Friday, April 13, 2007

Population Effect of Increased Access to Emergency Contraceptive Pills

Ann alerts me to the fact that there are some questions about one of Dr. Trussell's studies, Population Effect of Increased Access to Emergency Contraceptive Pills.

As I mentioned to Ann, I don't see anything particularly noteworthy about it, but I don't have time now to get up to speed on who's claiming what about the study, so here are some excerpts to assist you with your own research:

Of the 3.1 million annual unintended pregnancies in the United States, most result from situations in which pregnancy risk is immediately evident: a condom breaks, birth control pills are taken inconsistently, or incorrectly, or no contraception is used at all.1 Increasing women's ability to obtain and use emergency contraceptive pills promptly after such events should improve their chances of averting pregnancy. A modeling exercise published in 1992 suggested that widespread use of emergency contraceptive pills had the potential to cut the number of induced abortions in the United States by at least half.2

During the past several years, this potential has been assessed in numerous studies that evaluated various strategies designed to enhance women's access to emergency contraceptive pills. Here we review the findings of these studies, focusing particularly on the strategies' effect on emergency contraceptive pill use and pregnancy rates. This paper will be supplemented by a planned review on this topic for the Cochrane collaboration, which will include statistical meta-analyses that estimate the combined effect of advance provision of emergency contraception on a variety of outcomes among all randomized controlled trials.3


In August 2006, we used the following strategies to search MEDLINE, POPLINE, EMBASE, and LILACS databases for studies that presented data comparing the effects of different levels of access to emergency contraceptive pills. The strategies included terms signifying emergency contraception combined (except for the LILACS database) with terms signifying access and had no time restriction.


We also examined reference lists from articles identified through the research and spoke with other experts in the field to find additional published or unpublished studies.


Our searches identified a total of 717 articles. Two authors (E.G.R. and J.T.) independently reviewed all information about these articles provided by the databases. We selected studies of any design that yielded primary data comparing effects of interventions or programs with different levels of access to emergency contraceptive pills. We excluded 694 from this review.......23 articles met our selection criteria. Two or three authors (E.G.R., J.T., and C.P.) then independently abstracted relevant data from the selected studies. We contacted authors of five of the studies to clarify the information presented in their publications. We resolved any discrepancies through consensus. We assessed the quality of the design, implementation, and analysis of each study, and when possible we tabulated the data on pregnancies and emergency contraceptive pill use.15 We did not perform statistical meta-analysis, because study design of our included studies differed substantially.


The 23 included studies were conducted in 10 countries, and all were reported between 1998 and 2006. Ten randomized trials and four cohort studies prospectively allocated women to either an increased access intervention group or a control group and followed up the women individually to ascertain outcomes (Table 1). These 14 studies enrolled a total of 13,564 women. In each of these studies, the intervention included provision of one or more free emergency contraceptive pill packages at enrollment that women could keep at home and use later if needed. The interventions differed in their aggressiveness....Thus, the difference in access between the intervention and control groups was less in some studies than in others.


Most of the 14 individually allocated studies had low power to detect even moderate differences in pregnancy rates between groups. Only four were designed with pregnancy as a planned primary outcome. In some of the studies, the study population was not at high risk for unintended pregnancy or abortion, as demonstrated by low rates in the control groups or nonintervention settings.20,21 Outcome ascertainment approaches also varied. Pregnancies were detected by participant report, supplemented in some studies by chart review and scheduled or indicated pregnancy testing. In four studies, pregnancies were not reported for the intervention and control groups. Data on emergency contraceptive pill use were obtained exclusively by participant report and was thus not verifiable. Planned follow-up duration ranged from 3 to 12 months. The proportion of enrolled women who did not complete the planned follow-up was very high (more than 30%) in at least four of the studies (Table 2).


An additional nine studies included in our review did not assign individual participants to intervention or control groups (Table 3). Rather, these studies compared population-level statistics in settings where the intervention was implemented to corresponding statistics in other times or settings. Interventions included legislation allowing distribution of emergency contraceptive pills over the counter or behind the counter from a pharmacist, provision of an advance supply of emergency contraceptive pills, implementation of a telephone prescription service, and introduction of a dedicated product by a pharmaceutical company. In the reports of these studies, information on factors other than the intervention that might have influenced the findings was limited or nonexistent. Some reports provided no evidence about the penetration of the intervention in the study community, so the proportion and characteristics of the population actually exposed was therefore unclear. These studies also suffered from other weaknesses akin to those noted for the studies using individual allocation, including populations at low risk for pregnancy, relatively easy emergency contraceptive pill access in the control settings, and possibly incomplete reporting of outcomes.


The specific emergency contraceptive pill regimens used in the 23 studies included the Yuzpe regimen (two doses of 100 mcg ethinyl estradiol plus 0.5 mg levonorgestrel taken 12 hours apart), the levonorgestrel-only regimen (two doses 0.75 mg levonorgestrel taken either together or 12 hours apart), and the mifepristone regimen (10 mg mifepristone).

