OBJECTIVE: To determine whether multiple courses of emergency contraceptive therapy supplied in advance of need would tempt women using barrier methods to take risks with their more effective ongoing contraceptive methods.
METHODS: We randomly assigned 411 condom users attending an urban family planning clinic in Pune, India, to receive either information about emergency contraception along with three courses of therapy to keep in case of need, or to receive only information, including that about the locations where they could obtain emergency contraception if needed. For up to 1 year, women returned quarterly for follow‐up, answering questions about unprotected intercourse, emergency contraceptive use, pregnancies, sexually transmitted infections, and acceptability.
RESULTS: Women given advance supplies reported unprotected intercourse at rates nearly identical to those among women given only information (0.012 versus 0.016 acts per month). Among those who did have unprotected intercourse, however, supply recipients were nearly twice as likely (79% versus 44%) to have taken emergency contraception, although numbers were too small to permit statistically significant inferences. No women used emergency contraception more than once during the study, even though everyone in the advance‐supplies group had extra doses available. All women found knowing about emergency contraception useful, and all those receiving only information wished they had received supplies as well.
CONCLUSION: Multiple emergency contraception doses supplied in advance did not tempt condom users to risk unprotected intercourse. After unprotected intercourse, however, those with pills on hand used them more often. Women found advance provision useful.
Condom users given emergency contraception in advance are not more likely to have unprotected intercourse; when such intercourse occurs, however, they more often take emergency therapy.
Population Council, Mexico City, Mexico (CE), K.E.M. Hospital, Pune, India (SA, KC), Office of Population Research, Princeton University, Princeton, New Jersey (AH, JT), and the Population Council, Johannesburg, South Africa (KB).
Address reprint requests to: Charlotte Ellertson, Population Council, Escondida 110, Colonia Villa Coyoacan, Mexico DF 04000, Mexico; E‐mail: firstname.lastname@example.org.
The David and Lucile Packard Foundation (Los Altos, California) funded the study.
Anna Glasier suggested several features of the study design. Martha Campbell and Beverly Winikoff reviewed the draft protocol and gave helpful comments. Sunil Patel, Annik Sorhaindo, and Sarah Thomas organized, checked, and helped to interpret the data. John Skibiak pointed out an interpretation error in the earlier literature critical to our own analysis. German Rodriguez guided our statistical analysis.
Received November 20, 2000. Received in revised form May 30, 2001. Accepted June 7, 2001.
Emergency contraceptives can prevent pregnancy even after unprotected intercourse. The best‐studied regimen (Yuzpe) consists of ordinary combined oral contraceptives containing ethinyl estradiol and norgestrel, taken briefly in a higher dose than daily regimens. Conventional clinical guidelines recommend a first dose within 72 hours after unprotected intercourse and a second dose 12 hours later.1
In most countries, emergency contraception is available only by prescription.2 However, recent public policy debates in the United Kingdom, France, the United States, and elsewhere are re‐examining this requirement, as are health researchers.3 The debates center not on the therapy's safety and efficacy, because these are established, but rather on women's potential patterns of using emergency contraception. Those opposing prescription requirements make several points. Women frequently have trouble locating a doctor within 72 hours, and some are too embarrassed to request help after an accident. The World Health Organization also reported recently that delays increase pregnancy risk eightfold when women started therapy 61–72 hours after unprotected intercourse, rather than within the first 12 hours.4 Not surprisingly, women having emergency contraceptive pills on hand start therapy sooner after unprotected sex than those who must seek, or even simply fill, prescriptions (Skibiak JP, Ahmed Y, Ketata M. Testing strategies to improve access to emergency contraception pills: Prescription vs. prophylactic distribution. Nairobi: Population Council, 1999).
Family planners supporting prescription requirements, however, cite different arguments. Some worry that convenience would lead women to take chances with, or even abandon entirely, their regular family planning methods. Emergency contraceptives, although much better than no treatment after unprotected intercourse, are less effective than any ongoing contraceptives. In addition, if used frequently rather than occasionally, the methods would entail the same contraindications as ongoing oral contraceptives, rather than the single existing and relative contraindication: existing pregnancy.5 At least some ongoing family planning methods (eg, condoms) also protect against sexually transmitted infections; emergency contraceptives do not.
