{ "background": "Community health centres are critical for primary care delivery in sub-Saharan Africa, yet robust methods for evaluating their long-term performance and forecasting health systems yield are underdeveloped. ", "purpose and objectives": "This study aimed to develop and validate a time-series forecasting model to measure and project health systems yield—defined as the composite output of service coverage and quality—in Tanzanian community health centres. ", "methodology": "We utilised longitudinal administrative data on facility operations, staffing, and service outputs. The core forecasting model is a seasonal autoregressive integrated moving average with exogenous variables (SARIMAX), specified as \ (B) \ (Bˢ) \ᵈ\Ds yt = \ (B) \ (Bˢ) \ + \ Xt, where Xₜ represents covariates including drug stock levels and trained workforce. Model fit was assessed using rolling-origin cross-validation, with forecast uncertainty quantified via 95% prediction intervals. ", "findings": "The model forecasts a significant upward trend in systems yield, with a projected mean increase of 18. 7% (95% PI: 14. 2, 23. 1) over the forecast horizon. The analysis identified drug supply continuity as the most influential exogenous driver, with its coefficient estimated precisely (β = 0. 23, robust SE = 0. 04). ", "conclusion": "The proposed SARIMAX model provides a statistically robust tool for evaluating and projecting health systems performance, demonstrating its utility for strategic resource planning. ", "recommendations": "Health planners should integrate such forecasting models into routine health management information systems to anticipate resource needs and prioritise investments in pharmaceutical supply chains. ", "key words": "health systems strengthening, forecasting, time-series analysis, primary health care, health services research, Tanzania", "contribution statement": "This paper presents a novel application of the SARIMAX framework for forecasting composite health systems yield, providing a replicable methodological tool for long-term performance evaluation in low-res
Rashidi et al. (Thu,) studied this question.