HIV Prevalence, Risk, and Partner Serodiscordance Among Pregnant Women in Bangkok FREE

OUTSIDE the United States and western Europe, the human immunodeficiency virus (HIV) epidemic is largely heterosexual and is growing. The largest percentage increase in new cases is occurring in Asia and in women of childbearing age. Within 3 to 4 years, Asia may have the largest number of persons with HIV infection.^1- 3

In Asia, Thailand has had a relatively well-characterized epidemic. After an explosive increase in 1988 in HIV seroprevalence in persons who inject drugs, HIV seroprevalence increased rapidly in female sex workers (FSWs) and their male clients, and then spread into the general population of women of childbearing age.^4- 5 Based on national sentinel surveillance data, it is estimated that there are 23000 births (2.3%) to HIV-positive women of 1 million annual births in a total population of 60 million (C. Kunanusont, MD, PhD, Ministry of Public Health [MOPH], Thailand, oral communication, May 1997). Although data have shown a decline in HIV seroprevalence in male military recruits^6- 7 due to behavior changes such as increased condom use and decreased FSW patronage,^7- 9 sentinel seroprevalence data for antenatal women are less clear.^10- 11

Epidemiologic studies in Thailand have shown that FSW patronage is the leading HIV risk behavior for men.^12- 15 It is widely presumed that most transmission to non-FSW women is from a husband or a partner and that women are at risk because of the risk behavior of their partners.¹⁶ However, preliminary data from our study hospitals suggested a higher-than-expected serodiscordance among male partners of pregnant seropositive women. We thus investigated whether independent risk factors for infection could be identified in antenatal women in the general population, and how these risks were related to HIV serodiscordance among current partners.

The study was conducted in Bangkok at the 2 largest maternity services in Thailand. Siriraj Hospital is a 2000-bed university teaching hospital and Rajavithi Hospital is a leading MOPH hospital. At each hospital, about 20000 women register for antenatal care each year and both serve predominantly lower-income populations in the surrounding communities.

At both study hospitals, as in most hospitals in Thailand, HIV testing of pregnant women is a routine part of antenatal sexually transmitted disease (STD) control. Most women are tested for HIV, syphilis, and hepatitis B infection at first antenatal clinic (ANC) visit and at third trimester. About 85% of women delivering at the study hospitals register for antenatal care before their third trimester. The ANC HIV testing was introduced in 1991. Since 1992, all women registering for antenatal care are offered group HIV pretest counseling and asked for written consent for confidential HIV testing as part of routine hospital procedures (routine antenatal testing is offered throughout most of the country). The HIV testing is performed using enzyme immunoassay, and positive results are confirmed by particle agglutination or Western blot testing. Women testing HIV positive are offered individual, confidential posttest counseling. For enrolled women (below), hospital HIV test results were confirmed by enzyme immunoassay and Western blot testing at The HIV/AIDS Collaboration laboratory.

After posttest counseling, HIV-positive pregnant women were offered enrollment (about 45% of all seropositive pregnant women) in a prospective perinatal transmission study¹⁷ if they registered for antenatal care before their third trimester, resided in the Bangkok area, had a Thailand national identification card, intended to continue the pregnancy, planned to deliver at the study hospital, and were planning to have follow-up care for themselves and their children at the hospital. At enrollment, a pretested baseline questionnaire was administered covering demographic, obstetric, and behavioral information for the women and their perception of current partner risk behavior. After enrollment of a seropositive woman, the next consenting HIV-negative woman attending the ANC, matched by trimester, was offered enrollment as a control. Control women were given the same questionnaire but not followed.

At study hospitals, current partners of women testing positive for HIV or other STD tests are routinely asked to visit the hospital for confidential counseling and STD testing. Separate or joint confidential counseling by hospital and study staff is offered to the woman and her partner. Confirmed partner HIV test results were used for analysis of partner HIV status and serodiscordance.

We considered couples married if they reported themselves as married according to legal or traditional criteria. Because 99% of both HIV-positive and HIV-negative women reported being married or having a regular partner, current partner denotes the husband or regular partner of the woman at enrollment. Sexual partner was defined as a male partner with whom the woman had had sexual intercourse. Lifetime sex partners was defined as the sum of different men with whom a woman had had sexual intercourse. Commercial sex work was defined as exchange of sex for money, and women were considered as FSWs if they had engaged in commercial sex work. History of STD was based on reported diagnosis of any STD (unspecified) or syphilis or gonorrhea, specifically.

Hospital HIV test results for first ANC visits were summarized to evaluate temporal trends. Associations with HIV seropositivity of demographic, obstetric, and behavioral risk factors were determined by χ², χ² for trend, and Fisher exact test (SAS Version 6, SAS Institute Inc, Cary, NC, and Epi Info Version 6, Centers for Disease Control and Prevention [CDC], Atlanta, Ga). Continuous data were compared with the Wilcoxon rank sum test. All factors associated with HIV infection on univariate analysis at P <.1 were included in unconditional, stepwise multivariate logistic regression analysis. Continuous variables were dichotomized for logistic analysis. Two final models were generated to predict women's seropositivity: the first model included women's behavioral factors; the second model also included women's reported behaviors of their current partner. In a separate analysis of seropositive women whose partners' HIV status was known, relative risk for partner serodiscordance was calculated on the basis of women's risk behavior.

