Assessment and prediction on technical efficiency for medical institutions in China based on network data envelopment analysis and mixed effect model

  Objective  To analyze the efficiency of resource allocation and to predict health resources and the efficiency of health resource input and output for next five years in medical institutions in China for providing references to policy-making by health administrative agencies.

  Methods  We extracted data on health resource allocation and the input and output of health resources of the medical institutions in 31 provinces and municipalities between 2007 to 2016 from China Health Statistics Yearbook. Network data envelopment analysis (DEA) method was used to evaluate the efficiency of health resource allocation at province level and panel data fixed effect model and random effect model were adopted to predict the efficiency in next five years. Principal component analysis was used to evaluate indicators used in the study.

  Results   Based on DEA results, the overall technical efficiency of health resources allocation was good only for health institutions in 3 of 31 province-level regions (Shanghai municipality, Henan and Guangdong province) during the 10-year period; the DEA results also demonstrated higher technical efficiency value for medical service capacity output (named node 2) than that for reception capacity output (named node 1) of the medical institutions in 15 (48.34%) of the 31, with higher technical efficiency value for medical institutions in southeastern province-level regions and the technical efficiency fluctuated greatly during the period in the western province-level regions. The average annual efficiency predicted for next five years for medical institutions in all the 31 province-level regions is 0.95 and higher than that predicted for next ten years; the average annual efficiency predicted for next five years is within a fair level for medical institutions in 13 province-level regions.

  Conclusion  The efficiency of health resource allocation is at a low level for medical institutions in the majority of province-level regions and there is a great regional disparity in the efficiency, with higher efficiency for medical institutions in southeastern province-level regions. The DEA based efficiency prediction could be used in planning health resources allocation.