Spermicides are unique among contraceptive options. They are coitally dependent but are not dependent on a male partner's cooperation. Their low cost, availability, and ease of use may be especially desirable for women who have intercourse infrequently and want to avoid hormonal methods. Currently, spermicides are among the least commonly used methods of contraception in the United States,1 yet there is a potentially high level of demand.2
All currently available spermicides contain the active ingredient nonoxynol-9 in a carrier such as a gel, foam, or film. Nonoxynol-9 is a surfactant that immobilizes or kills sperm by destroying the sperm cell membrane. Because this action is not specific to sperm, it was hoped that nonoxynol-9 use would reduce the risk of sexually transmitted infection.3 However, more recent clinical trial results show that it can cause genital irritation4 and may increase the likelihood of human immunodeficiency virus (HIV) transmission in very frequent users.5 Thus, the development of alternative spermicides has become a research priority.
C31G is a spermicidal mixture of two surfactants. Studies indicate that repeated use is safe with less cervicovaginal toxicity than nonoxynol-9.6 C31G has in vitro activity similar to that of nonoxynol-9.7 Phase 1 studies of C31G indicated that a 1.0% concentration was optimal.8 C31G effectively prevents sperm from penetrating midcycle mucus.9 A male tolerance study showed that male partners of women using C31G did not suffer penile irritation from the product.10 Taken together, these studies indicate that C31G is well tolerated, safe, and effective.
We conducted a multicenter, randomized, double-masked trial of C31G spermicidal gel and a commercially available nonoxynol-9 spermicide to compare contraceptive efficacy, acceptability, and safety. The primary objective was to estimate whether the contraceptive efficacy of C31G is noninferior to that of nonoxynol-9. We hypothesized that the contraceptive efficacy of C31G would be noninferior to that of nonoxynol-9. Outcomes included pregnancy rates, adverse events, continuation rates, and acceptability.
MATERIALS AND METHODS
This study was a phase III randomized, double-masked, noninferiority trial to evaluate the contraceptive efficacy of C31G over 6 months of use (six cycles and 183 days) compared with a commercially available, nonoxynol-9-based spermicidal gel. Participants had the option to participate in an extension study for a total treatment period of 12 cycles or 365 days. The primary study outcome was contraceptive efficacy. Secondary outcomes included acceptability and safety. Safety assessments included incidence of urinary tract infections (UTI), bacterial vaginosis, yeast infections, gonorrhea, and Chlamydia and occurrence of adverse events and serious adverse events. The definition of an adverse event was any symptom, illness, or experience that developed or worsened in intensity after randomization. A serious adverse event was defined according to Food and Drug Administration regulations.
The study was conducted at 14 sites of the National Institutes of Health's Eunice Kennedy Shriver National Institute of Child Health and Human Development Contraceptive Clinical Trials Network between 2004 and 2008. The protocol was initiated after obtaining approval from the Institutional Review board of each center and the National Institutes of Health. Eligibility requirements included being a healthy, sexually active woman, 18–40 years old, at risk for pregnancy and desiring contraception, having 24- to 35-day menstrual cycles, and being at low risk for HIV or other sexually transmitted infection. “Low risk” was defined as having a single male sexual partner for at least 4 months before study enrollment who was also at low risk for sexually transmitted infection. Study participants were asked to engage in at least four acts of vaginal sexual intercourse per month, use the study product as the primary method of contraception, and keep a diary of coital activity, product use, use of other vaginal products, and adverse events. Exclusion criteria included: allergy or sensitivity to either study product, three or more UTIs in the previous year, history of infertility, contraindication to pregnancy, use of shared drug injection needles in the previous 12 months, recent diagnosis (less than 3 months) or frequent outbreaks (less than three per year) of herpes simplex virus, sexually transmitted infection diagnosis within 6 months before enrollment, HIV infection, or abnormal cervical cytology confirmed by colposcopy within the previous 12 months.
The two study drugs were C31G gel in a 1% concentration and nonoxynol-9 gel in a 4% concentration. Randomization was performed in a 3:2 ratio of C31G to nonoxynol-9 (block size of 10) to obtain more information about adverse effects and acceptability in the C31G group. Study drugs were provided in single-use, prefilled applicators with identical overwraps to assure allocation concealment. These were packaged at a central site according to the randomization schema in sequentially numbered opaque boxes. Interventions were assigned by opening the next sequentially numbered box.