The study protocol was approved by the Thailand MOPH Ethical Review Committee and the CDC Institutional Review Board. Voluntary, written, informed consent was obtained from all study participants (except as described).

Nearly complete (>98%) ANC seroprevalence data were available for 1991 through 1996, representing about 20000 antenatal registrants per hospital per year (1 hospital started routine testing in July 1991). Seroprevalence was similar at the hospitals (Table 1). For the 2 hospitals combined, ANC HIV prevalence increased from 1.0% to 2.3% during 1991 to 1996 (P<.001), an increase from about 400 to 800 HIV-positive women among those registering for ANC annually.

A total of 342 HIV-positive and 344 HIV-negative pregnant women were enrolled from November 1992 through March 1994. Most (75%) were enrolled before their third trimester. Of eligible women asked to enroll, 95% of 360 HIV-positive and 97% of 355 HIV-negative women participated. One additional HIV-positive woman did not complete enrollment and one was later excluded because seropositivity could not be confirmed. For HIV-positive women, reasons for participation included counseling support, expedited medical care for study visits, and reimbursement for transportation and medical expenses; reasons for nonparticipation included not wanting to be part of an HIV study, to sign a consent form, or to answer personal questions.

As seen in Table 2, compared with HIV-negative controls, HIV-positive women were younger at time of first pregnancy, had been with their partner for less time, and were more likely from a rural area, to have separated from a prior partner, and to be primigravida and nulliparous. Most women (>95%) were literate, and duration of time in Bangkok was similar. Most were married and nearly all (except 2 HIV-positive women) were living with the current partner. Family income was comparable and consistent with salaries for unskilled work.

Seropositive women had a higher sexual-risk profile than seronegative women (Table 3). The HIV-positive women were younger at first sexual intercourse and more likely to have had more than 1 lifetime sex partner (50% vs 23%), to have been pregnant by more than 1 partner (among multigravida, 53% vs 27%), and to have an STD history (22% vs 5%). Commercial sex work, although uncommon (10% vs 2%), was strongly associated with seropositivity. Injection drug use was rare (1% in HIV-positive women), but having a sex partner who used injection drugs was associated with HIV seropositivity.

More than 1 lifetime sex partner was associated with seropositivity, and for greater analytic precision, partner number was categorized as 1, 2, or more than 2 partners (Table 3). Odds ratios for seropositivity increased from 2.7 with 2 lifetime sex partners to 6.9 with more than 2. Excluding FSWs, odds ratios increased from 2.6 for 2 lifetime partners to 4.5 for more than 2.

Past contraceptive use was similar for HIV-positive and HIV-negative women (Table 3). About two thirds reported use of oral contraceptives, and one fifth, injectable contraception and condoms as contraception. According to women's perceptions, partners of HIV-positive women were more likely to have behavioral risks for HIV (eg, sex with an FSW in past 5 years, STD history, and sex with an FSW while married).

Factors associated with HIV seropositivity on univariate analysis were analyzed by multivariate logistic regression (Table 4). Model 1 includes women's behavioral risk factors; model 2 also includes behavioral risk factors of male partners (reported by the women). In both models, more than 2 lifetime sex partners for the women was most strongly associated with seropositivity, while 2 lifetime sex partners was also an independent risk factor. In model 2, several women's behavioral factors dropped out (STD history, partner who uses injection drugs, and with partner less than 1.5 years), and were replaced by 2 partner behavioral risk factors: STD history and sex with FSWs in past 5 years. Nulliparity and migration from rural area were also statistically significant in both models. Commercial sex work and younger age did not change final models.

Of 307 HIV-positive women (90%) with current partners tested at the study hospitals, 81 partners (26.4%) were HIV negative. Concordant couples were together longer than discordant couples (mean, 2.2 vs 1.9 years, P=.04). Partner serodiscordance was evaluated by women's risk exposure (Table 5). Women reporting more than 1 lifetime sex partner were more likely to have a serodiscordant partner than those reporting 1 (45% vs 8%; relative risk, 5.47). The same association was found when excluding FSWs, with slightly lower relative risk. Female sex workers were more likely than non-FSWs to have a serodiscordant partner. About half (52%) of women did not report any risk apart from sex with current partners. For these women, only 10% of partners were HIV negative vs 44% of partners of women reporting 1 or more risk behaviors.

Seroprevalence of HIV has increased steadily in pregnant women in Bangkok and sex with current or previous partner is the only identified HIV risk for most HIV-positive women in ANCs. This study confirms the common assumption that most Thai women with HIV infection are infected by their husbands, who bring HIV infection into the home after FSW contact.^4,16 However, more women than previously thought have had more than 1 sexual partner, and multiple partners is a strong risk factor for infection in pregnant women in the general population.

