Evaluating the Effectiveness of Barrier Contraceptive Methods: Impact of Emergency Contraception

Afternoon Discussion

The afternoon discussion was a freewheeling dialogue. Designed to focus on a series of questions raised in the final morning presentation, the discussion evolved in a fluid way, with participants speaking from their own experience and interests. The originally posed questions merged in different ways, depending on the viewpoint and scientific issue that participants raised. The discussion, in the end, fell into five general categories, presented below. The sequence generally follows the practical decisions researchers have to make:

• what question to address
  
• how to provide emergency contraception (EC)
  
• how to calculate study results
  
• how study designs and analyses affect labeling
  
• whether a new study design is a viable option for consideration.

What Do We Want To Know about the Effectiveness of Barrier Contraceptives?

The group reached a near consensus that the most important function of a clinical trial of a barrier method is to assess the effectiveness of the barrier method alone. The effectiveness of the combined method of a barrier method with EC as a backup was a secondary goal. Since the primary purpose is to provide information to consumers on the effectiveness of barrier methods alone, researchers need to measure what some refer to as the method specific effectiveness (MSE).

Knowing the effectiveness of barrier methods alone is important. Women who use barrier methods may not want to use hormonal methods, including emergency contraceptive pills, so they need to know the MSE for barrier methods. Users of barrier methods may not know about EC or have access to it. Thus, MSE for barrier methods is the primary piece of information these users need.

However, the group also agreed that researchers cannot ethically prohibit a woman from using EC. Participants felt that researchers have to provide EC during a trial if a woman wants it. This ethical mandate exists even if a woman agrees at the beginning of the trial to use only the method being tested and not EC. Thus, measuring MSE will be difficult.

If barrier method effectiveness alone should be measured but EC should be provided during the trial, what information would a trial need to gather? The trial would need accurate data on use of the barrier method, on EC requests and use, on menses, and on consistent and correct use during all cycles, not just those when pregnancies occurred. Also, it should document and report what women were told about EC during the informed consent process and how they were counseled about EC. Some participants also expressed interest in seeing if the decision to request EC as a backup varied by when in the cycle unprotected intercourse occurred.

Even if detailed and accurate data on all coital acts and EC uses are collected, it may not be possible to extract meaningful estimates of the MSE from trials where EC is available (see How Should Researchers Calculate Study Results). Despite this warning, the majority of participants wanted to evaluate the MSE of barrier methods in trials where EC is provided as backup. The analysts were charged with the task of doing the best they could to extract meaningful MSE measures from trials where EC is used.

A long discussion took place regarding how to characterize the type of trial the group seemed to want. Would it be best to think of such a trial as an MSE trial? Would it be preferable to consider it as a combined method effectiveness (CME) trial, with a primary analysis that adjusts somehow for EC use? Would it be better to consider this type of trial to be a multiple option method (MOM) trial, with the options being either the barrier method alone or the barrier method plus EC? Finally, one participant commented: "We're getting caught up in semantics." The group concluded it wanted to get method specific effectiveness while allowing for use of EC.

Should EC be Provided in Advance or upon Request?

The group was about evenly divided on whether to provide EC in advance or upon request. Participants made strong arguments for both points of view and raised questions about research issues that need addressing.

One argument for advance provision concerned the issue of controlling the study environment. By providing EC in advance during a research trial, the study can standardize access to EC. If EC is not provided ahead of time, women may use EC in haphazard ways, depending on their perception of need and availability in the community where they live. By standardizing EC provision, study results may be easier to interpret.

Some participants argued that advance provision would make research results relevant to the real-world environment (at least in the United States) that women may encounter when the results are available in the future. Several proponents of advance provision felt that the availability of EC was increasing rapidly. Consequently, EC should be provided ahead of time in the trial to simulate what the practical world may be like when the study results are available. "We're trying to establish what the standard of care should be," said one proponent. (It is worth noting that this argument also reflects the group's desire to obtain meaningful measures of CME even though MSE was judged to be of primary interest.)

Skeptics of advance provision in a trial pointed out that women overseas will not have access to EC in the near future. Thus, the trial should not provide it in advance because this is not standard practice among providers. By providing it in advance, researchers would be adding an unnecessarily artificial element to the trial. Also, providing EC in advance might tend to increase its use, thus making the results for the MSE more difficult to calculate.

Participants expressed interest in studying how provision of EC would affect the use of barrier methods with EC as a backup. In fact, research on advance provision might be more of a service-delivery issue than a research question for a clinical trial. How to provide EC in a research trial in order to arrive at the most accurate measure of MSE in an ethical way may be a different issue than how we want to provide it in clinical practice.

