Secondary Logo

Contraceptive Effectiveness and Safety of Five Nonoxynol-9 Spermicides: A Randomized Trial

Raymond, Elizabeth G. MD, MPH*; Lien Chen, Pai PhD*; Luoto, Joanne MD

doi: 10.1097/01.AOG.0000113620.18395.0b
Original Research

OBJECTIVES: To estimate and compare the effectiveness and safety of 5 spermicides over 6 and 7 months of use, respectively. The spermicides included 3 gels containing 52.5 mg, 100 mg, and 150 mg of nonoxynol-9 per dose and a film and a suppository, each containing 100 mg of nonoxynol-9 per dose.

METHODS: Women wishing to use only spermicide for contraception for 7 months were randomly assigned to use 1 of the 5 spermicides with emergency contraception backup. Participants were followed up for up to 30 weeks after admission.

RESULTS: Of 1,536 women enrolled, 868 (57%) either relied on the spermicide for 6 months or became pregnant. The probability of pregnancy during 6 months of typical use of the spermicide was 22% (95% confidence limits 16%, 28%) in the 52.5-mg gel group, 16% (10%, 21%) in the 100-mg gel group, 14% (9%, 19%) in the 150-mg gel group, 12% (7%, 17%) in the film group, and 10% (6%, 15%) in the suppository group. The pregnancy risk in the 52.5-mg gel group was significantly different (P < .05) from that in either of the other gel groups. The pregnancy risks in the three 100-mg product groups were not significantly different (P = .35). No significant differences among groups were found in the 7-month probability of specified urogenital conditions.

CONCLUSION: The gel with the lowest amount of nonoxynol-9 was less effective than the 2 higher-dose gels. Among 3 products containing 100 mg of nonoxynol-9, formulation did not significantly affect pregnancy risk. All products were safe.


Among users of 5 spermicides, pregnancy risk varies by amount of nonoxynol-9 per dose but not by product formulation, and all products are safe.

From *Family Health International, Research Triangle Park, North Carolina; and the †National Institute of Child Health and Human Development, Rockville, Maryland.

Received August 14, 2003. Received in revised form October 28, 2003. Accepted November 13, 2003.

Address reprint requests to: Elizabeth Raymond, MD, MPH, Clinical Research Division, Family Health International, P.O. Box 13950, Research Triangle Park, NC 27709; e-mail:

*For other members of the Spermicide Trial Group, see the Appendix.

Funding Support: Support for this study was provided by Family Health International with funds from the National Institute of Child Health and Human Development contract number N01-HD-7-3271. The views expressed in this article do not necessarily reflect those of Family Health International or the National Institute of Child Health and Human Development.

Spermicides were donated by the following companies: Columbia Laboratories, Inc, Miami, FL; Advanced Care Products, Personal Products Company, a Division of McNeil-PPC, Inc, Skillman, NJ; Thompson Medical Company, Inc, West Palm Beach, FL. Supplies used for emergency contraception were donated by Wyeth Ayerst, Philadelphia, PA, and Women's Capital Corporation, Washington, DC.

Vaginal spermicides containing nonoxynol-9 have been available over the counter in the United States for almost 50 years. Their popularity has varied over time; in 1995, the last year for which national data are available, approximately half a million women in this country relied on these products to prevent pregnancy.1 Despite the long experience with these products, knowledge about their contraceptive effectiveness is limited. The many published reports of spermicide use have yielded disparate results. Pearl indices, for example, have ranged from 0 to 67 pregnancies per 100 woman-years.2 This variability may reflect differences in the doses and formulations of the spermicides studied, the underlying fecundity of the study populations, compliance with instructions for use of the spermicide, or coital patterns. Many of the published studies had serious flaws in design, execution, or analysis.3

In 1995, recognizing this deficiency, the U.S. Food and Drug Administration published a proposed rule that would require spermicide manufacturers to submit data demonstrating the contraceptive effectiveness of each marketed spermicide product in order for the product to remain on the market.4 This rule has not yet been finalized. Meanwhile, the National Institute of Child Health and Human Development sponsored the study reported here to evaluate the effectiveness, safety, and acceptability of 5 marketed nonoxynol-9 products.

