Risk factors for HIV-1 transmission from HIV-seropositive male blood donors to their regular female partners in northern Thailand

Objective: To describe risks for HIV transmission from male blood donors to their regular female sex partners in Chiang Mai, Thailand.

Design: Cross-sectional study.

Methods: From March 1992 through September 1995, 405 HIV-seropositive male blood donors (index cases) and their regular female partners were enrolled in the study. Women with risk factors for HIV infection other than sexual contact with the index male were excluded. Couples were interviewed and examined; specimens were collected for laboratory analysis.

Results: Overall, 46% of the 405 women enrolled were HIV-positive. Ninety-eight per cent of male index cases had a history of sex with a female prostitute; 1.5% reported always using condoms with their regular partner. History of sexually transmitted disease (STD) and swollen inguinal lymph nodes in the female partner were associated with an increased risk of HIV infection in the female. History in the female of genital herpes [odds ratio (OR), 3.46; 95% confidence interval (CI), 1.50-8.78], gonorrhea or chlamydia infection (OR, 2.71; 95% CI, 1.39-5.53), and stable relationship of longer than 24 months (OR, 2.28; 95% CI, 1.02-5.09) were associated with an increased risk of HIV infection in the female. Consistent condom use in the past 2 years (OR, 0.10; 95% CI, 0.01-0.79) was associated with a decreased risk of HIV infection in the female.

Conclusions: Married women in northern Thailand who appear otherwise to be at low risk for HIV infection may be exposed to this virus by their husbands. High rates of sex with commercial sex workers among men and low use of condoms within stable relationships may be important factors promoting the transmission of HIV in married couples. Programs to increase the regular use of condoms among married couples could be an important public health intervention to prevent transmission of HIV and other types of STD in northern Thailand.

Introduction

A decade has passed since the first case of AIDS was reported in Thailand. As of 31 March 1997, 59 782 persons in this country had developed AIDS, of whom 16 051 (26.8%) had died [¹]. The epidemic of HIV infection in Thailand has been described as occurring in sequential waves [²]. The first wave occurred among injecting drug users (IDU), the second among female commercial sex workers (CSW) and the third among young men who frequently visited female CSW [^3,4]. Heterosexual transmission is the dominant mode of transmission of HIV infection in Thailand, as in many other countries in Asia, Africa, and the Caribbean.

Studies of HIV-1 isolates from infected persons in Thailand have revealed two genetic subtypes of HIV-1 [^1,2]: subtype E, which has been found primarily among female CSW and sexually transmitted disease (STD) clinic patients throughout the country, and subtype B, which has been found more commonly in the central and southern regions of Thailand, primarily among IDU and their contacts. Subtype E accounts for an estimated 85% of HIV-1 infections in Thailand.

Since June 1989, the Royal Thai Ministry of Public Health has conducted twice-yearly sentinel seroprevalence surveys to estimate HIV-seroprevalence for several groups, including IDU, female CSW, male patients attending STD clinics, women attending antenatal clinics, and blood donors [^5-10]. Also in 1989, a nation-wide program for screening all blood donors for HIV-1 antibody was implemented. In Chiang Mai, the largest city in the north, the prevalence of HIV infection among blood donors in 1993 was 3.34% [¹¹].

Numerous reports have described the HIV epidemic in Thailand, and several cross-sectional and longitudinal studies of various Thai populations at high risk for HIV infection have been published [^12-17]. Because blood donors are screened routinely and infected donors are generally asymptomatic, insights into the risk behavior patterns of individuals who do not know they are seropositive may be obtained by studying married blood donors and their wives. In addition, because the blood bank should notify HIV-infected donors, counsel them about the risks of further transmission, and screen and counsel their regular sexual partners, investigators may have an opportunity to collect additional information.

In this paper we present data on risk factors for HIV transmission from infected male blood donors to their regular sexual partners in northern Thailand.

