Objective
We sought to compare complication and continuation rates of the levonorgestrel intrauterine system (LNG-IUS) with the subdermal etonogestrel (ENG) implant across the United States among girls and women 15-44 years of age.
Study Design
A retrospective study of health insurance claims records from 2007 through 2011 identified a cohort of women who had LNG-IUS (n = 79,920) or ENG implants (n = 7374) inserted and had insurance coverage for 12 months postinsertion. Claims for complications were examined 12 months after insertion, or until removal of either device within each of 3 age groups.
Results
After its introduction in 2007, the frequency of ENG implants increased each year and almost a third of all insertions were in teenagers. However, among women ≤24 years old who had delivered an infant in the prior 8 weeks, a LNG-IUS was more likely to be inserted than an ENG implant ( P < .05). The most frequent complications with both methods were related to abnormal menstruation, which was more likely to occur among ENG implant users. Overall, 83-88% of the entire sample used their chosen method for at least 12 months. The odds of continuation were similar for both methods among teenagers, but ENG implants were more likely to be removed prematurely among women 20-24 years old (odds ratio, 1.21; 95% confidence interval, 1.06–1.39) and 25-44 years old (odds ratio, 1.49; 95% confidence interval, 1.35–1.64).
Conclusion
Both of these long-acting contraceptive methods are well tolerated among women of all ages, and demonstrate high continuation rates.
During 2008, >50% of all pregnancies in the United States were unplanned. This high rate of unintended pregnancies could be significantly reduced through increased use of highly effective long-acting reversible contraception (LARC), such as subdermal etonogestrel (ENG) implants and intrauterine devices (IUDs). The ENG implant has a failure rate approaching 0% while the levonorgestrel intrauterine system (LNG-IUS) has a low failure rate of 0.2% over the course of 1 year. In spite of its demonstrated high efficacy, relatively few women in the United States are prescribed LARC each year. Although increasing in recent years, only 5% of sexually active adolescent and young women reported using IUDs from 2002 through 2010. In 2009, <1% of US women using contraceptives reported using ENG implants, although this may be due to the limited time they had been available. As a result of LARC’s meager use, most women of reproductive age who desire contraception are still prescribed less effective methods.
One reason for the low rate of LARC utilization may be concerns that it has side effects or increases the risk of pelvic inflammatory disease (PID). For example, ENG users often experience irregular or frequent bleeding. Moreover, concerns about future fertility appear to be the reason that providers infrequently recommend IUDs, even though the American Congress of Obstetricians and Gynecologists states that IUDs are safe to use in teens and nulliparous women. As a result of these concerns, more than half of US health practitioners provide IUDs rarely or never to nulliparous women or teenagers. Two recent studies partially addressed these concerns by demonstrating that PID, uterine perforation, and ectopic pregnancies occur infrequently (<1%) during the first year of use among women using modern IUDs. These studies did not, however, directly compare side effects of the ENG implant with LNG-IUS using age-stratified analyses. As one Cochrane review pointed out, these types of comparisons are important to help providers counsel their patients on the best type of LARC for them.
Finally, it is important to understand how long women continue to use different LARC methods. Prior studies have shown that LARC is used significantly longer than oral contraceptive pills, patches, rings, or barrier methods, and satisfaction with LNG-IUS and the ENG implant is high. However, these research studies were conducted in populations that may differ from the general population. Further, these studies did not directly compare continuation rates between different LARC methods within different age groups, but instead assessed continuation of each method separately. The objective of this study was to compare 2 different methods of LARC (LNG-IUS vs ENG implant) by examining: (1) relative use trends over time, (2) complication rates, and (3) continuation rates among teenagers 14-19 years of age vs women 20-24 years and 25-44 years of age.
