Condom use by married or cohabiting couples in populations with high rates of HIV infection has become a significant public health issue. The HIV pandemic has prompted massive efforts to promote condom use, but these efforts have targeted high-risk behaviors and focused on premarital and extramarital sexual relationships. A fair degree of success in increasing condom use in these circumstances has been achieved.1 By comparison, the needs of married and cohabiting couples have been neglected despite the fact that, in generalized HIV epidemics, many infections occur within marital and cohabiting unions because of either prior infection by one partner or infidelity.2

Research in diverse settings has shown that condoms are often regarded as more appropriate for nonmarital than marital relationships.3 In countries with generalized HIV epidemics, only 8% of married contraceptive users report condom use, and this rate has shown no increase over the last 20 years.4 Most studies have also found widespread resistance to the use of condoms in stable, long-term relationships because of their association with lack of trust and illicit sex.5 A study in Zaire found that women were not able to insist on condom use—even when they suspected their husbands of having multiple partners—because of their fear of an angry reaction.6 It is therefore not surprising that condoms are one of the least frequently used contraceptive methods by married couples in African countries.7 Moreover, many couples throughout Sub-Saharan Africa want to have large families; thus, use of condoms—particularly if it is prolonged—may be considered unacceptable.8

This study assesses the level of condom use by married or cohabiting couples and identifies factors that promote or impede use in two sites in the province of KwaZulu-Natal, where about one-third of antenatal clinic attendees are HIV-positive.9 The explanatory or predictor factors were selected on the basis of a conceptual framework that drew upon social learning theory and the theory of reasoned action.10 Knowledge of condoms and belief in their preventive efficacy are necessary, though insufficient, preconditions for use. Other factors likely to influence condom use are personal acquaintance with AIDS sufferers, perceived severity of the consequences of HIV infection, perceived risk of HIV infection from a partner and perceived self-efficacy to prevent infection. Attitudes toward condom use and partner discussion of condoms also influence use.11

Though most studies have focused on men and women separately, it is becoming increasingly evident that the dynamics of sexual negotiation cannot be accurately understood unless researchers consider the attitudes and behavior of both partners.12 The collection of cognitive, attitudinal and behavioral data for partners in marital or cohabiting relationships allows a more complete understanding of a couple's contraceptive use.13


Sample Design and Data Collection

The data for this analysis were derived from a household survey conducted in 1999–2000 in a rural area 80 km southwest of Durban and an urban area 15 km north of Durban. The populations of both areas are predominantly black, with isiZulu being the main home language. In both sites, 23 households were randomly selected from each of 20 enumeration areas. In each household, one index adult was randomly selected using a Kish grid. Those individuals in marital or cohabiting unions, and their partners, were then interviewed. The participation rates for index respondents were 70% and 87% for men and women, respectively, and 84% and 94% for their male and female partners, resulting in a final sample of 238 couples (122 urban and 116 rural). The main reason for the lower participation of men was difficulty in contacting them at home, a problem exacerbated by the migration of men out of KwaZulu-Natal to find work. Of the participating couples, 60% were married and 40% were not married but cohabiting.* Polygamous marriages do not appear to be very common. In the household survey, only one man reported having more than one wife.

Respondents were interviewed independently, and where possible concurrently, by specially trained field staff of the same sex. The sample was not designed to be representative of KwaZulu-Natal—and there was oversampling in the urban area—so for the logistic regression analysis the data were weighted to conform to the province's rural–urban proportions recorded in the 1996 population census. The study was granted ethical clearance from the University of KwaZulu-Natal in Durban.

Definition of Variables

For the survey question on frequency of condom use with the named spouse or cohabiting partner, possible responses were "always," "occasionally," "at the start of the relationship only" and "never." Although only consistent use offers effective protection against HIV infection, because of the low number reporting "always" using condoms, the categories "always" and "occasionally" were combined for the analysis. The purpose of condom use (pregnancy and/or disease prevention) was not elicited.

Some of the predictors are self-explanatory, but the derivation of others requires explanation. The salience of HIV/AIDS was defined as "high" if the respondent had had a family member or friend who had suffered or died from AIDS, or had attended the funeral of someone who had died of AIDS. Attitude toward use of condoms within marriage (understood as including cohabiting relationships) was derived from agreement or disagreement with three statements: "It is acceptable for a married couple to use condoms"; "It is acceptable for a married woman to ask her husband to use condoms"; and "To protect themselves against HIV and STIs, a married couple can use condoms every time they have sex." A summary scale was created from these responses and was divided into three categories: favorable, neutral and unfavorable.