The hospital prevalence data described herein are based on routine counseling and testing of about 40000 pregnant women each year. We believe they are a reliable indicator of HIV prevalence for the general antenatal population. During the past 6 years, ANC HIV seroprevalence at the 2 largest maternity services in Bangkok has increased from 1% to 2.3%. Our findings are consistent with national surveillance data showing increases in ANC seroprevalence in most regions; 1995 median prevalence rates were estimated at 2.7% in central Thailand and 2.3% nationwide, double 1992 estimates.^10- 11 Seroprevalence rates for antenatal women and male military recruits are thought to reflect most closely the rates in the general population. In Bangkok, prevalence rates for 21-year-old military recruits were 2.9% in 1992 and 2.6% in 1994.^6- 8 However, our data show that prevalence in antenatal women continues to rise in Bangkok. One explanation, suggested by our study, is that young women may have partners several years older than themselves. Thus, prevalence in antenatal women may increase for several years before reflecting falling prevalence rates in young men.

Few study women could be classified as practicing high-risk behaviors, such as drug use, commercial sex, or having many partners. Although most were infected via sex with partners, more women than commonly thought were likely infected by a previous partner. Demographic and behavioral risk profiles of most HIV-positive antenatal women were modest: young, recently married, nulliparous, and with 1 or 2 previous partners. Despite the lack of high-risk behavior according to traditional criteria, we show that even modestly increased sexual exposure was a risk factor for infection. In the Bangkok setting of seroprevalence rates of 2% to 5% in young men, women who are monogamous but who have had 1 or more previous partners are at demonstrably increased risk for infection. In our multivariate analysis of women's risk factors, having 2 lifetime partners (1 previous partner) increased the odds ratio for HIV infection to almost 3, and having more than 2 increased the odds ratio to nearly 6. This risk persisted even when we tried to account for the current partner's risk behavior.

Although traditional Thai cultural values emphasize premarital abstinence for women, 23% of HIV-negative and 50% of HIV-positive women in our sample reported more than 1 lifetime partner. This is much higher than reported in a national survey done in 1990 (less than 5% reported more than 1 partner),¹⁶ and in a recent MOPH survey (6% to 8% reported more than 1 partner),¹⁸ indicating that the number of sex partners may be commonly underreported.¹⁶ In addition, there may be differences in urban areas with large in-migrations. Traditional sexual patterns in women may be changing rapidly while the HIV epidemic is spreading. Although our study involved HIV-positive antenatal women enrolled in a prospective perinatal transmission study, we believe the study population is representative of the hospital catchment area because enrollment criteria were largely geographic, registration rates were high, and controls were drawn systematically from the clinic.

Elsewhere, in Kigali, Rwanda, in 1988, where HIV prevalence in women of childbearing age was 32%, one third of women reported more than 1 lifetime partner and were at increased infection risk.¹⁹ In Belle Glade, Fla, during 1989 to 1991, where antenatal seroprevalence was 5%, two thirds of women reported more than 1 lifetime partner and also were at increased risk.²⁰

One fourth of the HIV-positive pregnant women in this study had an HIV-negative partner. This continues to be a consistent finding at the study hospitals. Despite many reports on HIV testing of pregnant women worldwide, there have been few reports of HIV test results of their partners. In Thailand, it is commonly assumed that in a given relationship men are the index cases. However, the male serodiscordance data indicate that even in an epidemic largely characterized by transmission from FSWs to male clients to regular partners, it cannot be assumed the male partner is the index case, even when both partners are seropositive. In terms of biological risk, it is not surprising that some male partners in this study were HIV negative. Female-to-male HIV transmission is relatively inefficient.^14,21- 25

Apart from biologic issues, our finding of a 26% male serodiscordance rate for antenatal women with HIV infection raises complex counseling issues and indicates additional stresses of HIV on the family unit.²⁶ At these study hospitals, women with HIV infection and their partners were counseled separately and were offered joint counseling; both concordant and serodiscordant couples were encouraged to discuss results with each other. Although partner testing provides important information to the individual and to the family, a better understanding of the impact of this testing²⁷ and how best to provide counseling support is needed.

Although the 100% condom campaign in Thailand has been remarkably successful in increasing condom use in commercial sex encounters,²⁸ preventing HIV transmission between partners will be more challenging,⁹ especially in young couples desiring children. Sexually active persons, including adolescents, should be educated about HIV risks, encouraged to know their partner's and their own HIV status, and encouraged to avoid high-risk sex and to use condoms as a primary form of STD and birth control. These precautions should not be limited to commercial sex exchanges. Indeed, commercial sex interventions must not inadvertently encourage increased noncommercial casual, unprotected sex.¹⁶ General intervention strategies must acknowledge that young women in rapidly changing cultures such as in Thailand are likely to be more sexually active than reported. Although raising complex social issues, antenatal HIV counseling and testing programs, along with partner counseling and testing, are valuable for patient care, preventive education, and monitoring of the heterosexual HIV epidemic in the general population.