A randomized-controlled pilot study on the effects of naproxen for the treatment of bleeding and spotting among new Copper T 380A IUD users

Background: The most common reason women report discontinued use of the Copper T 380A (TCu380A) IUD is because of bleeding irregularities after method initiation. The objective of this study was to determine whether an over-the-counter NSAID, naproxen sodium, reduces the number of bleeding and spotting days and heavy or prolonged bleeding episodes among new users of the TCu380A IUD compared to placebo. Methods: In this double-blind pilot trial, we randomized 28 new TCu380A IUD users who reported menstrual cycle changes within 4-6 weeks after IUD placement to either naproxen 440 mg or placebo twice daily for 7 days for three consecutive 28-day cycles and one additional 28-day cycle without treatment. Participants completed a daily bleeding and other symptom diary, and monthly questionnaires. Results: Although not statistically signicant, participants in the naproxen arm reported more mean number of spotting-only days during the three treatment cycles compared those in the placebo group (13.5 days (SE 5.1) vs.7.5 days (SE 1.7), respectively). Otherwise, the mean number of bleeding-only days during treatment and post-treatment and spotting-only post-treatment were similar between the groups. During each treatment cycle, fewer participants in the treatment group reported heavy bleeding compared to placebo group (30.7% fewer in cycle 2; 5.4% fewer in cycle 3). More women in the treatment group, however, reported prolonged menstrual bleeding of greater than 7 days for each cycle, although the percentage reporting prolonged bleeding decreased each consecutive treatment and post-treatment cycle (+27.5%, +19.8%, +9.1%, +4.5%, cycles 1-4 respectively). Other symptoms experienced by new TCu380A users did not differ between the study arms. Conclusions: Treatment with naproxen did not signicantly reduce the number of bleeding or spotting days among new TCu380A users. Although not statistically different, proportionally fewer participants in the naproxen group reported moderate-heavy or heavy periods, but more reported prolonged menstrual bleeding compared to placebo. Participants tolerated oral naproxen use well. Trial registration: subtracting three times each participant’s reported baseline monthly days of bleeding from their reported bleeding/spotting, spotting-only, bleeding-only in study months 1 through 3. We compared differences between the treatment groups using independent t-tests, assuming unequal variances.

broids, or endometrial polyps, use of depo-medroxyprogesterone acetate or hormonal IUD within the last three months, vaginal or cesarean birth in the last four weeks, breastfeeding, or prior TuC380A IUD use.

Recruitment
We recruited participants with advertisements and automated weekly reports using Current Procedural Terminology codes and device J codes (J7300) of patients aged 18-49 years who received a TCU380A IUD at one of the UW-a liated health clinics within the prior week. From these reports, research staff called potential participants to inform them about the study, and if interested, reviewed the study eligibility criteria. Potential participants who met the eligibility criteria were scheduled for a study visit within 7 days of their TCu380A IUD insertion.
Once consented, enrolled participants completed several baseline questionnaires that included contact and demographic information, medical history, menstrual and pelvic pain history questions adapted from the World Endometriosis Foundation (WERF-EPHect Questionnaire) [9]. Study data were collected using REDCap (Research Electronic Data Capture) a secure, web-based application designed to support data capture for research studies [10].

Subject Randomization
Research staff contacted enrolled participants 4-6 weeks after their TCu380A IUD insertion to determine eligibility for randomization to a treatment arm. Participants were eligible to be randomized if they reported any change in menstrual bleeding since their TCu380A IUD insertion, were willing to forgo the use of NSAIDs for the remainder of the study other than the assigned study medication, and were able to complete a daily bleeding diary.
The UW Investigational Drug Service (IDS) assigned the treatment arm using a 4-block 1:1 randomization scheme. Kelley-Ross pharmacy in Seattle, WA compounded the study medication and delivered it directly to IDS for packaging. IDS labeled all study medication bottles, regardless of allocated cohort, as 440 mg naproxen-sodium tablets in order to maintain double-blinding.
Randomized study participants received their study drug, either 440 mg naproxen sodium or identical placebo capsule, which was to be taken the rst day of their next menstrual period, twice daily for 7 days during three consecutive cycles. Each cycle was de ned as 28 days [11]. During the fourth cycle, participants did not take any study medication, but continued to complete a daily bleeding diary. We chose 7 consecutive days because this is consistent with current U.S. national contraceptive guidelines for NSAID treatment of unscheduled spotting or light bleeding or heavy or prolonged menstrual bleeding with Cu-IUD use [12].

