Profiles of copper intrauterine devices and levonorgestrel intrauterine systems users in France in 2019: A national observational population-based study.

NR and MZ conceived and designed the experiments. NR, LD, RSD, AW, and MZ analyzed the data. NR and MZ wrote the paper. LR, RSD, and AW revised the work critically for important intellectual content. NR performed the literature review. BB extracted data from the French hospitalization database. Each author has confirmed compliance with the journal's requirements for authorship.

In 2019, a total of 477 705 Cu‐IUDs and 355 242 LNG‐IUSs were dispensed to French women aged 13–49 years. The sociodemographic characteristics and medical history of the women who received a Cu‐IUD/IUS in 2019 are shown in Table  1 . The mean age of the women was 32.5 ± 7.3 years for Cu‐IUDs and 36.4 ± 7.7 years for LNG‐IUS. Sociodemographic and medical characteristics of the women ( n  = 832 947) at inclusion according to insertion of Cu‐IUD or LNG‐IUS in 2019 a 4027 (0.8) Abbreviations: Cu‐IUD, copper‐intrauterine device; IQR, interquartile range; LNG‐IUS, levonorgestrel‐intrauterine system; SD, standard deviation. Values are given as number (percentage) unless otherwise indicated. In the 10 years before insertion. At least two dispensations in the previous year. In the Cu‐IUD group, 56 612 (11.9%) women were affiliated to the Universal Health Insurance coverage (vs 36 282 [10.2%] in the LNG‐IUS group), and 71 911 (15.1%) belonged to the most deprived social group (vs 58 141 [16.4%] in the LNG‐IUS group). A larger proportion of LNG‐IUSs than of Cu‐IUDs was prescribed by gynecologists (55.0% vs 45.6%), whereas prescriptions by midwives were more frequent in the Cu‐IUD group (195 282 [21.2%] vs 217 718 [11.8%]). The proportion of prescriptions by general practitioners was similar for the two groups. A large proportion of the women who had a Cu‐IUD or LNG‐IUS in 2019 had a history of device dispensation 5–6 years previously: 232 969 (48.8%) in the Cu‐IUD group and 206 891 (58.2%) in the LNG‐IUS group. More women in the LNG‐IUS group than in the Cu‐IUD group had a history of arterial disease. The proportion of women with a history of thromboembolic disease was similar in the two groups. The use of anxiolytics and antidepressants in the preceding year was more frequent in the LNG‐IUS group than in the Cu‐IUD group. The multivariate model included the following variables after the stepwise selection: age (in four age groups); gynecological history; previous dispensation of an IUD (5–6 years before); Universal Health Insurance cover; deprivation index; number of deliveries in the last 10 years; type of physician; history of breast/ovarian cancer, venous disease, or arterial disease in the last 10 years; and dispensation of anxiolytics and antidepressant in the preceding year. Dispensing of LNG‐IUS rather than a Cu‐IUD was associated with being aged older than 35 years compared with being aged 25–35 years (OR 35‐44 1.85, 95% CI 1.83–1.87, p  < 0.001; OR ≥45 3.66, 95% CI 3.59–3.73, p  < 0.001), having a prescription from a gynecologist rather than from a general practitioner (OR 1.10, 95% CI 1.09–1.11, p  < 0.001), and having a gynecological history (OR 2.28, 95% CI 2.20–2.36, p  < 0.001), after full adjustment (see Table  2 ). Associations between receiving a levonorgestrel‐intrauterine system (LNG‐IUS) rather than a copper‐intrauterine device (Cu‐IUD) in 2019 and sociodemographic and medical characteristics a 0.75 (0.74–0.76) 0.91 (0.90–0.92) 1.11 (1.10–1.12) Abbreviations: CI, confidence interval; Cu‐IUD, copper‐intrauterine device; LNG‐IUS, levonorgestrel‐intrauterine system; OR, odds ratio. Values in parentheses are 95% CI. High blood pressure, dyslipidemia, myocardial infarction, stroke. Endometrial hyperplasia, endometriosis, polyp/myoma, heavy menstrual bleeding. By contrast, having a history of pregnancy in the