Characteristics of Sex Events, Partners, and Motivations and Their Associations with HIV-Risk Behavior in a Daily Diary Study of High-Risk Men Who Have Sex with Men (MSM)

Abstract

In the United States, men who have sex with men (MSM) continue to be at high-risk for HIV and other STIs, and condoms represent the most popular, affordable, and accessible method of prevention. Although a vast body of research has explored various factors associated with condom use in MSM, fewer studies have explored situation-level characteristics that affect their decisions about sex partners and condom use. Daily diary studies are well-suited to help improve our understanding of these event-level factors in detail, including the sex events themselves, partner characteristics, and motivations. As part of a larger study using ecological momentary assessment methods, high-risk MSM completed daily diary surveys about their sexual behavior on their smartphones each day for 30 days. This study explored detailed descriptive characteristics of sex events, partner characteristics, and motivations for sex and condom use, and examined whether specific aspects of these characteristics were associated with having condomless anal sex (CAS) with high-risk partners. High-risk CAS was common among MSM, with the majority of participants having met their partners online and many reporting sex the same day they met. Results showed that the odds of CAS were not higher with partners met online versus those met in other ways, but MSM were more likely to have asked online partners about their HIV status and testing history before sex. The odds of engaging in high-risk CAS was higher when MSM reported intimacy or self-assurance motives. Not having condoms readily available was a CAS motivation reported more commonly when MSM had sex with high-risk partners. Findings suggest that interventions should incorporate strategies that help MSM be safer specifically when meeting partners online and when having sex for intimacy or reassurance. Interventions that remind MSM to carry condoms at opportune moments may also help reduce some HIV risk.

INTRODUCTION

In the United States (US), HIV incidence has declined overall in recent years, but rates of new infections remain high particularly among men who have sex with men (MSM; ^{1, 2}), and continue to increase among certain subgroups of MSM, including young MSM aged 25–34 (3). If this trend continues, the US Centers for Disease Control and Prevention (CDC) suggests that 1 in 6 MSM are at risk for becoming infected with HIV in their lifetimes (4). Nearly all of these new infections occur as a direct result of sexual risk behavior, which among MSM involves having insertive or receptive anal sex without using an effective method of prevention (5, 6). As such, behavioral approaches to prevention continue to play a key role in overall HIV prevention efforts (7). Although recent biomedical innovations like pre-exposure prophylaxis (PrEP) have provided MSM with more prevention options than ever before (8–10) and may ultimately be as popular as other methods, condoms are currently the most widely accessible, cheapest, and most frequently used form of HIV prevention globally (11). Condoms are also the only form of HIV prevention that is also effective in preventing many other common sexually-transmitted infections (STIs; ¹²) which contribute to onward HIV infections and disease burden in general (13, 14). Together, this landscape points to the need for continued research that improves our understanding of condom use and the factors that affect it, particularly among at-risk populations like MSM.

A vast body of existing research has explored various factors that are associated with condom use among MSM. However, most of this research has focused on person-level factors, like broad demographic characteristics (e.g., whether certain age groups tend to use condoms more so than others), general behavioral tendencies, and traits (e.g., sexual compulsivity). Yet, the decision to use condoms during anal sex with a particular partner is highly situational, and the same person often choses to use condoms with some partners and not others, or during some sex acts versus others (e.g., receptive anal sex versus insertive; ¹⁵). Several studies from the comparatively smaller event-level literature have suggested that 71–83% of the overall variance in condom use occurs at the event-level (16–18), meaning that condom use varies much more across situations than across individuals. Given these findings, studies focused on understanding situational factors that affect condom use among MSM have become increasingly common in recent years. Most of this research has used longitudinal survey techniques that involve asking participants to recall the last few times they had sex and the circumstances involved, so as to identify consistent patterns (15). Recall-based, event-level studies like these have so far identified a number of situational and contextual factors that influence whether MSM use a condom during anal sex. These include their relationship with their partner (19, 20), where/how they met (21–23), and substance use (15, 24). So far, alcohol and stimulant drug use have emerged as two of the most consistent predictors of HIV-risk behavior in MSM (25–28), but few event-level studies have focused on understanding prevention decisions among MSM who use these drugs. Although recall-based event-level studies like these have made important contributions to the content of risk reduction interventions (15), their reliance on participants’ ability to accurately remember characteristics of sex events that may have happened weeks or months in the past limits the amount of detail that can be validly assessed about them. Several situational factors that may be important to decisions about condom use, including thoughts and “motivations” about sex and condom use, may be impossible to accurately assess at any significant delay, because they may be subject to later rationalization (29).

