Prevalence of Same-Gender Sexual Behavior and HIV in a Probability Household Survey in Mexican Men.

For more than a decade the impact of the human immunodeficiency virus (HIV) infection has been felt in many countries. Mexico is not an exception, and AIDS is among the leading causes of death in the young adult population. The fact that HIV is mainly transmitted by sexual activity makes it imperative that we have a better understanding of sexual behavior, in order to be able to favorably influence the trend of the epidemic.

Case reports suggest that the epidemiological pattern of AIDS in Latin America is different from that observed in other parts of the world. Within this region, Mexico seems to have unique characteristics (Mann, Tarantola, & Netter, 1992; Valdespino, Izazola, & Rico, 1988); 95% of the cases to date among men and 45% among women are due to sexual transmission. Among men with sexually acquired AIDS, 60% were classified as homosexuals, 25% as bisexuals, and 15% as heterosexuals. HIV transmission due to injection drug use is negligible in Mexico ([is less than] 1% of reported cases). As of December 31, 1994, 20,796 AIDS cases had been reported in Mexico, most of which have occurred in men (85%), and in Mexico City (56%) (Instituto Nacional de Diagnostico y Referencia Epidemiologicos [INDRE], 1995).

Scientific reports of same-gender sexual behavior and risks for HIV in Spanish-speaking countries are rare. Available reports on male sexual behavior in Mexico, and concerning men in general from Latin-America and Mediterranean countries, have suggested that a substantial number of men engage in homosexual practices, and that most of these men sustain mainly heterosexual relationships (i.e., they are behaviorally bisexual) (Boulton, 1991; Carrier, 1985). It has been hypothesized that a high prevalence of bisexual behavior will have a major impact on the risk of HIV infection among women and children (Carrier, 1989).

A previous study in six Mexican cities documented substantial bisexual behavior among men interviewed in a cross-sectional convenience sample of gay gathering places (Izazola et al., 1991). Because the study of sexual behavior using population-based probability sampling has been rare, convenience surveys of sexual behavior have frequently been the main sources for estimates for the planning of AIDS-related activities (Turner, Miller, & Moses, 1989).

Projections of the AIDS epidemic and design of intervention strategies critically depend on accurate estimates of the magnitude and distribution of the population engaging in risky sexual behavior. Therefore, we conducted a study to provide quantitative population-based estimates of adult males who engaged in exclusive same-gender behaviors and in bisexual behaviors, to estimate HIV-1 prevalence and to discuss implications for HIV-1 sexual transmission in a population with low injection drug use.

METHODS

Data Collection

A household probability survey was carried out in the Mexico City metropolitan area (MCMA) from June 1992 to March 1993. A sampling framework of sanitary jurisdictions from the National Health Surveys System of the Mexican Ministry of Health was used, which employs a multi-stage stratified probability area sampling design. The primary sampling units were clusters of blocks (equivalent to census tracks in the United States). Compact segments of 10 contiguous households were drawn using a systematic selection procedure.

A household listing was performed as the first step of the survey in order to obtain a census of all persons living in each household. Sociodemographic information was collected: age, gender, relationship to the head of the household, schooling, and occupation. For eligible men, information about living with a stable female partner or with their children was also obtained. Respondents were asked to participate in this health survey without being told that AIDS or sexual behavior was the focus of the questionnaire, since these topics were not included in the household listing. Interviewers were instructed to obtain the household information from the first adult found in the selected households.

Informed consent to participate in a survey of sexual behavior and to potentially donate blood and saliva for HIV-1 screening was obtained. Households were revisited up to 10 times. Face-to-face interviews were carried out among eligible individuals using a 25-minute structured questionnaire. Since inquiring about sexual behavior is socially sensitive, interviewers were instructed to question the respondent alone. If the conditions of the interview did not guarantee confidentiality, the interviewers were instructed to postpone or terminate the interview.

An initial total of 13,057 households were included in the sampling frame. One third of the households were excluded for several reasons: vacancy at the time of the survey (13%), not being a household at the time of the survey (6%), nonresponse to the household interview (1.6%), and having no eligible men between 15 and 60 years old in the household (13%). Consequently, 8,759 households were considered eligible, and 13,713 men between 15 and 60 years old lived in these households. Of these, 8,600 (63%) were able to be contacted on a person-to-person basis and were asked to respond to the individual questionnaire: 532 (6%) refused to answer the questionnaire and 8,068 (94%) were successfully interviewed. The overall response rate of this survey (59%) was lower than the national response rate found in a 1993 multipurpose national survey of chronic diseases in Mexico that used the same sampling framework (response rate = 67%). In this survey, the lowest response rate obtained was for males living in Mexico City (64%) (Secretaria de Salud, 1994).

