Correlates of mpox vaccination among sexual minority men in the United States: Sexual behavior, substance use, and main partner relationships

Abstract

Sexual minority men (SMM) have accounted for the majority of infections during the 2022 outbreak of the orthopox virus known as mpox (previously “monkeypox”) in the US. This study examined correlates of mpox vaccination. Between July 28 and September 22, 2022, adult cisgender SMM (n=2,620) not previously diagnosed with mpox responded to recruitment advertisements on social networking applications and completed an online survey. Of these, 730 (27.9%) received at least one vaccine dose. Logistic regression indicated sex with a casual partner was positively associated with vaccination. Stimulant drug use was negatively associated with vaccination; meanwhile, the use of ecstasy, ketamine, gamma-hydroxybutyrate (GHB) or psychedelics was positively associated with vaccination. Among partnered SMM, non-monogamous sexual agreements, relationship length of ≥2 years, and relationship functioning were positively associated with vaccination. Even at low levels of relationship functioning, SMM in non-monogamous relationships of ≥2 years were more likely to be vaccinated than single SMM. At very high levels of relationship functioning, partnered SMM were more likely to be vaccinated than single SMM regardless of sexual agreement or relationship length. Findings are discussed in relation to prior research on HIV, other STI prevention, and theories of dyadic functioning and health in this population.

INTRODUCTION

Since May, 2022 the orthopox virus known as mpox (previously known as “monkeypox”) has infected more than 84,000 people in over 100 countries, causing at least 80 deaths (Centers for Disease Control and Prevention – CDC, 2022a; World Health Organization, 2022). The US has documented more than 30,000 cases of mpox and over 20 deaths related to the disease as of January, 2023 (CDC, 2022a). Sexual minority men (SMM) have born a disproportionate burden of infection. Men constitute 97% of identified cases and more than 75% of those men infected disclosed a history of sexual contact with other men (Hazra et al., 2022). The epidemic has so far concentrated disproportionately in Black SMM in the Southern US. Florida, Georgia and Texas all rank among the top five states reporting cases, along with California and New York (CDC, 2022b).

Mpox is spread most often by close physical contact with an infected individual involving exposure o respiratory droplets or direct contact with mucocutaneous lesions (Kaler et al., 2022). As a result, sexual contact has substantial transmission potential as the result of mere proximity and skin contact. Although there is evidence that mpox can be detected in semen, the potential for transmission through exposure to semen has not been clearly established (Lapa et al., 2022; Peiró-Mestres et al., 2022; Reda et al., 2023).

Mpox Vaccination

While there is substantial evidence that immunity to smallpox provides protection against mpox infection, the exact extent of the protection conferred by smallpox vaccination is difficult to determine (Poland et al., 2022). Two vaccines for smallpox, one of which is also FDA-approved for mpox, are currently available. According to recent observational reports and unpublished data from the CDC, disease incidence in unvaccinated but vaccine-eligible men between 18–49 in the US was approximately 7 to 10 times higher than the incidence among those who received either one dose or two doses of the JYNNEOS vaccine (the latter defined as “fully vaccinated”), putting preliminary estimates for vaccine effectiveness at 37% and 69%, respectively (CDC, 2022c, 2022d).

The concentration of mpox infections among SMM, and the potential for the virus to be transmitted sexually, have focused vaccination efforts. In part due to limited vaccine supplies (CDC, 2022b), vaccination in the US was prioritized to those diagnosed with or exposed to mpox as well as those who met sexual behavior criteria (e.g., multiple partners). Those restrictions have been loosened in some jurisdictions as vaccines have become more available (Kaiser Family Foundation, 2022).

Given eligibility criteria for vaccination included a sexual behavior component for a substantial period of the outbreak, it is reasonable to expect that mpox vaccination would be associated with sexual behavior. It is also plausible that vaccination would be associated with sexual behavior because it alters risk perception. CDC guidelines for preventing mpox infection include avoiding close physical contact with others (CDC, 2022e). This recommendation encompasses sexual interactions. Some mpox prevention messages have directly suggested limiting casual or one-time sexual partners (Building Healthy Online Communities, 2022).

Factors associated with mpox vaccine uptake

Relatively little is known about factors associated with mpox vaccine uptake beyond the limitations imposed for eligibility by health departments. Previous research on HIV infection risk would suggest that associations with drug and alcohol use are likely but may be complex. On the one hand, SMM who use substances may be more likely to meet sexual behavior criteria for mpox vaccination eligibility. The use of cannabis and illicit drugs is a well-established predictor of sex with casual male partners (e.g., Cain et al., 2021; Starks et al., 2020). Although more varied, some evidence suggests alcohol use is associated with sexual behavior with casual partners (e.g., Robbins et al., 2020; Rowe et al., 2016; Zhang & Wu, 2017). On the other hand, substance use has also been associated with decreased concern for mpox (Wang et al., 2022), decreased engagement with healthcare providers (Gwadz et al., 2016; Salway et al., 2019) and missed HIV care appointments among SMM living with HIV (Batchelder et al., 2021), potentially decreasing the likelihood those using substances get vaccinated.