In the last decade, more intensive event-level studies that assess situation-level factors closer to when they happen, such as daily diary studies, have become more common in part because the growth of technology has made these studies easier and cheaper to conduct. In these studies, technology-assisted methods like e-mail, text messages, or native smartphone apps are often used to prompt or collect surveys about participants’ sexual behavior each day (30). This approach has the advantage of assessing aspects of sexual situations within 24 hours of their occurrence, increasing the accuracy of self-report (31, 32) and allowing researchers to capture more events (and as a result, more variation in those events over time). In principle, it also allows researchers to capture more detailed information about sex events that would not be possible through recall (29, 32, 33), including precise data about partner characteristics (e.g., how long participants had known each one, where they met), the sex events themselves (e.g., what time of day sex occurred), correct/complete use of condoms, and participants’ motivations around the time sex occurred (e.g., important reasons for having sex, not using condoms). Although at least two such studies have used this approach to study the influence of participants’ motivations for sex and not using condoms (19, 34), the vast majority of daily diary studies have been focused on understanding the contribution of affect and substance use before sex to later choices about condom use (35–38). These studies have undoubtedly made important discoveries about how these states can influence condom use decisions, but most have provided limited data about the details of sex events themselves or about partner characteristics and whether aspects of events/partners may be associated with choices about sex and condom use. Providing thorough situation-level data about whether factors like these are associated with risk could help direct guidance about situations that lead to risk of exposure to HIV and other STIs.

This study used daily diary surveys from a broader 30-day ecological momentary assessment (EMA) study of high-risk MSM to provide basic descriptive information about sex events, including partner characteristics and participants’ motivations for sex and condom use around the time sex occurred. We also explored whether aspects of these events, partners, and motivations were associated with ultimately having condomless anal sex (CAS) when their partner was high-risk. Although we also collected data on several other possible antecedents and precipitants of HIV-risk behavior in this study via other concurrent sampling techniques (e.g., intentions, affect, attitudes), it is not feasible to test the influence of all of these states in a single manuscript. As such, this manuscript focuses on describing the detailed data provided in daily diary surveys about sex events themselves, and future manuscripts will address the influence of earlier related behaviors and states in more detail.

METHODS

Participants

Participants (N = 100) were recruited from gay-oriented smartphone dating apps (e.g., Grindr, Scruff), social networking sites (e.g., Facebook, Instagram), and via in-person outreach (e.g., flyers, business cards) in the northeastern US from January 2016 to October 2018. Eligible participants were: (1) 18+ years old, (2) assigned male sex at birth, (3) currently male gender, (4) HIV-negative or unknown status, (5) able to read and speak English fluently, and (6) not currently prescribed or taking PrEP. They also reported (7) having had CAS with a non-exclusive partner at least once in the past 30 days, and (8) having consumed five or more drinks on a single occasion at least once in the past 30 days. Given that our study period was 30 days, we focused on individuals with very recent risk behavior so as to increase the probability that at least some sex events would occur during the study period. We focused on those who were not currently taking PrEP because this study was focused on understanding in-the-moment decision-making among MSM who rely on condoms as their primary method of prevention. We focused on those with a recent heavy drinking day because heavy drinking is a key risk factor for HIV infection (39–41) and this level of drinking is common among MSM (42–44). Since we wanted to study sexual decisions with at least some “casual” partners, participants were excluded if they (1) had been in a mutually sexually-exclusive relationship for three months or longer. Participants were also excluded if they (2) were currently receiving counseling or medications for alcohol or drug problems, since our aim was to study non-treatment-seeking MSM. Finally, for safety reasons, participants were also excluded if they were (3) currently receiving treatment for serious mental illness (e.g., schizophrenia, bipolar disorder) or (4) had injected drugs within the last three months.

