Emergency contraception or oral postcoital contraception prevents pregnancy after unprotected intercourse. The Yupze regimen, consisting of elevated doses of combination oral contraceptive pills initiated within 72 hours of unprotected intercourse, reduced the risk of unintended pregnancy by at least 74% in clinical trials.1 About 3.5 million unintended pregnancies occur each year in the United States.2 The effect of increased emergency contraception could be substantial.
Emergency contraception for a woman who has unprotected intercourse, requires several steps to reduce the risk of unintended pregnancy. She must perceive that her risk of pregnancy is real and must be motivated to prevent it. She must be aware that a postcoital contraceptive method exists that could reduce her risk of conception. She also must have specific knowledge about how to obtain it and time its administration. She must have access to it, including a place to get it, a way to pay for it, and the time to invest in getting it. Any of those steps can be barriers to actually using emergency contraception after unprotected sex.
Most women age 18–44 years (66%) report general knowledge or “having heard of emergency contraception,” but few (1–2%) report having ever used it. Adolescents and minorities are less likely to know about it or use it.3,4 Most health-care providers consider it safe and effective; however, few routinely discuss it with patients, prescribe it more than a few times a year, or prescribe it prospectively.4 Some providers and policy makers believe that easy access to emergency contraception will encourage risky sexual and contraceptive behaviors, although there are few empirical data to support that idea.5
It is important to evaluate advance provision of emergency contraception to improve its use, given the limited time frame for use (within 72 hours of unprotected intercourse) and the higher efficacy of the regimen the earlier treatment begins.6 The only trial that evaluated the effects of increased accessibility on use was conducted in Scotland on women aged 16–44 years. The investigators found that advance provision of emergency contraception was safe and increased its use.7 Those results might not be generalizable to populations in the United States. Sexual risk behaviors, including unprotected intercourse, are far more prevalent in the United States among low-income urban youths, as are unintended pregnancies and sexually transmitted infections.2,8,9 Contraceptive method choice is likewise different, with women in the United States under age 30 years relying more on reversible methods, particularly condoms.10 More research in the United States focused on young women at high risk of unintended pregnancy is necessary to answer questions about potential risks and benefits of increased accessibility and to help formulate evidence-based standards of care to guide providers.
We conducted a controlled trial to examine the effects of education and advance provision of emergency contraception on knowledge and use of the method, as well as on contraceptive behavior, in a culturally diverse group of adolescents and high-risk young women who attended a publicly funded family planning clinic in San Francisco. Our hypothesis was that advance provision of emergency contraception with routine education would increase knowledge and use without compromising contraceptive behavior.
The study protocol was reviewed and approved by the Committee on Human Research at the University of California, San Francisco, and participants gave informed consent at enrollment. Participants attended an affiliated, publicly funded, family planning clinic of the Department of Obstetrics, Gynecology, and Reproductive Sciences of San Francisco General Hospital between June and November 1998. Women were eligible if they were 16–24 years of age, spoke English or Spanish, and were available for follow-up in 4 months. Exclusion criteria included pregnancy, use of contraceptive implants or intrauterine devices, presentation for emergency contraception, and medical contraindications to the use of oral contraceptive (OC) pills.
Participants were assigned on an alternating basis to one of two study groups. Women assigned to the treatment group received education and an advance provision of a single treatment dose of emergency contraception, and women in the control group received education on emergency contraception alone. Research assistants interviewed participants at enrollment and at follow-up using a questionnaire to obtain demographic information and to measure outcomes. To measure knowledge of emergency contraception, they were asked if they had heard of or knew what emergency contraception or the morning-after pill was. If they responded yes, they were asked five true-false questions that tested specific knowledge on what emergency contraception is, timing of administration, how to obtain it, protection from sexually transmitted diseases, and its main adverse effect, nausea. They were also asked about frequency and timing of emergency contraception.
We collected information on current contraceptive methods and patterns of use. Subjects answered multiple-choice questions about unprotected sex (intercourse without any form of contraception including condoms) and condom use in the past 4 months. The most consistent responses, respectively, were “I never had unprotected sex” and “I used condoms every time I had sex.” Oral contraceptive users were asked to describe the frequency of missed pills in the last 4 months with “I never missed pills” the most consistent response category.
A research assistant administered follow-up questionnaires either by telephone or at a subsequent clinic visit, 4 months after enrollment. A bilingual research assistant administered a translated version of the questionnaire to Spanish-speaking women and translated versions of all printed materials were used.
All subjects received education on emergency contraception from regular clinic counselors who followed a protocol information handout, and they were given the handout. Women in the treatment group received advance provision of emergency contraception from a clinic physician or nurse practitioner. The treatment regimen comprised eight OC pills that contained 0.15 mg of levonorgestrel and 30 μg of ethinyl estradiol given with verbal and written instructions to be taken in two divided doses 12 hours apart within 72 hours of unprotected intercourse. Women in the treatment group also received four 25-mg diphenhydramine tablets for nausea. Women in the treatment group completed clinic-specific consent forms for emergency contraceptive pills and received additional instructions with the medication, so they, and the research and clinical staff, could not be masked to the study arms.
