Given the AIDS pandemic across Africa, health professionals and researchers are trying to understand the social context in which people are likely to take precautions to protect themselves against HIV infection. Because HIV has spread across Africa mostly through sexual contact, research has focused on condom use—particularly in southern and eastern Africa, where as many as one in four adults are HIV-positive.1 Although the prevalence of HIV infection among adults in the West African nation of Côte d'Ivoire had reached approximately 10% by 1999,2 and only about 1% of Ivoirian adults reported using condoms,3 the determinants of condom use in the country have not been investigated extensively.

Descriptive statistics from the 1994 Demographic and Health Survey (DHS) conducted in Côte d'Ivoire reveal that when asked how HIV is contracted, 92% of men and 80% of women mentioned at least one means of sexual transmission. However, when asked about their inclination to use condoms, only 23% of Ivoirian men and 7% of Ivoirian women indicated that they had ever used a condom.4 Thus, information about links between knowledge and condom use is important for the development of appropriate interventions.


Although family planning and AIDS awareness programs in Africa promote the use of condoms, studies have revealed many obstacles to use. Negative attitudes about condoms are a major barrier. For many Africans, condoms suggest unfaithfulness and mistrust.5 Because of such beliefs, personal and emotional concerns often supersede the choice to use condoms. And even if a man has a positive attitude about condoms, they may not be readily available to him. For example, a study in South Africa found that condom availability varied greatly according to the type of distributor, and could be hindered by short business hours and the attitudes of providers.6

Another barrier to condom use is that African men control much of the decision-making regarding sexual encounters. For example, a cohort of men from Zimbabwe expressed the belief that women need their partners' permission to use contraceptives,7 and a sample of men from Uganda conveyed concerns about retaining control over their female partner.8 As a result of this male-female power differential, women may find themselves in situations that increase their risk for sexually transmitted infections (STIs), including HIV infection, despite knowledge they may have about how to protect themselves.

Because of patriarchy and a history of polygamy, extramarital relationships are common in many African societies.9 The risk of such relationships is increased by the frequent migration of African men to seek seasonal or long-term employment. Meanwhile, the stigma associated with condom use inhibits persons in extramarital sexual relationships from protecting themselves and their partners. In a study of extramarital relations that focused mostly on countries in Sub-Saharan Africa, regular condom use during sexual encounters was uncommon.10 Moreover, Mbizvo and Bassett have reported that the risk and incidence of STIs in Sub-Saharan Africa increased as labor migration of husbands became necessary and, consequently, the number of male and female sexual partners increased.11 In addition, in a study of male truck drivers in Nigeria, 72% of married men reported having multiple extramarital sexual partners, and 60% reported being unwilling to use condoms because they did not like using them or were unaware of the risks of unprotected sex.12

Extramarital liaisons include exchanges of sex for money or goods and may or may not involve partners who are sex workers. Although many prostitutes and their clients know that condoms can guard against STIs,13 condom use in commercial sex encounters is relatively rare. In one South African study, only 29% of urban prostitutes and 14% of suburban prostitutes reported condom use in sexual encounters with their clients.14 Clients of sex workers may use various tactics—from financial incentives to violence—to avoid using condoms in solicited sexual encounters. Female sex workers, who lack the leverage to convince clients to use condoms, may have little choice but to put themselves and their partners at risk for HIV and other STIs. Similarly, in situations in which men provide gifts or financial support to young women in exchange for sex, young women may compromise on condom use to receive material goods.15


For the most part, Ivoirian men's attitudes toward condom use and the social context of sexual relationships reflect patterns seen throughout Africa. Among women, the condom is the most widely known method of contraception;16 however, women have little power to advocate its use.

