Introduction
The male condom is the most effective barrier method currently available for the prevention of sexually transmitted diseases (STDs), including HIV infection [1]. It has proved to be effective in studies of couples with discordant HIV serological status, with virtually no HIV transmission occurring when it is used correctly and consistently [2,3]. However, women wanting to protect themselves must rely on their partners' willingness to use a condom. The recent development of the female condom may provide women with the first female-controlled barrier method which is effective against STDs, including HIV infection [4,5]. The female condom is a pre-lubricated, disposable, polyurethane vaginal sheath with two rings: one inside the closed upper end, which fixes around the cervix, and one at the wider open bottom which forms the external edge of the sheath and remains outside the vagina after insertion. Although the acceptability of the female condom in different population groups has been well studied [6-10], its use-effectiveness for STD prevention and contraception requires more evaluation.
Two studies have examined the use-effectiveness of the female condom as a contraceptive method, showing results similar to other female-controlled barrier methods [11-13]. However, the use-effectiveness of this new device in the prevention of STDs has been little studied [14]. The physical properties of the female condom make it likely that it will be as effective as the male condom in the prevention of STDs, and in vitro studies of polyurethane have shown no leakage of HIV and cytomegalovirus after use in an artificial coital model [15]. However, due to the risk of HIV infection, ethical considerations prevent an evaluation of the female condom in the prevention of STDs without attempting to use male condoms, a method with established efficacy, first or simultaneously. Accordingly, this intervention focused on the commercial sex establishments of Thailand, where male condom use has already been successfully promoted through the '100% condom program' [16,17]. In these establishments, there would always be a small proportion of clients refusing to use male condoms, leaving sex workers with no alternative but to take an unacceptable risk by having unprotected sex, or else lose their income by turning away these clients. Using a randomized controlled trial design, this study estimated the additional protection against STDs offered to sex workers by giving them the option of using a female condom when clients refused to use a male condom.
Methods
Study design
Sex establishments from four different cities in Thailand participating in a programme of early diagnosis and treatment of STDs were considered for this study. Only sex establishments where sex workers reported having several Thai clients per day were included in the trial. Randomization was carried out by sex establishments and not by individuals, so that sharing of female condoms across study groups would be minimized. The randomization was into two groups: (1) the male condom group, in which sex workers would propose a male condom to their clients, and, if clients refused or could not use male condoms, they were instructed not to have sex; and (2) the male/female condom group, in which women would propose a male condom to their clients as a first choice, but would have the option of using a female condom when clients refused or could not use a male condom. The randomization procedure relied on a table of random numbers matched with the list of sex establishments, and took into account the size of the sex establishments when assigning them to each of the two study groups. Recruitment of sex workers took place between October 1994 and January1995. An explanation of the purpose of the study was given to all women. Eligible women were at least 18 years of age; were not using a diaphragm, a cervical cap nor vaginal spermicides; had no evidence of intravenous drug use; were willing to use the condoms as instructed; and were willing to keep a pictorial coital log recording all sexual acts with clients and noncommercial sexual partners. After informed consent was obtained, women were interviewed regarding demographic variables, sexual practices and medical history. A physical examination, including a pelvic examination, was performed. Vaginal specimens were obtained for trichomonas and yeast, cervical and urethral specimens for Neisseria gonorrhoea culture, cervical specimens for Chlamydia trachomatis test by enzyme-linked immunoassay (ELISA), and urine specimens for a pregnancy test. All women were counselled regarding transmission of HIV and other STDs and were given a free supply of condoms.
Effective enrolment started 1 week after the screening visit, after results of clinical and laboratory examinations confirmed that women were not pregnant and had no evidence of trichomoniasis, candidiasis, genital ulcer disease, gonorrhoea or chlamydial infection. During the week prior to enrolment, women from both groups were taught how to fill in the coital log and women from the male/female condom group learned how to use the female condom, following instructions that were provided during a training session at the screening visit. It was assumed that all women knew how to use the male condom prior to the study through the '100% condom program'. For women who were diagnosed with an STD during the screening visit, enrolment was delayed until a cure was obtained.
