Raymond, Elizabeth G. MD, MPH; Chen, Pai-Lien PhD; Dalebout, Sandra M. MPH
Emergency contraceptive pills are indicated for use after unprotected sexual intercourse to prevent pregnancy. The most effective regimen currently available in the United States consists of one dose of levonorgestrel, 0.75 mg, taken as soon as possible within 72 hours after unprotected sex, followed by a second identical dose 12 hours later.1 This regimen was approved by the United States Food and Drug Administration as a prescription product on July 28, 1999. Shortly after approval, the distributor of this product (Plan B, distributed by Women's Capital Corporation, Washington, DC) decided to apply to the Food and Drug Administration to market this regimen over the counter. The company sponsored two studies to collect data relevant to this application. The first evaluated women's comprehension of a prototype over-the-counter label for the product. That study, in which 663 women were interviewed in malls and family planning clinics in eight US cities, showed that by reading the label, most subjects could understand the key information necessary for safe and effective use of the product.2
This article reports the results of the second study, which was an “actual use” study designed to determine whether women would use the product appropriately and safely when it was dispensed using a simulated over-the-counter approach.3 This approach had three main components. First, the study subjects were not evaluated by a health care provider; each subject had to determine on her own by reading the package label whether or not the product was appropriate for her. Second, no unsolicited counseling or instruction was provided to subjects other than that on the package itself. Third, unlike in most drug studies, most study subjects paid for the product. The goals of the study were to assess use patterns, side effects, and pregnancies after use.
MATERIALS AND METHODS
This study was conducted between November 2001 and April 2002 at eight Planned Parenthood clinics located in or near five cities in the United States (Boston, Massachusetts; Grand Rapids, Michigan; Houston, Texas; Phoenix, Arizona; Seattle, Washington) and at five pharmacies near Seattle, Washington. The pharmacies had collaborative agreements with physicians, allowing pharmacists to dispense emergency contraceptive pills directly to women without a physician's prescription. The study protocol and the informed consent forms signed by subjects were approved by institutional review boards associated with each site and with Family Health International. Procedures followed were in accordance with the ethical standards for human experimentation established by the Declaration of Helsinki of 1975, revised in 1983. The Food and Drug Administration also reviewed the protocol, and a number of that agency's comments were incorporated into the study design and analysis plan.
During the study enrollment period, all study sites refrained from providing unsolicited counseling about emergency contraceptive pills or from evaluating any woman clinically to determine whether the treatment was indicated, unless she had declined to join the study.
The study admission procedures were implemented mostly by nonclinical staff. After providing brief oral information about the study, the staff invited women who presented with a request for emergency contraceptive pills to examine a sample study package (described below). Each woman was given a questionnaire on which she recorded demographic and study eligibility information and indicated whether or not she thought she should receive the product on that day. The staff reviewed her responses to assess the following study eligibility criteria: She wanted emergency contraception for her own use; she had not previously participated in a survey at a mall related to emergency contraceptive pills; she could read English; and she thought she should receive the product on that day. In two cities (Houston and Phoenix), women 15 years and younger were excluded from the study because the institutional review boards would have required parental consent for such women.
Staff gave each woman who met the criteria an informed consent form to read. This form included a statement that study participants would receive monetary compensation; this information had been deliberately withheld up to that point. The consent form did not include any details about the product, such as indication, contraindications, instructions for use, or side effects; this information was provided only on the package or in response to a specific question. Women who signed the consent form recorded additional background information and then were each provided one study package. Each subject paid the same amount for the package that the facility would have charged her for emergency contraceptive pills outside the study. The range of charges for individual subjects was $0 to $71. Each subject also received a data card on which to record information about use, side effects, menses, and pregnancy tests after using the product.
Site staff contacted each subject who received the product 1 week and 4 weeks later and asked about her use of the product (including the date and time when she had sex and when she took each pill dose), side effects, allergies, unusual bleeding, and evidence of pregnancy (menses, pregnancy tests, or other indications of pregnancy). Subjects were asked to refer to the data card during the contact, if available. Staff made additional contacts if a subject's pregnancy status was unclear or if she had unresolved adverse events at the last contact, or if she had used the product within 1 week before the last contact. At the first contact after use of the product, subjects were instructed to send their completed data cards to the site. After all contacts had been made, the sites provided compensation ($40 at the clinics; $45 at the pharmacies). In some cases, this compensation was less than the amount the subject had initially paid for the product. Women were allowed to enroll in the study an unlimited number of times.
Each study package contained two tablets of levonorgestrel (0.75 mg). The study package label had the same content as the approved prescription label for this product (Plan B), but the label was reformatted to be more understandable to users and to be compliant with Food and Drug Administration guidelines for over-the-counter drugs. Most importantly, the required “Drug Facts” panel was added to the back of the package. The packages were similar to the packages illustrated in our previous publication.2 No physician package insert was included.