One study reported that married men were more likely to use condoms with extramarital sexual partners than with their wives.17 Another study, however, found that men often seek out prostitutes who do not insist on condom use.18 With almost 70% of sex workers carrying HIV,19 the consequences of this lack of precaution are staggering: One study estimated that the odds of acquiring HIV were 1.7-2.7 times as high among men who had had unprotected sexual contact with a female sex worker as among those who had not.20 Not surprisingly, in a study of 490 HIV-infected men in Abidjan, Côte d'Ivoire, approximately half of the participants' wives were infected with HIV.21

Because of its relatively strong economy, especially in the agricultural and industrial sectors, Côte d'Ivoire has seen substantial increases in the number of migrant workers from neighboring West African nations. As a result, many women come to Côte d'Ivoire to earn money as commercial sex workers.22 In cities such as Abidjan, where approximately 40% of the total population is foreign-born,23 the mix of customs and cultures challenges the capabilities of AIDS awareness programs, which may not be suited to effectively convey their messages in multiple languages and dialects. Furthermore, the large number of migrant workers creates conditions for widespread dissemination of disease. The country's HIV epidemic is especially concentrated in the main agricultural and industrial centers—Abidjan, Bouake and Daloa—largely as a consequence of transmission involving transient populations.24


No previous research in Côte d'Ivoire has explored the effects of Ivoirians' knowledge of AIDS on their likelihood of using condoms. Elsewhere in Africa, research on links between AIDS knowledge and condom use has produced mixed results. An examination of AIDS knowledge and sexual behavior among secondary school and college students in Tanzania found that participants with "good" knowledge of how HIV is transmitted were unlikely to use condoms despite their awareness of the increased risk for HIV infection from such behavior.25 Conversely, among adults in a cross-sectional study in rural Senegal, persons with "good" knowledge* of AIDS were more likely than persons without "good" knowledge to use condoms.26 Data from longitudinal studies provide insight into the effect of acquisition of AIDS knowledge on condom use over time. For example, a study of gold miners in South Africa showed that as the male participants became more knowledgeable about AIDS, their use of condoms increased with all types of sexual partners.27 Moreover, a study in Senegal of female sex workers and their male clients showed that after two years of intervention, clients' AIDS knowledge and consistent condom use significantly increased, as did sex workers' efforts to use condoms.28

Why would AIDS knowledge and condom use be linked? From a theoretical standpoint, knowledge can be used as a form of self-empowerment. According to the AIDS Risk Reduction Model, knowledge of AIDS is necessary to recognize one's behavior as high risk and then to take action to change that behavior.29 Accuracy of knowledge may therefore influence the degree to which persons acknowledge their behavior as risky and the types of precautions they take to reduce their risk. For example, persons who believe they are at risk for HIV infection and know that condom use reduces that risk can adopt measures to protect themselves, whereas persons who think that HIV is spread by mosquitoes will not know of appropriate measures.

This article contributes to the literature on AIDS knowledge and condom use, focusing on the data on sexually active respondents in the 1994 Côte d'Ivoire DHS. The current study started with the hypothesis that having more-accurate knowledge about how HIV is transmitted would be associated with a greater likelihood of having used a condom at last sex.


The 1994 Côte d'Ivoire DHS collected basic information on households as well as demographic data and information on contraceptive practices and knowledge from nationally representative samples of women and men. A supplement included in the survey also inquired about knowledge of AIDS, protective sexual behavior and condom use.

To compare the data on men and women and to obtain relevant information on television and radio ownership, data from the DHS women's and men's individual surveys were merged with those from the DHS household survey. Because the current study examines AIDS knowledge and condom use at last sex, the analysis was restricted to Ivoirians who responded that they had heard of AIDS and had had sexual intercourse in the two months before the survey. The sample for the current analysis comprised 5,653 participants (4,429 women and 1,224 men).

