Online and Offline Sexual Health-Seeking Patterns of HIV-Negative Men who have Sex with Men

Abstract

To inform health information targeting, we used cross-sectional data from 2577 HIV-negative MSM to identify groups of men who access similar sources. Offline, more men reported talking to a physician about HIV than about having sex with men; fewer than half attended a safer sex workshop. Online, men sought information primarily through Internet search engines, GLBT websites, or health websites. A latent class analysis identified four groups of health seekers: minimal health seekers, those who accessed online sources only, those who sought information mostly from health professionals, and those who sought information from diverse sources. Minimal health seekers, 9% of the sample, were the group of greatest concern. They engaged in unprotected anal sex with multiple partners but infrequently testing for HIV or sought sexual health information. By encouraging health seeking from diverse sources, opportunities exist to increase men’s knowledge of HIV/STI prevention and, when necessary, access to medical care.

INTRODUCTION

Since the first cases of HIV/AIDS in 1981, the sexual health of men who have sex with men (MSM), and in particular the sexual health of men who identify as gay or bisexual, has become a public health concern. Much of the research has focused on interconnected health disparities present in this population, including higher rates of HIV and sexually transmitted infections (STIs),^1–3 mental health concerns,^4–6 and substance use.^7–10 While researchers have studied individual mechanisms by which MSM seek out sexual health information, missing from the current body of knowledge is an understanding of how men simultaneously access both offline and online sources of information. A more nuanced understanding of how men access this information can inform referral processes and intervention development.

Throughout the HIV literature, early research referred to members of the study population as gay and bisexual men and later as MSM. Although the former term implies identity and the latter behavior, the terms are often used synonymously. In the discussion that follows, we use the term gay or bisexual men for studies that explicitly used the term; otherwise, we deferred to the behavioral description, MSM.

In the early 1990s, research examined the sexual health-seeking practices of gay and bisexual men with varying HIV-related diagnoses, finding that gay-identified men living with HIV or testing for the virus were more likely to seek out information and support than HIV-negative, non-testing, or bisexual men.^11–16 More recent research, focusing on the influence of internalized homonegativity on disclosure of gay and bisexual men’s sexual orientation or on their overall health, suggests that approximately 80% of men are open about their sexuality to their healthcare provider.^{4, 17, 18} However, depending on the study, only 15%–30% of men are comfortable discussing their same-gender sexual behaviors and HIV/STI risk-reduction strategies with the provider.^{4, 17, 19, 20} Further, the majority of gay and bisexual men report no change in the specificity of care or dialogue about MSM-specific health issues after disclosure.^21,22

Community-based organizations (CBOs) and AIDS-service organizations (ASOs) have formed, among other reasons, to prevent the spread of HIV/STIs among gay and bisexual men. One study found that 30% of gay men had talked to an employee of a community-based organization about HIV/STI prevention.¹⁹ The most difficult men to reach with prevention messages or to engage in care include those who are young, identify as bisexual, in a presumed monogamous sexual relationship, using or involved with someone using intravenous drugs, HIV-negative, unaware of their HIV-status, or members of a racial/ethnic minority.^23–27 In response, many CBOs/ASOs and healthcare providers with gay and bisexual male clients have attempted to remove structural barriers to accessing care^28–30 by strengthening the provider-patient relationship,^{31, 32} and expanding beyond traditional individual-, group-, and community-level interventions to using informal social networks^{21, 33} and web-based communication, ^34–37 including instant messaging.^38–40

MSM who do not seek help from a community-based organization or healthcare provider might get information from the Internet. Internet-using MSM report seeking information online about how to be a better lover, men’s physical health, and relationships.³⁶ HIV-positive MSM report using the Internet to learn about highly active antiretroviral therapy⁴¹ and clinical trials.⁴² HIV-positive online health seekers tend to have a higher income, be more educated, have no history of injection drug use, have a CD4 counts greater than 200 cells/mm³, and have higher adherence to antiretroviral medication regimens than persons with HIV who do not seek sexual health information online.^43–45

To date, much of the health-seeking research has focused on the usefulness of single-source health information rather than the concurrent use of diverse sources. In addition, much of the research has focused on HIV-positive MSM, whose motivation for accessing health information is likely to be different than that of HIV-negative MSM. Given these gaps in understanding, this study had two aims. First, we sought to identify how HIV-negative MSM access health information from diverse sources; second, we identified predictors of health-seeking behaviors.