We assumed a 5% baseline emergency contraception use rate, based on chart review at the clinic, and calculated that approximately 100 women would be needed in each study arm to have a 90% power to detect a 15% difference between treatment arms. We allowed for 20% attrition from enrollment to follow-up. All analyses were conducted using the intent-to-treat population, with all study subjects analyzed according to their initial group assignment.
We tested for any significant differences between treatment and control groups at enrollment, then analyzed prospective data to determine whether there were differences between treatment and control groups at follow-up. Contingency tables and χ2 statistics were used for significance tests with categorical variables, and t tests were used for continuous variables. We used multiple logistic regression analysis to determine the effect of advance provision of emergency contraception on use at follow-up. Two models were used, one controlling for current contraceptive method and pattern of contraceptive use at enrollment, and the other controlling for current contraceptive method and frequency of unprotected sex reported at enrollment. Odds ratios with 95% confidence intervals and P values are reported. Significance was at the .05 level. Data were analyzed using Stata 6.0 (Stata Corporation, College Station, TX).
Two hundred sixty-three women were enrolled, 133 to the treatment group and 130 to the control group. Follow-up interviews were completed for 213 (81%) study participants, 111 treatment and 102 control. An analysis of study attrition showed no significant differences in baseline traits of participants who completed the study and those who were lost to follow-up. There were no statistically significant differences in demographic or sexual characteristics between treatment and control groups at enrollment.
Information on the sample at enrollment is shown in Table 1. The mean age was 19 years, and most participants were adolescents (64%) and ethnic minorities (76%). Fifteen percent spoke Spanish as their primary language. Most participants presented for pregnancy testing (41%) or contraception (39%). For 40% of participants it was their first visit to the clinic.
About half (52%) the participants reported having been pregnant and 19% had children. Over one third of participants (38%) reported having had one or more elective abortions. About one in five (18%) had histories of prior sexually transmitted diseases (STDs). At enrollment, frequency of sexual activity varied. Five percent reported no sexual activity in the past 4 months, and over half (58%) reported having sex once a week or more. About one in five participants (22%) reported having a new sexual partner in the past 4 months (Table 1).
Condoms were the most frequently reported current contraceptive method at enrollment (43%), followed by OC (26%). Almost 20% of participants reported using no contraception. More participants in the control group (24%) than the treatment group (15%) reported using no contraception, but the difference was not statistically significant.
Subjects were asked about knowledge of emergency contraception and its use in the past 4 months. At enrollment, 68% had heard of or knew what emergency contraception was; however, only 15% correctly answered all five knowledge questions (Table 2). At enrollment, about 4% of women in both groups reported having used emergency contraceptive pills in the past 4 months.
At follow-up, essentially all participants reported having heard of emergency contraception. However, women in the treatment group (47%) were significantly more likely (P = .03) than women in the control group (32%) to correctly answer all five specific knowledge questions. Women in the treatment group (20%) were almost three times as likely as women in the control group (7%) to report using emergency contraception during the study period (P = .006).
We did multivariable analysis to determine whether the treatment group was more likely to use emergency contraception during the study, after we controlled for contraceptive method and patterns of use at enrollment. Results presented in Table 3 show that women in the treatment group still were almost three times as likely as women in the control group to use emergency contraception, after we controlled for current contraceptive method and either pattern of contraceptive use or frequency of unprotected sex.
In Table 4, participants are categorized by contraceptive method used at follow-up, compared with enrollment. Depot medroxyprogesterone acetate was considered the most effective method, followed by OC. All other methods, including condoms, spermicides, diaphragms, and withdrawal, were considered less effective than depot medroxyprogesterone acetate or OC. Not using a contraceptive method was considered the least effective.11
Only 7% of participants reported using no contraception at enrollment and at follow-up. Almost half (46%) reported using the same method at enrollment and follow-up. Subjects were as likely to switch to a more effective method at follow-up (22%) as they were to switch to a less effective method (24%). However, women in the treatment group (28%) were significantly more likely than women in the control group (17%) to report using a less effective method at follow-up (P = .05).
Patterns of condom and contraceptive use are presented in Table 5. At enrollment and at follow-up, there were no significant differences in the proportion of women in the treatment and control groups who reported that they never had unprotected sex (P = .92 and .42, respectively). The overall proportion of study participants reporting “never had unprotected sex” increased from enrollment to follow-up (33% versus 56%).
Although the differences were not statistically significant, women in the control group were more likely than women in the treatment group to report consistent condom and OC use at enrollment (P = .08). At follow-up, there was no difference in the proportion of women in the treatment and control groups who reported consistent condom use (P = .71). The overall proportion of subjects who reported consistent condom use increased from enrollment to follow-up (18% versus 49%). However, at follow-up women in the control group (58%) were significantly more likely than women in the treatment group (32%) to report consistent OC use (P = .03). The overall proportion of subjects who reported consistent OC use increased from enrollment to follow-up (34% versus 45%).