Procedures

Participants were first screened online before being contacted by staff to schedule an appointment to learn more about the study and, if interested, to enroll. Participants who lived nearby our offices in Providence, RI could elect to attend this appointment in-person, or participants could choose to meet with staff via videoconferencing. Another manuscript has shown that there were no substantive differences in response rates, reactivity, or haphazard responding among participants who enrolled remotely versus in-person (45). During these enrollment appointments, staff reviewed the study’s procedures, obtained informed consent, and then walked participants through downloading the MetricWire app to their personal smartphones, which was used to collect survey data throughout the study. Staff then provided thorough training on how to use the app, the types of surveys they would be asked to complete, and walked participants through a typical day on the study, demonstrating how to initiate various types of assessments. These walkthroughs also explained the meaning of each question in each survey. This study focuses on daily diary surveys, which participants were asked to complete each morning as soon as possible after they woke up. Responses submitted after 5 p.m. each day were considered missing for that day. Participants were also coached to achieve response rates of 100% for the daily diary assessments and received feedback throughout the study period via email about their response rates. If participants fell below the target of 100% for these prompts in a given week, staff would contact participants by phone to discuss ways of improving, including keeping their phone in view, making sure their ringer was on, and setting alarms for morning surveys. Participants were compensated based on their response rates, earning $2 for each daily diary survey they completed, plus a “bonus” of $10 for every 10 days they submitted 100% of these surveys. Together with compensation provided for other study surveys, participants could earn a total of $210. All procedures were reviewed and approved by Brown University’s IRB.

Measures

Baseline surveys

After providing informed consent and enrolling in the study, participants completed a thorough baseline assessment to assess several person-level characteristics. These surveys included basic demographic information, as well as the Alcohol Use Disorders Identification Test (AUDIT; ^{46, 47}) and Drug Abuse Screening Test (DAST-10; ^{48, 49}), which are reported here to help describe the sample. AUDIT scores of 8 or higher were considered evidence of possible drinking problems (50) and DAST-10 scores of 3 or higher were considered evidence of possible drug-related problems (51).

Daily diary surveys

Each day, participants reported sexual behavior that had occurred since the previous morning survey. Participants were asked to indicate the number of partners they had oral, anal, or vaginal sex with (0–4 partners per day). Then, for each partner, they were asked to report whether this had occurred with a new partner (someone they had not had oral, anal, or vaginal sex with before), a partner they were mutually sexually exclusive with, how long they had known this partner (we just met yesterday to a year or longer), where they met this partner (e.g., bar/club, party, online, etc.), and this partner’s gender (male, female, transwoman, transman). Participants were also asked to report whether, before sex began, they had asked each partner about whether they were taking PrEP, and if so, what their response was (yes, no). They were also asked to report whether, before sex began, they had asked each partner about their HIV status, and if so, what their partner’s response was (positive, negative, don’t know). Items also inquired about whether participants had asked each partner about the last time they were tested for HIV. Then, participants were asked to report the approximate time at which sex began with this partner (using a digital time question type), which sex acts they engaged in (oral, insertive anal, receptive anal, vaginal sex) and whether they used a condom for each act. For each partner, participants also reported whether they drank or used drugs with each partner, and if so, what types of drugs were used. Participants could select from among ten categories of drug types including marijuana, cocaine, methamphetamine, heroin, ecstasy, MDMA, psychedelics, and prescription stimulants, painkillers, and sedatives, among others, and could select multiple types. We also assessed participants’ motivations for having sex with each partner by asking them to identify “reasons that were important in [their] decision to have sex with [this partner].” Six checkbox response options were available that were adapted to reflect each of the six types of motives identified in Cooper, Shapiro, and Powers (52), including intimacy (“to feel emotionally close to my partner”), enhancement (“because it feels good”), self-affirmation (“to re-assure myself that I am attractive ”), coping (“to cope with feeling upset”), partner approval (“to keep my partner content or interested”), and peer approval (“friends will think less of me if I don’t”). Participants could identify multiple motives. Finally, when participants reported not using a condom during anal or vaginal sex, they were asked to respond to a follow-up checkbox item asking which reasons were most important in their decision not to use a condom with that partner. Response options were adapted from the most common reasons reported by MSM in past studies assessing motivations (or “justifications”) for using/not using condoms (53, 54), including “I trusted that he/she was ‘clean’,” “I didn’t have condoms with me,” and “I didn’t want to ruin the moment.” Participants could select multiple reasons per act.

