Xia, Qiang MD, MPH*†; Osmond, Dennis H PhD‡; Tholandi, Maya MPH*; Pollack, Lance M PhD‡; Zhou, Wei MS§; Ruiz, Juan D MD, MPH, DrPH*; Catania, Joseph A PhD‡
In 2002, California ranked second in the country for the cumulative number of reported AIDS cases and accounted for approximately 14.9% of all cases in the United States and Puerto Rico.1 Men who have sex with men (MSM) continue to be the primary risk category for AIDS cases in California, comprising 60% and 53% of male and all reported AIDS cases in 2002, respectively.2,3 Given that only 5% to 7% of American men have had sex with another man during adulthood,4,5 these figures illustrate that HIV/AIDS continues to be an epidemic among MSM in California.
Studies conducted throughout the state have shown a high level of HIV prevalence among younger and older MSM samples,6-9 and recent reports of increased sexual risk behaviors have resulted in an increase in the incidence of sexually transmitted diseases (STDs) and HIV infection among MSM.10-16 Studies based on convenience sampling may produce biased estimates, however, by overrepresenting men with high levels of risk. Time-location random sampling provides a probability sample of MSM but may be biased toward men who frequently visit gay venues.7,17,18 Results from population-based studies conducted in metropolitan areas, such as San Francisco and Los Angeles,8,19 may not be generalizable to the rest of the state.
A statewide population-based sample of MSM from metropolitan and rural areas would provide a better picture of the HIV/AIDS epidemic across California to guide intervention efforts. The ability to obtain a statewide population-based sample of MSM has always been restricted by the high costs of screening a large number of households to obtain a large enough sample of MSM for useful estimates. The California Health Interview Survey (CHIS),20 a random-digit dial survey, interviews more than 50,000 adults on a variety of health topics in California every other year and includes a question on sexual orientation. We have taken advantage of the CHIS sample to obtain a statewide population-based sample of MSM. In this study, we present data on HIV prevalence, sexual risk behaviors, and HIV testing among MSM.
Participants and Procedures
The sample included men aged 18 to 64 years who participated in CHIS 2001, self-identified as gay or bisexual, and gave permission to be recontacted. These criteria yielded a sample of 741 men. Contact information from CHIS 2001 was limited to only a first name (or an initial or other method to refer to the individual in the household) and addresses obtained from reverse directory services for a portion of the sample. Of the 741 records, 617 men had a mailing address. An initial letter was sent to the 617 men with an address, informing them of the study. The letter explained the nature and importance of the study and informed the men that an interviewer would be calling them in the near future.
The interviews were conducted from May 5 through June 23, 2002. With a computer-assisted telephone interviewing (CATI) program, MSM were interviewed in English or Spanish for 30 to 45 minutes. Respondents who completed the interview and self-reported as HIV-negative or of unknown HIV status were asked to take an HIV test. Those men who consented to take an HIV test were sent a home urine specimen collection kit (Calypte Biomedical Corporation, Berkeley, CA). The kit included instructions for collecting and mailing the urine sample and where to obtain the HIV test result. A card provided as part of the test kit contained the respondent's study identification (ID) number and instructions to call for test results 3 weeks after sending the urine sample. The card also provided a toll-free telephone number, the days of the week, and times of day when respondents could call for test results. Respondents identified themselves on the telephone by ID number only. Test results and counseling were provided over the telephone by a certified HIV test counselor. Samples were tested for HIV antibodies by the San Francisco Department of Public Health microbiology laboratory, using an enzyme immunoassay (EIA; Calypte Biomedical Corporation). Positive specimens were confirmed by Western blot analysis (Calypte Biomedical Corporation).
The study protocol was reviewed and approved by the Committee for the Protection of Human Subjects at the University of California, San Francisco, and the California Health and Human Services Agency.
Participants were asked about their age, race/ethnicity background, education, income, city/rural area currently residing in, and length of residency.
HIV Testing and HIV Infection Status
Participants were asked whether they had ever been tested for HIV and, if so, the dates of the first and last tests and the results. Those who self-reported as HIV-negative or of unknown HIV infection status were asked to take an HIV test using a home urine specimen collection kit. The self-reported HIV infection status and biologic testing results were measured.