preceding year (OR 0.65, 95% CI 0.64–0.66, p  < 0.001) and having prescription from a midwife rather than from a general practitioner (OR 0.65, 95% CI 0.64–0.66, p  < 0.001) was statistically associated with choosing a Cu‐IUD rather than a LNG‐IUS. Among the LNG‐IUS group, 217 515 (61.2%) women received a LNG‐IUS 52 mg, 109 018 (30.7%) received a LNG‐IUS 19.5 mg, and 28 709 (8.1%) received a LNG‐IUS 13.5 mg (Table  3 ). The women receiving a LNG‐IUS 13.5 mg were the youngest (mean age 31.3 ± 8.5 years). Those receiving a LNG‐IUS 52 mg were the oldest (mean age 38.4 ± 6.8 years) and the most likely to be affiliated to the Universal Health Insurance. The LNG‐IUS 19.5 mg was more often prescribed by gynecologists than devices with the two other doses. More than half of the LNG‐IUS users had a history of using a copper or hormonal IUD 5–6 years previously. Women who used the dose of 52 mg were more likely to have a gynecological history than those using the other doses. Sociodemographic and medical characteristics of the women ( n  = 355 242) at the time of dispensation of LNG‐IUS, by LNG dose, in 2019 a Abbreviations: Cu‐IUD, copper‐intrauterine device; IQR, interquartile range; LNG‐IUS, levonorgestrel‐intrauterine system; SD, standard deviation. Values are given as number (percentage) unless otherwise indicated. In the 10 years before insertion. At least two dispensations in the previous year. Details of the continuation rates are presented in Table  4 . One year after dispensation, 412 982 (86.4%) women were considered to still be using an IUD and 304 636 (85.7%) a LNG‐IUS (84.3% for 52 mg, 83.9% for 19.5 mg, and 86.9% for 13.5 mg). Among the women who were no longer using either device, 63 892 (13.4%) in the Cu‐IUD group and 48 974 (13.8%) in the LNG‐IUS group switched for another form of contraception (mostly oral contraception). Switch rates were highest for the LNG‐IUS 19.5 mg group (15.5%). Continued use of Cu‐IUD and LNG‐IUS (by dose) 1 year after insertion: contraceptive switch, indicators of pregnancy and removal acts ( n  = 832 947) a Abbreviations: Cu‐IUD, copper‐intrauterine device; LNG‐IUS, levonorgestrel‐intrauterine system. a Values are given as number (percentage). No other contraceptive delivered, no pregnancy indicator, and no removal act code in the year after dispensation of the device. Second trimester ultrasound ≥180 days or abortion ≥60 days after dispensation of the device. The multivariate model concerning the continued use 1 year after insertion for all devices included the following variables after the stepwise selection: age (in four age groups); type of device; gynecological history; arterial history; history of breast cancer; Universal Health Insurance cover; deprivation index; history of prior use of IUDs 5–6 years before; deliveries in the last 10 years; history of pregnancy in the preceding year; type of physician; and dispensation of anxiolytics or antidepressants in the preceding year. Having a dispensation of a Cu‐IUD in 2019 rather than a LNG‐IUS (OR 1.20, 95% CI 1.19–1.22, p  < 0.001), being aged older than 35 years rather than 25–35 years (OR 35–44 1.30, 95% CI 1.28–1.32, p  < 0.001; OR ≥45 1.87, 95% CI 1.81–1.93, p  < 0.001), having previous use of an IUD (OR 1.47, 95% CI, 1.44–1.49, p  < 0.001), and having a device prescription from a gynecologist rather than a general practitioner (OR 1.29, 95% CO 1.27–1.31, p  < 0.001) are significantly associated with a higher continuation rate 1 year later. ORs are described in Table  5 . Associations between use of IUD 1 year after dispensation and sociodemographic and medical characteristics a Abbreviations: CI, confidence interval; Cu‐IUD, copper intrauterine device; IUD, intrauterine device; LNG‐IUS, levonorgestrel intrauterine system; OR, odds ratio. Unless otherwise stated, p  < 0.001. Values in parentheses are 95% CI. At least two dispensations in the preceding year.

