If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
We compared the 48 and 60 month continuation rates of levonorgestrel (LNG) and copper (Cu) intrauterine devices (IUDs) among women enrolled in the Contraceptive CHOICE Project (CHOICE). Our primary outcome was continuation at 48 months.
This is a prospective cohort study of women who received an IUD through CHOICE. We randomly selected women who had either LNG or Cu IUDs inserted between January 2008 and June 2009 and contacted them by telephone. Once contacted and consented, they were asked whether they were still using their IUD. Women who reported discontinuation of the IUD were asked for the reasons and subsequent contraceptive use. Survival analysis using Cox proportional hazards was performed to assess for factors associated with discontinuation and to calculate hazard ratios.
Of the 460 women we attempted to contact, 321 (70%) were reached for interviews. Continuation data on the remaining 139 women were available from CHOICE and its substudies. Continuations at 48 and 60 months were 62.3% and 51.7% for LNG IUD and 64.2% and 55.9% for the Cu IUD, respectively. Continuation at 48 months was highest among women older than 29 years of age at insertion (LNG IUD, 72.5%; Cu IUD, 77.1%). Women younger than 24 years of age had the lowest 48 month continuation (LNG IUD, 55.4%, and Cu IUD, 53.2%). In univariable and multivariable analysis, demographic characteristics, menstrual profile, and pregnancy history were not associated with discontinuation. Age older than 29 years was associated with less discontinuation than those 24–29 years of age (hazard ratio, 0.67, 95% confidence interval, 0.47–0.96).
IUD continuation remains high (> 60%) at 48 months with no difference between Cu and LNG IUDs.
Highly effective, long-acting reversible contraceptive (LARC) methods have the potential to decrease unintended pregnancies and reduce health disparities. As was shown in the Contraceptive CHOICE Project (CHOICE), women who use short-acting contraceptives have a 20-fold increase in unintended pregnancy as compared with those using LARCs.
randomized parous women from multiple countries to copper- or levonorgestrel-containing IUDs. The copper-containing IUD (Cu-IUD) has been associated with short-term increases in bleeding and cramping, whereas the levonorgestrel-containing IUD (LNG-IUD) generally makes cramping and bleeding lighter.
The Long-Term Utilization and Continuation of Intrauterine Devices (LUCID) study sought to describe continuation of the LNG- and Cu-IUD at 48 and 60 months. Given the improved bleeding profile associated with the LNG-IUD, our hypothesis was that women choosing the LNG-IUD would have higher long-term continuation than those using the Cu-IUD at 48 months.
Materials and Methods
The LUCID study is a prospective cohort study of women who participated in CHOICE and received an IUD. CHOICE was a longitudinal, observational study that offered women in the St Louis area contraceptives of their choice without cost. The goal of CHOICE was to decrease the unintended pregnancy rate at a regional level by emphasizing the most effective methods of contraception and by eliminating the barrier of cost. The methods of CHOICE have been described previously.
This study was approved by the Human Research Protection Office at Washington University in St Louis.
The primary outcome of this study was the continuation of IUDs at 48 months; the secondary outcome was the continuation at 60 months. To determine long-term use of IUDs, we included any CHOICE participant aged 14–45 years at the time of the IUD insertion who received the IUD though CHOICE between Jan. 1, 2008, and June 30, 2009. This time frame was chosen because it would allow participants to be contacted approximately 60 months after the IUD insertion.
A random sampling of women who had an IUD inserted during this period was generated. Thus, women who both continued and who discontinued their IUDs during CHOICE would be included in the sample. We excluded the following women: (1) participated in CHOICE but did not consent to future studies (< 5% CHOICE participants); or (2) refused to give informed consent.
CHOICE participants provided multiple methods of contact including home and mobile phone numbers, e-mail addresses, a physical address, and updated contact information from hospital records as well as contact information of friends and family that could be used by the study team for follow-up.
Participants were followed up for 2–3 years, with telephone surveys at 3 and 6 months after enrollment and every 6 months thereafter. For this substudy, potential participants were attempted to be reached using the following contact algorithm: they were called during a weekday morning, a weekday afternoon, a weekday evening, and during a weekend day; an e-mail was sent to any addresses provided during CHOICE participation; and finally, a letter was sent to their home address. Attempts were made to reach participants’ alternate contacts as well when necessary.
