Patterns of vaginal bleeding are an important factor in the acceptability of contraceptive methods. The analysis of data obtained from daily menstrual diary records is a major methodological problem to which no satisfactory solution exists. This review describes approaches to the analysis of bleeding patterns among contracepting and non-contracepting women and reviews the difficulties involved. The reference period method, introduced to avoid the arbitrary rules and definitions required for an analysis based on the concept of a menstrual cycle, is discussed and its limitations presented. The review draws on reports of meetings convened by the World Health Organization and the University of Exeter Family Planning Unit to discuss issues in the analysis. Previously unpublished methods are summarized and areas of controversy and topics for further research are identified.
PIP: Methods of analyzing vaginal bleeding patterns, both of normally menstruating non-contracepting and contracepting women, are reviewed, and the reference period method is described in greater detail. For women using long-acting hormonal methods, the concept of cycle is no longer appropriate. The reference period method adopts the woman as the unit of analysis, divides her menstrual diary into consecutive periods, and summarizes vaginal bleeding patterns within each period. Data are presented in bar plots showing median length of bleeding, 5th and 95th percentiles, means and standard deviations. Natural, oral contraceptive, and IUD patterns are similar, while 3-monthly injections, vaginal rings and minipill patterns are disrupted patterns with no notion of cyclicity. Dealing with events that overlap reference period boundaries is still a problem, and is still being studied by working groups, in particular at WHO and the University of Exeter. For example, the effects of different definitions have not yet been systematically evaluated, and recommendations have not been tested empirically. The WHO has published a set of data on 4817 women's diaries to stimulate statistical research on this topic.