Analyses

We first calculated descriptive statistics for all reported sex partner characteristics, the sexual behaviors reported for each, and rates of prevention method use (condoms, PrEP) with each partner and act. For these analyses, we characterized “high-risk partners” as those that (1) were not on PrEP, (2) were partners that participants were not mutually sexually-exclusive with, or (3) were partners that participants were sexually-exclusive with, but were of unknown HIV status (either because participants had not ever asked about HIV status or were unsure for other reasons). CAS with these partners was often the focal outcome of interest, which we compared with condom-protected anal sex and/or oral sex only. This approach construes CAS events with mutually sexually-exclusive partners who had been recently tested as involving less risk, in addition to condom-protected anal sex and oral sex only. Although we did not ask participants to report whether the partners they asked about HIV status and were HIV-positive were undetectable, we elected not to classify CAS events with sexually-exclusive HIV-positive partners as “high-risk CAS events” because all participants indicating that these partners were “HIV positive” meant that participants had also reported having asked about their HIV status and testing history in the past, and that these conversations were also likely to have involved a conversation about treatment status. However, only one CAS event with a HIV-positive, mutually sexually-exclusive partner was reported during the study period. Three additional CAS events with HIV-positive, non-mutually exclusive partners were also reported, but were classified as “high-risk” CAS given the non-mutually sexually exclusive criterion.

Next, we calculated frequencies and percentages to explore whether there were differences in CAS across several partner characteristics, including new partners, how long participants had known each partner, sexually non-exclusive partners, and partners on PrEP. We then used similar descriptive statistics to examine whether there were differences in sexual behavior/condom use across the times of day that sex occurred, where participants met their partners, common motivations for sex, and alcohol/drug use with sex partners. Since these daily diary data were clustered within individuals, we also fit random-effects logistic models to test whether high-risk CAS events (versus lower-risk sex) varied across specific characteristics of sex events (e.g., length of time partners were known, sex motives). Since high-risk CAS is a binary outcome, each specified a binomial distribution and used a logit link function and exchangeable correlation structure. Although each model originally included demographic covariates including age, racial/ethnic minority status, education, and income, none of these outcomes were associated with focal outcomes in any of these models. As such, all models simply adjusted for all person-level characteristics. Then, we explored whether specific motivations for not using condoms were identified more often after CAS events that occurred with lower-risk versus high-risk partners. However, given that this sample focused on MSM who were mostly not in sexually-exclusive relationships, CAS events that occurred with lower-risk partners (e.g., those who were on PrEP, were mutually sexually-exclusive) accounted for a relatively small percentage of all CAS events (27%, N = 80). In logistic models focused only on CAS events, this resulted in large standard error estimates. Given this, we used unadjusted odds ratios to explore whether motivations for condomless sex differed across high-risk versus lower-risk partners. All analyses were conducted in Stata 14.

RESULTS

Participant demographic characteristics are reported in Table 1. Overall response rates in this study were very high. Participants, on average, did not complete a daily diary survey on 1.87% of possible study days and submitted a survey too late (after 5 p.m.) on 4.01% of study days. As such, a total of 5.81% of all possible study days were missing. This provided an overall total of 2,890 person-days of data for analysis. On average, participants provided 29.7 study-days of data (SD = 1.8, Range = 17–32).

TABLE 1.

Demographic Characteristics and Key Variables (N = 100)

Characteristics	Mean (SD) or N (%)
Age (Range: 18 – 54)	27.1 (7.7)
Race
White	76 (76.0)
Black or African American	4 (4.0)
Asian	8 (8.0)
American Indian/Alaska Native	1 (1.0)
Multiracial	6 (6.0)
Chose not to respond	5 (5.0)
Ethnicity (Hispanic or Latino)	16 (16.0)
HIV-status (self-reported)
Negative	83 (83.0)
Don’t know	17 (17.0)
Currently in sexually-exclusive relationship	5 (5.0)
Avg. length of relationship (months)	1.4 (0.9)
College degree	54 (54.0)
Low income1	29 (29.0)
Unemployed	13 (13.0)
Sexual orientation/identity
Gay	81 (81.0)
Bisexual	12 (12.0)
Other	5 (5.0)
Not sure	2 (2.0)
Days since most recent CAS2 (at baseline)	11.7 (9.7)
History of any PrEP use	5 (5.0)
Problem alcohol use (AUDIT3 > 8)	67 (67.0)
Problem drug use (DAST4 > 3)	24 (24.0)

Note.

¹Represents those with a household annual income <$30,000/year.

²CAS = condomless anal sex.

³AUDIT = Alcohol Use Disorders Identification Test.

⁴Drug Abuse Screening Test.