Women were asked to return for follow-up every 14 days for 24 weeks. At each follow-up visit, the women answered a questionnaire on medical history, sexual practices and, for those in the male/female condom group, acceptability of the female condom. A pelvic examination was performed; vaginal specimens for trichomonas and yeast, and endocervical specimens for chlamydia ELISA and gonorrhoea culture were collected. The genital mucosa was systematically inspected for the presence of erosions, ulcerations, and bruises which might have been related to the use of male or female condoms. Investigators were instructed to tell women to stop the female condom use if they thought that genital lesions might be related to its use; they were requested to inform the trial monitor immediately if such events occurred. After vaginal examination, the coital log was reviewed with the interviewer, and the women were given a new supply of condoms. If infection with an STD was diagnosed, the women were treated free of charge, with antibiotics, according to the national standards. Women were re-enrolled into the study only if there was no clinical or laboratory evidence of STD at the next follow-up visit.
This protocol was reviewed and approved by the Secretariat Committee on Research Involving Human Subjects (SCRIHS) of the World Health Organization and by the Thai National Ethics Committee.
Statistical methods
The baseline characteristics of the study participants were described using means and proportions, and compared between trial groups using Student's t test and χ2 test where appropriate.
The coital logs filled out by the sex workers were used to compute the proportion of male and female condoms used during sexual acts, and whether the condoms tore or slipped in or out. Since the sex establishments were the units of randomization, these proportions were calculated for each sex establishment. A weighted regression analysis was performed using the logit-transformed proportions of condom use as the dependent variables, and trial groups and study sites as the independent variables. The weights used in the analysis were the number of sexual acts per sex establishment. The proportion of unprotected sexual acts (defined as any sexual act in which condoms were not used, tore, or slipped in or out) was compared between the two study groups using the same statistical methods.
Incidence rates of STDs were expressed as the number of STD events (gonorrhoea, chlamydial infection, trichomoniasis or genital ulcer disease) per 100 person-weeks of follow-up. Study participants contributed to person-time until they were diagnosed with one STD event, they were lost to follow-up, or they had completed the 6-month follow-up. For the women diagnosed and treated for an STD, person-time was counted again from the day of the visit during which a cure was confirmed. Loss to follow-up was defined as missing two consecutive visits plus 6 days (i.e., 34 days in total). After being lost to follow-up, a woman could re-enter the trial (i.e., person-time could be counted again) if she had no clinical nor laboratory evidence of STD on the day she was returning to the trial centre. As the sex establishments were the unit of randomization and not the individuals, incidence rates of STDs were calculated for each sex establishment (i.e., number of STDs diagnosed in each sex establishment divided by the total weeks of follow-up of all sex workers operating in that sex establishment). The sex establishments in the two trial groups were then compared using a weighted t test on the log-transformed incidence rates of STDs. The weights used in the analysis were the person-time of follow-up per sex establishment. The ratio of the weighted geometric mean incidence rate of STDs in the male/female over the male condom group was obtained by taking the antilogarithm of the difference between the weighted mean log-transformed incidence rates. A weighted regression analysis was also performed, where the dependent variable was the log-transformed incidence rate of STDs, and the independent variables were the trial group, study site, and baseline characteristics of the study participants averaged over the sex establishment (e.g., mean age of women working in the sex establishment; mean proportion of women who would have sex with clients refusing to use condoms, etc.). The analysis was completed under an 'intention-to-treat' basis. Similar analyses were carried out on the candidiasis episodes in the two groups.
The sample size calculation used assumptions based on 6-month cumulative incidences of STDs in the two trial groups due to the availability of common methods of calculation. The study was designed to detect a 33% reduction in risk of STDs in the male/female condom group when compared to the male condom group (α = 0.05, β = 0.20, two-sided, design effect = 2), with several options for the baseline cumulative incidences of STDs in the male condom group. Under these assumptions, the calculated sample size was 600 (300 in each group).