The operational definition of same-gender sexual behavior was based on the response to questions that included physical contact with other men, such as practicing masturbation with a male partner or having oral or anal intercourse. This operational definition of homosexual behavior excludes activities in which no direct contact was reported; For example, masturbation in the presence of other males without further indication of direct physical involvement was excluded.

Collection of Specimens for HIV Screening

A subsample of the respondents was asked to provide blood and/or crevicular fluid (CF) samples for the detection of HIV; CF was collected using the Epitope[TM] oral fluid device (Parry, Perry, & Mortimer, 1987). Due to the small number of men expected to report homosexual behavior, they were over-sampled by making revisits (blinded to the interviewers) after the individual questionnaire was completed and analyzed; systematic samples of one of every three men who reported homosexual behavior and of one sixth of the heterosexual men were drawn. Both CF and serum were screened for HIV antibodies using second generation ELISA reagents (Frerichs, Htoon, Eskles, & Lwin, 1992; Mortimer & Parry, 1994). Repeated positive samples were tested with Western-Blot assays and interpreted according to the ASTHLD/CDC criteria (Centers for Disease Control, 1989). (ELISA: Abbott Laboratories, Chicago, IL, USA; HIV-1 Western-Blot Assay: Organon-Technika Corp., Durham, NC, USA.) Only the participants who provided blood samples were offered test counseling and their serologic results at the National HIV/AIDS Information Centers. The response rate was 69% for the collection of CF and 42% for blood. After discarding inadequate specimens, 1,200 CF and 406 blood samples were analyzed (290 paired) from 1116 individuals.

Statistical Analysis

We weighted survey data with sample weights to account for the differential probabilities of selection and nonresponse (Kish & Frankel, 1974). Standard errors of prevalence estimates were computed with a two-stage variance formula using SUDAAN software (Research Triangle Institute, 1990), a family of statistical procedures designed for analysis of data from complex sample surveys. The weighted odds ratios and the 95% confidence intervals were calculated taking into account the complex sample design, using multivariate logistic regression. Clustering in sampling designs may lead to larger standard errors because of the similarity among individuals in contiguous areas. This loss of precision may be measured by the design effect, defined as the multiplier to be applied to the variance of a survey estimate under simple random sampling to take account of the complex sample design. Given the low prevalence of same-gender sexual behavior in the general population, small design effects were expected (Kalton, 1993) even though this study used a multistage stratified and clustered sample. The estimated design effects for variables concerning homosexual and bisexual behaviors were less than 1.3, indicating little impact in the estimation of standard errors.

RESULTS

The demographic characteristics of the survey participants were similar to the male population as reflected in the 1990 Population Census (Instituto Nacional de Estadistica, Geografia e Informatica [INEGI], 1992)(see Table 1). However, men aged 45 and older were slightly underrepresented in the sample household census, compared to the 1990 census (15% vs. 17% respectively); there were fewer men with 4 years of college or more and graduate education (17% vs. 22%); and there were fewer economically inactive men (5% vs. 7%).

Table 1. Comparison of Selected Sociodemographic Characteristics of Mexico City Adult Males

Characteristics             National population    Survey household
                                   census               census
                              (N = 2,973,266)        (N = 13,713)

Age
    15-19                           19.8                 20.7
    20-24                           18.1                 19.0
    25-34                           27.5                 27.0
    35-44                           18.1                 18.6
    45-54                           11.5                 10.4
    55-60                            5.0                  4.2
Education
    Elementary                      28.7                 29.4
    Jr. high school                 27.9                 33.2
    High school                     21.6                 19.5
    University                      19.8                 15.9
    Graduate                         2.0                  0.7
Marital status
    Married                         54.8                 54.2
    Single                          42.8                 43.3
    Divorced, separated,             2.1                  2.4
      & widowed
Occupation
    Working                         73.6                 75.2
    Unemployed                       2.2                  2.3
    Students                        16.9                 17.7
    Economically inactive            7.3                  4.8

Note. According to the 1990 Population Census and Survey Respondents, Mexico City 1992-93. Included are the populations of 16 Sanitary Jurisdictions in the Mexico City metropolitan area: 14 in the Federal District (Milpa Alta and Tlahuac were excluded) and 2 in the State of Mexico (Naucalpan and Netzahualcoyotl). Results shown as percentage.