We performed a noninferiority sample size calculation using nQuery Advisorâ Release 4.0 with binomial distribution assumption.11 We estimate the 6-month cumulative probability of pregnancy using Kaplan-Meier methods to be 15% in women using nonoxynol-9 as their primary contraceptive for 6 months. We defined noninferiority as a pregnancy probability in the C31G group no more than 5% higher than in the nonoxynol-9 group. At the 2.5% level of significance (one-sided), a sample size of 1,002 women in the C31G treatment arm and 668 women in the nonoxynol-9 treatment arm was sufficient to reject, with 80% statistical power, the null hypothesis that the 6-month pregnancy probability in the C31G treatment arm was more than five percentage points greater than that in the nonoxynol-9 arm.
Once eligibility was confirmed, each participant returned for an admission visit and was randomly assigned to receive either the C31G or nonoxynol-9 gel. Each participant received a kit labeled with a randomized identification number containing a supply of single-use, prefilled applicators sufficient to last until her next scheduled study visit. Study products had similar appearance, texture, and smell.
Participants were scheduled for additional study visits after cycles 1, 3, and 6 of product use and after cycle 12 for those in the study extension. At each visit, a gynecologic examination was performed, including wet mount and assessment for bacterial vaginosis. Cervical cancer screening was performed after cycles 6 and 12. Urine was collected for pregnancy test (all visits) and dipstick analysis (cycles 1, 3, and 9) and urine culture (cycle 6 and 12). Coital diaries and compliance with product use were reviewed, and an acceptability questionnaire was completed at cycles 1, 6, and 12.
Four analysis populations were defined for the final statistical analysis: 1) the intent-to-treat population included all women randomized into the study; 2) the all-treated population, defined as intent-to-treat participants who applied study drug at least once, was used to determine safety and adverse event results; 3) modified intent-to-treat population consisted of intent-to-treat participants who had at least one episode of coitus while using the assigned study product and for whom there was at least one report of pregnancy status; and 4) the efficacy-evaluable subset of the modified intent-to-treat population included only those participants whose diaries indicated correct and consistent use of the assigned study product for at least one menstrual cycle. Perfect-use estimates were calculated from this population.
The primary efficacy end point was the cumulative probability of pregnancy in each group at 6 months, as determined from the modified intent-to-treat population using the Kaplan-Meier method. Pregnancies that occurred before randomization or after discontinuation of study method were excluded. The Peto method12 for calculating standard error, which focuses on the observed number of events in the experimental group, was used to construct 95% confidence intervals. This method was used to avoid underestimation of the standard error.
Secondary efficacy and effectiveness end points included 6-month correct and consistent use pregnancy probabilities, 12-month pregnancy probabilities, and 6-month Pearl rates. Correct and consistent 6- and 12-month pregnancy probabilities were calculated from the correct and consistent use population using the Kaplan-Meier method, whereas 6-month Pearl rates with confidence intervals13,14 were calculated from the modified intent-to-treat population.
Emergency contraception use was allowed in this study. The effects of emergency contraception use on the correct and consistent use pregnancy probability estimate were accounted for by subtracting the length of the cycle in which emergency contraception was used from the total length in study for nonpregnant participants and for pregnant participants if pregnancy did not occur in the cycle where emergency contraception was used. A Cochran-Mantel-Haenszel test was used to analyze emergency contraception use, and site was included in the model to account for center-to-center variation.
Safety outcomes were determined for the all-treated population. Data were collected from all participants regarding adverse events and serious adverse events. To be recorded as an adverse event or serious adverse event, an event had to begin or worsen between randomization and 14 days aftr last use of the study product. Categorical methods were used to evaluate the incidence of UTI, bacterial vaginosis, vaginal yeast infections, and sexually transmitted infections. For categorical data, either Cochran-Mantel-Haenszel test or Fisher's exact test was used. Acceptability of both products was compared using responses to a three-item questionnaire.
