The Annual American Men’s Internet Survey of Behaviors of Men Who Have Sex With Men in the United States: 2015 Key Indicators Report

Introduction

The American Men’s Internet Survey (AMIS) is an annual online behavioral survey of men who have sex with men (MSM) who live in the United States. The methods have been previously published [1,2]. This supplemental report updates that previous manuscript with the most current data available from AMIS (AMIS-2015). Methods in AMIS-2015 are unchanged from the previously published manuscript unless otherwise noted.

Methods

Recruitment and Enrollment

As in the prior year, AMIS participants were recruited through convenience sampling from a variety of websites using banner advertisements or email blasts to website members (hereafter referred to generically as “ads”). The survey was not incentivized. Data on the number of clicks on all banner ads were obtained directly from the websites. In AMIS-2014, data on the number of clicks on geospatial social networking banner ads were instead obtained by counting the number of clicks on the survey landing page. Men who clicked on the ads were taken directly to the survey website hosted on a secure server administered by SurveyGizmo (Boulder, CO, USA). Participants were also recruited by emailing participants from the previous cycle of AMIS (AMIS-2014) who consented to be recontacted for future studies. To be eligible for the survey, participants had to be 15 years of age or older, consider themselves to be male, and report that they had oral or anal sex with a man at least once in the past (hereafter referred to as MSM). Persons who reported being younger than 15 years of age or refused to provide their age were not asked any other screening questions. Those MSM who met the eligibility criteria and consented to participate in the study started the online survey immediately. The full questionnaire for AMIS-2015 is presented in Multimedia Appendix 1.

AMIS-2015 ran from September 2015 through April 2016, and resulted in 137,608 persons clicking on the ads and landing on the study’s recruitment page (Table 1). Most persons who clicked on the ads were from general social networking websites (66,500/137,608, 48.33%). Of the 1248 participants who completed the AMIS-2014 survey and were emailed links to the AMIS-2015 survey, 9.13% (114/1248) clicked on the link. One-third (33.58%, 46,207/137,608) of those who landed on the study’s page started the screening process and 56.09% (25,919/46,207) of those were eligible. The most common reason for ineligibility was not ever having male-male sex. More than three-quarters (78.52%, 20,351/25,919) of those who were eligible consented to participate in the survey. There were 2291 of 20,351 (11.26%) surveys determined to likely be from duplicate participants. Deduplication of survey responses was performed in the same manner as in previous AMIS cycles [1,2]. Among unduplicated surveys, almost two-thirds (64.21%, 11,597/18,060) were considered successful (ie, observations with no missing values for the first question of at least two consecutive sections). Most successful surveys were among men who reported having sex with another man in the past 12 months (89.07%, 10,330/11,597). Finally, 1.09% (113/10,330) of the sample was found to have provided an invalid ZIP code and was excluded from the final analytical sample.

Almost all these surveys (10,217/10,330, 98.91%) provided a valid US ZIP code. ZIP codes provided by participants were validated by merging them with the 2015 ZIP code-to-county crosswalk files created by the US Department of Housing and Urban Development [3]. Any ZIP codes that could not be matched to this list were then hand-validated by checking against the ZIP code locator tool on the United States Postal Service website [4]. ZIP codes that could not be found were classified as invalid. Overall, the completion rate was 7.4% (10,217/137,608), with an analytical sample consisting of 10,217 surveys out of 137,068 clicks.

Measures and Analyses

For AMIS-2015 analyses, participants were categorized as either being AMIS-2014 participants who took the survey again or new participants from website/app types based on target audience and purpose: gay social networking (n=1), gay general interest (n=2), general social networking (n=4), and geospatial social networking (n=2). Recruitment outcomes and demographic characteristics for the AMIS-2014 participants are presented and for all behavioral outcomes, they are recategorized according to their original recruitment source. We do not provide the names of the websites/apps to preserve operator and client privacy, particularly where a category has only one operator. The participants who were eligible, consented, unduplicated, successful, reported male-male sex in the past 12 months, and provided a valid US ZIP code were included in analyses of participant characteristics and behavior.

For AMIS-2015, we created a more refined population density variable for each participant’s county of residence as determined by their ZIP code. The levels of the population density variable are from the National Center for Health Statistics (NCHS) Rural-Urban classification scheme [5]. The NCHS classifies counties into six categories: central (ie, inner city) or fringe (ie, suburban) areas of large metropolitan statistical areas (MSAs; population size ≥1,000,000), medium-sized MSAs (population size 250,000-999,999), small MSAs (population size <250,000), micropolitan area (counties that contain all or part of a city of 10,000 or more), and noncore (counties that do not contain any part of a city of 10,000 or more). We further collapsed these categories into a four-level urbanicity variable: urban (central), suburban (fringe), medium/small metropolitan, and rural (micropolitan and noncore).