This trial included 3 gels, a film, and a suppository. Gels A, B, and C contained 52.5 mg, 100 mg, and 150 mg of nonoxynol-9 per dose, respectively, and the film and the suppository each contained 100 mg of nonoxynol-9 per dose. This array of products was chosen to allow us to assess separately the effect of 2 key aspects of spermicides: the amount of active ingredient (by comparing the 3 gel products) and the formulation (by comparing the film, suppository, and gel B). The 2 primary null hypotheses tested were that the probability of pregnancy did not differ during 6 months of typical use of either the 3 gels or the three 100-mg products. The probabilities of selected urogenital conditions were similarly compared over 7 months of spermicide use. This trial differs from all previous spermicide-effectiveness trials in that for ethical reasons, emergency contraceptive pills were explicitly made available to all participants when needed.

Back to Top | Article Outline


The trial was conducted at 14 sites in the United States between June 1998 and August 2002. The Institutional Review Boards at each site and at Family Health International approved the study. All participants signed written informed consent forms before enrollment. In addition, at one center, the Institutional Review Board required that male partners of participants sign an informed consent form. The study was monitored by a Data and Safety Monitoring Board, which reviewed one interim pregnancy analysis in October 2000. The results of this analysis did not meet the predefined stopping guideline, which specified that a study arm might be stopped if the 6-month pregnancy probability during typical use of the product was significantly higher than 30%, and the board suggested no change in procedures.

Healthy, sexually active women aged 18–40 years who had no history suggestive of subfecundity, who were at low risk for sexually transmitted infections, who wished to rely on a spermicide as their only contraceptive method for 7 months, and who were willing to accept a moderate risk of pregnancy were eligible for the study. Women were disqualified if they had contraindications to nonoxynol-9 use or to pregnancy, were pregnant or within 6 weeks after the end of their last pregnancy, were breastfeeding, or had a partner who to their knowledge had risk factors for sexually transmitted infections.

After providing informed consent, each volunteer had an interview, pelvic examination, Pap test, wet prep, and urine pregnancy test sensitive to 25 mIU/mL human chorionic gonadotropin. A urine culture was done if she had symptoms of a urinary tract infection. If she was eligible, site staff telephoned a central randomization center to obtain an assignment to 1 of the 5 study spermicide groups. The randomization scheme was stratified by site with randomly ordered block sizes and was generated by computer at Family Health International before the start of enrollment at each site. The participant was informed about proper use of her assigned spermicide and about the availability and use of emergency contraceptive pills free of charge from the study site in case of unprotected or inadequately protected intercourse. Each participant completed a self-assessment quiz designed to evaluate and enhance her understanding of this information. She was given a supply of her assigned spermicide and a daily diary on which to record all menstrual periods, acts of intercourse, and use of contraception.

Follow-up visits were scheduled at 4, 17, and 30 weeks after admission. Each participant was also asked to return if she had medical problems or if she wished to stop relying on the assigned spermicide as her primary contraceptive method earlier than 30 weeks. At each visit, a pregnancy test was done. The participant was asked about specified symptoms of vulvar or vaginal irritation or of urinary tract infection within the previous week, about other adverse events, and about complaints her partner may have had about the spermicide. A pelvic examination and wet prep were performed at the final visit and at any visit when the subject reported any irritation symptoms. Spermicide group assignment was not concealed from clinical examiners. A urine culture was performed if the participant had symptoms of a urinary tract infection. At the 4-week visit, the participant repeated the self-assessment quiz, and at the 4-week and final visits, she completed a questionnaire about acceptability of the spermicide. Each participant was asked to perform a pregnancy test at home 2, 10, and 23 weeks after admission. If the site investigator determined that continued use of spermicide as a sole contraceptive method was unsafe or the participant decided that continued use was undesirable, the participant was discontinued from the study. She was asked to perform a home pregnancy test 2 weeks later, and she was then contacted at 30 weeks after admission and asked about interim contraception use and pregnancies.