The measure of perceived self-efficacy for HIV prevention was derived from agreement or disagreement with two statements: "There is not much use in trying to prevent HIV; if you are going to get it, you will get it eventually no matter how much you try"; and "If a wife/husband gets HIV or STIs from outside the marriage, there is nothing the husband/wife can do to avoid getting infected him/herself." Respondents who disagreed with both statements were defined as possessing high self-efficacy.

Differences in condom knowledge, attitudes and use between men and women across three subgroups were analyzed using chi-square tests. Relationships between explanatory factors and condom use were assessed using logistic regression.


Female respondents ranged in age from 15 to 49 years (mean, 33) and male respondents from 20 to 55 years (mean, 38). Most respondents had some formal education, with 39% of men and 38% of women having completed secondary schooling.

A low proportion of respondents reported that they consistently or occasionally used condoms with their spouse or partner (15% of men and 18% of women). Of these, only 5 men and 12 women said that they always used condoms (a group we will refer to as consistent users), with the rest of the respondents indicating occasional use.

Consistency of reporting within couples was high: In 77% of couples, both partners said they were not using condoms, while in 10%, both said they were using condoms always or occasionally. Because most of the women reporting condom use were also using another method of contraception (94%), typically the injectable, there is little doubt that condoms were being used by women primarily for disease prevention.

Table 1 (page 26) gives a profile of condom use and associated knowledge and attitudes among three subgroups of the study population: urban couples in which the woman had a secondary or higher education; rural couples in which the woman did not attain secondary schooling; and an intermediate group of urban couples in which the woman had less than secondary schooling and rural couples in which the woman had secondary or higher schooling. A sharp gradient in consistent or occasional condom use is evident across the three subgroups: Reported use was significantly higher in the urban, more educated subgroup than in the intermediate and the rural, less educated subgroups (34%, 18% and 11%, respectively, for women, and 29%, 16% and 8% for men).

Belief in AIDS fatality was high among both men and women, and HIV/AIDS was defined as salient for about 40% of all respondents. Knowledge of condoms was almost universal: About 90% had heard of the method and knew a potential source of supply. Condoms were also widely recognized as a highly effective method of preventing HIV infection. In general, women were more likely than men to have a positive attitude toward condom use in marital or cohabiting relationships (47% vs. 29%), but only about 45% of either sex had ever discussed the method with their partner.

A higher proportion of women than of men felt at risk of HIV infection from their partner (57% vs. 22%). Not surprisingly, perceived risk of HIV infection was strongly and positively associated with the woman's belief that her partner was unfaithful (not shown). Information on HIV status was not obtained. Given the proportions of men and women who reported ever being tested for HIV (23% and 35%, respectively), it is reasonable to assume that most did not know their status. Fewer than one-third of the respondents were classified as possessing high self-efficacy regarding HIV prevention.

For women, the most pronounced differences between subgroups were in attitude toward condom use and discussion of condoms with partners, which paralleled the pattern in reported use. Lower proportions of rural, less educated women had a positive attitude toward condom use than did urban, more educated women (35% vs. 58%), and they reported dramatically less discussion of condoms (28% vs. 80%). Rural women were also less likely to report high self-efficacy than were their urban counterparts (22% vs. 38%).

For men, the most pronounced difference was in condom discussion with partners: 31% in the rural, less educated subgroup, compared with 76% in the urban, more educated subgroup. The two subgroups also differed significantly in the proportion who believed that AIDS is fatal (95% rural vs. 76% urban). This surprising difference may reflect greater awareness of drug therapy for AIDS among urban than rural men. The proportion who perceived a risk of HIV infection from their partner was higher among rural than urban men (25% vs. 10%).

Predictors of consistent or occasional condom use, as reported by women, are presented in Table 2. In exploratory analysis, a woman's belief that AIDS is fatal and the salience of HIV/AIDS were found to be unrelated to use and were therefore dropped. Because it is difficult to demonstrate a causal relationship between condom use and either a woman's attitude toward condoms or whether she discusses them with her partner, these factors were also excluded. However, the man's attitude toward condoms was retained in the analysis.

