depicts serostatus-stratified proportions of participants' HIV prevention strategies since their last core study visit. Among MLWOH, the most common strategies reported were discussing theirs and their partners' HIV serostatus as well as abstaining from having anal sex. Biomedical HIV prevention strategies were low among MLWOH (<15% for PrEP and PEP indicators). The most common strategies reported among MLWH were talking about HIV status, discussing level of viral load, and using condoms (>25% for all indicators). Compared with MLWOH, MLWH were more likely to report talking with a partner about their viral load, using condoms, having a partner they knew was on PrEP, and pulling out before cumming. MLWOH were more likely to abstain from anal sex than MLWH.

FIG. 2.

Proportion of self-reported individual HIV prevention strategies enacted by midlife and older adult MSM, N = 566. “Thinking of the sex you have had with other men since your last visit. Which of the following safer sex methods have you used? (check all that apply).” Participants reported their prevention strategies as well as the prevention strategies used by their male partner(s). MLWH, men living with HIV; MLWOH, men living without HIV; MSM, men who have sex with men.

Latent classes: MLWOH

We conducted four LCAs with survey responses from MLWOH (Table 2). Given model fit criteria, we concluded that the 3-class solution was the best fitting model. Based on the conditional response probabilities (Fig. 3), we identified Class 1 as high CHP overall (n = 126, 43%). Notably, biomedical prevention tools were predominantly reported by men in Class 1. Men in Class 2 (n = 44, 15.0%) exhibited high conditional response probabilities for self and partner abstention from anal sex (high anal sex abstention). Men in Class 3 (n = 123, 42%) reported limited prevention behaviors all around (low prevention overall).

Table 2.

Latent Class Analysis Model Comparison

Classes	MLWOH					MLWH
	BIC	ABIC	BLMR likelihood ratio test	p	Entropy	BIC	ABIC	BLMR likelihood ratio test	p	Entropy
1	3128.23	3087.01	NA		NA	3502.66	3464.61	NA		NA
2	2810.92	2725.30	396.83	<0.0001	0.896	3123.77	3044.50	451.81	<0.0001	0.922
3	2802.44	2672.42	88.00	<0.0001	0.947	3115.36	2994.87	81.34	<0.0001	0.866
4	2825.33	2650.91	56.63	0.333	0.899	3118.66	2956.95	69.62	<0.0001	0.838
5	—	—	—	—	—	3131.42	2928.49	60.16	<0.0001	0.872

ABIC, adjusted BIC; BIC, Bayesian Information Criterion; BLMR, Bootstrapped Lo-Mendell-Rubin; MLWH, men living with HIV; MLWOH, men living without HIV; NA, not applicable.

FIG. 3.

Latent class analysis: conditional response probabilities for 3-class solution of CHP among MLWOH. Class 1 = high CHP overall; Class 2 = high anal sex abstention; Class 3 = low prevention overall. CHP, combination HIV prevention; MLWOH, men living without HIV; PEP, postexposure prophylaxis; PrEP, pre-exposure prophylaxis.

We observed bivariate between-class differences in condomless anal intercourse; specifically, there were lower rates of condomless receptive (F = 6.94, p = 0.031) and condomless insertive anal intercourse (F = 7.12, p = 0.028) among men in the anal abstention class compared with men in the high CHP and low prevention classes, respectively. Men in the high CHP class reported higher rates of popper use than those in the anal abstention and low prevention classes (χ² = 9.09, p = 0.011). There were no bivariate between-class differences by age, race/ethnicity, educational attainment recruitment cohort, alcohol consumption, marijuana use, and other recreational drug use.

Multinomial logistic regression model (Table 3), adjusted for race/ethnicity, indicated that compared with Class 2, men in Class 3 were 2.5 [95% confidence interval (CI): 1.01–6.21] times more likely to report any recent condomless insertive anal intercourse. Compared with Class 1, men in Class 3 were less likely to report using poppers [adjusted odds ratio (AOR) = 0.42, 95% CI: 0.24–0.75].

Table 3.

