@article{Tsui_50_45,
author = {Tsui, Amy and Anglewicz, Phil and Sarnak, Dana and Ahmed, Saifuddin and Makumbi, Fredrick and Guiella, Georges and Mosso, Rosine and Gichangi, Peter},
title={{Predictive utility of key family planning indicators on dynamic contraceptive outcomes: Results from longitudinal surveys in Burkina Faso, Kenya, Uganda, and Côte d'Ivoire}},
journal = {Demographic Research},
volume = {50},
number = {45},
pages = {1301--1352},
doi = {10.4054/DemRes.2024.50.45},
year = {2024},
abstract = {Background: Many health and demographic surveys routinely collect information on women’s exposure to family planning (FP) messages, counseling on contraceptive side effects, discussions about FP with providers, contraceptive decision-making autonomy, and the desire for additional children. Several studies have shown significant associations with current contraceptive use status from these cross-sectional data. However, the predictive utility of these indicators on contraceptive use and its dynamics over time is not well-known, primarily due to the lack of longitudinal data. 

Objective: To empirically assess the predictive utility of key family planning indicators on contraceptive use outcomes using longitudinal data from women participating in national annual surveys in four sub-Saharan African countries. 

Methods: This study utilizes nationally representative longitudinal Performance and Monitoring for Action data collected in Kenya, Burkina Faso, Uganda, and Côte d’Ivoire from reproductive-age women to model the predictive influence of indicators measured at baseline on their subsequent contraceptive use (including intentions to use and method type) and contraceptive dynamics (including adoption, discontinuation, and switching) over an approximately three-year period. 

Results: Some indicators measured at baseline (e.g., FP media exposure and fertility preferences) consistently influenced women’s contraceptive adoption and intentions over two to three years of observation, but the predictive utility of most indictors was largely weak for dynamic outcomes such as switching and discontinuation. 

Contribution: Our study highlights a programmatic need to reconsider long-standing FP indicators given the mixed evidence of their influence on contraceptive behavior over time. We recommend the identification of key measures that can predict dynamic contraceptive behaviors established through longitudinal data to inform actionable interventions. },
URL = {https://www.demographic-research.org/volumes/vol50/45/},
eprint = {https://www.demographic-research.org/volumes/vol50/45/50-45.pdf}
}