A recent study by Glasier and Baird in Edinburgh, Scotland, examined advance provision as a service delivery issue. They compared a treatment group of 549 women who were supplied EC pills to keep at home with a control group of 522 women who could use EC by visiting a doctor in case of an emergency. After a year, almost half of the treatment group (47 percent) used EC at least once, compared to about one fourth (27 percent) of the control group. There were 18 unintended pregnancies in the treatment group compared to 25 in the control group. The women in the treatment group were not more likely to use EC repeatedly, and the two groups used other methods in a comparable fashion. "Making emergency contraception more easily obtainable does no harm and may reduce the rate of unwanted pregnancies," the study concluded. Research on the issue of advance provision of EC is currently under way in Ghana by FHI and in India and Zambia by the Population Council.

The topic of protocols about advance provision emerged, but the issues involved were not discussed in-depth. For example, one of the questions not addressed directly was the condition under which women should be told to use EC as backup. For example, should the provider tie this advice to when in the cycle the misuse, nonuse, or problem with the method occurred? If a trial provides EC in advance, researchers should not communicate that there is a "safe period" for unprotected intercourse when EC use is not necessary.

How Should Researchers Calculate Study Results?

Two general ways exist to calculate pregnancy probabilities in barrier method trials where EC is sometimes used. First, calculations can estimate the effectiveness of using a barrier method with EC as a backup if women choose to use it. This calculation approach leaves women who use EC in the primary analysis and does not have to make statistical adjustments. Put simply, it ignores the use of EC. If a woman who uses EC becomes pregnant, she will be counted as having a pregnancy. Otherwise, she will be counted as not having a pregnancy. Labeling of the barrier method would have to be adjusted, because use of EC would presumably lower the pregnancy rates in the trial. The consumer would have to be informed of that fact. Moreover, if EC is used frequently, this calculation approach provides more of a CME measure than an MSE measure.

The second general approach is to attempt to calculate a pregnancy probability for using a barrier method alone -- that is, to extract a measure that better reflects MSE. This requires a statistical adjustment for each time EC is used. Statistically, one counts all EC uses as a fraction of a possible pregnancy; the fraction would be equal to the probability of a clinically detectable pregnancy occurring after unprotected intercourse on the cycle day of the act leading to EC use.

A major concern of some participants is the potential for large-scale use of EC, which would cloud results so that they are not meaningful as a measure of MSE. In the only large-scale trial to date that has used a statistical adjustment approach, summarized by Terri Walsh of the California Family Health Council in the morning session, only 33 women used EC. Adjustment for EC use did not substantially alter estimated pregnancy probabilities. In that trial, EC was not provided in advance. Participants were informed about EC use at enrollment, but study staff neither encouraged nor discouraged EC use. If EC is provided in advance and discussed with participants more frequently, some think that much higher use may occur during a trial. This could result, as one participant put it, in "the tail wagging the dog." That is, the amount of EC use could be a greater determinant of the reported pregnancy probability than the issue at stake: the effectiveness of the barrier method itself.

One way to address the larger-scale use of EC is to stratify the women at the beginning of the trial according to their intent to use EC. The FHI spermicide trial being conducted in 11 U.S. research centers will use this approach in one secondary analysis. To estimate the effectiveness of spermicides alone, a secondary analysis will exclude women from the analysis who say at admission that they intend to use EC if they have a problem with the spermicide.

Neither the adjustment nor the stratifying approach will solve the statistical problem perfectly. The background paper by Dominik et al. provides a more detailed discussion of these issues.

How Do the Studies Affect Product Labeling?

Several important points emerged during this discussion. Researchers pointed out that the best we may be able to offer from trials is an effectiveness range rather than an exact percentage. Because of EC use, the best calculation may be a range. Participants brainstormed about what labels might say, depending upon the audience. A women's advocate pointed out that men would not be interested in labeling on condoms about EC use because women use EC. A participant from the FDA suggested that it might be useful to do research to see what women and men want to see on a label.

Several participants from the FDA also made the point that they were open-minded about how to approach labels for new barrier methods. They welcomed advice from the research community on how to design and analyze barrier method trials and seemed to welcome suggestions for a standard protocol for barrier method trials. It is important to devise something that is reasonable, even if it is not what the FDA is currently using.

Research designs need to consider what the future is likely to be, because the study results may affect labeling in the future. Study designs could even affect how a manufacturer chooses to market a condom or spermicide product -- whether, for example, the marketing is linked with EC.

Is the True Efficacy Study a Viable Option for Consideration?

The participants discussed the potential role in trials of the true efficacy design described in a morning presentation. A true efficacy trial is a randomized trial comparing a barrier method to no method (or to another barrier method); it includes women who desire pregnancy but are willing to delay pregnancy for a short period. The group was generally optimistic about the future of this research approach.

Several participants questioned whether the ease of using a method would affect results. As currently proposed, the trial records use of a barrier method during just one cycle. Is just one use enough for measuring true efficacy, or would more practice in using the method correctly be needed? For example, a spermicidal product might be more difficult to use correctly than a condom. One suggestion was to lengthen the study to two cycles rather than one. Another was to allow a practice period with the barrier device. Others pointed out that this approach would not allow for testing other issues, such as safety, which require longer-term use. Participants thought it would be useful to try this approach with the products being used in the spermicide trial that FHI is now coordinating.

return to table of contents