Multinomial Logistic Regressions for Combination HIV Prevention Classes

Variables	MLWOHa			MLWHb
	0: Class 1	0: Class 1	0: Class 2	0: Class 1	0: Class 1	0: Class 1	0: Class 2	0: Class 2	0: Class 3
	1: Class 2	1: Class 3	1: Class 3	1: Class 2	1: Class 3	1: Class 4	1: Class 3	1: Class 4	1: Class 4
	AOR (95% CI)	AOR (95% CI)	AOR (95% CI)	AOR (95% CI)	AOR (95% CI)	AOR (95% CI)	AOR (95% CI)	AOR (95% CI)	AOR (95% CI)
Age	—	—	—	1.03 (0.98–1.08)	1.00 (0.95–1.06)	1.05 (1.00–1.11)	0.98 (0.93–1.03)	1.02 (0.98–1.07)	1.06* (1.01–1.11)
Race/ethnicity
Non-Hispanic White Person of color	Ref. 1.45 (0.59–3.57)	Ref. 0.82 (0.41–1.63)	Ref. 0.50 (0.19–1.34)	Ref. 0.30** (0.15–0.69)	Ref. 0.93 (0.41–2.09)	Ref. 0.77 (0.36–1.68)	Ref. 3.07* (1.38–6.85)	Ref. 2.57* (1.18–5.62)	Ref. 0.89 (0.41–1.95)
Viral load
Detectable Undetectable	— —	— —	— —	Ref. 0.39 (0.15–1.03)	Ref. 0.31* (0.12–0.78)	Ref. 0.53 (0.21–1.39)	Ref. 0.79 (0.35–1.78)	Ref. 1.35 (0.59–3.09)	Ref. 1.92 (0.86–4.32)
Condomless receptive anal intercourse
None Any	Ref. 0.47 (0.18–1.23)	Ref. 1.09 (0.61–1.96)	Ref. 2.21 (0.86–5.70)	Ref. 0.98 (0.43–2.22)	Ref. 1.06 (0.45–2.48)	Ref. 1.17 (0.52–2.65)	Ref. 1.05 (0.47–2.35)	Ref. 1.15 (0.54–2.46)	Ref. 1.07 (0.47–2.45)
Condomless insertive anal intercourse
None Any	Ref. 0.48 (0.20–1.13)	Ref. 1.06 (0.60–1.87)	Ref. 2.50* (1.01–6.21)	Ref. 1.30 (0.57–2.97)	Ref. 0.84 (0.36–1.98)	Ref. 0.61 (0.27–1.38)	Ref. 0.65 (0.29–1.48)	Ref. 0.48 (0.22–1.04)	Ref. 0.71 (0.31–1.65)
Taken poppers
None Any	Ref. 0.62 (0.28–1.37)	Ref. 0.42** (0.24–0.75)	Ref. 0.76 (0.33–1.77)	Ref. 0.94 (0.44–2.00)	Ref. 0.87 (0.40–1.91)	Ref. 0.42* (0.19–0.92)	Ref. 0.89 (0.42–1.88)	Ref. 0.45* (0.22–0.94)	Ref. 0.47 (0.21–1.03)

^a Class 1 = high CHP overall; Class 2 = anal sex abstention; Class 3 = low prevention overall.

^b Class 1 = high CHP overall; Class 2 = high CHP/low condom use; Class 3 = high condom reliance; Class 4 = low prevention overall.

^* p < 0.05; ^**p < 0.01.

AOR, adjusted odds ratio; CHP, combination HIV prevention; CI, confidence interval; MLWH, men living with HIV; MLWOH, men living without HIV; Ref., referent group.

Latent classes: MLWH

We conducted five LCAs with survey responses from MLWH (Table 2). Although the 5-class solution provided better fit statistics than the 4-class solution, we selected the 4-class solution based on the interpretability of conditional response probabilities (Fig. 4) and ensuring adequate sample sizes across all classes to make between-group comparisons. Class 1 (n = 57, 20.9%) and Class 2 (n = 74, 27.1%) were both identified as high CHP groups but exhibited noticeable differences in reports of condom use and pulling out before cumming (Class 1: high CHP overall; Class 2: high CHP/low condom use). Class 3 (n = 61, 22.3%) was identified as a group primarily reliant on condom use (high reliance on condoms). Men in Class 4 (n = 81, 29.7%) were those who self-reported very limited prevention behaviors overall (low prevention overall).