Methods

Study population

The study population consisted of HIV-seropositive males who donated blood at the Red Cross blood bank in Chiang Mai, at Chiang Mai University Hospital, or at the Lampang Provincial Hospital from 1 March 1989, through 30 September 1995, and their regular female sexual partners. We also included 13 couples recruited from an STD clinic in Chiang Mai who met the study criteria. Previously identified seropositive male blood donors were contacted by mail and asked to return to the blood bank for notification of their test results; those who returned were informed that they were HIV-positive and were counseled. A repeat blood sample was drawn for confirmatory HIV-1 antibody testing [enzyme linked immunosorbent assay (ELISA) and Western blot].

This group of HIV-infected donors was interviewed by trained personnel who used a screening questionnaire to determine eligibility for our study; subjects had to be aged 18 years or older and have at least one regular female sexual partner (in most cases, this was the subject's wife). Only men who gave voluntary informed consent and agreed to bring their regular female partner (or partners) to participate in the study were enrolled. Female partners with other risks for HIV infection (e.g., sex with males other than their current partner, injecting drug use, commercial sex work) were not enrolled in the study. For clarity, we will refer to an index case as a 'husband' and a female sexual partner as a 'wife' throughout this manuscript.

Data collection procedures

Husbands and wives were interviewed separately by trained interviewers of their own sex. The baseline interview included questions about demographic characteristics (medical history, including transfusions and injections, STD, and sexual practices), frequency and types of sexual contact with the regular partner, contraceptive use, and history of smoking and parenteral drug use.

The husbands were asked about age at first intercourse, number and frequency of sexual contacts with different partners, frequency of sex with CSW, homosexual contact, types of sexual practices and condom use. A pictorial guide to STD was used to assist the interviewers in identifying the specific type of STD syndrome, e.g., genital ulcer or urethral discharge, that was reported.

The husbands and wives underwent a complete physical examination (including pelvic examinations for the women), which detected clinical signs and symptoms of AIDS, STD and fungal infections. Laboratory examinations included: Papanicolaou test; cervical/vaginal lavage for human papillomavirus testing; bacterial cultures for gonorrhea, chlamydia, and trichomonas; serologic tests for syphilis; complete blood count; and enumeration of lymphocyte subsets by flow cytometry.

Antibody to HIV-1 was measured by ELISA (Abbott Laboratories, North Chicago, Illinois, USA). Specimens that were repeatedly reactive were tested by the Biotech HIV-1 Western blot (DuPont, Wilmington, Delaware, USA) and were considered HIV-1-seropositive if more than one band from a structural gene product was present (core, polymerase, or envelope) [¹⁸]. A blot with none of these bands was considered negative; blots with bands that did not meet the criteria for positivity were considered equivocal and excluded from statistical analysis. In addition, HIV-1 antibodies from HIV-positive individuals were typed using a peptide enzyme immunoassay to differentiate antibodies to subtypes E and B [¹⁹].

To test for differences in the rate of HIV-positivity of the wives associated with various risk factors, we used χ² tests of proportion (with Yates's correction in 2 × 2 tables) or Fisher's exact test where appropriate. Logistic regression was used to estimate the adjusted odds ratio for HIV-positivity for each risk factor. Backward elimination was used to include only risk factors that were statistically significant (P ≤ 0.05). However, some nonsignificant factors were retained in the logistic regression model if they were believed likely to confound the estimated odds ratios for other risk factors of interest, or if their removal caused parameters for other risk factors to change by 10% or more. We also examined interactions between the CD4 lymphocyte level of the index case and the use of oral contraceptives and medroxyprogesterone acetate (Depo Provera; Upjohn, Kalamazoo, Michigan, USA) as well as interactions between the estimated numbers of sexual contacts in the past 2 years and the other risk factors in the model. Likelihood ratio tests were performed and likelihood ratio-based 95% confidence intervals were computed for the odds ratios. SAS Proc Logistic [²⁰] and Proc Genmod [²¹] were used for multivariate modeling. The Hosmer and Lemeshow method [²²] was used to evaluate goodness of fit of the model to the data.

The study was reviewed and approved by the Institutional Review Board of The Johns Hopkins University School of Hygiene and Public Health, Baltimore, Maryland, USA, and the Medical Ethics Committee of Chiang Mai University, Chiang Mai, Thailand.