One middle ground between requiring a prescription and selling emergency contraceptives over the counter is to give women prescriptions for the pills, or actual pills themselves, ahead of time during routine family planning visits.3 Our study explored the effects of giving women multiple advance courses of emergency contraception, and tracking the resulting activity for up to 1 year. We also collected explicit information about condom use (and unprotected intercourse), and we tried to gather preliminary data on sexually transmitted infections. Our hypothesis was that women having supplies on hand would not be more likely than women without such supplies to engage in unprotected intercourse. We suspected, however, that when unprotected intercourse did occur, women with supplies on hand would be more likely to use them than would women who knew about the method but had to seek the pills after an accident.
MATERIALS AND METHODS
With approval from the Population Council's institutional review board, we recruited 411 consenting barrier‐method users at a large urban family planning clinic in Pune, India. Condoms are frequently the method of choice there for postpartum clients, and many such women enrolled in our study. (Ours was not an efficacy trial). Women affirmed their intent to use their chosen barrier methods (all had selected condoms) during the study, and they agreed to quarterly follow‐up (either at the clinic or during home visits) for up to 1 year. Data collection ran from April 22, 1997 to June 30, 2000.
Using an individual randomization schedule prepared at the Population Council office in New York with statistical software (Statistical Package for the Social Sciences, v 6; SPSS, Chicago, IL), clinic staff assigned each participant to one of two groups. Women in the first group received only information (oral and written) about emergency contraception, consistent with the new standard practice in the clinic. The information explained what the therapy is, where to obtain it, and how to use it. Women in the second group received the same information, along with three courses of the Yuzpe regimen to keep in case of need. Each course consisted of eight tablets of a widely available combined oral contraceptive brand; each tablet contained 30 μg ethinyl estradiol and 30 mg norgestrel. Women were instructed to take four tablets within 72 hours of unprotected intercourse, and to repeat the same dose 12 hours later.
Concerning the randomization, a clerical error resulted in a string of women being assigned to the wrong study group. Twelve were incorrectly assigned to receive only information, and 16 incorrectly received information plus supplies. Of those who were wrongly assigned, only two women (both of them incorrectly assigned to receive information plus supplies) had unprotected intercourse during the study period. Neither used emergency contraception. Because the mistake was clerical, we do not suspect bias that would invalidate the study. To be sure, however, we did all analysis twice–once with the women the way they were treated and once with the women they way they should have been treated–and none of our conclusions were affected.
At each follow‐up interview, study participants answered questions about unprotected intercourse, emergency contraceptive use, and sexually transmitted infections. Women who had used emergency contraception characterized the circumstances leading to its need, the way in which they had taken the regimen (to assess correct use), any adverse effects they experienced, and their willingness to use emergency contraception again. Participants receiving advance supplies also assessed whether having emergency contraception on hand changed their propensities for condom use. All participants also reported their attitudes toward emergency contraception (particularly toward ready access) and noted whether they had discussed the method with others. Women who had or wished to become pregnant and those who switched to non‐barrier contraceptives left the study, as did women who could no longer comply with the follow‐up schedule.
We analyzed the data using Stata (v 6.0; Stata Corporation, College Station, TX). To evaluate differences between proportions, we used Fisher's exact tests. Z‐tests detected differences between means, and Wilcoxon rank‐sum tests analyzed the difference between two medians calculated from small samples. All significance tests were at 5%.