Table 2 provides estimates of the frequency of lifetime and past-year sexual behavior for the total sample as well as population estimates for the MCMA, according to the gender of the respondents' sexual partners. Almost 16% of participants had not initiated their "active sexual life" (i.e., never had anal, oral, or vaginal intercourse): The majority of these are the youngest respondents included in the survey (15-19 years of age). Thirty-seven men (0.4%) reported exclusive lifetime homosexual behavior and 173 (2.1%) reported having had sexual intercourse (defined as anal or vaginal penetration) with both men and women in their lifetime. The rest (81.6%) reported exclusive heterosexual behavior.

Table 2. Percentage Distribution by Gender of Lifetime and Past-Year Sexual Partners Among Men

Gender of sexual                %                Population
partners                     (95% CI)        estimates (95% CI)

Lifetime partners
  None                         15.8                470,200
                           (13.0-18.6)        (387,200-553,200)
  All females                  81.7               2,426,500
                           (68.9-94.3)      (2,049,000-2,804,000)
  Males & females              2.1                 637,000
                            (1.7-2.4)          (50,100-70,600)
  All Males                    0.4                 12,900
                            (0.3-0.6)          (7,600-18,200)
Total                         100.0               2,973,300
Past year partners
  None                         22.6                671,700
                           (22.6-26.5)        (555,200-788,300)
  All females                  76.7              22,279,300
                           (64.7-88.6)      (1,925,000-2,633,600)
  Males & females              0.3                  8,200
                            (0.1-0.4)          (4,100-12,300)
  All males                    0.5                  4,000
                            (0.3-0.7)          (7,700-20,300)
Total                         100.0               2,973,300

Note. Results shown for male respondents to a household probability survey, Mexico City, 1992-1993. Ages ranged from 15 to 60 years.

Of the 210 men reporting homosexual practices, 68% had the last homosexual encounter within the previous 10 years, 52% within the previous 5 years, 34% in the past 2, years and 27% in the previous year. The estimated prevalence of same-gender sexual behavior was 0.8% in the year previous to the interview: 0.5% exclusive homosexual behavior and 0.3% bisexual.

Of the men who were already sexually active, 8% of the heterosexuals, 10% of the bisexuals, and 27% of the exclusively homosexual men reported being celibate in the previous year (p [is less than] 0.0001). Of the 173 lifetime bisexual men, 156 (90%) reported sexual activity in the previous 12 months: 70% reported having sex only with women, 7% only with men, and 13% with both men and women.

Among married individuals, or men living in a consensual heterosexual partnership, 2% reported homosexual behavior at least once in their lifetime. Of these, only 5% reported being homosexually active in the past year (i.e., 0.1% of total married men reported a male partner in the 12 months previous to the interview).

Even while only a very small proportion of married men reported being homosexually active while being married or while living in a consensual heterosexual relationship, they represented half of the 173 lifetime bisexuals. Of the rest of the bisexual men, 5% were separated or divorced and 45% were single.

Homosexually Active Men in the Previous Year

In Table 3, the main behavioral characteristics of the 61 men who were homosexually active in the year previous to the interview are presented. Engagement in steady relationships was infrequent. Also, when compared to previous studies using convenience samples, there was a relatively low number of sexual partners, low frequency of anal intercourse in lifetime and in the past year in sexual encounters with males, and low frequency of condom use. While little more than one third of the homosexually active men reported sex with both men and women, 52% of them identified as heterosexual or bisexual. The small number of respondents in this category limited further analysis.

