The polyurethane female condom is approximately as effective a contraceptive as other barrier methods [1,2] It is impermeable to pathogens that cause sexually transmitted infections (STIs) , and two studies have reported substantially lower STI incidence in female condom users [4,5]. Numerous acceptability studies have had encouraging results , and adding the female condom as a preventive option has resulted in less unprotected intercourse among Thai sex workers  and in a Zambian STI clinic . To date, however, these efforts have generally been small and short-term, or have involved intense clinical involvement and follow-up.
We sought to conduct a large-scale intervention that would incorporate widespread female condom distribution to many women at risk of STI, employing a replicable information and motivation campaign. A cluster-randomized community intervention trial was designed to test the impact of female condom introduction at agricultural units in Kenya. Standard means of STI prevention have limited value for many women in Africa, as monogamy and male condom use are beyond their control. The consistency of male condom use remains low in most groups there, even among those at high risk of contracting STI. Providing female-controlled, or at least female-inserted, methods that protect against STI may offer a more practical solution.
Details of intervention trial design appear elsewhere . In summary, the study measured and compared the prevalence of sexually transmitted infections (STI) in women with and without access to female condoms, accounting for clustering effects. We hypothesized that the proportion of protected coital acts would be doubled at the intervention sites offering both male and female condoms, causing a reduction in STI prevalence. We identified six matched pairs of communities from tea, coffee and flower plantations with primary health care clinics and large numbers of female employees. One community within each pair received the female condom intervention, a program that relied on large and small group meetings, video presentations, puppetry and other folk media, printed materials, individual counseling, and female and male condom availability for all adults. Control communities received male condoms only, and a similar prevention program excluding information on female condoms. The program was intended to be replicable and affordable , if successful.
The prevention program operated through clinic service providers, outreach workers and plantation managers, all of whom underwent varying degrees of pre-study training. Before the study began, 5-day training sessions were organized for clinic staff, covering the study protocol and questionnaires, the informed consent process, and syndromic management of STIs. Staff was taught the proper placement and use of the female condom using anatomical models, the benefits of condom use, and ways to motivate use of the device. A comprehensive list of problems and the anticipated range of user and partner reactions were discussed, with role-playing to model client concerns. Three-day training courses instructed outreach workers at the sites on the same topics, excluding STI management.
The plantations were located near Kericho, in the Nandi Hills, near Naivasha, and near Thika (Fig. 1). They were matched on three criteria: same agricultural product; same geographical area; and no more than five-fold difference in number of permanent employees. One site per pair was randomly assigned to the intervention condition and the other to the control condition using computer-generated random numbers. The sites were distant enough to limit social or sexual contacts between the intervention and control participants.
Using randomly sorted employee rosters, we recruited permanent female employees, 18–50 years of age, who were sexually active, not pregnant and did not desire pregnancy in the coming year. At the baseline visit, we screened each woman, administered informed consent, received written agreement from each woman to participate, and tested her for gonorrhea, chlamydia and trichomoniasis. Infected women were treated and retained in the cohort. Using pelvic models to demonstrate the female condom, and penis models for male condoms, study participants were trained in proper condom use at baseline. They were encouraged to return to the clinic at any time to discuss problems incurred with condom use.
We deemed 160 women per site as a feasible number to enroll in two weeks. With 12 sites and 1920 women, the study would have 80% power to detect a prevalence difference of 10% between intervention and control sites at each time point, with baseline STI prevalence of 20%, one-sided alpha=0.05, 25% loss to follow-up, and an intracluster correlation coefficient (ICC) of 0.03 (a measure of variability within clusters compared with total variability, which can reduce study power) [10,11].
Plantation management provided numbers of residents of reproductive age at each site, which we used as denominators for condom distribution rates. The total at the intervention sites was 8972 adults, and the total at the control sites was 17 050, so that our study cohort comprised less than 10% of the adult residential population. The prevention program and condom distribution was applied at the community level, whereas the interview and STI data collection was limited to the sample cohort.
