不同数据包络分析模型在公立医院技术效率评价中的应用

Application of different data envelopment analysis models in assessing hospital technical efficiency

目的比较不同数据包络分析（dataenvelopmentanalysis，DEA）模型评价四川省县级及以上公立医院技术效率的结果，分析全要素生产率变化及原因，为医院技术效率和全要素生产率改善提供建议。方法采用DEA的CCR模型、BCC模型、超效率模型和非任意控制变量模型评价四川省县级及以上医院的技术效率；用Malmquist指数分析研究全要素生产率和总技术效率的变化及原因。结果传统DEA模型测得2003—2007年平均总技术效率为0．7630，平均纯技术效率为0．8042，平均每年有23．55％的医院为纯技术有效；二级医院和县级医院纯技术效率低于其他级别医院。平均每年有59．04％的医院全要素生产率提高，总技术效率变化和技术水平变化对全要素生产率的影响差别不大。超效率模型实现了对总技术有效医院的排序。非任意控制变量模型和传统DEA模型所得不同类别、不同级别医院纯技术效率比较结果有差异。结论2003～2007年四川省县级及以上公立医院达到技术有效的医院比例较低，且5年内变化平缓，有很大改善空间。尤其县级医院应引起重视，通过合理配置资源，加强技术培训，提升效率。公立医院应实施效率与技术并重的医院发展模式，最大程度提升全要素生产率。通过超效率模型，达到有效的医院管理者可了解医院所处相对位置。采用非任意控制变量模型对有不可控变量的公立医院进行效率评价，所得结果更为客观、合理。

Objective To compare the results of different Data Envelopment Analysis（DEA） models in evaluating the techni- cal efficiency at and above county level public hospital in Sichuan province, and analyze the changes of total factor productivity （TFP） and reasons for the change in order to provide suggestions for improving technical efficiency and TFP. Methods Models of CCR, BCC and super efficiency were used to evaluate technical efficiency, and Malmquist Index model to analyze the change of TFP. Nondiscre- tionary model was used to discuss the impact of government input on the efficiency of hospitals. Results Between 2003 and 2007,the average annual total technical efficiency was 0. 7630, and the pure technical efficiency was 0. 8042 assessed by traditional DEA model. Only 23.55 percent of hospitals were total technical efficient. Pure technical efficiency of level II and county level hospitals was less than that of other levels hospitals. There was an average of 59.04% of hospitals＇ TFP increased each year. The change of total technical effi- ciency and technical level had on significant effect on the TFP. Super efficiency model could rank the total technical efficient hospitals. After dividing hospitals into different groups according to their types and levels,it was found that there were differences in ranking re- suits between nondiseretionary model and traditional DEA models. Conclusions During the period of 2003 to 2007, the proportion of technical efficient hospitals at and above the county level was low, although there is a smooth changing, suggesting that there is a great space for improvement. County hospitals should pay more attention to improve the efficiency by allocating the resources rationally and strengthening the technical training. The public hospital should implement the development model in which the attention will be paid e- qually to the efficiency and technology to promote the TFP. The efficient hospitals＇ supervisor can understand the relative position by super efficiency model. The results of using nondiscretionary model in evaluating public hospitals＇ technical efficiency are more objec- tive and reasonable.