Results
Seventy-one sex establishments were selected from the four study sites, of which 37 were randomized to the male condom group, and 34 to the male/female condom group. Between October 1994 and January 1995, all sex workers operating in these establishments and who were willing to participate were screened for the study. Of the 644 women screened, 548 met the inclusion criteria and were assigned to their respective trial groups, depending on the sex establishments where they were operating: 266 women in the male condom group, and 282 in the male/female condom group. Forty-four women had no follow-up at all, 11 (4.3%) from the male condom group, and 33 (11.7%) from the male/female condom group (P = 0.001). There were no statistically significant differences on the baseline characteristics of the women without follow-up in the two trial groups. During the week prior to enrolment, the women of the male/female condom group who had no subsequent follow-up had tried a similar number of female condoms when compared to other women of the same trial group (mean = 4 and 2.2, respectively, P = 0.07) and had reported a similar satisfaction rate with the device (59% versus 63%, P = 0.81).
Results are presented for 255 women in the male condom group and 249 in the male/female condom group. Tables 1 and 2, respectively, show the distribution of demographic and selected characteristics, and sexual practices and sexual history, of study participants by trial group. The two groups were similar with respect to age, education level, marital status, history of prostitution, attitudes towards clients who refused condoms, vaginal cleansing practices, history of STDs, and number of STDs diagnosed at baseline. Only contraception use differed between the two groups (P = 0.02): 112 (45%) women assigned to the male/female condom group were using the oral pill compared to 82 (32%) of the male condom group, and 37 (15%) of the male/ female condom group were using injectable hormones compared to 67 (26%) of the male condom group.
The mean duration of follow-up was 12.9 weeks in the male condom group, and 12.6 weeks in the male/female condom group (P > 0.05). The baseline characteristics of the women who did not reach the median follow-up (12.6 weeks) versus those who did were compared for each of the two trial groups. There were no differences, except for two variables: there was a larger proportion of Thai nationals and of sex workers charging more than 500 bahts (US$ 20) per sexual act among the women lost to follow-up in the male/ female condom group. The number of women completing 24 weeks of follow-up was 60 (24%) in the male condom group and 68 (27%) in the male/female condom group.
Table 3 compares the use of male and female condoms between the two trial groups. Data recorded in 41 out of 2928 (1.4%) coital logs were not consistent (i.e. the sum of sexual acts with and without condoms was not adding up to the total number of acts recorded for the corresponding period) and were therefore removed from the analysis. Including these coital logs in the analysis despite their inconsistencies does not change significantly any of the comparisons described below (data not shown). A total of 105 942 sexual acts were analysed, which correspond to a mean (SD) number of sexual acts per working day of 3.31 (2.00) in the male condom group and 3.63 (2.39) in the male/female condom group (P > 0.05). There was a higher proportion of male condoms used in the male condom group when compared with the male/female condom group (97.5% versus 88.2%, P < 0.0001). However, when use of female condoms was considered, the overall proportion of sexual acts in which any condom was used was similar in the two groups (97.3% in the male condom group versus 97.9% in the male/female condom group, P > 0.05). Nevertheless, when considering the occurrence of tears and slippage of condoms, there was a 17% reduction in the proportion of unprotected sexual acts in the male/female condom group when compared with the male condom group (5.9% versus 7.1%, P = 0.16).
Male condoms tore more often than female condoms (2.8% versus 0.5% P < 0.0001), but were less likely to slip during use (1.3% versus 5.7%, P < 0.0001). However, among women with complete follow-up, the proportion of female condoms slipping in or out decreased from 10 of 95 (10.5%) at the first follow-up visit to zero of 129 (0%) at the last visit (P < 0.001). Interestingly, the proportion of male condoms torn was higher in the male condom group compared with the male/female condom group (3.2% versus 2.3%, P = 0.05).
