{ "background": "Weak governance structures in community health centres can impede the adoption of essential diagnostic protocols, undermining health system performance. In Senegal, decentralised management of these centres presents a critical but understudied governance challenge affecting clinical practice.", "purpose and objectives": "This study aimed to quantify the influence of specific health systems governance factors—financial autonomy, supervisory regularity, and supply chain integrity—on the adoption rate of WHO-recommended diagnostic algorithms for common febrile illnesses.", "methodology": "A cross-sectional, multilevel regression analysis was conducted using audit data from a stratified random sample of community health centres. The adoption rate was modelled as a function of centre-level governance variables and individual health worker characteristics. The statistical model was specified as: $\\text{logit}(p{ij}) = \\beta0 + \\beta1 X{1ij} + \\beta2 Z{2j} + uj$, where $p{ij}$ is the probability of algorithm adherence for health worker $i$ in centre $j$, $X$ and $Z$ are individual and centre-level covariates, and $u_j$ are centre random effects. Robust standard errors were used for inference.", "findings": "Centres with high supply chain integrity demonstrated a 34% higher mean adoption rate (95% CI: 22% to 46%) compared to those with low integrity, after controlling for health worker training. Financial autonomy showed a non-significant association.", "conclusion": "Governance mechanisms directly affecting service delivery, particularly supply chain reliability, are stronger drivers of diagnostic optimisation than broader financial decentralisation in this context.", "recommendations": "Policy should prioritise strengthening logistical systems and embedding algorithm adherence within routine supervisory checklists at the community health centre level.", "key words": "health systems governance, diagnostic algorithms, multilevel modelling, community health workers, Senegal", "contribution statement": "This paper provides a novel methodological application of multilevel regression to isolate the effect of distinct