Background: In 2013, Kenya introduced a free maternity policy in all public healthcare facilities. In 2016, the Ministry of Health shifted responsibility for the program, now called Linda Mama, to the National Hospital Insurance Fund (NHIF) and expanded access beyond public sector. This study aimed to examine the implementation of the Linda Mama program. Methods: We conducted a mixed-methods cross-sectional study at the national level and in 20 purposively sampled facilities across five counties in Kenya. We collected data using in-depth interviews (n = 104), administered patient-exit questionnaires (n = 108), and carried out document reviews. Qualitative data were analysed using a framework approach while quantitative data were analysed descriptively. Results: Linda Mama was designed and resulted in improved accountability and expand benefits. In practice however, beneficiaries did not access some services that were part of the revised benefit package. Second, out of pocket payments were still being incurred by beneficiaries. Health facilities in most counties had lost financial autonomy and had no access to reimbursements from NHIF for services provided; but those with financial autonomy were able to boost facility revenue and enhance service delivery. Further, fund disbursements from NHIF were characterised by delays and unpredictability. Implementation experiences reveal that there was inadequate communication, claim processing challenges and reimbursement rates were deemed insufficient. Conclusions: Our findings show that there are challenges associated with the implementation of the Linda Mama program and highlights the need for process evaluations for programs to track implementation, ensure continuous learning, and provide opportunities for course correcting programs’ implementation.
In 2010, Kenya transitioned from a centralised system of governance to a devolved system comprising a national government and 47 semi‐autonomous counties. 23 National government retained policy development and regulatory functions, management of the national referral health facilities, capacity building, and technical assistance to counties; County governments on the other hand are responsible for service delivery and management of county health facilities. 24 The public sector is organised hierarchically into 4 tiers (6 levels): (I) Community health services (level 1) (II) Primary care provided by dispensaries (level 2) and health centres (level 3) (III) County referral services including first referral sub‐county hospitals (level 4) and second referral county hospitals (level 5) (IV) National/tertiary referral hospitals (level 6). 24 We conducted a mixed‐methods cross‐sectional study using qualitative and quantitative approaches at the national level and in five counties of Kenya. Data were collected between June and August 2019. We selected counties purposively in consultation with the NHIF. Selected counties included two sites with Universal Health Coverage (UHC) initiatives. One was among the country’s four UHC pilot sites implemented by the national government where user fees at public hospitals were removed; securing commodities through the Kenya Medical Supplies Authority was to be ensured coupled with conditional grants from the national government for lost revenue. 25 In the public sector, this UHC pilot program had an overlap with the Linda Mama Program as its focus was on primary health care including immunisation, maternal and child health (MCH), family planning, TB, HIV, sexually transmitted diseases, and improved nutrition of pregnant women until the first five years of a child’s life. 26 The other had a county‐run UHC program in all public facilities in the county and covered inpatient services and outpatient services, but excluded some specialised services as well as family planning, maternal, neonatal and child services. 27 The county‐run UHC program had no overlap with the Linda Mama program. These counties were selected because the NHIF was keen to understand Linda Mama implementation in both public and private facilities in the context where parallel initiatives were in place. Table 1 presents the demographics and health indicators of the study counties. County demographics and health indicators Across the five counties, 20 healthcare facilities were purposively sampled for data collection. These included one public county referral hospital, one public sub‐county hospital, one public health centre, and one faith‐based hospital/health centre in each of the sampled counties. As a process evaluation, 35 , 36 , 37 our research sought to examine how a program was implemented and to generate evidence that could explain observed policy outcomes. We also sought to highlight opportunities for course‐corrections in implementation that may help the program attain its intended outcomes. 35 , 36 , 37 In this study, we assessed the emergence of the Linda Mama program, its implementation fidelity, and implementation experiences of various actors. An assessment of the emergence of the program examined factors that led the previous free maternity policy to morph into the Linda Mama program. Assessing the fidelity, entailed comparing what the intended program design was (‘de jure’ policy) against what was being implemented in practice (‘de facto’ policy). 35 , 37 We examined implementation fidelity across key dimensions of the program namely the program beneficiaries, the benefit package, financing arrangements, and facility contracting for quality. We also explored the experiences of various actors involved in the implementation of the policy. After obtaining written consent from participants, we collected qualitative data using in‐depth interviews and document reviews. We conducted interviews (n = 104) with purposely selected participants drawn from the national and county level who had knowledge on the Linda Mama program either because of their roles and/or experience in implementing the program. Data collection was discontinued upon data saturation. At the national level, participants were drawn from Ministry of Health, NHIF national and regional branch officials, private sector umbrella organisations, and developmental organisations supporting health financing interventions. At the county level, we selected participants from the county health management teams (CHMT) and healthcare facilities. Table 2 outlines the distribution of study participants. Number and type of study participants Abbreviation: NHIF, National Hospital Insurance Fund. We reviewed policy documents including circulars or policy communications from NHIF to the facilities and the implementation manual. We collected quantitative data using structured questionnaires and data abstraction tools. We used a researcher administered structured questionnaire to carry out patient‐exit interviews to determine the level of OOP costs paid by Linda Mama beneficiaries to access maternal services: Participants were consented prior to this. A sample size was determined based on a 95% confidence level, 10% margin of error and assuming 50% reported OOP payment by Linda Mama beneficiaries in the population. 38 , 39 We randomly selected patients seeking maternal care (n = 108): antenatal, delivery, and postnatal care from the selected facilities. Table 3 outlines the demographic characteristics of patient‐exit interview participants. Characteristics of patients that participated in patient exit interviews We also assessed the structural quality of care offered in the selected healthcare facilities by collecting data on the availability of tracer medicines and medical equipment that are essential for MCH. Review of administrative data from the facility records and NHIF claims management system was done to examine funding flows for the Linda Mama program and determine patterns of claims versus reimbursements. Audio recordings were transcribed verbatim in Microsoft Word and those that were in Kiswahili were translated to English. A framework approach was used for the qualitative analysis; it entails familiarisation, identifying a thematic framework, coding, charting, and interpretation of results. 40 Familiarisation of the data, through listening to the audios and reading the transcripts, ensured transcription and translation accuracy. Following verification, coding of transcripts was done based on a thematic framework developed and agreed upon by the investigators that was derived from the conceptual framework. After indexing of the transcripts and ensuring new emerging themes were captured, charting was done. Charting involved summarising the findings of the transcripts based on the identified themes and identifying illustrative quotes. Data analysis was completed by identifying associations between the themes and providing explanations relevant to the objective of the study. Quantitative data was entered in Microsoft Excel then imported to Stata Version 15.0 for data cleaning and analysis. Data cleaning was based on logic checks and frequency distributions. Descriptive analysis was done using frequency distributions, measures of central tendency and dispersion, using means and 95% confidence intervals or medians and interquartile ranges, as appropriate.
N/A