Figure 1 shows the proportion of condom use by trial group over the 24 weeks of follow-up: the use of female condoms remained limited in the male/female condom group (12.0% of all sexual acts), and did not vary much with time (from 9.1% of sexual acts at the first follow-up visit to 15.4% at the last visit). The mean (range) proportion in women of the male/female condom group reporting no female condom use at the 2-weekly follow-up visits was 41.5% (29.0-56.0%); those reporting female condom use in 1-10% of sexual acts: 15.0% (6.0-27.5%); in 11-49% of sexual acts: 31.1% (16.0-39.4%); and in more than 50% of sexual acts: 12.1% (5.4-18.0%). The female condom was well tolerated by study participants and no genital lesions attributed to the female condom were reported during the study period.
The incidence rate of STDs by trial group is shown in Table 4. There was a 24% reduction in the weighted geometric mean incidence rate of STDs in the sex establishments of the male/female condom group when compared with the male condom group, although the difference was not statistically significant (2.81 versus 3.69 per 100 woman-weeks, P = 0.18). In a weighted regression analysis using the log-transformed incidence rate of STDs as dependent variable, controlling for study site did not affect the coefficient associated with trial group. However, the decrease in weighted geometric mean incidence rate of STDs in the male/female condom group was only 16% (P = 0.32) after adding into the model the two variables significantly associated with the log-transformed incidence rate of STDs: the mean cost of sex (there was a lower incidence rate of STDs in the sex establishments that charged their clients more), and the proportion of women agreeing to have sex with clients who refused to use condoms (there was a higher incidence rate of STDs in the sex establishments with higher proportion of women having sex anyway). There were no interaction terms between study sites and trial groups with P-values less than 0.05. For one site (number 3), the weighted geometric mean incidence rate of STDs was significantly lower in the male/female condom group when compared to the male condom group (rate ratio = 0.47; 95% confidence interval = 0.25-0.91). At that same site, the use of male condoms in the male condom group was high (99.2%), and the use of female condom in the male/female condom group remained limited (5.6% and significantly lower than at the three other sites, P < 0.0001), resulting in a 25% decrease in the proportion of unprotected sexual acts in the male/female condom group compared to the male condom group. In site number 4, which had the lowest mean proportion of male condom use in the male condom group (89.2%, significantly lower than in the three other sites, P < 0.0001), female condom use was high (17.9%), resulting in a 26% decrease in the proportion of unprotected sexual acts, and in a 27% decrease in the weighted geometric mean incidence rate of STDs in the male/female condom group compared with the male condom group (P = 0.29). Women reporting 100% condom use over their follow-up period (161 in the male/female condom group and 137 in the male condom group) had an incidence rate of STDs of 2.27 and 3.03 per 100 woman-weeks in the male/female and male condom groups, respectively.
The incidence rates of candidiasis episodes in the two groups were also measured. The weighted geometric mean incidence rate of candidiasis episodes by sex establishment was 0.92 per 100 woman-weeks in the male/female condom group compared to 0.86 in the male condom group (P > 0.05).
Discussion
Women were instructed to use the female condom only as a back up when clients refused to use male condoms. From the observed difference between the two study groups in the proportion of male condoms used (88% versus 97% in the male/female versus male condom groups, respectively), it is apparent that women in the male/female condom group did not insist on male condom use as much as women in the male condom group did. However, this reduction in male condom use was counterbalanced by the use of female condoms in 12% of all sexual acts, contributing to a 17% reduction in the proportion of unprotected sexual acts in the male/female condom group (P = 0.16). The replacement of male condoms by female condoms in the male/female condom group suggests that some sex workers, and/or their clients, preferred using the female condom. This phenomenon, which was sustained over the entire study period, is likely to have been even more common had the sex workers been instructed to use the female condoms whenever they wanted.
Slippage and tears of condoms may seriously affect their efficacy. In this study, the female condom proved to be more resistant to tearing than male condoms (0.5% torn against 2.8%, P < 0.0001). On the other hand, slippage of female condoms was common during the first two weeks of use, as has been reported in previous acceptability studies [6,10]. This problem disappeared over time among the women who continued using the female condom, stressing the need for enhanced training and counselling of any woman willing to use the female condom. It is interesting to note the higher proportion of male condoms torn in the male condom group (3.2%) when compared with the male/female condom group (2.3%), P = 0.05. As the clients of the two groups were made comparable by the randomization procedure, it is likely that the higher proportion of male condoms torn in the male condom group was observed among clients with whom a female condom would have been used in the male/female condom group. Since sex workers have a preference for using female condoms with clients who are drunk or violent (C.S. personal communication), it may well be that male condoms were more often torn with this category of clients.
