Lagarde, E.; Auvert, B.; Chege, J.; Sukwa, T.; Glynn, J. R.; Weiss, H. A.; Akam, E.; Laourou, M.; Caraël, M.; Buvé, A.the Study Group on the Heterogeneity of HIV Epidemics in African Cities

Introduction

The effectiveness of condoms in reducing transmission of HIV and other sexually transmitted diseases (STDs) has been widely reported ^[1]. For example, a European study of HIV discordant couples found no seroconversion among those using condoms consistently, and found an incidence rate of HIV infections of 4.8 per 100 per year among intermittent users ^[2]. A meta-analysis of condom effectiveness suggests that use of condoms decreases per-contact probability of HIV transmission by 69% ^[3]. However, when limited to consistent use only, per-contact probability of HIV male-to-female transmission is estimated as decreased by 95% ^[4].

In Africa, studies conducted among commercial sex workers demonstrated that condom use reduced incidence of HIV infection ^[5-8]. Repeated population-based surveys conducted in Uganda found a dramatic increase in condom use associated with a decline in HIV seroprevalence among pregnant women attending antenatal clinics ^[9]. There are, however, few published longitudinal studies directly assessing condom impact in general populations. Allen et al. found an association between condom use and a decrease in HIV incidence ^[10]. A study conducted in Malawi found no effect of condom use on STD incidence rates but consistent use was very rare (1%) ^[11]. A high rate of consistent condom use seems to be necessary to impact on population HIV/STD infection levels.

The present study aims to compare levels of condom use in four urban populations of sub-Saharan Africa with different levels of HIV/STD infection, and to assess the association of frequent condom use with risks of infection with HIV and other STDs at an individual level.

Populations

A standardized cross-sectional population survey was conducted in four cities in sub-Saharan Africa. Two cities (Yaoundé, Cameroon and Cotonou, Benin) were selected because of their relatively low and stable level of HIV infection. The two other cities (Ndola, Zambia and Kisumu, Kenya) were chosen because of their high HIV prevalence level.

Households were randomly selected and, within these households, all adults aged 15-49 were asked to participate in the study. After consent, participants were interviewed using a standardized questionnaire on sexual behaviour and socio-demographic characteristics by trained local interviewers. Blood samples were collected for detection of HIV, herpes simplex virus type 2 (HSV-2) and syphilis infection, and samples of urine for detection of chlamydial and Gonorrhoea infections. The detailed design and main results of this study are described elsewhere ^[12].

Methods

The survey questionnaire was designed to collect information on all spousal and non-spousal partners during the past 12 months. For each spousal partner, frequency of condom use in the past 12 months was asked, using the categories Always/Often/Half of the time/Rarely/Never. For each reported non-spousal partner of the past 12 months (up to a maximum of eight), condom use was assessed by two items: use of condom in the last episode of sexual intercourse and overall frequency of condom use with the partner (recorded as Always/Most of the time/Rarely/Never).

Rates of condom use

Rates of condom use were compared between cities using three indicators. First, the proportion of those reporting frequent use of condoms ('Always' or 'Most of the time') with all non-spousal partners of the past 12 months. This indicator has been chosen because: (i) sexual networks are extended beyond households by means of non-spousal partnerships; and (ii) only consistent or frequent use of condoms has been shown to provide an effective protection against STDs and HIV ^[4]. Although consistent use of condoms provides the most effective barrier against HIV/STDs, the low number of those reporting 'always' using condoms prompted us to group categories 'Always' and 'Most of the time' to increase statistical power.

Second, overall condom use in the population was assessed, disregarding the kind of partnership, by computing proportions of those reporting ever having used condoms with their spouse or with any of their non-spousal partners of the past 12 months

Finally, we computed proportions of married respondents reporting frequent use ('Always' or 'Often') of condom with their spouse (or with one of their spouses for polygamous men).

Validity

Consistency of answers was assessed using the two items addressing condom use within non-spousal partnerships by comparing proportions of reported condom use in the last intercourse with reported overall frequency of condom use (Always/Most of the time/Rarely/Never). If both answers were consistent, the proportion using a condom for the last sexual intercourse should increase with increasing frequency of condom use. Answers were considered inconsistent if either (i) condoms were reported to be used 'Always' but the last sexual intercourse was not protected, or (ii) condom use was reported as 'Never' but the last sexual intercourse was protected. In addition, reliability of reports of condom use within married couples was assessed using reports from monogamous couples for whom information was available from both members.