Materials and Methods
This retrospective cohort study used 2007 through 2011 health insurance claims from Clinformatics DataMart (OptumInsight Life Sciences Inc, Eden Prairie, MN). The data set contains information on >45 million individuals, of which approximately 80% purchase their health insurance through their employer. The dataset has been deidentified and does not contain any information on socioeconomic status or race/ethnicity. In general, this dataset is roughly representative of the US working population, with some overrepresentation in the South. Data reflect only health claims that have been paid by the insurance company, based on submissions from doctors across the United States. This study was exempted from full review by the University of Texas Medical Branch Institutional Review Board.
For our study, we included only the LNG-IUS and the ENG implant LARC devices to directly compare these methods. We did not include the copper IUD because it is used rarely among teenagers, and the continuation and side effects of this method have already been compared to the LNG-IUS in different age groups. We identified 131,634 girls and women between 15-44 years of age who had a claim for the insertion of a LNG-IUS and 12,381 for an ENG implant from 2007 through 2011, with 1 year of follow-up. Including follow-up time, our results extended into 2012. Subjects were chosen if they had a Healthcare Common Procedure Coding System code (J7302, S4989, S4981) combined with either a Current Procedural Terminology ( CPT ) code (58300) or International Classification of Diseases, Ninth Revision ( ICD-9 ) code (69.7, V25.1, V25.42) that indicated insertion of the LNG-IUS. Insertion of the ENG implant device was determined by a Healthcare Common Procedure Coding System code (J7307).
Of all women who had a LNG-IUS inserted during 2007 through 2011, a total of 79,920 were included in the study after excluding those who were immunocompromised (042, 043, 044, 279.0, 279.1, 279.2, 279.3, 795.71, V08), autistic or mentally impaired (299, 299.0, 299.00, 299.01, 317, 318), and not enrolled for 12 continuous months from the date of LARC insertion ( Appendix ; Supplementary Figure ). After applying the same criteria, 7374 women who had the ENG implant device inserted were included in the study. Mean age of included LNG-IUS users was 31.6 ± 6.3 years compared to 32.1 ± 8.3 years among the excluded. Mean age of ENG implant users was 24.1 ± 6.9 years compared to 24.6 ± 7.4 years among the excluded.
Other data included in this study were year of birth, year of device insertion, type of provider, removal, and whether it was inserted within 8 weeks of childbirth. To determine age at insertion, birth year was subtracted from year of LARC insertion. Provider specialties were categorized and included: obstetricians/gynecologists, pediatricians, family practitioner/internal medicine/general practitioners, clinics, nonphysician providers, and specialists. The specialist category includes physicians with specialties that do not fit the other categories. In the absence of a code, providers were categorized as “unknown” because it was not known if the provider was a physician or a nonphysician provider.
Claims for LNG-IUS or ENG implant insertion within 8 weeks of delivery were examined by age group. Complication claims and removal of the device were examined across the 12 months following device insertion. Claims for complications were only examined until removal of either LARC method. Complications included: pain associated with female genital organs (dyspareunia, dysmenorrhea, or premenstrual tension), disorders of menstruation (excessive menstruation, dysfunctional or functional uterine hemorrhage, postcoital bleeding, endometrial hyperplasia, metrorrhagia, irregular menstrual cycles, absence of menstruation, scanty or infrequent menstruation), and inflammation/infection (PID, including inflammatory diseases of the ovary, fallopian tube, pelvic cellular tissue, and peritoneum; and inflammatory disease of the uterus except cervix; cervicitis and endocervicitis; and cystitis). The frequency of normal intrauterine pregnancy and abnormal pregnancy (ectopic pregnancy and molar pregnancy/abnormal products of conception/missed or spontaneous abortion) that occurred between device insertion and removal was also examined.
Early discontinuation of LARC (within 12 months of insertion) was examined using ICD-9 or CPT codes for LNG-IUS removal ( ICD-9 codes 97.71, V25.12, V25.13; CPT codes 58301, 58562) or for ENG implant removal ( CPT codes 11976, 11982). Associations among age, LARC method, and removal of LARC within 30 days of a code indicating abnormal bleeding were also investigated.