In previous research on AIDS knowledge,30 the degree of AIDS knowledge was evaluated by awarding a respondent a point for each accurate response given about AIDS. However, that method, which measures accuracy on the basis of the quantity of correct responses, does not account for incorrect responses that could potentially inhibit respondents from effectively protecting themselves and others from contracting HIV. To better examine the varying degrees of AIDS knowledge, the independent variable in this study was derived from two open-ended survey questions that measured the accuracy of knowledge respondents had about AIDS: 1) "How can one become infected with AIDS?" and 2) "How can one avoid contracting AIDS?"31 Individual-level data were coded according to the accuracy of the participants' responses: all accurate, partly accurate or all inaccurate.

To examine the influence of sources of AIDS information, an independent variable was created on the basis of study participants' responses to the question "Where did you hear about AIDS?"32 The responses fell into seven main categories: television or radio; friends, family or neighbors; newspapers or magazines; posters or public campaigns; health clinics or health workers; spouse or partner; and "other" sources.

Because respondents were permitted to identify multiple sources of information, the outcome data included numerous combinations of these sources. The majority of these combinations fell into five categories, leaving a residual group of "other" combinations that was used as the reference category (Table 1 [pdf]).

Table 2 [pdf]shows the distribution of respondents by each of the social and demographic characteristics that served as independent variables. Women constituted almost four-fifths of the total sample. The majority of women (72%) and men (60%) in the sample were married; married persons are more likely than their unmarried counterparts to have a steady sexual partner. Approximately three-quarters of respondents were Muslim or Christian, and about half lived in rural areas. One-third of men and one-half of women had had no formal schooling. Two-thirds of men and women lived in a household with a radio, and almost one-third owned a television.

The dependent variable in the logistic regressions measures condom use at last sex. Age was included as an independent variable because previous research has shown that age plays a significant role in condom use;33 this variable is categorized according to five-year intervals. Education was measured by comparing respondents who had had no education with those who had had primary or secondary schooling; the influence of this characteristic on condom use previously has been shown to significantly predict condom use.34 The largest cities in Côte d'Ivoire have had the highest prevalence of HIV infection and have probably seen the most efforts to prevent the spread of HIV;35 therefore, regional comparisons are included in the model.§ The model also examines the effects of television and radio ownership; because AIDS-awareness programs are often broadcast on television and radio, exposure to these media has proven to be an important factor.36 Finally, marital status is coded as a binary variable; previous research suggests that married couples are less inclined than other couples to use condoms.37

Because the characteristics vital to condom use may differ between men and women, a combined analysis for men and women was compared with models run separately according to sex to determine which of these two options would be most appropriate for the logistic analyses. According to log likelihood ratios, the separated models displayed a better fit to the data (1205.2 and 1834.9, -2 log likelihood) than did the combined model (3347.8, -2 log likelihood). Thus, the association between accuracy of AIDS knowledge and use of condoms was analyzed separately for each sex. Further analyses were performed to control for social and demographic characteristics, education, radio and television ownership and sources of AIDS information. All analyses were conducted by using SAS software, version 8.0.


Table 3 [pdf]shows results from the analyses of men's condom use at last sex. In the initial analysis, no data are reported on the group of men who had no accurate knowledge about AIDS, because none of the respondents in this category had used a condom at last intercourse. The odds of having used a condom did not differ significantly by accuracy of information about AIDS. The addition of social and demographic variables to the model had little effect on the relationship between accuracy of AIDS information and condom use. However, the odds of use were significantly lower among married men than among single men (odds ratio, 0.4). The odds decreased as age increased and were significant for men aged 35 or older (0.3-0.5). Living in the urban savannah or in Abidjan decreased the odds of condom use by 44% and 18%, respectively, whereas being Catholic increased the odds by 44%. Only the effects of age and marital status retained their significance in the final model, however.

Although the effect of the accuracy of men's AIDS knowledge on condom use was virtually unchanged by the addition of education to the analysis, men's level of education, by itself, strongly predicted use. Men with a secondary or higher education were significantly more likely than men without formal schooling to have used a condom at last sex (odds ratio, 2.0); the effect of this characteristic remained highly significant even after all other variables had been added to the analysis (1.7).