Enrollment, allocation, and follow-up numbers for participants are shown in Figure 1. Participants were primarily young, unmarried, non-Hispanic white women with at least some college education who lived with their partners (Table 1). Contraceptives most commonly used in the 6 months before enrollment included condoms (reported by 75% of participants), withdrawal (49%), and the rhythm method (14%). Only 17% had used hormonal contraception in the 6 months before enrollment, whereas 13% had used a spermicide within that time period. Over half (55%) had never used spermicide before this study.
All participants in the all-treated population reported at least one coital act during the study. The study product was used according to protocol (ie, it was the only method used and was used correctly) for 76% of reported coital acts among the C31G group and 78% of coital acts among the nonoxynol-9 group (P=.23). The study method was used with another method, used incorrectly, or not used at all in 3%, 5%, and 16% of coital acts, respectively, without any difference in the percentages for both treatment groups. An equal percentage of women in each group reported at least one departure from study instructions (43.9% in the C31G group compared with 43.0% in the nonoxynol-9 group). Slightly more than half of participants in each group discontinued before study completion (Fig. 1). Reasons for discontinuation are summarized in Figure 1. A large number of women in both groups withdrew consent, most commonly because of dissatisfaction with the study product (3.3% of C31G and 4.6% of nonoxynol-9 users), moving (1.9% of C31G and 2.5% of nonoxynol-9 users), and “other” reasons (11.7% of C31G and 11.8% of nonoxynol-9 users). None of these reasons was statistically different between groups. Older age (odds ratio 1.29; 95% confidence interval [CI] 1.19–1.41, for every 5 years older than age 18) and being married (odds ratio 1.44; 95% CI 1.17–1.78) were significantly associated with study completion. Treatment group had no effect on the likelihood of study completion.
For all participants, the mean±standard deviation number of days that the study product was used was 50.3±37.32 days in the C31G group and 49.4±34.9 days in the nonoxynol-9 group. For those in the 6-month study, mean number of days of use was 40.3±27.1 days for C31G and 39.9±25.5 days for nonoxynol-9, and in the 12 month extension study, the mean was 96.9±42.1 days for C31G and 96.9±36.5 days for nonoxynol-9.
Table 2 and Figure 2 show the 6-month and 12-month pregnancy probabilities for both products. Pregnancy probabilities for both the modified intent-to-treat and correct and consistent use populations demonstrate noninferiority of C31G compared with nonoxynol-9. Pearl index pregnancy rates (modified intent-to-treat) were 26.0 for C31G and 26.1 for nonoxynol-9, a difference of −0.1 (95% CI −8.9 to 8.7) based on 3,925 months of pregnancy risk exposure for C31G and 2,622 months for nonoxynol-9. Analyses of pregnancy probabilities in the correct and consistent use population were based on 2,346 months in the C31G group and 1,585 in the nonoxynol-9 group. There were no significant differences in pregnancy outcomes between groups. Within the modified intent-to-treat population, emergency contraception was used by 8.2% of women and 1.4% of cycles in the C31G group and 5.4% of women and 0.9% of cycles in the nonoxynol-9 group (P=.04).
Results of analyses of selected secondary outcomes are summarized in Table 3. There were no significant differences between the two groups in the frequencies of UTI, bacterial vaginosis, or vaginal yeast infections. Rates of genitourinary discomfort were also similar, reported by approximately one fifth of participants in each group. There were no significant changes in Papanicolaou testing or wet mount findings from baseline to study exit with use of either product. Furthermore, there were no significant differences between groups in these outcomes for the extension period, cycle 6 through cycle 12, of product use.
The proportion of participants who experienced an adverse event considered possibly, probably, or definitely related to study contraceptive was significantly different between groups (35% of C31G participants, 41% of nonoxynol-9 participants; P=.02). Related adverse events reported by at least 2% of participants are listed in Table 4. Percentage of participants with a serious adverse event exclusive of congenital anomaly was low in both groups (10 [1.2%] in C31G compared with 12 [2.1%] in nonoxynol-9 groups, P=.19) (Table 5). None of the serious adverse events in the C31G group were deemed related to the drug; of the serious adverse events listed for the nonoxynol-9 group, one case of hypersensitivity to the product and one case of pelvic inflammatory disease were, respectively, reported as definitely and probably related to product use.
