Comparing modelled with self-reported travel time and the used versus the nearest facility : modelling geographic accessibility to family planning outlets in Kenya

Bouanchaud, Paul and Macharia, Peter M and Demise, Eden G and Nakimuli, Doreen (2022) Comparing modelled with self-reported travel time and the used versus the nearest facility : modelling geographic accessibility to family planning outlets in Kenya. BMJ Global Health, 7 (3): e008366. ISSN 2059-7908

Abstract

Introduction Geographic access to family planning (FP) services has been characterised through a variety of proximity metrics. However, there is little evidence on the validity of women’s self-reported compared with modelled travel time to an FP outlet, or between different distance measures. Methods We used data from four urban sites in Kenya. A longitudinal FP outlet census was directly linked with data from cross-sectional FP user surveys. We combined characteristics of outlet visited to obtain FP, transport mode, self-reported travel time and location of households and outlets with data on road networks, elevation, land use and travel barriers within a cost-distance algorithm to compute modelled travel time, route and Euclidean distance between households and outlets. We compared modelled and self-reported travel times, Euclidean and route distances and the use of visited versus nearest facility. Results 931 contraceptive users were directly linked to their FP source. Self-reported travel times were consistently and significantly higher than modelled times, with greater differences for those using vehicles rather than walking. Modelled and Euclidean distances were similar in the four geographies. 20% of women used their nearest FP outlet while 52% went to their nearest outlet when conditional on it offering their most recently used FP method. Conclusion In urban areas with high facility density and good road connectivity, over half of FP users visited their nearest outlet with their chosen method available. In these settings, Euclidean distances were sufficient to characterise geographic proximity; however, reported and modelled travel times differed across all sites.