Systematic review of efficacy with extending contraceptive implant duration

Abstract Background Extending contraceptive implant duration of use increases accessibility by maximizing the lifetime of devices. Objectives To review the contraceptive efficacy during extended use of progestin implants. Search strategy PubMed and EMBASE were searched for articles in any language, 1996–2017, utilizing terms for devices and contraceptive efficacy. Selection criteria Randomized clinical trials (RCTs), cohort studies, and case–control studies were included; abstracts, posters, and presentations were excluded. Studies evaluating Norplant and implants currently in pre‐marketing trials were excluded. Data collection and analysis Titles and abstracts of articles were reviewed; those that met inclusion and exclusion criteria underwent full text review and data abstraction. Main results The search identified 2951 articles; six met inclusion and exclusion criteria. Five studies evaluated the etonogestrel implant (Implanon), and one the levonorgestrel implant (Jadelle). One RCT randomized to method, not duration; the remaining studies were prospective cohort studies. Three studies analyzed efficacy among women beyond currently approved duration separately. All studies were of poor to fair quality by United States Preventative Services Task Force (USPTF) grading. Limitations include lack of generalizability and control of important confounders. Conclusion These studies provide limited data for extended duration of contraceptive implants.

include etonogestrel (ENG)-releasing devises, including Implanon and Nexplanon (Merck, Kenilworth, NJ, USA) and levonorgestrel (LNG)releasing devices, including the Sino-Implant II (Dahua Pharmaceutical, Shanghai, China) and Jadelle (Bayer, Berlin, Germany). 12,13 The only difference between the two ENG implants is that Nexplanon contains a radiopaque dye used to locate the device in the event that it cannot be palpated. The key aspects of product information for each of these devices is included in Table 1.
Pharmacokinetic studies provide information about the serum levels exhibited throughout use of contraceptive implants as well as what levels are required to inhibit ovulation. Prior work indicates that ovulation is suppressed at serum ENG levels greater than 90 pg/mL. 14 McNicholas et al. 15 found that median ENG levels at the end of 3 years of use of the implant are beyond this value at 207.7 pg/mL (range 63.8-802.6 pg/mL). Median serum ENG levels remain high at 166.1 pg/mL (range 25-470.5 pg/mL) at 4 years of use and 153.0 pg/mL (range 72.1-538.8 pg/mL) at 5 years of use. 15 Furthermore, this study failed to show a difference in median serum ENG levels across body mass index (BMI) groups, including obese women, at the end of the fifth year of implant use. 15 However, women may exhibit wide ranges of ENG, with some values below 90 pg/mL, as McNicholas found in 23 participants (written communication, July 2017).
Among women using the LNG implant, variations in serum LNG levels exist. Sivin et al. 16 measured serum LNG concentrations among women extending duration. Pregnancies were identified in five women using the Jadelle LNG implant beyond 5 years. In all women, the serum LNG levels fell to or below 180 pg/mL, indicating that this may represent a threshold below which contraceptive efficacy cannot be guaranteed. Further analysis revealed that study site, body weight and ponderal index, and duration of use were the most important variables affecting serum LNG levels. 16 Importantly, the pharmacokinetic data on ENG and LNG have revealed that serum concentrations between participants and even within the same participant over multiple collections vary widely. 15,16 Among women who have serum point values of progestin below presumed thresholds, pregnancies may not occur. It remains unclear what progestin level is needed to provide effective contraception, given the implant's other mechanisms of action including cervical mucus thickening and endometrial atrophy.
Barriers to obtaining a progestin-only implant are reported among women who desire them, including access to trained professionals for placement and removal, and high up-front device costs. Efforts have been made to take advantage of convenient opportunities for placement, such as immediately post pregnancy. 17 Extending the duration beyond the currently approved length of use, if shown to be efficacious, could offer another means to enhance accessibility by maximizing the lifetime of the implant.
The purpose of this systematic review is to examine current evidence to answer the question: "What is the contraceptive effectiveness of progestin implants beyond currently approved durations of use (i.e. extended use)?"