Sex behaviors

Participants reported a total of 620 sex events across the 30-day study period. Eight participants did not report any sex events. Of these events, 384 involved anal sex (64.6% involved insertive, 50.8% involved receptive), for an average 3.84 anal sex events per participant (SD = 3.5). Fourteen participants reported no anal sex. Thirty percent of all anal sex acts involved condom use (28.7% of insertive and 27.6% of receptive anal sex acts), and 22% of participants who reported any anal sex also reported using condoms during these acts every time they occurred. Figure 1 shows the time of day when participants reported that sex events occurred. There were no systematic differences in condom use during anal sex or having CAS with high-risk partners based on the time of day sex occurred.

Figure 1.

Percentage of sex events by time of day they began.

Sex partners

Table 2 shows the percentage of various sex acts that were reported by partner characteristics. Seventy-five percent of all anal sex acts occurred with “high-risk” partners (i.e., non-exclusive, exclusive but unknown HIV status, or not on PrEP), and 66% percent of these acts did not involve condom use (a total of 189 CAS events with high-risk partners). For 26% of CAS acts, participants reported that their partners were on PrEP. Seventy-nine percent of these events occurred within partnerships that were not mutually sexually exclusive, and 21% occurred with sexually exclusive partners that were of unknown HIV status. Thirty-six percent of all CAS events occurred with partners that participants had never had sex with before. Participants reported not asking about HIV testing history or status 61% of the time when they had anal sex with a new partner. Over a third of all sex acts reported occurred with partners that participants had met on the same day they had sex (36.9%), with 21.6% of sex events occurring with partners they had known for 6 months or more. Compared with partners known more than two months, the odds of using a condom during anal sex was 1.7 times higher with partners known only a few days to a few months, and 2.9 times higher among partners that participants reported meeting the same day. Unsurprisingly, participants reported meeting most of their partners online, while meeting through friends, at bars/clubs, and at parties were the next most common methods of meeting partners. Anal sex that occurred with partners that participants met online were 96% less likely to involve CAS with a high-risk partner compared to partners met in other ways, although this difference was not statistically significant (p = .07). Similarly, although partners met at bathhouses or other sex clubs were nearly seven times more likely to involve CAS with high-risk partner, these meeting partners through these venues was especially rare (n = 15) and only a few participants ever met partners this way. This resulted in especially large standard errors and suggested that these results are not likely to be meaningful. There were no other systematic differences in condom use during anal sex or having CAS with high-risk partners based on where participants met their partners. However, participants were 3.1 times more likely to ask new partners about their HIV status and testing history when meeting them online, compared to all other locations/methods. No other differences emerged in asking new partners about HIV testing/status emerged. Finally, participants reported drinking with their sex partners 34.9% of the time when high-risk CAS occurred, versus 23.4% of the time when it did not, but alcohol use with partners was not significantly associated with engaging in high-risk CAS in regression models. Participants reported having used drugs with their partners during 19.6% of high-risk CAS events, versus 13.2% of lower-risk sex events. The most common drugs used during these events were marijuana (84.2% of all co-use events) and stimulants (15.8% of all co-use events, with 7.4% involving methamphetamine, 5.3% involving cocaine, and 2.1% involving prescription stimulants). Neither using marijuana nor stimulant drugs with a sex partner were significantly associated with engaging in high-risk CAS during that event. See Table 3.

TABLE 2.

Percentages of sex acts occurring by partner characteristics

	Any sex (N = 620)	Anal sex1 (N = 384)	CAS2 (N = 269)
Sexually exclusive partner	24.7	25.8	30.5
New partner	45.5	43.0	36.1
Partner HIV-status
Negative	42.9	43.2	40.5
Don’t know	56.9	56.5	59.1
Partner on PrEP	21.2	22.5	26.3
Length known partner
Just met yesterday	36.9	35.4	29.7
A few days	5.3	5.7	5.6
A week or two	7.6	8.1	7.8
2–4 weeks	6.1	6.3	8.6
A month or two	8.1	9.1	8.9
2–6 months	13.6	15.1	17.8
6–12 months	6.5	6.8	6.7
1+ years	16.0	13.5	14.9
Location met partner
Through friends	8.6	9.4	8.2
Party	3.7	5.3	6.7
Porn theater/Video booth	1.6	0.3	0.4
Gym/health club	0.4	0.3	0.0
Online	68.2	69.8	68.6
Bar/club	7.2	7.5	9.0
Bathhouse/sex club	2.6	2.5	3.1
Cruising spot	1.2	0.0	0.0
Other	6.5	5.0	3.9
Alcohol use w/ partner	26.9	28.4	32.0
Drug use w/ partner	15.2	16.7	17.5

Note.