Another concern about the safety and efficacy of condoms would be their propensity to induce genital lesions, which themselves could be used as a portal of entry for HIV if condoms were to be used inconsistently. However, there was no significant macroscopic genital lesion associated with long-term use of the female condom in our study, suggesting that the female condom was safe, as was observed in previous studies [18]. Nonetheless, vaginal colposcopy, which is difficult to interpret in sex workers, would be required to rule out lesions not visible by speculum examination.
There was a 24% decrease in the weighted geometric mean incidence rate of STDs in sex establishments in which female condoms were used, although not statistically significant (P = 0.18). The parallel decrease observed in the proportion of unprotected sexual acts (17%) and in the incidence rate of STDs (24%) in the male/female condom group compared to the male condom group suggests that the female condom was as effective as the male condom in the prevention of common STDs. Since the female condom tended to be more resistant to tears than male condoms, and since polyurethane has been shown to be impermeable in vitro to HIV in an artificial coital model [15], it is likely that a similar protection against HIV infection would be obtained. Finally, there was no increase in the incidence of candidiasis in the group using the female condom, in line with the findings of an earlier study [18].
Although these findings have potentially important implications, one should note that there are several limitations to this study. Sex workers are a mobile population and their follow-up in prospective studies is difficult. Indeed, almost half of the study participants were lost to follow-up after 3 months. However, the loss to follow-up affected both groups, and was not related to the group assignment. Another limitation is that the reliance on coital logs filled out by the sex workers for the analysis of condom use may lead to incorrect data: there is a 100% condom use policy which is strictly enforced by the government in commercial sex establishments in Thailand, and therefore some women may be reluctant to report lack of condom use. This would result in false reports of condom use, which may be suspected given the remarkably high condom use observed in our study. However, 1995 data obtained from clients with sex workers in the same setting showed a similar high proportion of condom use of 92% [19]. In addition, the comparability of the reported proportion of male condoms torn in the male condom group with what was observed by direct examination in a previous study in Northern Thailand (3.2 versus 3.9%, respectively) adds to the reliability of coital log data [20]. The high incidence rate of STDs (close to two STDs per year per woman) observed in this study, despite the high level of condom use, is surprising. In a previous study among sex workers of Bangkok in 1990, the incidence rate of gonococcal and/or chlamydial cervical infections was 6.4 per 100 woman-weeks in the group with the highest condom compliance (defined as > 75% of sexual acts), which is compatible with our findings [21]. The fact that in our study, sex workers who reported consistent condom use still experienced high incidence rates of STDs could have several explanations: condom failures may play an important role in the transmission of common STDs; there was a large over-reporting of condom use; sexual acts with non-client partners (e.g., husbands) were not reported (although sex workers were instructed to do so). Finally, the conclusions drawn in this study are relevant to a specific study population (sex workers and their clients in Thailand), in which a very high level of male condom use had already been obtained through the '100% condom program'. Similar studies should be repeated in other settings to re-assess the efficacy of the proposed strategy in places where condom use and incidence rate of STDs are different.
In this study, for the first time, sex workers were offered an additional choice of barrier method to protect themselves against STDs. The approach in this study, where the use of the female condom was suggested only when male condom use was not possible, meant that the female condom did not substantially affect the use of the male condom in the study population, and thereby did not threaten prevention efforts and male condom promotion in Thailand. Promotion of such a strategy, rather than promoting the female condom as a replacement to the male condom would optimize the public health benefit of the female condom, especially since the current price of the female condom is much higher than that of the male condom. An assessment of the applicability of this combined strategy in other countries would require further studies since cultural factors, as well as the rate of use of the male condom may have an impact on the results. Finally, other safe and effective preventive methods to be used by women, such as vaginal microbicides, are urgently needed in order to increase the choices available to women who seek to protect themselves against STDs and HIV infection [22].
