|

Factors Associated with Use of the Female Condom in Zimbabwe

Dominique Meekers Kerry Richter

First published online:

Abstract / Summary
CONTEXT

Because women can initiate use of the female condom, the method is believed to make it easier for women to protect themselves against sexually transmitted infections (STIs), including HIV infection. Evidence is lacking about factors associated with trying the female condom and using it consistently.

METHODS

A sample of 1,740 sexually active consumers visiting retail outlets in urban Zimbabwe that sell male or female condoms were surveyed in 1998, one year after a social marketing campaign had begun. Logistic regression analyses were conducted to assess factors associated with ever-use of the female condom and consistent use (always or often) with marital and regular nonmarital partners.

RESULTS

Perceived ease of use and affordability of the product and prior use of the male condom were associated with men's and women's ever-use. Consistent use with marital partners was negatively associated with reporting multiple partners in the past year (odds ratio, 0.3) and positively associated with using the device for pregnancy prevention (5.4) and previously using the male condom (8.0). Consistent use with regular nonmarital partners was associated with numerous variables, including perceived ease of use (1.9) and effectiveness for STI prevention (3.8), low HIV risk perception (2.4), and use for pregnancy (2.9) and STI (2.3) prevention.

CONCLUSIONS

Perceived affordability and ease of use may encourage couples to try the female condom but may not lead to consistent use. Because the reasons for use can vary between marital and nonmarital relationships, the female condom may need to be positioned differently for different target populations.

International Family Planning Perspectives, 2005, 31(1):30–37

Because use of the female condom can be initiated by women, the device provides a way for women to protect themselves from sexually transmitted infections (STIs), including HIV infection.1 When used correctly, the female condom is as effective as the male condom in reducing HIV transmission;2 in addition, it can be inserted hours before intercourse,3 and it is therefore less likely than the male condom to reduce sexual spontaneity.

In Zimbabwe, where the prevalence of HIV infection is high and male promiscuity is common,4 there has been considerable interest in the female condom. More than 30,000 women petitioned the government in the mid-1990s to make female condoms widely available, to give women greater protection against STIs.5 In a context in which it is considered improper for a woman to refuse to have sex with her husband, the female condom may provide an acceptable solution.6 In initial acceptability studies in Zimbabwe, virtually all women liked the female condom, as did most males; most women and men preferred the female condom to the male condom.7 Men liked the product because it does not interrupt the sexual act and it reduces their responsibility for protection.

The factors that affect use and consistency of use in the general population are unclear, however. Several studies suggest that the female condom is most likely to be popular among married women, because it allows them to initiate protection.8 Nevertheless, some studies have shown that use is lowest with regular partners9 and that users of the female condom are likely to have already used male condoms.10 Thus, increased use of the female condom may not necessarily lead to increased levels of protection against HIV infection or other STIs.

Reasons given in acceptability research for discontinuing female-condom use include partner disapproval, difficulty using the product and unplanned pregnancy.11 The device is considerably more expensive than the male condom (which is often available free of charge); thus, cost may be another reason for discontinuation.12 On the other hand, the female condom provides protection against both pregnancy and STIs, and women can initiate its use. Peer support and other social support appear to stimulate use.13 Positive promotion, publicity and support from health care workers are believed to improve correct use.14

Our analysis uses data from an exit survey of 1,753 randomly selected consumers at retail outlets in urban Zimbabwe to examine factors associated with the likelihood that men and women had ever used the female condom and factors associated with consistency of use with marital and regular nonmarital partners. At the time of our survey, a social marketing program promoting and distributing the female condom had been conducted in urban areas of Zimbabwe for a little more than a year.

The Social Marketing Program

In 1997, Population Services International (PSI) started social marketing subsidized female condoms in urban areas of Zimbabwe. The social marketing program was conducted on behalf of the National AIDS Coordination Program and the Zimbabwe National Family Planning Council as part of a larger social marketing program promoting and distributing the male condom.15 The social marketing program is funded by the U.S. Agency for International Development and the British Department for International Development.

