Objective
To systematically review the literature to determine if extended use of intrauterine devices, including the copper or levonorgestrel intrauterine device, beyond approved durations is effective and safe for preventing pregnancy.
Study Design
We searched PubMed, Embase, Cumulative Index to Nursing and Allied Health Literature, and Global Index Medicus from inception through June 2019. We considered primary studies of women using the T380A copper or currently available levonorgestrel intrauterine devices past their approved duration (extended use). The primary outcome was pregnancy. Safety outcomes included adverse events and side effects that led to discontinuation. We abstracted study data into evidence tables and assessed study quality as related to the risk of bias, precision, and external validity. For each intrauterine device type, we calculated a pooled pregnancy rate during the first 2 years of extended use using the Poisson inverse variance method.
Results
Of 4068 studies identified by our search, 4 good-to-poor-quality studies of the 52 mg levonorgestrel intrauterine device (approved for 5 years), with a total of 2098 women starting extended use, and 2 good-to-fair-quality studies of the T380A copper intrauterine device (approved for 10 years), with 245 women starting extended use, met inclusion criteria. For the levonorgestrel intrauterine device, the pooled pregnancy rate was 0.02 per 100 person-years (95% confidence interval, 0.00–0.45) in year 6, 0.03 per 100 person-years (95% confidence interval, 0.00–0.71) in year 7, and 0.02 per 100 person-years (95% confidence interval, 0.00–0.29) in years 6 and 7 combined. During this same time, annual rates of adverse events or discontinuation owing to side effects ranged from 0 to 3.7 per 100 participants. For the copper intrauterine device, the pooled pregnancy rate for years 11 and 12 was 0.0 per 100 person-years (95% confidence interval, 0.0–0.8), and annual rates of adverse events and discontinuation owing to side effects during extended use ranged from 0 to 4.6 per 100 participants.
Conclusion
The available evidence suggests that rates of pregnancy, adverse events, and discontinuation owing to side effects during the first 2 years of extended use of the 52 mg levonorgestrel intrauterine device and the T380A copper intrauterine device are low and comparable to rates during approved duration of use. Data were limited in quantity and quality, and may not be generalizable to all intrauterine device users.
Why was this study conducted?
This study was conducted to systematically review the literature to determine if the use of modern intrauterine devices beyond approved durations is effective and safe.
Key findings
Current evidence for the levonorgestrel 52 mg and the T380A copper-containing intrauterine devices suggests that rates of pregnancy and other complications during the first 2 years of extended use are comparable to rates during approved duration of use; data were limited in quantity and quality, and may not be generalizable to all intrauterine device users.
What does this add to what is known?
This study provides pooled estimates of pregnancy rates for the first 2 years of extended use of the levonorgestrel 52 mg and the T380A copper-containing intrauterine devices, and summarizes adverse events and side effects during extended use.
Initial evidence suggests that IUDs remain both safe and effective beyond their approved durations of use. A 2014 literature review supported the use of the following IUDs past their approved duration: TCu380A, TCu220, Multiload Cu-375 and Cu-250, frameless GyneFix (330 mm 2 ), and the LNG 52 mg IUD. Additionally, recent clinical guidelines support the extended use of the TCu380A IUD and the LNG 52 mg IUD. If the use of IUDs could be extended without compromising effectiveness or safety, women could benefit from expanded contraceptive protection. Extended use could reduce the number of IUD removals and reinsertions, thereby decreasing procedural risks and increasing safety and convenience. Extended use could also increase cost effectiveness by reducing device and procedure costs, time and travel for additional clinic visits, and medical staff time.
Our systematic review expands on the 2014 literature review on extended IUD use by including safety outcomes (ie, adverse events and side effects leading to discontinuation), incorporating new effectiveness data, and quantifying effectiveness rates during extended use. We examined the research question: Among women using either the Cu or LNG IUD, is extended use beyond the approved durations effective and safe for preventing pregnancy?
Methods
We report this systematic review according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. We prospectively registered a protocol for this review in PROSPERO, an international prospective register of systematic reviews (CRD42017076903).