FIG. 4.

Latent class analysis: conditional response probabilities for 4-class solution of CHP among MLWH. Class 1 = high CHP overall; Class 2 = high CHP/low condom use; Class 3 = high condom reliance orientation; Class 4 = low prevention overall. CHP, combination HIV prevention; MLWH, men living with HIV; PEP, postexposure prophylaxis; PrEP, pre-exposure prophylaxis.

We observed bivariate between-class differences by age, race/ethnicity, condomless insertive anal intercourse, and poppers use. Men in the high CHP/low condom use and low prevention groups were on average older than the high CHP overall and condom-reliant groups (F = 3.48, p = 0.016). Men of color exhibited lower rates of being in the high CHP/low condom use and higher rates of being in the condom-reliant group compared with being in the overall high CHP and low prevention groups (χ² = 14.83, p = 0.002). Men in the low prevention group exhibited lower rates of condomless insertive anal intercourse (χ² = 8.58, p = 0.036) and poppers use (χ² = 10.46, p = 0.015) than all other CHP groups. No between-class differences were observed by educational attainment, viral load status, MACS enrollment cohort, condomless receptive anal intercourse alcohol consumption, marijuana use, and other recreational drug use.

In multivariable logistic regression models (Table 3), being virally undetectable was associated with lower odds of being in Class 3 compared with Class 1 (AOR = 0.31, 95% CI: 0.12–0.78). Using poppers was associated with lower odds of being in Class 4 compared with Classes 1 (AOR = 0.42, 95% CI: 0.19–0.92) and 2 (AOR = 0.45, 95% CI: 0.22–0.94). Men of color were less likely to be in Class 2 compared with Class 1 (AOR = 0.30, 95% CI: 0.15–0.69). Compared with Class 2, men of color were more likely to be in Classes 3 (AOR = 3.07, 95% CI: 1.38–6.85) and 4 (AOR = 2.57, 95% CI: 1.18–5.62).

Discussion

Our current study extends the field's understanding of how contemporary MSM in midlife and older adulthood living with and without HIV incorporate CHP into their sexual practices. Similar to prior research, our findings exhibit diverse prevention strategies that exist between and within MSM by serostatus.²³ We observed three latent classes for MLWOH and four latent classes for MLWH, which reinforces the concept that HIV prevention is not “one-size-fits-all” for MSM.

CHP in MLWOH

Almost as many men living without HIV (MLWOH) were classified in the high CHP class as men classified in low prevention class. For MLWOH in Class 2, there was a 100% conditional probability for abstaining from anal intercourse. The prevention strategies among these men may reflect a low perceived need for CHP to align with a low perceived risk of HIV transmission. Multivariable models provided limited insight into sociodemographic and behavioral differences that distinguish participants between CHP classes. Among MLWOH, participants in the overall low prevention group (Class 3) were roughly 2.5 times more likely to engage in condomless insertive anal intercourse compared with the high prevention/low condom group (Class 2).

Given that men in the overall low prevention group engaged in limited prevention behaviors, this finding may reflect strategic sexual positioning based on individuals' understanding of transmission risk levels associated with condomless sex as a receptive partner versus an insertive partner may be at play. Those who engaged in condomless insertive anal intercourse may do so as a risk-limiting strategy; however, motivations and preferences for sexual positioning as it relates to personal CHP warrants ongoing investigation.

In addition, poppers use was associated with being in high CHP groups relative to the low prevention groups MLWOH. These findings are especially interesting in light of prior analyses conducted with men from the MACS before the advent of PrEP, PEP, and TasP, which indicated poppers as a strong risk factor for HIV seroconversion.^35,37 Poppers are frequently used among MSM to enhance sexual pleasure and facilitate sexual activities (e.g., engaging in receptive role during anal intercourse).³⁸ The finding that men who used poppers were more likely to use multiple prevention modalities may suggest that midlife and older MSM are balancing or incorporating pleasure into personal conceptualizations of HIV prevention by taking additional measures to reduce HIV transmission through sex when using poppers.