Before the female condom was introduced in Zimbabwe, market research had been conducted to obtain information on consumer perceptions of the product. The device was marketed—under the brand name care—as a "contraceptive sheath" instead of a condom to avoid the stigma associated with STI prevention. The image of the product was supported by the slogans "the care contraceptive sheath is for caring couples" and "for women and men who care."16 The product's image was promoted through an extensive mass media campaign that included radio, magazine and newspaper advertisements. In addition, the communication campaign provided information on how to use the device through question-and-answer magazine columns, a weekly 15-minute radio call-in show that allowed consumers to ask questions about the product, and a detailed brochure that was available wherever the product was sold. Given that care was marketed as a contraceptive instead of a disease-prevention product, the campaign targeted women in long-term relationships.

Initially, care was sold only through selected pharmacies and clinics. Distribution has since expanded to other outlets, including large supermarkets, convenience stores, private doctors' offices and clinics. Care is sold at a retail price of US$0.24 (Z$3 each) for a box of two condoms. At the time of our survey, female condoms were also provided free of charge at government hospitals, family planning clinics and other public health institutions in two districts in each of the country's 10 provinces.17

At the onset of the program, PSI forecast sales of 4,000 female condoms per month,18 yet during the first four months of the program, 95,000 condoms were sold.19 Although this novelty use has since decreased, sales remain higher than expected: From July 1997, when the program was launched, to December 1997, 120,720 care condoms were sold; 119,650 were sold in 1998, 165,769 in 1999, 187,049 in 2000, 455,566 in 2001, and 683,700 in 2002.20

METHODS

Aim of the Survey

This analysis uses data from a survey of 1,753 male and female consumers in urban Zimbabwe visiting retail outlets that sell male or female condoms. The sampling methods were designed to obtain a sample representative of the population of consumers (who constitute the target population for social marketing condoms). However, because use of the female condom is low, users of the female condom were oversampled. Our analyses are weighted to correct for this oversampling. We excluded from our analysis 13 respondents who had not had vaginal intercourse in the year before the survey, which reduced our working sample to 1,740.

Sample Selection

The survey covers Harare, Bulawayo, Chitungwiza, Gweru, KweKwe, Mutare, Masvingo and several small towns. The study was also conducted in rural areas; however, because female condoms were sold only in urban areas, we restricted our analysis to residents of urban areas.

Sampling was conducted in two stages. In the first stage, the retail outlets were selected. Four outlet types were included: pharmacies, supermarkets, other traditional outlets (e.g., small stores) and nontraditional outlets. Outlets were selected systematically from a list of all outlets that sell Protector Plus male condoms or the care female condom. Outlets without any female-condom sales in the past three months were excluded. Fieldwork hours for each region (Harare, Bulawayo, other urban) and outlet type were allocated proportional to the sales volume of care and Protector Plus condoms (based on PSI—Zimbabwe 1998 sales records).21

The consumers (potential respondents) were selected during the second stage. A screening questionnaire was used to determine whether consumers had ever used the female condom and, for those who had not, whether they had used a male condom in the past year. All consumers who had ever used the female condom were selected for the female-condom sample. Among the remaining persons screened, one in 10 male-condom users and one in 10 individuals who had never used either the female condom or the male condom were systematically selected for interviewing. After the target sample sizes for the male-condom user and the condom-nonuser groups were reached, recruiting of and interviewing with female-condom users continued.

Same-sex interviewers conducted the interviews, using separate questionnaires for users of the female condom, users of the male condom and condom nonusers. Interviewers worked in pairs and kept a tally of the number of consumers who were not screened (i.e., persons exiting the retail outlet while an interview was in progress). The data are weighted to correct for the oversampling of female-condom users and for the differential sampling probability across outlet types.*

Measures

We use three variables to examine female-condom use. Ever-use of the female condom is measured by a dummy variable that equals one if the respondent reported having used the female condom at least once in the 12 months before the survey, and zero otherwise. For respondents who had used the female condom, we defined two measures of consistency of use: The first of these two measures, which applies to married persons only, is a dummy variable that equals one if respondents reported often or always using the female condom with their spouse, and zero otherwise; the other variable measures the same behavior with regular nonmarital partners and is restricted to consumers who reported having had at least one such partner in the past year. (Participants were also asked about use with casual partners; however, the number of respondents who reported having such a partner was insufficient for statistical analysis.)

Several predictor variables are used. To measure perceptions of the product, we use five dummy variables indicating whether the consumer thinks the female condom is effective for STI and HIV prevention, effective for pregnancy prevention, affordable, easy to use and reusable. As measures of exposure to information about the female condom, six dummy variables indicate whether the respondent had ever heard about the female condom from radio, TV, newspapers or magazines, doctors or clinics, a sexual partner or friends; this information is available only for consumers who ever used the female condom.