The dispensing of IUDs under real‐life conditions in France over an entire year and in the whole French population were described. Two types of user profiles emerged from the study results depending on the type of device: women aged over 35 years with a gynecological history or a gynecological prescription were more likely to be prescribed a LNG‐IUS rather than a Cu‐IUD; and women with a prescription from a midwife or a pregnancy in the preceding year were more likely to be prescribed a Cu‐IUD. High continuation rates 1 year after dispensation were found for both IUDs: six out of seven women are still using the IUD 1 year after dispensation. After adjustment, having a dispensing of a Cu‐IUD rather than a LNG‐IUS was associated with a higher chance of still using the device 1 year later. In the present study, women who received a LNG‐IUS were older than those receiving a Cu‐IUD, and were less likely to be affiliated to the Universal Health Insurance. On the contrary, women with a LNG‐IUS were more likely to live in the most deprived areas, but there was only a very slight difference between the groups, and living in a deprived area is not necessarily a relevant notion to define the socioeconomic position of people. Previous national French surveys have shown that IUDs/IUSs have appeared to replace the pill as the contraceptive method of choice, mostly in older women and women with a higher socioeconomic status. 4 These surveys did not distinguish between types of IUDs. Lower levels of financial constraint are known to be associated with increased use of IUDs, including for younger women. 14 , 15 In France, dispensation of IUDs/IUSs is reimbursed at a rate of at least 65% by the National Health Insurance. However, access to health professionals for a placement of a device may depend on several factors (heterogeneous medical demography, more expensive consultations for the gynecologist than the general practitioner because of an extra fee charge). The financial incentive for general practitioners to insert an IUD could be a lever to promote dispensation of Cu‐IUD/LNG‐IUS and to decrease unequal access to long‐acting reversible contraception as observed in the United Kingdom. 16 Promoting consultations about contraception with midwives outside the perinatal period could be another option. 17 Lastly, training other health professionals, such as nurses, on the insertion of IUDs is a possibility that should be considered. 18 In the present study, women who had a pregnancy history in the year before dispensation of a device were more likely to get a Cu‐IUD rather than a LNG‐IUS. A recent Brazilian monocentric study has only showed that users of Cu‐IUDs were less likely to be nulliparous compared to users of LNG‐IUSs, but it was not focused on the postpartum period. 19 In the present study, the LNG‐IUS appeared to be prescribed according to French professional recommendations. The National College of French Gynecologists and Obstetricians recommends the use of LNG‐IUS in absolute/relative hyperestrogenic environments, endometriosis, adenomyosis, and endometrial hyperplasia. 10 Phase II and III studies showed the 13.5 and 19.5 mg LNG‐IUSs to result in lower systemic exposure to LNG with a similar impact on the endometrium and cervical function as the 52 mg device. 20 However, evidence of efficacy for conditions other than contraception has been established for the 52 mg LNG‐IUS only, and only the 52 mg LNG‐IUS is recommended in France for heavy menstrual bleeding and endometriosis. In the contraceptive CHOICE project, the 1‐year continuation rates for LNG‐IUS and Cu‐IUD were 88% and 84%, respectively. 21 The continuation rates that were found are comparable with these figures. It was shown that dispensing of Cu‐IUDs rather LNG‐IUSs is associated with a higher chance of continued use 1 year later, and users of the low‐dose LNG‐IUS had a lower continuation rate than those who used the 52 mg LNG‐IUS. The 12‐month continuation rate of the 13.5 mg LNG‐IUS was 80.4% in the Phase III European multicenter study. 22 As of 2021, the 19.5 mg LNG‐IUS continuation rate has not been studied or compared to that of the other available methods of contraception. The lower continuation rate for the low‐dose LNG‐IUS than for the 52 mg LNG‐IUS could be explained by dissatisfaction related to bleeding disturbances or because low‐dose devices concerned younger women, although a previous study found no difference between continuation rates of LNG‐IUS 52 mg for younger and older women. 23 The present study has some limitations. One is that placement and removal of Cu‐IUDs/IUSs might not be correctly coded within the SNDS database. Additional information was used to measure effective use of IUDs (removal act, other contraceptive dispensation, and pregnancy indicators) but this could have led to errors in measurement and to an overestimation of the continuation rates. Nevertheless, the present results are consistent with previous studies. Moreover, the misclassification bias seems to be non‐differential, so the associations between use of Cu‐IUDs/LNG‐IUSs 1 year after dispensation and medical characteristics could be underestimated. Lastly, diagnoses such as heavy menstrual bleeding are made at the hospital and are not always available within the SNDS database. Indeed, women who consult outside hospitals for this condition may not be recorded and the frequency of diagnosis would therefore be underestimated. Beside these limitations, this is the first study of a nationwide cohort that precisely reflects the use of Cu‐IUDs and the various LNG‐IUSs in France. These data will make it possible to study changes in the dispensation of these contraceptive devices since the French pill crisis in 2012. Lastly, these data may be of particular importance as they were collected the year before the COVID‐19 pandemic, which had a major impact on reproductive services and the use of contraceptives worldwide. 24 , 25

The vast majority of women aged 15–49 years worldwide were using contraceptive methods in 2019. 1 In France, seven out of ten women of childbearing age use medical contraception, 2 a rate consistent with the high rates of contraceptive coverage in Western Europe. 3 Three major cross‐sectional phone surveys have been performed in France to assess changes in the use of contraception from 2010 4 to 2016. 2 These studies showed an important change in contraceptive behaviors since 2012. That year, a woman brought a lawsuit against a pharmaceutical company after suffering a stroke while taking a combined pill. 5 , 6 , 7 Consequently, the use of oral contraception decreased from 50% in 2010 to 33.2% in 2016. Meanwhile, the use of intrauterine devices (IUDs) increased from 18.7% to 25.6%. This trend has been observed throughout the world. 8 Little is known about the use of IUDs in real‐life conditions, and the two types of devices have been rarely studied separately worldwide, although they do not have the same therapeutic indication. Copper IUDs (Cu‐IUDs) are a first‐line contraception, whereas, in France, levonorgestrel intrauterine systems (LNG‐IUSs) are a second‐line contraception (i.e. prescribed in case of dissatisfaction with a first contraception) and are recommended for some gynecological conditions such as endometriosis 9 or heavy menstrual bleeding. 10 It was hypothesized that users of Cu‐IUDs and LNG‐IUSs may be different and that is not relevant to describe the intrauterine contraceptives as a homogeneous group, while their indications differ. The fact that the two types of devices are confounded in previous studies does not send a clear message, as if the two types of users could be confounded. Highlighting two types of users' profiles may encourage both researchers and clinicians to study the two types of devices separately in the future to outline users' specific behaviors and needs. The aim of the present nationwide register‐based study was to analyze separately the use of Cu‐IUDs and LNG‐IUSs in France in 2019 and to assess the association between the choice of device and the women's sociodemographic and medical characteristics. A secondary aim was to assess the continuation rates of the Cu‐IUDs and the LNG‐IUSs up to 1 year after dispensation of the device and its underlying correlates using a medico‐administrative database.

