The transmission of HIV-1 infection among adults in sub-Saharan Africa is predominantly heterosexual and is facilitated by the presence of other sexually transmitted diseases (STDs) [1,2]. Although data from the United States and Europe have suggested that the efficacy of male-to-female transmission of HIV is greater than of female-to-male transmission [3-6], evidence from African populations, where infections with other STDs are more prevalent, has been inconclusive. Incidence rates for seroconversion were at least twice as high among female partners of HIV-infected men than male partners of HIV-infected women in Rwanda, Zaire and Martinique, but these differences were not statistically significant [7-9]. A cohort study carried out in Rakai district, Uganda , found no difference in sero- incidence rates between men and women in serologically discordant relationships.
In general, longitudinal studies in Africa reporting on the relative efficacy of male-to-female HIV transmission versus that of female-to-male transmission have been based on small numbers of seroconversions, either because the follow-up period was short or the number of couples studied was small. Although data on male-to-female transmission are more abundant for developed countries, information on female-to-male transmission of HIV-1 infection are still relatively scarce because, unlike in developing countries, many more men are infected than women . In Africa, additional data on the prevalence of serologically discordant couples, and background incidence rates for couples in concordant negative marriages, are important for assessing the public health impact of HIV transmission associated with having an HIV-positive spouse, but have rarely been reported .
We report the HIV incidence rates of 2200 HIV- negative men and women according to the HIV status of their spouses from a 7 year study of a rural general population cohort in Masaka district, Uganda.
The Medical Research Council (MRC) general population cohort is located in rural south-west Uganda and comprises approximately 10000 people residing in 15 neighbouring villages in Masaka district. The inhabitants are mainly subsistence farmers. The population was first surveyed in 1989/1990 and has been re-surveyed annually since then. A detailed description of the cohort design and methods is given elsewhere . Briefly, after each annual census, a medical team interviewed and obtained a blood specimen from all adults (aged 13 years and over) who provided informed consent. Sera were tested for HIV-1 antibodies by two independent enzyme-linked immunosorbent assays and confirmed by Western blot where necessary [12,13]. On average, 65% of adults in this population were bled at any given annual sero-survey. HIV test results were provided to study participants, on request, by experienced counsellors employed by MRC and stationed in the study area. During annual serological surveys, all adults were actively encouraged to visit a counselling station. Condom promotion activities have been ongoing throughout the study area since the first survey round as part of a general community development programme. Apart from the counsellors and MRC statisticians based in Entebbe (100km away), all field survey and research staff remained blind to the HIV status of the study participants.
Marital status information was collected at each annual census and all married women were individually linked to their husbands if resident in the study area. Polygyny is practised but is uncommon, and no woman in this population reported more than one husband at one time. Using data from the first eight survey rounds, all resident adults with one or more spouses censused as resident were classified according to their own HIV sero-status together with that of their spouse(s). Individuals with one or more spouses present were classified as being in a ‚concordant negative‚ marriage if both they and their spouse(s) were HIV-negative, ‚concordant positive‚ if both they and their spouse(s) were HIV-positive, and ‚discordant‚ if they were HIV- negative and (one or more of) their spouse(s) was HIV- positive or if they were HIV-positive and (one or more of) their spouse(s) was HIV-negative. Married individuals who could not be classified as being in concordant or discordant marriages, because either their own or their spouse(s) sero-status was unknown, were described as having ‚incomplete sero-status‚. Where possible, the sero-status of individuals censused as resident but not bled at a particular survey round was determined from the serological status at earlier or later survey rounds: those who were resident with a definitive negative sero-status at a later round were assumed to be HIV-negative, whereas those resident with a definitive HIV-positive test result at an earlier round were assumed to be HIV-positive, at that round.
Interviews conducted at the fourth annual survey round included questions regarding sexual behaviour (condom use and numbers of sexual partners) and reported STDs in the past 12 months (genital ulcer and discharge). Data on Muslim religion, which has previously been found to be associated with a reduced risk of HIV infection in this population [14,15], was available from the baseline, fifth and sixth censuses. These data were used to compare risk factor profiles for adults in concordant negative and discordant marriages resident at the fourth survey round.