Women who were given emergency contraception in advance were significantly more likely to use it than women in the control group who received education only. To assess whether women in both groups initially were equally likely to have unprotected sex and were equally motivated to prevent pregnancy, we used multivariate analysis to control for differences in contraceptive methods, patterns of use, and unprotected sex reported at enrollment. For example, if women in the control group were more motivated to prevent pregnancy and used contraception more consistently, they might have had less need to use emergency contraception. Advance provision was the strongest predictor of emergency contraception use at follow-up, even after controlling for contraceptive method, pattern of contraceptive use, and frequency of unprotected sex. This finding implies that women who had unprotected sex were more likely to take action if they had emergency contraception on hand. Specific measures of motivation to prevent pregnancy would be helpful to get a more accurate estimate of the strength of intervention effect on use.
Our data also suggest that women can be educated effectively about emergency contraception with routine clinic counseling, but education alone might not be sufficient to increase use of this important contraceptive method. Although more women in the control group were knowledgeable about emergency contraception at follow-up, the proportion of women who used it at follow-up did not increase. Women in the control group who had unprotected sex had to go through traditional routes, ie, phone, clinic, or physician, to obtain emergency contraception. Studies on contraceptive education show that increased knowledge alone does not necessarily correlate with behavioral changes.12 It is possible that education alone might not increase use of emergency contraception, as shown in this study; however, it is also possible that the effects of education might be observed in other outcomes, such as unintended pregnancy rates. However, advance provision could eliminate access barriers that might prevent a knowledgeable woman from using emergency contraception.
Women given emergency contraception in advance had more specific knowledge and greater use of it. Clinic counselors followed the same protocol for educating women in both groups. Women in the treatment group received additional verbal and written instructions with the pills, which might have been a more powerful educational tool. Women in the treatment group might have had better retention of information because they used the pills.
To test our hypothesis that educating high-risk women about emergency contraception and increasing use would not compromise contraceptive behavior, we looked at changes in current contraceptive methods from enrollment to follow-up and reported patterns of use at follow-up. During the study, a number of contraceptive changes occurred, with about half (47%) the women in both groups reporting using a different contraceptive method at follow-up compared with enrollment (Table 4). That finding was consistent with other studies that showed discontinuation rates of OC and depot medroxyprogesterone acetate in adolescents as high as 50% at 6 months.13,14
Of concern is the finding that significantly more women in the treatment group reported using less effective methods at follow-up. It is possible that women in the treatment group relied on the advance provision of emergency contraception, which could be beneficial or detrimental depending on previous method and the pattern of use. For example, an individual who used OC inconsistently might benefit from changing to condoms with emergency contraception as a back-up. Longer term follow-up and measurement of unintended pregnancy rates would assess the net effect of changes in methods and increased use of emergency contraception.
A positive finding was that there was no significant increase in proportion of women in the treatment group who reported unprotected sex or decrease in consistent condom use at follow-up. More women in both groups reported not having unprotected sex and using condoms consistently at follow-up compared with enrollment. That finding might indicate a benefit associated with attending the clinic over the intervention, which is consistent with a previously reported benefit of clinic interventions on contraceptive behaviors.15 The proportion of women in the treatment group who reported consistent OC use was significantly less than the control group at follow-up (32% versus 58%), but the proportion of treated women who reported consistent OC use at follow-up was slightly higher than at enrollment (32% versus 25%). Changes to less-effective contraceptive methods and patterns of OC use are potentially negative and need to be explored in relation to observed benefits of advance provision of emergency contraception.
Bias might have been introduced because participants were assigned to groups systematically and not randomly. Although that design had logistical advantages, the predictability of the study group assignment might have permitted manipulation of the sequence of enrollment. However, comparisons of the treatment and control groups at enrollment did not find significant differences, and we also attempted to control for measurable confounders in the multivariate analysis. We attempted to limit unintended interventions and ascertainment bias by using standardized questionnaires and scripted guidelines for enrollment and education.
Our findings are based on self-reported data, which might not indicate actual behaviors; however, studies using the Youth Risk Behavior Surveillance questionnaires for adolescents have shown high reliability across an array of health-risk behaviors.16 To determine the relationship between advance provision of emergency contraception and risky sexual behavior more accurately, data should also include biologic markers of sexually transmitted disease acquisition and unintended pregnancy. Our sample was too small and the study was not designed to evaluate the effect of advance provision on pregnancy rates. A larger scale, randomized intervention that incorporates longer follow-up time and multiple doses of emergency contraception is needed to validate our findings and evaluate efficacy with respect to reducing unintended pregnancies. Our results also might not be generalizable to low-risk and nonclinic populations.