Using balancing weights to compare performance across facilities providing family planning services in Kenya

Abstract Assessing the extent to which the quality of family planning (FP) delivery in facilities makes a difference for key outcomes such as service satisfaction or contraceptive discontinuation is of key interest to the family planning field. However, assessment of this relationship is methodologically challenging due to differences in populations served across facilities. Furthermore, data that connect facilities to the populations served are limited. We use novel data from the Performance Monitoring for Action (PMA) project and a new methodological approach to examine the relationship between facility level characteristics and FP outcomes. The PMA data consist of facility surveys and client exit interviews, and capture women’s FP outcomes and include information on characteristics of the individual woman, the facility where the woman obtained her family planning services, and follow-up information on contraceptive use. We use a design-based direct standardization method to balance the distribution of populations served across facilities while controlling for the additional variability induced by the balancing weights. We find significant evidence of variation in FP outcomes across facilities that cannot be accounted for by differences in women characteristics. The type of facility (e.g., dispensary), their size, the proportion of staff present, and whether the facility was public were associated with more positive service satisfaction. A higher ratio of staff to FP visits was predictive of lower contraceptive discontinuation.