Despite the availability of safe and effective contraception, approximately half of all pregnancies in the United States are unplanned or unintended, and nearly half of these end in pregnancy termination.1–5 Women who continue unintended pregnancies are less likely to seek early prenatal care, more likely to unknowingly expose the fetus to potentially harmful substances, and more likely to have a child at risk for low birth weight, developmental delay, and abuse.6 In particular, the unintended pregnancy rate among United States adolescents remains the highest in the industrialized world.7,8 Although efficacious methods of contraception are available, women seeking to use them encounter many systematic and individual barriers to long-term effective use.
Although United States teens are using contraception more,9,10 fewer adults are using contraception regularly.3,10 Currently, half of all unintended pregnancies in the United States occur in the 10% of fertile women who use no contraception.11–13 Among adolescents and adult women together, half report having used contraception in the month in which they conceived.3,7,13,14 This suggests that a number of factors must be addressed to reduce the number of unintended pregnancies. A model of feasibility for hormonal contraception should include 1) the access to therapy, 2) acceptability of therapy, 3) compliance with therapy, and 4) tolerability for continued use. Furthermore, feasibility concerns will likely differ between these two groups of contraceptive users and nonusers.
Access to contraception refers to the ability of the patient to receive therapy and depends on system-level factors such as insurance coverage and sociocultural norms.13,15 Both those starting and those continuing contraception may face access barriers. Current users may experience barriers preventing continuation of therapy. Nonusers of contraception, defined as those who have not regularly used contraceptive therapies for a minimum of three continuous cycles, may experience obstacles to initiation of therapy. Acceptability of therapy is another barrier for nonusers. Women initiating hormonal therapy must find a contraceptive fit for their lifestyle, personal habits, and personal diligence or exactness of routine. Acceptability is assumed to be established after three continuous cycles of use.
Feasibility must also include discussion of negative adverse effects. Such adverse effects lead to discontinuation of therapy, or nontolerance. Pharmaceutical manufacturers have been focusing on formulations and delivery methods to increase tolerability by developing newer contraceptive methods such as “the patch,” “the ring,” and triphasics. Better adverse-effect profiles may make contraceptive therapy more feasible over the long term.
Lastly, long-term feasibility of hormonal contraception must include data on compliance with the method. Actual use efficacy rates are proxy measures of compliance. A significant proportion of women are noncompliant in their consistency or correct usage of contraceptive methods. Thus, ideal use and typical use failure rates diverge. For example, oral contraceptives are extremely effective when used perfectly, but 15% of oral contraceptive pill users report difficulty taking the pill correctly and regularly,16,17 and an estimated 22% miss two or more pills per cycle.18,19 Typical use of the oral contraceptive pill results in a failure rate of 5–18%.7,18,20 Compliance with hormonal therapy is therefore critical to long-term feasibility and may be better served by a method that separates efficacy from behavior.
The contraceptive patch is one therapy that has shown substantial promise for increasing feasibility of hormonal contraception. Studies have reported good compliance in adults21–23 and older adolescents,7,24,25 although social desirability bias, resulting from medial portrayals of the patch, may cloud the picture. Previous trials have also examined compliance with the contraceptive patch and reported high compliance rates in controlled populations, although none has shown a significant improvement in unintended pregnancy rate.22,23 Other research has demonstrated acceptability of the patch among adolescents.24,26,27 Such favorable initial reports make the contraceptive patch an ideal candidate for use in a high-risk population, where improved feasibility may reduce the rate of unintended pregnancy.3,4,28,29
Our aim in this study was to evaluate the feasibility of the contraceptive patch in a community-based prospective cohort study among contraceptive-naïve women at high risk for a subsequent unintended pregnancy and abortion. We also hoped to identify factors associated with effective use of the patch as compared with the pill.
MATERIALS AND METHODS
We gathered data from three Planned Parenthood clinics representing approximately 95% of the total clinic volume seen by Planned Parenthood of the Rochester and Syracuse Region, encompassing a wide geographic catchment area. Clients attend these clinics predominately for contraception, routine gynecologic care, pregnancy termination, and rape crisis services.
