Comprehensive measurement of health system performance at district level in India: Generation of a composite index

Abstract

There have been limited attempts at measurement of health system performance at decentralized levels in low- and middle-income countries. This study was undertaken to develop a composite indicator to measure health system performance at district level in India. Primary data were collected from 377 public health facilities in 21 districts of Haryana state in India using health facility surveys. In addition, 1700 health care providers and 800 clients visiting health facilities were interviewed. Routine health management information system data at district and state level were also analyzed. These data were used for computing 67 input and process indicators covering six health system building blocks. Indicators were normalized and aggregated to generate domain-specific and overall composite health system performance index (HSPI) for each district. Several sensitivity analyses were performed to assess robustness of results. Overall, Panchkula and Ambala districts were found to be the best performing in the state (with HSPI scores of 0.64 and 0.62 out of 1), while Mewat, Faridabad, and Palwal districts had the poorest performance (with HSPI scores of 0.46, 0.49, and 0.48 out of 1). Significant variation in performance was observed for each health system building block. Sensitivity analyses results showed that study findings were robust to variations in methods of aggregation of indicators. Our study provides a framework and methods to measure health system performance at district level in a comprehensive manner. The composite indicator provides a summary snapshot to benchmark performance, while building block and domain scores provide critical information for programmatic action.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 27 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.