The primary outcome was cervical and vaginal infections occurring to study participants. We diagnosed chlamydial and gonococcal infections by ligase chain reaction (LCx; Abbott Laboratories, Abbott Park, Illinois, USA)  using urine specimens. We diagnosed vaginal trichomoniasis with the InPouchTM TV test system (Biomed Diagnostics, San Jose, California, USA)  using self-administered swabs . Specimens were transported to the University of Nairobi Department of Medical Microbiology within 24 h of collection, with urine specimens in cool boxes and InPouch cultures at ambient temperature. These three infections were combined into a single STI outcome for analysis purposes. Infected participants were treated with the antibiotics included in the Ministry of Health STI kits.
The prevalence of each specific STI was calculated at each time point, and in the aggregate. Although the 6- and 12-month prevalences approximate the cumulative incidence, the conservative term prevalence is used as we did not have accurate data on incident infections between study visits. We calculated and compared distributions of various features of women in the intervention and control arms at baseline and during follow-up. We calculated condom distribution rates by dividing numbers of male and female condoms distributed each month by the estimated numbers of resident adults of reproductive age. We calculated the ICC for aggregate STI at baseline, and for each of the STI outcomes, at 6 and 12 months .
We used a marginal approach to longitudinal data for the statistical analysis: the average of an outcome at each time point in each study arm was estimated and its variance calculated with the increase in variance due to clustering at the levels of site and person considered a nuisance variable. The matched pairs of sites were considered strata and sites were primary sampling units in estimating variances and testing the effect of interventions on the outcomes of interest.
Estimates and statistical tests of hypotheses on the effect of female condom introduction on the STI outcomes were made using SUDAAN's logistic, descriptive and categorical data analysis procedures . A logistic model was fitted for each of the STI variables, including study arm, follow-up visit, and visit-by-study arm interaction as explanatory variables. We found no interaction effects and so the results from main-effects models with study arm and visit as explanatory variables are presented. In addition, estimates of changes in prevalence from baseline to each of the follow-up periods were compared between intervention and control arms. We calculated t-statistics to test whether the changes in prevalence across visits differed statistically from zero, and used the t-distribution with corresponding degrees of freedom and P-values.
To estimate the association of categorical variables like frequency of male condom or female condom use with study arm, study visit, and covariates like marital status or age groups, we used the Grizzle–Stamer–Koch procedure . The proportions of study participants who reported certain frequency of product use were modeled as a linear function of the covariates of interest. We report adjusted Wald F-statistics to test for the significance of the association of the covariate with product use. Similarly, we calculated adjusted Cochran–Mantel–Haenszel test statistics when the association of covariates such as marital status and age group was of interest while controlling for study arm and visit.
The multivariable logistic models for each STI were first fitted using women treated as the primary sampling unit, which allowed testing of the significance of the contributions of the covariates. For the final model, in which we specified no more than six covariates, we used clustering by site to adjust the variance of estimates.
A total of 3031 women were screened of whom 1931 were eligible and volunteered to participate in the study. Of this enrolled population, 1929 (64%) had valid baseline data and comprise the analysis population. A toal of 1752 women (91% of analysis population) completed at least one scheduled follow-up visit and had STI test results and therefore appear in follow-up analyses; 1580 women (82%) completed both follow-up study visits with STI data. (STI testing was not carried out during unscheduled follow-up visits, which were excluded from the current analysis.) In total, 969 women in the analysis population were in the intervention arm: 887 had at least one valid follow-up visit, 881 completed 6-month follow-up and 797 completed 12-month follow-up. A total of 960 women were in the Control arm: 865 had at least one valid follow-up visit, 856 completed the 6-month follow-up and 798 completed the 12-month follow-up.
Features of the cohort
The characteristics of the study participants are detailed elsewhere . Two-thirds of women screened were eligible and volunteered. Women at the intervention and control sites were similar, with mean age of 33.1 years and a mean of 3.6 living children. The majority was married (65% at intervention sites and 55% at control sites), although not all were living with their spouses. About 90% had completed primary school or less. More than half the women (58%) used family planning, with the most popular method being the injectable hormone depot-medroxyprogesterone acetate. Only 22% of women had ever used male condoms. Less than 10% reported more than one sex partner in the past 3 months.