The primary analyses included all enrolled participants who contributed any follow-up data (the primary-analysis population). The primary pregnancy analyses estimated and compared the cumulative 6-month probabilities of pregnancy during typical use of the spermicides (ie, while the participant was relying primarily on the spermicide for contraception). Each participant contributed until the earliest of the following dates: the estimated date of fertilization of a pregnancy; the date she stopped relying primarily on the assigned spermicide for contraception (her “stop-use date”); the latest date her pregnancy status could be reliably determined; and 183 days after randomization. Pregnancies were included regardless of the means by which they had been detected (pregnancy test, participant report, or medical record review). An obstetrician (E.G.R.) unaware of treatment group estimated the date of fertilization by adding 14 days to the onset date of the last menstrual period or by considering ultrasound results or other obstetric information when available. The probability of pregnancy in each spermicide group was estimated by using the Kaplan-Meier product-limit method, and 95% confidence limits were calculated by the method of Peto et al.5 The Fisher least significant difference approach with the logrank test was used to compare pregnancy risks over 6 months of use among the 3 gel groups and among the 3 groups using spermicides containing 100 mg of nonoxynol-9 per dose. If a significant result (P < .05) was found in either of these comparisons, pairwise comparisons among the constituent groups were performed.

In secondary exploratory pregnancy analyses, comparisons among groups were adjusted for center cluster (grouped geographically) and other baseline characteristics found to be related to pregnancy (P < .1) in a Cox regression model that included 9 characteristics but not spermicide group. A sensitivity analysis compared the pregnancy probabilities among groups in a predefined “per protocol” subset of participants without admission violations that might have affected fecundity. This subset excluded women who were older than 40 years, who were taking medications that might have decreased fecundity, or who might have had a vaginal abnormality that interfered with spermicide retention. Pregnancy probabilities were also calculated in other subsets felt to be of interest.

Similar methods were used to estimate and compare the probabilities of pregnancy through 6 complete menstrual cycles of typical use, consistent use, and perfect use of each product. Rules for identifying complete cycles were defined before the analysis. These rules excluded the enrollment cycle unless enrollment occurred during the first 5 days of that cycle and also excluded all data after the first of 8 consecutive days on which a participant failed to record on her diary whether she had experienced any bleeding. Consistent use was defined as use of the assigned spermicide at every act, with or without another contraceptive method. Perfect use was defined as consistent use of only the assigned spermicide, which was inserted into the vagina within the time specified in the manufacturer's product label (gels, 0–60 minutes before coitus; suppository, 10–60 minutes before coitus; film, 15–60 minutes before coitus).

An intent-to-treat analysis to estimate the probabilities of pregnancy over 6 months after admission (regardless of spermicide use) was performed by using the same methods as the primary analyses except that the stop-use date was not used in the determination of each participant's time in analysis. In an additional analysis, the 6-cycle pregnancy probability was estimated in each group according to previously described methods as if emergency contraceptive pills had not been used (Dominik R. Estimating the probability of pregnancy among barrier contraceptive users: removing the effect of emergency contraceptive back-up: statistical approaches to the evaluation of barrier contraceptive effectiveness [dissertation]. Chapel Hill (NC): University of North Carolina; 2000).

The primary safety analyses focused on 5 conditions: vulvovaginal candidiasis, bacterial vaginosis, vulvar or vaginal irritation without evidence of concurrent vulvar or vaginal infection, urinary tract infection or urinary tract infection symptoms, and culture-proven symptomatic urinary tract infection. Each participant contributed to these analyses through the earlier of 7 months or 1 week after the date she stopped using the spermicide, which was considered to be the later of the date of her last spermicide use and the date when she stopped considering the spermicide to be her primary method. All cases of each of the 5 conditions ascertained from all relevant sources (participant interview, medical record review, examination and wet prep performed by study staff, Pap test, and urine cultures) were included. Wet preps done by study staff were considered diagnostic for bacterial vaginosis only if either vaginal pH higher than 4.5 or amine odor was noted along with other evidence of this condition. Vulvar or vaginal irritation was defined as any of the following: vaginal or vulvar itching, pain, discharge, dryness, loss of sensitivity, or puffiness, unspecified vaginitis or cervicitis, allergy to spermicide, or specified genital lesions or signs of irritation. Concurrent infections included vulvovaginal candidiasis, bacterial vaginosis, vaginal trichomoniasis, vulvovaginal condyloma, or genital herpes within 2 weeks before or after the irritation symptoms or signs. Culture-proven symptomatic urinary tract infection was diagnosed by identification of a single organism or a “positive” report on culture of specimen obtained from participants with dysuria, frequency, or other urinary tract infection symptoms. The same analytic approach described for the pregnancy analyses was used to estimate the probability of each of the 5 conditions in each spermicide group and to compare the probabilities among the three gel groups and the three 100-mg product groups over 7 months of typical spermicide use.