All HIV-negative individuals identified with HIV-negative or HIV-positive spouses during the first seven survey rounds who had a further definitive test result at one or more subsequent rounds were eligible for inclusion in the incidence analyses. Incident cases identified up to the eighth serological survey were included in the analysis. Person-years (PY) at risk for seroconversion commenced at the date when an individual was first tested HIV-negative and ceased at their estimated date of seroconversion or date last bled, whichever was earliest. Dates of seroconversion were estimated as midway between the date of the last negative and first positive HIV test result. HIV incidence rates were determined by the method of person-years as in previous incidence analyses . PY at risk and events (seroconversions) were stratified according to sex, age group (13-24, 25-34, 35-44, 45+ years) and the sero-status of the spouse (HIV-negative or HIV-positive). Both age and sero-status of marriage were treated as time-dependent variables. Individuals with HIV-negative spouses identified at a subsequent survey round as having an HIV-positive spouse (either because their spouse seroconverted or because of remarriage) contributed PY at risk as concordant negative until the survey round at which they were first identified with their HIV-positive spouse, after which they contributed PY at risk as discordants.
Incidence rates, and rate ratios (RRs) for incidence in women relative to men, were estimated by the method of maximum likelihood, with adjusted RRs being obtained by the Mantel-Haenszel technique . Approximate 95% confidence intervals (95% CI) for rates and RRs were estimated using the Gaussian approximation to the log-likelihood, and tests of significance were based on the chi-squared distribution . Confidence intervals for age-adjusted RRs were derived from the binomial distribution. All statistical analyses were performed using the STATA computer package, using standard programs for the analysis of cohort studies .
At the baseline survey round 42% of adults censused were married and 91% of these had one or more spouses censused as resident (Table 1). Eighty-six per cent of adults were in marriages in which complete HIV-sero-status information was available. Among those in monogamous marriages with complete sero-status information, 88% were in concordant negative, 5% were in concordant positive and 7% were in discordant marriages. Corresponding percentages for adults in polygamous marriages were broadly similar: 89% concordant negative, 2% concordant positive and 9% discordant. The overall prevalence of being in a discordant marriage was 7% (95% CI 6%-8%).
Using data for all survey rounds, 2442 married adults were identified as resident, of whom 2424 (99%) could be classified according to the sero-status of their marriage at some point. Eighty-nine per cent of adults in concordant negative marriages at baseline were eligible for follow-up and an additional 605 were identified at later survey rounds, yielding 2079 concordant negative adults contributing to the incidence analyses (969 men and 1110 women). A total of 121 HIV-negative adults (58 men and 63 women) entered into follow-up in discordant marriages, of whom 55 were identified at baseline and 66 were identified at later survey rounds.
Table 2 shows marital status at the end of follow-up for all 2200 HIV-negative men and women contributing to the incidence analyses. Adults with HIV-negative spouses were followed for an average of 4.8 years, and those with HIV-positive spouses for 3.6 years. Compared with those with HIV-negative spouses, adults entering into follow-up with HIV-positive spouses were less likely to remain married to the same spouse and were more than twice as likely to become widowed, separated or divorced (Table 2). The latter pattern was particularly notable among men. Compared with women with HIV-positive spouses, significantly fewer men remained married to the same spouse (c2 for difference (df=1)=4.3, P=0.04).
Seventeen men and eight women seroconverted while under follow-up with HIV-negative spouses, and 12 men and 22 women seroconverted while with HIV-positive spouses (Table 3). HIV incidence rates were highest among women who had HIV-positive spouses (105.5 per 1000 PY) and lowest among women with HIV-negative spouses (1.5 per 1000 PY) with rates for men being intermediate (51.7 per 1000 PY for HIV-positive spouse and 3.6 per 1000 PY for HIV-negative spouse) (Table 3). Four of the 22 seroconversions observed among women in discordant marriages occurred among those recorded as having been in a polygamous marriage at one or more rounds. Their incidence rate was 89.6 per 1000 PY, only slightly less than that for all women with HIV-positive spouses.
Twenty-nine adults (20 women and 9 men) were under follow-up for seroconversion when their spouse seroconverted and six of these also went on to seroconvert (data not shown). Using the spouse‚s estimated date of seroconversion as the date of entry into follow-up, the 29 adults contributed 85 PY at risk with an overall incidence rate of 70.6 per 1000 PY (95% CI 31.7-157.2). Three of the six seroconversions had estimated dates of seroconversion within 1 year (119.8 per 1000 PY), two within 2 years (110.2 per 1000 PY) and one within 3 years (60.5 per 1000 PY) of the estimated date of seroconversion of their spouse.
Men with HIV-positive spouses were on average 14.5 times more likely to seroconvert than those with HIV-negative spouses, whereas women with HIV-positive spouses were 69.6 times more likely to seroconvert (Table 4). The corresponding age-adjusted RRs were 11.6 for men and 105.8 for women, both being highly statistically significant (P<0.001). Among men, the RRs for those with HIV-positive spouses varied significantly with age, being highest in the older age groups (c2 for heterogeneity (df=3)=9.95, P=0.02).