Condom use and sexually transmitted infections

Condom use as a risk factor for sexually transmitted infections (STIs) including HIV infection was assessed by comparing individual infection indicators with reports of condom use with all non-spousal partnerships. STI indicators included: (i) biological tests providing HIV and HSV-2 serological status, markers of active syphilis (Treponema pallidum haemagglutination [TPHA] and rapid plasma region [RPR] titre performed on blood samples) and evidence of infections by Chlamydia trachomatis and Neisseria gonorrhoea (using polymerase chain reaction tests performed on urine samples); and (ii) men's reports of genital pain/discharge and of any genital sores in the 12 months preceding the interview (women were not asked to report any symptoms). Only the symptom histories cover the period corresponding to the 12-month reference period for recorded partnerships. Logistic regression models were fitted to adjust for potential confounding factors and other HIV/STI risk factors. Potential variables considered for inclusion were age, marital status, educational level, occupation, religion, migration, circumcision status before first sex (men), number of lifetime partners, sex with commercial sex workers (men), one-off sex, sex in exchange for money (women), and dry sex practices (women). Variables were retained in the model if they were significantly associated with the outcome (P < 0.05), or if they acted as a confounder for the association between the outcome and condom use by changing the odds ratio (OR) for this association substantially.

Results

Structured interviews were completed by 1021 men and 1095 women in Cotonou, 973 men and 1116 women in Yaoundé, 829 men and 1060 women in Kisumu, and 720 men and 1010 women in Ndola. The proportion who reported one or more non-spousal partners in the past 12 months ranged from 32 to 66% for men and from 13 to 44% for women (Table 1). The rate was significantly higher in Yaoundé than any other city (P < 0.001 for any pair-wise comparison). The proportion of men who reported one or more spousal partners in the past 12 months ranged from 34 to 53% for men and from 43 to 57% for women (Table 1).

Table 1

Rates of condom use

Proportions of those reporting frequent condom use with all non-spousal partners were similar in all cities for men (20-25%), but varied between 10.8 and 24.4% among women (Table 1). There was no evidence that individuals in cities with lower HIV levels (Cotonou and Yaoundé) had higher levels of condom use with non-spousal partners. On the contrary, rates of reported condom use were lowest among women in these cities. The proportion of those who had ever used condoms with any partners showed no association with the level of HIV in the city, and was highest in Yaoundé, where HIV levels are moderate and STI levels are relatively high (P < 0.001 for all pair-wise comparisons). Finally, rates of frequent condom use with spouses were low and very similar from one site to another, ranging from 4.1 to 7.4% among men and from 2.4 to 6.0% among women.

Validity

As expected, the proportion of non-spousal partnerships for which last sexual intercourse was protected increased with reported frequency of condom use in all sites for both men and women (Table 2). All trends were highly significant (P < 0.001). Inconsistent answers were rare, ranging from 0 to 2.9 per 100 partnerships. However, the difference in proportions of those who used condoms in the last contact between the overall categories 'Most of the time' and 'Rarely' is small among men from Cotonou, and among men and women from Kisumu. Condom use data were available from 915 monogamous couples (Table 3). The proportion of members of these couples reporting ever having used a condom with their spouse ranged from 9 to 18%. Concordance as measured by the kappa value was good in Cotonou (63%) and Yaoundé (64%) but relatively low in Kisumu (28%) and Ndola (18%). In every site except Yaoundé, men tended to declare condom use more frequently than their wife, but discordant responses were seen in both directions in all sites.

Table 2

Table 3

Association with STDs

Associations between frequent use of condoms with all non-spousal partners and STIs are shown in Figure 1, adjusted for HIV risk factors and confounders of the association with condom use (see notes in Fig. 1). Among biological indicators of STIs, condom use was associated with a significantly decreased prevalence of chlamydial and HSV-2 infection among women in Yaoundé only [adjusted OR = 0.29, 95% confidence interval (CI) = 0.09-0.96; and adjusted OR = 0.52, 95% CI = 0.28-0.96, respectively]. No other significant associations with biological indicators of STIs were found and Figure 1 does not reveal any particular trends.