STI prevalence by study arm and visit
STI prevalence at the sites ranged from 13.4 to 31.2% at baseline . The corresponding ranges were 8.0 to 27.4% at 6 months, and 13.3 to 24.3% at 12 months. No crude association was found between any specific or aggregate STI prevalence and study arm during follow-up (Table 1). The prevalence of each STI decreased between baseline and the 6-month follow-up visit in both intervention and control groups, and then stabilized from the 6-month to the 12-month follow-up. The lack of an intervention effect, and the substantial effect of study visit, were confirmed in simple logistic models for each STI outcome (Table 2).
The change in STI prevalence from baseline to 6-month follow-up did not differ by study arm except for gonorrhea. In the control arm, the baseline level of over 3% gonorrhea prevalence decreased to about 1% at 6 months. In the intervention arm, however, the decrease was from 2% at baseline to 1.6% at 6 months. Changes in prevalence from 6 to 12 months were small; trichomoniasis prevalence increased in that interval.
The ICC for STI prevalence at baseline was low, indicating sufficient study power to detect the expected 10% prevalence decrease at intervention sites . During follow-up, with an average of 144 study participants per site at 6 months and with ICC ranging from 0.0033 for trichomoniasis to 0.0113 for chlamydia (Table 3), the design effects for detecting differences in STI prevalence ranged from 1.5 to 2.6. That is, the variance of the prevalence estimates increased by more than half, relative to variance from a simple random sample. For month 12, we found negative ICCs, which would lead to decreased variance.
Male condom distribution and use
The average number of condoms distributed via clinic and outreach in the control sites steadily increased from about 300 to over 500 devices per 100 adults per month by study's end (i.e. five per person per month). At the intervention sites, the average number of male condoms distributed varied more over time and averaged about 400 devices per 100 adults per month.
The baseline prevalence of any male condom use was about 10% among study participants in the control arm and over 13% among those in the intervention arm. Any use of male condoms increased to over 60% in both study arms at 6 months, consistent use was 15–20% at 6 months, and over 20% at 12 months. The proportion of consistent male condom users in the control arm was higher than in the Intervention arm at both visits, 23 versus 14% at 6 months and 24 versus 22% at 12 months. Yet the women at the control sites were also more likely to report not using male condoms in the preceding months at both visits (43 versus 29% at 12 months at intervention sites).
Female condom distribution and use
Female condom distribution also increased during follow-up, most smoothly in the second half of the study, reaching slightly over 200 devices per 100 adults per month. At 6 months, 39% of women at intervention sites reported never using the female condom, and 11% reported consistent use. At 12 months, the corresponding percentages were 58 and 7%. The main reason given by women for not using female condoms consistently was partner objection (29% at 6 months and 30% at 12 months). The second most common reason was ‘didn't need protection/mutually faithful with partner’ (8 and 9%, respectively).
At 6 months, 10% of women at intervention sites reported not using male or female condoms in the preceding 3 months; 15% reported not using either condom at 12 months. Total male and female condom distribution at the intervention and control sites were roughly similar within matched pairs.
Other selected behaviors relevant to STI risk
Various aspects of sexual behavior of study participants did not differ greatly by study arm or visit. The prevalence of contraceptive use remained at close to 60% in both study arms throughout follow-up. About 3–5% of women at control sites and 2–3% at intervention sites reported not having sex since the last visit (P < 0.05). Close to 40% of all women reported vaginal douching at 6- and 12-month visits. Self-reported STI or STI treatment in the period between visits did not differ meaningfully between study arms.
Self-reported symptoms of STIs
At each clinic visit, we collected self-reported data on STI symptoms as indicators of infections for which we did not test. Few women reported relevant symptoms, and there were no meaningful differences in the rates by study arm or study visit (Table 4).
Logistic models of STI
Full logistic models including all main effects hypothesized to affect STI acquisition showed few significant associations, so simpler models that included only variables with significant effects were fitted in the full model. The coefficient for study arm did not differ from zero in any of the logistic models, confirming the null effect of female condom introduction seen in crude analysis. These models all revealed the protective effect of current (at the time of follow-up visit) antibiotic use against each of the specific STIs, and the aggregated STI outcome. For trichomoniasis and for aggregated STI, being married was associated with lower risk. Self-reported consistent use of either male or female condoms was not associated with protection against infection.