None of the included studies found clinically or statistically significant differences between intervention and control groups in pregnancy or abortion rates (Tables 2 and 3). Of those that provided data on emergency contraceptive pill use, all but one9 showed that a substantially higher proportion of women in the intervention group than in the control group used emergency contraceptive pills. Several studies suggested that the intervention also increased promptness of use of emergency contraceptive pills,7,10,16,22,23,25 although one large trial did not confirm this effect.20


A substantial body of research is now available on the effects of improved access to emergency contraceptive pills. These data demonstrate convincingly that greater access increases use. However, to date, no study has shown that increased access to this method reduces unintended pregnancy or abortion rates on a population level. The specific interventions varied among the studies, and the quality of some of the studies was poor. Nevertheless, the consistency of their primary findings is hard to ignore.

Further research is clearly needed to inform funding agencies, family planning program managers, policy makers, and women in making decisions about spending limited resources on emergency contraceptive pills. One obvious question is prompted by the observation in many studies that poor access to the pills is not the only impediment to their use: even when women have emergency contraceptive pills available, they often fail to take them after the most risky coital acts. For example, four trials reported the proportion of pregnant women who took emergency contraceptive pills in the menstrual cycles in which pregnancy occurred—ie, who may have tried to prevent the pregnancy by taking the pills.16,20,21,24 In each trial, this proportion was less than 30% in both intervention and control groups. Qualitative studies have found that commonly reported reasons for nonuse of emergency contraceptive pills include failure to recognize the risk of pregnancy, neglect of perceived risk, stigma, and misperceptions about the method. Certainly, emergency contraceptive pills cannot reduce the risk of pregnancy unless they are actually used. Therefore, development and evaluation of strategies to overcome these barriers and maximize emergency contraceptive pill use seem imperative.

In addition, more precise estimates of the efficacy of emergency contraceptive pills are desirable. A recent analysis has confirmed that the levonorgestrel regimen, which is the form of emergency contraceptive pills most commonly used worldwide, is significantly more efficacious after a single coital act than no treatment. Specifically, this analysis suggested that we can be 95% confident that it reduces pregnancy risk by more than 23%.36 But just how much more remains poorly defined; the published efficacy figures calculated from currently available data on this regimen—on average, approximately 80%37,38—may overstate actual efficacy, possibly quite substantially.36,39 Clearly, if the method is only weakly efficacious, it is unlikely to produce a major reduction in unintended pregnancy rates no matter how often women use it. The best way to obtain reliable efficacy data would be by conducting a placebo-controlled randomized trial. Such a trial would need to be large to provide precise results, and would therefore be expensive. Furthermore, because it is already clear that emergency contraceptive pills are more effective than placebo, ethical issues and recruitment methods would need careful consideration.

Some have speculated that easier access to emergency contraceptive pills might promote risk taking, thereby offsetting the reduction in pregnancy risk when emergency contraceptive pills are actually used. However, the comparative trials have found almost no evidence of differences between intervention or control groups on either unprotected sex or compliance with primary contraceptive use.13,14,16,18–22,24 In addition, the three studies that collected data on sexually transmitted infections, an indicator of sex unprotected by condoms, also found no effect of the intervention.10,16,19 Additional research on this topic therefore does not seem a high priority.

The study's bottom line:

Because emergency contraception can safely reduce the risk of unintended pregnancy for individual women who use it, improved awareness of and access to the medication is certainly appropriate and desirable. Thus, the recent United States Food and Drug Administration decision to allow adults to purchase emergency contraceptive pills behind the counter is a welcome step for our country. However, previous expectations that improved access could produce a direct, substantial impact on a population level may have been overly optimistic. Ultimately, emergency contraception may contribute its greatest public health benefit indirectly, by providing an opportunity to encourage women who may be in a particularly receptive frame of mind to adopt a more effective contraceptive method or to use their current method more correctly and consistently. Numerous studies have found that women who use emergency contraceptive pills are subsequently likely to adopt more effective contraception,14,21,22,24,25 but rigorous research on this possibility remains to be conducted.

Labels: ,


At 2:27 PM, Anonymous Anonymous said...

ema - My concern about this piece is one I have about any systematic review - it's not original research. It is dependent on the ability of the authors to appropriately search for, identify, include or exclude, and interpret the results. To know for certain whether they are doing a good job with this piece, you'd really have to vet their search strategies and whether they are representing the conclusions of the original articles accurately. Additionally, the studies come from 10 different countries and had varying methods, which may reduce the applicability to the situation in the U.S. It may be that their conclusions are dead on, but those are the concerns I would have going into it.

At 4:57 PM, Blogger ema said...


I thought the problem was that some people were making unsubstantiated claims about the study--something to the effect that it "shows" that EC/Plan B doesn't decrease pregnancy rates and that Plan B doesn't work all that well, right? (Of course, as you noted, the study is a review, and it shows no such thing.)

In any case, I haven't had time to look at this properly, but I plan to.

At 4:28 PM, Anonymous Anonymous said...

Yes, I think some people may be making claims that this study doesn't fully support, but I wanted to point out that there may be methodological problems in how it was done in the first place. I can't help it - I'm a medical librarian. ;)


Post a Comment

<< Home