预测生殖内分泌与不孕症专科培训申请者成功匹配的独立预测因素
该回顾性队列研究分析了2019至2023年(不含2020年)发送至宾夕法尼亚大学的286份生殖内分泌与不孕症(REI)专科培训申请,占国家住院医师匹配项目(NRMP)REI申请者近全部。199份(69.6%)成功匹配。多变量模型显示,住院医师培训项目附属REI专科培训、住院医师排名、USMLE成绩、至少一篇第一作者论文、进行中项目及壁报展示是匹配成功的独立预测因素,就读国际医学院为负向预测因素。模型曲线下面积0.883,正确分类超过75%的申请者匹配结果。
要点
- 住院医师培训项目附属REI专科培训、住院医师排名和研究产出是匹配成功的强预测因素。
- 就读国际医学院与匹配成功呈负相关(aOR 0.32)。
- 预测模型曲线下面积0.883,灵敏度88.5%,特异度65.8%。
核心结论
- 住院医师培训项目附属REI专科培训、住院医师声誉和研究产出强烈预测匹配成功。
- 性别、种族和民族不是主要预测因素,但某些种族和民族群体的代表性不足限制了检测潜在差异的统计效能。
- 预测模型正确分类了超过75%的申请者匹配结果。
研究设计与样本
- 回顾性队列研究,纳入2019至2023年(不含2020年)发送至宾夕法尼亚大学的286份REI专科培训申请,占NRMP REI申请者99.0%。
- 暴露因素包括人口学特征、教育背景、培训经历和学术成就。
- 主要结局为匹配结果,通过在线检索和与项目管理人员沟通确认。
- 采用单变量分析筛选与匹配相关的变量,再纳入多变量模型识别独立预测因素;使用自举法评估模型区分度和校准度。
主要发现
- 286份申请中199份(69.6%)成功匹配。
- 模型曲线下面积0.883,灵敏度88.5%,特异度65.8%。
临床怎么用
- 作者认为这些发现可能帮助申请者优化申请材料并估计成功匹配进入REI专科培训的机会。
- 作者认为这些发现可能协助项目批判性审查其专科培训匹配的选拔标准。
局限性
- 某些种族和民族群体的代表性不足限制了检测潜在差异的统计效能。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
OBJECTIVE: To identify independent predictors of a successful match to reproductive endocrinology and infertility (REI) fellowships, and to develop and internally validate a prediction model for REI match results. DESIGN: Retrospective cohort study. SUBJECTS: Reproductive endocrinology and infertility fellowship applications sent to the University of Pennsylvania from 2019 to 2023 (excluding 2020), which represented nearly all REI applicants nationally according to National Resident Matching Program data. EXPOSURE: Demographics, education, training, and academic achievements. MAIN OUTCOME MEASURE(S): Match result, confirmed through online search and communication with program administrators. Univariate analyses identified variables associated with match, which were then included in multivariable models to identify independent predictors. Bootstrapping was used to assess model discrimination and calibration. The final model was integrated into a web-based tool. RESULT(S): Of 286 applications (99.0% of REI applications to the National Resident Matching Program), 199 (69.6%) resulted in a successful match. In univariate analyses, variables associated with match were younger age, attendance at an allopathic US medical school, United States Medical Licensing Examination (USMLE) and Council on Resident Education in Obstetrics and Gynecology scores, residency rank, residency affiliation with a fellowship, research experiences, first-author publications, abstracts/articles in progress, and poster presentations. In the adjusted model, independent predictors of match included residency affiliation with an REI fellowship (adjusted odds ratio [aOR], 5.43; 2.02-14.64), residency rank (aOR, 1.77; 1.25-2.50), USMLE score (aOR, 1.05; 1.02-1.08), at least one first-author publication (aOR, 2.32; 1.08-4.96), projects in progress (aOR, 1.26; 1.02-1.55), and poster presentations (aOR, 1.07; 1.00-1.15). Attendance at an international medical school was a negative predictor (aOR, 0.32; 0.11-0.88). The model achieved an area under the curve of 0.883, with 88.5% sensitivity and 65.8% specificity. A refined model without USMLE scores maintained strong performance (C-statistic, 0.85; 0.81-0.91; calibration slope, 0.91; 0.72-1.24). CONCLUSION(S): Affiliation with an REI fellowship, residency reputation, and research output strongly predicted match success. Gender, race, and ethnicity were not major predictors, yet underrepresentation of certain racial and ethnic groups limited the power to detect potential differences. Our prediction model correctly classified >75% of candidates' match results. These findings may help candidates optimize applications and estimate chances of a successful match into REI fellowship, as well as assist programs in critically reviewing their selection criteria for fellowship match.
引用
- Humphries LA, Applebaum J, Mainigi MA, et al. Predicting a successful match among applicants to reproductive endocrinology and infertility fellowship. Fertil Steril. 2024. PMID: 39187046. doi:10.1016/j.fertnstert.2024.08.337
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。