LSE Research Proposal Example: Health services researcher to systems policy (Score 93)

Health systems in low- and middle-income countries routinely adopt evidence-based policies that, once implemented, produce outcomes markedly different from those observed in trial or pilot conditions. The distance between what a policy prescribes and what a health system delivers is not simply an administrative failure; it reflects structural tensions between the conditions under which evidence is generated and the organisational realities in which policy must operate. This proposal investigates that distance in a bounded, tractable way. The central research question is: To what extent does variation in implementation fidelity explain divergence between expected and observed health service quality outcomes in district-level primary care systems? Two subsidiary questions follow. First, which organisational and governance factors are most consistently associated with lower fidelity in resource-constrained settings? Second, do health workers' reported understanding of policy rationale moderate the relationship between fidelity and service quality, independently of resource availability? These questions are grounded in a practical observation from a workplace policy memo examining a quality-improvement initiative across district health facilities: facilities that received identical protocol training showed substantial variation in measured service quality at six-month review. That variation was not fully explained by staffing ratios or supply-chain disruption alone, which suggests that implementation processes themselves carry explanatory weight. The research problem is therefore not whether evidence-based policy works in principle, but under what organisational conditions its translation into practice holds or breaks down. Two bodies of scholarship bear directly on this question. The first is implementation science, which has developed frameworks — most prominently the Consolidated Framework for Implementation Research and the Normalisation Process Theory — to map the mechanisms by which interventions become embedded in practice. This literature has generated substantial evidence in high-income clinical settings, particularly in the United States and the United Kingdom, but its application to district-level primary care in sub-Saharan Africa and South Asia remains uneven. Studies that do engage with these settings tend to focus on single-disease programmes such as HIV treatment or maternal health, and relatively few examine cross-cutting service quality metrics that would allow comparison across programme areas. The second relevant body of work is health policy analysis, which has examined how policy design, bureaucratic capacity, and political economy shape what reaches the front line. Scholars working in this tradition — including those associated with the health systems strengthening literature — have documented the importance of governance structures and accountability mechanisms. However, this literature often treats implementation as a dependent variable explained by political or fiscal factors, rather than examining the internal organisational processes through which fidelity is maintained or eroded at facility level. The gap between these two literatures is methodological as much as substantive. Implementation science tends to use qualitative or mixed designs focused on single interventions; health policy analysis tends to use cross-national or cross-district quantitative comparisons that aggregate away facility-level variation. Neither approach, taken alone, can answer the question of how facility-level organisational processes mediate the relationship between policy design and service quality outcomes. A study that combines structured facility-level data on implementation processes with quantitative service quality indicators, and that treats health worker cognition as a potential moderator, would occupy a position that neither literature currently fills. The proposed design is a cross-sectional comparative study of primary care facilities within a single district health system, using a combination of structured observation, health worker survey, and routine service quality data. Limiting the study to one district controls for macro-level governance and fiscal variation while preserving sufficient within-unit heterogeneity to test the main relationships. The sampling frame would comprise all facilities within the district meeting a minimum patient-volume threshold, estimated at between thirty and fifty facilities depending on the district selected. For each facility, implementation fidelity would be operationalised using a structured observation checklist adapted from existing validated tools in the implementation science literature, covering protocol adherence, supervision practices, and documentation completeness. Health worker surveys would measure self-reported understanding of policy rationale using a short instrument adapted from the Normalisation Process Theory toolkit, with items translated and back-translated following standard procedures. Service quality outcomes would be drawn from routine health management information system data, specifically composite scores derived from antenatal care, outpatient consultation, and referral completion indicators. These are standard outputs in most district health information systems and do not require primary data collection beyond what is already recorded. The analysis would proceed in two stages: first, a multivariate regression examining the association between fidelity scores and service quality outcomes, controlling for facility size, staffing, and supply-chain variables; second, a moderation analysis testing whether health worker understanding of policy rationale conditions the fidelity-quality relationship. This design is appropriate because it matches the level of analysis — the facility — to the theoretical claim, which is about organisational processes rather than individual behaviour or national policy architecture. A purely qualitative design would not generate the variation needed to test moderation; a purely quantitative design using national-level data would aggregate away the facility-level processes that are the object of study. The principal feasibility constraint is access to routine health management information system data and permission to conduct structured observation in facilities. This would require formal approval from the relevant national or sub-national health authority, which typically involves a research ethics review and a data-sharing agreement. These processes can take three to six months and represent the main scheduling risk. A contingency plan would involve restricting the study to facilities where a prior quality-improvement programme has already established a data-sharing relationship, reducing the access negotiation burden. Ethics considerations include the anonymisation of facility-level data to prevent identification of individual facilities or workers in published outputs, and the voluntary nature of health worker survey participation. No patient-level data would be collected, which substantially reduces the ethics risk profile. Survey instruments would be reviewed for cultural appropriateness before deployment. A provisional timeline for a one-year MSc research project would allocate the first term to literature consolidation and instrument development, the second term to data collection and preliminary analysis, and the third term to final analysis and write-up. The cross-sectional design is deliberately chosen to fit within this constraint; a longitudinal design tracking fidelity over time would be more powerful but is not feasible within a single academic year. The scope boundary is explicit: this study does not aim to evaluate the effectiveness of any specific health policy, nor does it seek to generalise findings beyond the district studied. Its contribution is to demonstrate a replicable analytical approach for examining implementation fidelity as a mediating variable between policy design and service quality. The LSE Department of Health Policy brings together researchers working on health systems governance, policy analysis, and health economics, with particular strength in applied quantitative methods and comparative health systems research. The MSc Health Policy programme's methods training — covering causal inference, health data analysis, and policy evaluation — directly supports the analytical approach proposed here. The programme's emphasis on connecting policy analysis to implementation evidence aligns with the research question, which sits at the intersection of health systems governance and implementation science. The department's access to health policy datasets and its connections to international health organisations would support the data access negotiations that represent the main feasibility risk. The health policy applied project component of the programme provides a structured opportunity to pilot the observation instrument and refine the survey items before full data collection, which would strengthen the eventual dissertation. This proposal does not name a specific supervisor, as the appropriate match would depend on current faculty availability and research focus. Faculty working on health systems strengthening, implementation research, or health service quality in low- and middle-income settings would represent the most relevant supervisory fit, and the applicant intends to identify and contact appropriate faculty during the application process using publicly available research profiles.