Among those with HIV-negative spouses, rates of seroconversion in women were significantly lower than those of men (age-adjusted RR=0.4, P=0.02). In contrast, among those with HIV-positive spouses, women had twice the rate of seroconversion of men, this excess being of borderline statistical significance after adjustment for age (P =0.07). The majority of the excess risk was among women aged 13-24 years, in whom 11 seroconversions were observed in 71.8 PY compared with none in 32 PY for men (P=0.002).
A total of 1392 adults with HIV-negative spouses, and 60 with HIV-positive spouses, were resident and surveyed during the fourth annual round (data not shown). Compared with those with HIV-negative spouses, men with HIV-positive spouses reported significantly higher rates of genital ulcer (13% versus 3%; P=0.07) and genital discharge (20% versus 1%; P=0.002) whereas, among women, those with HIV-positive spouses were significantly less likely to be Muslim (8% versus 29%, P=0.02). The reported rates of condom use were less than 10% among those with HIV-negative spouses whereas none of those with HIV-positive spouses reported ever having used a condom.
Data on Muslim religion were available for 97% of adults contributing to the incidence analyses. Incidence rates for non-Muslims were significantly higher than those of Muslims for men (age-adjusted RR=3.5 95% CI 1.0-11.7) and women (age-adjusted RR=2.7 95% CI 1.0-7.9). Adjustment for Muslim religion in addition to age had little effect on the RRs presented in Table 4.
In this 7-year cohort study of 2200 HIV-negative married men and women residing in rural Masaka district in Uganda, women with HIV-positive spouses experienced approximately twice the incidence rate of men. The excess was most marked among women aged 13-24 years, and incidence rates were significantly higher than those for men in this age group (P=0.002). The estimated twofold excess risk of male-to-female HIV-1 transmission compared with that for female-to-male transmission supports the majority of available data from Europe, the USA and Africa [3-9]. The apparently greater susceptibility of women to infection from heterosexual intercourse has also been observed for other sexually transmitted diseases, such as herpes simplex virus type 2 , and is likely to be explained by biological factors. The larger and more vulnerable mucosal surface of the female genital tract, and the higher intensity of exposure to HIV infection during heterosexual intercourse, have been postulated as reasons for this .
In contrast, among those with HIV-negative spouses, men had twice the incidence rate of women. This is likely to be explained by behavioural factors: men reported higher total numbers of sexual partners and higher rates of STDs than women, suggesting that they are more likely to engage in extramarital sexual partnerships. Similar patterns have been noted elsewhere in rural Uganda and West Africa, where men are more commonly found to acquire HIV infection from outside marriage [10,20].
Cross-sectional baseline data found 7% of adults in serologically discordant marriages, with the number of marriages in which the man was HIV-positive being roughly equal to the number in which the woman was HIV-positive. This finding is compatible with the incidence data: although men are twice as likely to be the source of HIV infection in marriage, once discordant their marriages progress to being concordant positive twice as fast as those in which the woman is the source of infection. These observations illustrate that inferring information from prevalence data about the relative efficacy of male-to-female HIV transmission versus female-to-male transmission can be highly misleading, and underline the importance of incidence data for this purpose.
Very few studies have directly compared HIV incidence rates in adults in serologically discordant marriages with those in concordant negative marriages . Age-adjusted HIV incidence among women with HIV-positive spouses was 105.8 times that of women with HIV-negative spouses, the equivalent figure for men being 11.6 times. The excess rate for women is substantially higher than the 11-fold excess reported in the neighbouring Rakai district of Uganda, but the latter figure was based on few seroconversions . The excess rate for men increased significantly with the age of the man, perhaps reflecting reduced extramarital sexual activity with increasing age. Our findings suggest that married women and older married men in discordant marriages in rural Uganda are at very high risk of acquiring HIV infection from their spouses.
Sixteen per cent of women surveyed at baseline were identified as being in polygamous marriages. Women in such marriages were treated as being at statistically independent risk of sero-conversion. As the risk of HIV transmission is highly related to the degree of infectivity of the HIV-positive spouse, this assumption is not strictly correct. However, incidence rates among women in discordant polygamous marriages were close to those of all women in discordant marriages, whether polygamous or monogamous.
The present incidence analyses confirm previous reports of an increased risk of HIV infection associated with not being Muslim from cross-sectional and case-control study data in this population [14,15]. This may reflect the effect of male circumcision; 100% of Muslim men, but fewer than 10% of non-Muslim men, are circumcised. The findings described above were not materially altered when adjustment was made for Muslim religion in addition to age. There were insufficient numbers of incident events for confounding factors other than religion to warrant further adjustments and their effects could not therefore be quantified.