Coital frequency was compared among groups by using the Kruskal-Wallis test. Except where otherwise noted, a P value less than .05 was considered significant for all primary and secondary analyses. All participants were analyzed in the groups to which they were assigned.

The target enrollment was 1,800 enrolled women. The National Institute of Child Health and Human Development determined this number based on precision and power estimates for a range of reasonable pregnancy probabilities, sample sizes, and expected retention rates, and on logistical and financial considerations. Recruitment was slower than expected, however, and enrollment was stopped before the target was reached.

Gel A (Advantage S; Columbia Laboratories, Inc., Miami, FL) contained 52.5 mg of nonoxynol-9 per dose, polycarbophil, glycerin, carbomer 934P, water, mineral oil, hydrogenated palm oil glyceride, sorbic acid, and methylparaben. Gels B and C (Ortho Options Conceptrol Vaginal Contraceptive Gel; Advanced Care Products, Skillman, NJ) contained 100 mg and 150 mg of nonoxynol-9 per dose, respectively, plus lactic acid, methylparaben, povidone, propylene glycol, purified water, sodium carboxymethylcellulose, sorbic acid, and sorbitol solution. Each film (Ortho Options Contraceptive Film; Advanced Care Products) contained 100 mg of nonoxynol-9, polyvinyl alcohol, and glycerin. Each suppository (Encare; Thompson Medical Company, Inc., West Palm Beach, FL) contained 100 mg of nonoxynol-9, polyethylene glycol, sodium bicarbonate, sodium citrate, and tartaric acid.

Back to Top | Article Outline


Altogether, 1,536 women enrolled in the spermicide study, ranging from 6 to 300 women at each site. Fifty enrolled participants failed to meet all admission criteria. Admission violations included the following: the participant was not screened on the randomization date (n = 3), the participant had risk factors for sexually transmitted infections (n = 3), the participant's admission pregnancy test was positive (n = 1), the participant was using medication that might interfere with fecundity (n = 1), the participant may have had an abnormality that could interfere with retention of spermicide (n = 1), and the absence of such an abnormality was not documented (n = 41). Three women (1 from each gel group) received a spermicide (gel) at follow-up visits other than the one assigned. Eight women (3 in the gel A group, 2 in the gel C group, 2 in the suppository group, and 1 in the film group), received spermicide that expired before her last day in the analysis. Of the 1,536 enrolled participants, 51 (n = 8–12 in each group) had no follow-up after admission and were thus excluded from the primary analysis population, which thus consisted of 1,485 women.

Baseline characteristics of participants in the 5 groups were similar (Table 1). The median age was 26 years. The main contraceptive methods used before admission were male condoms, withdrawal, spermicides, and periodic abstinence (data not shown). Participants reported a median of 8 coital acts in the previous 4 weeks. Thirty-six percent of the analysis population reported vaginal candidiasis, bacterial vaginosis, or other vulvar or vaginal irritation or infection within the prior 6 months or had symptoms or signs of such a condition at admission.