¹Represents percentage of all sex events.

²Represents percentage of all anal sex events.

TABLE 3.

Results of random effects logistic models of key outcomes

	Condom use during anal sex
	OR	SE	p	95% CI
Length known partner1
A few days to a month or two	1.71	0.50	.065	0.09–3.04
Met the same day	2.90	0.75	< .001	1.73–4.82
	Asked about HIV status
Location/method met partner
Through friends	0.60	0.51	.542	0.11–3.14
At a party	0.11	0.17	.145	0.01–2.12
Online	3.09	1.56	.024	1.16–8.26
Bar or club	0.40	0.31	.237	0.09–1.83
Bathhouse or sex club	2.04	2.64	.580	0.16–25.71
	High-risk CAS
Location/method met partner
Through friends	1.52	0.79	.415	0.55–4.23
At a party	1.86	1.40	.413	0.42–8.17
Online	0.51	0.19	.072	0.25–1.06
Bar or club	1.72	1.09	.397	0.49–5.97
Bathhouse or sex club	6.79	7.94	.101	0.69–67.19
Sex motives
My friends will think less of me if I don’t	4.39	3.35	.052	0.99–19.54
To re-assure myself that I am attractive	2.40	0.96	.029	1.10–5.27
To keep my partner content or interested	0.51	0.22	.122	0.22–1.20
To feel emotionally close to my partner	2.09	0.78	.048	1.01–4.35
To cope with feeling upset	0.93	0.72	.929	0.21–4.23
Because it feels good	1.00	0.58	.994	0.32–3.01
Alcohol use w/ partner	1.38	0.36	.218	0.83–2.31
Marijuana use w/ partner	2.15	1.99	.410	0.35–13.23
Stimulant drug use w/ partner	2.89	2.74	.264	0.45–18.64

Motivations for sex and not using condoms

Figure 2 shows the percent of all sex events in which participants reported that a specific motive was involved in their decision to have sex. No motivation was identified for only one reported sex event (0.2%), while participants reported one primary motivation for 67.3% of sex events, two key motivations for 22.1%, and three or more for 10.5%. Although “because it feels good” was identified as one motivation for sex for 94% of all reported sex events, this reason was identified as the sole motivation in 62.9% of sex events. That is, participants reported deciding to have sex for reasons other than “because it feels good” for 37% of sex events. Figure 3 shows odds ratios for sex events that involved high-risk CAS when specific motivations for sex were identified (versus those when it was not). This figure shows that the odds of engaging in CAS with a high-risk partner were 2.1 times higher when participants identified wanting “to feel closer to [their] partner” as a key motivation for having sex than when they did not. The odds of high-risk CAS occurring were 2.4 times higher when participants reported having sex to “re-assure [themselves] that [they were] attractive” versus when they did not. Figure 4 shows the percent of CAS events in which participants identified specific reasons for not using condoms by partner risk level. Unadjusted odds ratios suggested that when participants reported not using a condom because none were available, the odds were 2.9 times higher that a CAS event involved a high-risk partner (χ²[1]=6.4, p = .01).

Figure 2.

Percent of sex events in which specific motivations for sex were endorsed.

Figure 3.

Odds ratios of event-level associations between endorsing specific motivations for sex and high-risk condomless anal sex (CAS).

Figure 4.

Percent of condomless anal sex (CAS) events in which specific motivations for not using condoms were endorsed by partner risk level.

Note. *p < .05.

DISCUSSION

As part of a larger EMA study, 100 high-risk MSM who were not on PrEP used their personal smartphones to complete detailed diary surveys of their sexual behavior each day over a 30-day period. In this study, we focused on describing the sexual behaviors, partner characteristics, and event-level motivations for sex and condom use that MSM reported in these diaries. We also explored whether specific characteristics and motivations were associated with HIV-risk behaviors, including CAS with high-risk partners and not asking partners about their HIV testing history or status. Since diary data were collected each day, this method ensured that the detailed data collected about these events could be reported more accurately than many past studies relying on recall-based methods alone.