Whereas over 99% of married adults in the population were classified according to the serological status of their marriage at one or more rounds, the percentage of the study population bled at any given survey round was 65%. This resulted in gaps in serological data during follow-up and some adults with HIV-negative spouses who later had an HIV-positive spouse could have been wrongly classified as remaining in a concordant negative marriage during follow-up. In the analysis, adults identified as having an HIV-positive spouse at a particular survey round remained in this category even if they were censused with a (different) HIV-negative spouse at a later round. Our lack of data on regular sexual partners other than spouses, and of regular partners or spouses residing outside the study area, may also have resulted in some adults being wrongly classified in concordant negative marriages. Whereas the approach used will have resulted in some misclassification, the overall effect would have been to dilute, rather than inflate, any difference in incidence between adults with HIV-negative and HIV-positive spouses.
HIV-negative adults with HIV-positive spouses were more likely to have remarried, and more than twice as likely to be widowed, separated or divorced during follow-up as those with HIV-negative spouses. Although higher rates of widowhood among those in discordant marriages are most likely to be a direct consequence of HIV infection, the higher rates of divorce, separation and remarriage suggest a higher underlying rate of partner change. This pattern was particularly marked among men with HIV-positive spouses who were significantly less likely to remain married than HIV-negative women in discordant marriages.
The probability of heterosexual HIV transmission depends on a range of different factors relating to the infectiousness of the index case and the susceptibility of the exposed, including the clinical stage of HIV infection, the type and extent of sexual exposure, the presence of other STDs, viral load and HIV subtype [4,6,21-25]. Although we were unable to examine the effect of these factors directly, we investigated the effect of a proxy measure for infectivity using time since infection of the index case for 29 adults who had a spouse with an estimated date of sero-conversion. Rates of seroconversion were 116 per 1000 PY during the 2 years after the date of sero-conversion of the spouse. Although based on only five events, these data suggest that individuals are at particularly high risk of infection around the time of sero-conversion of their spouse. The 23 spouses remaining HIV-negative at the end of follow-up warrant further surveillance.
Intensive counselling of discordant couples who know their HIV sero-status has been observed to result in increased rates of condom use and reduced rates of HIV-1 sero-conversion in some African settings [7,8]. Members of the present study population were encouraged to obtain their HIV results through the counselling service provided free of charge within the community . Despite these efforts, less than 10% of all adults make use of counselling . Married couples rarely attend for counselling together and we do not know whether individuals who come for counselling share their test results with their spouse, although they are encouraged to do so. Although we had only limited data on condom use, it is noteworthy that none of the HIV-negative adults in discordant marriages reported using a condom. In many sub-Saharan populations, the desire to reproduce may conflict with the need for preventative measures such as condom use . Moreover, whereas reported condom use is becoming more acceptable with casual partners, it remains less acceptable within marriage in this population. Low rates of condom use, together with the lack of knowledge of sero-status, are likely to have contributed importantly to the high incidence rates among discordant couples in this population.
An important factor that cannot be examined due to the nature of the study design was frequency of sexual contact among partners. Such sensitive information is best obtained through in-depth interview, which is not feasible in the context of large-scale epidemiological studies. Like other heterosexual transmission studies it has been assumed that adults with an HIV-positive spouse who sero-converted were infected by their spouse. The proportion of infections acquired from outside the marital relationships cannot be quantified.
In this rural population cohort of 2200 married adults located in south-west Uganda, the rate of male-to-female HIV transmission was on average twice that of female-to-male transmission. Women and older men with HIV-positive spouses are at very high risk of acquiring HIV infection compared with those with HIV-negative spouses. Married men and women in regular sexual partnerships in this population should continue to be encouraged to attend for HIV counselling with their spouses in order that those at high risk of infection can be identified and advised about appropriate preventative measures.
The authors are grateful for the continued support and co-operation of the population of the study area. They would like to thank the Director of the Uganda Virus Research Institute and the Director of Medical Services, Ministry of Health, Uganda, for their support and for their permission to publish the results of this study. They would also like to thank Maria Quigley for comments on an earlier draft of this paper.