Several factors suggest that relationship status might be associated with mpox vaccination. Partnered SMM in monogamous agreements (which prohibit sex with partners outside the relationship) are less likely to have had sex with casual partners compared to single men and those partnered SMM in non-monogamous relationships (wherein sex with outside partners is permitted in some way) (Starks et al., 2020; Starks, Robles, Bosco, et al., 2019). They are therefore less likely to meet sexual behavior eligibility criteria for vaccination. It is also plausible that relationship status is associated with risk perception for mpox. Research on HIV testing behavior indicates that partnered SMM – particularly those in monogamous relationships – perceive themselves to be at lower risk for HIV infection (Stephenson, White, & Mitchell, 2015). They are less likely to test routinely for HIV (Dellucci et al., 2022; Mitchell & Petroll, 2012; Stephenson, White, Darbes, et al., 2015) and be on pre-exposure prophylaxis (PrEP) for HIV prevention (Starks et al., 2020; Stephenson et al., 2021).

At the same time, interdependence theory (Kelley & Thibaut, 1978; Rusbult & Van Lange, 2003) and related theories of dyadic coping and health (Lewis, McBride, et al., 2006) would suggest that partnered SMM would be more likely to pursue vaccination – especially those with higher levels of relationship quality. Lewis et al. (2006) suggested that partners in high quality relationships are more likely to perceive a health concern – such as mpox infection – as a shared threat or stressor. This shared threat perception motivates joint coping. Partners are more likely to work together and support one another in the implementation of strategies to reduce the health concern (e.g., mpox vaccination). Consistent with these assumptions, Starks et al. (2021) tested a model of dyadic coping and COVID-19 prevention behavior using data collected prior to COVID-19 vaccine availability. They found that partners’ consensus about the value of joint effort – working together as a couple – to reduce COVID-19 risk was associated with COVID-19 prevention behaviors. Consensus about joint effort was in turn positively associated with measures of relationship functioning.

Current Study

The purpose of this study was to identify correlates of mpox vaccination among SMM. Because sexual behavior with casual partners contributed to the determination of vaccine eligibility (Kaiser Family Foundation, 2022), and because vaccination might reduce concerns about engaging in sex with casual partners, we hypothesized that sexual behavior with casual partners would be positively associated with vaccination in any cross-sectional data. Aligned with research on the associations between drug use and sexual behavior (e.g., Starks et al., 2020) as well as healthcare engagement (Gwadz et al., 2016; Salway et al., 2019), we hypothesized that substance use would be positively associated with the odds of mpox vaccination in a bivariate context; however, we anticipated it would be negatively associated with vaccination in any multivariable context that controls for sexual behavior. Finally, consistent with theories of dyadic coping (Lewis, McBride, et al., 2006) and prior research on HIV risk perception (Stephenson, White, & Mitchell, 2015), we anticipated that partnered SMM in non-monogamous agreements would be more likely to be vaccinated for mpox (compared to single SMM and those in monogamous relationships) and that relationship functioning would be positively associated with mpox vaccination among partnered SMM.

METHOD

Participants

Participants were included in these analyses if they met the following eligibility criteria: identified as cisgender men, indicated they were aged 18 or older, and provided a zip code indicating residence in a US state or the District of Columbia. Those in relationships were included if they reported that their relationship partner was a cisgender man aged 18 or older. Participants were excluded if they reported that they had been diagnosed with mpox by a physician in the past 6 months. All participants completed consent and responded to the survey in English.

Design

The study utilized a cross-sectional survey design and convenience sampling. The data were generated by an online survey whose primary purpose was to assess eligibility for several NIH-funded research studies.

Procedure

Participants were recruited between July 28, 2022 and September 22, 2022 through geosocial networking applications for gay, bi, trans, and queer people seeking sexual partners (including, but not limited to, Jack’d and Scruff). Respondents who clicked on the advertisement were directed to consent information. Those who consented indicated this by clicking a button to proceed with the survey. A waiver of written consent documentation was obtained. Consistent with its primary function as a study screener, the survey was brief (completion time approximately 11 minutes). Participation was not compensated. The Institutional Review Board of The City University of New York approved all procedures.

Measures

Demographics

Participants reported their age, sexual identity, HIV status, and relationship status as well as race and ethnicity. Response options for sexual identity included: gay, bisexual, queer, pansexual, demisexual, and heterosexual. Participants also had the option to select “other identity.” Those who did were given a text box to write-in a preferred response. Race and ethnicity were assessed separately. Participants indicated their racial identity in one of 5 nominal categories: White/Caucasian, Black/African America/Afro-Caribbean, Asian/Pacific Islander, Multiracial, and Not Listed. Those who selected “Not Listed” were given a text box to enter their racial identity. Ethnicity was assessed using a single item, “Do you consider yourself to be Hispanic or Latino.” In analyses, participants were categorized as Latino if they responded “yes” to the ethnicity item irrespective of their response to the racial identity item. Zip code determined geographical region of residence in the US (Northeast, Southeast, Midwest and West).

Consistent with previous studies comparing single SMM to those in relationships (e.g., Starks et al., 2020; Starks, Robles, Bosco et al., 2019), participants were initially asked, “Are you currently in a relationship?” Those who responded “yes” were subsequently asked, “Do you consider this partner to be a main partner? By main partner, we mean someone you feel committed to above anyone else. This would be someone you call a boyfriend, partner, significant other, or spouse?” Participants were classified as being “partnered” if they responded yes to both questions and as “single” if they responded no to either.

Among partnered SMM, sexual agreements were assessed using a single item “How do you and your partner handle sex with outside partners?” This item has been commonly used in prior research with this population (e.g., Starks et al., 2020; Starks, Robles, Bosco et al., 2019). Participants were classified as monogamous if they indicated (“neither of us has sex with others, we are monogamous” or “I don’t have sex with other people, I don’t know what my partner does”). Participants were classified as non-monogamous if they selected any response indicating that either they or their partner had sex with outside partners. Relationship length was assessed with a single item using a series of nominal responses (less than 1 month, 1 or 2 months, 3 to 6 months, more than 6months but less than 1 year, 1 or 2 years, 2 years or more) and dichotomized to distinguish between couples who had been together for 2 years or more (versus less than 2 years).