Measures of family status and fertility preference are current marital status (married vs. not married) and whether the respondent indicated not wanting to have a child in the next two years. As measures of sexual risk behavior and personal risk perception, our analyses use dichotomous variables indicating whether respondents had had two or more sexual partners in the past 12 months and whether they perceived their risk of HIV infection to be low or nonexistent (vs. moderate, large and do not know).

Two measures of male-condom use are used. The first indicates whether the respondent had used the male condom at least once in the past year. The second measure—used only for persons who had a marital or regular nonmarital partner and who reported using the female condom at least once with any type of partner in the past 12 months—indicates consistency of male-condom use (often or always vs. occasionally or never) in the past 12 months, generally before the respondent had tried the female condom.

The following background characteristics are used as control variables: current age (a continuous variable), level of schooling (secondary or higher education vs. less than a secondary education), residence (Harare vs. other urban) and socioeconomic status (low, medium or high, depending on reported household amenities and possessions). The socioeconomic measure is an index that splits the sample into three roughly equal groups. Further details about the survey methods and sample are described elsewhere.22

Statistical Analysis

We performed logistic regression analyses for men and women separately to test the hypothesis that positive perceptions about the female condom would be associated with an increased likelihood of having used the female condom at least once in the past 12 months.23 Additional logistic regression analyses were conducted to test our hypothesis that perception of the product's effectiveness for pregnancy prevention, affordability and ease of use would be associated with improved consistency of use; separate analyses were conducted to examine consistency of use with marital partners and consistency of use with regular nonmarital partners. Because the female condom was marketed in Zimbabwe as a "contraceptive sheath," it is unclear whether the device was used primarily as a means of preventing pregnancy. Positioning the product as a contraceptive, rather than as a condom, may have allowed women in marital and other regular relationships to use it covertly for protection against HIV and other STIs. It is unclear whether the usual measures of risk (for example, multiple partners) will be associated with higher levels of female-condom use, as monogamous women placed at risk by their partners may be the most likely users.

RESULTS

Participants

After adjustment for the oversampling of female-condom users, 2% of participants in our analysis reported ever having used the female condom, 47% the male condom only and 51% neither condom type (Table 1). Although weighting has a large effect on the percentage of female condom users, it has little effect on the distribution of respondents by background characteristics.

The majority of the consumers surveyed were in their 20s; the restriction of the sample to sexually experienced individuals may partially explain the small proportion of teenagers in the unweighted sample (6%). Males made up more than half the sample (55%); a majority of respondents (60%) were either married or cohabiting. Seventy percent of the sample reported not wanting to have a child in the next two years. Sixty-two percent believed they had a low risk or no risk of HIV infection, and 61% had used a male condom in the past year.

On average, respondents were highly educated: Some 82% had at least a secondary education. This finding was expected, because mean educational and socioeconomic levels tend to be higher for consumers (i.e., those who shop and, therefore, have money) than in the general population.24

Ever-Use

Our first two sets of logistic regression analyses assess the likelihood—for men and women separately—of having tried the female condom (Table 2). For each gender, the first model examines respondents' perceptions concerning the female condom, without adjustment for other characteristics. The second model includes controls for family status, risk behavior and perceived level of risk, prior use of the male condom and background characteristics.

•Males. In the first model, perceived effectiveness for pregnancy prevention, affordability, ease of use and reusability were all positively associated with the odds of female-condom use. In the second model, however, only perceived effectiveness in pregnancy prevention, perceived affordability and perceived ease of use remained significant (odds ratios, 2.9, 4.2 and 6.1, respectively).

The odds of having used the female condom were about 30 times as high among males who had prior experience with the male condom as they were among other males. Findings regarding risk perceptions and behaviors were unexpected. Male respondents who had had two or more sexual partners in the past year were significantly less likely than other men to have tried the female condom (0.5). Likewise, men with a low perceived risk of HIV infection were significantly more likely than others to have tried the female condom (2.2). This finding may reflect more frequent use of the female condom with marital or regular nonmarital partners—the type of use emphasized by the social marketing campaign—than with casual partners. Moreover, consistent users of male and female condoms may be likely to state that they are at low HIV risk because of their consistent use rather than because of their sexual risk behavior. Finally, male consumers of average socioeconomic status were more likely than those of low socioeconomic status to have tried the female condom (2.2).