The analysis was conducted according to a plan written before initiation of the analysis. The two primary outcomes were contraindicated use and incorrect use of the product. The operational definitions of these outcomes were based on the package label. Contraindicated use was defined as having occurred if the subject was pregnant (with fertilization having occurred at least 14 days before product use because this is the earliest time that a pregnancy would likely be detectable in a prescription setting), had unexplained vaginal bleeding before use, or was allergic to any ingredient in the tablets. Pregnancies were identified by an obstetrician who reviewed data from all subjects who had any indication of possible pregnancy. Dates of fertilization used in the analysis were the earliest of the dates estimated independently by three obstetricians after reviewing each pregnant participant's data. Pregnancy status was unclassifiable in some subjects (such as those without documentation of menses after product use but no other positive signs of pregnancy). Unexplained bleeding was identified by a gynecologist after review of subjects' reports of “unusual” bleeding and adverse events occurring before pill use. Incorrect use was analyzed according to multiple definitions. In the primary definition, which was mandated by the Food and Drug Administration, use was considered incorrect if the first tablet was taken more than 72 hours after intercourse or if the second tablet was not taken exactly 12 hours and 0 minutes after the first. Multiple secondary analyses allowed more realistic interpretations of this schedule; in the one presented here, which is the most conservative, use was defined as incorrect if the first tablet was taken more than 72 hours after intercourse or if the second tablet was taken more than 16 hours after the first. Some subjects' use patterns were not classifiable because of missing or inconsistent data. The proportions of subjects with contraindicated and incorrect use were estimated by dividing the number of subjects whose first use was contraindicated or incorrect by the total number of subjects with classifiable first-use patterns.
Secondary outcomes included repeat use, pregnancy, and adverse events. Pregnancy rates were calculated as the number of pregnancies per user among all subjects whose pregnancy status could be determined.
The incidences of the primary and secondary outcomes were calculated in subgroups defined by the following variables: eligibility status, age (16 years and younger, 17 years and older), race (nonwhite, white), ethnicity (Hispanic, not Hispanic), educational level (less than high school, high school or more), and prior emergency contraception use (ever, never). Statistical comparisons of subgroups were not made. In all analyses, data from the cards sent in by subjects were used only if the corresponding information had not been obtained by directly interviewing the subject. All confidence intervals (CIs) were computed using exact methods.
The sample size for this study was calculated to estimate the frequency of each of the two primary study outcomes, contraindicated and incorrect use of the product, with 97.5% CIs of ±5 percentage points around the point estimate, assuming that both proportions were truly 15%. These calculations indicated that 256 subjects who used the study product would be needed. To ensure diversity of the study population, allow for subgroup analyses, and account for potential underestimation of follow-up rates at each site, more subjects than needed were actually enrolled.
Altogether, 665 women were screened at least once for the study. Of these, 80 (12%) did not meet study entry criteria at first screening: 41 indicated that they were “not sure” that they should receive the product because of a desire for more information or various concerns about it, one indicated that she had not come to the facility for emergency contraceptive pills (although her other data suggested that she had in fact come for that reason), and the other 38 declined to sign the consent form. Nine of the 80 ineligible women were enrolled in the study because of a misunderstanding at the study sites. Nine of the 585 eligible screened women decided not to participate in the study after having signed the informed consent form. Thus, a total of 585 subjects received a study package at first screening. All of these women had signed the consent form.
The 585 enrolled subjects had a median age of 21 years at first screening. Twenty percent classified themselves as at least partly nonwhite, and a quarter had no education after high school (Table 1). Almost 40% had annual incomes of $15,000 or less. Forty percent had used emergency contraceptive pills at least once in the past, including 10% in the past 3 months (data not shown). The characteristics of subjects who were screened but did not receive a study package, those who provided any follow-up data, those who later reported having used the product, and those who were lost to follow-up were similar to the characteristics of the entire population who received the product (data not shown).
Ninety-two subjects (16% of those who received the study product) asked to speak with the clinician or pharmacist at the first screening. The most common questions that these subjects asked related to the safety of the product, including side effects and contraindications (34 subjects), the instructions for use (18 subjects), whether or not to take it with food (13 subjects), or other questions about the product (ten subjects) or about other contraception (ten subjects).
A total of 543 subjects (92.8%) provided follow-up data after the first screening. Three reported that they took neither pill. The main reasons for initial product use cited by the other 540 subjects were that a condom broke (45%) or that no contraception had been used during sex (40%). All of the reasons given for product use were consistent with the labeled indication for use of the product (emergency contraception). In most cases, the reason for pill use was also consistent with the reason that the subject gave at the admission visit for wanting to receive the pills.