Eligible subjects were women of any age seen for contraceptive counseling and who were prescribed either the patch (Ortho Evra; Ortho-McNeil Pharmaceutical, Raritan, NJ) or the pill (any brand) between November 2003 and March 2005. To achieve a sample population with as little selection bias as possible and because women change contraceptive methods for a variety of reasons, we limited our study sample to women who were contraceptive naïve or first-time users of hormonal contraception. Data were collected through the standard clinic forms, which were modified for this study. Data were collected on sociodemographic factors, insurance coverage, and obstetric history. We collected data prospectively to capture dates of follow-up visits, tolerance, and compliance. For cases of discontinuation and/or adverse events, we recorded access to therapy concerns, any adverse effects prompting discontinuation, compliance problems, and any reported pregnancies.
Bivariate analyses were performed using χ2 tests. Patients were categorized as high risk if at risk for a subsequent unintended pregnancy and abortion. High-risk subjects were defined as those who were under age 16 years, who were nonwhite and/or Hispanic, who had government health insurance, and/or who had a prior pregnancy termination.4,30 Age was categorized into tertiles. Parity and the number of prior pregnancy terminations were categorized as none, one, or more than one.
Insurance plans were categorized into self-pay, private insurance (including Blue Cross-Blue Shield), Medicaid, and the Family Planning Benefit Program. The New York State Family Planning Benefit Program is similar to Medicaid but allows for a higher income limit of up to 200% of the Federal Poverty Level. Because health services at Planned Parenthood are guaranteed regardless of ability to pay, those with no insurance were provided contraception at little to no cost, based on a sliding fee scale tagged to self-reported income. Maximum cost for a month of pills was $34 and $43 for the patch. For those under age 18 years who preferred that their parents' insurance not be billed, contraception was prescribed confidentiality based on self-reported personal income. These subjects were also coded as self-pay.
The primary outcome measure was time in days to discontinuation of the patch or pill. Failure rates were assessed based on those who chose to discontinue the patch or pill. Survival analyses with life tables and Cox proportional hazards were used to assess acceptability and compliance. Life tables were used to account for loss to follow-up and competing reasons for termination of therapy. Based on actuarial tables, this method calculates a separate failure rate for each month of contraceptive use rather than the overall rate for the study. Subjects contribute data so long as they remain in the study and report no contraceptive failures. Cox proportional hazards were used to assess confounding and effect modification by age, black versus white race, insurance, student and work status, parity and abortion history, high-risk status, and prior contraceptive use. Risk factors for early discontinuance of the oral contraceptive pill and contraceptive patch were identified in this manner. Stepwise, backward, and forward selection procedures yielded the same model for discontinuance assessed by the likelihood ratio. Assumptions of the Cox proportional hazard function were checked by examining Schoenfeld residuals.
We also included other proxy measures of acceptability and tolerability in the study design. The initial dropout or loss-to-follow-up rate served as a proxy measure for acceptability. Acceptability was defined as continued use beyond the first three cycles; we used 70 days as a cutoff value. Tolerability was assessed by the adverse-effect profile. We recorded reasons for discontinuation and the rate at which subjects switched to other contraceptive therapies. The primary compliance outcomes were percentage pregnancy rate and time to discontinuation as described by the survival curves. All analyses used two-sided tests with α=0.05 and were performed with SAS 8.02 (SAS Institute, Cary, NC). Pearl indices, the number of pregnancies per 100 woman-years of use, were also calculated for each contraceptive method.31
The medical staff was not informed of the study a priori to minimize Hawthorne effects, although the clinical history forms were modified. Clinicians provided contraceptive counseling according to standard clinical practice, without intervention by the investigators or Planned Parenthood administration. Subjects self-selected their contraceptive method. Effects by clinic location were examined as a safeguard against bias by clinic staff in all multivariate models. Planned Parenthood of the Rochester and Syracuse Region and the University of Rochester Internal Review Board approved this study. This article does not necessarily reflect the views of Planned Parenthood Federation of America or that of the above local affiliate.
