To the Editor Huynh et al1 described a new model to predict 30-day death or readmission in patients with heart failure. Using data from 430 patients with heart failure, the model predicted 30-day death or readmission (C statistic = 0.82) or readmission alone (C statistic = 0.80) with superior accuracy compared with their test of a published claims model (C statistic = 0.56).1 First, there is an issue of generalizability in a model with a relatively small sample size and from 1 geographic location. For example, in the model, living alone (hazard ratio, 2.05; 95% CI, 1.12-3.76) and being discharged during the winter (hazard ratio, 1.61; 95% CI, 1.00-3.31) were among the strongest predictors of 30-day death or readmission.1 The effect of these variables may vary considerably by location and country, depending on weather, culture, and social services.
Ranasinghe I, Krumholz HM. Predicting 30-Day Readmission or Death in Patients With Heart Failure: Looking Beyond the C Statistic. JAMA Cardiol. 2016;1(8):965. doi:10.1001/jamacardio.2016.3101
Artificial Intelligence Resource Center
Customize your JAMA Network experience by selecting one or more topics from the list below.