METHODS

Study Design

Internet-using MSM (n=2716) completed an online survey about their online and offline sexual behavior. Participants were recruited during three months in 2005 through banner advertisements placed on the most highly subscribed gay website in the US. Eligibility criteria included being male, 18 years of age or older, a resident of the US, and acknowledging having had sex with another man at least once during their lifetime. Men of color were deliberately over-sampled to provide approximately equivalent groups of Asian, Latino, Black, and White men for analyses.

Study procedures are described in detail elsewhere.^{46, 47} Briefly, by clicking on a study banner advertisement, prospective participants were transported to the study website. After completing a screening and consent process, participants answered questions about their sexual health. A refuse to answer option was provided for each question. The mean survey completion time was 45 minutes. Participants were initially compensated $10, which in the third month was raised to $20 in order to speed recruitment. The institutional review board of the researchers’ home institution approved study procedures.

Measures

Sexual health-seeking strategies

For this study, sexual health-seeking was defined as seeking information about any of the following: HIV, STIs, having sex with men, or sexual health. We conceptualized the outcome as a latent variable measured by twelve indicator questions. Three of the questions measured the proportion of participants who sought information off-line: talking to a health professional about HIV, talking to a health professional about having sex with men, and attending an HIV-prevention workshop. Response options were within the last 12 months, ever but not including that last 12 months, never, or refuse to answer. Nine questions asked if participants had accessed sexual health information from one of the following online source within the last twelve months: an online HIV expert (web-based health advice such as “Ask Dr. K”), Internet search engines, (e.g., Google, Yahoo!, AOL), bulletin boards or blogs, health websites (e.g., WebMD, Yahoo! Health, Ask Dr. K), gay, lesbian, bisexual, and transgender websites (e.g., Gay.com, PlanetOut.com), federal government websites (e.g., National Institutes of Health (NIH), Centers for Disease Control and Prevention (CDC)), national radio and television websites, (e.g., New York Times, Washington Post, Public Radio), and health insurance websites (e.g., Blue Cross and Blue Shield, Health Partners). Response options for the online questions were yes, do not know, no, or refuse to answer. Responses to the twelve sexual health-seeking questions were dichotomized into never or at least once in their lifetime.

Other measures

Demographic measures included age and education – both measured as continuous variables – and a series of categorical variables: race/ethnicity, city size, and region of the USA. Behavioral variables included gender of sexual partners in the last three years, ever testing for HIV, ever being diagnosed with an STI, having two or more UAI partners in the last 12 months, and degree of openness as a gay, bisexual, or a man attracted to other men (outness). We decided a priori to treat each of these variables as predictors and as covariates of sexual health-seeking in an explanatory multinomial logistic regression model.

Statistical Methods

Prior to the regression analysis, we examined the distribution of each of the health-seeking items using SAS, version 9.1.⁴⁸ We estimated iterative latent class models using Mplus, version 5.1.⁴⁹ We chose the latent class approach because of its ability to estimate a number of classes that produce optimal fit to the observed data by maximizing between-class heterogeneity and within-class homogeneity.⁵⁰ We excluded HIV-positive men from this analysis because the limited number of participants (n=119) was not conducive to a latent class analysis. Of the 2,578 HIV-negative men, 2,577 were used for the latent class analysis and 2,491 were used to calculate adjusted prevalence odds ratios; excluded men had missing data on the variables of interest. For each model, we used the Akaike information criterion (AIC),⁵¹ the sample-size-adjusted Bayesian information criterion (SABIC),⁵² entropy, the Lo-Mendell-Rubin adjusted likelihood ratio test (LMR-LRT),⁵³ and the parametric bootstrap likelihood ratio test (BS-LRT) to test the fit of the n-class model to the n-1-class model. A better fitting model was defined as one with a lower AIC and SABIC, a higher entropy (range-0–1, where 1=perfect classification), and a significant (p<0.05) LMR-LRT and BS-LRT.

