In South Africa, women who know someone who has received a diagnosis of HIV infection or who died of AIDS are no more likely than other women to have used a condom at last sex, according to an analysis of nationally representative data from 1998.1 However, the odds of use with a casual partner are slightly more than twice as high as those for use with a spouse (adjusted odds ratio, 2.3). Knowing that condoms can prevent transmission of a sexually transmitted infection (STI) also significantly increases a woman's odds of use (1.9).
To identify factors that affect condom use, researchers examined data collected from more than 11,000 women aged 15-49 in the 1998 South Africa Demographic and Health Survey (DHS). Measures related to sexual behavior or AIDS-related knowledge included knowing someone who had received a diagnosis of HIV infection or had died of AIDS, the type of partner at most recent sexual intercourse, knowing that condom use can prevent STI transmission, and knowing that a person can have HIV infection and yet look healthy and be symptom-free.
The analysis also considered three social variables--the participants' employment status, their response to a survey question asking whether they alone decided how to spend their earnings, and whether they had experienced or been threatened with violence in the previous year. The demographic variables assessed were age, level of education, rural versus urban residence and race.
Nearly eight in 10 survey respondents were black (78%); the rest were biracial (10%), white (8%), or Asian or Indian (4% combined). Six in 10 were aged 20-39, and a similar proportion lived in urban areas. Seven in 10 had a secondary or higher education. Thirty-one percent of women were currently working; 70% decided for themselves how to spend their earnings. Nine percent reported having been mistreated in the previous year.
Seventeen percent of respondents had personally known someone who had received a diagnosis of HIV infection. Eighty-five percent knew that condoms are effective in preventing HIV transmission; however, only 30% knew that an HIV-infected person could appear healthy.
Among women who were sexually experienced, 50% said their most recent sexual partner was their spouse; 43% a regular, nonmarital partner; and 7% a casual partner. Overall, 12% had used a condom at their most recent intercourse--7% with their husband, 16% with another type of regular partner and 23% with a casual partner.
Variables associated with condom use in the bivariate analysis at p<.001 were examined in a multivariate logistic regression analysis. According to this criterion, all of the variables were included in the multivariate analysis except for the social characteristics and the variable of knowing that HIV-infected persons can appear healthy. Despite being significant only at p=.05, having known someone who received a diagnosis of HIV infection was included in the logistic regression model because it was considered the key explanatory variable. After excluding the data of women who had reported not using condoms because they wanted to become pregnant, the researchers analyzed the data of 8,014 sexually experienced women who had provided responses about contraceptive use.
Condom use was associated in the multivariate analysis with partner type and AIDS knowledge, but not with having known someone who had received a diagnosis of HIV infection or died of AIDS. Compared with women who had last had sex with their husband, those who had done so with a regular, nonmarital partner or a casual partner had approximately twice the odds of condom use (odds ratio, 1.8-2.3). Understanding that condom use can protect against HIV infection nearly doubled the odds of use (1.9).
Women who were white, biracial, Asian or Indian had lower odds of condom use than did black women (0.4). (A subsequent calculation of estimated probabilities revealed that much of this difference could be explained by more frequent condom use during marital sex by black women than by nonblack women.) An interaction term combining race and partner type showed that nonblack women who had last had sex with a regular, nonmarital partner had higher odds of condom use than did black women whose most recent partner was their husband (2.5).
Women younger than 30 had significantly higher odds of condom use at last sex than did those aged 40-49 (odds ratio, 1.5-2.0). The odds increased with level of schooling and were significantly elevated for all women who had completed at least a primary education (2.0-7.3). Living in a rural area was negatively associated with condom use (0.6).
The researchers note that their analysis is limited by the absence of DHS data on consistency of condom use, as well as on self-perceived risk of HIV infection. In addition, they say, a few hundred women who indicated that they had never been married reported that their most recent sexual intercourse had been with their husband; they point out, however, that the number of women reporting these contradictory responses was relatively small and probably did not affect the overall findings.
According to the investigators, the significant association between condom use and knowledge of the protective effect of condoms suggests that "women do have some agency in determining whether condoms are used." Although they believe their data show that certain types of women have heard and heeded messages about condom use for HIV prevention, they conclude that "efforts are needed to ensure that prevention campaigns, including condom promotion, reach rural, less-educated women and their partners."--C. Coren
1. Camlin CS and Chimbwete CE, Does knowing someone with AIDS affect condom use? an analysis from South Africa, AIDS Education and Prevention, 2003, 15(3):231-244.