Relative Risk of Cervical Neoplasms Among Copper and Levonorgestrel-Releasing Intrauterine System Users

Abstract

Objective: To evaluate the relative risk of cervical neoplasms among copper intrauterine device (Cu IUD) and levonorgestrel-releasing intrauterine system (LNG-IUS) users.

Methods: We performed a retrospective cohort analysis of 10,674 patients who received IUDs at Columbia University Medical Center. Our data were transformed to a common data model and are part of the Observational Health Data Sciences and Informatics network. The cohort patients and outcomes were identified by a combination of procedure codes, condition codes, and medication exposures in billing and claims data. We adjusted for confounding with propensity score stratification and propensity score 1:1 matching.

Results: Before propensity score adjustment, the Cu IUD cohort included 8,274 patients and the LNG-IUS cohort included 2,400 patients. The median age for both cohorts was 29 years at IUD placement. More than 95% of the LNG-IUS cohort used a device with 52 mg LNG. Before propensity score adjustment, we identified 114 cervical neoplasm outcomes. Seventy-seven (0.9%) cervical neoplasms were in the Cu IUD cohort and 37 (1.5%) were in the LNG-IUS cohort. The propensity score matching analysis identified 7,114 Cu IUD and 2,174 LNG-IUS users, with covariate balance achieved over 16,827 covariates. The diagnosis of high-grade cervical neoplasia was 0.7% in the Cu IUD cohort and 1.8% in the LNG-IUS cohort (2.4 [95% CI 1.5-4.0] cases/1,000 person-years and 5.2 [95% CI 3.7-7.1] cases/1,000 person-years, respectively). The relative risk of high-grade cervical neoplasms among Cu IUD users was 0.38 (95% CI 0.16-0.78, P<.02) compared with LNG-IUS users. By inspection, the Kaplan-Meier curves for each cohort diverged over time.

Conclusion: Copper IUD users have a lower risk of high-grade cervical neoplasms compared with LNG-IUS users. The relative risk of cervical neoplasms of LNG-IUS users compared with the general population is unknown.

Fig. 1.. Propensity score distribution of copper intrauterine device (Cu IUD) and levonorgestrel intrauterine system (LNG-IUS) users. Area under the curve: 0.76; 75.5% is in equipoise.

Spotnitz. Intrauterine Device Use and Cervical Neoplasm Risk. Obstet Gynecol 2020.

Fig. 2.. A plot of traditional and calibrated significance testing for the propensity score stratification analysis. Estimates below the dashed line (gray area) have a P<.05 using traditional P-value calculation. Estimates in the orange areas have a P<.05 using the calibrated P-value calculation. Blue dots indicate negative controls.

Spotnitz. Intrauterine Device Use and Cervical Neoplasm Risk. Obstet Gynecol 2020.

Fig. 3.. Covariate balance showing the absolute standardized difference of a mean before and after propensity score stratification (A) and propensity score 1:1 matching (B). Each blue dot represents a covariate. A. Number of covariates: 20,086; after propensity score adjustment maximum (absolute): 0.11. B. Number of covariates: 16,827; after propensity score adjustment maximum (absolute): 0.10.

Spotnitz. Intrauterine Device Use and Cervical Neoplasm Risk. Obstet Gynecol 2020.

Fig. 4.. Kaplan-Meier plot of high-grade cervical neoplasm–free survival compared with time (days) by intrauterine device (IUD) type as calculated by propensity score 1:1 matching. The Kaplan-Meier curves are shown with CI shading. The time at risk was from 30–2,530 days. Day 0 corresponds with 30 days after the index events. The number of patients at risk in each cohort as a function of time is shown below and parallel to the x-axis. Cu IUD, copper IUD cohort; LNG-IUS, levonorgestrel-releasing intrauterine system cohort.

Spotnitz. Intrauterine Device Use and Cervical Neoplasm Risk. Obstet Gynecol 2020.

Figure

No available caption