Our population of interest was women using modern Cu or LNG IUDs past their approved duration. We included studies on the copper T380A as it is the only Cu IUD included in the World Health Organization’s Prequalification program. This program generally determines which medicines are procured by international family planning organizations such as the United States Agency for International Development and United Nations Population Fund. We included all currently marketed LNG IUDs with the following dosages: 52 mg, 18.5 mg, and 13.5 mg. The intervention of interest was extended use beyond approved duration, and we determined each device’s “approved duration” by examining package labeling or governmental regulatory organization (eg, the United States Food and Drug Administration) approvals. A comparison group was not required; however, the relevant comparison group for this research question would be women using the same IUD during approved duration. Our primary outcome was pregnancy occurring among IUD users during the extended use period. Safety outcomes assessed included adverse events (ie, expulsion, displacement, perforation, and infection) and side effects (ie, bleeding and pain) that led to discontinuation during extended use.
We included randomized controlled trials (RCTs), cohort (comparative and noncomparative), and case-control studies investigating the contraceptive effectiveness and safety of Cu and LNG IUDs used beyond their approved duration of use (ie, extended use). We excluded case reports, case series, conference abstracts, and unpublished studies.
With the help of research librarians, we searched PubMed, Embase, Cumulative Index to Nursing and Allied Health Literature, and Global Index Medicus from database inception through June 2019 ( Appendix: Supplemental Table ). We searched for articles that included either Cu or LNG IUDs and had data on effectiveness or safety outcomes during extended use. We did not apply any language restrictions. We hand-searched the references of relevant review articles.
Using Covidence systematic review software, 2 authors evaluated the title and abstract of each article. If an article was potentially eligible, 2 authors screened the full text for inclusion; any conflicts were resolved through discussion. We documented reasons for exclusion, such as incorrect exposures, outcomes, or study types. Authors then extracted the following information from articles meeting eligibility criteria: study title, author, publication year, funding source, type of IUD(s), study design, population, duration of extended use, pregnancy outcomes, adverse events, and side effects leading to discontinuation. We focused only on outcomes that occurred during extended use. Study design was determined as relevant to our research question. For example, if a study was reported as an RCT but did not contain a comparison group relevant to our research question, we considered the study to be a noncomparative cohort study. If multiple publications reported on the same study, we included the most recent relevant publication. We resolved disagreements through discussion.
Two authors independently assessed study quality using a framework modeled after the United States Preventive Services Task Force. We rated study quality (good, fair, or poor) based on assessments of risk of bias, precision, and external validity. We assessed risk of bias (low, high, or unclear) using the Cochrane Risk of Bias Assessment tool for RCTs, and we modified the tool to reflect domains relevant to observational studies, including information bias and confounding, and removed domains not relevant to nonrandomized studies. To assess precision, we considered sample size and confidence intervals; and to assess external validity, we considered study population and setting.
We attempted to extract data on the annual rates of outcomes for each year of extended use. If so specified by study authors, we reported results as either gross or net rates, to distinguish rates that accounted for discontinuation and loss to follow-up (ie, net rates) and those that did not (ie, gross rates). If studies provided only cumulative rates for each additional year of observation, we calculated year-specific results when possible. We calculated the 95% confidence intervals (CIs) for the rates of the individual studies using the exact Poisson method and a pooled pregnancy rate with associated CIs for the first and second years of extended use for each IUD using the Poisson inverse variance method ( Appendix: Supplemental Methods ). If precise person-time was not available, we assumed that participants who became pregnant did so at the midpoint of the time period, and those who were lost to follow-up during that time period did not contribute any person-time to the denominator.