Figure 1 depicts the numbers of study participants who were randomized and followed in each study arm. Table 1 characterizes participants, whereas Table 2 summarizes follow‐up patterns. Participants receiving advance supplies completed more follow‐up visits on average than did women who received only information (P = .04). Loss to follow‐up was higher than we hoped (19%), but comparable to that (about 30%) from the only existing similar study with a comparably long follow‐up period.6 Because many participants were postpartum, child‐care duties may have hampered follow‐up. Women left the study mainly to switch to more reliable ongoing contraceptives than condoms (n = 96, or 23%). Of these, 58% obtained intrauterine devices (IUDs), 39% switched to oral contraceptives, and 3% had tubal ligations. Thirty‐eight women enrolled less than a year before the study ended and so could not possibly have contributed a complete year of follow‐up.
We excluded women who never completed any follow‐up from the remainder of the analysis. In addition, 24 women reported that they wanted to leave the study to have another child. We excluded these women from analyses of unprotected intercourse and emergency contraceptive use because we could not be sure the women wished to avoid pregnancy.
Among the relevant population for analysis, 23 women reported acts of unprotected intercourse during the study (Table 3). Although some women reported multiple acts of unprotected intercourse during a follow‐up period, no woman reported unprotected intercourse at more than one follow‐up visit. Women receiving information plus supplies were not significantly more likely than women receiving only information to have had unprotected intercourse, even adjusting for differing average lengths of study participation. Women with advance supplies, however, were almost twice as likely (79% vs 44%) to follow unprotected intercourse by using emergency contraception (Table 3). The difference was not, however, statistically significant, perhaps because of small numbers. No one used emergency contraception more than once.
Among the 15 emergency contraceptive users, only one took her therapy incorrectly. This member of the advance supplies group omitted her second dose. We observed no pregnancies, as expected among such a small number of users, and side effects were roughly consistent with the literature. However, 13% rated the regimen as unsatisfactory, and 27% said that they would not use it again. All of these women switched from condoms to IUDs. Of the four users given only information, three had obtained the pills from a medical store and one from the hospital.
Ninety‐eight percent of women given advance supplies said having supplies on hand did not make them take chances with their condoms. The three women who said their supplies had tempted them to take chances each reported unprotected intercourse during the first quarter of follow‐up. All three took emergency contraception after the unprotected intercourse. A “curiosity effect” might explain this behavior. One of the three women subsequently switched to oral contraceptives and left the study. The other two each reported in later follow‐up interviews that their advance supplies no longer tempted them to take chances.
Our attempt to study sexually transmitted infections, even in a preliminary way, was severely handicapped by the reliance on self‐reported data. Only one participant reported being treated for a sexually transmitted infection, with this infection occurring during the same follow‐up period in which she had reported unprotected sex.
Nearly all (99%) respondents said they would recommend emergency contraception to a friend or relative, if needed. Ninety‐six percent of participants reported talking about emergency contraception with friends or relatives. The rest said that either the topic never arose or that they were not allowed outside their houses to converse with others. No one reported having a friend or relative who had used emergency contraception during the study period, however. Having the supplies on hand did not make women more likely to talk with others about emergency contraception (P = .12).
Every respondent found it useful to know about emergency contraception. All women given advance supplies liked having emergency contraception readily available. All women receiving only information indicated at least once that they wished they had also received supplies.
Previously, only three published6–8 articles and one unpublished (Skibiak JP, Ahmed Y, Ketata M. Testing strategies to improve access to emergency contraception pills: prescription vs. prophylactic distribution. Nairobi: Population Council, 1999) study have explored advanced provision. The main limitation of these earlier studies is that each gave women only one packet of emergency contraception in advance. All but one also had very short follow‐up periods, and none attempted to track sexually transmitted infections.
The first and most influential study enrolled over 1000 Scottish women seeking abortions or emergency contraception itself. It randomized them to receive either information on emergency contraception or information supplemented by supplies.6 After tracking women for 1 year, researchers concluded that women given supplies were not more likely to use them repeatedly than women given just information. A problem, however, was that women received only one course of therapy. After using the pills once, women in the advance supplies group were identically situated to their counterparts in the information‐only group: all had to return to the clinic for supplies. Giving women multiple courses of therapy would have better approximated over‐the‐counter access. Directly measuring effects on unprotected sex, and on sexually transmitted infections, would also have been useful.