Bleeding Diary
Participants tracked their daily bleeding, medication use and side effects on paper diaries along with daily text messages (Mir3, San Diego, CA, USA,) or using the "Clue" tracking smartphone application, a free, period-tracking app available through Apple or Google-Play [13].

Monthly Phone Calls
Study staff called participants monthly to complete a questionnaire about study medication intake, IUD removal or expulsion, last menstrual period duration and ow, expected and unexpected vaginal bleeding or spotting in the last month, and any adverse events. Questions about menstruation were adapted from the St. Louis CHOICE project questionnaire [14]. De nition of heavy ow was based on the participants' subjective experience of ow as "light," "moderate," or "heavy" compared to their menstrual ow just prior to their TCu380A insertion.

Final 6-month Study Visit
The nal study visit was scheduled 6 months after each participant's IUD insertion. Participants returned their study medication bottles and any unused study medication, and submitted their completed paper bleeding diaries. Participants completed the same menstrual questionnaires as were used at baseline, along with 6-month satisfaction and TCu380A IUD continuation questionnaires.

Safety Assessment
A meta-analysis found that OTC naproxen sodium is as safe as ibuprofen, acetaminophen, and placebo and that there is no evidence of overall dose-related increase in adverse events [15,16]. Nonetheless, study staff informed participants at the start of the study that they were to report all adverse events or concerns related to the treatment. Study staff queried each participant during monthly calls for side effects that may require modi cation of the intervention. Speci cally, participants were assessed for gastrointestinal, cardiovascular or hepatic effects and allergic reactions. Each participant was advised to contact the research site immediately if any signi cant medical problems or complaints occurred during the study. The PI, a family planning-trained family physician, was available 24 hours a day and able to evaluate all complaints and assure that appropriate health care or referral was provided for all study participants. Adverse events were de ned by federal regulations as any new medical problem or exacerbation of an existing problem experienced by a participant while enrolled in the study, whether or not it was considered treatment related by the investigator.
Sample size: We originally planned to enroll 60 subjects based on other similar studies that evaluated treatment for bleeding among women using contraception [17][18][19]; as reported by Cohen et al. [17], a sample size of 42 provided 80% power to detect a difference of 7 bleeding/spotting days in 30 days by two-sample t-test, allowing for an expected 20% drop-out. We designed the study to include another site, in addition to the UW site enrolled participants. After 6 months of failed active recruitment at the other site, we discontinued that site from the study. Resources did not permit us to contract with an alternative recruitment site, and thus the protocol was revised to enroll 32 participants at the Seattle site only.

Statistical Analysis
We compared baseline characteristics of the women in the naproxen and placebo groups using Chi-squared, Fisher's Exact, and independent-t-tests as appropriate (Table 1). For days with bleeding/spotting outcomes (Table 2), we used independent t-tests to compare the unadjusted mean difference between the naproxen and placebo groups. Bleeding was de ned as any bleeding that requires > 1 panty liner, tampon, or pad in a day. Spotting was de ned as minimal blood loss that requires 1 or no use of any protection including panty liners. Bleeding episodes were de ned as days of bleeding or spotting that were bounded by 2 days of no bleeding or spotting. Bleeding/spotting episodes were classi ed by the following bleeding pattern: prolonged (i.e., length of a single bleeding/spotting episode 14 days or more per 84-day reference period), frequent (more than 5 bleeding/spotting episodes per 84-day reference period), infrequent (i.e., less than 3 bleeding/spotting episodes per 84-day reference period) or amenorrhea (i.e., absence of bleeding/spotting for at least 84 consecutive days). Participants who reported a 14-day episode and more than 5 episodes within a 84-day period were classi ed as "both prolonged and frequent bleeding." Subjects who reported between 3-5 bleeding/spotting episodes were considered as having a "normal" bleeding pattern. We calculated a change in the number of days bleeding/spotting, spotting-only, bleeding-only in study months 1 through 3 and at baseline by subtracting three times each participant's reported baseline monthly days of bleeding from their reported bleeding/spotting, spotting-only, bleeding-only in study months 1 through 3. We compared differences between the treatment groups using independent t-tests, assuming unequal variances.
We calculated the proportion of women who reported heavy bleeding, frequent cramping, extended menstrual bleeding and unexpected bleeding on their monthly questionnaires, and compared the naproxen and placebo groups using chi-square and Fisher exact tests (Table 3). We calculated the number of days in which a woman reported each potential side effect during the treatment months (months 1-3) and compared the resulting means of the naproxen and placebo groups using a two-sided independent t-test (Table 4). For analysis of satisfaction, we calculated median satisfaction score and interquartile range (Table 5). We analyzed participants in their assigned treatment group. Participants who had missing bleeding diary data were excluded from the analysis. Analyses were conducted using SAS version 9.4 (SAS Institute, Inc., Cary, NC). Nominal p-values are reported without adjustment for multiple comparisons.