Table 3. Partnership and Sexual Practices of Homosexually Active Men

Variable                                                       %

Gender of sexual partners
     Had sex with males and females                           36%
     Had sex only with males                                  64%
Partnership in past year
     Have a steady female partner                             15%
     Have a steady male partner                               30%
Lifetime male partners(a)
     One lifetime partner                                     13%
     Never had anal intercourse                               23%
     Never had oral receptive intercourse                     43%
Behavior in last intercourse with a male
     Oral insertive sex                                       46%
     Condom use for oral insertive sex                        11%
     Oral receptive sex                                       41%
     Condom use for oral insertive sex                        17%
     Anal insertive sex                                       32%
     Condom use for anal insertive sex                        50%
     Anal receptive sex                                       56%
     Condom use for anal receptive sex                        65%
     Anal sex with a casual partner                           27%
     Used a condom for anal sex with a casual partner         30%
Condom use in last sexual encounter with a male
     With last sexual partner                                 46%
     With steady male partner                                 58%
Self-reported sexual identity
     Heterosexual                                             39%
     Bisexual                                                 13%
     Homosexual                                               48%

Note. Data for the previous year, in a household probability sample in Mexico City, Mexico, 1992-93, (n = 61).

(a) Median number of lifetime partners = 20. Median number of lifetime partners with anal intercourse = 15.

HIV Screening

Blood samples were collected and processed for HIV screening from 406 respondents; 1 out of 41 samples from the homosexual/bisexual men tested positive for HIV, and none of 365 blood samples of the exclusively heterosexual men was positive (Fisher's exact test p-value = 0.10). One of 46 CF samples of the homosexual/bisexual men and 1 of 1,243 CF samples of the heterosexual men tested positive for HIV (Fisher's exact test p-value = 0.08). When individuals provided both serum and saliva, HIV testing was performed in both specimens; all results were concordant. Additionally, two individuals reported that they were actively homosexual and HIV infected, but refused to provide new specimens to be retested for HIV because they had tested positive for HIV with ELISA and Western Blot tests. In summary, the overall weighted estimate of the prevalence of HIV infection in Mexico City males aged 15 to 60 years old was 0.2%. The prevalence was significantly lower in the 1,145 men who reported exclusive heterosexual behavior compared to the 71 homosexual or bisexual men tested (0.09% vs. 4%, Fisher's p [is less than] 0.001).

Evaluation of Nonresponse Bias

The sensitive nature of the topic likely ensured that this survey would provide lower-bound estimates of same-gender sexual behavior. The format of a face-to-face structured interview may have led some individuals to conceal present or past homosexual behaviors; this misclassification would likely yield lower estimates. The extent and likelihood of nonresponse bias was tested with a logistic regression model using the demographic variables included in the household questionnaire, as shown in Table 4. While 5,846 individuals did not have a completed individual questionnaire (including 532 refusals), information on demographic variables was available for all individuals from the household interview. The variables were entered simultaneously in a multivariate analysis, a backward selection of significant variables was performed in order to provide adjusted estimates.

Table 4. Logistic Regression Models to Estimate the Probability of Reporting Same-Gender Sexual Behavior

                                             Homosexual
                                             behavior

Characteristics                n (%)        Bi-       p
                                          variate

Age
  15-19                         1668        1.3     0.0005
                               (1.7)
  20-29                         2732        2.5
                               (3.1)
  30-39                         1812        2.8
                               (3.5)
  40-49                         1113        2.0
                               (2.5)
  50-60                         726         1.0
                               (1.3)
Position in the household
  Head                          4098        0.7       NS
                               (2.5)
  Son                           3047        0.8
                               (2.6)
  Other                         826         1.0
                               (3.3)
Education

  Primary or less               1812        1.0       NS
                               (2.4)
  High school                   4323        1.1
                               (2.7)
  [is less than or equal
    to] 3 years of college      617         1.4
                               (3.3)
  [is greater than or equal
    to] 4 years of college      1156        1.0
                               (2.3)
Have children

Yes                             4193        0.6     0.0006
                               (2.0)
No                              3817        1.0
                               (3.2)
Occupation
Unemployed                      248         1.0     0.0033
                               (3.4)
Lower levels                    842         0.8
                               (2.8)
Blue collar                     2544        0.7
                               (2.6)
Students                        1422        0.3
                               (1.2)
White collar                    1780        0.9
                               (3.2)
Higher levels                   1225        0.8
                               (2.80

                               Relative odds

Characteristics                Multi-       P
                              variate

Age
  15-19                         1.1        NS

  20-29                         1.7        0.01

  30-39                         2.4        0.0002

  40-49                         5.1        0.0001

  50-60                                    1.0

Position in the household
  Head                          1.1        NS

  Son                           0.8        NS

  Other                         1.0

Education

  Primary or less               1.0

  High school                   1.1        NS

  [is less than or equal
    to] 3 years of college      1.5        NS

  [is greater than or equal
    to] 4 years of college      0.7        NS

Have children

Yes                             0.5        0.0001

No                              1.0

Occupation
Unemployed                      1.0        NS

Lower levels                    1.0        NS

Blue collar                     1.0        NS

Students                        0.4        0.0001

White collar                    1.0        NS

Higher levels                   1.0        NS

Note. Reported according to demographic variables.