Statistical analyses
Descriptive statistics included: (1) the distributions of year of insertion, age, and provider specialty by LARC method; and (2) the frequencies of complications, failure, and early discontinuation by age group and LARC method. The ratio of insertion of ENG implants compared to LNG-IUS across time was plotted by age group. Logistic regression models estimated the associations of the type of contraceptive inserted (ENG implant vs LNG-IUS) with age, year of insertion, and physician specialty. Multivariate logistic regression models directly compared outcomes for the ENG implant with LNG-IUS. Each outcome (complication, discontinuation) was coded using a binary response, with 1 indicating that the outcome occurred, and 0 indicating that the outcome was not present. Specifically, the models in this study evaluated the effects of: (1) age on the type of contraceptive used within 8 weeks of a delivery; (2) age, type of contraceptive, and their interactive effect on the outcome variables; (3) age group; and (4) the association of abnormal bleeding with discontinuation. Multivariate logistic regression models controlled for age at insertion, contraceptive type, provider type, and year of contraceptive insertions. Statistical software (SAS, version 9.3; SAS Institute, Cary, NC) was used to conduct all statistical analyses.
Results
The proportion of total ENG implants that were inserted each year increased greatly from 2007 through 2011 ( Table 1 ). In contrast, the proportion of total LNG-IUSes inserted each year was relatively evenly distributed across each year of study, although the proportion inserted in 2007 was lower than the following years. LNG-IUSes were more frequently placed in women 25-44 years of age than women ≤24 years of age, while most ENG implants (60%) were placed in women ≤24 years of age. Among all provider types, obstetrician/gynecologists most frequently inserted either device. The majority of LNG-IUS and ENG implant users did not pay a copay for device insertion (98.2% and 98.7%, respectively.) For those with copays, the mean cost was $28.70 for LNG-IUS and $26.80 for ENG implants (results not shown).
Variable | ENG implant (n = 7374) | LNG-IUS (n = 79,920) | ||
---|---|---|---|---|
n | Column % | n | Column % | |
Year | ||||
2007 | 504 | 6.8 | 12,559 | 15.7 |
2008 | 1150 | 15.6 | 18,140 | 22.7 |
2009 | 1330 | 18.0 | 18,007 | 22.5 |
2010 | 1847 | 25.0 | 15,245 | 19.1 |
2011 | 2543 | 34.5 | 15,969 | 20.0 |
Age, y | ||||
15-19 | 2388 | 32.4 | 2204 | 2.8 |
20-24 | 2014 | 27.3 | 8988 | 11.2 |
25-44 | 2972 | 40.3 | 68,728 | 86.0 |
Provider type | ||||
Obstetrics/gynecology | 6386 | 86.6 | 67,502 | 84.5 |
FP/GP/IM/pediatrics | 459 | 6.3 | 3853 | 4.8 |
Clinics | 55 | 0.7 | 6083 | 7.6 |
Nonphysician provider | 366 | 5.0 | 1893 | 2.4 |
Specialists | 64 | 0.9 | 406 | 0.5 |
Unknown | 44 | 0.6 | 183 | 0.2 |
Among girls and women who had either LARC method inserted within 8 weeks of delivery, those 15-19 years old and 20-24 years old were less likely to have an ENG implant inserted than the LNG-IUS ( Table 2 ). The interaction term between age and type of device confirmed that the IUD was preferred by younger women within 8 weeks of delivery while preference for ENG implant contraception within 8 weeks of delivery increased with age.