Of the 540 subjects who used the study product provided at the first screening, 509 (94%) provided usable information about the timing of both the sex act that motivated first pill use and the ingestion of the first pill. These subjects took the first pill a median of 36 hours after the sex act. The range was 1 to 175 hours, except for two subjects who indicated that they took the first pill before the coital act that motivated use. Among the 523 subjects who provided usable information about the timing of ingestion of both pill doses after first screening, the median interval between doses was 12 hours, and the range was 0 to 36 hours. A total of 387 women (74%) took the pills exactly 12 hours 0 minutes apart, 51 (9%) took them less than 12 hours apart, and 22 (4%) took them more than 16 hours apart.
A total of 523 subjects provided sufficient data to allow a determination of whether or not the first use was contraindicated (Table 2). Seven of these subjects were considered to have possibly used the product counter to indications: One might have been pregnant at the time of use, and six had unexplained vaginal bleeding before use. The proportion of classifiable subjects whose first use of the product was contraindicated was 1.3% (97.5% CI 0.5%, 3%).
Of the 506 subjects who provided sufficient data to allow evaluation of correctness of first use according to the primary definition, 140 used the pills incorrectly by this definition. The most common error, made by 136 subjects, was failure to take the two pills exactly 12 hours, 0 minutes apart. Ten subjects took the first pill more than 72 hours after sex. The proportion of classifiable subjects whose first use was incorrect according to the primary definition was 28% (97.5% CI 23%, 32%). When incorrect use was alternately defined as taking the first tablet more than 72 hours after sex or the second tablet more than 16 hours after the first, the proportion of classifiable subjects whose first use was incorrect was 6.6% (97.5% CI 4.3%, 9.5%).
Of the 665 women screened for the study, ten (1.5%) were screened more than once. All ten received study product at each screening. The ten second screenings occurred between 4 and 80 days (median 8 days) after the first screening, and two third screenings occurred 14 and 26 days after the second screenings. In all cases, the reasons for the emergency contraception requests were consistent with the labeled indication for use of the product. Follow-up was accomplished after the repeat screenings of eight subjects, including both who had been screened three times. In each case that was followed up, the subject had used the product. None of the eight subjects had any contraindications to use at any of their recorded uses. Six of their 18 total uses were incorrect by the primary definition, all because they took the second pill other than 12 hours 0 minutes after the first. None of the repeat uses were incorrect by the alternate definition.
Ten subjects who used the product were found to be pregnant during the study. None of these subjects received the product more than once. As noted, one of the pregnancies may have been fertilized 14 or more days before product use; however, because estimating pregnancy onset dates may be inexact, we considered that all ten pregnancies may have occurred after use. None of the pregnant subjects had other contraindications to product use. All took the first pill within 72 hours after sex, and the second pill 12 or 13 hours after the first. For 516 subjects, data indicated that pregnancy had not occurred, and for 14 subjects, insufficient evidence was collected either to confirm or to exclude pregnancy. The proportion of subjects who may have had a product failure is ten of 526 classifiable subjects, or 1.9% (95% CI 0.9%, 3.5%).
Of the 540 subjects who used the study product, 246 (46%) reported at least one adverse event. None of the events were serious. Events reported by at least 5% of the population were abdominal pain (reported by 14% of subjects), nausea or vomiting (14%), headache (11%), and fatigue or weakness (8%). Subjects whose use was contraindicated or incorrect did not differ notably from the total population with respect to the frequency or nature of adverse events.
Only 31 subjects reported having consulted a provider after receiving the product. The most common reasons for consulting the provider were simply to inform him or her about product use (ten subjects), to discuss medical problems or side effects (nine subjects), or to discuss ongoing contraception (five subjects).
No important differences were observed in the incidence of the primary and secondary outcomes between subgroups of the population (Table 3).
In this study of emergency contraceptive pills provided with a simulated over-the-counter approach, nearly all subjects used the product appropriately and safely. The incidence of contraindicated use was extremely low (1.3%). The incidence of incorrect use was 28% by our primary definition, but clearly this definition was overly strict in allowing absolutely no variation in the interval between the two doses of medication. By an alternate definition that allowed up to a 4-hour delay in ingestion of the second pill, the incidence of incorrect use was 6.6%. No serious adverse events occurred, and the pattern of events that were reported was consistent with previous studies of the regimen. Fewer than 2% of subjects became pregnant. These findings were consistent across all subgroups examined. Particularly reassuring was the observation that potentially vulnerable groups, such as minors and less-educated women, were not substantially more likely than others to use the product in a contraindicated or incorrect manner and did not have notably higher risks of adverse events or pregnancy.