1.N‚Galy B, Ryder RW. Epidemiology of HIV infection in Africa. J Acquired Immune Defic Syndr
2.Mann JN. Heterosexual transmission of HIV: a global review a decade later. Int J STD AIDS
3.Mastro TD, de Vincenzi I. Probabilities of sexual HIV-1 transmission. AIDS
1996, 10 (Suppl. A)
4.European Study Group on Heterosexual Transmission of HIV. Comparison of female to male and male to female transmission of HIV in 563 stable couples. BMJ
5.Padian NS, Shiboski SC, Glass SO, Vittinghoff E. Heterosexual transmission of human immunodeficiency virus (HIV) in northern California: results from a ten-year study. Am J Epidemiol
6.Nicolosi A, Correa Leite ML, Musicco M, et al
. The efficiency of male-to-female and female-to-male sexual transmission of the human immunodeficiency virus: a study of 730 stable couples. Epidemiology
7.Allen S, Tice J, Van de Perre P, et al
. Effect of serotesting with counselling on condom use and seroconversion among HIV discordant couples in Africa. BMJ
8.Kamenga M, Ryder RW, Jingu M, et al
. Evidence of marked sexual behavior change associated with low HIV-1 seroconversion in 149 married couples with discordant HIV-1 serostatus: experience at an HIV counselling centre in Zaire. AIDS
9.Neisson-Vernant C, Quist D, Delaunay C, Sobesky G, de Thore S, Chout R. Seroconversion in heterosexual partners of HIV infected patients in Martinique. International Conference on AIDS
. 1992 [abstract PoC 4163].
10.Serwadda D, Gray RH, Wawer MJ, et al
. The social dynamics of HIV transmission as reflected through discordant couples in rural Uganda. AIDS
11.Mulder DW, Nunn AJ, Wagner HU, Kamali A, Kengeya-Kayondo JF. HIV-1 incidence and HIV-1-associated mortality in a rural Ugandan population cohort. AIDS
12.Nunn AJ, Biryahwaho B, Downing RG, van der Groen G, Ojwiya A, Mulder DW. Algorithms for detecting antibodies to HIV-1: results from a rural Ugandan cohort. AIDS
13.Nunn AJ, Biryahwaho B, Downing RG, Ojwiya A, Mulder DW. Computer-assisted quality assurance in an HIV serology laboratory. Methods Inf Med
14.Malamba SS, Wagner HU, Maude G, et al
. Risk factors for HIV-1 infection in adults in a rural Ugandan community: a case-control study. AIDS
15.Nunn AJ, Kengeya-Kayondo JF, Malamba SS, Seeley JA, Mulder DW. Risk factors for HIV-1 infection in adults in a rural Ugandan community: a population study. AIDS
16.Kengeya-Kayondo JF, Kamali A, Nunn AJ, Ruberantwari A, Wagner HU, Mulder DW. Incidence of HIV-1 infection in adults and socio-demographic characteristics of seroconverters in a rural population in Uganda: 1990-1994. Int J Epidemiol
17.Clayton D, Hills M. Statistical methods in epidemiology.
New York: Oxford University Press; 1993.
18.Clayton D, Hills M. Analysis of follow-up studies. Stata Technical Bulletin
19.Langeland N, Haarr L, Mhalu F. Prevalence of HSV-2 antibodies among STD clinic patients in Tanzania. Int J STD AIDS
20.N‚Gbichi JM, De Cock KM, Batter V, et al
. HIV status of female sex partners of men reactive to HIV-1, HIV-2 or both viruses in Abidjan, Cote d‚Ivoire. AIDS
21.Cameron DW, D‚Costa LJ, Maitha GM, et al
. Female to male transmission of human immunodeficiency virus type 1: risk factors for seroconversion in men. Lancet
22.Kunanusont C, Foy HM, Kreiss JK, et al
. HIV subtypes and male-to-female transmission in Thailand. Lancet
23.Hayes RJ, Schultz KF, Plummer FA. The cofactor effect of genital ulcers in the per-exposure risk of HIV transmission in sub-Saharan Africa. J Trop Med Hyg
24.Ragni MV, Faruki H, Kingsley LA. Heterosexual HIV-1 transmission and viral load in hemophilic patients. J Acquired Immune Defic Syndr Human Retrovirol
25.de Vincenzi I, for the European Study Group on Heterosexual Transmission of HIV. A longitudinal study of human immunodeficiency virus transmission by heterosexual partners. N Engl J Med
26.Seeley J, Wagner U, Mulemwa J, Kengeya-Kayondo J, Mulder D. The development of a community-based HIV/AIDS counselling service in a rural area in Uganda. AIDS Care
27.Ruberantwari A, Mugula F, Bachengana C, Nunn AJ. Who comes for HIV counselling? International Conference on AIDS and STD in Africa.
1995 [abstract WeD860].
28.van Devanter NL, Cleary PD, Moore J, Stuart A, O‚Brein TR. Reproductive behaviours among heterosexual couples with HIV infection.International Conference on AIDS in Africa.
1993 [abstract PO-C22-3150].