The second study, from Ghana,7 concluded that spermicide users given emergency contraceptive pills in advance were not more likely to have unprotected intercourse than those who received only information. Women in the Ghana study, however, were not randomized. Two of the four participating clinics offered advance provision, while two offered only information. In addition, the study was quite small (n = 211), followed women for just eight weeks, and collected no information on sexually transmitted infections. The study also provided only single packets of emergency contraception.
The third and most recent study,8 enrolled female adolescents (n = 263) in San Francisco, California, and randomized them to receive information plus supplies or information only. Although adolescents in the two groups reported identical frequencies of unprotected sex after 4 months' follow‐up, those given supplies were three times as likely to have used them when needed. Like the other studies, however, the adolescents received only single courses of therapy, and investigators did not track sexually transmitted infections.
A fourth study, conducted in Zambia, also gave women just one packet of emergency contraception in advance (Skibiak JP, Ahmed Y, Ketata M. Testing strategies to improve access to emergency contraception pills: prescription vs. prophylactic distribution. Nairobi: Population Council, 1999). The Zambian women were randomized to one of three groups: information only, information plus supplies, or information plus prescriptions for emergency contraception. Participants were followed for only 3 months. Although the Zambian study did enroll 50% condom users, the population most relevant to concerns about sexually transmitted infections, it also enrolled 50% oral contraceptive users.
We found no evidence that easy access to emergency contraception caused condom users to abandon their primary method. Women who received information about emergency contraception as well as three courses of therapy to keep on hand reported no more unprotected intercourse than did women who received only information. When women did have unprotected intercourse, however, those with supplies on hand were far more likely to use them (79% vs 44%). Perhaps because of small numbers, though, this finding is not statistically significant.
We were struck that fully one quarter of those with emergency contraceptives on hand did not use them when needed, perhaps because the method was very new in the population. Not one participant had heard of emergency contraception before the study.
Although the number of women who used emergency contraception was small, as noted, only one did not use her pills according to instructions. The vast majority (93%) used the regimen correctly, undermining arguments that physicians need to explain the rules exactly when administering the pills. Conveying this information when supplies are first provided appears adequate.
Nobody used emergency contraception more than once during the study, indicating that women do not appear to substitute emergency contraceptives for ongoing methods. Ours was designed explicitly to make repeat use as easy (or easier) than it would be under conditions of over‐the‐counter access, and yet we found no repeat use. Emergency contraceptives, either alone or as a back‐up to barrier methods, are apparently not attractive enough to negate other considerations, such as efficacy. The leading reason for leaving the study was a wish to switch to a family planning method more effective than condoms with emergency contraceptive backup. Interestingly, women with supplies on hand were slightly less likely to switch than were those who had only information to back their condoms in case of failure. Again, however, our numbers are small. Future studies should track this dynamic, as it bears on whether promoting emergency contraception as an adjunct to condoms can rationalize condom use for women worried about the condom's high contraceptive failure rate, but who are also fearful of sexually transmitted infections.
Nearly all (98%) women given advance supplies said the pills did not tempt them to take chances with unprotected intercourse. Indeed, unprotected intercourse (just under 0.5% of the estimated total number of acts) and emergency contraceptive use in our study were extremely low, even below figures expected, based on published condom‐breakage rates.1 Because only one sexually transmitted infection was reported, women's assessments seem plausible. A limitation is that we cannot distinguish those acts of unprotected intercourse caused by risk‐taking from those that resulted from broken condoms. Only women who used emergency contraception reported the circumstances of the unprotected intercourse. These women cited both broken condoms and lack of condom use as reasons for using emergency contraception, but we had no way to classify comparable acts of unprotected intercourse among women who did not use emergency contraception. Our most extreme assumption is that all of the reported unprotected intercourse resulted from no condom use. Even under this assumption, however, risk‐taking was extremely low.
Women clearly liked having emergency contraceptives readily available. All women who received advance supplies considered them useful; all women given only information wished they had also received supplies.