Overall, participants provided remarkably complete data over the study period and had response rates that exceeded many other past daily diary studies of similar populations (35, 37, 55). Most participants (86%) also reported at least one anal sex event across the study period, and an overall average of about four such events per participant over the 30 days suggests that this method successfully captured enough events to meaningfully analyze within-person variability. Less than a third of all anal sex events reported in this study involved condom use, a rate that is similar but somewhat lower than has been reported in some nationally-representative studies of MSM in the US (56, 57). Three-quarters of all anal sex events also occurred with partners that would have posed high risk for transmitting HIV or other STIs to the participant, either because they were not mutually sexually exclusive, or if they had agreed to be exclusive, because the participant was still unsure about their partner’s HIV status. Our results suggest that the vast majority of sex events and CAS events occurred with partners that were not sexually exclusive, and that over a third of all sex events occurring in a 30-day period were with new partners. As such, consistent with past studies (58–60), having concurrent sexual partnerships involving many new partners is likely a key source of risk for these men. Similarly, over a third of all sex events occurred with partners that participants met the same day, and participants reported having talked with these partners about their HIV status and testing history before having sex less than a third of the time. These findings suggest that yet another source of risk involves MSM having sex with partners without first having had a conversation about their HIV statuses or the last time they were tested. Although interventions for reducing HIV risk among MSM who are not on PrEP have largely focused on encouraging condom use (61, 62), helping these men initiate discussions with their partners about HIV testing history and commit to having sex with one partner at a time that they know well could provide other options for reducing future risk.

Since most participants in this study were recruited via the internet, it is not surprising that they reported meeting the vast majority of their partners online. However, this rate (68%) is also similar to that reported in more representative samples of MSM (63, 64). Although meta-analyses have suggested that MSM who use the internet to meet partners are more likely to have engaged in sexual risk behaviors (65, 66), participants in our study were no more likely to engage in CAS with partners they met online than with those they met elsewhere. They were also more likely to have asked partners that they met online about their HIV status and/or testing history prior to sex, which could suggest that MSM may feel more comfortable asking partners about these issues via online chats, perhaps before agreeing to meet prospective partners. So, all else equal, an upside of meeting partners online may be that this medium provides a “safer” platform through which MSM can initiate these conversations and agree upon the details of sex without the barriers normally present when meeting someone in person (e.g., less fear of rejection). However, it is important to note that our analyses focused on comparing participants’ prevention behaviors (e.g., condom use, discussing HIV with partners) prior to/during sex across different partner types within the same participant, meaning that we compared what an individual participant did in one situation versus another. There is compelling evidence that, across MSM, those who meet partners online tend to have more sex partners, which may still be a key source of risk for HIV and STIs for these men (67). Next, consistent with past studies (68–71), our results also showed that the odds of engaging in CAS with a high-risk partner were over four times higher when participants reported meeting their partner at bathhouses or other sex clubs. Although this association was not statistically significant, this was most likely due to the low number of partners met at these locations. This finding is consistent with bathhouses as locations that some MSM may seek for high-risk sex (70, 71). Finally, a larger percentage of high-risk CAS events involved participants using alcohol and stimulant drugs specifically with their partners when compared to sex that involved less risk (e.g., any sex, low-risk CAS), a finding that is consistent with a large body of research suggesting that using these drugs is associated with HIV-risk behavior in MSM (25, 26, 72, 73). However, this difference was also not statistically significant, and likely reflects that this model tested whether alcohol/drug use specifically with sex partners played a role in HIV-risk behavior. A more dedicated, thorough study of participants’ alcohol and drug use prior to sex (whether with their sex partners or not) and their effects on decisions about prevention is currently being prepared.

Since participants submitted diary surveys within 24-hours of sex events occurring, we also explored participants’ primary motivations for sex and, if they engaged in CAS, their primary reasons for not using condoms while these thoughts could be more easily and accurately reported. Although past research has shown that specific motivations may be uniquely associated with sexual risk behavior (52, 74, 75), nearly all of these studies have used recall-based survey methods that require participants to both remember what their mindset was (sometimes weeks or months in the past) and aggregate “how often” they had sex for that reason over a given period of time. This approach can result in considerable bias (31), and does not recognize that these motivations likely vary considerably across situations. In this study, participants noted a pleasure motivation for sex for nearly all reported sex events, but also endorsed at least one other motivation for over a third of sex events. Among the most common of these additional motives were intimacy (“to feel close to [my] partner,” endorsed on 20% of sex events), self-affirmation (“to re-assure [yourself] that [you] are attractive,” 14%) and partner satisfaction (“to keep [my] partner content,” 12%). Our results also showed that the odds of engaging in CAS with a high-risk partner were about two times higher when participants reported being motivated to have sex by intimacy and self-affirmation motives. These findings are consistent with past studies showing that MSM who endorse intimacy motives are less likely to use condoms during anal sex (34, 76). However, they contrast with the only event-level study of sex motives in MSM we are aware of, which found that participants were less likely to engage in CAS when they endorsed self-affirmation motives (34). Together, these results suggest that risk reduction interventions should specifically acknowledge these additional motivations for sex as potential risk factors, and help recipients develop the skills to apply strategies to reduce their risk specifically when they feel motivated to be close to their partner or to affirm their own attractiveness.