First, respondents were asked whether they had had sex with men, women, or both in the past 12 months, followed by a series of questions about their sexual behaviors with male partners, including the number of male sexual partners and number of partners with whom they had engaged in various sexual behaviors (insertive and receptive anal intercourse with and without a condom). They were also asked about their sexual behaviors in the past 12 months on a partner-by-partner basis (frequency of insertive and receptive anal intercourse with and without a condom and their partner's HIV status) for their current primary partner and 3 most recent other male partners or for their 4 most recent male sexual partners if they did not have a primary partner.
Partner-by-partner information, including knowledge of the partner's HIV status, was used to determine whether unprotected anal intercourse (UAI) had occurred with a serodiscordant partner. We describe serodiscordant partnerships as (1) “the serodiscordant risk to the receptive partner” when the receptive partner is HIV-negative and the insertive partner is HIV-positive or of unknown HIV status or the insertive partner is HIV-positive and the receptive partner is HIV-negative or of unknown HIV status and as (2) “the serodiscordant risk to the insertive partner” when the insertive partner is HIV-negative and the receptive partner is HIV-positive or of unknown HIV status or the receptive partner is HIV-positive and the insertive partner is HIV-negative or of unknown HIV status (Table 1).
We used a 6-level variable categorizing a subject's sexual risk in the past 12 months from low to high risk: no male sexual partners, no anal intercourse, protected anal intercourse (100% condom use), 100% HIV-seroconcordant UAI, serodiscordant risk to the insertive partner for UAI, and serodiscordant risk to the receptive partner for UAI.
Participants reported if they or their sexual partners had ever injected or currently inject any recreational drugs.
Sexually Transmitted Diseases
Participants were asked if they had ever been diagnosed with an STD and whether they had been tested for an STD in the past 12 months.
In our analysis, each case was weighted to account for the probability of selection, nonresponse, and undercoverage. The replicate Jackknife method was used to estimate standard errors by using 80 replicate weights, which were created with the paired Jackknife method. All data were weighted unless indicated otherwise.
The prevalence of HIV infection, sexual risk behaviors, and HIV testing, with 95% confidence intervals (CIs), were computed for the entire sample and subgroups.
Chi-square tests were performed to examine demographic and behavioral risk factors associated with HIV infection, sexual risk behaviors, and HIV testing. We did not perform multivariate analysis, because the sample size of our study was small (n = 398) and our goal was to present a description of HIV prevalence, sexual risk behaviors, and HIV testing within the study sample rather than to identify independent predictors of HIV infection and sexual risk behaviors. All analyses were conducted with SUDAAN,21 which accounts for the complex sample design.
The target sample was defined by men aged 18 to 64 years who self-reported as gay or bisexual in CHIS 2001 but, regardless of their orientation, were screened for same-gender sexual behavior in the past 10 years. Of a total of 875 men who self-identified as gay (n = 593) or bisexual (n = 282) in CHIS 2001, 741 (84.7%) agreed to participate in the CHIS MSM follow-up study. No statistical difference was found between the men who agreed to participate and those who did not in terms of age, race, or metropolitan statistical area (MSA) status (Los Angeles MSA, San Francisco MSA, MSA1 [other large MSA], MSA2 [smaller MSA], and Not MSA).
Among 741 respondents who gave consent for follow-up, 193 (26.0%) were not reachable and 114 (15.4%) men were excluded because of no sex with a male partner in the past 10 years, leaving a total of 434 (58.6%) respondents eligible for the study. Thus, the obtained sample was composed of sexually active (in the past 10 years) MSM aged 19 to 65 years. Of the 434 eligible respondents, 398 (91.7%) completed the interview.
The characteristics of the participants are presented in Table 2, with unweighted and weighted analysis results. The final sample of 398 MSM represents MSM throughout the state of California. Many (67%) were non-Hispanic white. Men with annual incomes of less than $20,000 and more than $80,000 accounted for 19%, respectively. Fifty-five percent of the men were from the Los Angeles and San Diego areas, 30% were from San Francisco and the San Francisco Bay Area, and 15% were from the rest of the state. Thirteen percent of the men reported ever being married to a woman.