•Females. For women, perceptions about the female condom's effectiveness for STI and pregnancy prevention, affordability and ease of use were positively associated in the unadjusted model with having tried the female condom. However, after adjustment for background and other characteristics, only two of these variables remained significant: Women who found the female condom affordable and those who considered it easy to use were significantly more likely than other women to have used the device (odds ratios, 3.6 and 3.2, respectively).

Like men, female respondents who had had experience with the male condom had significantly elevated odds of female-condom use (odds ratio, 8.4). Currently married women were less likely than nonmarried, sexually active women to have tried the female condom (0.4); the desire not to have a child in the next two years was not associated with having tried the device. These findings suggest that women may generally try the female condom not for family planning, but rather for STI prevention. Women's number of recent sexual partners and perceived HIV risk level were not associated with use, after adjustment for other variables. Again, this may reflect the fact that care was positioned as a family planning product, which may encourage women in long-term relationships to try the product.

Consistency of Use

In further analyses restricted to participants who had ever used the female condom, we assessed potential associations between selected variables and consistency of use with marital partners (Table 3) and regular nonmarital partners (Table 4). Because the number of female-condom users is relatively small, both analyses were conducted for male and female respondents combined.

•With marital partners. In the first of five models (Table 3), sources of information about the female condom were not associated with consistency of use with marital partners when background characteristics were controlled. Similarly, in the second model, perceptions about the female condom were not associated with consistency of use.

In the third model, the intention to postpone childbearing was not associated with consistency of use, suggesting that respondents relied on other methods for family planning. Married respondents who reported having had two or more partners in the past year (which generally implies that they had had at least one extramarital relationship) were less likely than others to have used the female condom consistently (odds ratio, 0.3). This finding may in part reflect that these persons are much more likely to be consistent users of the male condom, as it loses significance after adjustment for male-condom use (not shown). Moreover, this finding is consistent with the idea that the female condom is particularly attractive to persons who are monogamous, regardless of whether they believe their partner is faithful.25

Model 4 examines the association between each of two reasons for using the female condom—to protect against pregnancy and STIs—and consistency of use (Table 3); the reference category for each dummy variable is all other response options. Those who used the product for pregnancy prevention were more likely than others to have used the female condom consistently (odds ratio, 5.4). In contrast, use of the product for STI prevention was not significantly associated with consistency of use.

Consistent with findings from earlier acceptability studies,26 the results for Model 5 indicate that respondents who had often or always used male condoms with their marital partner before they started using the female condom were significantly more likely than other respondents to have used the female condom consistently (8.0).

•With regular nonmarital partners. In the first of five models examining consistency of use with regular nonmarital partners (Table 4), respondents who had obtained information about the female condom from physicians or clinics were more likely than other respondents to have used the method consistently (odds ratio, 2.6). In the second model, perceived effectiveness for STI/HIV prevention and perceived ease of use were significantly associated with consistent use (3.8 and 1.9, respectively). As we had anticipated, perceived effectiveness for pregnancy prevention was not significantly associated with consistency of use with nonmarital partners.

In the third model, consumers who perceived themselves to have a low risk or no risk of HIV infection were more likely than others to have used the female condom consistently with their regular nonmarital partners (2.4). This finding is consistent with the notion that respondents who are monogamous—but who may have promiscuous partners—are more likely than others to use the female condom. In addition, this finding may indicate that use of the female condom reduces one's perceived risk, or that respondents with high perceived risk are more likely than others to use the male condom,27 thereby reducing their need for consistent female-condom use.

In the fourth model, respondents who had initially tried the female condom to prevent pregnancy or protect against STIs had significantly elevated odds of having used it consistently (2.9 and 2.3, respectively). Interestingly, prior male-condom use with regular nonmarital partners was not associated with consistent use of the female condom with such partners.

DISCUSSION

In the weighted sample in our study, conducted one year after a mass marketing campaign for the female condom had begun in Zimbabwe, 2% of sexually active consumers had used the female condom in the previous year. Our use of a sample of consumers is appropriate considering that consumers comprise the maximum target group for the social marketing program that distributes and promotes the female condom. However, our findings for this population of consumers may not be generalizable to the population at large, since it is known that consumers tend to be wealthier than average. As with all cross-sectional surveys, our study can illustrate factors associated with use of the female condom, but it cannot prove causality.