During the 3–4 months of study enrollment, six subjects used the study product twice and two subjects used it three times. Each of the repeat users reported a legitimate need for the product each time she used it. Considering that 7.5% of sexually active women at risk of pregnancy who do not desire pregnancy in the United States use no contraception4 and that more than 1.6% of condoms break or slip during use,5 the repeated need for the product in this study by eight of 540 users (1.5%) over the course of the 3-month enrollment period is not surprising. In a trial in Scotland in which emergency contraceptive pills were made available to women by prescription for a year, 42 of the 87 women (48%) who used the pills did so more than once.6
The proportion of subjects who became pregnant in our study (1.9%) is comparable to the proportions reported in two previous trials of the levonorgestrel emergency contraception regimen (1.1% and 2.9%).1,7 However, this comparison should be interpreted with caution because of differences in the baseline pregnancy risk of the three study populations. Both prior trials excluded women who had recently used hormonal contraception, whereas one quarter of our subjects used hormonal contraception in the month before admission. On the other hand, the two previous trials excluded women who had had more than one unprotected coital act before admission to the study. Our study had no such restriction.
Forty percent of our study subjects had previously used emergency contraceptive pills, which is likely a higher proportion than would exist in a real over-the-counter setting. Because these women presumably had received counseling about the method from a clinician at some point in the past, one might have expected that their outcomes would have been better than those of naive users. However, stratified analyses indicated that such was not the case. Thus, the inclusion of these women was not an explanation for our favorable results. Furthermore, our stratified results suggest that prior clinician counseling did not improve use of the product or reduce the likelihood of adverse events.
Only 7% of our subjects were lost to follow-up after receiving study product. This proportion is quite low, considering the sensitive nature of the treatment and the questions asked. The baseline characteristics of our lost subjects were indistinguishable from those of the total study population, and their use of the product was thus likely similar to that of the total population.
Our definitions of contraindicated and incorrect use were based on the Food and Drug Administration–approved prescription package label for the product. Recent data suggest, however, that this label may be excessively restrictive. The two pills can be taken at the same time without a loss of efficacy or an unacceptable increase in the nature or frequency of side effects.8,9 Also, although the method is more effective the sooner after sex it is used, it retains substantial efficacy as long as 120 hours after sex.8 Furthermore, no medical rationale exists for withholding this method from women with bleeding abnormalities or even from women who might be pregnant.10 If we altered our definitions in accord with these considerations, the proportion of incorrect use in our study would be even lower than presented here, and the proportion of contraindicated use would be zero.
Our study did not include a comparison group. Ideally, we would have conducted a randomized trial comparing outcomes of women who received the product in an over-the-counter manner with outcomes of women who received it in an actual prescription setting. However, we did not think that such a trial was feasible. Inadvertent contamination of the over-the-counter group would be difficult to prevent (and to detect) if both approaches were employed at a single site. More importantly, we were not confident that an actual prescription setting could be simulated in the context of a study, because study investigators are by necessity specially selected and trained. They may therefore provide better care than clinicians outside a study, leading to improved outcomes. As it turned out, the incidence of contraindicated and incorrect use in our study are both low enough on an absolute scale to be meaningful without direct comparison with results from a prescription setting.
In our study, we included as many key elements of the over-the-counter setting as possible: Subjects were not evaluated by a clinician for contraindications; clinicians did not provide unsolicited counseling; subjects were required to pay for the study drug; and subjects could return to the site for additional packages as often as desired. Subjects were permitted to ask questions of the clinician or pharmacist, as they could in a drugstore, but the ultimate decision of whether they obtained the product was theirs. In addition, the admission questionnaires and procedures were carefully designed not to give any unsolicited hints to subjects about proper product use. All of the study sites were facilities known in the community as places where emergency contraception was available. However, subjects obviously had to be informed that they were in a study and would be followed up, which may have influenced the care with which they read the instructions and used the product. Also, all the women in our study initially came to the study sites expecting to obtain treatment from a clinician. Therefore, we have no information about outcomes among women who might be newly attracted to use the product if it became available over the counter. Scant information is available about how well “actual use” studies of other drugs targeted for over-the-counter use predict subsequent use of these drugs in the real over-the-counter setting. Recent data on over-the-counter antifungal agents and smoking cessation products have been interpreted as showing that “actual use” studies overestimated the eventual success of these therapies,11–13 although these interpretations have been criticized.14–16
In summary, women do not need provider intervention to use the levonorgestrel regimen of emergency contraceptive pills safely and effectively. These findings provide further strong support for the proposal that this therapy should be made available to women without a prescription.10,17 Because this regimen may be more effective the sooner it is taken after sex, enhancing women's ability to obtain it quickly and easily over the counter could have a major benefit to women and to the public health.