Fig. 1

In Cotonou, Kisumu and Ndola, the proportions of men reporting genital pain or discharge during the past 12 months were significantly lower among those reporting frequent condom use. Adjusted ORs were 0.28 (95% CI = 0.09-0.94), 0.34 (95% CI = 0.14-0.83) and 0.33 (95% CI = 0.12-0.90), respectively. Similar associations were found for reported genital sores in Cotonou and Kisumu with adjusted ORs of 0.14 (95% CI = 0.02-1.02) and 0.18 (95% CI = 0.04-0.75), respectively. As some STI indicators may reflect infection that occurred many years ago, we repeated the analysis after restricting data to those aged 15-24 years. The sample size allowed us to perform such an analysis for HSV-2, HIV and syphilis only. We found no significant protective effect of condom use on any of the latter three infections in that subsample. Finally, to investigate the lack of association between condom use and men's reported STI symptoms in Yaoundé, we assessed whether this finding could stem from higher rates of condom use as a consequence of STI symptoms in that city. Among the subsample of men reporting a history of STI symptoms, the proportions of those reporting having done something to avoid passing infection to their partner were 46% (n = 129), 76% (n = 234), 64% (n = 125), 54% (n = 167) in Cotonou, Yaoundé, Kisumu and Ndola, respectively. Among them, 8.5, 11, 6.0, and 8.5%, respectively, reported using condoms for that purpose. This suggests that condoms were more often used by men in Yaoundé than in the other sites as a response to STI symptoms.

Discussion

No correlation between higher rates of condom use and lower aggregate levels of HIV/STI infections was found in our study. At an individual level, the probability of infection by HIV, HSV-2, syphilis, C. trachomatis, N. gonorrhea and Trichomonas vaginalis were generally not decreased by reported frequent use of condoms in non-spousal partnerships, although a decrease was found among women in Yaoundé for chlamydial and HSV-2 infections. The probability of men reporting symptoms of STIs over the past 12 months was significantly decreased by reported frequent use of condoms in non-spousal partnerships in each city except Yaounde.

Available data suggest that actual condom use has remained very low in most of the settings where it has been measured. Most of the studies conducted in Africa have measured the proportion of individuals who report ever having used condom, and this measure is consequently the only one that can be compared from one site to another. In the rural area of Mwanza, Tanzania, a study conducted in 1992 found that the proportion who had ever used condoms was 20% among men and 3% among women ^[13]. In the Arusha region of Tanzania, the equivalent figures were 34% among men and 14% among women ^[14].

In the past 10 years, the awareness of African populations of condoms as a protective device for HIV and other STIs has increased dramatically. It is possible that variations in condom use reflect differences in AIDS awareness in relation to local HIV infection levels and that the levels of use that we can measure now are at least partly a result of the different levels of HIV in the community. Condom use may have been introduced when epidemics were already mature, whereas modelling exercises suggest that prevention efficacy is maximum at the early stages of an epidemic ^[15]. It is therefore necessary to look at historical data on condom use to see whether levels of condom use in the 1990s could explain current observed HIV infection levels. In 1989-1990, a multicentre study was coordinated by the World Health Organization in several developing countries ^[16]. A total of eight countries from sub-Saharan African were involved (Central African Republic, Côte d'Ivoire, Guinea-Bissau, Togo, Burundi, Kenya, Tanzania and Zambia). The proportions of all men who reported having ever used condoms ranged from 14 to 36%. This proportion was consistently lower for women, from 6 to 24%. Levels of condom use measured in 1989-1990 in these African countries were not notably lower among countries with large HIV epidemics today. Conversely, the same study found that condom use among men was high in Thailand (> 40%), and it has been shown ^[17] that condom use played an important part in controlling the HIV epidemic in that country. Another source of historical data on condom use is Demographic and Health Surveys. In the early 1990s, Demographic and Health Surveys were conducted in several countries throughout the world, including three of the four countries of our study. These surveys were conducted in 1991 in Cameroon ^[18], in 1993 in Kenya ^[19] and in 1989 in Zambia ^[20]. No survey was conducted in Benin at the same period. At that time, data on condom use concerned contraception and not protection against STIs. However, these surveys confirm that the level of condom use was very low. In Cameroon, Kenya and Zambia, respectively, 9, 4 and 9% of female respondents reported having ever used condom for contraception. In urban areas of the same respective countries, 2, 1 and 3% of female respondents reported current use of condom for contraception. Here again, condom use levels were not higher in Kenya and Zambia than in Cameroon. In conclusion, it seems that variations in levels of condom use in African populations, including those of our study, all ranged below the necessary threshold to achieve a significant impact on the level of the HIV epidemic, and that the slight variations we observed were not sufficient to modulate overall levels of HIV/STI infections.