One-thousand two-hundred and thirty persons aged 12–44 met our inclusion criteria. Most (89%) of the study population fell under our a priori definition of being high risk for an unintended pregnancy. The subjects were a racially diverse group of primarily single women. Just over half the sample described themselves as being a student in some capacity. Of the students, 11.4% worked full-time and 48.1% were unemployed. Of those not students, 37.8% worked full-time and 40.9% were unemployed. Insurance and work status were significantly correlated (P<.001), and no effect modification was seen by age, race, or marital status.
Although more than half of the women in our study described themselves as employed, only 22.5% of these were able to use a private insurance plan to obtain contraception. One third of those who were employed relied on self-payment to obtain contraception. Half of the women in the sample reported using condoms for birth control at least some of the time. Tables 1 and 2 contain descriptive statistics.
Thirty-eight percent (n=465) of the women had no further follow-up in any of the clinics after their initial visit in which they were prescribed the contraceptive patch or oral contraceptive pill. The initial loss-to-follow-up rate was higher among those using the patch (45.2% versus 29.5%, P<.001). Acceptability was defined as continued use for three cycles or more. Of those with verified follow-up visits, continued use beyond the first 70 days was achieved in 67% of patch users and 89% of pill users (P<.001). Neither race (P=.06) nor clinic location (P=.14) was associated with differences in acceptability.
The most common reason to discontinue either the patch or the pill was to switch to another contraceptive method. Forty-six percent of discontinuing pill users and 20% of discontinuing patch users cited this as their reason to discontinue (P<.001). Pregnancy, contraceptive preferences, skin reactions, compliance issues, and increased or irregular vaginal bleeding were the most common reasons to discontinue the patch. Table 3 demonstrates data on tolerability or self-reported adverse effects of the pill and patch.
We captured a total of 3,206 cycles in this study, resulting in a Pearl index of 3.62 for the oral contraceptive and 14.84 for the transdermal contraceptive in our high-risk population. By time period, the unintended pregnancy rate was higher for those using the patch than for those using the using the pill: before 70 days, relative risk (RR)=0.73 (95% confidence interval [CI] 0.41–4.13), P=.722; and after 70 days, RR=3.23 (95% CI 1.43–7.31), P=.005. For subjects who used the patch after acceptability was established at 70 days, the absolute risk for unintended pregnancy was 3.46%. This is presumably a direct measure of “typical use” effectiveness, given that most failures are likely attributable to user error.
By life tables survival analysis, pill users had a significantly higher continuation rate than patch users (P=.028) as shown in Figure 1. At 1 year of follow-up, 76% of pill users and 57% of patch users continued their method (P=.004). Among patch users, those with private insurance fared the best, followed by those self-paying, those in the Family Planning Benefit Program, and lastly those in Medicaid (Fig. 2). Full-time workers also did better than part-time workers (Fig. 3).
Among patch users, the multivariate Cox proportional hazards model showed no difference in continuation by clinic location (P=.865), black versus white race (P=.199), marital status (P=.652), high-risk status (P=.098), age (P=.109), parity (P=.887), or abortion history (P=.209). Although those at high risk for a future unintended pregnancy demonstrated lower continuation rates than those not at high risk (P=.016), in the multivariate model this was not significant and did not alter the findings. In the multivariate model, only work status was a significant predictor of discontinuance (P=.033).
In this study, we found the patch to have significantly higher discontinuation and unintended pregnancy rates in this high-risk population. We hypothesized that the patch would be equal or superior to the contraceptive pill in this population. The patch claims simplicity, reversibility, and the convenience of weekly dosing with a generous fail-safe period, providing a defense against imperfect dosing.32,33 We believed any concerns with the contraceptive patch would be minimal compared with expected increases in acceptability, tolerance, and compliance. However, we found the opposite.