Data were collected using a telephone-based survey once consented for participation in this study. Women were asked whether they were still using the IUD they had received through CHOICE. If they had discontinued, they were asked to recall the date it was removed and the reason for discontinuation. When a participant could recall the removal year and month but not the date, the date was imputed randomly. When they could recall the year but not month, then the month and date were imputed randomly.
In many cases, the actual removal date was available through the CHOICE clinical data. When there was a discrepancy between the self-reported removal date and the available clinical data, the date available from the medical record was used.
Study data were entered by the primary investigator (J.T.D.) and trained research assistants directly into the REDCap electronic data capture tools hosted at Washington University in St Louis.
We used frequencies, percentages, means, and SDs to describe the demographic and reproductive characteristics of the study participants. A χ2 test was performed for categorical data and a Student t test was performed for continuous data. The Kaplan-Meier survival function was used to estimate the continuation rates for the 2 IUD types at 48 and 60 months after insertion. A Cox proportional hazard model was used to estimate hazard ratios (HRs) of variables associated with discontinuation. The last point of contact in CHOICE was used as the endpoint for those who could not be contacted for status of method use.
Participants were censored if their IUD was removed because of a pregnancy or an attempted pregnancy. They were also censored at 60 months if they had their IUD removed at that time. If a patient presented with a partial expulsion of her IUD and it was removed and a new one placed during that month, this was not considered a discontinuation. Tests for collinearity were performed between variables in the final model.
Confounders were identified as variables that changed HRs by more than 10% when included in the model. Confounding variables were then included in the final multivariable analysis. An alpha level was set to 0.05 for all statistics.
Analyses were performed using Stata version 13 (StataCorp, College Station, TX). We made the following assumptions in determining our sample size: 50% continuation of LNG-IUD at 48 months and 35% continuation of Cu-IUD at 48 months. We also assumed that we would be unable to contact 20% of potentially eligible participants. To detect a 15% difference in groups with an 80% power and an alpha of 0.05, we would need 230 participants in each group for a total of 460 participants.
Of the 1,925 women who had IUDs placed in CHOICE between January 2008 and June 2009, 460 were selected randomly for participation. We were able to contact 321 women in this group. All except 6 (98%) agreed to participate in the survey. Of the 139 who could not be reached, 58 had enrolled in substudies of CHOICE; therefore, their IUD continuation status was known. The remaining 81 participants were censored at their last point of contact in CHOICE to supplement the data collected by telephone interviews. In all, continuation data were available for 379 women (82.4%).
We compared participants who completed the telephone survey with women who could not be reached. Women surveyed were slightly older (27.4 vs 26.0 years; P = .02) and were more likely to have a prior unintended pregnancy (79.5% vs 69.3%; P = .03). Otherwise, the 2 groups were demographically similar.
Baseline characteristics of the cohort are shown in Table 1 and are stratified by IUD type. When we compared the 2 groups of IUD users, age and race were the only variables that differed significantly. LNG-IUD users had a mean age of 26.1 years compared with a mean age of 27.8 years for Cu-IUD users (P = .003). LNG-IUD users were more likely to be black than Cu-IUD users (44.4% vs 34.1%, respectively, P = .04).
Table 1Baseline characteristics stratified by type of IUD
Overall (n = 460)
LNG-IUD (n = 234)
Cu-IUD (n = 226)
High school or less
College or graduate degree
Married/living with partner
Underweight (< 18.5)
Normal weight (18.5–24.9)
Obese (> 30.0)
3 or more
3 or more
History of STI
BMI, body mass index; Cu, copper; IUD, intrauterine device; LNG, levonorgestrel; SES, socioeconomic status (this is defined as receiving government aid such as welfare or food stamps or having trouble paying for basic necessities such as food, medical care, transportation, or housing); STI, sexually transmitted infection.
Diedrich. Long-term continuation of IUDs. Am J Obstet Gynecol 2015.
Continuation at 48 months among LNG-IUD users was 62.3% (95% confidence interval [CI], 55.1%–68.7%) and among Cu-IUD users was 64.2% (95% CI, 57.0–70.5%). There was no difference between continuation at 48 months and IUD type by log-rank test (P = 0.08). At 60 months, estimated LNG-IUD continuation was 51.7% and Cu-IUD continuation was 55.9%.