There were a total of three congenital anomalies recorded for women who got pregnant during the study. One pregnancy in the C31G group resulted in a fetus with renal and cardiac malformations. This pregnancy was terminated, and the abnormalities were deemed unrelated to the study drug. No additional congenital anomalies were identified among the 57 live births in the C31G group. In the nonoxynol-9 group, 2 of 46 viable fetuses (4.3%) had congenital anomalies, one of whom was born with cardiac anomalies and the other with gastroschisis.
Acceptability of product was similar with approximately three fourths of women in each group reporting that they strongly or somewhat liked their assigned method after cycle 6 (or premature study exit). A similar proportion of women in each group said that they definitely or probably would use the product again. Acceptability was also high (89% in C31G compared with 86% in nonoxynol-9, P=.18) among the women who continued use of either product for 12 months.
We compared the contraceptive efficacy, acceptability, and safety of spermicidal products containing C31G and nonoxynol-9. We found that C31G was noninferior to nonoxynol-9 for efficacy and that side effects were similar. Strengths of this study include the randomized, masked design, the large number of participants, and the rigorous and detailed follow-up. In addition, participants were similar to those expected to be typical users. They were young, sexually active women who tended to be users of nonhormonal contraception. More than two thirds of the participants had proven fertility. Previous studies of spermicide efficacy show a wide range of pregnancy rates, from 3.8% to as high as 29%.15 The 6-month pregnancy rates of approximately 12% for each spermicide in the modified intent-to-treat population are consistent with rates seen in other spermicide studies.16–18 Perfect-use pregnancy rates were approximately 5% with both C31G and nonoxynol-9, consistent with the rates reported for some other N-9 formulations.12 This demonstrates that spermicides can provide efficacious contraception in women able to use them consistently and correctly.
Figure 2 indicates that a separation in pregnancy probability seems to occur at approximately 8 months, although the reasons for this are unclear. One might expect that those who continued in the study like their assigned product and therefore would be more likely to use it correctly and consistently; however, this should have been equally true for both groups. The sample size for the 6- to 12-month follow-up period was relatively small (Fig. 1); thus, the cumulative pregnancy probability estimates are less precise, and the power to detect a real difference is low.
Although the percentage of women reporting emergency contraception use was significantly higher among the C31G users, this unexpected finding is likely spurious given the double-masked nature of the trial. Because cycles of emergency contraception use were excluded both from the numerator (pregnancies) and denominator in the efficacy analyses, this difference would not have a differential effect on pregnancy rates.
The incidence of serious adverse events was low with both products. None were related to C31G use, and only two were potentially related to nonoyxnol-9 use. The proportion of women experiencing any adverse event (serious or not) potentially related to the study product was higher in the nonoxynol-9 group. This seems attributable, at least in part, to higher reported rates of genital symptoms (Table 4). However, rates of yeast vaginitis, bacterial vaginosis, and UTIs were low and similar between groups. Only 2% to 3% of users cited adverse events as a specific reason for discontinuation. The rate of major congenital anomalies seen in this study is within the range expected in the general population.19 The number of anomalies in either group was low, and no statistically significant difference was detected between the two groups.
Women who completed the study found the C31G product to be acceptable with the majority reporting that they liked it and would use it again. A recent review suggested that real-world acceptability and consistent use of spermicides are affected by social contexts such as partnership characteristics and common beliefs in women's social groups, which are not usually addressed in clinical trials.20 It is notable that even in this clinical trial setting, over 20% of coital acts were associated with nonuse or incorrect use of either assigned spermicide. Factors such as those listed previously may contribute to this inconsistency of use and in turn to the low real-world effectiveness21 of spermicides. Consideration of social contexts may be increasingly relevant as new spermicidal products become available.
Although male partners of participants in our study were not surveyed directly, few women reported that their partners had any adverse effects. Tolerability for male partners is obviously important, and evidence suggests that men will use spermicides if they are shown to be safe and effective.22
Limitations of the study include the high rate of discontinuation, which may also reflect some level of unstated dissatisfaction with either study product. Approximately 50% of participants discontinued before six cycles of use, similar to rates reported from previous randomized trials of spermicides.15,23 Such high rates may raise concern for external validity.24 In our study, the rate of discontinuation not related to pregnancy was similar to that seen in other studies of spermicide efficacy.15,16 In a previously published spermicide trial, women who did not complete the study were younger or unmarried and had intercourse less frequently than those who did complete.23 Findings in our study were similar: older and married women were significantly more likely to complete the study than younger or unmarried women.