Table 1

Table 1

Almost 60% of the 1,485 participants in the primary-analysis population either relied on the spermicide for 6 months or became pregnant earlier while relying on the assigned spermicide (Table 2). The rest discontinued early, were later discovered to have been pregnant on admission, or were lost to follow-up. The proportion of women in each category was similar in the 5 spermicide groups, except in the gel A group, which included more participants who became pregnant and fewer who discontinued early or were lost than the other groups. Of the 252 participants who discontinued earlier than 6 months, 20% did so at least in part because she or her partner had a medical problem. These problems included vulvar or vaginal irritation or other local reactions, menstrual abnormalities, urinary tract infection, vulvovaginal candidiasis, other medical conditions unrelated to spermicide, and penile irritation. Women who were lost to follow-up were on average slightly younger, less educated, and less likely to consider avoiding pregnancy to be “very important” and had sex more often than other women in the primary-analysis population (data not shown). The total amounts of time contributed by the 5 groups both to the primary pregnancy analysis and to the safety analysis were similar.

Table 2

Table 2

The 1,485 participants in the primary-analysis population recorded diary information on coitus or vaginal bleeding on an average of 89% of their days in the primary pregnancy analysis. An average of 97% of the diary entries were recorded on the day that the information pertained to or within 1 week after that. Participants in the suppository group had intercourse slightly less often than participants in the other groups (Table 3), but no statistically significant difference in coital frequency among groups was found. Participants reported having used the assigned product at almost all coital acts.

Table 3

Table 3

The cumulative 6-month probability of pregnancy during typical use ranged from 10% to 22% in the 5 groups (Table 4). No statistical difference was found among the probabilities in the 3 groups assigned products containing 100 mg of nonoxynol-9 per dose. However, the probabilities among the 3 gel groups were found to be statistically different. In pairwise comparisons, the probability in the gel A group was significantly greater than the probability in either of the other 2 gel groups. Adjustment for center cluster, age, marital status, parity, importance of avoiding pregnancy, and prestudy coital frequency did not substantially change these findings, except that the difference between the gel A and B groups was no longer statistically significant. In 4 of the 5 groups the risk of pregnancy appeared to decrease over time as expected (data not shown). We suspect that the observed increase in the risk over time in the gel B group was more likely due to chance than to a real effect.

Table 4

Table 4

The probabilities of pregnancy among women without admission violations that might have affected fecundity (the “per protocol” subset) were almost identical to those among all participants (Table 4). Probabilities varied in other subsets defined by individual baseline characteristics, however. To identify factors independently related to pregnancy, a Cox regression model was fitted that included age, marital status, race/ethnicity, education, parity, importance of avoiding pregnancy, prestudy coital frequency, prior spermicide use, and desire for additional children, controlling for center cluster and spermicide group. Only coital frequency was related at the preselected significance level (P < .005).

The probability of pregnancy through 6 menstrual cycles of typical use was lower in each group than the probability through 6 months of use (Table 4). The 6-cycle probabilities during consistent and perfect use were similar to the 6-cycle probability during typical use in each group, except the suppository group. In that group, consistent and perfect use were associated with substantially lower probabilities of pregnancy than typical use. No statistically significant differences were found among groups in any of the cycle-based analyses. The rules for defining cycles resulted in the exclusion of substantial numbers of women from the analysis in each spermicide group, and the amount of time contributed by each woman was generally lower than in the month-based analyses. The exclusion of the incomplete admission cycles reduced the number of pregnancies in the analysis. Therefore, the cycle-based probability estimates are less precise than the month-based estimates and the comparisons among groups are less powerful.

The intent-to-treat analysis of the probability of pregnancy through 6 months after admission regardless of when each participant stopped using her assigned spermicide included 1,038 women who contributed a full 6 months to the analysis or became pregnant earlier than 6 months. The findings were very similar to those of the primary analysis (Table 4).

A total of 82 women used emergency contraceptive pills 109 times during their time in the primary pregnancy analysis, and in 57 of the menstrual cycles included in the cycle-based analysis. Because the pills were used so infrequently, they had little effect on estimates of the probability of pregnancy (data not shown).

A total of 34 serious adverse events occurred in 31 study participants during or after spermicide use, none of which were felt to be related to the spermicide. The 7-month probability of vaginal candidiasis, bacterial vaginosis, vulvar or vaginal irritation without infection, urinary tract infection or urinary tract infection symptoms, or culture-proven symptomatic urinary tract infection did not differ among groups (Table 5). Large proportions of the incident cases of candidiasis and bacterial vaginosis (65% and 42%, respectively) were ascertained by participant report or other sources and were not confirmed by wet prep at the study clinics. Complaints among partners of participants included penile burning, itching, and urinary tract infection symptoms. A significant difference was found among the 7-month probabilities of partner complaints in the 3 groups that received products containing 100 mg of nonoxynol-9, and pairwise comparisons showed a significant difference between the suppository group and the gel B group. None of the safety results differed materially when the analyses were repeated including only women who had not used any nonoxynol-9 product within the week before admission (data not shown).