CHP in MLWH

In MLWH, almost half of the participants were classified into the high CHP classes (Classes 1 and 2), although between high CHP groups, there was a stark difference in self-reported condom use. Nearly one-third of MLWH were classified as being in an overall low prevention class (Class 4). Interestingly, although MLWH in Class 2 were classified as high prevention regarding CHP, condom use was higher and the primary strategy for HIV prevention among men in Class 3. These findings may reflect variations in beliefs and preferences regarding strategies to reduce risks of HIV transmission. MLWH in Class 2 may rely on CHP strategies given the limited propensities to use condoms, whereas men in Class 3 may be less inclined to incorporate additional HIV prevention strategies given a high perceived effectiveness of condoms as a solo prevention strategy.

This finding suggests that the ways in which CHP intersects with levels of risk are more nuanced; specifically that men in high CHP groups may still exhibit higher levels of HIV-related risk behaviors than those who rely on individual prevention strategies if the individual prevention strategy is highly effective (e.g., condom use). However, a contextualized understanding of midlife and older adult MSM's strategy preferences and decisions, especially as it relates to individual-level constructions of CHP, requires further exploration.

Multivariable models also indicated that compared with non-Hispanic White MLWH, men of color were more likely to be in the low or condom-reliant prevention groups compared relatively with the high prevention groups. These findings likely reflect ongoing systemic inequalities related to discrimination, medical mistrust, and insurance coverage that disproportionately burden communities of color, particularly Black and Brown people, in accessing medical care and other sexual health resources needed to sustain antiretroviral uptake for primary and secondary HIV prevention.^39–43 Lastly, similar to participants living without HIV, poppers use was associated with being in a high CHP group compared with the overall low prevention group. How using poppers fits into MLWH's sexual behaviors at the intersection of HIV prevention and pleasure warrants further exploration.

Biobehavioral HIV prevention

Among MLWOH, conditional probabilities for personal and partner PrEP use, although low, were predominantly reported by men in the high CHP group. Men in the high CHP group were likely to integrate PrEP with other strategies, particularly condom use, which counters the notion of risk compensation (i.e., intention or willingness to engage in increased HIV risk behaviors, such as condomless sex, given the protections that the medication provides against HIV).⁴⁴ Among MLWH, conditional probabilities for partner PrEP use were highest in both the high prevention groups compared with the condom-reliant and low prevention groups; however, the conditional probabilities for condom use that distinguish the two high CHP groups suggest there may be differing attitudes toward combining biobehavioral methods such as PrEP with sociobehavioral methods such as condoms.

We cannot infer risk compensation for MLWH men in Class 2 (high CHP/low condom use group) from these findings because the cross-sectional nature of our analysis prohibits insight into behavioral changes over time (i.e., PrEP may have been added as a prevention strategy rather than as a substitute for condom use). Across all groups for both MLWOH and partners of MLWH, the integration of PEP was low. These findings align with prior studies that indicate historically low awareness and uptake among MSM.^10,45

Although an imperfect proxy for TasP, discussions about viral load were high in both high CHP groups among MLWH. Multivariable analyses, however, indicated that compared with MLWH with a detectable viral load, having an undetectable viral load was associated with being in the overall condom-reliant prevention group (Class 3) relative to the overall high CHP group (Class 1). Irrespective of viral load, men in Classes 1 and 3 had high conditional probabilities for condom use, but men with a detectable viral load had a higher propensity for utilizing additional prevention strategies than men who were undetectable. These findings suggest that men with a detectable viral load may be largely aware of the risk implications attributed to having a detectable viral load and utilize multiple strategies to maximize preventing transmission.

On the contrary, men with an undetectable viral load may be less likely to use strategies beyond condoms with an understanding that they have a virtually zero risk of transmitting the virus to a partner. Given our measurement of CHP, MLWH in the overall low prevention group (Class 4) who have an undetectable viral load may in fact rely on TasP as a solo prevention strategy. For MLWOH, discussions about viral load were low across all CHP groups, which may reflect irrelevancy assuming a low number of known partners living with HIV.