Finally, when participants reported CAS, we also asked them to report their primary reasons for declining to use condoms. Helping MSM understand and challenge the motivations they have for not using prevention strategies plays a key role in many of the most effective available interventions for reducing HIV risk behavior. One especially relevant example is Personalized Cognitive Counseling, which refers to these motivations as “self-justifications” that prevent MSM from realistically assessing the risk involved (77, 78). Using an abbreviated list of some of the most common self-justifications and other reasons MSM give for not using condoms (79–81), we asked participants to report which were most important in their decision not to use a condom when they engaged in CAS. The most common reasons reported involved believing their partner was low-risk, that condomless sex feels better, and that they did not want to ruin the moment. In general, most of these reasons were endorsed just as frequently with high-risk partners as they were with lower-risk ones, suggesting (like many past studies have (e.g., ^{82, 83, 84})) that challenging risk perceptions and teaching quick and delicate ways of suggesting condom use with partners could help MSM avert some risk. However, one important exception was that not having condoms available was endorsed much more often when participants had sex with high-risk partners. Indeed, it was the third most common reason reported for not using a condom during sex with these partners. This pattern of findings could suggest that MSM may often be unprepared for sex with higher-risk partners, but may persist in having anal sex anyway. Interventions that increase the accessibility of condoms and/or remind MSM to carry condoms when they are likely to have access (so-called “just-in-time” interventions) could help increase their use specifically with high-risk partners.

Limitations

Although this study has many strengths, several limitations should also be noted. First, this sample was primarily composed primarily of younger and higher-risk MSM who were recruited online. As such, the findings reported here may be unique to this specific subset of MSM. Similarly, most participants who completed this study lived in the northeastern US, so similar research conducted with MSM in other locations may also yield different results. Next, although we attempted to collect data about sex events and partners that was as detailed as possible, questions and categories intended to characterize the risk involved in sex with particular partners will inevitably be imperfect. However, we believe that our approach appropriately characterized many high-risk events and afforded meaningful comparisons with lower-risk sex. Finally, though our assessment approach was also specifically designed to capture participants’ thoughts and motivations about sex as soon as possible afterward to minimize bias, assessing concepts like these may only be possible in the moment participants make their choices, rather than after the event has happened. That is, the motivations or reasons for sex and condom use MSM report in the lead up to these behaviors may be different than those they report afterward, because motivations assessed afterward may more aptly reflect “rationalizations” or “self-justifications” of their choices after the fact. As such, assessing these motivations around the time when MSM are making decisions about sex but before these events occur may provide a more accurate picture of the role they play in behavior.

In summary, this study used daily diary methods over a 30-day period to assess detailed characteristics of sex events, sex partners, and key motivations for decisions about sex and condom use in a sample of high-risk MSM in the northeastern US. Our results showed that CAS with non-sexually-exclusive partners was common, and that MSM asked a minority of their partners about their HIV status or testing history prior to sex. MSM also met the majority of their partners online, with a sizable minority of sex events occurring on the same day that participants met their partners. Although CAS was no more likely with partners met online than those met in other ways, MSM were more likely to ask partners they met online about their HIV status and testing history prior to sex. Participants also frequently identified motives for having sex other than pleasure alone, and the odds of engaging in CAS with a high-risk partner was significantly higher when they reported having sex to boost intimacy with their partner and to re-assure themselves of their attractiveness. Finally, other than low risk perceptions, beliefs that condoms dampen the pleasure involved in sex and a desire to avoid ruining the moment were among the most common reasons given for not using condoms with all partners, but not having condoms readily available was identified as a reason much more commonly when MSM had sex with high-risk partners. Together, these findings suggest that interventions should incorporate strategies that help MSM be safer specifically when meeting partners online and when engaging in sex for intimacy or re-assurance. They also suggest that interventions intended to remind participants to carry condoms at opportune moments may help reduce some HIV risk behavior.