In Yaoundé, rates of ever using condoms with any partners were much higher than in any of the other three cities. However, rates of frequent use of condoms with non-spousal partners were not higher in men and women from Yaoundé, and a concomitant study of female sex workers found that 28% of them used a condom during the most recent contact with a client in Yaoundé, compared with 68% in Cotonou, 50% in Kisumu and 28% in Ndola ^[21]. It is therefore unlikely that rates of condom use in Yaoundé are sufficient to explain why HIV levels remained moderate while STI rates and risk behaviour are high.

An important concern is the validity of self-reported condom use that may be subject to recall bias, reluctance to declare risky unprotected sexual intercourse, or to over-reporting of condom use in the context of a survey addressing HIV and sexuality. Consistency of responses among couples was satisfactory in Cotonou and Yaoundé but rather low in Kisumu and Ndola. However, as far as condom use with non-spousal partners is concerned, very low and similar levels of discrepancy were found in the four cities between reported use of a condom during last sexual intercourse with a given partner and reported overall frequency of condom use with the same partner, although only the extreme responses of 'Never' or 'Always' could be checked. Also, the two questions followed each other on the questionnaire, so discrepancies may suggest misunderstanding rather than deliberate misreporting. In addition, one could expect a potential social desirability bias would be much stronger among young people than among older people. The proportion of frequent condom use with all non-spousal partners was, however, not significantly higher among young people except among women from Kisumu (data not shown).

Except for female HSV-2 and chlamydial infections in Yaoundé, biological infection status was not found to be associated at an individual level with lower rates of condom use. Conversely, men's reports of a history of STI symptoms was associated with a lower frequency of condom use with non-spousal partners in three out of the four cities. Questions on symptoms referred to the past 12 months, while infections were measured as point prevalence at time of survey and for some reflect lifetime exposure. This association may have been found because reported STI symptoms and questions on partnerships and condom use covered exactly the same period of time, namely the 12 months preceding the interview. Several authors have argued that condom use is rarely related to a significant reduction in STI rates ^[22,23]. Our study suggests that this lack of association may stem from dissociation between periods of reference for infection and condom use data. Condoms may also be used to prevent onward transmission of STIs and condom use may even be a marker for high-risk behaviour. Figure 1 shows a slight (but not significant) association between condom use and increased HIV prevalence. Finally, it is possible that men who declared condom use may have hesitated subsequently to declare symptoms of STIs because questions on condom use and questions on STI symptoms were separated by only 1 page over the 23 pages of the questionnaire.

Although effectiveness of condoms in protecting against HIV/STIs transmission is well established, we failed to illustrate differential rates of condom use between cities with high level of HIV infection and cities with low levels of HIV infection. Rates of condom use may not have been sufficient to impact on aggregate HIV/STI levels in the four cities even though some individual impact could be found on men's self-reported history of STI symptoms.

References

Section Description

This publication is sponsored by the Joint United Nations Programme on HIV/AIDS (UNAIDS) and the Agence Nationale de Recherches sur le SIDA (ANRS) Paris, France. The Editors of this supplement wish to acknowledge the referees who provided peer review of the manuscripts.

The study was supported by the following organizations: UNAIDS, Geneva, Switzerland; European Commission, Directorate General XII, Brussels, Belgium; Agence Nationale de Recherches sur le SIDA/Ministère français de la coopération, Paris, France; DFID, London UK; The Rockefeller Foundation, New York, USA; SIDACTION, Paris, France; Fonds voor Wetenschappelijk Onderzoek, Brussels, Belgium; Glaxo Wellcome, London, UK; and BADC, Belgium Development Cooperation, Nairobi, Kenya.