The Creeping Pearl Index:
Distinguishing Between Clinical Trial Efficacy and Actual-Use Effectiveness
Reversible contraception is widely used by virtually all reproductive-aged women at some point in their lives, yet many patients and clinicians are unaware of how effective their chosen methods actually are and how efficacy is determined in both clinical trials and the real world.1 While the effectiveness of contraception is a major determinant in a woman’s decision, other factors are also key considerations. In the following article, reversible combined hormonal contraception (CHC) methods, historical trends in contraceptive clinical trial design, the impact of obesity on efficacy, US Food and Drug Administration (FDA) guidance for trial conduct, perfect and actual-use data, and shared decision-making around contraceptive choice will be summarized. The focus of this article will be on reversible CHCs, as LARC, progestin-only, and non-hormonal methods are beyond the scope of this brief review.
In 2013, Trussell and Portman published “The Creeping Pearl” in the journal Contraception to address how trends in trial design and changing populations have led to an increase in pregnancy rates in modern contraceptive studies as defined by the Pearl Index (PI).2 Much has changed in the contraceptive world since then, with the approval of several new contraceptives, updated FDA guidance, and new data emerging on the role of obesity in contraceptive efficacy.
The selection of a particular method is a preference-sensitive decision—each woman will weigh various factors that are important to her when making an individual choice. These factors include effectiveness, dose, desire (or not) for hormonal methods, delivery route and level of invasiveness, convenience and frequency of administration, and non-contraceptive benefits that may be particularly important. No single method is right for all women. Nonmedical issues, such as access and availability of methods, are often determining factors. Varying preferences and tolerability issues lead to different contraceptive choices. This is true not only from one person to another, but even within an individual woman’s reproductive years. A woman may stay with a method for longer and is more likely to be consistent with its use if it is a method of her choosing and fits her lifestyle.3
Contraceptive options align along a spectrum and are tiered based on “effectiveness,” which is based on actual use in real users, not “efficacy” as defined in contraceptive trials (Figure 1).4 The most effective user-independent options, Tier 1, include permanent and semi-permanent methods requiring invasive procedures and products such as sterilization, implants, and IUDs. Tier 2 includes hormonal methods that for the most part rely heavily on user compliance. Tier 3 includes barrier and behavioral methods that require use during each episode of sexual activity and are thus more prone to user failure. Without any contraception, the annual rate of pregnancy is 85%.
Figure 1. Contraceptive Options Tiered Based on Effectiveness
The hormonal methods in Tier 2 are among the most popular largely because they give the user some independence and control, are less invasive, and are highly effective, with pregnancy rates estimated to be in the 4%‒7% range. However, some Tier 2 methods, such as Depo-Provera with its injection, delayed return to fertility, and reversible bone loss, as well as contraceptive rings that require vaginal placement and need to be removed monthly, can pose challenges to some users.
Methods of combined hormonal contraception, particularly those with combined oral contraception (COC), are widely used, and contain both a progestin and an estrogen component. The progestin component prevents pregnancy primarily by inhibiting ovulation, thickening cervical mucus, and through other mechanisms. The progestins used in CHCs have differing pharmacologic and tolerability issues such as bleeding patterns and metabolic impacts. Estrogen is largely added for cycle control and to optimize the bleeding profile. Estrogen doses of 35 mcg or less are used to minimize possible estrogen-related side effects—such as breast tenderness, headache and nausea—and to improve the overall safety profile. Healthcare providers seldom prescribe contraceptives containing 50 mcg of estrogen per day. Consistent with this trend in lowering estrogen dose, the vast majority of recently approved and available daily oral CHCs contain 35 µg or less of estrogen.
