作者: Mario A. cleves , Nena Sanchez , Mayumi Draheim
DOI: 10.1016/S0895-4356(97)00091-7
关键词:
摘要: Abstract The performance and predictive power of the Deyo-Charlson Romano-Charlson comorbidity indices were compared when short-term mortality after hospitalization was outcome interest. These are commonly used to adjust for effect comorbidities using administrative data in comparative studies. In-hospital Medicare claim patients admitted one six medical categories (back pain, stroke, pneumonia, hip replacement, transurethral radical prostatectomy, or lysis peritoneal adhesion), selected analyses. Logistic regression models employed evaluate relative importance explanatory these predicting 30, 90, 180 days admission. contribution each index assessed via likelihood ratio chi-square statistic (G 2 ), area under receiver operator characteristic (ROC) curve assess power. Indices evaluated weights suggested by Charlson et al . empirically derived weights. Both improved base model equally, although both poor with values C ranging from 0.60 0.78. Our results suggest limited applicability approaches examining mortality. A slight increase observed calculated empirical our data.