All authors have no conflict of interest to disclose. NR, RDS, and MZ are employees of the French National Agency for Medicines and Health, BB and AW of the French National Health Insurance Fund. LD is employee of the Public Hospitals of Paris (APHP).

The French National Health Data System (SNDS) provides information on health insurance claims for 99.5% of the population living in France (about 67 million people). Each individual living in France is given a unique anonymous identifier that is linked to information from two principal data sources in the SNDS database: the national health insurance claims database (DCIR) and the national hospital and discharge database (PMSI). The DCIR database includes individual information on outpatient medical care and reimbursed drugs. Drugs are coded according to the Anatomical Therapeutic Chemical (ATC) classification, and medical devices according to the Reimbursed Products and Benefits List (LPP). The PMSI database contains details of all hospital admissions in France and hospital diagnoses according to the International Statistical Classification of Diseases and Related Health Problems (ICD‐10). These databases have been fully described elsewhere and are regularly used for the monitoring of drugs. 11 , 12 All dispensations of Cu‐IUDs and LNG‐IUSs concerning women living in France and aged 13–49 years were extracted in 2019. The following information was collected about the women on the date of dispensation of Cu‐IUD/LNG‐IUS in 2019: age; affiliation to the Universal Health Insurance (which provides health insurance cover for people with low incomes); social deprivation index (a geographic indicator of social deprivation); arterial (high blood pressure, dyslipidemia, myocardial infarction, stroke); venous (thrombosis, pulmonary embolism); gynecological history (gynecological conditions diagnosed according to the ICD‐10 for which LNG‐IUSs are recommended in France: heavy menstrual bleeding, endometrial hyperplasia, endometriosis, polyp, and myoma); oncological history (breast, endometrium, and ovarian cancers); number of deliveries in the 10 years before; prescriber's specialty; use of Cu‐IUD/LNG‐IUS before 2019; pregnancy; and the use of antidepressants and anxiolytics in the previous year. The following information was also collected during the year after the device was dispensed: existence of a removal act code; pregnancy indicator; or dispensation of another contraceptive. In the absence of these three criteria, a woman was considered to still be using a Cu‐IUD/LNG‐IUS 1 year after dispensing. Analyses were performed with SAS Enterprise Guide software 7.1 (SAS Institute). The frequency of dispensations of Cu‐IUD and LNG‐IUS were first described in a bivariate analysis with social, demographic, and health characteristics. Missing data were included. Logistic regression was then performed to study the association between dispensation of Cu‐IUD/LNG‐IUS and women's characteristics by calculating odds ratios (OR) and their 95% confidence intervals (CI), first with bivariate models, then with models adjusted for age in four age groups (13–24, 25–34, 35–44, ≥45 years), and, finally, in a multivariate model using stepwise regression with the backward selection of variables (selection criterion p  < 0.05). The proposed variables before selection were as follows: age (in four age groups); gynecological history; previous dispensation of an IUD (5–6 years before); Universal Health Insurance; deprivation index; pregnancy in the year before dispensation; pregnancy in the last 10 years; type of physician; history of breast/endometrial/ovarian cancer, arterial disease, or venous disease in the last 10 years; dispensation of anxiolytics or antidepressants in the preceding year. Finally, associations between use of Cu‐IUD/LNG‐IUS 1 year after dispensation and sociodemographic and medical characteristics were characterized by logistic regressions, with a bivariate analysis, a model adjusted for age in four age groups, and a multivariate model using backward stepwise regression to select variables ( p  < 0.05). The proposed variables before selection were the same as those detailed before. The present study was authorized by decree 2016–1871 on December 26, 2016. 13 As a permanent user of the SNDS, the author's team was exempt from approval from the institutional review board.