Limitations and future research

Our findings should be evaluated in the context of study limitations. Our measure of individual-level prevention strategies may not be fully comprehensive of alternative strategies reported by MSM in prior studies. Recognizing the various levels of effectiveness, our measure did not include strategies such as HIV testing, sero- or PrEP-sorting (e.g., seeking partners of the same serostatus, partners living with HIV who have an undetectable viral load, or partners living without HIV who are on PrEP), strategic sexual positioning (e.g., being the receptive versus insertive partner), or how HIV prevention in sexual health contexts intersect with contexts related to injection drug use (e.g., syringe exchange) for either the participants or their male partners. Therefore, participants in low prevention groups may be enacting prevention strategies not captured by our measure.

Our CHP measure also is limited in its ability to capture structural interventions for HIV prevention, such as having access to high-quality sexual health services or insurance policies that cover testing services. Further, the current study was implemented before the availability and implementation of next-generation PrEP and HIV treatment modalities, such as long-acting injectable PrEP and long-acting injectable antiretrovirals for MLWH. Uptake of emerging modalities in the biobehavioral HIV prevention pipeline, when available, may inform individuals' new constructions of CHP practices.^46,47

The Healthy Aging data set does not provide insight into the role that monogamy or consensual nonmonogamy may have in shaping personal practices around CHP. Relationship dynamics and agreements may be relevant to MSM given the widespread acceptability and practices of consensual nonmonogamy.⁴⁸ Even among participants in romantic relationships who may have only had one sexual partner in the past 6 months, we were unable to ascertain whether the prevention strategies reported by each participant were specific to primary partners, casual partners, or both. As it relates to CHP, individuals may engage in different combinations of HIV prevention strategies with a primary romantic partner compared with a casual or secondary partner. Another conceptual limitation is that our study's findings were intended to be descriptive of latent classes based on CHP. Although we found diverse classifications of CHP in our sample of midlife and older adult MSM, we are limited in our understanding of how and in what contexts participants enact CHP. In addition to expanding measurement of CHP and accounting for participants' relationship dynamics, future studies should utilize mixed-methods approaches to better understand how individual-level CHP is constructed among midlife and older adult MSM. These studies should attend to factors, such as access to care and sexual health resources, how prevention strategies intersect with sexual pleasure, and perceptions of HIV risk. In addition, the cross-sectional nature of our study prevents our ability to evaluate changes in participants' CHP strategies and factors over time.

Limitations in our analytic approach include the inherent threats to accuracy and validity imposed by self-report and social desirability. Surveying about sexual behaviors, sexual health, and HIV is often deemed a sensitive topic and may affect participants' willingness to provide any or completely accurate responses.⁴⁹ Many of our study's items that asked participants to respond based on a time-specified window period, including our items on HIV prevention strategies, may be susceptible to recall bias. Lastly, our findings may not be generalizable to communities beyond our sample. Our sample was predominantly non-Hispanic White and college educated. Limited diversity across racial and ethnic groups in our sample limited our ability to analyze racial and ethnic disparities sufficiently. Further, the prevalence of recreational drug use was low in our sample. Future studies should evaluate the robustness of our study-identified latent classes and their correlates in more diverse samples.

Practice and policy implications

We recommend scaling up sexual health communication practices among clinical providers to facilitate HIV serostatus-neutral discussions about patient strategies for primary and secondary HIV prevention as appropriate.⁵⁰ These discussions offer opportunities for providers to discuss the effectiveness of individual strategies, including when combined with other strategies, how attitudes toward different prevention strategies change over time, promote strategies that are effective based on participants' preferences, and facilitate patients' abilities to initiate biobehavioral prevention methods (e.g., prescriptions, financial assistance programs), including newly emerging strategies as they become available. Further, these discussions must be undertaken in ways that are nonjudgmental and that recognize HIV prevention is not a one-size-fits-all approach.

Our findings also suggest that biobehavioral prevention strategies are not being implemented in ways that are equitable for MSM in midlife and older adulthood, particularly for men of color. Although researchers continue to observe increases in the adoption of PrEP and improvements in outcomes along the HIV care continuum among MSM, progress in communities similar to midlife and older adult MSM continues to lag behind.¹⁰ A recent study suggested a need for policies that support race/ethnicity- and age-targeted allocation of PrEP resources to generate substantial cumulative reduction in new diagnoses.⁵¹ Strategic and scaled-up allocation of PrEP resources toward midlife and older MSM may have a major impact on slowing the increase in new cases in this age cohort.