Using this multivariate regression, we calculated that the mean estimated probability of reporting homosexual behavior was 2.5% among the men who answered the individual questionnaire (respondents) and 2.9% for those who only answered the household questionnaire (nonparticipants). The weighted mean, using information from respondents and estimates for nonrespondents, was 2.7%, thus showing a statistically significant nonresponse bias of limited size (0.2%) in the total estimate of same-gender sexual behavior.

A higher number of re-visits to interview potential respondents was statistically associated with a higher prevalence of reported lifetime homosexual behaviors (linear trend p-value = 0.03). The prevalence of reports of lifetime same-gender sexual behavior was 2.5% (95% CI: 2.1 - 2.9%) if the interview took place within the first 5 visits, 4.1% (95% CI: 3.0 - 5.2%) if it was performed during the sixth and seventh revisit, and 6.5% (95% CI: 1.5 - 11.5%) if response occurred between 8 and 10 revisits (p for trend = 0.04). Seventy-eight percent of the interviews took place during the first 3 visits to the households, and only 1.2% between revisits 8 and 10. This information was not included in the model to estimate the probability of reporting homosexual behavior because the information was censored.

DISCUSSION

This study provides quantitative estimates of the prevalence of same-gender sexual behavior of males 15 to 60 years old living in the Mexico City metropolitan area. The estimated prevalence of homosexual and bisexual behavior and the estimated prevalence of HIV infection were lower in this population-based probability survey than results from other nonprobability studies.

To our knowledge, this is the first population-based survey using probability sampling methodology for the study of male same-gender sexual behavior in a Latin American or Spanish-speaking country.

While anthropologic studies have suggested that homosexual and, particularly, bisexual behavior among Mexican men is very common (Carrier, 1985), this study indicates a low prevalence. The different methodologies used to obtain the estimates may be accountable, on their own, of yielding to over- or under-estimations; for example, the Carrier study cited above used qualitative and ethnographic methods, and this fact alone might account for the whole difference between the estimates.

Two other hypotheses may also account for this discrepancy. First, the main focus of this study was to quantify the frequency of behaviors with potential for HIV transmission, and other dimensions of sexual orientation were not taken into consideration. For instance, there was no attempt to measure homosexual attraction, arousal, or fantasies, since these complex constructs are difficult to be measured and are not necessarily related to HIV transmission. As has been reported recently, "Sexual fantasy, in contrast to sexual activity, refers to private psychological imagery associated with feelings that are explicitly erotic or lustful and with physiologic responses of sexual arousal" (Friedman & Downey, 1994, p. 923). Furthermore, some behaviors that could be interpreted as features of homosexual orientation were not included in the operational definition of homosexual behavior for this study. For instance, 29% of the sample had practiced masturbation in the presence of other males, although no physical contact was reported. If we had defined same-gender sex based on sexual identity, the prevalence estimates would also be different: 0.3% of the total sample self-identified as bisexual. Of these, one fourth reported that they were sexually inactive in the previous 5 years, one third had sex exclusively with women, one third had sex with both men and women, and the rest had sex only with men. Among those self-identified as homosexual (0.4%), one third also had sex with women in the past 5 years. In sum, the issue of sexual orientation goes beyond the focus of the present communication, but may help explain some differences in the definition and estimation of male same-gender sex.

Second, stigma could have led some individuals to conceal their homosexual or bisexual behavior when answering the questionnaire. Others may have been reluctant to participate in the survey, as was suggested by the trend noted in the reporting of same-gender sexual behavior and the number of times visited until response was obtained. However, the internal validity of the estimates of same-gender sexual behavior in this survey is supported by two facts: (a) The analysis of nonresponse bias using a logistic regression model provided an estimate of lifetime same-gender sexual behavior of 2.7% if all eligible men had responded, compared to reports of 2.5% among respondents, and (b) a higher HIV prevalence was found among those who reported same-gender sexual behavior compared to exclusive heterosexuals (4% and 0.09% respectively).

