Abstract:

 Objective　To investigate the quality and key influencing factors of doctoral dissertations in clinical medicine, so as to provide empirical evidence for improving the training quality assurance system.Methods　A retrospective analysis was conducted on 282 double-blind peer review reports from 94 clinical neurosurgery doctoral candidates from the graduating classes of 2021 to 2025. Data on demographic characteristics, key training stages (proposal defense, mid-term inspection), and research topic sources were collected. Spearman correlation analysis, univariate analysis, and multiple linear regression analysis were used to screen independent factors influencing the overall score.Results　The median overall score of the doctoral dissertations was 85.00, indicating a medium-to-high overall quality. Among the sub-evaluation indicators, "innovation of dissertation achievements" scored the highest and showed the strongest correlation with the overall score (r=0.859, P<0.001). Multiple linear regression analysis indicated that an "excellent" result in the mid-term inspection was an independent positive influencing factor for the overall score (β=3.318, P=0.021). Compared with national-level projects, dissertation topics derived from "other projects" were independent negative influencing factors for the overall score (β=-2.718, P=0.025). Variables such as year of admission, study mode, and whether graduates received honors had no statistically significant effect on the total score (all P>0.05).Conclusion　The overall quality of clinical medicine doctoral dissertations is satisfactory, with innovation being the core evaluation element. Mid-term inspection results and topic sources are independent predictors of dissertation quality. It is recommended that training institutions strengthen the screening and incentive functions of the mid-term assessment, and optimize scientific resource allocation and process guidance for topics derived from non-national projects, so as to systematically improve the quality of doctoral dissertations.