Relationship functioning

Three subscales from the Perceived Relationship Quality Component (PRQC) scale (Fletcher et al., 2000) were used to assess satisfaction (“How content are you with your relationship?”), commitment (“How devoted are you to your relationship”) and intimacy (“How intimate is your partner?”). Each subscale was comprised of three items. Participants indicated their response on a Likert-type scale ranging from 1 = “not at all” to 7 = “extremely.” The total score produced by summing the nine items from all three subscales had excellent reliability (Cronbach’s α = .95)

Problematic alcohol use

The Alcohol Use Disorders Identification Test – Concise (AUDIT-C) is a brief, screening measure assessing problematic alcohol use (Bush et al., 1998). Participants indicated their drinking frequency (never, monthly, 2–4 times a month, 2–3 times a week, 4 or more times a week), the number of standard drinks typically consumed per day (1–2, 3–4, 5–6, 7–9, 10 or more), and how often they have six or more drinks on one occasion (never, less than monthly, monthly, weekly, daily or almost daily). Total scores were created by summing item responses (Cronbach’s α = .77), with higher scores indicating more problematic drinking.

Drug use

Participants indicated whether or not they had used cannabis and a number of individual illicit drugs, including cocaine or crack, 3,4-methylenedioxy-methamphetamine (MDMA a.k.a. ecstasy), gamma-hydroxybuterate (GHB), ketamine, amphetamines, or psychedelics (lysergic acid diethylamide – LSD, phencyclidine – PCP, mescaline or mushrooms) in the past 30 days. Data were aggregated into three dichotomous variables indicating whether the participant used any cannabis, stimulants (including cocaine, crack, or amphetamines), or other illicit drugs (ecstasy, GHB, ketamine, or psychedelics).

Sexual behavior

Participants were initially asked if they had ever engaged in sexual intercourse (penile-anal or penile-vaginal) in their lifetime. Those who indicated yes were then asked whether they had anal sex (insertive or receptive) with a casual partner (someone who was not their main partner) in the past 30 days. Responses were used to create a dichotomous variable indicating whether the participant had any anal sex with a casual partner (versus none) in the past 30 days.

Mpox diagnosis and vaccination for mpox and COVID-19.

Recent mpox diagnosis was assessed using a single item. “In the past 6 months, have you been diagnosed with mpox by a medical provider?” Responses options were dichotomous (yes or no). Regardless of diagnostic history, respondents were asked, “As of today, have you received a vaccine to prevent mpox infection?” Response options included: 1) Yes I have received the first dose; 2) Yes, I have received the first and the second dose; 3) No, but I have an appointment to receive the first dose; 4) No, and I do not have an appointment; and 5) No, and I do not plan on getting the vaccine.

Participants also provided information about severe acute respiratory syndrome (SARS)-CoV-2 (COVID-19) vaccination. They responded to a single item, “Have you received a COVID-19 vaccine?) by indicating either “Yes – I have received at least 1 dose” or “no.”

Analytic Plan

Data management and analyses were conducted using SPSS version 28. Qualtrics’ automated fraud detection monitored for mass completion of surveys by an automated response generator or human respondent. In addition, survey responses were inspected for duplicates based upon convergent information from IP addresses, contact information, and demographic responses. Responses were also screened for patterned or improbable responses.

For descriptive analyses, participants were organized into 3 groups based on their reported mpox vaccination status: those who had received at least one dose of a mpox vaccine; those who had not yet been vaccinated but who had an appointment to receive one; and those who had not received a vaccine and had no appointment scheduled. Analyses of variance and χ² tests of independence evaluated bivariate associations between mpox vaccination status and demographic characteristics as well as constructs of primary interest (substance use, relationship functioning, and sexual behavior).

Subsequently, a multivariable logistic regression model was calculated. The model utilized a dichotomous mpox vaccination variable, which distinguished between those who had received at least one dose of the vaccine and those who had not, as the dependent variable. Standard errors were initially calculated using a model-based estimator and subsequently cross-checked for consistency against those produced using robust estimation procedures. Overall model fit was evaluated by the Hosmer-Lemeshow Goodness of Fit Test and a log-likelihood goodness of fit test. The latter compared the model specified to a null model. Predictors in the specified model included: demographic characteristics, relationship variables (relationship status and sexual agreement as well as relationship functioning) and substance use indicators (AUDIT -C scores, cannabis use, and other illicit drug use) and the occurrence of anal sex with casual partners.

Relationship functioning and relationship length were incorporated into the model as “hurdle covariates.” This novel analytic procedure draws upon the established convention of hurdle modeling (Hu et al., 2011). Typical hurdle modeling is applied to an outcome that has two components – one binary (modeled using a logistic distribution) and one count (modeled as a Poisson or Negative Binomial distribution). The binary component differentiates between those who had no instances of the outcome and those who had at least one instance. The count component quantifies the number of times the event occurred among people who had at least one occurrence.

In these analyses, we applied the logic of hurdle modeling to predictor variables that are relevant only to partnered SMM. To achieve this, we first mean centered relationship functioning. Then, individuals who identified as single – and therefore were missing data on these variables because they had no relationship to report on – were assigned values of zero. The modified predictor variable was then incorporated into the model along with dummy codes indicating relationship status (single, partnered-monogamous, or partnered non-monogamous). Analogously, dichotomous relationship length was incorporated as a dummy code (0 = < 2 years and 1 = 2 years or more) with single individuals assigned a score of zero.