The findings of this quantitative study do not contradict findings from the qualitative studies cited above (Carrier, 1985; 1989). The way of selecting participants in convenience samples may affect the external validity of some studies. For example, a study designed to recruit bisexual men could gather a sample in which half were married men. However, the findings of this type of study must not be interpreted as an indication of frequent bisexual behavior among married men, because of the nonpopulation base of the sample.

Kinsey reported, based on a sample of U.S. White men from 1938 to 1948, that 37% had had at least some overt homosexual experience, that 10% were more or less exclusively homosexual for at least three years between the ages of 16 and 55, and 4% were exclusively homosexual throughout their lives after the onset of adolescence (Kinsey, Pomeroy, & Martin, 1948). However, since this was not a probability sample, the data did not allow inference to the population level and is not comparable to population-based studies. More recent analysis of U.S. national representative samples collected in 1970 (n = 3,018) and 1988 (n = 1,481) showed that approximately 2% of men engaged in same-gender sexual behavior in the previous year (Fay, Turner, Klassen, & Gagnon, 1989). Results from other U.S. national probability surveys conducted by NORC/General Social Surveys in 1988, 1989, and 1990 showed a lifetime same-gender sexual behavior pooled estimate of 6.0%, and 1.8% in the previous year (Billy, Tanfer, Grady, & Klepinger, 1993; Rogers & Turner, 1991). The most recent national U.S. survey (1994) estimated a prevalence of same-gender sexual behavior since puberty of 7.1% (Lauman, Gagnon, Michael, & Michaels, 1994).

A French national telephone survey that included 9,928 men (response rate of 66%) showed a male same-gender sexual behavior lifetime estimate of 4.1% and a past year estimate of 1.1% (ACSF Investigators, 1992). A face-to-face survey of 8,384 males in Great Britain (response rate of 65%) showed a lifetime estimate of male same-gander sexual behavior of 3.6%, and 1.1% for the past-year estimate (Johnson, Wadsworth, Wellings, Bradshaw, & Field, 1992). The results from these studies and the survey in Mexico City show similar past-year estimates of same-gender sexual behavior. A higher proportion of men who reported same-gender sexual behavior had sex with both men and women in France than in Great Britain, as was also observed in the Mexican sample (Bajos et al., 1995).

An analysis of the 1994 U.S. survey and other NORC/GSS surveys indicated differential same-gender sexual behavior prevalence in the previous year according to place of residence: 10% in the 12 top central cities, 3.6% in the next 88 central cities, and 1% in rural areas (Billy et al., 1993). Similarly, in Paris the prevalence was 4.7 times higher than in rural French communities (Lauman et al., 1994), and the lifetime estimate for London was 8.6% and 3.5% for the past year (ACSF Investigators, 1992). The Mexico City estimates of same-gender sexual behavior are comparable to the estimates in the U.S., France, and the United Kingdom at the national level, but are not comparable to the estimates from the largest cities in those countries. There was no evidence of clusters of homosexual and bisexual men in neighborhoods in Mexico City in the household survey. Consultation with leaders in the gay community in Mexico City pointed to the lack of gay neighborhoods, as found in the major cities in the U.S. and in London and Paris, as a potential explanation for this incongruence.

The 0.2% HIV prevalence found in this study was similar to a 0.06% HIV seropositivity rate in 400,000 voluntary blood donors in Mexico (national level) in 1993 (INDRE, 1993), or to a 0.08% seroprevalence rate in 187,311 blood donors in Mexico City during 1994 (INDRE, 1994).

There was a lower HIV prevalence in lifetime homosexual and bisexual men in this study (4%) compared to an estimate of 25% based on a sample in gay gathering places in Mexico City in 1988 (Izazola-Licea et al., 1991), or to estimated rates of 21% for bisexual and 34% for homosexual men who attended the AIDS Center for HIV testing in Mexico City in 1988-1989 (Hernandez et al., 1992). These differences may be explained by the self-selection of the participants of the convenience samples, where a higher frequency of homosexual and bisexual men reported riskier behavior. For example, one fourth of the homosexually active men in the household survey had never engaged in anal intercourse, and 13% of those who practiced anal sex had only one sexual partner.