Acknowledgements
We thank Dr A. Malek for his participation in the preparation of the study protocol, Dr P. Feldblum for his advice on the study design, Dr A. Ascherio, Dr A. Donner, Dr S. Kerry and Dr R. Hayes for their useful suggestions on the data analysis; we also thank the personnel involved in the follow-up of the sex workers in Thailand, and the sex workers for their participation in the study.
References
1. Update: barrier protection against HIV infection and other sexually transmitted diseases. MMWR 1993, 42:589-591.
2. de Vincenzi for the European Study Group on Heterosexual Transmission of HIV: A longitudinal study of human immunodeficiency virus transmission by heterosexual partners. N Engl J Med 1994, 331:341-346.
3. Roper WL, Peterson HB, Curran JW: Commentary: condoms and HIV/STD prevention - clarifying the message. Am J Public Health 1993, 83:501-503.
4. Rosenberg MJ, Gollub EL: Commentary: methods women can use that may prevent sexually transmitted disease, including HIV. Am J Public Health 1992, 82:1473-1478.
5. Gollub EL, Stein ZA: Commentary: the new female condom - item 1 on a women's AIDS prevention agenda. Am J Public Health 1993, 83:498-500.
6. Bounds W, Guillebaud J, Stewart L, Steele S: A female condom (FemshieldTM): a study of its user-acceptability. Br J Family Planning 1988, 14:83-87.
7. Sakondhavat C: The female condom. Am J Public Health 1990, 80:498.
8. Jivasak-Apimas S: Acceptability of the vaginal sheath (Femshield*) in Thai couples. Contraception 1991, 44:183-190.
9. Ford N, Mathie E: The acceptability and experience of the female condom, Femidom among family planning clinic attenders. Br J Family Planning 1993, 19:187-192.
10. Apipunyanon W, Srichan P, Magkalaviraj C: Acceptability of female condoms in sex workers in Udon Thani. J Health Sci (Thai Journal of Health Science) 1996, 5:375-383.
11. Bounds W, Guillebaud J, Newman GB: Female condom (Femidom). A clinical study of its use-effectiveness and patient acceptability. Br J Family Planning 1992, 18:36-41.
12. Trussel J, Sturgen K, Strickler J, Dominik R: Comparative contraceptive efficacy of the female condom and other barrier methods. Family Planning Perspectives 1994, 26:66-72.
13. Campbell P: Efficacy of the female condom. Lancet 1993, 341:1155.
14. Soper DE, Shoupe D, Shangold GA, et al.: Prevention of vaginal trichomoniasis by compliant use of the female condom. Sex Transm Dis 1993, 20:137-139.
15. 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.
16. Hanenberg R, Rojanapithayakorn W, Kunasol P, Sokal D: The impact of Thailand's HIV control program, as indicated by the decline of sexually transmitted diseases. Lancet 1994, 344:243-245.
17. Rojanapithayakorn W, Hanenberg R: The 100% condom program in Thailand. AIDS 1996, 10:1-7.
18. Soper DE, Brockwell NJ, Dalton HP: Evaluation of the effects of a female condom on the female lower genital tract. Contraception 1991, 44:21-29.
19. Nelson KE, Celentano DD, Eiumtrakol S, et al.: Changes in sexual behavior and a decline in HIV infection among young men in Thailand. N Engl J Med 1996, 335:297-303.
20. Rugpao S, Pruithithada N, Yutabootr Y, et al.: Condom breakage during commercial sex in Chiang Mai, Thailand. Contraception 1993, 48:537-547.
21. Niruthisard S, Roddy RE, Chutivongse S: Use of nonoxynol-9 and reduction in rate of gonococcal and chlamydial cervical infections. Lancet 1992, 339:1371-1375.
22. Pauwels R, de Clerck E: Development of vaginal microbicides for the prevention of heterosexual transmission of HIV. J Acquir Immune Defic Syndr 1996, 11:211-221.
© 1998 Lippincott Williams & Wilkins, Inc.