RESULTS

Demographic characteristics of the sample are summarized in Table 1. Briefly, 63.0% of the sample was under age thirty, 72.8% were men of color, and 86.4% had one or more years of college education. About half of the sample reported living in a downtown or suburban area (55.3%), and more participants came from the south (32.5%) than from each of the three other census regions of the US. Most participants reported having had sex with only men in the last 3 years (83.0%), had been tested for HIV at least once (81.8%), had never been diagnosed with an STI (59.3%), and had not engaged in unprotected anal intercourse with two or more partners in the past 12 months (69.0%). Over half (61.9%) were open about their sexuality to most or all of the people they knew.

Table 1.

Demographic Characteristics of Participants (N=2,578)

	n (%)
Age (in years)
18–24	959 (37.2)
25–29	664 (25.8)
30–39	670 (26.0)
40+	282 (11.0)
Race/Ethnicity
Latino American	638 (24.8)
Asian American	496 (19.2)
Black or African American	408 (15.8)
White	701 (27.2)
Other	335 (13.0)
Years of Education
12 or less	351 (13.6)
13–16 years	1556 (60.4)
17 or more	671 (26.0)
City Size
Rural	127 (5.0)
Small town	366 (14.3)
Medium-sized city	656 (25.6)
Suburb of large city	621 (24.2)
Downtown/Central District of large city	798 (31.1)
Region of USA
Northeast	385 (15.1)
Midwest	605 (23.7)
South	829 (32.5)
West	735 (28.8)
Sexual Partners in Last 3 Years
Only Men	2129 (83.0)
Men and Women	436 (17.0)
Ever Tested for HIV
Yes	2098 (81.8)
No	468 (18.2)
Ever Diagnosed with an STI
Yes	1044 (40.67)
No	1523 (59.3)
2+ UAI Partners in Last 12 months
Yes	796 (31.0)
No	1769 (69.0)
Outness
Not out	219 (8.5)
Out to a few people they know	469 (18.2)
Out to about half the people they know	291 (11.3)
Out to most people they know	650 (25.3)
Out to almost all/all people they know	942 (36.6)

Nearly all (97.1%) of the men used at least one of the twelve sources of health information at least once to learn about HIV, STIs, having sex with men, or their sexual health. As shown in Table 2, the most endorsed offline-sources were talking to a health professional about HIV (80.2%) and talking to a health professional about having sex with men (62.1%). Fewer than half (46.6%) had attended a safer sex workshop or other HIV-prevention program. The most endorsed online sources were using a search engine (76.8%), a GLBT website (67.9%), and a health website (55.9%). Few men endorsed seeking information from a health insurance website (12.7%) and emailing a health provider (9.0%), so we dropped these items from the latent class analysis.

Table 2.

Frequency of Sources Participants Ever Used to Seek out Information about HIV, STIs, Having Sex with Men, and Sexual Health (n=2,578)

		Yes n (%)	No n (%)
Off-line Sources	Talked with a doctor or health professional about HIV	2065 (80.2)	509 (19.77)
	Talked with a doctor or health professional about having sex with men	1598 (62.1)	977 (37.9)
	Attended a safer sex workshop or other HIV-prevention program	1199 (46.6)	1377 (53.5)
On-line Sources	Sought out information using a search engine	1965 (76.8)	594 (23.2)
	Sought out information using a GLBT website	1740 (67.9)	821 (32.1)
	Sought out information using a health website	1432 (55.9)	1128 (44.1)
	Sought out information using a bulletin board	771 (30.1)	1789 (69.9)
	Sought advice from an online HIV expert	728 (28.3)	1846 (71.7)
	Sought out information using a federal government website	728 (28.4)	1833 (71.6)
	Sought out information using a national radio and television website	528 (20.6)	2032 (79.4)
	Sought out information using a health insurance website	324 (12.7)	2235 (87.3)
	Emailed a health provider	229 (9.0)	2330 (91.1)

Measures of sexual health-seeking listed in Table 2 (with the exception of the least frequent: seeking out information on a health insurance website and emailing a health provider) were used as indicators of the latent variable. Estimation of the latent class model resulted in a four-class solution (Table 3). The addition of the covariates strengthened the fit of the four-class solution and did not alter the classification quality (AIC=25529.38; SABIC=25809.56; entropy=0.80).

Table 3.