While there are a myriad number of COCs to choose from, only four non-daily combined hormonal options, all of which employ ethinyl estradiol (EE) are available—the EE/etonogestrel NuvaRing monthly single-use vaginal ring; the EE/segesterone acetate, Annovera, a newly approved monthly reusable vaginal ring with a year of use; the once-weekly Xulane (previously Ortho Evra) patch delivering approximately 56 mcg of EE per day with norelgestromin; and the newly approved once-weekly Twirla patch with approximately 30 mcg of EE per day with levonorgestrel. There are advantages to transdermal delivery. Transdermal methods provide a controlled release of hormones over time that offers the potential to reduce the incidence or severity of side effects such as breast tenderness, headache, and nausea potentially associated with drug peak and trough levels and hormone withdrawal.Transdermal delivery also avoids reduced bioavailability seen with oral drug administration.3 In addition, this delivery system may help women who are unable to take oral medications, which is particularly common among younger users.5 Non-daily, non-oral options have the potential to reduce the daily pill-taking burden some women associate with OCs. In fact, in a multinational questionnaire, 49% of contraception users reported a preference for a non-daily method, and 52% were frustrated with taking a pill daily.6
The approval of the reusable Annovera vaginal ring adds convenience and is a welcome additional non-daily method. However, there is very limited data from its multinational trials in women with obesity, an important population particularly here in the US and a major factor recently associated with decreased contraceptive efficacy in clinical trials.7 In fact, at approximately 50% enrollment, women with BMI >29.0 kg/m2 were no longer enrolled in the two Annovera trials, and all women with a BMI >29.0 kg/m2 were discontinued from the trials.8 The recent U.S. Twirla patch trial included a significant number of women with a BMI ≥30kg/m2—35% of the overall study population, and roughly equivalent to the obesity rate in the U.S. It is important to note this enrollment provided prospective data that enabled calculation of efficacy by BMI. To address potential efficacy and safety issues, FDA labeling indicates that the use of Twirla is for women with a BMI < 30kg/m2. In April 2020, the FDA revised the labeling for Xulane regarding its use in women with a BMI ≥30 kg/ m2. Xulane is now contraindicated for use in women with a BMI ≥ 30 kg/ m2. These women may have a higher risk of venous thromboembolic events compared with women with a lower BMI.9 While having a variety of options allows for individualized choice and shared decision-making for patients and providers, it is imperative to put contraceptive efficacy and effectiveness into perspective—given how well a particular method works for certain populations is a critical factor in contraceptive selection.
In clinical trials, the Pearl Index is the most common regulatory endpoint for contraceptive efficacy. The Pearl Index (PI), defined as the number of on-treatment pregnancies multiplied by 13 cycles, divided by the number of on-therapy cycles times 100, provides an estimate of the number of pregnancies per 100 woman-years of product use (Figure 2).
Figure 2. Pearl Index Calculation
The number of cycles in the denominator directly impacts the Pearl Index and the resulting width of the 95% confidence interval. Adding more patients or more cycles may not affect the point estimate but can drive down the upper bound of the 95% CI.
The Pearl Index (PI) calculation, while a regulatory standard, is highly sensitive to study design, duration and population factors, and certain factors used in earlier CHC trials are known to yield low pearl indices, including: enrolling women in European trial sites (with lower BMI and more uniform ethnicity and socioeconomic status); restricting enrollment based on BMI or weight; recruiting more affluent, educated, nulliparous women; not requiring women to anticipate or record regular sexual activity; nor accounting for cycles without sexual activity in efficacy analyses.2,10 These studies produced PIs that were artificially low and not generalizable to current US women using contraception. As a result, there has historically been a wide gap between clinical trial efficacy and actual-use effectiveness.
As pharmaceutical sponsors have begun to broaden enrollment criteria and modify analysis methods, the Pearl Indices from contemporary CHC trials have been rising. This is the concept Trussell and Portman termed as the “The Creeping Pearl,” and it has become all the more apparent in the last decade.
Contemporary CHC trials include multiple factors known to increase Pearl Indices; these include: limiting enrollment to women living in the US; fewer to no restrictions on body weight or BMI; documenting and removing sexually inactive cycles; and more frequent pregnancy testing with more sensitive tests. This resulted in more inclusive and representative study populations and a “Creeping Pearl” more reflective of actual-use effectiveness.
In fact, the identical contraceptive’s initial Pearl Index from its own registration trial often increases dramatically when it is used as a comparator arm in trials conducted more recently. Shown in Figure 3 are just three examples of this phenomenon with Loestrin, Levlite and Nordette.
Figure 3. Pearl Indices in Initial FDA Registration Studies Increased in Later Trials
As shown in Figure 4, an important population to consider in contraceptive studies are women with obesity, and FDA has encouraged sponsors to study women with obesity prospectively in clinical trials. A meta-analysis performed by the FDA looking at individual patient data from seven combination oral contraceptive trials demonstrated that overall, there was a 44% increased risk of pregnancy for women with BMI >30 kg/m2 compared to women with lower BMIs.11
Figure 4. FDA Meta-Analysis: Relationship Between Obesity and Contraceptive Effectiveness
The desogestrel/EE study was one of the few that included no restriction on BMI and had the highest percentage of obese women in this analysis. The hazard ratio for risk of pregnancy was more than 2.5 times greater for women with obesity compared to women without. And in the Ortho Evra patch, a nearly 9-fold greater risk of pregnancy was observed. When combining oral contraception and Ortho Evra patch data, the overall hazard ratio for pregnancy on CHCs in obese women is 65% higher than in a normal weight cohort. When including a higher proportion of these women in a prospective trial, the Pearl Index will certainly increase.
