Impact of data source and time reference of functional status on hospital mortality prediction (BMC Health Services Research)

Background: The study objective was to compare physical function documented in the medical recordswith interview data, and also to evaluate hospital mortality predictions using pre-admissionand on-admission functional status derived from these two data sources. Methods: A prospective cohort study of 1402 subjects aged 65 years and older to the general medicinedepartment of an acute care hospital was conducted. Patient-reported pre-admission and onadmissionfunctional status for impairment in any of the five activities of daily living (ADLs)items (feeding, dressing, grooming, toileting and bathing), transferring and walking, werecompared with those extracted from the medical records. For the purpose of mortalityprediction, pre-admission and on-admission impairment in transferring from the two datasources were included in separate multivariable logistic regression models. We used avariable selection method that combines bootstrap resampling with stepwise backwardelimination. Results: For all ADL categories, the agreement between the data sources was good for pre-admissionfunctional status (k: 0.53-0.75) but poor for on-admission status (k: 0.18-0.31). Onadmissionimpairment was higher in the medical records than at interview for all basic ADLs.Using interview data as the gold standard, although sensitivity for pre- and on-admissionADLs was high (59-93%), specificity for on-admission status was poor (30-37%). The preadmissionmodels using interview data predicted mortality better than the model usingmedical records (c-statistic: 0.83 versus 0.82). Similar results were found for modelsincorporating on-admission functional status (c-statistic: 0.84 versus 0.81). However, thedifferences between the four models were not statistically significant. Conclusion: Medical records can be a good source for pre-admission functional status but on-admissionfunctional impairment was over-reported in the medical records. The discriminatory power ofthe hospital mortality prediction model was significantly improved with the incorporation offunctional status information but it was not significantly affected by their time reference orsource of data.