Latent Class Analysis for the Health-Seeking Variable (N=2577)

Model	Log Likelihood	df	SCF	AIC	SABIC	Entropy	LMR-LRT	df	P Value	BS-LRT	df	P Value
1 class	−15449.09	10	1.00	30918.18	30944.95	--	--	--	--	--	--	--
2 classes	−14187.95	21	1.11	28417.90	28474.12	0.71	2493.42	11	0.00	2522.28	11	0.00
3 classes	−13820.98	32	1.15	27705.96	27791.63	0.76	725.54	11	0.00	733.94	11	0.00
4 classes	−13595.76	43	1.11	27277.52	27392.63	0.76	445.29	11	0.00	450.44	11	0.00
5 classes	−13453.23	54	1.36	27014.46	27159.02	0.76	281.80	11	0.35	285.06	11	0.00

Note: Bolding indicates preferred class solution. df=degrees of freedom, SCF=scaling correction factor, AIC=Akaike information criterion, SABIC=sample-size adjusted Bayesian information criterion, LMR-LRT=Lo-Mendell-Rubin adjusted likelihood ratio test, and BS-LRT=parametric bootstrap likelihood ratio test.

The distribution of the endorsements by class for the ten-item sexual health-seeking latent variable is listed in Table 4. Men in the minimal health-seeking class (8.9% of the total sample) did not endorse any of the items used as indicators of sexual health-seeking at or above 50%. The online only class (16.9%) showed strong endorsement for using search engines, health websites, and GLBT websites. Men in the mostly health professional class (33.1%) endorsed talking to a doctor or health professional about HIV, talking to a doctor or health professional about having sex with men, and using Internet search engines. The diverse sources class (41.1%) endorsed seeking information from all three off-line sources, Internet search engines, bulletin boards, health websites, GLBT websites, and federal government websites.

Table 4.

Percentage Endorsement of Sources for Sexual Health Information by Latent Class (N=2,491)

		Minimal Health Seeking (n=222)	Online Only (n=421)	Mostly Health Professional (n=824)	Diverse Health Seeking (n=1024)
Off-line Sources	Talked with a doctor or health professional about HIV	10.0	37.0	99.0	100.0
	Talked with a doctor or health professional about having sex with men	4.0	7.0	75.0	89.0
	Attended a safer sex workshop or other HIV-prevention program	16.0	26.0	45.0	63.0
On-Line Sources	Sought advice from an online HIV expert	5.0	24.0	17.0	45.0
	Sought out information using a search engine	38.0	97.0	52.0	98.0
	Sought out information using a bulletin board	6.0	39.0	8.0	51.0
	Sought out information using a health website	5.0	72.0	21.0	90.0
	Sought out information using a GLBT website	33.0	88.0	44.0	89.0
	Sought out information using a federal government website	1.0	30.0	6.0	53.0
	Sought out information using a national radio and television website	0.0	23.0	4.0	4.0

Note: Bolding indicates that 50% or more of the participants in the class endorsed the item.

The diverse sources class was used as the reference group for the between class multivariate logistic regression because it contained the largest number of men in it. There were no statistically significant differences between groups in terms of age, race/ethnicity, city size, or region of the USA. Men with less education however, had lower odds of using diverse sources (Table 5).

Table 5.

Odds of Being in a Sexual Health-Seeking Group Based on a Multinomial Logistic Regression (N=2,491)