Results
Between February 2016 and May 2017, 34 women enrolled into the study, of whom 28 were randomized one month after TCu380A insertion. Twenty-ve participants completed all four study cycles; 20 participants submitted data on bleeding. Fig. 1 depicts participant ow according to CONSORT guidelines [8]. There were no signi cant differences of baseline characteristics between the study arms at baseline (Table 1).
Among 20 participants who submitted their bleeding diaries, participants in the naproxen arm reported more mean number of spotting-only days during the three treatment cycles compared those in the placebo group (13.5 days (SE 5.1) vs.7.5 days (SE 1.7), respectively), although this nding was not signi cant (Table 2). When adjusted for baseline bleeding, naproxen participants had 0.9 bleeding-only days/cycle, or 2.7 total more bleeding-only days (95% CI -4.1-9.5 days) during the rst 3 months and only one additional bleeding-only day during the post-treatment observational month, suggesting that naproxen did not contribute to additional days of bleeding compared to placebo.
There were no statistically signi cant differences between the naproxen and placebo group among the participants who responded to the monthly questionnaire about their period being heavier than baseline, lasting more than 7 days, frequent cramping or unexpected bleeding. Although not signi cant, fewer women in the naproxen group reported moderate-heavy or heavy periods during the second and third treatment months (-30.7% month 2, -5.4% month 3) (Table 3). On the other hand, more naproxen users reported menstrual bleeding length ≥ 7 days throughout the study period, but the difference between naproxen and placebo groups narrowed with each consecutive treatment month and into the nal post-treatment observational month (+27.5%, +19.8%, +9.1%, +4.5%, months 1-4 respectively). Although not signi cant, more than 25% fewer participants reported often/always cramping in the naproxen group only during the rst treatment month than the placebo group. Otherwise, complaints of cramping were similar during the second and third treatment months (Table 3). Unexpected bleeding varied throughout the study, but overall, a smaller percentage of naproxen users reported this nding than those who took placebo (Table 3).
Overall, naproxen was well tolerated by the participants. There was no statistically signi cant difference in the percentage of naproxen users and placebo users who reported symptoms (Table 4). Cramping was the most frequently reported side effect by women in both groups. No participants reported any major adverse outcomes during the treatment or post-treatment months. None of the women who completed the nal survey discontinued their TCu380A IUD, except for a placebo group participant whose IUD fell out prior to the second treatment month of the trial. Two women, one in each group were lost to follow up and thus, their continuation status of TCu380A IUD is unknown.
Both groups were similarly very satis ed with the TCu380A IUD, with 3=very satis ed (naproxen median score =3, IQR 2-3; placebo median score=3, IQR 2.25-3) ( Table 5). Only one person, assigned to the naproxen group, was not satis ed with her IUD, due to prolonged menstruation and cramping during the second half of the cycle. The scores indicating the likelihood of continuation of the TCu380A IUD were similar between the groups.