The analysis methods for AMIS-2015 did not substantively differ from those previously published but are repeated in this report for clarity [1]. Overall, chi-square tests were used to identify whether participant characteristics significantly differed between recruitment sources. Multivariable logistic regression modeling was used to determine significant differences in behaviors based on self-reported human immunodeficiency virus (HIV) status while controlling for race/ethnicity, age group, National HIV Behavioral System (NHBS) city residency, and recruitment website type. MSAs included in NHBS in 2015 were Atlanta, GA; Baltimore, MD; Boston, MA; Chicago, IL; Dallas, TX; Denver, CO; Detroit, MI; Houston, TX; Los Angeles, CA; Miami, FL; Nassau-Suffolk, NY; New Orleans, LA, New York City, NY; Newark, NJ; Philadelphia, PA; San Diego, CA; San Francisco, CA; San Juan, PR; Seattle, WA; and Washington, DC. Self-reported HIV status was categorized as either HIV-positive or HIV-negative/unknown status, consistent with surveillance reports produced by NHBS [6]. HIV testing behaviors were only examined among those who did not report that they were HIV-positive and were also presented by participant characteristics. Multivariable logistic regression results are presented as Wald chi-square P values to denote an independently significant difference in the behavior for each subgroup compared to a referent group. Statistical significance was determined at P<.05.

Results

Approximately seven in 10 (7291/10,217, 71.36%) participants included in this report were white and non-Hispanic, less than half were 40 years of age or older (4326/10,217, 42.34%), and their most common region of residence was the South followed by the West (Table 2). Participants were recruited from all US states and there were at least 100 participants from each of 28 states (Figure 1). Overall, 9.35% (955/10,217) of participants reported being HIV-positive, 69.91% (7143/10,217) reported being HIV-negative, and 20.74% (2119/10,217) reported having an unknown HIV status. All participant characteristics differed significantly based on where they were recruited (Table 2).

Most participants reported having anal sex without a condom with another man within the past 12 months (Table 3). Compared to HIV-negative/unknown status participants, those who were HIV-positive were significantly more likely to report anal intercourse without a condom (adjusted OR [AOR] 1.86, 95% CI 1.59-2.18), including with male partners who were of discordant or unknown status (AOR 2.75, 95% CI 2.36-3.20). Within each serostatus group, anal intercourse without a condom differed significantly by age group (HIV-positive and HIV-negative/unknown status participants), and recruitment website (HIV-negative/unknown status participants only). Anal intercourse without a condom with partners of discordant or unknown HIV status differed significantly by race/ethnicity (HIV-positive participants only), recruitment website (HIV-positive participants only), and age (HIV-negative/unknown status participants only).

Almost one-quarter (235/955, 24.6%) of HIV-positive participants reported using marijuana in the past 12 months (Table 4). Compared to HIV-negative/unknown status participants, HIV-positive participants were significantly more likely to report use of marijuana (AOR 1.43, 95% CI 1.22-1.69) and other illicit substances in the past 12 months (AOR 2.20, 95% CI 1.88-2.59). Within each serostatus group, use of marijuana and other illicit substances differed significantly by age (HIV-positive and HIV-negative/unknown status participants), residence in an NHBS city (HIV-negative/unknown status participants only), and recruitment website type (HIV-negative/unknown status participants only). Marijuana use also differed significantly by recruitment website among HIV-positive participants. Use of other illicit substances differed significantly by race/ethnicity among HIV-negative/unknown status participants.

HIV testing behaviors were examined among those who did not report being HIV-positive (Table 5). Most participants (7327/9262, 79.11%) reported having been previously tested for HIV infection, and just over half (5158/9262, 55.69%) reported being tested in the past 12 months. HIV testing behavior, both ever tested and tested in past 12 months, differed significantly by age, residence in an NHBS city, and recruitment website type.

Compared to HIV-negative/unknown status participants, HIV-positive participants were significantly more likely to report sexually transmitted infection (STI) testing (AOR 4.00, 95% CI 3.43-4.68) and STI diagnosis (AOR 3.83, 95% CI 3.20-4.59) in the past 12 months (Table 6). The most common STI diagnosis among HIV-positive participants was syphilis (144/955, 15.1%), whereas gonorrhea was the most common STI diagnosis among HIV-negative/unknown status participants (427/9262, 4.61%). Among HIV-negative/unknown status participants, STI testing differed significantly by race/ethnicity, age, and residence in an NHBS city. Among both HIV-positive and HIV-negative/unknown status participants, STI testing differed significantly by recruitment website type and STI diagnosis differed significantly by age, NHBS city residence, and recruitment website type.

‎View this figure