	PORAdj. (95% CI)
	Minimal Health Seeking (n=222)	Online Only (n=421)	Mostly Health Professional (n=824)	Diverse Sources (n=1024)
Age (Continuous)	1.01 (0.98, 1.05)	1.00 (0.97, 1.02)	1.04 (1.02, 1.06)	Ref.
Years of Education (Continuous)	0.86 (0.79, 0.93)	0.91 (0.85, 0.97)	0.91 (0.87, 0.95)	Ref.
Race/Ethnicity
Latino	1.24 (0.63, 2.45)	0.89 (0.55, 1.44)	0.9 (0.65, 1.26)	Ref.
Other	1.06 (0.45, 2.46)	0.78 (0.45, 1.37)	0.95 (0.64, 1.41)	Ref.
Asian	1.57 (0.79, 3.11)	0.88 (0.53, 1,47)	0.97 (0.67, 1.40)	Ref.
Black	1.58 (0.77, 3.23)	0.68 (0.38, 1.23)	1.23 (0.85, 1.79)	Ref.
White	Ref.	Ref.	Ref.	Ref.
City Size
Rural	2.58 (1.00, 6.64)	1.19 (0.57, 2.48)	0.92 (0.53, 1.60)	Ref.
Small	1.32 (0.65, 2.65)	1.63 (0.94, 2.80 )	1.32 (0.92, 1.90)	Ref.
Medium	1.59 (0.85, 2.95)	1.06 (0.65, 1.71)	1.02 (0.75, 1.38)	Ref.
Suburb	0.84 (0.43, 1.61)	1.21 (0.76, 1.91)	1.01 (0.75, 1.37)	Ref.
Downtown	Ref.	Ref.	Ref.	Ref.
Region of the USA
Northeast	0.58 (0.29, 1.14)	0.76 (0.43, 1.32)	1.20 (0.84, 1.72)	Ref.
South	0.89 (0.50, 1.58)	1.26 (0.80, 1.98)	0.83 (0.60, 1.13)	Ref.
West	0.77 (0.42, 1.42)	1.08 (0.66, 1.78)	1.15 (0.83, 1.58)	Ref.
Midwest	Ref.	Ref.	Ref.	Ref.
Sexual Partners Last 3 Years
Only Men	1.86 (1.03, 3.33)	1.06 (0.68, 1.65)	1.07 (0.77, 1.50)	Ref.
Men and Women	Ref.	Ref.	Ref.
Ever Tested for HIV
Yes	0.02 (0.01, 0.03)	0.05 (0.03, 0.08)	1.20 (0.61, 2.39)	Ref.
No	Ref.	Ref.	Ref.	Ref.
Ever Diagnosed with an STI
Yes	0.17 (0.10, 0.30)	0.38 (0.26, 0.56)	0.59 (0.46, 0.74)	Ref.
No	Ref.	Ref.	Ref.	Ref.
2+ UAI Partners in Last 12 Months
Yes	1.12 (0.68, 1.84)	0.63 (0.42, 0.95)	1.20 (0.95, 1.53)	Ref.
No	Ref.	Ref.	Ref.	Ref.
Outness
Out to almost all or most people I know	0.07 (0.03, 0.15)	0.1 (0.05, 0.21)	0.72 (0.38, 1.39)	Ref.
Out to most people I know	0.12 (0.05, 0.27)	0.19 (0.09, 0.39)	1.02 (0.54, 1.96)	Ref.
Out to about half the people I know	0.2 (0.08, 0.48)	0.17 (0.08, 0.39)	1.02 (0.51, 2.03)	Ref.
Out to a few people I know	0.47 (0.22, 1.01)	0.38 (0.18, 0.79)	1.02 (0.53, 1.96)	Ref.
Not out	Ref.	Ref.	Ref.	Ref.

Note: Bolding indicates significance at p<=0.05.

As compared to health seekers using diverse sources, those using minimal sources were less educated and had a greater odds of having only male sexual partners and a lower odds of testing for HIV, ever being diagnosed with an STI, and being open (out) about their sexuality. The online only group was similar, except that the gender of their sexual partners in the last three years did not differ from those using diverse sources. The online only group had lower odds of reporting two or more UAI partners in the last 12 months than the other groups. Those relying mostly on health professionals were less educated and had lower odds of being diagnosed with an STI than the diverse sources group.

DISCUSSION

This study described four classes of sexual health-seeking behavior by HIV-negative MSM. MSM who used diverse sources to seek sexual health information tended to be open about their sexuality, to engage in UAI with multiple partners, to have had an STI diagnosis, and to have tested for HIV regularly. This group appeared to contain high-risk men who followed recommendations about HIV testing and were actively seeking information both online and offline. The second group, men who primarily sought information from health professionals, also used a search engine. They appeared just as open about their sexuality as those in the diverse sources group, reported similar number of UAI partners, and were testing for HIV at a similar rate. However, they appeared less motivated to seek out sexual health information from diverse sources and were less likely to report ever having an STI diagnosis. We see this group as at the same high risk as the diverse sources group but less vigilant in their health seeking, possibly because they have had fewer consequences such as an STI or have not attended a safer sex workshop or other HIV-prevention program. The online only group was less open about their sexuality, was less likely to have engaged in UAI, and was less likely to have had an HIV test or to have received an STI diagnoses than the diverse sources group. This was consistent with them being at lower risk and hence, less active in engaging in sexual health-seeking. The minimal health-seekers were the group of greatest concern to public health. Representing 9% of this sample, they appeared to be engaging in UAI with multiple partners as frequently as the other high-risk groups. However, they were not being tested for HIV and not engaged in health seeking. Since these men appear less open about their sexuality, reaching them is an identified challenge. By being less out, these men are less likely to receive the HIV prevention messages targeting the gay and bisexual male community and less likely to be contacted by traditional CBO/ASO outreach efforts, e.g., outreach at bars and public sex environments.