Age, y | ENG implant n (%) | LNG-IUS n (%) | OR (95% CI) of ENG implant vs LNG-IUS |
---|---|---|---|
15-19 | 153 (6.4) | 438 (19.8) | 0.28 (0.23–0.34) a |
20-24 | 240 (11.9) | 1803 (20.0) | 0.54 (0.47–0.62) a |
25-44 | 415 (13.9) | 9809 (14.3) | 0.98 (0.88–1.08) |
Serious complications among women using either the LNG-IUS or ENG implant were rare, with <1% of the women in this study having experienced PID ( Table 3 ). Ectopic pregnancy and molar pregnancy were also rare. ENG implant users from all age groups were more likely to experience metrorrhagia or irregular menstrual cycles within 1 year of insertion compared to LNG-IUS users. Women who were 20-24 years old and 25-44 years old were more likely to discontinue the ENG implant within 1 year of insertion, while there were no significant differences in the likelihood of discontinuing either LARC method in the first year among 15- to 19-year-old girls and women. Overall, normal pregnancies were rare during use of either device. Normal pregnancy occurred in 0.9% of women using LNG-IUS, and 0.8% using ENG implants. The odds for normal pregnancy were lower for 15- to 19-year-old ENG implant users compared to those using the LNG-IUS. The odds of pregnancy were similar for both devices among 20- to 24-year-old and 25- to 44-year-old women. Young ENG implant users (15-19 years old) were less likely to experience dyspareunia, inflammatory disease of the uterus, or cervicitis compared to those using LNG-IUS. ENG implant users 20-24 years of age were less likely to experience dysmenorrhea or inflammatory disease of the uterus compared to those with LNG-IUS. An abnormal pregnancy, although extremely rare for both methods, was twice as likely among ENG implant users as LNG-IUS users only in the 20- to 24-year-old group. ENG implant users aged 25-44 years were more likely to experience excessive menstruation, uterine hemorrhage, cervicitis, or to have the device removed within 1 year. Among LNG-IUS users, early removal more frequently occurred among those who had insertions in clinics (18.7%) and by specialists (16.3%) followed by obstetrician/gynecologists (11.3%), general practitioners (11.2%), and nonphysician providers (10.7%). Early discontinuation of ENG implants occurred more frequently among those who had their device inserted in a clinic (16.4%) or by obstetrician/gynecologists (15.2%) followed by nonphysician providers (12.6%), general practitioners (12.6%), and specialists (12.5%).
Outcomes | Age 15-19 y | Age 20-24 y | Age 25-44 y | ||||||
---|---|---|---|---|---|---|---|---|---|
n (column %) | OR (95% CI) c | n (column %) | OR (95% CI) c | n (column %) | OR (95% CI) c | ||||
ENG implant | LNG-IUS | ENG implant | LNG-IUS | ENG implant | LNG-IUS | ||||
Total, N | 2388 | 2204 | – | 2014 | 8988 | – | 2972 | 68,728 | – |
Outcome a | |||||||||