Results

Descriptive statistics

During the period between March 1992 and September 1995, 405 couples were enrolled in the study (Table 1). A comparison of the 405 index cases and 452 eligible donors who were not enrolled showed no significant differences by age, education or occupation (data not shown); the groups had similar risk factors for HIV infection. Ninety-eight per cent of the index cases reported a history of sexual contact with a female CSW. More than half these men were repeat blood donors, and about 25% had CD4 lymphocyte counts below 200 × 10⁶ cells/l at baseline. Other abnormal physical findings in this group included: tinea infection, 18 cases; oral thrush, 10 cases; herpes zoster, 20 cases; lymphadenopathy, 55 cases; hepatomegaly, 21 cases; splenomegaly, three cases; anal scarring, two cases; genital herpes, 14 cases; and genital warts, four cases. Six men were circumcised and 12 had a penile glass bead inlay. Only 28 (7%) of the men were aware of their seropositivity prior to notification by the blood bank.

Table 1

Overall, 187 (46%) of the 405 women were found to be HIV-positive at baseline; none reported any risk factor other than sexual contact with their husbands. More than half the couples had been married for at least 24 months (Table 1); most participants were less than 30 years old (median ages: men, 29 years; women, 27 years) and most had 8 years of education or less. About half the wives had their first intercourse before age 20, and almost 70% had more than 100 sexual contacts with their husbands in the past year. Use of hormonal contraceptives was common; 62% of the women had used oral contraceptives; 30% had used medroxyprogesterone acetate; and 17% had used both during the past 2 years. Condom use was rare; using data from the 5-year contraceptive calendar, we found that only six (1.5%) of the women reported that condoms were used consistently during the past 2 years.

Peptide analyses to determine the subtype of HIV were performed on 526 specimens from index cases and their regular female partners: 487 (92.6%) were reactive to subtype E, five (1.0%) to subtype B only, eight (1.5%) reacted to both E and B subtypes, and 26 (4.9%) were nontypable samples.

Univariate analysis

HIV-positive women did not differ from HIV-negative women by age, education, age at first intercourse, or type or frequency of sex with their husbands in the past year. None of these women had ever been an IDU. A history of STD was significantly associated with an increased risk of HIV infection in these women (Table 2). We also found that a history of gonorrhea, chlamydia, or genital herpes was frequently associated with increased risk, but histories of syphilis, trichomoniasis, bacterial vaginosis, and candidiasis were not. Significant positive associations were also seen for ever having swollen inguinal lymph nodes (a symptom of STD) and for having an abnormal vaginal discharge on the baseline physical examination.

Table 2

Women whose relationship had lasted 2 years or more were at significantly higher risk of HIV infection. Women who had used medroxyprogesterone acetate for contraception in the last 2 years were at slightly higher risk for HIV infection, but this difference was not significant. The odds ratio (OR) for women who had consistently used condoms in the past 2 years was 0.23, but this finding was not significant. Neither the type of sexual practices in the past 2 years nor having had more than 100 sexual contacts with an HIV-positive husband in the past year was associated with increased risk.

The only characteristic of the husbands that was significantly associated with an increased risk for HIV infection among their wives was a history of gonorrhea in the past 2 years (Table 2).

Multivariate analysis

This analysis presented adjusted OR and 95% confidence interval (CI) values for the seven variables included in the final logistic regression model (Table 3). Abnormal vaginal discharge was not included because it was collinear with histories of genital herpes and gonorrhea or chlamydia. As neither the interactions between the male CD4 T-lymphocyte level and the use of oral contraceptives or medroxyprogesterone acetate, nor those between sexual contacts and other risk factors were significant, they were not included in the final model.

Table 3

Three variables were found to be significantly associated with an increased risk for HIV infection: a history of genital herpes (OR, 3.46; 95% CI, 1.50-8.78), a history of gonorrhea or chlamydia (OR, 2.71; 95% CI, 1.39-5.53), and a relationship that lasted 2 years or more (OR, 2.28; 95% CI, 1.02-5.09). Consistent condom use in the past 2 years was significantly associated with a decreased risk for HIV infection (OR, 0.10; 95% CI, 0.01-0.79). Although the adjusted OR values for medroxyprogesterone acetate use in the past 2 years and a male CD4 T-lymphocyte count of less than 200 × 10⁶ cells/l were both above 1.00, neither finding was significant. Oral contraceptive use was not associated with an increased risk of HIV infection.