Prior controlled studies showed greater compliance with the contraceptive patch in both adults18,34 and teens.7,35 The 1995 National Family Growth Survey found perfect and typical failure rates of 0.1% and 8.1%, respectively, for oral contraceptive users.20 Pearl indices (pregnancies per 100 woman-years use) for the oral contraceptive pill are reported at 0.79 and 7.58 in perfect and imperfect trials, respectively. Our population using the contraceptive pill was similar, with 89% demonstrating acceptability and a Pearl index of 3.62. Although our pill outcomes were similar, patch outcomes were disappointing. Pearl indices for the patch had been reported at 0.73 with perfect usage and at 4.46 with imperfect usage.18 Our patch users demonstrated only 67% acceptability and had a Pearl index of 14.84. The anticipated benefit from convenient weekly dosing apparently did not demonstrate effectiveness.
We observed a discontinuation rate of 24% at 1 year with the pill. This rate is similar to that reported in the general population where the discontinuation rate for pill-related adverse effects is 32% at 1 year.36 Thus, our continuation rates may be generalizable. Our continuation rate for the patch, 57% at 1 year, represents the first report of typical use continuation in the literature. The lower continuation rate for the patch may be due, in part, to a lower baseline familiarity with transdermal delivery system, given its introduction in 2001, compared with the pill introduced in 1960.
We were also interested in the tolerability of the patch, given previous reports of skin irritation for those using the patch. Skin irritation or site reactions were cited as the treatment-limiting factor by only 3.3% of patch users. It may have been a deterrent from perfect compliance in many more users, however. Prior trials have reported skin site reactions in 4.7% of users34 and treatment-limiting skin reactions in only 2.6% of users.22 Additionally, although the patch has been tested in humid climates,37 it has not been evaluated in colder and drier northern climates such as Western New York. In such climates, widespread use of lotions for dry skin may reduce the effectiveness of the patch. It is unclear if the lotion use and the drier climate was a factor in our poorer outcomes with the patch. It has not been investigated, to our knowledge.
Other factors may also explain the discrepancy between the literature and our results. First, because 86% of our high-risk population was employed, a student, or otherwise had access to health insurance, individuals may have continued contraceptives through other health care sources without our knowledge. However, this is likely to be a source of nondifferential error, given that both the pill and patch groups had a financial incentive to continue with contraception through Planned Parenthood with its discounted prescription plan. Moreover, our loss-to-follow-up rate was similar to other noninvasive open-label trials,7,38 and our use of survival analysis controlled for differential loss to follow-up between the groups.
Our study was also limited by selection bias. Baseline differences in the two groups may account for behavioral differences. Baseline differences in the contraceptive attitudes and practices of the two groups are highlighted by the prior abortion rate. We presume that many patch users began using the patch postabortally. Users self-selected their contraceptive regimen. Patch selectors may have been less likely to select strict contraceptive regimens and may be more likely to fail, even with a weekly regimen. This is mitigated, however, by the fact that both groups were equally high risk and that multivariate modeling showed no significant effect by baseline characteristics other than work status. Moreover, heavy patch advertising at the time of this study may have reminded users of their patch, falsely increasing the compliance rate secondary to greater social desirability surrounding the product. This differential error would only bolster our findings, however. A future randomized trial would eliminate selection bias and minimize social desirability bias.
The greatest strength of this study is that it is a community-based cohort study of contraceptive feasibility. Our population was largely at high risk for future unintended pregnancy and pregnancy termination. Although our definition is not universal, it does distinguish our population from others studied. By examining feasibility of the contraceptive patch in a high-risk population, we believe this study contributes significantly to understanding the generalizability of contraceptive therapies. This study's noninvasive strategy was also a strength. Our methodology took efforts to avoid Hawthorne effects and disruption of standard care practices. By evaluating use by a community-based high-risk population of contraceptive-naïve women, our study represents the actual outcomes of women starting hormonal contraception for the first time, unbiased by prior experience.
We found work status to be significantly associated with continuation. We also chose to highlight insurance for its relevance to the funding of contraception in high-risk populations. Both serve as markers of socioeconomic status. We presume that most unintended pregnancies were due to user-related failure and that identification of risk factors can help in identifying those who will succeed. In our population, being under- or unemployed, and thus of lower income status, predicted a lower patch-continuation rate. Receiving government health insurance also demonstrated this trend. Thus, financial burdens associated with patch use may precipitate discontinuation for poor, high-risk women.