Acceptability studies have shown that prior experience with the male condom is associated with use of the female condom and may improve consistency of use of the female condom.28 Some studies suggest that persons who have used barrier methods may be more comfortable touching their bodies.29 Our results confirm that perceived affordability and ease of use of the female condom and prior use of the male condom are strong predictors of female-condom use for both male and female consumers.

Among female-condom users, monogamy, the desire to prevent pregnancy and previous male-condom use are all associated with consistent use with marital partners. Perceived ease of use, perceived effectiveness for STI/HIV prevention, low or no perceived HIV risk, and the desire to prevent pregnancy and STIs are associated with consistent use with regular nonmarital partners. These findings suggest that in marital unions, the female condom is largely used as a mechanism for pregnancy prevention, which is consistent with the way the product has been marketed. In nonmarital relationships, however, the female condom appears to be used for both pregnancy and STI prevention.

These findings have important policy implications. First, the results indicate that campaign strategies that succeed at getting people to try the female condom will not necessarily lead to consistent use. Second, because factors associated with female-condom use vary for people in marital and nonmarital relationships, programs may need to position the female condom differently for different target populations.

Footnotes

*The first weight equals 10 for consumers who used the male condom (but not the female condom) and for condom nonusers. For the female- condom users, the corresponding weight equals 273/492, to account for the fact that 219 female-condom users were interviewed after the interviewing and screening of the other two groups had stopped. The second weight equals the ratio of the total number of consumers to the number of screened consumers.

References

1. Gollub E, The female condom: tool for women's empowerment, American Journal of Public Health, 2000, 90(9):1377–1381.

2. AIDS Weekly Plus, U.N. promotes female condom against AIDS in Africa, AIDS Weekly Plus, May 4, 1998, p. 16; Cecil H et al., The female condom: what we have learned thus far, AIDS and Behavior, 1998, 2(3):241–256; Trussell J et al., Comparative contraceptive efficacy of the female condom and other barrier methods, Family Planning Perspectives, 1994, 26(2):66–72; and Young A, The Female Condom: A Review, Geneva: World Health Organization, 1997.

3. AIDS Weekly Plus, UNAIDS promotes female condoms in developing countries, AIDS Weekly Plus, July 28, 1997, pp. 10–11.

4. Young A, 1997, op. cit. (see reference 2); De Zoysa I, Sweat MD and Denison JA, Faithful but fearful: reducing HIV transmission in stable relationships, AIDS, 1996, 10(Suppl. A):S197–S203; and Mbizvo EM et al., HIV prevalence in Zimbabwean women: 54–67% knowledge and perceived risk, International Journal of STD & AIDS, 2003, 14(3):202–207.

5. AIDS Weekly Plus, Zimbabwe women petition the state on female condom, AIDS Weekly Plus, Dec. 23–30, 1996, p. 10; and AIDS Weekly Plus, Female condom launched in Zimbabwe, AIDS Weekly Plus, July 28, 1997, pp. 9–10.

6. AIDS Weekly Plus, 1996, op. cit. (see reference 5).

7. Ray S et al., Acceptability of the female condom in Zimbabwe: positive but male-centered responses, Reproductive Health Matters, 1995, Vol. 3, No. 5, pp. 68–79; Francis-Chizororo M and Natshalaga NR, The female condom: acceptability and perception among rural women in Zimbabwe, African Journal of Reproductive Health, 2003, 7(3):101–116; and Ray S and Maposhere C, Male and female condom use by sex workers in Zimbabwe: acceptability and obstacles, in: Beer M, ed., Beyond Acceptability: Users' Perspectives on Contraception, London: Reproductive Health Matters, 1997.

8. El-Bassel N et al., Acceptability of the female condom among STD clinic patients, AIDS Education and Prevention, 1998, 10(5):465–480; and Sly DF et al., Factors associated with use of the female condom, Family Planning Perspectives, 1997, 29(4):181–184.

9. Ray S and Maposhere C, 1997, op. cit. (see reference 7).

10. El-Bassel N et al., 1998, op. cit. (see reference 8); Agha S, Intention to use the female condom following a mass-marketing campaign in Lusaka, Zambia, American Journal of Public Health, 2001, 91(2):307–310; and Agha S, Patterns of use of the female condom after one year of mass marketing, AIDS Education and Prevention, 2001, 13(1):55–64.