Our model fitted well to the data. The Hosmer and Lemeshow goodness-of-fit χ² square was 2.16 (P = 0.98).

Discussion

This cross-sectional analysis of 405 couples in northern Thailand, in which all the males and almost half the females were infected with HIV, produced several notable findings: the apparent rate of male-to-female transmission was very high; and as seen in several other studies, [^23-24] an increased risk of HIV infection among the women was associated with histories of their having either ulcerative (chancroid or herpes) or nonulcerative (gonorrhea or trichomoniasis) STD, whereas consistent condom use was associated with a decreased risk of HIV transmission to the women. Medroxyprogesterone acetate injection and oral contraceptive use were not associated with an increased risk of HIV-positivity.

Because we excluded all couples in whom the wives had risk factors for HIV infection other than sexual contact with their husbands, it seems likely that the male subjects were the source of HIV infection for all of the women in this study. For the males, our data suggest that CSW were the source of infection in the great majority of cases; 98% had a history of CSW contact, whereas rates of bisexuality and injecting drug use were each 3%. Therefore, we are confident that virtually all the cases of HIV-positivity found among our female subjects represented transmission through a monogamous sexual relationship and not from other sources.

The demographic characteristics and medical histories obtained at screening were similar for the 452 HIV-seropositive male donors who were not enrolled and the 405 men who participated in the study. We believe that there were not important biases associated with study enrollment. In addition, because the men and the women were interviewed separately, we could compare their individual reports on sexual practices to assess their reliability. These data will be reported in a separate manuscript.

Studies of HIV-1 transmission in different populations have revealed considerable variation in the rate of heterosexual transmission. A 1988 report from the USA of persons who had acquired HIV through blood transfusion found that only about 10% of their regular sex partners were infected [²⁵]. Early reports from the Multicenter Hemophilia Cohort Study indicated that infected male hemophiliacs had a similarly low rate of transmission to their wives [²⁶]. However, among infected hemophiliacs in the study who continued to practice unprotected sex, the rate of HIV transmission to their partners increased when the men became immunosuppressed [²⁷]. Various studies have found an HIV prevalence of 30-50% among the sexual partners of HIV-infected IDU [^28-30] but, many such partners either injected drugs themselves or had sex with other partners. Thus, these data may not be a reliable estimate of the rate of sexual transmission.

Important differences in infectivity between subtypes of HIV-1 in these reports and those in our study might help to explain the high rate of apparent heterosexual transmission we found. However, little information is available on the relative infectivity of the various subtypes of HIV-1. Mastro et al. [³¹] used a mathematical model and historical data from Thai military recruits to estimate the rate of heterosexual transmission of subtype E from infected female CSW to young men in Thailand; their estimate was about 10-fold higher than rates estimated from populations in the USA and Europe who were presumably infected with subtype B. However, this high rate could have been influenced by a high prevalence of STD in this setting or a high rate of early HIV infection among the CSW which may be associated with high viral burden in the genital secretions. More recently, Kunanusont et al. [³²] reported a significantly higher infection rate among the regular female sexual partners of persons in Bangkok, Thailand infected with subtype E than among those infected with subtype B. However, adequate adjustment for all the possible confounding variables, including degree of immunosuppression, was not possible in this study. Finally, the report by Soto-Ramirez et al. that HIV-1 subtype E is more infectious than subtype B for Langerhans cells in vitro [³³] was interpreted by those investigators as an indication that subtype E, which is responsible for large heterosexual epidemics in Asia, is transmitted more efficiently through sexual activity than is subtype B. In North America and Europe, subtype B has been primarily responsible for HIV infection among male homosexuals and IDU. Clearly, further investigation of infectivity rates for HIV subtypes and their implications for the epidemiology of HIV/AIDS is warranted.

