Sponge versus diaphragm for contraception: a Cochrane review

1. Introduction 

Although vaginal contraceptives have been used for millennia, little is known about their contraceptive efficacy [1]. Contemporary methods include spermicide, the female condom, the cervical cap, the diaphragm and the contraceptive vaginal sponge. The Today sponge (VLI Corporation, Irvine, CA, USA) is a single-use, soft polyurethane device approximately 2.5 cm thick with a diameter of 5.5 cm and an indentation on one side to facilitate its placement against the cervix. The sponge is designed to prevent pregnancy primarily through the gradual release of the spermicide nonoxynol-9 and, to a lesser extent, by blocking the cervix and absorbing semen.

Advantages of the vaginal sponge and the diaphragm include a lack of hormonal side effects and medical contraindications; control of the method by the woman rather than by her partner; and the ability to insert the device in advance of intercourse and to leave it in place for repeated acts of intercourse. Both devices can be used for more than one coital act within 24 h, but, unlike the diaphragm, the sponge does not require the insertion of additional spermicide for repeat coitus. The sponge also has the advantage over the diaphragm of not requiring a prescription or fitting from a health care provider. Although the cost of the diaphragm and its fitting make the device more expensive than the sponge initially, the diaphragm can be used for years. The sponge is relatively inexpensive, but the recurrent cost of the disposable method could be prohibitive for some. The sponge and diaphragm must be used correctly and consistently with each act of intercourse in order to be effective in preventing pregnancy. Also, some women consider the two contraceptive methods intrusive. Limited evidence [2] suggests that the sponge, as well as the diaphragm [3], could increase the risk of toxic shock syndrome. Furthermore, high concentrations of nonoxynol-9 in the vagina from sponge or diaphragm use could potentially cause adverse effects [4].

Participants in the two large trials of the sponge were instructed to leave the device in place for no more than 48 h. However, when the US Food and Drug Administration approved the device for use in the US, they used a more conservative limit of 24 h due to a lack of data from both trials relating to the duration of sponge use. The recommended 24-h limit is not based on data showing an increase in pregnancy or complications related to longer use, but is due to a lack of evidence.

The Today sponge has been approved for sale in numerous nations including Singapore (1981); the United Kingdom, the Netherlands and Norway (1982); and the United States and Switzerland (1983). The sole US manufacturer ceased production of the Today sponge in 1995 after the US Food and Drug Administration imposed new manufacturing standards. However, the Today sponge was purchased by Allendale Pharmaceuticals in 1998 and the sale of the product in the US is expected to resume.

2. Materials and methods 

The objective of the present review was to compare the contraceptive efficacy of and continuation with the vaginal contraceptive sponge impregnated with nonoxynol-9 vs. the latex diaphragm used with spermicide. The primary outcome measures of interest were overall pregnancy rates and discontinuation rates. Neither the randomized controlled trials of the sponge nor this review included sexually transmitted infection protection as an outcome measure [5].

We included all randomized controlled trials comparing the vaginal contraceptive sponge and the diaphragm with spermicide. Trials included women of reproductive age who wanted contraception with either the sponge or diaphragm. The Today contraceptive vaginal sponge and its predecessor, the Collatex sponge, are moistened with water to activate the spermicide before being inserted by the user in the vagina and positioned over the cervix. The sponge releases about 125 mg of nonoxynol-9 over a 24-h period of use. It can be inserted in advance of intercourse and should remain in place for a minimum of 6 h after intercourse and a maximum of 24 h after insertion. It does not need to be replaced or supplemented with spermicide for any additional coital acts. The sponge has a polyester loop to grasp for its removal. Any latex diaphragm assigned for use with spermicide could be the comparison method.

We searched MEDLINE via PubMed back to its inception using the following search strategy: (vaginal diaphragm OR (contraceptive devices, female[mh] AND diaphragm)) AND (contraceptive sponge OR (contraceptive devices, female[mh] AND sponge)).