The equation resulting from the use of these modified hurdle covariates is depicted below. Note, other covariates are omitted for simplicity and the natural log of Y is used to acknowledge the linking functioning associated with the linear prediction of a binary outcome.

$$\text{Ln}[\mathrm{E}(\mathrm{Y})]={\mathrm{B}}_{\text{intercept}}+{\text{BX}}_{\text{partnered-monogamous}}+{\text{BX}}_{\text{partnered-nonmonogamous}}+{\text{BX}}_{\text{modified}\text{relationship}\text{functioning}}+{\text{BX}}_{\text{modified}\text{relationship}\text{length}}$$

In the resulting equation, the intercept (${\mathrm{B}}_{\text{intercept}}$) is necessarily the expected value of the outcome for single men because only they have values of zero on all predictors. The coefficients associated with dummy codes indicating relationship status (${\mathrm{B}}_{\text{partnered-monogamous}}$ and ${\mathrm{B}}_{\text{partnered-nonmonogamous}}$) are calculated at average levels of relationship functioning and relationship duration of less than 2 years – because these are the zero values on each respective predictor. The value or level of relationship functioning at which the coefficients for relationship status dummy codes are calculated can then be modified by re-centering relationship functioning at various levels above or below the mean and then assigning single individuals again to be a score of 0. This is analogous to the established procedure of testing simple-main effects (of one predictor across levels of another) when two predictors are involved in an interaction. In a hurdle covariate context, relationship functioning is always re-centered first – while those who are single have missing values on the variable. Analogously, coefficients for relationship status dummy codes can be calculated at the value of relationship length equal to 2 years or more by reversing the dummy coding for this variable (such that 0 = 2 years or more) and assigning single individuals a score of zero after the reversal.

RESULTS

In total, 9088 unique participants began the survey after viewing consent information and 3411 (37.5%) of them completed it. Among these participants, 2671 met eligibility criteria for inclusion. Examination of data from these respondents revealed that 51 had been diagnosed with mpox by a physician in the past 6 months. Because a positive diagnosis for mpox would substantively alter factors associated with vaccination, these participants were excluded from subsequent analyses. This resulted in a final sample of 2620.

Table 1 contains descriptive data for the analytic sample. While most participants identified as White (62.5%), 10.5% identified as Black, African American, or Afro-Caribbean, 18.1% as Latino/x or Hispanic; 3.4% as Asian or Pacific Islander, and 5.5% identified in some other way. The majority of respondents indicated an HIV negative or unknown sero-status (79.1%). Most identified as gay (83.5%) or bisexual (11.5%). The remaining 131 (5%) participants identified as queer (1.9%), pansexual (2.1%), demisexual (0.3%), heterosexual (0.3%) or selected an “other identity” option (0.3%). Most participants were partnered in either a monogamous (12.3%) or non-monogamous (57.1%) relationship. Among partnered SMM, 74.7% had been in their relationship for 2 years or more.

Table 1.

Sample characteristics

	Overall	Vaccinated (at least 1 dose)	Scheduled (but unvaccinated)	Unvaccinated	χ2 (df)	p
n(%)	2620 (100.0)	730 (27.9)	124 (4.7)	1766 (67.4)
Race and Ethnicity					13.22 (8)	.105
White/Caucasian	1637 (62.5)	440 (60.3)	70 (56.5)	1127 (63.8)
Black/African American	274 (10.5)	69 (9.5)	16 (12.9)	189 (10.7)
Latino	474 (18.1)	147 (20.1)	25 (20.2)	302 (17.1)
Asian/Pacific Islander	90 (3.4)	33 (4.5)	2 (1.6)	55 (3.1)
Other	145 (5.5)	41 (5.6)	11 (8.9)	92 (5.3)
Geographic Region					53.09 (6)	<.001
Northeast	592 (22.6)	217 (29.7)a	34 (27.4)a	341 (19.3)b
Midwest	468 (17.9)	111 (15.2)a	22 (17.7)ab	335 (19.0)b
Southeast	932 (35.6)	2.5 (28.1)a	37 (29.8)a	690 (39.1)b
West	628 (24.0)	197 (27.0)a	31 (25.0)ab	400 (22.7)b
Sexual Identity					61.19 (4)	<.001
Gay	2188 (83.5)	659 (90.3)a	109 (87.9)a	1420 (80.4)b
Bisexual	301 (11.5)	29 (4.0)a	12 (9.7)b	260 (14.7)b
Other	131 (5.0)	42 (5.8)a	3 (2.4)a	86 (4.9)a
HIV status					283.54 (4)	< .001
Negative - no PrEP	1220 (46.6)	117 (24.2)a	43 (34.7)b	1000 (56.6)c
Negative - PrEP	852 (32.5)	399 (54.7)a	43 (34.7)b	410 (23.2)c
Positive	548 (20.9)	154 (21.1)a	38 (30.6)b	356 (20.2)a
COVID vaccination					73.08 (2)	< .001
No vaccination	219 (8.4)	10 (1.4)	5 (4.0)	204 (11.6)
At least one dose	2401 (91.6)	720 (98.6)a	119 (96.0)b	1562 (88.4)c
Relationship status					86.53 (4)	<.001
Single	818 (31.2)	159 (21.8)a	38 (30.6)b	621 (35.2)b
Partnered - monogamous	322 (12.3)	54 (7.4)a	19 (15.3)b	249 (14.1)b
Partnered - non-monogamous	1480 (56.5)	517 (70.8)a	67 (54.0)b	896 (50.7)b
Anal sex with a casual partner	1496 (57.1)	515 (70.5)a	76 (61.3)b	905 (51.2)c	79.49 (2)	<.001
Drug Use
Cannabis	1183 (45.2)	364 (49.9)a	63 (50.8)ab	756 (42.8)b	12.06 (2)	.002
Stimulant drugs	507 (19.4)	115 (15.8)a	24 (19.4)ab	368 (20.8)b	8.556 (2)	.014
Other illicit drugs	420 (16.0)	144 (19.7)a	23 (18.5)ab	253 (14.3)b	11.801 (2)	.003
	M(SD)	M(SD)	M(SD)	M(SD)	ANOVA
Age	42.30 (13.00)	42.19 (12.16)	42.58 (12.87)	42.32 (13.35)	F(2,2617) = 0.06	.943
AUDIT-C	2.13 (2.59)	3.40 (2.53)a	3.27 (2.60)ab	3.01 (2.61)b	F(2, 2617) = 96.06	.002
Relationship Functioning	49.25 (12.10)	51.39 (10.85)a	48.42 (11.37)ab	48.32 (12.59)b	F(2, 2082) = 14.31	<.001