In the final model, which included the source of AIDS knowledge and ownership of a television or radio, Ivoirian men reporting friends, family or neighbors as their exclusive source of AIDS information had significantly decreased odds of condom use at last intercourse (0.5).

For women, the characteristics predicting condom use were similar to those for men (Table 4 [pdf]). In the initial, unadjusted analysis, however, Ivoirian women with inaccurate knowledge were significantly less likely than those with accurate knowledge to have used a condom at last sexual intercourse (odds ratio, 0.3); nevertheless, after adjustment for social and demographic factors, education and sources of knowledge, the difference was no longer significant (0.6). The variables of age, marital status, religious affiliation and region of residence showed effects similar to those in the men's analyses. For example, the odds of condom use decreased as age increased: Compared with female respondents aged 15-19, women aged 25-50 had significantly reduced odds of condom use (0.2-0.6). Being Catholic or living in the urban savannah each increased the female participants' odds of condom use by 50%, but the effect lost significance once all other variables were added to the analysis. Condom use was strongly related to marital status: The odds of use were significantly lower among married women than among single women (0.2), an effect that remained significant in the final model.

As in the men's analysis, adding education to the model further explained the effect of accuracy of AIDS knowledge on women's odds of using a condom. Compared with women who had had no formal education, women who had had primary schooling were more likely to use a condom (odds ratio of 1.6), but this difference was insignificant after adjustment for sources of AIDS knowledge (1.4). Having had secondary or higher schooling significantly increased the odds of condom use (2.9), even after adjustment for sources of knowledge (2.2).

The inclusion of variables for sources of AIDS information also helped to explain the relationship of the accuracy of women's AIDS knowledge to their use of condoms, as the effect of accuracy of knowledge lost significance. Like men, women who had learned of AIDS only from friends, family or neighbors were significantly less likely to have used condoms at last sex (odds ratio, 0.3). Having heard of AIDS exclusively from television or radio decreased the odds of condom use to 0.6; having learned about AIDS from print and broadcast media and family, friends or neighbors decreased the odds to 0.4.


Of the two previous studies on AIDS knowledge and condom use that included men and women,38 neither analyzed data according to sex. In the current study, accuracy of AIDS information seemed to play a mixed role in predicting condom use. For men, the weakness of the association found between correct information about AIDS and condom use at last sexual intercourse may have resulted from the lack of a comparison group of male condom users without knowledge of AIDS. Among women, the connection between accuracy of AIDS knowledge and condom use was statistically significant; however, after controlling for all other variables, the relationship lost significance.

Secondary or higher education appears to have been an important characteristic in determining condom use. With the ability to better comprehend the complex information about and implications of both HIV infection and family planning, formally educated persons may take more precautions.39 Marriage strongly predicted condom use for men and women. Married men's greater likelihood of condom use compared with that of married women in Côte d'Ivoire may reflect men's greater tendency to engage in extramarital affairs; Ivoirian men have been shown to use condoms more often with their casual partners than with their spouses.40 The finding that older adults had reduced odds of condom use may also be associated with marital status, because older adults are more likely to be in a stable relationship—a situation in which condoms are not frequently used.

The source from which participants learned about AIDS also significantly predicted condom use. Men and women who had heard of AIDS through word of mouth only (i.e., friends, family or neighbors) had decreased odds of using condoms. These sources might not emphasize taking precautionary measures to prevent transmission of HIV. In addition, these social networks may emphasize procreation, thus discouraging use of condoms. Adult Ivoirian women, most of whom are married, may have no control over contraceptive decisions or may not want to use condoms because of a desire to become pregnant. Further research is needed to explore these hypotheses.

At the time this research was done, Sara L. Zellner was a researcher at the Population Research Center, University of Texas, Austin, Texas, USA. She is now an independent researcher residing in Washington, DC.