RESUMEN
Objective:To comprehensively evaluate the input and output efficiency of scientific research in hospital by bootstrap data envelopment analysis, to provide useful information for optimization of scientific performance appraisal and hospital discipline development strategy.Methods:37 disciplines were included as decision making unit, input variables include research expenditure and number of research personnel, and output variables include number of science and technology awards, research projects, patent transfer, paper, composition, and academic influence. The bootstrap-DEA method was used to evaluate the efficiency of all DMUs.Results:The main of overall efficiency and pure technical efficiency in basic DEA model are 0.858 and 0.909, but are 0.804 and 0.853 in Bootstrap DEA model, the differences between two models have statistically significant ( P<0.001). There are 11 DMUs with an overall efficiency in 0.9~1.0, 14 DMUs in 0.8~0.9, 7 DMUs in 0.6~0.8, 5 DMUs lower than 0.6. There are 3 DMUs are increasing return to scale, 16 DMUs are constant return to scale, 18 DMUs are decreasing return to scale. No statistically significance was observed between different types of DMUs( P>0.05). There are 4 DMUs reveal input slacks in number of research staffs and 10 DMUs reveal output slacks. Conclusions:The results of Bootstrap-DEA are more accurate than the basic methods for the evaluation of the input-output efficiency of hospital scientific research, so that it is worth popularizing and applying. According to the evaluation results, the hospital management department and disciplines could optimize their discipline development strategies and put forward targeted improvement measurements.