NOTE: AUDIT-C = Alcohol Use Disorders Identification Test-Concise; PrEP = pre-exposure prophylaxis; ref = referent; Stimulant drugs included cocaine, crack, and amphetamines; other illicit drugs included ecstasy, GHB, ketamine, and psychedelics.

Bivariate associations between mpox vaccination and substance use variables partially conformed to those hypothesized. SMM who received at least one dose of the mpox vaccine scored significantly higher on the AUDIT-C and were significantly more likely to report the use of cannabis and other illicit drugs (ecstasy, GHB, ketamine, or psychedelics) compared to those who had not received a mpox vaccine and did not have an appointment to receive one. Contradicting our initial hypothesis, the reverse association was observed for stimulant drugs (cocaine, crack, or amphetamines). SMM who had received at least 1 dose of the mpox vaccine were significantly less likely to report stimulant drug use compared to those who had not received a dose and did not have an appointment to receive one.

Beyond this, several other bivariate associations were noteworthy. SMM who received a mpox vaccine were more likely to report having anal sex with a casual male partner and to indicate being in a non-monogamous relationship compared to those who were unvaccinated or who were scheduled for (but had not yet received) a vaccination. Among men in relationships, those who received a mpox vaccine had significantly higher scores on relationship functioning compared to those who had not received a vaccine and did not have an appointment to receive one. Finally, COVID-19 vaccination was significantly associated with mpox vaccination. Although a substantial majority (91.6%) of the sample had received at least one dose of a COVID-19 vaccine, 93.2% of those who had not received a COVID-19 vaccination indicated that they had not received a mpox vaccine and did not have an appointment to receive one.

Mpox vaccination: Logistic regression results

Table 2 contains the results of multivariable logistic regression analyses predicting the odds of having received at least one dose of the mpox vaccine at the time of survey completion. The overall model was significant (omnibus χ²(20) = 458.367, p < .001) and the Hosmer and Lemeshow goodness-of-fit test was non-significant (χ²(8) = 7.21, p = .514); both indicating adequate model fit. Consistent with hypotheses, participants who reported anal sex with casual partners in the past 30 days were significantly more likely to have received at least 1 dose of the mpox vaccine. We initially hypothesized that substance use would be negatively associated with mpox vaccination in models controlling for sexual behavior. Consistent with this, participants who reported stimulant drug use were less likely to be vaccinated. Contrary to hypotheses, AUDIT-C scores and cannabis use were not significantly associated with mpox vaccination status. Also contrary to hypotheses, the use of other illicit drugs (ecstasy, GHB, ketamine, or psychedelics) was positively associated with mpox vaccination.

Table 2.

Multivariable logistic regression: Mpox vaccination

	B	95% Confidence Interval	Odds Ratio	p
Intercept	−2.639	(−3.196, -2.081)	0.071	<.001
Age	−0.002	(−0.010, 0.006)	0.998	.563
Race and Ethnicity (ref = White/ Caucasian)
Black/African American	0.359	(0.025, 0.693)	1.432	.035
Latino	0.205	(−0.051, 0.460)	1.227	.116
Other	0.247	(−0.171, 0.665)	1.280	.246
Asian/Pacific Islander	0.386	(−0.122, 0.895)	1.472	.136
Sexual Orientation (ref = bisexual or other)	0.595	(0.298, 0.892)	1.813	<.001
Geographic Region (ref = Northeast)
Midwest	−0.708	(−1.007, -0.410)	0.492	<.001
South	−0.757	(−1.008, -0.506)	0.469	<.001
West	−0.303	(−0.569, -0.038)	0.738	.025
Current PrEP prescription	1.441	(1.215, 1.666)	4.223	<.001
HIV status (ref = negative or unknown)	0.879	(0.604, 1.153)	2.407	<.001
Relationship status (ref = single)
Partnered - monogamous	0.106	(−0.307, 0.519)	1.112	.616
Partnered - non-monogamous	0.386	(0.070, 0.703)	1.472	.017
Relationship functioning	0.025	(0.015, 0.036)	1.026	<.001
Relationship length (ref = < 2 years)	0.368	(0.086, 0.651)	1.445	.011
AUDIT-C	0.032	(−0.005, 0.070)	1.033	.092
Drug use
Cannabis	0.073	(−0.126, 0.272)	1.076	.473
Stimulant drugs	−0.597	(−0.901, -0.294)	0.550	<.001
Other illicit drugs	0.328	(0.028, 0.628)	1.388	.032
Anal sex with a casual partner	0.525	(0.305, 0.746)	1.691	<.001

NOTE: AUDIT-C = Alcohol Use Disorders Identification Test-Concise; PrEP = pre-exposure prophylaxis; ref = referent; Stimulant drugs included cocaine, crack, and amphetamines; other illicit drugs included ecstasy, GHB, ketamine, and psychedelics.