Table 5

Table 5

Back to Top | Article Outline


Our analysis was explicitly structured to examine 2 factors that may potentially influence the effectiveness and safety of nonoxynol-9 spermicide products: amount of nonoxynol-9 per dose and product formulation. In this trial, we found that a spermicidal gel containing 52.5 mg of nonoxynol-9 per dose was associated with a higher probability of pregnancy over 6 months of typical use than gels containing 100 and 150 mg of nonoxynol-9 per dose. In contrast, we found no difference in the 6-month probability of pregnancy during typical use of a film, suppository, or gel each containing 100 mg of nonoxynol-9 per dose. These results are consistent with the interpretation that dose may affect effectiveness—in fact, there may be a threshold effect—but that formulation does not.

However, this interpretation may be an oversimplification. The 52.5-mg gel differed from the 2 higher-dose gels in several respects, including inactive ingredients, gel volume, and applicator design; any of these attributes could affect contraceptive effectiveness. The 2 higher-dose gel products consisted of the same substance in identical applicators, but the 150-mg applicators contained 50% more volume than the 100-mg applicators. Our finding of no difference in pregnancy probability associated with these 2 products may reflect the uniform concentration of nonoxynol-9 rather than an absence of difference in contraceptive potential between 100 mg and 150 mg of nonxynol-9. Similarly, our failure to observe a difference in effectiveness among the three 100-mg products used in this study does not prove that spermicide formulation is always unrelated to effectiveness. This study had limited power to detect small differences among groups.

Comparing our results to those of previous studies is hampered by the substantial variations in study designs and populations among the studies. The study most similar to ours was the recently published trial of a nonoxynol-9 film and foaming tablet, also designed and managed by Family Health International.6 That trial found much higher probabilities of pregnancy than we present here: the 6-month typical-use pregnancy probabilities in the film and tablet groups were 25% and 28%, respectively. The earlier trial was done largely in countries where fertility rates are higher than in the United States. The participants individually may have been at higher risk for pregnancy than our participants: they were younger and were more likely to be married, parous, and desirous of additional children. They also reported more frequent intercourse during the study than our participants (about 13 acts per 30-day interval, compared with 9.2 in our trial). Our results are comparable to the findings of the 1995 National Survey of Family Growth and the 1994–1995 Abortion Patient Survey, which together indicated that about 16% of spermicide users became pregnant in 6 months.7 That similarity is somewhat surprising because our participants were selected to be at risk for pregnancy and our approach to detection of pregnancy was certainly more sensitive than the methods used in the surveys. Our participants received repeated, detailed instruction on how to use the products consistently and properly, which may have reduced their risk of pregnancy compared with the general population of spermicide users.

Except in the suppository group, the probabilities of pregnancy during consistent and perfect use were similar to the probability during typical use. The same observation was made in the earlier trial of film and foaming tablets.6 One explanation may be that in both trials, the product was reportedly used at almost all coital acts. Alternatively, it is possible that inconsistent and imperfect use tended to occur at times in the cycle when fecundity was low.

The safety of the 5 spermicides in women did not appear to differ materially. This result is not unexpected. Previous studies have indicated that nonoxynol-9 does not affect the risk of vaginal candidiasis8,9 and that it may not cause significant vaginal pathologic changes except when used more often than in this trial and in high doses (eg, 150 mg).10 Spermicides have been found in some but not all studies to decrease the risk of bacterial vaginosis11 and to increase the risk of urinary tract infection,12,13 but apparently any differences in exposure in the 5 groups in this study were not sufficient to have an effect. No serious adverse events related to the spermicides occurred in our study, and the cumulative risks of the local effects evaluated in our analysis were not alarmingly high in any spermicide group.