| MATERIALS AND METHODS
This review uses the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines for reporting. 18 Our primary question is: among women using a contraceptive implant, are pregnancy rates different for women who continue using the same device beyond its currently approved duration from pregnancy rates among women who initiate use of a new contraceptive implant at its expiration date? Because we did not anticipate identifying many studies that met this primary question, we also included studies that assessed pregnancy rates during any duration of extended use of contraceptive implants.
We included studies published between January 1, 1996, and We excluded from this review studies evaluating the levonorgestrel implant Norplant, as this device is no longer available, as well as contraceptive implants currently in pre-marketing trials. Conference abstracts, posters, and oral presentations were excluded.
We searched published studies among PubMed and EMBASE.
To locate articles, we used the following search terms for devices, developed together with a Health Science Librarian: "levonorgestrel", "norplant", "Jadelle", "etonogestrel", "Implanon", "Nexplanon" and "sino implant," as well as search terms for pregnancy outcomes (to assess T A B L E 1 Description of devices. contraceptive failure) and study designs including clinical trials or cohort studies (full search strategy available recorded in File S1). guidelines, each article's study quality was rated as "poor," "fair," or "good". 19 This rating was based on methodological quality as assessed by the reviewers along the following domains: assembly of appropriate comparison groups (including adequate description and execution of randomization in RCTs and assessment of potential confounders in cohort studies such as participant age); maintenance of comparable groups and participant attrition during the study; and description of how pregnancy was assessed. Additionally, quality of data analysis and interpretation was assessed according to whether adjustments were performed for confounders, and whether contraceptive efficacy was measured cumulatively or analyzed separately among women beyond approved duration.

| RESULTS
We identified 2951 citations and selected 38 articles for full-text screening ( Fig. 1). Thirty-two studies were excluded because they did not meet inclusion or exclusion criteria (File S2). We included six published studies in this systematic review, resulting in a total of 1075 women who self-selected extended duration of their con- T A B L E 2 (Continued) completed 4 years of use and there were no reported pregnancies. 21 Two noncomparative cohort studies evaluated women with 1 year of extended use of the ENG implant. Women who were breastfeeding or recently using another method of contraception were excluded.
Sample sizes were small for women completing the fourth year of implant use (n=47 and n=151). No pregnancies were reported in either of these studies. 22,23 We identified a multicenter pharmacokinetic study investigating extended duration of the LNG implant. Jadelle implants were inserted in 199 women across multiple sites over 7 years. The participants underwent regular blood draws to assess serum LNG levels. Over the course of the study, five women became pregnant: two in the fifth year and three in the seventh year. All five women had serum LNG levels that fell to or below 180 pg/mL and the two women who conceived in the fifth year demonstrated consistently low LNG levels in the three measurements prior to pregnancy. 16

| DISCUSSION
This systematic review identified six studies published from 1996 to 2017 that met inclusion and exclusion criteria. Five studies evaluated the ENG implant (Implanon) and one evaluated the LNG implant (Jadelle). There were no studies of the Nexplanon or Sino-Implant II.
One study was an RCT, however randomization was to contraceptive method, not duration; the remaining five studies were prospective cohort studies. Three studies separately analyzed contraceptive efficacy among women beyond currently approved duration. All studies were of poor to fair quality by USPTF grading.
These studies enrolled a total of 1075 women and assessed con- The lack of randomization and small population size within these studies invite concern that participants may not be representative of the population. Inclusion and exclusion criteria for all studies defined participants as medically healthy, limiting generalizability. Additionally, women who elect to extend duration may differ widely from those who replace their implant. Future research should randomize women to extended duration or critically evaluate potential confounders in continuation. Additionally, more pharmacokinetic research is needed on ENG and LNG serum levels as they relate to the outcome of interest: contraceptive failure. Variability in BMI and potential drug interactions should be further evaluated. Moreover, future studies should investigate whether or not extension of duration is a service women desire. While outside the scope of this review, durability of the devices should also be investigated.
If found to be efficacious, the ability to extend duration of con-