Our findings of high satisfaction and acceptability at 6 months may, in part, reflect attrition of dissatisfied participants. Those who did not like their assigned contraceptive (or whose partners disliked it) may have been more likely to drop out. Acceptability data were also collected from women who exited the study early, and this bias may be small; however, we have no acceptability information for the approximately 11–12% of women who were lost to follow-up in either group.
We also lack data about any adverse events that may occur with use of longer than 12 months. However, no poststudy safety concerns have been reported. Finally, a potential benefit of spermicides is sexually transmitted infection prevention. C31G has broad-spectrum in vitro activity against Chlamydia, HIV, and herpes simplex virus-2.25–28 However, published evidence suggests that neither nonoxynol-9 nor C31G confers protection against HIV transmission.29,30 Thus, C31G may not offer an advantage in this regard. In our study, rates of acquisition of other sexually transmitted infections were too low to draw any conclusions about infection prevention.
Results of this randomized trial indicate that the contraceptive efficacy of C31G is noninferior to that of nonoxynol-9. C31G is also safe and highly acceptable. The commercial availability of spermicidal products containing C31G will provide more options to women who seek a coitally dependent, nonhormonal, nonprescription method of contraception.
1. Mosher WD, Jones J. Use of contraception in the United States: 1982–2008. National Center for Health Statistics. Vital Health Stat 2010;23(29).
2. Darroch JE, Frost JJ. Women's interest in vaginal microbicides. Fam Plann Perspect 1999;31:16–23.
3. Cook RL, Rosenberg MJ. Do spermicides containing nonoxynol-9 prevent sexually transmitted infections? A meta-analysis. Sex Transm Dis 1998;25:144–50.
4. Stafford MK, Ward H, Flanagan A, Rosenstein IJ, Taylor-Robinson D, Smith JR, et al. Safety study of nonoxynol-9 as a vaginal microbicide: evidence of adverse effects. J Acquir Immune Defic Syndr Hum Retrovirol 1998;17:327–31.
5. Van Damme L, Ramjee G, Alary M, Vuylsteke B, Chandeying V, Rees H, et al; COL-1492 Study Group. Effectiveness of COL-1492, a nonoxynol-9 vaginal gel, on HIV-1 transmission in female sex workers: a randomised controlled trial. Lancet 2002;360:971–7.
6. Catalone BJ, Ferguson ML, Miller SR, Malamud D, Kish-Catalone T, Thakkar NJ, et al. Prolonged exposure to the candidate microbicide C31G differentially reduces cellular sensitivity to agent re-exposure. Biomed Pharmacother 2005;59:460–8.
7. Catalone BJ, Miller SR, Ferguson ML, Malamud D, Kish-Catalone T, Thakkar NJ, et al. Toxicity, inflammation, and anti-human immunodeficiency virus type 1 activity following exposure to chemical moieties of C31G. Biomed Pharmacother 2005;59:430–7.
8. Mauck CK, Weiner DH, Creinin MD, Barnhart KT, Callahan MM, Bax R. A randomized Phase I vaginal safety study of three concentrations of C31G vs Extra Strength Gynol II. Contraception 2004;70:233–40.
9. Mauck CK, Creinin MD, Barnhart KT, Ballagh SA, Archer DF, Callahan MM, et al. A phase I comparative postcoital testing study of three concentrations of C31G. Contraception 2004;70:227–31.
10. Mauck CK, Frezieres RG, Walsh TL, Schmitz SW, Callahan MM, Bax R. Male tolerance study of 1% C31G. Contraception 2004;70:221–5.
11. Machin D, Campbell MJ. Statistical tables for design of clinical trials. Oxford (UK): Blackwell Scientific Publications; 1987.
12. Peto R, Pike MC, Armitage P, Breslow NE, Cox DR, Howard SV, et al. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. II. analysis and samples. Br J Cancer 1977;35:1–39.