We searched Popline (CD-ROM) using the following strategy: (global)(vaginal diaphragm OR (diaphragm AND latex)) AND (vaginal sponge OR sponge).

We searched LILACS using the following strategy: contraceptive agents, female [Words] and diaphragm [Words] and sponge [Words].

We also searched EMBASE using the following strategy: (contraception AND diaphragm) AND ((contraception AND sponge) OR Today sponge).

We supplemented the computer search with the reference lists of review articles and textbook chapters on barrier contraceptives. We also searched the Cochrane Controlled Trials Register. In addition, we contacted the authors of randomized controlled trials in this field to identify other reports we may have missed.

Two reviewers evaluated the titles and abstracts found during the literature searches, and we photocopied all potentially relevant articles. Each reviewer then independently examined each study for possible inclusion and assessed the methodological quality using the Cochrane guidelines. We focused on the method of randomization, the use of allocation concealment, the use of blinding, and the exclusion of participants after randomization.

After the inclusion of the study, three reviewers (Kuyoh, Grimes and Gallo) abstracted the data. No disagreements about the inclusion of studies or the abstracted data occurred. We contacted an author of one included study [6] for details about this and the related trial [7] sponsored by the same organization. We asked for supplementary information about the method of randomization and allocation concealment. We then entered the data into RevMan 4.1. We calculated Peto odds ratios using women as the denominators with 95% confidence intervals for the outcomes overall pregnancy and discontinuation of the method. We did not include in our analyses women who were excluded in the original trials regardless of the appropriateness of these exclusions.

3. Results 

Two randomized controlled trials [6], [7], both sponsored by Family Health International, met the inclusion criteria. The larger trial [6] recruited women from 13 clinical sites in the US; these data were part of the New Drug Application to the U.S. Food and Drug Administration for the contraceptive sponge. The trial compared the Today sponge with a diaphragm (All-Flex, Ortho Pharmaceutical Corporation, Raritan, NJ, USA), used with a contraceptive cream containing nonoxynol-9 2% (Ortho Creme, Ortho Pharmaceutical Corporation). Follow-up was planned every 3 months, with the 6- and 12-month follow-up to be clinic visits.

The smaller trial [7] recruited women from only one clinical site, the Margaret Pyke Center in London, UK. Although the Collatex sponge was the predecessor to the Today sponge, the version of the Collatex sponge evaluated in this trial was identical to the Today sponge. The arcing spring diaphragm had not previously been used in this clinic and was provided by Family Health International. Nonoxynol-9 cream 2% was used as the spermicide with the diaphragm. Follow-up visits were scheduled at 1, 3, 6 and 12 months.

Both trials supplied participants with a free supply of the devices. Thus, the trials were not designed to measure method discontinuations related to cost.

Both trials had similar methods. Randomization was done centrally using random permuted blocks based on computer-generated random numbers. The block sizes, and any possible variations in sizes, are unknown. Allocation concealment was accomplished by use of method indicator cards, contained in sequentially numbered, sealed, opaque envelopes. Whether these envelopes were impervious to deciphering is unclear. Neither published report described an a priori hypothesis or sample size and power calculation.

The larger US trial was originally described in an article by Edelman et al. in 1984 [6]. The cases categorized as “unprocessable,” the pregnancies, and the protocol deviations and inconsistencies were later reanalyzed and submitted as an addendum to the final report to the US National Institutes of Health [8]. Since we believe the addendum was based on more appropriate analytic decisions, we used the data from this source instead of from the original article by Edelman and colleagues. However, the reanalysis in the addendum still excluded 13 women (9 sponge users and 4 diaphragm users) because of eventual discovery of ineligibility due to pregnancy at enrollment. These exclusions were inappropriate since the failure to conform to an intent-to-treat analysis can lead to biased results [9]. Loss-to-follow-up rates were high in both groups: 35.2% of the sponge users and 43.6% of the diaphragm users.