Many researchers and public health advocates view the female condom as a method that can enhance the ability of women to protect themselves from STI . The need for risk reduction is especially great in the developing world, where many women cannot control their sexual exposures, and the STI/HIV burden is high. The female condom can be a means of improving sexual communications, equalizing imbalances in sexual decision-making, and empowering women [18,19]. Unfortunately, this promise failed to materialize in our intervention: we found no difference in the prevalence of the three STIs during follow-up at the intervention versus the control sites. The null effect was clear in crude and multivariable analyses. STI prevalence was similar even within matched pairs that most favored the intervention site in terms of numbers of educational events and numbers of condoms distributed.
We have no evidence that the lack of an intervention effect is due to bias. The cluster-randomization created a sufficient number of comparable agricultural units within which to implement the intervention . Our data on permanent employees excluded transient employees, but proximate plantations growing a given crop hire the same sorts of temporary workers in the same seasons. Employment stability at the sites led to high follow-up rates in both study arms. The STI test methods were accurate and unlikely to introduce misclassification. The high baseline STI prevalence, coupled with relatively low intra-cluster correlation, yielded ample power to detect a prevalence difference of 10%. Nevertheless, we found almost no difference in follow-up STI prevalence between intervention and control arms. Various factors help explain the lack of an intervention effect.
Our baseline measures of male condom use are consonant with national level data showing that in 1998 less than 10% of women had ever used male condoms, and 2% of married women and 8% of single women currently used them . Male condom use reported in our study interviews increased substantially during follow-up, but did not differ systematically between the intervention and control sites. The increases were not for contraceptive reasons: although we did not collect data to distinguish condom use with regular versus casual partners, condom use for family planning purposes remained below 4% throughout follow-up. The accuracy of condom use reports from the questionnaire items must be questioned, however, as the STI prevalences did not differ between women who reported consistent condom use compared with those who reported less frequent use or non-use.
Use of female condoms was lower than we anticipated, with close to 10% of women reporting consistent use. We expected a doubling in the proportion of protected coital acts at the intervention sites, which did not occur. The female condom may require more practice and more support for new users than was practicable here, and the early training in the clinic along with advice from outreach workers may not have been sufficient. However, these female condom findings are also consistent with other data. A more intensive intervention among US women at high risk of STI to whom male and female condoms were promoted found that about 25% of coital acts were protected by female condoms , but few women were able to maintain consistent use . In another US study of high-risk women, women assigned to use one type of condom (male or female) had the same STI reinfection rate as women assigned to receive a hierarchically-presented choice of multiple barrier methods .
We relied on, and paid stipends to, extant clinic service providers to disseminate information about STIs and condoms, counsel users, and record data on condom distribution and educational events. We also upgraded their skills in STI syndromic management, and separate case management data files indicated that appropriate care was offered in 75–80% of cases presenting at both intervention and control sites.
Despite the standardized training they had received, however, providers’ enthusiasm for the project varied dramatically. In a parallel qualitative study to ascertain providers’ knowledge, attitudes and practices regarding condoms and STIs, two notable findings emerged: (1) in observations of client visits, providers at intervention sites rarely promoted female condoms; and (2) in interviews, providers stated that the female condom was unsuitable for the most women. Although the bulk of male and female condom distribution was done by outreach workers and not clinic-based providers, the latters’ negative perceptions undoubtedly hindered uptake of the female devices.
Rumors about and suspicion of study purposes, staff, and devices plagued the project. Allegations of witchcraft were rife early in data collection. The female condoms were said to cause adverse health effects and changes in body shape of users, whereas some claimed the male condoms contained pinholes or were laced with pathogens. Despite male involvement in all group educational events, and the indirect effect of improved STI management, resident men complained that they benefited insufficiently from the study. Earlier preparation with community leaders might have minimized these damaging perceptions.
We found that STI prevalence at these rural sites, although variable, rivals that reported for urban Kenyan cohorts. [24–28] Condom promotion had become rather lethargic at the sites, but was re-invigorated by our fieldwork. STI management was generally poor at the plantations before the study began, and past policies at some sites required employees with STI to arrange and pay for treatment themselves. Multiple STIs could be grounds for dismissal. Such stigmatizing policies are a thing of the past at these sites.
