Assessing the community-level impact of a decade of user fee policy shifts on health facility deliveries in Kenya, 2003-2014

BACKGROUND: The long-term impact of user fee removal policies on health service utilization in low- and middle-income countries may vary depending on the context in which they are implemented, including whether there are policy actions to support implementation. We examined the community-level impact of a decade of user fee policy shifts on health facility delivery among poorest and rural women and compared the changes with those among the richest and urban women in Kenya using data from three rounds of nationally representative surveys. METHODS: Data are from births occurring in the 5 years preceding the survey to women aged 15-49 years who were interviewed in the 2003, 2008-2009 and 2014 Kenya Demographic and Health Surveys. A total of 5949, 6079 and 20,964 births were reported in respective surveys. We conducted interrupted time series analysis predicting changes in quarterly proportions of births occurring in public and private health facilities as well as at home before and after the 2004, 2007 and 2013 user fee policy shifts in Kenya. RESULTS: There were no statistically significant immediate changes in the proportion of births occurring in public facilities following the 2004, 2007 and 2013 user fee policy shifts among poor or rural women. There was, however, a statistically significant increase in home deliveries among all women and among those from the poorest households immediately following the 2004 policy. There was also a statistically significant increase in public facility deliveries among women from the two top quintiles, which was accompanied by a statistically decline in home deliveries immediately after the 2007 policy shift. Differences in trends in public facility deliveries between pre- and post-policy periods were not statistically significant for all sub-groups of women, indicating that even among the sub-group that experienced significant immediate increase after the 2007 policy shift, this pattern was not sustained over time. CONCLUSION: The findings of this paper provide empirical evidence that poorly implemented user fee removal policies benefit more well-off than poor women and in cases where there are significant immediate effects on uptake of facility delivery, this trend is not sustained over time. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (10.1186/s12939-018-0774-4) contains supplementary material, which is available to authorized users.