Effects of community-based health insurance on modern family planning utilization in Ethiopia

Dear Joseph Obure and Carl Mhina
Thank you for your detailed review and attention dated 30 August, 2019. We have carefully checked the comments and revised the manuscript accordingly. Our responses are given in a point by point manner below.
 
Sincerely, 
Author Team

Methodology:
Reviewer comment (RC) 1: The data from this study is a result of a cross section survey conducted in well-sampled population of Women in areas where CBHI program was implemented using the routinely collected monitoring checklist. The sampling and study recruitment is well described. However, the tool used to collect monitoring data was limited to the CBHI purposes and did not include other variables that may have an effect on the use of FP services beyond the CBHI. Furthermore, the lack of comparison data (intervention vs control or pre vs post intervention) limited the ability to establish any causal relationship between the CBHI and FP use. The statistical analysis is descriptive and is limited to married women excluding unmarried women. Other factors that may have an effect on the FP use were not accounted for in the analysis.
 
Response (R) 1:
Thank you for providing the opportunity to clarify. We agree with your point, but it is important to understand the nature of the data. USAID Transform: Primary health care project implemented different thematic areas jointly (12 different thematic areas together). The primary mechanism for outcome monitoring and evaluation relied on the baseline and end line data conducted by external evaluator. In 2017, following an external evaluation, program implementer and USAID mission in Ethiopia agreed that if TRANSFORM: PHC relied on facility and community level surveys to measure the program outcomes; it would have to wait until mid- or end-of-project evaluations to understand how to optimize the contributions made to reproductive health in Ethiopia. By then, the opportunity to adapt in response to the results would have passed. For this program, meaningful monitoring would require engaging a range of actors with the power to turn this evidence into learning and action in real time. The follow up visit focused on progress outcome, not focused with comparison groups (interventions vs control). The checklist allows the project to gain insight into progress in service delivery (e.g., FP by method, CBHI related, maternal child health, gender related, quality improvement related, malaria related major indicators etc. for decision-making) and change in health outcomes in target areas at the household level (e.g., modern contraceptive prevalence rates, percent of women with children between 0 and 11 months old who had at least four or more antenatal care visits at a health facility, percent of women with children between 0 and 11 months old who had delivered with the assistance of skilled birth attendants and the percent of children between 12 and 23 months old that are fully vaccinated). The checklist is integrated into routine program activities, and contains indicators that can be measured objectively. The checklist has focused program related indictors but no individual characteristic because of the objective of the follow up visit. We have mentioned in the method section about the primary role of the checklist for monitoring and inform its project. In this analysis we extracted only FP and CBHI related indictors. The project has remained three years in implementation and in the future we will compare yearly progress. In the method of analysis because of the nature of the variable, we have limited our analysis in descriptive statistics because we don’t have individual characteristics to do analytical or regression analysis. We have mentioned all this as a limitation of the study. But know I have tried to do simple traditional logistic regression analysis based on your recommendation, but not that comprehensive to see the causal effect of CBHI on family planning utilization.

Results:
Reviewer comment (RC) 2: The results section was brief, reflecting on less variables/factors the authors chose to explore. Few misrepresentations including a mismatch of the contents in the table and the results explanation, the use of FP methods reported to be 47.5% is not shown in Table 1 as outlined. It would also make sense for me to present tables with;
A). include the descriptive statistic table for participants with their demographic, social, economic and reproductive health related information comparing between FP use status (users vs non users),
B). Combine table 2 and 3,
C). Ensure the data presented include both CBHI users and non-users,
D). Account for other variables that may have an effect on the use of FP beyond the CBHI status, and
E). use regression analysis to establish independent factors for the use of FP methods.
It will be clearer to the readers if the authors outline the following terms clearly, women of reproductive age, young women, married women and unmarried women it gets confusing with which group within the participants are we explaining. The fact that married women were first described rather than all participants it deviates the scope (maybe a systematic description of the whole sample then focus on age groups the married vs unmarried, these factors themselves do affect the use of FP regardless of being in an insurance scheme)
 
Response (R) 2: Thank you for pointing this out. Yes, we observe fewer variables addressed in this analysis because of the limited number of variables in the follow up checklist and also have mismatched the content of the table, this mismatch was corrected accordingly.  
A). Yes, now corrected with limited variables. Table 1 explains the characteristics of the respondents.
B). Corrected, table 2 and 3 merged together for reader clarity.
C-E). Table 3 present logistic regression analysis to see the difference of CBHI users vs non-users and their FP utilization. For clarity about the study participants, the reviewer comment is totally accepted and corrections have been made accordingly. The study participants are all reproductive age group women (15-49 years), they might be married or unmarried. Our analysis focused on 3313 reproductive age women; we have mentioned in the method section and also the first paragraph of the results section. The details of the point you highlighted have been addressed accordingly. But the survey check list only focused on merged indicators rather than individual characteristics, as we have mentioned above, so this analysis has a shortage of important factors to analyse and mention as a limitation of the survey checklist.
 
Discussion & Conclusion:
Reviewer comment (RC) 3: The brief discussion on this section reflects on the few variables/factors explored to answer the study question and hence difficult to draw a meaningful conclusion. The study limitations should be taken into account as the conclusion statements are presented. The use of the economic term “life returns” should be clearly elaborated rather than just using it (it masks a lot of exactly what are these returns which could be a stronger talking point, also it risks a questionable thought of returns to scale in demand for health care in insured people). The conclusion is addressing the overview provided by the study, this could be widened as the paper limits itself on its initial role of just monitoring and inform its project (method section). Few grammatical errors are also of note and should be counter checked in the final paragraphs of both the discussion and conclusion sections.

Response (R) 3: Thank you for pointing this out and highlighting the importance of the inclusion of this information. This detail has been added in the revised version of the manuscript.