An important finding of our trial was that enrollment and follow-up were more difficult than anticipated. In planning the study, we projected that we would enroll 5.2 participants per site per month and that 33% would fail to complete the study without pregnancy. These projections were conservatively made based on our experience with many previous trials of spermicides or other barrier contraceptives. However, the average enrollment rate was only 4.0 participants per site per month, and 43% of women discontinued early or were lost to follow-up. The difficulty occurred at almost all sites: only 2 of the 14 sites enrolled faster than the expected rate, and only 3 had more than 67% of participants complete the study. In addition, the proportion of women who were lost to follow-up (26% overall) was substantially higher than has been seen in other recent studies of barrier contraceptive methods.6,14 Although the reasons for these problems are not clear, it is possible that the study procedures or the spermicides themselves deterred women from enrolling and staying in the study. Further analysis is planned to evaluate participant characteristics and other correlates of failure to complete the trial; this information should be useful to researchers planning studies of new spermicides and microbicides.

The high proportion of participants who failed to complete the study is a serious concern for the interpretation of the pregnancy probability estimates within each group: if women who were lost to follow-up had higher or lower risk of pregnancy after they were lost than other women, our reported pregnancy proportions would be underestimates or overestimates, respectively. The fact that the participants who were lost to follow-up had a higher coital frequency at baseline, a characteristic found to be associated with pregnancy among women who were not lost, suggests that the former possibility may be more likely. However, the proportion who did not complete was similar in all 5 groups. Therefore, we believe that our relative comparisons are credible.

All 5 products tested in this study were safe as used by our participants. The pregnancy probabilities we observed were in the range previously accepted for users of spermicides and other barrier methods,2,15 except perhaps in the 52.5-mg gel group. However, the risk of pregnancy in each group was higher than would be expected among users of other modern contraceptive methods, including hormonal methods and intrauterine devices. The medical community must take responsibility for ensuring that potential users fully understand effectiveness of spermicide products relative to other methods.

Back to Top | Article Outline


1.Trussell J, Kowal D. The essentials of contraception. In: Hatcher RA, Trussell J, Stewart F, Cates W, Stewart GK, Guest F, Kowal D, editors. Contraceptive technology. 17th ed. New York (NY): Irvington Publishers; 1998. p. 211–48.
2.Trussell J. Contraceptive efficacy. In: Hatcher RA, Trussell J, Stewart F, Cates W, Stewart GK, Guest F, Kowal D, editors. Contraceptive technology. 17th ed. New York (NY): Irvington Publishers; 1998. p. 799–844.
3.Cates W, Raymond EG. Vaginal spermicides. In: Hatcher RA, Trussell J, Stewart F, Cates W, Stewart GK, Guest F, Kowal D, editors. Contraceptive technology. 17th ed. New York (NY): Irvington Publishers; 1998. p. 357–70.
4.United States Food and Drug Administration, Department of Health and Human Services. Vaginal contraceptive drug products for over-the-counter human use. Fed Regist 1995;60:6892–902.
5.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 examples. Br J Cancer 1977;35:1–39.
6.Raymond E, Dominik R, Spermicide Trial Group. Contraceptive effectiveness of two spermicides: a randomized trial. Obstet Gynecol 1999;93:896–903.
7.Fu H, Darroch JE, Haas T, Ranjit N. Contraceptive failure rates: new estimates from the 1995 National Survey of Family Growth. Fam Plann Perspect 1999;31:56–63.
8.Barbone F, Austin H, Louv WC, Alexander WJ. A follow-up study of methods of contraception, sexual activity, and rates of trichomoniasis, candidiasis, and bacterial vaginosis. Am J Obstet Gynecol 1990;163:510–4.
9.Jones BM, Eley A, Hicks DA, Patel R, Wordsworth JM. Comparison of the influence of spermicidal and non-spermicidal contraception on bacterial vaginosis, candidal infection and inflammation of the vagina: a preliminary study. Int J STD AIDS 1994;5:362–4.
10.Roddy RE, Cordero M, Cordero C, Fortney JA. A dosing study of nonoxynol-9 and genital irritation. Int J STD AIDS 1993;4:165–70.
11.McGroarty JA, Reid G, Bruce AW. The influence of nonoxynol-9-containing spermicides on urogenital infection. J Urol 1994;152:831–3.
12.Hooton TM. Pathogenesis of urinary tract infections: an update. J Antimicrob Chemother 2000;46(suppl 1):1–7, discussion 63–5.
13.Handley MA, Reingold AL, Shiboski S, Padian NS. Incidence of acute urinary tract infection in young women and use of male condoms with and without nonoxynol-9 spermicides. Epidemiology 2002;13:431–6.
14.Steiner MJ, Dominik R, Rountree RW, Nanda K, Dorflinger LJ. Contraceptive effectiveness of a polyurethane condom and a latex condom: a randomized controlled trial. Obstet Gynecol 2003;101:539–47.
15.Steiner MJ. Contraceptive effectiveness: what should the counseling message be? JAMA 1999;282:1405–7.
Back to Top | Article Outline