11. Farr G et al., Contraceptive efficacy and acceptability of the female condom, American Journal of Public Health, 1994, 84(12):1960–1964; Hart GJ et al., Women's attitudes to condoms and female-controlled means of protection against HIV and STDs in south-western Uganda, AIDS Care, 1999, 11(6):687–698; Pool R et al., Men's attitudes to condoms and female controlled means of protection against HIV and STDs in south-western Uganda, Culture, Health and Sexuality, 2000, 2(2):197– 211; Ford N and Mathie E, The acceptability and experience of the female condom, Femidom, among family planning clinic attenders, British Journal of Family Planning, 1993, 19:(2):187–192; Ruminjo JK et al., Preliminary comparison of the polyurethane female condom with the latex male condom in Kenya, East African Medical Journal, 1996, 73(2):101–106; and Sapire K, The female condom (Femidom)—a study of user acceptability, South African Medical Journal, 1995, 85(10):1081– 1084.

12. Ray S et al., 1995, op. cit. (see reference 7); and Francis-Chizororo M and Natshalaga NR, 2003, op. cit. (see reference 7).

13. El-Bassel N et al., 1998, op. cit. (see reference 8); and Ankrah EM and Attika S, Adopting the Female Condom in Kenya and Brazil: Perspectives of Women and Men: A Synthesis, Arlington, VA, USA: Family Health International, 1997.

14. Ray S et al., 1995, op. cit. (see reference 7).

15. Population Services International (PSI), Social marketing with "care" in Zimbabwe: female condom sparks early interest, PSI Profile, Washington, DC: PSI, 1998.

16. Tsopotsa T, Marketing the female condom in Zimbabwe: new methods and challenges, in: Report from the Meeting on Changing Communication Strategies for Reproductive Health and Rights,New York: Population Council, 1998.

17. AIDS Weekly Plus, 1997, op. cit. (see reference 5); and Winter J, Female condom sales beat forecast by 10 times, AIDS Analysis Africa, 1997, Vol. 7, No. 5, p. 9.

18. Winter J, 1997, op. cit. (see reference 17); and Winter J, Social marketing of condoms (female, too) gets going (again), Zimbabwe, AIDS Analysis Africa, 1997, Vol. 7, No. 2, p. 6.

19. PSI, 1998, op. cit. (see reference 15).

20. Unpublished data for 1989–2002, PSI, Washington, DC, 2002.

21. Trigg C, 1999 Zimbabwe Condom Consumer Profile Survey, Harare, Zimbabwe: Target Research, 1999.

22. Kerrigan D et al., The Female Condom: Dynamics of Use in Urban Zimbabwe, Population Council, 2000, <http://www.popcouncil.org/ pdfs/horizons/fcz.pdf>, accessed Feb. 1, 2005.

23. Ford N and Mathie E, 1993, op. cit. (see reference 11).

24. Meekers D and Ogada EA, Explaining discrepancies in reproductive health indicators from population-based surveys and exit surveys: a case from Rwanda, Health Policy and Planning, 2001, 16(2):137–143.

25. AIDS Weekly Plus, 1996, op. cit. (see reference 5).

26. Cecil H et al., 1998, op. cit. (see reference 2); and Ray S et al., 1995, op. cit. (see reference 7).

27. Meekers D, Patterns of the use of the female condom in urban Zimbabwe, PSI Research Division Working Papers, 1999, No. 29.

28. Ray S et al., 1995, op. cit. (see reference 7).

29. Cecil H et al., 1998, op. cit. (see reference 2); and Ray S et al., 1995, op. cit. (see reference 7).

Author's Affiliations

Dominique Meekers is professor and chair, Department of International Health and Development, Tulane School of Public Health and Tropical Medicine, New Orleans, LA, USA. Kerry Richter is deputy research director, Population Services International, Washington, DC.

Acknowledgments

This analysis was funded by the Office of Health and Nutrition, Global Bureau, U.S. Agency for International Development (USAID) under cooperative agreement HRN-A-00-00021-00 and by core support from the British Department for International Development (DFID). The opinions expressed herein are those of the authors and do not necessarily reflect the views of USAID or DFID. Data collection was funded by USAID through the Horizons Project and by UNAIDS. The authors thank Guy Stallworthy for comments and suggestions on an earlier version and Cynthia Aragon for editing.

Disclaimer

The views expressed in this publication do not necessarily reflect those of the Guttmacher Institute.