Seventy-three men self-reported as HIV-positive. Based on our previous study of offering home testing, in which 100% of self-reported HIV-positive men were confirmed positive by HIV testing,22 only those who self-reported as HIV-negative or of unknown HIV status were offered HIV home urine specimen collection kits. Of the 325 respondents who were eligible, 213 (65.5%, unweighted) were tested for HIV antibodies, and, among those, 5 were HIV-positive.
Based on the combination of self-reported HIV-positive results, home urine specimen collection kit test results, and weighting those who self-reported as HIV-negative or unknown HIV status and were not tested by the home kits, the overall HIV prevalence was 19.1% (95% CI: 12.8% to 25.3%). Table 3 presents the point estimates and 95% CIs for HIV prevalence by participant characteristics without weighting those who did not return their home urine collection kits because of the small sample size. Subgroups of MSM with high HIV prevalence included men of mixed race (86.6%), men with a less than or equal to high school education (40.4%), men with an annual income less than $20,000 (35.0%), and men who ever injected recreational drugs (40.3%). The HIV prevalence in regions outside the major metropolitan areas was also high. Although the point estimate for the San Francisco Bay Area (14%) was lower than that for Los Angeles-San Diego (21%) and the rest of the state (20%), the 95% CIs overlapped the point estimates for the other 2 regions.
The estimate of HIV prevalence among the entire sample of MSM was based on self-reported seropositive status, urine testing results, and the assumption that those MSM who did not successfully return a home urine specimen collection kit had an HIV prevalence similar to those who did. The assumption was based on (1) a high proportion (93%, weighted; 94%, unweighted) of MSM in our sample who reported having ever had an HIV test before, (2) a high proportion (35%, weighted; 35%, unweighted) of 302 men who self-reported as HIV-negative had their last HIV-negative test in the past 12 months, (3) a low HIV prevalence (4.7%, weighted; 2.3%, unweighted) among those who self-reported as HIV-negative or of unknown status and successfully returned their home urine specimen collection kits, and (4) a similar (P = 0.67, weighted; P = 0.66, unweighted) distribution of testing history (tested negative more than 1 year ago, tested negative in the past 12 months, and never tested) between the respondents who did and did not successfully return their home urine specimen collection kits.
Sexual Risk Behaviors
In the past 12 months, 90% of respondents were sexually active with men and 9% had sex with women. In terms of sexual risk behaviors, 29% reported 5 or more male partners, 14% reported 2 or more UAI partners, 15% had UAI with a secondary partner, and 11% had serodiscordant UAI (Table 4).
Young men aged 19 to 29 years reported the lowest percentage (21%) of 5 or more male partners but the highest percentages of 2 or more UAI partners and serodiscordant UAI (21% and 14%, respectively). Men with a less than or equal to high school education reported the highest percentage (21%) of serodiscordant UAI, whereas all the other groups reported less than 10%. Nineteen percent of Hispanics and 17% of African Americans reported serodiscordant UAI. The percentage of men reporting UAI with a secondary partner was similar across all subgroups (age, education, race/ethnicity, income, and residence), with an overall prevalence of 15%. HIV-positive men reported more risky sex, including 5 or more male partners, 2 or more UAI partners, UAI with a secondary partner, and serodiscordant UAI (P < 0.05) (Table 5).
Overall, 93% of the sample had ever had an HIV test and received their test results. Among the 328 participants who were HIV-negative or of unknown HIV status 1 year before their 2002 follow-up interview, 108 had had an HIV test during the previous 12 months and 4 had received an HIV-positive test result.
We further assessed HIV testing behaviors among nonpositive men who were at high risk of HIV infection. By the definition in our study, men who were HIV-negative or of unknown HIV status and had 1 or more of the following behaviors in the past 12 months were considered to be at high risk: 5 or more male sexual partners, 2 or more UAI partners, UAI with a secondary partner, or serodiscordant UAI. One hundred thirty-one of the 328 nonpositive participants were found to be at high risk. Of these 131 men, 49% had had an HIV test during the past 12 months. Young men aged 19 to 29 years, men with an annual income between $40,000 and $59,999, and men living in San Francisco and the San Francisco Bay Area reported lower rates of HIV testing (Table 6).