Model results indicated that relationship status and relationship functioning were significantly associated with the odds of mpox vaccination in ways that largely conformed to hypotheses. Results indicated that these findings must be understood in the context of relationship length as well. Figure 1 displays the predicted odds of mpox vaccination among single and partnered SMM as a function of PRCQ scores and relationship length. At average levels of relationship functioning, partnered SMM in non-monogamous relationships who had been with their partners less than 2 years were significantly more likely to be vaccinated than single men. Single men and those in monogamous relationships who had been with their partners less than 2 years did not differ significantly from one another. The odds of mpox vaccination were positively associated with relationship length among partnered men. At average levels of relationship functioning, SMM in both monogamous (B = .474, 95%CI: 0.064, 0.885; OR = 1.607, p = .024) and non-monogamous (B = .755, 95%CI: 0.525, 0.985, OR = 2.217, p < .001) relationships of 2 years or longer were more likely to be vaccinated than single men.

Figure 1.

Predicted odds of mpox vaccination: relationship status, sexual agreements, relationship length and relationship functioning

NOTE: Predicted odds of vaccination for SMM in relationships are displayed on the left side of the x-axis along a gradient of relationship quality. The predicted score for single respondents is displayed by the bar on the right side of the x-axis.

Post hoc tests of the simple main effect of relationship status were then conducted at various levels of relationship functioning to elucidate associations with vaccination. At low levels of relationship functioning (1 standard deviation below the mean), SMM in non-monogamous relationships of duration 2 years or more were significantly more likely to be vaccinated than single men (B = 0.477, 95%CI: 0.184, 0.710, OR = 1.564, p < .001); however, those in relationships with duration less than 2 years did not differ significantly from single men. At low levels of relationship functioning, SMM in monogamous relationships did not differ significantly from single SMM regardless of relationship length. At very low levels of relationship functioning (2 standard deviations below the mean) partnered SMM did not differ significantly from single men, regardless of sexual agreement or relationship length. At high levels of relationship functioning (1 standard deviation above the mean), the pattern of significance was comparable to that observed at average levels of relationship functioning. Partnered SMM in relationships for 2 years or longer were significantly more likely to be vaccinated than single SMM, regardless of sexual agreement (B = 0.782, 95%CI: 0.374, 1.19, OR = 2.186, p < .001 and B = 1.063, 95%CI: 0.805, 1.320, OR = 2.895, p < .001 for monogamous and non-monogamous SMM, respectively). In addition, SMM in non-monogamous relationships of duration less than 2 years were significantly more likely to be vaccinated than single SMM (B = 0.694, 95%CI: 0.361, 1.028, OR = 2.003, p < .001). Those SMM in monogamous relationships less than 2 years did not differ significantly from single men. At very high levels of relationship functioning (2 standard deviations above the mean), partnered SMM in both monogamous and non-monogamous relationships were significantly more likely to be vaccinated than single SMM, regardless of relationship length. Across these analyses, SMM in monogamous relationships did not differ significantly from those in non-monogamous relationships (B = .281; 95%CI: .083, 0.644; OR = 1.324, p =.130).

With regard to demographic covariates, SMM who identified as gay were significantly more likely to be vaccinated compared to those who identified as bisexual or in some other way. Men on PrEP and those who were HIV positive were also more likely to be vaccinated compared to HIV negative men without a PrEP prescription. Those residing in the Midwestern, Western, and Southeastern US were all significantly less likely to be vaccinated than those living in the Northeast. Participants who identified as Black, African American, or Afro-Caribbean were significantly more likely to be vaccinated than White or Caucasian identified participants. Age was unrelated to vaccination status in multivariable analyses.

DISCUSSION

Findings from the current study provide substantive information about correlates of mpox vaccination among SMM during a critical period in the US response to the outbreak. Those SMM who reported sex with casual partners were more likely to be vaccinated. The use of stimulant drugs (cocaine, crack, or amphetamines) was associated with decreased odds of vaccination; meanwhile, the use of other illicit drugs (ecstasy, GHB, ketamine, or psychedelics) was positively associated with vaccination. Finally, mpox vaccination was positively associated with having a non-monogamous sexual agreement, relationship functioning, and relationship length among those SMM in relationships. These factors then contextualized differences in vaccination between single and partnered SMM.

A priori hypotheses anticipated that SMM who received a mpox vaccine would be more likely to report sex with casual partners. The association might arise from vaccination guidance that prioritized SMM with multiple sex partners (Kaiser Family Foundation, 2022). It could arise because SMM who were unvaccinated reduced their sexual partners, consistent with prevention messaging at that time (Delaney, 2022). It is also plausible that SMM who received the vaccine experienced a reduction in mpox-related concerns and subsequently increased their sexual activity. Findings from this cross-sectional survey align with all of these potential mechanisms and these mechanisms themselves are not mutually exclusive. Future studies, ideally with access to longitudinal data, are essential to elucidate the temporal order of events such as the receipt of prevention messaging, vaccination, and sexual behavior.