Nevertheless, there is evidence of significant risk for HIV transmission among the participants: 36% of the sample of men homosexually active in the previous year practiced unprotected anal receptive intercourse, 16% practiced unprotected anal insertive intercourse, and 8% had unprotected anal intercourse with casual partners. Given that the overall estimate of condom use in the last sexual encounter with a man was 46%, it is expected that this level of use will permit the further spread of the HIV infection. On the other hand, however, a significant number of respondents who reported same-gender sexual behaviors had never engaged in anal insertive or anal receptive intercourse (23%); also, 43% reported that they had never performed oral receptive behavior with other males. These findings imply lower risks for HIV infection in the sampled population.

In general, concurrent bisexuality was an infrequent behavior compared to serial bisexuality. Only a minority of bisexual men were homosexually active while being in a steady heterosexual relationship (i.e., married). This finding may imply that potential HIV transmission from bisexual men to women and to newborns could be infrequent.

Homosexually active men are a hard to reach population by traditional health interventions, particularly because of their lack of identity as gay or bisexual men. Therefore, substantive research must be developed to reach them properly with HIV/AIDS and other sexually transmitted disease prevention programs.

This population-based survey suggests that the possibility of further HIV infection spread to women and children might be facilitated by the low prevalence of condom use among bisexual men. However, the low prevalence of active bisexual men and the relatively low-risk behaviors with men suggests that a low frequency of transmission is expected. Social stigma cannot be excluded as a reason for the low reported prevalence of same-gender sexual behavior, even though representative surveys in other communities have also shown low estimates of same-gender sexual behavior, and the internal validity of this survey appears substantive. Further research needs to be performed to validate these findings in other representative samples of Spanish-speaking communities. Additional efforts must be made to evaluate the impact of a low frequency of same-gender sexual behavior in the evolution of the HIV/AIDS epidemic, to evaluate the possibility of a sustainable heterosexual epidemic in a community with low rates of injection drug use.

REFERENCES

ACSF Investigators (1992). AIDS and sexual behaviour in France. Nature, 360, 407-409.

Bajos, N., Wadsworth, J., Ducot, B., Johnson, A., Le Pont, F., Wellings, K., Spira, A., Field, J., & the ACSF Group. (1995). Sexual behaviour and HIV epidemiology: Comparative analysis in France and Britain. AIDS, 9, 735-743.

Billy, J., Tanfer, K., Grady, W. R., & Klepinger, D. H. (1993). The sexual behavior of men in the United States. Family Planning Perspectives, 25, 52-60.

Boulton, M. (1991). Review-of the literature on bisexuality and HIV transmission. In R. A. P. Tielman, M. Carballo, & A. C. Hendriks (Eds.), Bisexuality & HIV/AIDS: A global perspective (pp. 187-209). New York: Prometheus Books.

Carrier, J. M. (1985). Mexican male bisexuality. In F. Klein & J. Wolf (Eds.), Bisexualities: Theory and research (pp. 75-85). New York: Haworth Press.

Carrier, J. M. (1989). Sexual behavior and the spread of AIDS in Mexico. Medical Anthropology Quarterly, 10, 129-142.

Centers for Disease Control. (1989). Interpretation and use of the Western Blot Assay for serodiagnosis of human immunodeficiency virus type 1 infections. Morbidity and Mortality Weekly Report, 38(Suppl. 7), 1-7.

Fay, R. F., Turner, C. F., Klassen, A. D., & Gagnon, J. H. (1989). Prevalence and patterns of same-gender sexual contact among men. [Published erratum appears in Science, 244, 1531] Science, 243, 338-348.

Frerichs, R. R., Htoon, M. T., Eskles, N., & Lwin, S. (1992). Comparison of saliva and serum for HIV surveillance in developing countries. The Lancet, 340, 1496-1499.

Friedman, R. C., & Downey, J. I. (1994). Homosexuality. The New England Journal of Medicine, 331, 923-930.

Hernandez, M., Uribe, P., Gortmaker, S., Avila, C., DeCaso, L. E., Mueller, N., & Sepulveda, J. (1992). Sexual behavior and status for Human Immunodeficiency Virus type I among homosexual and bisexual males in Mexico City. American Journal of Epidemiology, 135, 883-894.