The Figure shows a Kaplan-Meier survival estimate by IUD type over the 60 months. Table 2 shows the specific continuation rates at 48 and 60 months between IUD users, which are further categorized by age group. Women younger than 24 years at the time of insertion had the lowest continuation overall, whereas women older than 29 years had the highest continuation. For those younger than 24 years of age, 45.5% of LNG-IUD users and 43.9% of Cu-IUD users were still using their method at 60 months. In contrast, more than two thirds of women older than 29 years of age continued to use their IUD (67.9% of LNG-IUDs and 67.2% of Cu-IUDs) at the 60 month time point.
Table 2Continuation of levonorgestrel and copper intrauterine devices
Univariable and multivariable analyses are presented in Table 3. A univariable analysis revealed no associations between demographics, socioeconomic status, or parity and discontinuation of IUD. The final multivariable model included IUD type, age category, education, and history of sexually transmitted infection (STI). Participants older than 29 years of age were less likely to discontinue as compared with our reference group of 24–29 years of age (HRadj = 0.67, 95% CI 0.47–0.96). a history of STI at baseline was associated with higher discontinuation (HRadj = 1.44, 95% CI 1.09–1.92).
Table 3Crude and adjusted hazard ratios for risk of iud discontinuation at 48 months
Our data show that IUD use remains high at 48 and 60 months. Greater than 60% of CHOICE IUD users are still using their IUD at 48 months, and approximately 50% are continuing at 60 months. Older age (older than 29 years) appears to be associated with higher continuation. When we compared continuation by IUD type, we found no difference in continuation at the 48 month time point.
The continuation rates seen in this study are higher than prior long-term IUD studies in the literature by Sivin et al.
We hypothesize that the higher continuation in our cohort is due to the fact that women could choose which method they wanted rather than having one assigned randomly, such as was done in the earlier study by Sivin et al,
which showed LNG-IUD continuation of 33% and Cu-IUD continuation of 41%.
We noted a similar continuation rate between the Cu- and LNG-IUD at 60 months. This finding is not consistent with our hypothesis.
Discontinuation of the LNG-IUD at or just prior to the 60 month time point most likely is due to concerns that contraceptive effectiveness may be waning. In fact, many (over 25%, see Table 4) of the women interviewed who discontinued their LNG-IUD between 54 and 60 months volunteered the reason for having it removed was because the device would expire after 5 years of use. More recent data indicate that the LNG-IUD is highly effective through the sixth year of use and may be effective for a longer period of time.
A major strength of this study is the use of the CHOICE cohort, in which the women could obtain their reversible contraceptive of choice at no cost. CHOICE included a large sample size of IUD users with a diverse participant population and high follow-up rates. Because of these high rates of follow-up during the 3 years of CHOICE and data available from CHOICE substudies, we were able to complement our data for the participants that we were unable to reach. Additionally, because women in CHOICE were not randomized to their IUD type, the continuation data we report are likely more representative of women in the United States where women choose which IUD to use.
There are limitations to our study. Less than 70% of our randomly selected cohort was reached for a phone interview. However, as stated in previous text, we were able to complement our continuation information with other data sources. Still, not all women in the cohort had complete updated information obtained from follow-up surveys.
Additionally, this is a subset of CHOICE participants. Although other continuation studies of CHOICE published data on the entire cohort, this study randomly selected women for inclusion to estimate continuation. CHOICE participants are from one geographic area in the Midwest and may not be representative of all US women.
Lastly, our sample may not have sufficient power to look at multiple subgroups and sociodemographic factors associated with discontinuation.
Health care expenditures are significant. By increasing LARC use, we can decrease the costs of unplanned pregnancy, reduce the need for abortion, and save health care dollars.
The views expressed herein are those of the authors and do not necessarily represent the official views of the National Institutes of Health.
The Contraceptive CHOICE Project was supported by an anonymous foundation. The Long-Term Utilization and Continuation of Intrauterine Devices study received funding from the Society of Family Planning. This study was also supported by the Washington University Institute of Clinical and Translational Sciences grant UL1 TR000448 from the National Center for Advancing Translational Sciences and award K23HD070979 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development .
Dr Diedrich and Ms Zhao report no conflicts of interest. Dr Peipert receives research support from Teva, Bayer Healthcare Pharmaceuticals, and Merck & Co, Inc and serves on advisory boards for Teva Pharmaceuticals. Dr Madden serves on a scientific advisory board for Bayer Healthcare Pharmaceuticals and a data safety monitoring board for phase 4 safety studies of Bayer contraceptive products.
Cite this article as: Diedrich JT, Madden T, Zhao Q, et al. Long-term utilization and continuation of intrauterine devices. Am J Obstet Gynecol 2015;213:822.e1-6.