The emergence of long-acting injectable PrEP and injectable antiretrovirals offers an opportunity to assist midlife and older adult MSM, especially those who report limited uptake of prevention strategies, to maximize HIV-related risk reduction where there may be little interest in engaging in sociobehavioral prevention strategies. The HIV workforce in clinical, community, and policy practice must support midlife and older adult MSM of diverse backgrounds to overcome barriers to biobehavioral prevention.

Promoting and enacting CHP remain a critical component for achieving benchmark objectives of the Ending the HIV Epidemic initiative in the United States. Monitoring CHP and the integration of emerging prevention strategies are critical for midlife and older adult MSM given this community's historical and disproportionate experience with the HIV epidemic in the United States. The findings from our study reinforce the importance of expanding definitions to address combinations of HIV prevention strategies undertaken in the context of sexual relationships rather than focusing on single strategies, such as condom versus no condom use, PrEP versus no PrEP use, and detectable HIV versus undetectable HIV viral loads. The diversity in midlife and older adults' engagement of CHP by and within the HIV serostatus groups is an important reminder that HIV prevention is not a one-size-fits-all approach.

Future HIV prevention research and programming for midlife and older adult MSM may benefit from a socially and behaviorally contextualized evaluation of factors that inform and support CHP.

Authors' Contributions

All authors have approved each section of the final article. S.M.: Conceptualization, data curation, formal analysis, writing, reviewing, and editing drafts. J.E.E.: Conceptualization, data curation, formal analysis, writing, reviewing, and editing drafts. D.W.: Conceptualization, supervising analysis, reviewing, and editing drafts. M.B.-I.: Conceptualization, reviewing, and editing drafts. S.A.H.: Conceptualization, reviewing, and editing drafts. R.D.: Conceptualization, reviewing, and editing drafts. F.P.: Conceptualization, reviewing, and editing drafts. M.R.F.: Conceptualization, funding acquisition, reviewing, and editing drafts. M.W.P.: Conceptualization, funding acquisition, supervising analysis, reviewing, and editing drafts.

Author Disclosure Statement

All authors have no competing financial interests to declare.

Funding Information

Data in this article were collected by the MACS Healthy Aging team (Co-PIs: Friedman and Plankey; 5R01MD010680-05), a subset of the MACS/WIHS Combined Cohort Study (MWCCS). The opinions, arguments, and conclusions from our study represent those endorsed by the authors and do not necessarily reflect those of the MACS or the National Institutes of Health (NIH). MWCCS (PIs): Data Analysis and Coordination Center (Gypsyamber D'Souza, Stephen Gange, and Elizabeth Golub), U01-HL146193; Chicago-Cook County CRS (Mardge Cohen and Audrey French), U01-HL146245; Chicago-Northwestern CRS (Steven Wolinsky), U01-HL146240; Los Angeles CRS (Roger Detels), U01-HL146333; Metropolitan Washington CRS (Seble Kassaye and Daniel Merenstein), U01-HL146205; Pittsburgh CRS (Jeremy Martinson and Charles Rinaldo), U01-HL146208.

The MWCCS is funded primarily by the National Heart, Lung, and Blood Institute (NHLBI), with additional cofunding from the Eunice Kennedy Shriver National Institute of Child Health & Human Development (NICHD), the National Human Genome Research Institute (NHGRI), the National Institute on Aging (NIA), the National Institute of Dental & Craniofacial Research (NIDCR), the National Institute of Allergy And Infectious Diseases (NIAID), the National Institute of Neurological Disorders and Stroke (NINDS), the National Institute of Mental Health (NIMH), the National Institute on Drug Abuse (NIDA), the National Institute of Nursing Research (NINR), the National Cancer Institute (NCI), the National Institute on Alcohol Abuse and Alcoholism (NIAAA), the National Institute on Deafness and Other Communication Disorders (NIDCD), and the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK).

Supplementary Material

Supplementary Table S1

Supplementary Table S2