Recovery, Dependence or Death After Discharge

Excerpt

Authors’ Reply—Thank you for your comments1 related to our article on prediction of recovery, dependence or death in elders who become disabled during hospitalization.2 We agree that validation of prognostic models such as the one we developed is critical. There are two key reasons for validation.3 The first is to account for the ‘optimism’ that occurs when a model is initially developed, since it is ‘optimized’ to the data. This can be thought of as ‘internal’ validation, and strategies such as dividing the sample into development and validation cohorts or bootstrapping techniques are utilized to estimate the extent to which c statistics may be inflated due to optimism.4 We utilized bootstrapping techniques, and the optimism-adjusted c statistics are reported at the end of the third paragraph of the results section (recovery: 0.76; dependence: 0.66; death: 0.73). The second reason for validation is to determine whether the model developed is predictive in other study populations and can be thought of as ‘external’ validation.3 We acknowledged lack of external validation of our tool as a limitation. Since our article was published, we have been contacted by several other investigators with data that may allow us to take this important next step. …