Dyspareunia b | 20 (0.8) | 36 (1.6) | 0.51 (0.29–0.88) d | 29 (1.4) | 180 (2.0) | 0.72 (0.48–1.06) | 45 (1.5) | 821 (1.2) | 1.27 (0.94–1.72) |
Dysmenorrhea | 59 (2.5) | 62 (2.8) | 0.88 (0.61–1.26) | 25 (1.2) | 191 (2.1) | 0.58 (0.38–0.88) d | 44 (1.5) | 1008 (1.5) | 1.01 (0.75–1.37) |
Premenstrual tension | 3 (0.1) | 6 (0.3) | 0.46 (0.12–1.85) | 5 (0.2) | 36 (0.4) | 0.62 (0.24–1.58) | 10 (0.3) | 467 (0.7) | 0.49 (0.26–0.92) d |
Excessive or frequent menstruation | 101 (4.2) | 74 (3.4) | 1.27 (0.94–1.73) | 80 (4.0) | 233 (2.6) | 1.55 (1.20–2.01) d | 132 (4.4) | 2561 (3.7) | 1.20 (1.00–1.44) d |
Dysfunctional or functional uterine hemorrhage | 122 (5.1) | 97 (4.4) | 1.17 (0.89–1.54) | 92 (4.6) | 343 (3.8) | 1.21 (0.95–1.53) | 132 (4.4) | 2494 (3.6) | 1.24 (1.03–1.48) d |
Postcoital bleeding | 6 (0.3) | 9 (0.4) | 0.61 (0.22–1.73) | 13 (0.6) | 47 (0.5) | 1.24 (0.67–2.29) | 8 (0.3) | 209 (0.3) | 0.89 (0.44–1.80) |
Endometrial hyperplasia | 0 (0.0) | 2 (0.1) | n/a | 0 (0.0) | 5 (0.1) | n/a | 5 (0.2) | 65 (0.1) | 1.78 (0.72–4.42) d |
Metrorrhagia | 68 (2.8) | 36 (1.6) | 1.77 (1.17–2.66) d | 45 (2.2) | 136 (1.5) | 1.49 (1.06–2.09) d | 65 (2.2) | 1156 (1.7) | 1.31 (1.02–1.68) d |
Irregular menstrual cycle | 178 (7.5) | 124 (5.6) | 1.35 (1.07–1.71) d | 151 (7.5) | 445 (5.0) | 1.56 (1.29–1.88) d | 165 (5.6) | 3012 (4.4) | 1.28 (1.09–1.51) d |
Absence of menstruation | 81 (3.4) | 75 (3.4) | 1.00 (0.72–1.37) | 58 (2.9) | 261 (2.9) | 0.99 (0.74–1.32) | 61 (2.1) | 1277 (1.9) | 1.11 (0.85–1.44) |
Scanty or infrequent menstruation | 2 (0.1) | 8 (0.4) | 0.23 (0.05–1.09) | 7 (0.3) | 26 (0.3) | 1.20 (0.52–2.77) | 3 (0.1) | 172 (0.3) | 0.40 (0.13–1.26) |
Pelvic inflammatory disease | 1 (0.0) | 4 (0.2) | 0.23 (0.03–2.06) | 1 (0.0) | 14 (0.2) | 0.32 (0.04–2.42) | 2 (0.1) | 41 (0.1) | 1.13 (0.27–4.67) |
Inflammatory disease of uterus except cervix | 2 (0.1) | 10 (0.5) | 0.18 (0.04–0.84) d | 1 (0.0) | 45 (0.5) | 0.10 (0.01–0.72) d | 3 (0.1) | 188 (0.3) | 0.37 (0.12–1.15) |
Cervicitis and endocervicitis b | 30 (1.3) | 45 (2.0) | 0.61 (0.38–0.97) d | 53 (2.6) | 206 (2.3) | 1.15 (0.85–1.56) | 64 (2.2) | 1003 (1.5) | 1.49 (1.15–1.92) d |
Ectopic pregnancy | 1 (0.0) | 1 (0.0) | 0.92 (0.06–14.8) | 1 (0.0) | 8 (0.1) | 0.56 (0.07–4.46) | 0 (0.0) | 36 (0.1) | n/a |
Abnormal pregnancy or spontaneous abortion | 7 (0.3) | 5 (0.2) | 1.29 (0.41–4.08) | 10 (0.5) | 20 (0.2) | 2.24 (1.05–4.79) d | 8 (0.3) | 121 (0.2) | 1.53 (0.75–3.13) |
Normal pregnancy b | 14 (0.6) | 32 (1.5) | 0.40 (0.21–0.75) d | 18 (0.9) | 118 (1.3) | 0.68 (0.41–1.12) | 28 (0.9) | 561 (0.8) | 1.16 (0.79–1.69) |
Hematoma of upper arm | 0 (0.0) | n/a | 0 (0.0) | n/a | 0 (0.0) | n/a | |||
Removal within 1 y b | 295 (12.4) | 257 (11.7) | 1.07 (0.89–1.28) | 305 (15.1) | 1152 (12.8) | 1.21 (1.06–1.39) d | 493 (16.6) | 8102 (11.8) | 1.49 (1.35–1.64) d |