Results
The search identified 4068 unique articles, which were then screened by title and abstract ( Figure ). We reviewed the full text of 119 articles and excluded the majority owing to either an incorrect exposure (eg, excluded IUD type or study not covering extended use) or excluded article type (eg, case report or commentary). We excluded 2 publications , because they reported on interim results of included studies. , We hand-searched the references of 2 relevant reviews, , and 1 additional study met inclusion criteria that was not identified by the search because the study’s indexing terms did not include extended use or time factors. In total, 4 studies of the LNG 52 mg IUD , and 2 studies of the Cu T380A , met inclusion criteria. The studies were either noncomparative cohort studies , , , or a single arm of a comparative study without a relevant comparison group. ,
LNG IUD
All 4 studies of LNG IUDs examined the 52 mg device, which is approved for 5 years of use ( Table 1 ). , Within these 4 studies, a total of 2098 participants started extended use (ie, entered a sixth year of use), with up to 10 years of extended use. Two studies were determined to be good quality, , 1 was fair, and 1 was poor. The first good-quality study was an arm of a multicountry RCT that included 717 women followed for up to 2 years of extended use, with low risk of bias, good precision, and good external validity. The other good-quality study was a noncomparative cohort study of 776 women in Brazil followed for up to 10 years in a chart review of extended use at a single center, with low risk of bias and good precision, though the external validity was unclear owing to limited demographic information. The fair-quality observational cohort study followed 496 women in the United States who were willing to extend the use of their LNG IUDs for 1 year. This study had good precision and external validity but a high risk of bias, as loss to follow-up was not described and outcomes were assessed by patient report over the phone. A poor-quality observational cohort study recruited 100 parous women in Sweden who were interested in extending the use of their LNG IUDs and were followed for up to 3 years. This study had an unclear risk of bias, poor precision, and unclear external validity, as its authors failed to describe outcome ascertainment or the characteristics of those who opted not to extend IUD use, and had a small sample size.
| Author, year, funding | Study details | Population, extended use | Follow-up | Cumulative results | Year-specific results | Quality |
|---|---|---|---|---|---|---|
| Ronnerdag and Odlind, 1999 Funding: The Family Planning Fund of Uppsala and Leiras Oy, Turku, Finland. (manufacturer of LNG-IUD) | Noncomparative cohort study Medical center in Sweden 1983–1990 | Women previously recruited as part of a multicenter trial, who were interested in continuing IUD use beyond initial study. Mean age 32.7 (range, 22–38) years at initiation of IUD; all parous Number starting extended use, n = 100 Extended use: mean 1.6 years (range, 0.3–3.0 years) | Every 3 months with history and exam Loss to follow-up: 0 Discontinuation: 18/100 owing to menopause, no longer desiring contraception, no further interest in study participation | N/A | Pregnancy Assessed at clinical follow-up: n 0 in total extended use Adverse events Expulsion, PID: 0 in extended use | Quality: poor Risk of bias: unclear Failed to describe characteristics of those who opted not to extend IUD use; failed to describe how outcomes were assessed during extended use Precision: poor Small sample size; confidence intervals for relevant outcomes not reported External validity: unclear Characteristics of those in extended use were not described |
| Rowe et al, 2016 Funding: United Nations Development Programme/ United Nations Population Fund/ WHO/World Bank, Special Programme of Research, Development and Research Training in Human Reproduction | Noncomparative cohort study (from RCT of LNG and Cu IUD) 20 medical centers in 9 countries (Brazil, Chile, China, Hungary, Mongolia, Philippines, Slovenia, Thailand, Tunis) Years of data collection not specified | Healthy parous women, 16–40 years old, relying on IUD for contraception, currently cohabitating Excluded if history of PID, ectopic or molar pregnancy, recent STI, recent delivery or abortion, undiagnosed vaginal bleeding, genital tract malformation, malignancy, fibroids, anemia Number starting extended use, n = 717 Extended use: up to 2 years | Follow-up visits in clinic 3, 6, 12 months, then yearly after IUD insertion for 7 years total Loss to follow-up: 56 Discontinuation: 260, owing to expulsion (12), bleeding (61), pain (5), other (182) 601 (83.8%) a , b completed 6 years 398 (55.5%) a , b completed 7 years | Pregnancy Assessed at clinical follow-up: n (rate c , SE) Year 