Comparison of the Charlson Comorbidity Index with the ASA score for predicting 12-month mortality in acute hip fracture

Abstract

BackgroundsThe ASA (American Society of Anaesthesiologists) Score is the current standard for measuring comorbidity in the Australian Hip Fracture registry, however it has never been validated for this purpose. Subsequently, a more appropriate and useful measure should be investigated. This study aimed to compare the ASA and Charlson Comorbidity Index (CCI) scores in predicting 12-month mortality following acute hip fracture. MethodsA retrospective analysis was performed on an audit database of patients who were admitted to an orthogeriatric unit in a public metropolitan hospital from November 2010 to October 2011. 12-month mortality data was linked through a dual search of Queensland Health and mortality registry data. The Charlson comorbidity index was retrospectively applied. Demographics (age, gender, admission residence) and covariates including ASA, CCI, fracture type, fixation type, cognitive impairment on admission, BMI and time to surgery were analysed with logistic regression. ROC curve analysis was performed to assess varying thresholds for each comorbidity system. ResultsA total of 320 patients were available for audit. Unadjusted bivariate analysis demonstrated significant difference between groups regarding increased age (p = 0.004), ASA score (p<0.001), CCI (p = 0.002), age-adjusted CCI (p = 0.002) and admission from a care facility (p<0.001). Logistic regression analysis demonstrated that only ASA (p<0.001) and admission from a care facility (p<0.001, OR=3.36, 95% CI = 1.9 - 6.0) independently predicted 12-month mortality; CCI was not a significant predictor in any models (p = 0.827, age-adjusted CCI: p = 0.864). Using ROC analysis, the ASA (AUC=0.668) outperformed either CCI (AUC=0.607 (CCI), AUC=0.614 (CCI age-adjusted). ConclusionsThe ASA score is independently associated with 12-month mortality; this was not replicated using either version of the CCI. The data does not suggest using the CCI in registry level datasets for the purposes of predicting 12-month mortality.