Discussion
Some women who use the TCu380A IUD report more cramping and heavier bleeding compared to baseline during the rst 3 months of use [20]. NSAIDs are the most widely studied medications for blood-loss reduction for Cu-IUD users. Evidence indicates that these medications can reduce menstrual bleeding for a current bleeding episode and may prevent heavy or prolonged menstrual bleeding [21]. Prior to this present pilot study, however, information regarding treatment with an OTC dose-equivalent NSAID medication for menstrual irregularities among new TCu380A IUD users with recent changes in menstruation was lacking. Determining whether OTC naproxen alleviates heavy or prolonged bleeding could be a simple and accessible intervention to improve associated bleeding and cramping of this popular and highly effective contraceptive method.
In this double-blinded pilot RCT, we randomized 28 new TCu380A IUD users who had reported menstrual changes after their rst month of use to determine whether naproxen 440 mg twice daily (880 mg/day) for 7 days during three consecutive menstrual cycles decreased overall number of days of bleeding/spotting, bleeding-only or spotting-only days compared to placebo. Among the 20 women who submitted their bleeding diaries, we found no signi cant differences between reported number bleeding/spotting, bleeding-only or spotting-only days between the groups. Among the larger group of participants, we also found no signi cant differences among the study participants' reporting moderately-heavy/heavy menstruation, prolonged bleeding, cramping or unexpected bleeding. Nonetheless, with each consecutive month of NSAID use, heavy menstruation, prolonged bleeding and unscheduled bleeding appeared to improve. A handful of small studies have evaluated the use of different types of NSAIDs, including naproxen, to treat heavy or prolonged menstruation in the presence of Cu-IUDs, with results favoring reduction in measured menstrual blood loss [22][23][24][25]. Similar to our ndings, two prior studies found no statistically signi cant differences between naproxen and placebo among IUD users regarding participants' self-reports of menstrual bleeding ow and duration [26,27]. Less consistent with other ndings, we found fewer participants in the naproxen group reported often/always cramping only in the rst treatment month compared to placebo [21].
Despite the body of evidence suggesting that NSAIDs treat menstrual blood loss among Cu-IUD users, the variation by which menstrual blood loss is measured makes comparisons with our ndings di cult [28]. Most prior studies used an objective measure of menstrual blood loss, which requires the burdensome task of study participants saving their used sanitary pads and transporting them back to the research team for processing by alkaline hematin technique [7]. Similar to more recent studies, our study utilized standardized methods of collecting and analyzing bleeding/spotting days [10]. Although subjective reporting of heavy menstrual bleeding has been shown to have limited accuracy, it remains important from the standpoint of the woman's perspective. Prior research suggests women who perceive a change to heavier blood loss or greater bleeding frequency with TCu380A IUD use are less likely to be satis ed with their method [20]. What is reassuring, however, is that to our knowledge no women in our study discontinued their TCu380A IUD method, except for a participant who reported it had fallen out. A growing body of evidence suggests that TCu380A IUD users continue their method at similar rates as other long-acting reversible methods, including those reporting heavy menses prior to insertion [6,29,30].
Strengths of our study include baseline similarities between study participants, as all randomized participants were rst time TCu380A IUD users and reported changes in their bleeding pattern within their rst month of use. A limitation includes our small sample size, which did not provide su cient power to detect differences between the groups. We were surprised by our di culty with recruitment. We made considerable effort to identify potential participants at the Seattle site by receiving weekly noti cations of women who had received a TCu380A IUD within our health system. The vast majority of women whom we "cold-called" about the study never responded. Our success was even less successful at our other potential recruitment site, where all women approached declined to participate in the study. While reasons for declining enrollment are not known, we believe that one barrier could have been that new TCu380A users did not want to forgo NSAID use after getting their method. Another potential barrier in Seattle was that women had to transport themselves to another location to pick up the study medication, since pharmacy dispensing laws in Washington State for research purposes did not allow the UW IDS to mail the pills either to enrolled participants' homes or to the clinical sites where baseline questionnaires were completed. Another limitation was missing bleeding diaries, occurring among more than 40% of placebo users and 15% of naproxen users. Because of the considerable missing data, there was not enough bleeding data to impute expected bleeding/spotting days. Thus, our ndings should be interpreted with caution.

Conclusions
The results of this study did not support our hypothesis that OTC (440 mg) naproxen would result in fewer bleeding and spotting days each month among new TCu380A users over a 3-month period. There were too few subjects to provide adequate power to determine differences between bleeding and spotting among naproxen and placebo groups. Based on the ndings and limitations of this study, further research is needed. An ideal study would entail a 3-arm comparison: tranexamic acid, naproxen and placebo. Anti brinolytic agents, which help prevent the disintegration of blood clots, have shown superior improvement of heavy menstrual bleeding over NSAIDs among non-IUD users [31]. Additional outcomes, such as satisfaction, quality of life and method continuation through the rst year of use would lend itself to a more robust study.  3.0 (0.8) *Missing 1 Naproxen case that did not report baseline bleeding days Participant ow according to CONSORT guidelines

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. CONSORT2010ChecklistBMCWH.pdf