Objective
The purpose of this study was to evaluate whether oral naproxen or transdermal estradiol decreases bleeding and spotting in women who are initiating the levonorgestrel-releasing intrauterine system.
Study Design
We conducted a randomized controlled trial of naproxen, estradiol, or placebo that was administered over the first 12 weeks of levonorgestrel-releasing intrauterine system use. Participants completed a written bleeding diary. We imputed missing values and performed an intention-to-treat analysis.
Results
There were 129 women who were assigned randomly to naproxen (n = 42 women), estradiol (n = 44 women), or placebo (n = 43 women). The naproxen group was more likely to be in the lowest quartile of bleeding and spotting days compared with placebo (42.9% vs 16.3%; P = .03). In the multivariable analysis, the naproxen group had a 10% reduction in bleeding and spotting days (adjusted relative risk, 0.90; 95% confidence interval, 0.84–0.97) compared with placebo. More frequent bleeding and spotting was observed in the estradiol group (adjusted relative risk, 1.25; 95% confidence interval, 1.17–1.34).
Conclusion
The administration of naproxen resulted in a reduction in bleeding and spotting days compared with placebo.
The levonorgestrel intrauterine system (LNG-IUS) is one of the most effective methods of reversible contraception that is available in the United States. The most commonly reported side-effect of the LNG-IUS is irregular bleeding; 20-33% of women report spotting or bleeding. Although the incidence of heavy bleeding, irregular bleeding, and spotting decreases over the first 6 months of use, irregular bleeding is the most frequently cited reason for discontinuation of the LNG-IUS. Up to 66% of women who request removal of the LNG-IUS will do so within the first 6 months of use. If women are counseled appropriately about the expected bleeding side-effects, they may be less likely to discontinue use because of unscheduled bleeding.
Effective treatments for irregular bleeding that is caused by progestin-only contraceptives have not been well-identified. Previous studies have shown that treatment with mefenamic acid, which is a nonsteroidal antiinflammatory drug (NSAID), decreases the number of bleeding days in users of both the levonorgestrel subdermal implant and depot medroxyprogesterone acetate (DMPA). Several studies have shown a reduction in irregular bleeding in levonorgestrel implant users with ethinyl estradiol administration ; another study failed to see an effect with transdermal estradiol. Likewise, estrogen treatment has not been shown to reduce the frequency of irregular bleeding consistently in DMPA users.
Decreasing the irregular bleeding that is associated with progestin-only contraceptives has the potential to improve satisfaction and increase continuation rates. Our primary objective was to compare bleeding and spotting in new users of the LNG-IUS who had been assigned randomly to oral naproxen, transdermal estradiol, or oral placebo over a 12-week treatment period. Secondary objectives included the comparison of bleeding patterns in the 4-week posttreatment period and participant satisfaction over the study period.
Materials and Methods
We conducted a randomized trial of women who were beginning the LNG-IUS (Mirena; Bayer Healthcare Pharmaceuticals, Montville, NJ) for contraception among women who were enrolled in a contraceptive cohort study that was conducted at a university research clinic. This study, the Contraceptive CHOICE Project, was described previously. Participants were recruited from November 2008 to January 2010, and follow up was conducted through May 2010. Women were eligible if they were English-speaking, willing to avoid additional use of exogenous hormones (eg, oral contraceptives) and NSAIDs for the duration of the study, and willing to comply with the study protocol by adhering to the medication regimen, keeping the bleeding diary, attending follow-up visits, and completing the telephone surveys as scheduled. Exclusion criteria included known or suspected pregnancy; contraindications to estrogen or NSAID use; current use of medications that alter estrogen metabolism; current regular use of an NSAID; current diagnosis of menorrhagia, metrorrhagia, symptomatic uterine fibroid tumors, or endometrial polyps; use of DMPA within the previous 6 months; postpartum in the past 4 weeks; induced or spontaneous abortion in the past 4 weeks; currently breastfeeding; or previous use of the LNG-IUS. We obtained approval from the Human Research Protection Office at Washington University in St. Louis before recruitment, and all women provided written consent before participation in the study. The trial was registered with clinicaltrials.gov ( NCT00789802 ).