To assess the participants' HIV testing behavior, we selected men who reported high-risk sexual behaviors in the past 12 months and who were HIV-negative or of unknown status 1 year before their interview. Complicating this issue was the fact that a number of men were found to be HIV-positive in the previous 12 months; we were thus unable to determine whether their high-risk sexual behaviors occurred before or after the test. If the high-risk sexual behavior only occurred after they were found to be positive, although this is unlikely to be the case, it would make them ineligible for the assessment of HIV testing behavior. Therefore, to examine the relation more conservatively, we excluded 4 men who were found to be positive during the past 12 months from the analyses. The results with and without these men were virtually the same, with 44% of men reporting an HIV test. Again, young men aged 19 to 29 years, men with an annual income between $40,000 and $59,999, and men living in San Francisco and the San Francisco Bay Area reported lower rates of recent HIV testing.
Our statewide population-based study indicates that HIV prevalence among MSM in California is still high: 19.1% (95% CI: 12.8% to 25.3%). This estimate was similar to some previous HIV prevalence studies conducted in metropolitan areas in California.8,19 We did not find evidence of lower HIV prevalence in other areas of California outside the San Francisco-Los Angeles-San Diego metropolitan areas. These findings underscore the historical shift of HIV from major urban epicenters to a statewide epidemic reflecting relative homogeneity of prevalence levels across the state, albeit incidence rates may vary from one location to another. Our findings may reflect the migration of the epidemic from urban centers to other locales, a process that is only poorly understood at present, or they may reflect that more men in outer areas are becoming infected. Both issues warrant additional research. To account for the evolving epidemic, HIV/AIDS prevention needs to be thought of in terms of the state as a whole rather than as selected venues within selected urban centers and HIV/AIDS surveillance requires a statewide design rather than a selected urban center/venue approach.
Generally, weighted and unweighted estimates of prevalence do not vary greatly, but there are a couple of notable exceptions. The estimate for young men 19 to 29 years old is 9% unweighted and 21% weighted. Because these men are especially mobile and our study was conducted a year after the original CHIS sample, losses to follow-up in that period account for the large effect of weighting in this age group. Probably for the same reason, similarly large increases attributable to weighting occur among Hispanic men and among men with a high school education or less. The loss to follow-up of these young and highly mobile populations suggests that weighted analysis may be more appropriate.
Although the response rate (91.7%) among those reachable and eligible men was high, approximately one quarter (26.0% [n = 193]) of men who originally agreed to participate in the follow-up study were not located because of time elapsed and the mobility of the population. Such a high nonresponse rate and the fact that we do not know the differences in HIV prevalence and risk behaviors between those who were reachable and those who were not raised a question as to whether the mobility introduced a significant bias. We believe that there was no strong association between mobility and HIV prevalence/risk behaviors after controlling for the factors (age, race, in household with individuals less than 18 years old, and MSA status) that we used for weighting, and our study provided good estimates of HIV prevalence and risk behaviors among MSM after weighting to correct some nonresponse bias.
We selected 4 variables as indicators of high sexual risk behavior: 5 or more male partners, 2 or more UAI partners, UAI with a secondary partner, and serodiscordant UAI. It is interesting that inferences about the differences in risk among subgroups-age, race, and education-depend on the variable chosen. There was little variation in the percentage of men (15%) who had UAI with secondary partners across subgroups, but the group at highest risk on other indicators varied with the measure. The lack of variation in the percentage reporting UAI with a secondary partner among age and racial groups contrasts with the substantial variation in the percentage reporting serodiscordant UAI. Younger men and Hispanic and African-American men reported more serodiscordant UAI, presumably the highest risk behavior. This difference could be attributable to the higher testing level (97.5%; the serodiscordant variable included partners of unknown HIV status as risky) and possibly greater use of serosorting among white MSM. When we assess sexual risk behavior among MSM, we should look at all indicators to avoid reaching broad conclusions based on a single indicator and to make intervention activities more specific to the needs of subgroups. HIV-positive men reported more risky sex, including 5 or more male partners, 2 or more UAI partners, UAI with a secondary partner, and serodiscordant UAI. Studies have demonstrated that disclosure of HIV serostatus to partners decreases sexual risk behaviors.23,24 We may need to promote mutual disclosure of HIV serostatus between sexual partners and use of condoms among young or less highly educated MSM to avoid potential serodiscordant risk.