It is unclear which aspects of the intervention – provider training, STI education, condom promotion, case management – were most closely associated with the decrease in STI prevalence from baseline to follow-up. Strictly speaking, the baseline prevalence cannot be compared with follow-up prevalences as the former is the result of years of person–time versus the 6-month prevalences during follow-up. However, we believe our intervention did have an impact at the sites. It remains to be seen whether these steps can be maintained after the close of our research project, a matter that we continue to discuss with plantation management. Based on their relatively high cost, female condoms are unlikely to be distributed there.
Female condom introduction has been more successful in shorter-term programs, and the devices have been well received by more selected cohorts such as sex workers and members of women's groups . Broader availability through social marketing in southern Africa has resulted in disappointingly flat sales [29,30], although sales are increasing again in Zimbabwe (M. Warren, personal communication, 2001). There does appear to be a niche for the device in all countries studied, and research on female condoms should continue along several lines, including more targeted introduction strategies, different prevention interventions, and new device designs. Myths about condoms must be addressed in any promotion programs in Kenya.
Mr Peter Mwarogo, formerly of the Family Planning Association of Kenya (FPAK), and Mr Stephen Mucheke of FPAK, helped design the intervention. Ms Kelley Ryan of FHI helped design the data collection forms, monitored the informed consent process, and analyzed the baseline data. We are particularly grateful for the untiring efforts of field study staff Dorcas Kungu, Ephel Khasandi, Nancy Maina, Catherine Wandera, and Joel Mutai. We thank Mario Chen-Mok and Hany Zayed for statistical support. FHI is an international nonprofit organization that conducts research and provides technical assistance in health, family planning, STIs and AIDS. This study was reviewed and approved by the FHI Protection of Human Subjects Committee, and the Kenyatta National Hospital Ethical Review Committee. All participants were verbally administered a study fact sheet in Kiswahili, and signed the study Volunteer Agreement form.
1. Farr G, Gabelnick H, Sturgen K, Dorflinger L. The Reality®
female condom: efficacy and clinical acceptability of a new barrier contraceptive. Am J Public Health 1994, 84: 1960 –1964.
2. Trussell J, Sturgen K, Strickler J, Dominik R. Comparative contraceptive efficacy of the female condom and other barrier methods. Fam Plann Perspect 1994, 26: 66 –72.
3. Drew WL, Blair M, Miner RC, Conant M. Evaluation of the virus permeability of a new condom for women. Sex Transm Dis 1990, 17: 110 –112.
4. Soper DE, Shoupe D, Shangold GA, Shangold MM, Gutmann J, Mercer L. Prevention of vaginal trichomoniasis by compliant use of the female condom. Sex Transm Dis 1993, 20: 137 –139.
5. Fontanet AL, Saba J, Chandelying V. et al
. Protection against sexually transmitted diseases by granting sex workers in Thailand the choice of using the male or female condom: results from a randomized controlled trial. AIDS 1998, 12: 1851 –1859.
6. Anon. Special Programme of Research, Development and Research Training in Human Reproduction. The Female Condom—A Review.
Geneva: World Health Organization; 1997.
7. Musaba E, Morrison C, Sunkutu MR, Wong EL. Long-term use of the female condom among couples at high risk of human immunodeficiency virus infection in Zambia. Sex Transm Dis 1998, 25: 260 –264.
8. Feldblum PJ, Bwayo JJ, Kuyoh M, Welsh M, Ryan KA, Chen-Mok M. The female condom and STDs: design of a community intervention trial. Ann Epidemiol 2000, 10: 339 –346.
9. Hayes R, Wawer M, Gray R. et al
. Randomised trials of STD treatment for HIV prevention: report of an international workshop. Genitourin Med 1997, 73: 432 –443.
10. Donner A, Donald A. Analysis of data arising from a stratified design with the cluster as the unit of randomization. Stat Med 1987, 6: 43 –52.
11. Bland JM, Kerry SM. Trials randomised in clusters. BMJ 1997, 315: 600. 600.
12. Lee HH, Chernesky MA, Schachter J. et al
. Diagnosis of Chlamydia trachomatis
genitourinary infection in women by ligase chain reaction assay of urine. Lancet 1995, 345: 213 –216.