In the UK trial, 15 women were excluded inappropriately from analysis after randomization because they never began their assigned method. Two participants were dropped from analysis because of eventual discovery of ineligibility: one sponge user (who had a pregnancy) and one diaphragm user (who did not become pregnant). Loss-to-follow-up was small; only one sponge user lacked follow-up information.

The sponge proved inferior to the diaphragm in terms of both efficacy and continuation rates. Both trials found higher rates of pregnancy with use of the sponge. The 12-month cumulative life table rate of pregnancy per 100 women was 17.4 for the sponge versus 12.8 for the diaphragm in the larger US trial [8] and 24.5 for the sponge versus 10.9 for the diaphragm in the UK trial [7]. The odds ratio of pregnancy for the sponge compared to the diaphragm was 1.5 (95% CI 1.1–2.1) in the US trial [8] and 2.6 (95% CI 1.3–5.4) in the UK trial [7].

Discontinuation rates were also higher with the sponge than with the diaphragm in both trials. In the US trial [8], the 12-month cumulative life table discontinuation rates per 100 women were 46.1 and 42.7 for the sponge and diaphragm, respectively. In the UK trial [7], the corresponding figures were 44.7 and 52.0. For both trials combined, the odds ratio of discontinuation was 1.3 (95% CI 1.1–1.6) for the sponge when compared with the diaphragm.

Allergic-type reactions were more common with the sponge as well. In the US trial [8] the 12-month cumulative life-table discontinuation rate for allergic-type reactions per 100 women were 4.0 versus 0.7 for the sponge and diaphragm, respectively. The corresponding figures from the UK trial were 0.9 and 0.0 [7].

Rates of discontinuation for discomfort differed in the two trials. In the US trial [8], the 12-month cumulative life table discontinuation rate for discomfort per 100 women was higher with the sponge (6.1 versus 2.8), while in the UK trial the pattern was reversed (4.8 versus 7.5). For other outcomes, both methods had similar results.

4. Discussion 

The contraceptive sponge proved less effective as a contraceptive than the diaphragm. This difference was statistically significant (p < 0.05) in each trial when analyzed in life table fashion.

A subgroup analysis of failures with the sponge led to a lengthy debate about the role of parity. McIntyre and colleagues analyzed the data from the US trial and found that while discontinuations due to pregnancy were comparable among nulliparous women using either method, they were markedly higher among parous women using the sponge [10]. Other observational data [11], [12] suggested that parity did not influence the risk of failure with the sponge. Instead, observed differences were attributed to “motivational factor related to future childbearing intentions” [13]. In the smaller trial [7], failure rates were high with the sponge in both nulliparous and parous women, and higher than the diaphragm in both groups. Since none of these subgroup analyses were planned in advance, they should be considered only hypothesis generating, at best. Moreover, subgroup analyses from randomized controlled trials usually hold little meaning [14]. These secondary data analyses should not detract from the overall finding of poorer efficacy with the sponge.

Discontinuation rates at 1 year were about 30% higher with the sponge than with the diaphragm. Among the allergic-type complaints associated with the sponge were dermatitis, erythema, and irritation; the predominant discomfort-related complaint was vaginal itching [8]. In the UK trial, more women discontinued use of the diaphragm because of messiness than discontinued the sponge for this reason (five women versus none). Subsequent survey research [15], [16] indicated that reasons for trying the sponge included its perceived effectiveness, convenience, comfort, and unobtrusiveness. Reasons expressed for discontinuing the sponge included accidental pregnancy, irritation or other vaginal discomfort, and “vaginal infection.”

The contraceptive sponge appears significantly less effective than the diaphragm, and its 12-month discontinuation rate is higher as well. Despite suggestions that nonoxynol-9 from the sponge may reduce the risk of some sexually transmitted diseases [5], [17], [18], a protective effect of this spermicide appears unlikely [19], [20]. Furthermore, the potential adverse effect of nonoxynol-9 in high concentration in the vagina from sponge or diaphragm use remains a concern [4]. Although trial recruitment might be difficult, further research regarding the contraceptive efficacy of the sponge should compare the sponge to spermicide alone.