In 2007, the Bone, Reproductive and Urologic Drugs Advisory Committee (BRUDAC) provided FDA with recommendations on clinical trial design, assessing the acceptability of risks and benefits and the role and impact of labeling. The panel delivered recommendations to: change entry criteria to reflect real-world prescribing—even if it results in rising pearl indices; conduct studies with active comparators; modify trial designs to provide results that reflect effectiveness in the real world; avoid arbitrary limits for the upper bound of the 95% CI in order to bring the widest range of new contraceptive options to market; and ensure that all relevant information be provided to the prescriber, including data on particular subgroups.12
The FDA’s 2019 draft guidance on contraceptive trials accepts most of these recommendations—specifically to enroll broadly representative populations, to avoid restrictions on BMI at enrollment, to enroll sexually active women, to exclude all sexually inactive cycles—but notes single-arm studies are sufficient.13 All of these factors have impacted recent contraceptive studies and approvals and should be viewed in light of shared decision-making and unmet contraceptive needs.
Putting all these factors into context with historic and modern contraceptive trials can be illustrative. As discussed, obesity and more generalizable patient populations have led to higher Pearl Indices in recent trials, and the Twirla study is no exception. The Twirla study incorporated nearly all of the considerations from the most recent FDA draft guidance, and not surprisingly, the Pearl Index was 5.83 and 4.34 in the overall and non-obese population, respectively. Broken down by BMI, for women <25 kg/m2 the PI was 3.5 (CI: 1.8-5.2), and 5.7 (CI: 3.0-8.4) for those >25 and <30 kg/m2.14 The Twirla patch is contraindicated in women with a BMI >30kg/m2 where the Pearl index was 8.6 (CI: 5.8-11.5). With respect to weight in the Xulane/Ortho Evra studies, 5 of the 15 pregnancies reported were among women with a baseline body weight ≥198 lbs, which constituted < 3% of the study population, suggesting that greater inclusion of obese women would have driven the overall PI higher. Additionally, the risk of venous thromboembolism (VTE) could be increased in women with BMI >30kg/m2—an established risk factor for VTE even in the non-contracepting individual—and is thus contraindicated in this population.9 In non-obese women, the cumulative annual pregnancy rate from life-table analyses for Twirla was 4.0%15, with a safety and efficacy profile similar to other Tier 2 methods. Uniquely, the Twirla study has provided important prospective information confirming how modern contraceptive trials—e.g, trials that include more obese women, study an exclusively US population, capture all sexually active cycles, exclude inactive cycles and those cycles where alternative contraception was used—can impact the Pearl Index. With more options available and accurate, prospective data informing labeling, patients and providers can better select the best option for the individual woman—the one she feels most comfortable using and that satisfies her own unique preferences and needs.
- CDC National Survey of Family Growth 2011-2015. Perspect Sex Reprod Health. 2017;49(1):7-16.
- Trussell J, Portman D. The creeping Pearl: Why has the rate of contraceptive failure increased in clinical trials of combined hormonal contraceptive pills? Contraception. 2013:88(5);604-610.
- Westhoff CL, Heartwell S, Edwards S, et al. Oral contraceptive discontinuation: do side effects matter? Am J Obstet Gynecol. 2007;196(4):412.e1‐412.e7.
- Hatcher RA. Contraceptive Technology. New York, NY: Ayer Company Publishers, Inc.; 2018.
- O’Connell K, Burkman R. The transdermal contraceptive patch: an updated review of the literature. Clin Obstet Gynecol. 2007;50(4):918-926.
- Hooper DJ. Attitudes, awareness, compliance and preferences among hormonal contraception users: a global, cross-sectional, self-administered, online survey. Clin Drug Investig. 2010;30(11):749-763.
- Annovera [Prescribing Information]. Boca Raton, FL: TherapeuticsMD, Inc. 2020.
- Edelman A, Trussell J, Aiken ARA, et al. The emerging role of obesity in short-acting hormonal contraceptive effectiveness. Contraception. 2018;97(5):371–377.
- Xulane [Prescribing Information]. Morgantown, WV: Mylan Pharmaceuticals, Inc. 2020.
- Gerlinger C, Trussell J, Mellinger U, et al. Different Pearl Indices in studies of hormonal contraceptives in the United States: Impact of study population. Contraception. 2014;90(2):142-146.
- Yamazaki M, Dwyer K, Sobhan M, et al. Effect of obesity on the effectiveness of hormonal contraceptives: an individual participant data meta-analysis. Contraception. 2015;92:445-452.
- Advisory Committee for Reproductive Health Drugs FDA Briefing Document. 2007.
- Establishing Effectiveness and Safety for Hormonal Drug Products Intended to Prevent Pregnancy Draft Guidance for Industry. Published July 1, 2019. https://www.fda.gov/media/128792/download
- Twirla [Prescribing Information]. Grand Rapids, MI: Agile Therapeutics, Inc. 2020.
- Agile Therapeutics [Data on File].