Our study disproved our initial hypothesis that the patch would have a better continuation rate than the pill in this high-risk population. We found the contraceptive patch to have lower rates of continuation and effectiveness. We could not, however, discern the difference between user failure and method failure. Our findings do not discount use of the contraceptive patch for high-risk populations. Continuance depends largely on cultural norms, personal values, product features, and personal responsibility.32,39,40 Clearly, the patch did work for some women. Practitioners should tailor contraceptive therapies to the individual, investigating which contraceptives are more feasible for the individual, taking into account each woman's socioeconomic concerns and risk status. Further research should investigate reasons for poorer real-world performance of the patch and investigate ways to assist women who are at high risk for unintended pregnancy in using the patch effectively.
1. Finer LB, Henshaw SK. Abortion incidence and services in the United States in 2000. Perspect Sex Reprod Health 2003;35:6–15.
2. Klerman LV. The intendedness of pregnancy: a concept in transition. Matern Child Health J 2000;4:155–62.
3. Henshaw SK. Unintended pregnancy in the United States. Fam Plann Perspect 1998;30:24–9, 46.
4. Besculides M, Laraque F. Unintended pregnancy among the urban poor. J Urban Health 2004;81:340–8.
5. Trussell J, Vaughan B, Stanford J. Are all contraceptive failures unintended pregnancies? Evidence from the 1995 National Survey of Family Growth. Fam Plann Perspect 1999;31:246–7, 260.
6. Henshaw SK. Abortion incidence and services in the United States, 1995–1996. Fam Plann Perspect 1998;30:263–70, 287.
7. Rubinstein ML, Halpern-Felsher BL, Irwin CE Jr. An evaluation of the use of the transdermal contraceptive patch in adolescents. J Adolesc Health 2004;34:395–401.
8. Westoff CF. Unintended pregnancy in America and abroad. Fam Plann Perspect 1988;20:254–61.
9. Abma JC, Martinez GM, Mosher WD, Dawson BS. Teenagers in the United States: sexual activity, contraceptive use, and childbearing, 2002. Vital Health Stat 2004;23:1–48.
10. Mosher WD, Martinez GM, Chandra A, Abma JC, Willson SJ. Use of contraception and use of family planning services in the United States: 1982–2002. Adv Data 2004;10:1–36.
11. Harlap S. Oral contraceptives and breast cancer. Cause and effect? J Reprod Med 1991;36:374–95.
12. Counseling to prevent unintended pregnancy. U.S. Public Health Service. Am Fam Physician 1994;50:971–4.
13. Brown SS, Eisenberg L, editors. The best intentions: unintended pregnancy and the well-being of children and families. Committee on Unintended Pregnancy, Division of Health Promotion and Disease Prevention, Institute of Medicine. Washington, DC: National Academy Press; 1995.
14. Petersen R, Gazmararian JA, Anderson Clark K, Green DC. How contraceptive use patterns differ by pregnancy intention: implications for counseling. Womens Health Issues 2001;11:427–35.
15. Sonfield A, Gold RB, Frost JJ, Darroch JE. U.S. insurance coverage of contraceptives and the impact of contraceptive coverage mandates, 2002. Perspect Sex Reprod Health 2004;36:72–9.
16. Rosenberg MJ, Waugh MS, Meehan TE. Use and misuse of oral contraceptives: risk indicators for poor pill taking and discontinuation. Contraception 1995;51:283–8.
17. Rosenberg MJ, Burnhill MS, Waugh MS, Grimes DA, Hillard PJ. Compliance and oral contraceptives: a review. Contraception 1995;52:137–41.
18. Archer DF, Cullins V, Creasy GW, Fisher AC. The impact of improved compliance with a weekly contraceptive transdermal system (Ortho Evra) on contraceptive efficacy. Contraception 2004;69:189–95.
19. Rosenberg MJ, Waugh MS. Oral contraceptive discontinuation: a prospective evaluation of frequency and reasons. Am J Obstet Gynecol 1998;179:577–82.