13. Borchardt KA, Hernandez V, Miller S. et al
. A clinical evaluation of trichomoniasis in San Jose, Costa Rica using the InPouch TV test. Genitourin Med 1992, 68: 328 –330.
14. Wawer MJ, McNairn D, Wabwire-Mangen F, Paxton L, Gray RH, Kiwanuka N. Self-administered vaginal swabs for population-based assessment of Trichomonas vaginalis
prevalence [letter]. Lancet 1995, 345: 131 –132.
15. Shah BV, Barnwell BG, Bieler GS. SUDAAN Users’ Manual
, Release 7.5. Research Triangle Park, North Carolina, USA: Research Triangle Institute; 1997.
16. Feldblum PJ, Kuyoh MA, Omari M, Ryan KA, Bwayo JJ, Welsh M. Baseline STD prevalence in a community intervention trial of the female condom in Kenya. Sex Transm Infect 2000, 76: 454 –456.
17. Feldblum PJ, Chen-Mok M, Bwayo JJ, Omari M, Kuyoh M, Ryan KA. Intracluster correlation of STD prevalence in a community intervention trial in Kenya. Lancet 1999, 354: 1356 –1357.
18. Gollub EL. The female condom: tool for women's empowerment. Am J Public Health 2000, 90: 1377 –1381.
19. Rivers K, Aggleton P, Elizondo J. et al
. Gender relations, sexual communication and the female condom. Critical Public Health 1998, 8: 273 –290.
20. Anon. Kenya Demographic and Health Survey 1998. Calverton, Maryland, USA: National Council for Population and Development, Central Bureau of Statistics, Macro International Inc.; 1999.
21. Artz L, Macaluso M, Brill I. et al
. Effectiveness of an intervention promoting the female condom to patients at sexually transmitted disease clinics. Am J Public Health 2000, 90: 237 –244.
22. Macaluso M, Demand MJ, Artz LM, Hook EW III. Partner type and condom use. AIDS 2000, 14: 537 –546.
23. Gollub EL, French P, Loundou A, Latka M, Rogers C, Stein Z. A randomized trial of hierarchical counseling in a short, clinic-based intervention to reduce the risk of sexually transmitted diseases in women. AIDS 2000, 14: 1249 –1255.
24. Laga M, Plummer FA, Piot P. et al
. Prophylaxis of gonococcal and chlamydial ophthalmia neonatorum: a comparison of silver nitrate and tetracycline. N Engl J Med 1988, 318: 653 –657.
25. Braddick MR, Ndinya-Achola JO, Mirza NB. et al
. Towards developing a diagnostic algorithm for Chlamydia trachomatis
and Neisseria gonorrhoeae
cervicitis in pregnancy. Genitourin Med 1990, 66: 62 –65.
26. Daly CC, Maggwa N, Mati JK. et al
. Risk factors for gonorrhoeae, syphilis, and trichomonas infections among women attending family planning clinics in Nairobi, Kenya. Genitourin Med 1994, 70: 155 –161.
27. Temmerman M, Njagi E, Nagelkerke N, Ndinya-Achola J, Plummer FA, Meheus A. Mass antimicrobial treatment in pregnancy: a randomized, placebo-controlled trial in a population with high rates of sexually transmitted diseases. J Reprod Med 1995, 40: 176 –180.
28. Thomas T, Choudhri S, Kariuki C, Moses S. Identifying cervical infection among pregnant women in Nairobi, Kenya: limitations of risk assessment and symptom-based approaches. Genitourin Med 1996, 72: 334 –338.
29. Meekers D. Patterns of Use of the Female Condom in Urban Zimbabwe.
PSI Working Paper No. 28. Washington, DC: Population Services International; 1999.
30. Mobley S, Fisher A, Rutenberg N. The female condom: dynamics of use in Zimbabwe.XIIIth International Conference on AIDS
. Durban, South Africa, July 2000 [abstract MoOrC132].
Keywords:© 2001 Lippincott Williams & Wilkins, Inc.
Female condoms; male condoms; sexually transmitted infections; cluster randomized trial; intervention trial; epidemiologic methods; Kenya