13. Higgins JE, Wilkens LR. Statistical comparison of Pearl rates. Am J Obstet Gynecol 1985;151:656–9.
14. Miettinen O. Estimability and estimation in case-referent studies. Am J Epidemiol 1976;103:226–35.
15. Grimes DA, Lopez L, Raymond EG, Halpern V, Nanda K, Schulz KF. Spermicide used alone for contraception. Cochrane Database of Systematic Reviews 2005, Issue 4. Art. No.: CD005218. DOI: 10.1002/14651858.CD005218.pub2.
16. Raymond EG, Chen PL, Luoto J; Spermicide Trial Group. Contraceptive effectiveness and safety of five nonoxynol-9 spermicides: a randomized trial. Obstet Gynecol 2004;103:430–9.
17. Barnhart KT, Rosenberg MJ, MacKay HT, Blithe DL, Higgins J, Walsh T, et al. Contraceptive efficacy of a novel spermicidal microbicide used with a diaphragm: a randomized controlled trial. Obstet Gynecol 2007;110:577–86.
18. Mauck CK, Freziers RG, Walsh TL, Peacock K, Schwartz JL, Callahan MM. Noncomparative contraceptive efficacy of cellulose sulfate gel. Obstet Gynecol 2008;111:739–46.
19. Cunningham FG, Leveno KJ, Bloom SL, Hauth JC, Rouse DJ, Spong CY. Prenatal diagnosis and fetal therapy. In: Cunningham FG, Leveno KJ, Bloom SL, Hauth JC, Rouse DJ, Spong CY, editors. Williams Obstetrics. 23rd ed. Available at: www.accessmedicine.com/content.aspx?aID=6021591
. Retrieved August 9, 2010.
20. Mantell JE, Myer L, Carballo-Diéguez A, Stein Z, Ramjee G, Morar NS, Harrison PF. Microbicide acceptability research: current approaches and future directions. Soc Sci Med 2005;60:319–30.
21. Trussell J. Contraceptive efficacy. In: Hatcher RA, Trussell J, Nelson AL, Cates W, Stewart FH, Kowal D, editors. Contraceptive technology. 19th ed. New York (NY): Ardent Media, Inc; 2007. p. 747–826.
22. Holmes WR, Maher L, Rosenthal SL. Attitudes of men in an Australian male tolerance study towards microbicide use. Sex Health 2008;5:273–8.
23. Raymond EG, Chen PL, Pierre-Louis B, Luoto J, Barnhart KT, Bradley L, et al. Participant characteristics associated with withdrawal from a large randomized trial of spermicide effectiveness. BMC Med Res Methodol 2004;4:23.
24. Schulz KF, Grimes DA. Sample size slippages in randomised trials: exclusions and the lost and wayward. Lancet 2002;359:781–5.
25. Thompson KA, Malamud D, Storey BT. Assessment of the anti-microbial agent C31G as a spermicide: comparison with nonoxynol-9. Contraception 1996;53:313–8.
26. Krebs FC, Miller SR, Malamud D, Howett MK, Wigdahl B. Inactivation of human immunodeficiency virus type 1 by nonoxynol-9, C31G, or an alkyl sulfate, sodium dodecyl sulfate. Antiviral Res 1999;43:157–73.
27. Corner AM, Dolan MM, Yankell SL, Malamud D. C31G, a new agent for oral use with potent antimicrobial and antiadherence properties. Antimicrob Agents Chemother 1988;32:350–3.
28. Wyrick PB, Knight ST, Gerbig DG Jr, Raulston JE, Davis CH, Paul TR, Malamud D. The microbicidal agent C31G inhibits Chlamydia trachomatis
infectivity in vitro. Antimicrob Agents Chemother 1997;41:1335–44.
29. Feldblum PJ, Adeiga A, Bakare R, Wevill S, Lendvay A, Obadaki F, et al. SAVVY vaginal gel (C31G) for prevention of HIV infection: a randomized controlled trial in Nigeria. PLoS One 2008;3:e1474.
© 2010 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
30. Peterson L, Nanda K, Opoku BK, Ampofo WK, Owusu-Amoako M, Boakye AY, et al. SAVVY (C31G) gel for prevention of HIV infection in women: a Phase 3, double-blind, randomized, placebo-controlled trial in Ghana. PLoS One 2007;2:e1312.