Demographic and Fertility Characteristics of Contraceptive Clusters in Burundi

Abstract

Examining women's reproductive experiences over time reveals a more dynamic view of women's behaviors and needs than current status measures alone. This study uses sequence and cluster analyses, which are designed for identifying patterns and subgroups in longitudinal data. We apply these methods to contraceptive calendar data in Burundi to identify discrete clusters of women based on contraceptive and pregnancy behaviors over the past 5 years. We identify six unique clusters; three characterized by no use of contraception (85 percent of women) and three by use (16 percent). The Quiet Calendar cluster (42 percent) comprise women who neither experience pregnancy nor use contraception. Family Builder 1 (25 percent) and 2 (18 percent) both include women who experience two pregnancies, but differ in unmet need and lifetime experience with contraception. Modern Mother (8 percent), Consistently Covered Mother (6 percent), and Traditional Mother (2 percent) clusters differ by type of contraception used following pregnancy. Factors associated with cluster membership are need for family planning, lifetime experience with contraception, marital status, pregnancy intention, and age. This clustering approach provides a new, more holistic way to measure the diverse needs across unique subpopulations and can inform the development of multifaceted, adaptable strategies to meet women's dynamic fertility needs over the reproductive life course.

Women’s contraceptive profiles in Burundi: Knowledge, attitudes, and interactions with media and health services

“…In contrast, women in the Family Builder 2 cluster are more likely to provide a non-numeric answer about their ideal number of children, suggesting that women who fall into this profile are in a concentrated family building stage but may not believe they can exercise much control over their reproductive lives [ 50 – 56 ]. This contrast comports with other analyses that indicate that women in the Family Builder 1 cluster have more experience with contraception than do women in Family Builder 2, are less likely to experience unmet need, and are more likely to participate in decisions related to contraceptive use [ 41 , 57 ].…”

“…Women with these calendar sequences are grouped in the cluster with the statistically closest medoid sequence. Additional details about the identified clusters are reported elsewhere [ 41 ].…”

“…Our analysis also indicates that the Quiet Calendar is bimodal in terms of its age composition: women who are younger than age 20 and older than age 35 are both more likely to belong to this cluster. Other analysis of the demographic composition of these clusters shows that this cluster is dominated by unmarried women with no children, but also that it includes some older women who are formerly married or have otherwise completed childbearing [ 41 ]. This heterogeneity in terms of women’s characteristics nonetheless results in a consistent, homogenous behavioral calendar sequence.…”

“…We can then use age as a covariate to detect the distribution of age groups that fall into each segment rather than imposing, a priori, any sample division based on age or other criteria. Methodological details of the sequencing and clustering procedures used to identify the 6 clusters are documented elsewhere [ 41 ].…”

“…Demographic and Health Surveys (DHS) contraceptive calendars are a rich source of data on contraceptive behavior and pregnancy experience covering the 5 years preceding the survey [ 17 ]. In contrast to this focus on attitudes in most prior segmentation analyses, a recent study has developed methods to identify discrete groups of women based on longitudinal behavioral data from contraceptive calendars in DHS [ 18 ]. Similarly, a model has been developed to cluster women into segments based on their contraceptive discontinuation behaviors recorded in DHS calendars [ 19 , 20 ].…”