Findings from this study suggest that substance use has a complex association with mpox vaccination – one that may vary across substances. This was most evident in the contrasting findings on stimulant (cocaine, crack, or amphetamines) and other illicit drug (ecstasy, GHB, ketamine or psychedelics) use. In bivariate and multivariable analyses (that included sexual behavior with casual partners as a covariate), the former was negatively associated with vaccination; meanwhile, the latter was positively associated with vaccination. There is some evidence to suggest that the use of stimulant drugs among SMM is a particularly salient barrier to receiving sexual-health related vaccinations generally (Nadarzynski et al., 2021). Further research is necessary to evaluate whether the impact of prevention messaging varies with specific drugs of abuse.

Differences in vaccination between single SMM and those in relationships varied as a function of sexual agreement (monogamous versus non-monogamous), relationship functioning (better relationship quality was associated with greater likelihood of vaccination), and relationship length (longer duration was associated with greater likelihood of vaccination). At the lowest level of relationship functioning (2 standard deviations below the mean), differences between single and partnered SMM were non-significant. At low levels of relationship functioning (1 standard deviations below the mean), differences began to emerge. SMM in non-monogamous relationships of 2 years or longer were significantly more likely to be vaccinated compared to single men. At average levels of relationship functioning and above, all SMM in non-monogamous relationships and those in monogamous relationships of 2 years or longer were significantly more likely to be vaccinated compared to single men. At the highest level of relationship functioning probed (2 standard deviations above the mean) all partnered SMM were more likely to be vaccinated than those who were single; meanwhile, at the lowest level of relationship functioning probed (2 standard deviations below the mean) all relationship status differences were non-significant.

These findings align with the idea that the health-related benefits of being partnered are predicated upon relationship quality and may emerge as relationships develop (i.e., the longer a couple has been together). This is entirely consistent with interdependence theory generally (e.g., Rusbult et al., 1998; Rusbult & Van Lange, 2003) and the theoretical framework put forward by Lewis et al. (2006). Across contexts, partners who are more satisfied with, committed to, and invested in their relationships are more likely to consider the impact of their behavior on one another and their relationship overall (Kelley & Thibaut, 1978; Rusbult & Van Lange, 2003). Specifically related to health, Lewis et al. (2006) would suggest that partners in higher quality relationships are more likely to consider mpox a shared health threat. Personal infection might pose a risk to one’s partner and the relationship overall (Lewis, Gladstone, et al., 2006). This provides an additional source of relational motivation for vaccination uniquely relevant to partnered SMM – one that would be expected to be most potent for those in the highest quality relationships. In addition, research on HIV (e.g., Gamarel & Revenson, 2015; Lewis, Gladstone, et al., 2006; Martins et al., 2021) and COVID-19 prevention (Starks et al., 2021) would suggest that SMM in higher quality relationships are more successful at joint coping or working together to reduce a health threat. They employ more positive (supportive) and fewer negative (aversive) social control techniques to influence one another’s health behavior (Lewis, McBride, et al., 2006).

Relationship length is commonly included as a covariate in research on relationship quality and health (e.g., Johnson et al., 2012; Stephenson, White, Darbes, et al., 2015). Its salience as a predictor in these analyses may arise for several reasons. It is plausible that relationship length serves as an additional, proxy indicator of relationship commitment. Agnew et al. (1998) for example suggested that one aspect of the larger construct of commitment was intention to persist in the relationship. It is also possible that relationship length serves as a proxy indicator of investment (Rusbult et al., 1998) or resource sharing. Rusbult et al. (1998) suggested that couples who share critical resources (e.g., time, finances, property etc.) are more likely to consider one another’s needs in times of conflict. Practically speaking, this kind of resource sharing requires time to accomplish in a relationship. Finally, it is also plausible that couples who have been together longer have simply had more opportunities to practice and develop effective communication skills. This might increase their chances of successfully discussing a shared health threat (such as mpox infection) and working together to address it.

Several demographic findings are noteworthy. First, these data highlight the vulnerability of SMM who do not identify as “gay”. Findings here align with Nadarzynski et al. (2021), who found that identifying as “gay or homosexual” (versus bisexual or in some other way) was associated with increased vaccine acceptability for human papillomavirus (HPV), meningitis C (MenC), and hepatitis A/B (HAV) (HCB). Aligned with the conclusions of Nadarzynski et al. (2021), several factors may diminish the effectiveness of health outreach to bisexual SMM. For example, SMM who identify as bisexual may experience stigma within the larger SMM community (Arriaga & Parent, 2019). Additionally, they may not feel comfortable discussing their same sex practices with their health care provider (Nadarzynski et al., 2021; Newman et al., 2013). Alternatively, public health messages formulated for gay-identified SMM may not resonate with or impact bisexual identified SMM to the same extent (Feinstein & Dyar, 2017).

A number of clinical and policy implications follow from these findings. First, these data point to the need to improve outreach to several key subgroups of SMM who may be experiencing stigma in healthcare settings. Most notably SMM who identify themselves as bisexual or another sexual identity besides gay may not feel comfortable discussing their sexual practices with a health care provider (Nadarzynski et al., 2021). In addition, these data highlight the need to reduce barriers to healthcare access associated with drug use – particularly stimulant drugs.