The only non-sexually related demographic difference between classes of HIV-negative sexual health-seeking MSM was years of education, confirming previous research that educated MSM are more likely to seek health information.^{45, 54} While previous research suggests that men of color are less likely to seek health information,^23–26 we did not detect a statistically significant difference nor did we detect a statistically significant difference based on city size. However, the POR for those living in a rural community bordered on significance, suggesting that a higher proportion of men in rural communities might be minimal health-seekers.

Whereas previous research found that bisexual men are less likely to seek out sexual health information than gay-identified men,^{13, 25} we found the opposite to be true when focusing on behavior over the past three years rather than identity. In our study, men who reported only having sex with men had greater odds of being in the minimal health-seeking group than behaviorally bisexual men.

Consistent with previous research,^{4, 18} our findings suggest that men who are not as open about their sexuality are less likely to seek sexual health information about HIV, STIs, and having sex with men. We found that those men who were not as open were less likely to talk to a health professional or to use diverse sources. Instead, they were either minimal health-seekers or relied on online sources only, perhaps because of the anonymity afforded by the medium.

The usual limitations of cross-sectional research apply to this study, particularly, temporal associations. Since this sample was recruited from a single website, we cannot know if these results generalize to all MSM. We also recognize that MSM’s Internet usage and health-seeking patterns might have shifted since we collected this data. Despite these limitations, this study adds to our knowledge of MSM’s sexual health-seeking behavior by identifying four distinct groups of HIV-negative sexual health-seekers and by identifying predictors of sexual health-seeking behavior.

This study has implications for physicians, researchers, and educators. Most of the offline sexual health information was received from a physician and not from a safer sex workshop or HIV-prevention program. Though 80.2% reported talking to a physician about HIV, only 62.1% talked about having sex with men, suggesting a need for providers to ask explicitly about the gender of men’s sexual partners. Since nearly twice as many men seek information from a physician than from a safer sex workshop or HIV-prevention program – an expected finding since individuals see their physician when ill or when needing annual exams – physicians with limited time to counsel patients on their sexual risk should consider referring patients to these workshops and programs. In our analysis, those seeking health information mostly from a professional was the second largest group. Potentially, a physician referral to online sources of sexual health information could move some of these men from the mostly health professional group to the diverse sources group.

Because of the Internet, MSM have relatively easy access to sexual health information. However, assuming that men will seek and find credible sexual health information without external guidance might be a naive assumption. Participants primarily used Internet search engines and GLBT websites. They were least likely to use a governmental, public media, or insurance website, or to email a healthcare provider. For example, only 28.4% of MSM in this sample used government websites such as www.cdc.gov, considered by many to be a “gold standard” of health information. Future research should focus on why men are not using these sites and identify strategies to increase their usage. In the interim, physicians and educators should refer men to credible websites. Educators should target hard-to-reach MSM whom appear to not be seeking out sexual health information. While we are aware of CBOs/ASOs conducting outreach on chat rooms and social networking websites, we found little research on the effectiveness of chat room interventions and none on social networking interventions. With the increasing use of these technologies, opportunities exist to learn if educators can increase MSM’s knowledge of their sexual health by directing them to credible websites via links and a brief explanation of the websites’ utility. Since a large number of MSM are Internet users, educators using chat rooms and social networking websites are likely to reach men in the minimal health-seeking and online only health-seeking groups.

This study provides a better understanding of the offline and online health-seeking patterns of MSM. With this information, physicians, researchers, and educators can identify strategies to increase men’s consumption of sexual health information. Though knowledge alone is insufficient to cause behavior change, with increased information, MSM are better equipped to avoid contracting HIV/STIs, to test regularly, and to identify symptoms and access care when necessary.