We randomly assigned participants to 1 of 3 arms: oral naproxen 500 mg, transdermal estradiol 0.1 mg, or oral placebo. Participants who were assigned randomly to either naproxen or placebo took the study medication twice daily for the first 5 days of each 4-week period, starting on the day after LNG-IUS insertion (day 1). We chose naproxen because it is inexpensive, is widely available in the United States, and can be taken twice daily. We administered the naproxen for the first 5 days of a 4-week period because this regimen had been investigated previously in several other studies of NSAIDS for the treatment of progestin-induced irregular bleeding. Women who were assigned to estradiol also started the patch the day after insertion (day 1) and used it continuously, changing the patch weekly. We chose transdermal estradiol as a non-oral route because changing the patch weekly likely would be preferable to women who would be choosing a long-acting method of contraception. Each group continued with the treatment regimen for the first 12 weeks of LNG-IUS use. A 90-day reference period typically is recommended for the study of bleeding patterns that are associated with hormonal contraception ; however, we used a 12-week or 84-day treatment period because the transdermal estradiol patch was changed weekly, and we administered the treatment for 3 consecutive 4-week blocks. Allocation of participants was equal between arms. A biostatistician who was not involved in the enrollment created the randomization scheme using computer-generated number tables. Participants were recruited, enrolled, and assigned to study group by a research nurse. The treatment arm assignments were contained in sequentially numbered opaque envelopes that were opened by the research nurse after the LNG-IUS insertion. The naproxen and placebo arms were blinded to the participant and the research team; the estradiol arm was open label because no transdermal placebo was available. A research pharmacist who was not involved in the study packaged and dispensed the study medications.
Individuals were provided with a 12-week supply of the study medication and study diaries to record bleeding and spotting and medication compliance. Bleeding was defined as any bleeding that required >1 panty liner, tampon, or pad in a day. Spotting was defined as any bleeding that required ≤1 panty liner, tampon, or pad in a day. Participants returned diaries by mail every 4 weeks for the 16 weeks of the study period. We conducted telephone surveys at 4, 8, and 16 weeks and an in-person follow-up visit at 12 weeks. Surveys asked about bleeding patterns, compliance with study medication, satisfaction with bleeding patterns and the LNG-IUS, and continuation of the LNG-IUS. Participants were reminded to return the bleeding diary at each telephone contact and were compensated for participation.
Statistical analysis
We compared baseline demographic and reproductive characteristics between groups using the chi-square test. Our primary outcome was the total number of bleeding and spotting days during the 12 weeks of the treatment period. Exploratory analysis of the data showed right-skewed distribution of bleeding and spotting days. Right or positive skew means that most of the values lie to the left of the mean and that the distribution is asymmetric rather then normally distributed. When data are skewed, comparison of means is not appropriate. We performed nonparametric testing with the Mann-Whitney test to investigate differences in the distribution of bleeding and spotting days between naproxen and placebo and estradiol and placebo. Because a greater percentage of bleeding and spotting days were concentrated in the first half of the study period, we divided the total bleeding and spotting days into quartiles and compared the quartile distribution of each study arm using Fisher’s exact test. The naproxen and estradiol arms were each compared with the placebo arm and not to each other; therefore, correction for multiple pairwise comparisons was not indicated. We performed univariate and multivariable Poisson regression to measure the relative risk of bleeding and spotting days in the treatment arms and to control for baseline characteristics that were not distributed equally in the study arms. We planned a priori to control for any covariates that were not distributed equally between the study groups. Poisson regression is appropriate in this setting because bleeding and spotting days are count data. Satisfaction and continuation were compared with the use of the χ 2 test. All analyses were performed using STATA software (version 10; StataCorp LP, College Station, TX).
To perform an intention-to-treat analysis of bleeding and spotting days, we imputed missing bleeding and spotting data for the 12-week treatment period using STATA software (version 10.0). The “impute” command estimates the missing values using selected covariates. We found that the following baseline characteristics were associated with the number of reported bleeding and spotting days: average amount of bleeding during menses, duration of menses in days, age of menarche, parity, and amenorrhea in previous 12 months. These covariates were included in the model for the impute command. Similar distributions of data were found between the dataset of actual data that were reported by participants and the combined imputed and reported dataset (data not shown).
We calculated that 99 women (33 in each group) would be required to detect a 25% reduction in the total number of bleeding and spotting days, with an alpha of .05 and 90% power. We estimated a mean number of 39 bleeding and spotting days during the 12-week treatment period, with a standard deviation of 12 days based on a previous study of bleeding and spotting in new LNG-IUS users. Assuming a 10% loss to follow up, we planned to enroll 38 women in each group, for a total of 114 participants. Approximately halfway through enrollment, we observed a greater than expected loss to follow up and added an additional 15 participants, for a final sample size of 129 subjects.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