Second, there are indications in these data that health prevention can be cross-promoted. SMM who were on PrEP for HIV prevention and those who were vaccinated for COVID-19 were more likely to be vaccinated for mpox. This tendency for prevention behavior to generalize across illnesses is supported by similar findings wherein PrEP has been positively associated with HPV vaccination among SMM (Pleuhs et al., 2022). Additionally, some SMM have indicated that being on PrEP facilitated further engagement in primary healthcare services (Sewell et al., 2021). Integrated services for SMM incorporating sexual health services with general primary care prevention may provide an added benefit for mpox vaccination.

Third, these data provide an initial opportunity to reflect on the impact of mpox prevention outreach. Although it is not possible to evaluate whether sexual behavior changed in direct response to prevention messaging or vaccination in this sample, the mere fact that unvaccinated SMM were less likely to report anal sex with casual partners is promising. At the broadest level, public health officials would likely hope that those SMM at highest risk of infection (those unvaccinated) would be the least likely to engage in behaviors that might transmit the virus. At the same time, these data suggest there is substantial room for improvement in prevention messaging (51.2% of SMM who were unvaccinated and had no appointment to be vaccinated still reported anal sex with a casual partner). Furthermore, bivariate analyses here suggest that those SMM who schedule a vaccination appointment (but have not yet received it) may experience some increased risk of exposure. In this sample, 61.3% of the SMM who had a vaccine appointment (but had not received a dose) reported anal sex with a casual partner. While significantly lower than the proportion of vaccinated SMM who reported sex with a casual partner, this proportion is higher than that observed among unvaccinated SMM who had no appointment. In cross-sectional data, it is not possible to determine if sexual exposure might be motivating SMM to make a vaccination appointment, or if merely making an appointment results in disinhibition of sexual behavior. Either way, the association observed here suggests that SMM may benefit from prevention messaging (to enhance self-protective behaviors and risk of onward transmission) delivered even at the point of scheduling a vaccination appointment.

Finally, brief interventions have been developed to reduce HIV-related risk and related drug use among SMM (Parsons et al., 2014). These effects appear particularly durable among single SMM (Starks & Parsons, 2018) and tailored adaptations have been developed for SMM in relationships (Starks, Robles, Pawson, et al., 2019). It is plausible that these interventions could be leveraged and adapted to guide the work of public health professionals seeking to enhance mpox vaccine uptake among single SMM at high risk for infection. In addition, these data suggest that sexual health interventions developed to address HIV infection risk among male couples (e.g., Grabbe et al., 2014; Newcomb et al., 2020; Starks, 2022; Starks et al., 2022; Starks, Dellucci, et al., 2019; Stephenson et al., 2022) might be leveraged to address mpox infection risk among partnered SMM.

These findings must be understood in the light of a number of limitations. First, these data must be understood in the context of the time period in which they were collected. Vaccination supply, and associated restrictions on vaccination, have fluctuated substantially in the months since the outbreak began in the US. Importantly, the data collection period (July 28 – September 22, 2022) was characterized by an acceleration in vaccination activity and broadening of vaccine eligibility. Approximately 80% of vaccine doses delivered in the US were administered in that period (CDC, 2022a). As access and urgency changes, correlates of vaccination may likewise change.

Second, the sample was restricted to cismale participants, findings may not generalize to trangender individuals at risk for mpox. Generalizability is also limited by recruitment venue. SMM were recruited online on social networking applications that cater primarily to those seeking sexual partners (e.g., Jack’d and Scruff). Findings may not generalize to SMM who are not active in such spaces. The term SMM encompasses a broad range of identity and behavior. While most of the sample (95%) identified as gay or bisexual, 5% identified in other ways, including a small number (n = 9; 0.3%) of heterosexually identified participants. Caution is warranted in generalizing inferences to subgroups of SMM who may be under-represented in the current sample and responses from heterosexual men recruited in venues that predominantly facilitate sex among men and with transgender people are likely not representative of the larger population of men who identified as heterosexual.

Third, there is variability across behavioral health studies in the way in which drug use is operationalized. In this paper, we distinguish between cannabis, stimulant drug use (cocaine, crack, and amphetamines), and the use of other illicit drugs (GHB, ketamine, ecstasy, and psychedelics). This decision is in keeping with a small body of sexual health research, wherein researchers have evaluated whether specific categories of illicit drugs (e.g., stimulants, “club drugs” or “chem-sex drugs”) are uniquely associated with sexual health outcomes. See Viamonte et al. (2022) for a review. At the same time, variations in recreational drug formulations complicate such distinctions. For example, some ecstasy tablets may contain stimulants such as methamphetamine, cocaine, or caffeine in addition to MDMA (Department of Justice, 2020).

Fourth, as mentioned previously, data were generated by a cross-sectional survey. This precludes any conclusions about causal associations or temporal order. For example, although these data suggest that those SMM who reported sex with casual partners were more likely to be vaccinated, it is not possible to determine whether the indicated sex with casual partners occurred before or after vaccination. In addition, due to the brief nature of the survey, data related to putative mediators (i.e., transformation of motivation or shared health threat) are not available.

Despite these limitations, the current study represents one of the most comprehensive examinations of correlates of mpox vaccination among SMM published to date. Broadly, these findings suggest that SMM who received the mpox vaccine are more likely to have anal sex with casual partners. They highlight the vulnerability of SMM who use stimulant drugs and also those who identify in ways other than gay. Finally, these findings provide a nuanced examination of the differences between single and partnered SMM. They suggest that the health-related benefits of being partnered emerge as a function of sexual agreement, relationship length, and also relationship functioning in ways that align with interdependence theory (Kelley & Thibaut, 1978; Rusbult & Van Lange, 2003).