6 b : 7 a (0.5, 0.2) Year 7: 7 (0.5, 0.2) Adverse events Expulsion (partial and complete): n (rate c , SE) Year 6 b : 97 a (6.9, 0.7) Year 7: 104 (8.1, 0.8) PID: n (rate c , SE) Year 6 b : 5 a (0.3, 0.1) Year 7: 5 (0.3, 0.1) Side effects Bleeding problems d : n (rate c , SE) Year 6 b : 516 a (36.1, 1.3) Year 7: 540 (39.2, 1.4) Pain: n (rate c , SE) Year 6 b : 74 a (5.7, 0.7) Year 7: 76 (6.1 0.7) | Pregnancy Assessed at clinical follow-up: n (rate e ) Year 6 b : 0 (0) a Year 7: 0 (0) a Adverse events Expulsion (partial and complete): n (rate e ) Year 6 b : 5 (0.7) a Year 7: 7 (1.2) a PID: n (rate e ) Year 6 b : 0 (0) a Year 7: 0 (0) a Side effects Bleeding problems d : n (rate e ) Year 6 b : 37 (3.7) a Year 7: 24 (3.1) a Pain: n (rate e ) Year 6 b : 3 (0.4) a Year 7: 2 (0.4) a | Quality: good Risk of bias: low Failed to describe enrollment rate or characteristics of those lost to follow-up; otherwise exposure and outcomes assessed reliably and consistently Precision: good Sample size calculated and enrollment targets met External validity: good Large, diverse, multicountry study |
| McNicholas et al, 2017 Funding: Society of Family Planning, anonymous foundation, Washington University Institute of Clinical and Translational Sciences National Center for Research Resources and the Eunice Kennedy Shriver National Institute of Child Health and Human Development | Noncomparative cohort study Family planning clinics, St. Louis, Missouri, USA 2012–2016 | Women who were English- or Spanish-speaking, sexually active with male partner, desired to avoid pregnancy for at least 1 year, willing to continue use of LNG-IUD for at least an additional year, within 57–63 months of initiation Number starting extended use, n = 496 Age range 18–45 years, 29.6% nulligravid Extended use: mean 19.3 months (range, 0.2–41.1 months) | Phone follow-up with surveys every 6 months for 36 months or until IUD removal Loss to follow-up: not reported Discontinuation: not reported | N/A | Pregnancy Assessed by patient report over the phone, confirmed clinically: n 2 in total extended time Year 6: 0.25 (95% CI, 0.04–1.42) per 100 woman-years Year 7: 0.43 (95% CI, 0.08–2.39) per 100 woman-years | Quality: fair Risk of bias: high Rates of loss to follow-up, discontinuation, and characteristics of those lost were not described, pregnancy assessment through patient report over the phone Precision: good Large cohort with narrow confidence intervals for the primary outcome External validity: good Cohort well described with racial and socioeconomic diversity, included both parous and nulliparous and obese and nonobese participants |
| Bahamondes et al, 2018 Funding: Partial support from the Fundação de Apoio à Pesquisa do Estado de São Paulo award and from the Brazilian National Research Council, and donations from the International Contraceptive Access Foundation. IUDs placed after 2007 donated by the International Contraceptive Access Foundation, Turku, Finland under an unrestricted grant | Noncomparative cohort study (chart review) Family planning clinic at the Department of Obstetrics and Gynecology at the University of Campinas Medical School, Campinas, São Paulo, Brazil 1990–2017 | Women who received an LNG-IUD for contraceptive purposes only Number starting extended time, n = 776 Mean age at IUD initiation 32 years (0.24 SE, range, 15–44 years); 45% with 0–1 prior pregnancy Extended time: median 13 months (range, 1–124 months) | Follow-up through routine care, attempted phone interview for those without follow-up in clinic Loss to follow-up: 42 (5.4%) Discontinuation reason (rates c during extended time): bleeding (2.3), expulsion (2.1), infection (3.2), menopause (5.7), other medical reason (1.8), planning pregnancy (7.1), change to permanent method (2.1), other personal reason (3.2), end of IUD life span (11.1) Extended: median 13 months (range, 1–184 months) | N/A | Pregnancy Assessed by chart review and phone: n 0 in extended use Adverse events Infection: rate f (95% CI) Years 6–7: 3.2 (0.7–5.6) Years 8–15: 0 Expulsion: rate f (95% CI) Years 6–7: 0.3 (0.0–0.7) Years 8–9: 0.6 (0.0–1.4) Years 10–11: 0 Years 12–13: 1.1 (0.0–2.4) Years 14–15: 2.1 (0.0–4.4) Side effects Removal for bleeding/pain: rate f (95% CI) Years 6–7: 0.2 (0.0–0.6) Years 8–9: 2.3 (0.0–4.6) Years 10–15: 0 | Quality: good Risk of bias: low Characteristics of those loss to follow-up not described, otherwise low risk of selection or attrition bias. Assessment of exposure and outcomes valid and implemented consistently. Precision: good Large sample size with generally narrow confidence intervals where calculated. External validity: unclear Single site in urban setting, publically funded clinic all from “low and middle strata of the population”; racial and medical comorbidities not described. |
b Numbers estimated from Appendix 1 of study.