All surveys of MSM underrepresent subsegments of the MSM population to some extent. Telephone surveys have been viewed with caution in this regard; however, numerous telephone surveys of sexual behaviors have been successfully conducted in the general population and among MSM.8,25,26 Given the high telephone coverage (96.18%) in California27 and the poststratification adjustment to correct the bias caused by nontelephone coverage, we believe that our probability sample of MSM is representative of MSM in California. Future telephone surveys may face the challenge of an increasing number of Americans switching from land lines to cell telephones.
Our estimates may be affected by sampling bias. We only sampled men who self-identified as gay or bisexual. HIV prevalence rates and HIV-related risk behaviors may differ among men who self-identified as heterosexual but had sex with men.28,29 Because studies have shown that a high percentage (88%-93%) of MSM self-identify as gay or bisexual,8,20,28 we do not expect that the exclusion of non-self-identified MSM biased our prevalence estimates markedly.
We excluded men (n = 114) who did not have sex with men in the past 10 years. Excluding non-MSM and those gay or bisexual men who had no homosexual risk of HIV infection in the past 10 years would make our study results more appropriate to guide HIV intervention programs that only target those who are at risk. The sample of our study was representative of sexually active adult MSM in California.
Overall, a high percentage (93%, weighted; 94%, unweighted) of MSM surveyed and living in California had ever had an HIV test and received their results. Among the non-positive MSM who were at high risk, 49% had had an HIV test in the past 12 months. Young men aged 19 to 29 years and men with an annual income between $40,000 and $59,999 reported lower rates of HIV testing. Men with a lower income may have more chances to access HIV testing through outreach services, whereas men with a high income may have better knowledge of HIV and test for HIV more frequently. In 2003, the Centers for Disease Control and Prevention launched a new initiative, Advancing HIV Prevention: New Strategies for a Changing Epidemic, which is aimed at reducing barriers to early diagnosis of HIV infection and increasing access to quality medical care, treatment, and ongoing prevention services.30 It is likely to be more effective if we can target interventions to those who are at high risk but less likely to have an HIV test.
In conclusion, HIV prevalence among MSM living in California continues to be high across the whole state, and population-based studies are needed periodically to complement findings from surveys using other sampling designs, such as convenience, time-location, and respondent-driven sampling.31,32
This study would not have been possible without the extensive cooperation of the respondents who were willing to participate in the project. The authors also thank E. Richard Brown at the University of California, Los Angeles, Center for Health Policy Research, for his technical assistance and J. Michael Brick, Ismael Flores Cervantes, W. Sherman Edwards, Alan Martinson, and Vasudha Narayanan at WESTAT Corporation for their data collection and statistical assistance.
4. Rogers SM, Turner CF. Male-male sexual contact in USA: findings from five sample surveys, 1970-1990. J Sex Res. 1991;28:491-519.
5. Binson D, Michaels S, Stall R, et al. Prevalence and social distribution of men who have sex with men: United States and its urban centers. J Sex Res. 1995;32:245-254.
6. Katz MH, McFarland W, Guillin V, et al. Continuing high prevalence of HIV and risk behaviors among young men who have sex with men. J Acquir Immune Defic Syndr Hum Retrovirol. 1998;19:506-512.
7. Valleroy LA, MacKellar DA, Karon JM, et al. HIV prevalence and associated risks in young men who have sex with men. JAMA. 2000;284:198-204.
8. Catania JA, Osmond DH, Stall RD, et al. The continuing HIV epidemic among men who have sex with men. Am J Public Health. 2001;91:907-914.
9. Dolcini MM, Catania JA, Stall RD, et al. The HIV epidemic among older men who have sex with men. J Acquir Immune Defic Syndr. 2003;33(Suppl 2):S115-S121.
10. Chen SY, Gibson S, Katz MH, et al. Continuing increases in sexual risk behavior and sexually transmitted diseases among men who have sex with men, San Francisco, 1999-2001. Am J Public Health. 2002;92:1387-1388.