Although we identified male blood donors who were infected with either HIV-1 subtype E or B in this study, the number of men infected with subtype B was too few to reliably compare the rates of transmission of these viral subtypes. Individuals who were dually reactive to the peptide ELISA probably had infections with only one subtype and were cross-reactive on the ELISA. We base this opinion on the fact that all of the dually reactive samples had ELISA optical density readings twofold higher with subtype E peptides than with subtype B peptides. However, they did not satisfy the arbitrary criterion of more than threefold-greater reactivity to a single subtype that is required to conclude that the infection is due to a specific subtype. This type of result has been reported frequently by investigators using this method at the Centers for Disease Control and Prevention and elsewhere but infection with two subtypes of HIV is rare, though it has been reported [³⁴].

Our study suggests a high frequency of sexual transmission of subtype E virus. Several characteristics of the population we analyzed contribute to this and inferences relating to infectivity of subtype E should be made with caution. A high rate of STD is associated with a greater likelihood of HIV transmission in both men and women [^9,16,31,35]. That many men reported having had an STD is not surprising, given the history of unprotected commercial sex in this group. However, the women's STD rate could be considered unexpected, as we excluded women who had sex with other males. Among our female subjects, we found that a history of STD was significantly associated with HIV-seropositivity; 35% of HIV-positive women had such a history versus 21% of those who were HIV-negative (P < 0.01). The low rate of condom use in the study population must also be considered in evaluating our results. More generally, we found that most couples had not been routinely practicing safe sex or following other procedures to prevent HIV transmission prior to notification of the male's seropositivity by study personnel. In addition, we found that consistent condom use was effective in preventing HIV transmission. This finding is similar to results previously reported by the European Group Study [³⁶].

Studies of the association between oral contraceptives and HIV have reported inconsistent findings. A positive association between HIV infection and oral contraceptive use has been reported by investigators in both cross-sectional and longitudinal studies of female CSW and female STD clinic attendees in Kenya [^36,37]. They suggested that increased cervical ectopy among women using oral contraceptives might allow easier passage of HIV through the columnar epithelium of the cervix [³⁷]; a similar phenomenon has been reported for Chlamydia trachomatis infection [³⁸]. In a subsequent investigation these researchers found a higher HIV infection rate in women who had cervical ectopy on pelvic examination [³⁹]. A study of female contacts of bisexual men attending several STD clinics in Brazil also found a significant association between oral contraceptive use and HIV prevalence [⁴⁰], but other investigators have shown no such association [^36,41,42]. One experimental study reported a strong relationship between progesterone and simian immunodeficiency virus infection in Rhesus macaques [⁴³]. We did not find a significant increase in HIV-positivity in women who had a history of medroxyprogesterone acetate use; the modest increase seen is of interest and deserves further evaluation.

Preliminary findings using a subset of the data on HIV infectivity and contraceptive method, also show no significant association between medroxyprogesterone acetate use and HIV transmission rate [⁴⁴]. A more detailed analysis of our data that uses contraceptive calendars and adjusts for the timing and frequency of exposures to HIV, male CD4 counts, female STD history, and the use of multiple contraceptives in this population is in progress.

In conclusion, much remains to be learned about the risks that women and men face for heterosexual transmission of HIV. As over 75% of HIV infections world-wide are believed to be transmitted heterosexually [⁴⁵], far more research should be undertaken to increase our understanding of both the epidemiology and the biology of this process. Careful epidemiological studies of couples in whom the risk factors for HIV infection can be well defined should yield important new information on the sexual transmission of HIV infection.

Acknowledgements

The authors acknowledge the collaboration of the Research Institute for Health Sciences, Chiang Mai University, and Lampang Provincial Hospital, Thailand. We thank K. Wongworapat, P. Vannareumol, S. Jeeyapant, P. Wongphan, S. Sennam, O. Chaiyarassamee, A. Sontiratana, B. Siriroj, S. Sriduangkaew, P. Khantataewee, C. Chanta, D. Kunangern, A. Wannarat, P. Dabbhasuta, P. Doncommul, S. Saeue and S. Niyomthai for their contributions and P.M. Gargiullo for assistance with statistical analysis