The Spermicide Trial Group consists of the following additional members: R. Wesley Rountree, MPH, Family Health International, Research Triangle Park, NC; Kurt T. Barnhart, MD, MSCE, University of Pennsylvania Medical Center, Department of Obstetrics and Gynecology, Philadelphia, PA; Lynn Bradley, MSCNM, Johns Hopkins Medical Services Corporation, Reproductive Medicine, Baltimore, MD; Mitchell D. Creinin, MD, University of Pittsburgh and the Magee-Womens Research Institute, Pittsburgh, PA; M. Wayne Heine, MD, University of Arizona Health Sciences Center, Division of Women's Research and Allied Studies, Tucson, AZ; Alfred Poindexter, MD, Baylor College of Medicine, Department of Obstetrics and Gynecology, Houston, TX; Livia Wan, MD, New York University School of Medicine, Department of Obstetrics and Gynecology, New York, NY; Mark Martens, MD, Minneapolis Medical Research Foundation, Hennepin County Medical Center, Minneapolis, MN (current affiliation: Department of Obstetrics and Gynecology, University of Oklahoma College of Medicine, Tulsa, OK); Robert Schenken, MD, The University of Texas Health Science Center at San Antonio, Department of Obstetrics and Gynecology, San Antonio, TX; Cate F. Nicholas, MS, PA, Vermont Women's Choice Program of Planned Parenthood, Burlington, VT (current affiliation: University of Vermont College of Medicine, Burlington, VT); Richard Blackwell, MD, University of Alabama at Birmingham, Department of Obstetrics and Gynecology, Birmingham, AL; David F. Archer, MD, Eastern Virginia Medical School, Department of Obstetrics and Gynecology, Norfolk, VA; and Melisa Holmes, MD, Medical University of South Carolina, Department of Obstetrics and Gynecology, Charleston, SC.

Cited By:

This article has been cited 3 time(s).

Obstetrics & Gynecology
Effects of Long-Term Use of Nonoxynol-9 on VaginalFlora
Schreiber, CA; Meyn, LA; Creinin, MD; Barnhart, KT; Hillier, SL
Obstetrics & Gynecology, 107(1): 136-143.
PDF (222) | CrossRef
Obstetrics & Gynecology
Noncomparative Contraceptive Efficacy of Cellulose Sulfate Gel
Mauck, CK; Freziers, RG; Walsh, TL; Peacock, K; Schwartz, JL; Callahan, MM
Obstetrics & Gynecology, 111(3): 739-746.
PDF (508) | CrossRef
Obstetrics & Gynecology
Contraceptive Efficacy of a Novel Spermicidal Microbicide Used With a Diaphragm: A Randomized Controlled Trial
Barnhart, KT; Rosenberg, MJ; MacKay, HT; Blithe, DL; Higgins, J; Walsh, T; Wan, L; Thomas, M; Creinin, MD; Westhoff, C; Schlaff, W; Archer, DF; Ayers, C; Kaunitz, A; Das, S; Moench, TR
Obstetrics & Gynecology, 110(3): 577-586.
PDF (520) | CrossRef
Back to Top | Article Outline
© 2004 The American College of Obstetricians and Gynecologists