Instituto de Diagnostico y Referencia Epidemiologica (INDRE). (1994, December). Medium term plan report of the Committee of Epidemiology, National Council for AIDS Prevention and Control. Mexico City, Mexico: Secretaria de Salud.

Instituto Nacional de Diagnostico y Referencia Epidemiologicos (INDRE). (1993). Enfermedades de transmision sexual y SIDA [Sexual transmitted diseases and AIDS], Mexico City, Mexico: Secretaria de Salud.

Instituto Nacional de Diagnostico y Referencia Epidemiologicos (INDRE).(1995, January). Boletin mensuel sobra SIDA/ETS (Mexico) enero de 1995. [Mensual Report about AIDS/STD (Mexico)]. Mexico City, Mexico: Secretaria de Salud.

Instituto Nacional de Estadistica, Geografia e Informatica (INEGI.) (1992). Resumen general XI censo general do poblacion y vivenda [XI National Population Census, 1990: General Abstract]. Mexico City, Mexico: INEGI.

Izazola-Licea, J. A., Valdespino-Gomez, J. L., Gortmaker, S. L., Townsend, J., Becker, J., Palacios-Martinez, M., Mueller, N. E., & Sepulveda Amor, J. (1991). HIV-1 seropositivity and behavioral and sociological risks among homosexual and bisexual men in six Mexican cities. Journal of Acquired Immune Deficiency Syndromes, 4, 614-622.

Johnson, A. M., Wadsworth, J., Wellings, K., Bradshaw, S., & Field, J. (1992) Sexual lifestyles and HIV risk. Nature, 360, 410-412.

Kalton, G. (1993). Sampling considerations in research on HIV risk and illness. In D. G. Ostrow & R. C. Kessler (Eds.), Methodological issues in AIDS behavioral research (pp. 53-74). New York: Plenum Press.

Kinsey, A. C., Pomeroy, W. B., & Martin, C. E. (1948). Sexual behavior in the human male. Philadelphia: W.B. Saunders.

Kish, L., & Frankel, M. R. (1974). Inference from complex samples. The Journal of the Royal Statistical Society, 36, 1-37.

Lauman, E. O., Gagnon, J. H., Michael, R. T., & Michaels, S. (1994). The social organization of sexuality. Sexual practices in the United States. Chicago: University of Chicago.

Mann, J. M., Tarantola, D. J. M., & Netter, T. W. (Eds.). (1992). AIDS in the world. A global report. Cambridge, MA: Harvard University Press.

Mortimer, P. P., & Parry, J. V. (1994). Detection of antibody to HIV in saliva: A brief review. Clinical and Diagnostic Virology, 2, 231-243.

Parry, J. V., Perry, K. R., & Mortimer, P. P. (1987). Sensitive assays for viral antibodies in saliva: An alternative to tests on serum. Lancet, 2, 72-75.

Research Triangle Institute. (1990). Software for survey data analysis (SUDAAN) (Version 5.31). Research Triangle Park, NC: Author.

Rogers, S. M., & Turner, C. E (1991) Male-male sexual contact in the USA: Findings from five sample surveys, 1970-1990. The Journal of Sex Research, 28, 491-519.

Secretaria de Salud. (1994). Encuesta nacional de enfermedads chronicas [National survey of chronic diseases]. Mexico City, Mexico: Secretaria de Salud.

Turner, C. E, Miller, H. G., & Moses, L. E., (Eds). (1989). AIDS, sexual behavior and intravenous drug use. Washington, DC: National Academy Press.

Valdespino-Gomez, J. L., Izazola-Licea, J. A., & Rico, B. (1988). AIDS in Mexico: Trends and projections. In Panamerican Health Organization, AIDS, profile of an epidemic (Scientific publication 512, pp 31-36). Washington, D.C.: Panamerican Health Organization.

Manuscript accepted July 30, 1999

Jose A. Izazola-Licea
Mexican Health Foundation, Mexico City, Mexico

Steven L. Gortmaker
Harvard School of Public Health

Kathryn Tolbert
The Putney School, Vermont

Victor De Gruttola
Harvard School of Public Health

Jonathan Mann

Dr. Jonathan Mann passed away in 1995.

Address correspondence to Jose A. Izazola-Licea, Periferico Sur 4809, Col. El Arenal Tepepan, 14610, Mexico, D. F., Mexico; e-mail: [e-mail].