In the unadjusted analysis, all factors except the woman's age and the woman's perceived HIV prevention self-efficacy were significantly related to reported condom use. Odds of consistent or occasional condom use were elevated among women who had a secondary or higher education (odds ratio, 4.4), those who perceived a risk of HIV infection from their partner (4.2) and those who believed in condom efficacy (5.5), as well as among urban women (3.2). Those whose partners had a positive attitude toward condom use also had higher odds of reporting their use (4.3), as did those married to men with higher education levels (3.8).

In the adjusted model, the strengths of these associations were attenuated, although their directions remained the same (Table 2). Only one statistically significant factor remained: Women who felt themselves to be at risk of HIV infection from their partner were more likely than other women to report condom use (odds ratio, 4.0).

To assess the robustness of these results, the multivariate analysis was repeated using men's reports of condom use as the outcome. The man's attitude toward condom use was replaced by the woman's attitude, and a woman's perceived self-efficacy was replaced by the corresponding measure for her partner. In addition, the man's perceived risk of HIV infection from his partner and his belief in condom efficacy were added to the model. The adjusted results (not shown) indicate that urban residence and level of education for men were significant predictors of condom use. An important result of this analysis is that the woman's perceived risk of HIV infection retained a strong and significant association with condom use (odds ratio, 5.1). This finding clearly demonstrates that a woman's perceived risk of HIV infection was significantly related to a couple's condom use, regardless of which partner's report was included as the measure of such use.


Any generalization from the results of this study must be cautious. Though the two study areas were selected to be typical of low-income areas of KwaZulu-Natal, the sample was not designed to be statistically representative. The number of couples interviewed was also relatively small. In addition, the results should not be interpreted as showing causal relationships. Offsetting these limitations, however, is the study's strength in having reports and perspectives of both partners in these marital and cohabiting relationships.

In some ways these findings are consistent with those of many other studies conducted in eastern and southern Africa: Condom use within marriage is uncommon, attitudes toward condom use remain rather negative (particularly among men) and women feel more vulnerable to HIV infection from their partners than do men. Yet this study's central message is more positive and challenges widely held views that resistance to condoms within marriage is immutable. Although no direct estimate of trends in condom use is possible, the results suggest that the infiltration of condom use in marital and cohabiting relationships in KwaZulu-Natal is relatively recent. The sharp gradient in behavior between urban and rural couples and across educational levels is typical of what is found in the early phase of behavioral change. Such change usually starts in the more privileged social strata and subsequently diffuses into the broader society.14 An optimistic outlook is that condom use will continue to rise in KwaZulu-Natal as less educated rural couples follow the lead set by their more advantaged counterparts.

The merit of this interpretation rests on the assumption that reports of condom use are valid. Skepticism is entirely justified, and the measurement of condom use in this study is not immune from possible error. However, the consistency between partners' reports is impressive, and the pattern of the statistically significant associations is coherent and convincing. The major limitation is the imprecision of the term "occasional use" and its lumping with "consistent use" in the analysis. Given the lack of evidence that irregular condom use protects against HIV infection, the implications of these results for HIV transmission are uncertain.15

Another challenge to conventional wisdom concerns the influence of women on condom use by their partners. The woman's perceived risk of infection from her partner emerges as the strongest predictor of use. Interpretation of this relationship is not straightforward: It could reflect open acknowledgment of the man's infidelity or even his HIV serostatus. However, the most plausible explanation is that the women in these relationships are able to translate their concerns into protective behavior. Although the negative attitude of many men to condom use within marriage or cohabiting unions no doubt serves as a barrier, it appears that the woman's perceived risk of HIV infection can override the man's objections. This interpretation contrasts sharply with the conclusions from many other studies, which have found that women are generally powerless to negotiate condom use with their husbands.16

Prevention programs have an important role to play in creating greater awareness of the risk of HIV infection within marital and cohabiting relationships. Indeed, the advent of antiretrovirals and the expansion and integration of voluntary counseling and testing into health delivery systems are likely to lead to a more accurate assessment of the risk and transmission of HIV infection. The inevitable identification of more and more HIV-discordant couples will further reinforce the need for greater condom use within stable relationships.

The results of this study confirm that some married and cohabiting couples are willing to use condoms at least some of the time if they perceive the risk of HIV infection. The promotion of condoms within such relationships has been a neglected component in HIV prevention programs, largely because of the widely held belief that resistance against condom use is too strong to overcome. By showing that modification of sexual behavior in response to HIV risk has actually begun in KwaZulu-Natal—an area with a high incidence of HIV—we hope that prevention programs will be encouraged to broaden their focus and strengthen efforts to meet the needs of married and cohabiting couples.