BACKGROUND: United States obstetrics and gynecology professional organizations recommend that women consider using intrauterine devices (IUDs) beginning in the immediate postpartum (PP) time period to prevent unintended pregnancy. However, risk of uterine perforation and IUD expulsion may vary by PP timing of IUD insertion
OBJECTIVES: To assess the risk of uterine perforation and IUD expulsion according to PP timing of IUD insertion, using a finer categorization of PP timing than has been previously attempted.
METHODS: Women receiving IUDs were identified at each of four sites (three Kaiser Permanente sites—Northern California, Southern California, and Washington—and Regenstrief Institute, Indiana) between January 2001 and April 2018 (start date variable by site) and assessed for outcomes through June 2018. PP timing and patient/procedure characteristics were assessed at the time of insertion. PP timing was dichotomized in two ways: at 14 weeks and separately at 36 weeks. Groups were further stratified into four and five-category PP exposure groups. Risk of uterine perforation and IUD expulsion was assessed via Cox models, including confounder adjustment by propensity score overlap weighting.
RESULTS: Among 326,658 women with IUD insertions, there were 1,008 uterine perforations and 8,943 IUD expulsions in 641,427 woman-years. After adjustment, the hazard ratio (HR) for uterine perforation with insertion ≤ 14 weeks PP was 3.44 (95% confidence interval [CI], 2.70-4.40) compared with insertion > 14 weeks PP, and HR ≤ 36 weeks PP was 4.36 (95% CI, 3.45-5.51) compared with insertion > 36 weeks PP. The adjusted HRs for IUD expulsion were 0.88 (CI, 0.83-0.94) and 0.95 (CI, 0.89-1.00) for the dichotomous 14- and 36-week PP comparisons, respectively. Consideration of heterogeneity in the dichotomous categories led to a separate, prespecfied four-category PP exposure variable. The adjusted HRs for uterine perforation decreased from ≤ 6 weeks, 6 to ≤ 14 weeks, and through 14 to ≤ 52 weeks compared with > 52 weeks (including no birth), respectively. The adjusted HRs for IUD expulsion were lowest at 6 to ≤ 14 weeks. Further concern of underlying heterogeneity in outcome risk led to splitting the earliest PP time period. With this split, lower rates of perforation and higher rates of expulsion were seen in IUD insertion 0-3 days PP compared with insertion 4 days to ≤ 6 weeks PP.
CONCLUSIONS: Underlying heterogeneity across the PP time period could lead to differences in risk interpretation based on the categorization of exposure. Clinical considerations of interpretation should be used alongside data-driven analyses to determine categorization of exposures.