11. Centers for Disease Control and Prevention. Increases in unsafe sex and rectal gonorrhea among men who have sex with men-San Francisco 1994-1997. MMWR. 1999;48:45-48.
12. Ekstrand ML, Stall RD, Paul JP, et al. Gay men report high rates of unprotected anal sex with partners of unknown or discordant HIV status. AIDS. 1999;13:1525-1533.
13. Chen SY, Gibson S, Weide D, et al. Unprotected anal intercourse between potentially HIV-serodiscordant men who have sex with men, San Francisco. J Acquir Immune Defic Syndr. 2003;33:166-170.
14. Gibson S, Klausner J, Kim A, et al. Continuing increases in sexual risk behavior, sexually transmitted diseases, and HIV incidence among men who have sex with men, San Francisco [abstract MoPeC3430]. Presented at: XIV International AIDS Conference; 2002; Barcelona.
15. Kim AA, Kent CK, Klausner JD. Risk factors for rectal gonococcal infection amidst resurgence in HIV transmission. Sex Transm Dis. 2003;30:813-817.
16. Katz MH, Schwarcz SK, Kellogg TA, et al. Impact of highly active antiretroviral treatment on HIV seroincidence among men who have sex with men: San Francisco. Am J Public Health. 2002;92:388-394.
17. Lemp GF, Hirozawa AM, Givertz D, et al. Seroprevalence of HIV and risk behaviors among young homosexual and bisexual men: the San Francisco/Berkeley Young Men's Survey. JAMA. 1994;272:449-454.
18. MacKellar D, Valleroy L, Karon J, et al. The Young Men's Survey: methods for estimating HIV seroprevalence and risk factors among young men who have sex with men. Public Health Rep. 1996;111(Suppl 1):138-144.
19. Stall R, Catania J, Osmond D, et al. The distribution of HIV infection among men who have sex with men: a household-based sample of four large urban centers in the USA [abstract ThOrC716]. Presented at: XIII International Conference on AIDS; 2000; Durban.
20. UCLA Center for Health Policy Research. California health interview survey. Available at: www.chis.ucla.edu
. Accessed October 20, 2004.
21. Research Triangle Institute. SUDAAN Language Manual [computer program]. Release 9.0. Research Triangle Park, NC: Research Triangle Institute; 2004.
22. Osmond DH, Catania JA, Pollack L, et al. Obtaining HIV test results with a home collection test kit in a community telephone sample. J Acquir Immune Defic Syndr. 2000;24:363-368.
23. De Rosa CJ, Marks G. Preventive counseling of HIV+/men and self-disclosure of serostatus of sex partners: new opportunities for prevention. Health Psychol. 1998;17:224-231.
24. Kippax S, Noble J, Prestage G, et al. Sexual negotiation in the AIDS era: negotiated safety revisited. AIDS. 1997;11:191-197.
25. Catania JA, Coates TJ, Stall RD, et al. Changes in condom use among homosexual men in San Francisco. Health Psychol. 1991;10:190-199.
26. Catania JA, Binson D, van der Straten A, et al. Methodological research on sexual behavior in the AIDS era. In: Rosen R, Davis C, Ruppel H, eds. Annual Review of Sex Research. Mt. Vernon, IA: The Society for the Scientific Study of Sexuality; 1995:77-125.
27. US Census Bureau. Current population survey, March 2001.
28. Goldbaum G, Perdue T, Wolitski R, et al. Differences in risk behavior and sources of AIDS information among gay, bisexual and straight-identified men who have sex with men. AIDS Behav. 1998;2:13-21.
29. Doll LS, Petersen LR, White CR, et al. The Blood Donor Study Group. Homosexually and nonhomosexually identified men who have sex with men: a behavioral comparison. J Sex Res. 1992;29:1-14.
30. Centers for Disease Control and Prevention. Advancing HIV prevention: new strategies for a changing epidemic-United States, 2003. MMWR. 2003;52:329-332.
31. Heckathorn D. Respondent-driven sampling: a new approach to the study of hidden populations. Soc Probl. 1997;44:174-199.
32. Heckathorn D. Respondent driven sampling. II: deriving valid population estimates from chain-referral samples of hidden populations. Soc Probl. 2002;49:11-34.
© 2006 Lippincott Williams & Wilkins, Inc.