c Cumulative termination rate per 100 participants
d Using WHO definitions of amenorrhea, infrequent and regular bleeds
e Year-specific termination rate per 100 participants, calculated from study data
Pregnancy
No pregnancies were observed in 3 of the 4 studies during 1–3 years of extended use beyond the 5-year approved duration, or 6–8 years of total use ( Table 1 ). The fourth study reported 2 pregnancies, with a pregnancy rate of 0.25 (95% CI, 0.04–1.41) per 100 woman-years in year 6, and a rate of 0.43 (95% CI, 0.08–2.39) per 100 woman-years in year 7. Pooling these data, we calculated the pregnancy rate in year 6 to be 0.02 per 100 person-years (95% CI, 0.00–0.45), in year 7 to be 0.03 per 100 person-years (95% CI, 0.00–0.71), and the rate for years 6 and 7 combined to be 0.02 per 100 person-years (95% CI, 0.00–0.29) ( Table 2 ).
| Study | Year 6 | Year 7 | Years 6 and 7 | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Number of participants starting | Number of pregnancies | Loss to follow-up | Total person-years a | Pregnancy rate (95% CI) c | Number of participants starting | Number of pregnancies | Loss to follow-up | Total person-years a | Pregnancy rate (95% CI) c | Pregnancy rate (95% CI) c | |
| Ronnerdag 1999 | 100 | 0 | 18 | 91 | 0.00 (0.00–4.05) | 82 | 0 | nr | nr | nc | nc |
| Rowe 2016 | 717 | 0 | 116 | 659 | 0.00 (0.00–0.56) | 601 | 0 | 203 | 499.5 | 0.00 (0.00–0.74) | 0.00 (0.00–0.32) |
| McNicholas 2017 | 496 | 1 | nr | 400 b | 0.25 (0.01–1.39) | nr | 1 | nr | 232.6 b | 0.43 (0.01–2.39) | 0.29 (0.03–1.04) |
| Bahamondes 2018 | 776 | 0 | 592 | 480 | 0.00 (0.00–0.77 | 184 | 0 | 70 | 149.0 | 0.00 (0.00–2.48) | 0.00 (0.00–0.59) |
| Overall (pooled) | 2089 | 1 | nc | 1630 | 0.02 (0.00–0.45) | nc | 1 | nc | nc | 0.03 (0.00–0.71) | 0.02 (0.00–0.29) |
a In Ronnerdag, Rowe, and Bahamondes, we assumed that pregnancies or loss to follow-up occurred mid-year, so these participants contributed half of a year to the denominator
b Total person-years was estimated using the formula: (Number of pregnancies)/(Pregnancy rate) × 100
c Rate per 100 person-years; 95% CIs of individual studies were calculated using the exact Poisson method. The overall rate with associated 95% CIs were calculated using the Poisson inverse variance method.
Adverse events
Three studies collected data on adverse events occurring during extended use, including IUD expulsion and IUD-related infections ( Table 1 ). For expulsions, 1 study reported no expulsions in up to 3 years of extended use ; another found expulsion rates of 0.7 and 1.2 per 100 participants in years 6 and 7, respectively ; and the third reported an expulsion rate of 0.3 (95% CI, 0.0–0.7) per 100 participants for years 6–7. For infections, 2 studies reported no infections during extended use, , and a third reported a rate of 3.2 (95% CI, 0.7–5.6) per 100 participants in years 6–7.
Side effects
Two studies reported on side effects leading to discontinuation ( Table 1 ). , One found discontinuation rates for bleeding of 3.7 per 100 participants in year 6 and 3.1 per 100 in year 7, and discontinuation rates for pain of 0.4 per 100 participants in years 6 and 7. The other study found 0.2 per 100 participants discontinuing owing to bleeding or pain during years 6–7.
Copper T380A
Two studies examined the Cu T380A, which is approved for 10 years of use ( Table 3 ). , These studies included 473 participants using the device for up to 6 years beyond the 10-year approved duration. , One good-quality study had 245 women start extended use with 2 years of follow-up. It was a multicountry trial with diverse participants, low risk of bias, good precision, and good external validity. The other study was of fair quality, and was a single-center study of 228 women in Brazil who started extended use at age 35 or older and were followed for up to 6 years. It had a low risk of bias and good precision; however, the older age of the participants made the external validity poor.
