Factors influencing variation in implementation outcomes of the redesigned community health fund in the Dodoma region of Tanzania: a mixed-methods study

Introduction Micro-health insurance (MHI) has been identified as a possible interim solution to foster progress towards Universal Health Coverage (UHC) in low- and middle- income countries (LMICs). Still, MHI schemes suffer from chronically low penetration rates, especially in sub-Saharan Africa. Initiatives to promote and sustain enrolment have yielded limited effect, yet little effort has been channelled towards understanding how such initiatives are implemented. We aimed to fill this gap in knowledge by examining heterogeneity in implementation outcomes and their moderating factors within the context of the Redesigned Community Health Fund in the Dodoma region in Tanzania. Methods We adopted a mixed-methods design to examine implementation outcomes, defined as adoption and fidelity of implementation (FOI) as well as their moderating factors. A survey questionnaire collected individual level data and a document review checklist and in-depth interview guide collected district level data. We relied on descriptive statistics, a chi square test and thematic analysis to analyse our data. Results A review of district level data revealed high adoption (78%) and FOI (77%) supported also by qualitative interviews. In contrast, survey participants reported relatively low adoption (55%) and FOI (58%). Heterogeneity in adoption and FOI was observed across the districts and was attributed to organisational weakness or strengths, communication and facilitation strategies, resource availability (fiscal capacity, human resources and materials), reward systems, the number of stakeholders, leadership engagement, and implementer’s skills. At an individual level, heterogeneity in adoption and FOI of scheme components was explained by the survey participant’s level of education, occupation, years of stay in the district and duration of working in the scheme. For example, the adoption of job description was statistically associated with occupation (p = 0.001) and wworking in the scheme for more than 20 months had marginal significant association with FOI (p = 0.04). Conclusion The study demonstrates that assessing the implementation processes helps to detect implementation weaknesses and therefore address such weaknesses as the interventions are implemented or rolled out to other settings. Attention to contextual and individual implementer elements should be paid in advance to adjust implementation strategies and ensure greater adoption and fidelity of implementation.