Objective: The Vanderbilt Evidence-based practice Center systematically reviewed evidence about interventions for symptomatic abnormal uterine bleeding (AUB), both irregular and cyclic. We focused on interventions that are suitable for use in primary care practice including medical, behavioral, and complementary and alternative medicine approaches.
Data sources: We searched MEDLINE®, CINAHL®, and Embase for randomized controlled trials (RCTs) published in English from January 1980 to June 2012 in women with symptomatic AUB. We also searched regulatory data and scientific publications for data about harms.
Review methods: Using dual review with a priori criteria, we excluded 1,734 publications because they did not address a Key Question, were not an eligible study design, or did not apply to the primary care treatment of AUB.
Results: Thirty-nine RCTs (6 good quality, 10 fair quality, and 23 poor quality) evaluated 12 distinct interventions. These included 7 studies of the levonorgestrel-releasing intrauterine system (LNG-IUS), 13 of nonsteroidal anti-inflammatory drugs (NSAIDs), 6 of tranexamic acid (TXA), and 5 of combined oral contraceptive pills (COCs). The majority of studies made direct comparisons to other drugs. Ten studies enrolled women with irregular uterine bleeding; the remainder focused on women with heavy cyclic bleeding. Among women with irregular menses, metformin, metformin with exenatide, and a tricyclic oral contraceptive improved menstrual regularity. Among women with heavy, cyclic menstrual bleeding all seven studies of LNG-IUS favored the intrauterine system in comparisons that included NSAIDs, COCs, progestogens and usual care. Reduction in menstrual blood loss ranged from 70 to 87 percent less bleeding than baseline. NSAIDs reduced bleeding in six of six studies when compared with placebo or progestogens. The degree of improvement was highly variable for individual women. TXA was more effective than progestogens and NSAIDs in three of four studies, and COCs provided benefit compared with placebo in two studies. Harms were rare and trials underpowered to assess harms for all interventions. For most interventions, surveillance studies of longer-term risks were not done in comparable populations.
Conclusions: Two interventions for irregular bleeding (metformin, COCs) and four for heavy cyclic bleeding (LNG-IUS, NSAIDs, TXA) have low or moderate strength of evidence for effectiveness, while COCs have high strength of evidence. Several common interventions (including diet and exercise and acupuncture) lack sufficient evidence. Across interventions, data are sparse to evaluate long-term improvements and risk of harms.
Limitations include a predominance of small, short trials lacking standard terminology and diagnostic criteria for identifying and including women with AUB. Tools for collecting outcome data are crude (e.g., collection of sanitary products to measure blood loss) and may contribute to a high rate of attrition. Emphasis on biologic outcomes may neglect the importance of patient-reported outcomes that assess whether symptoms are considered resolved by women themselves.