20. Fu H, Darroch JE, Haas T, Ranjit N. Contraceptive failure rates: new estimates from the 1995 National Survey of Family Growth. Fam Plann Perspect 1999;31:56–63.
21. Smallwood GH, Meador ML, Lenihan JP, Shangold GA, Fisher AC, Creasy GW. Efficacy and safety of a transdermal contraceptive system. Obstet Gynecol 2001;98:799–805.
22. Audet MC, Moreau M, Koltun WD, Waldbaum AS, Shangold G, Fisher AC, et al. Evaluation of contraceptive efficacy and cycle control of a transdermal contraceptive patch vs an oral contraceptive: a randomized controlled trial. JAMA 2001;285:2347–54.
23. Dittrich R, Parker L, Rosen JB, Shangold G, Creasy GW, Fisher AC. Transdermal contraception: evaluation of three transdermal norelgestromin/ethinyl estradiol doses in a randomized, multicenter, dose-response study. Am J Obstet Gynecol 2002;186:15–20.
24. Logsdon S, Richards J, Omar HA. Long-term evaluation of the use of the transdermal contraceptive patch in adolescents. ScientificWorldJournal 2004;4:512–6.
25. Balassone ML. Risk of contraceptive discontinuation among adolescents. J Adolesc Health Care 1989;10:527–33.
26. Falk G, Falk L, Hanson U, Milsom I. Young women requesting emergency contraception are, despite contraceptive counseling, a high risk group for new unintended pregnancies. Contraception 2001;64:23–7.
27. Zite NB, Shulman LP. New options in contraception for teenagers. Curr Opin Obstet Gynecol 2003;15:385–9.
28. Westoff CF. Contraceptive paths toward the reduction of unintended pregnancy and abortion. Fam Plann Perspect 1988;20:4–13.
29. Preventing unintended pregnancy in the U.S. Issues Brief (Alan Guttmacher Inst) 2004;3:1–4.
30. Jones RK, Darroch JE, Henshaw SK. Patterns in the socioeconomic characteristics of women obtaining abortions in 2000-2001. Perspect Sex Reprod Health 2002;34:226–35.
31. Trussell J, Kost K. Contraceptive failure in the United States: a critical review of the literature. Stud Fam Plann 1987;18:237–83.
32. Linn ES. Progress in contraception: new technology. Int J Fertil Womens Med 2003;48:182–91.
33. Pierson RA, Archer DF, Moreau M, Shangold GA, Fisher AC, Creasy GW. Ortho Evra/Evra versus oral contraceptives: follicular development and ovulation in normal cycles and after an intentional dosing error. Fertil Steril 2003;80:34–42.
34. Burkman RT. The transdermal contraceptive system. Am J Obstet Gynecol 2004;190:S49–53.
35. Harel Z, Riggs S, Vaz R, Flanagan P, Dunn K, Harel D. Adolescents' experience with the combined estrogen and progestin transdermal contraceptive method Ortho Evra. J Pediatr Adolesc Gynecol 2005;18:85–90.
36. Trussell J, Vaughan B. Contraceptive failure, method-related discontinuation and resumption of use: results from the 1995 National Survey of Family Growth. Fam Plann Perspect 1999;31:64–72, 93.
37. Abrams LS, Skee DM, Natarajan J, Wong FA, Leese PT, Creasy GW, et al. Pharmacokinetics of norelgestromin and ethinyl estradiol delivered by a contraceptive patch (Ortho Evra/Evra) under conditions of heat, humidity, and exercise. J Clin Pharmacol 2001;41:1301–9.
38. Faculty of Family Planning and Reproductive Health Care Clinical Effectiveness Unit. New Product Review (September 2003). Norelgestromin/ethinyl oestradiol transdermal contraceptive system (Evra). J Fam Plann Reprod Health Care 2003;30:43–5.
39. Freeman S. Nondaily hormonal contraception: considerations in contraceptive choice and patient counseling. J Am Acad Nurse Pract 2004;16:226–38.
40. Benagiano G, Shedlin MG. Cultural factors in oral contraceptive compliance. Adv Contracept 1992;8 suppl:47–56.