Top ten risk factors for morbidity and mortality in patients with chronic systolic heart failure and elevated heart rate: the SHIFT risk model

Aims We identified easily obtained baseline characteristics associated with outcomes in patients with chronic heart failure (HF) and elevated heart rate (HR) receiving contemporary guideline-recommended therapy in the SHIFT trial, and used them to develop a prognostic model. Methods We selected the 10 best predictors for each of four outcomes (cardiovascular death or HF hospitalisation; all-cause mortality; cardiovascular mortality; and HF hospitalisation). All variables with p < 0.05 for association were entered into a forward stepwise Cox regression model. Our initial analysis excluded baseline therapies, though randomisation to ivabradine or placebo was forced into the model for the composite endpoint and HF hospitalisation. Results Increased resting HR, low ejection fraction, raised creatinine, New York Heart Association class III/IV, longer duration of HF, history of left bundle branch block, low systolic blood pressure and, for three models, age were strong predictors of all outcomes. Additional predictors were low body mass index, male gender, ischaemic HF, low total cholesterol, no history of hyperlipidaemia or dyslipidaemia and presence of atrial fibrillation/flutter. The c-statistics for the four outcomes ranged from 67.6% to 69.5%. There was no evidence for lack of fit of the models with the exception of all-cause mortality (p = 0.017). Similar results were found including baseline therapies. Conclusion The SHIFT Risk Model includes simple, readily obtainable clinical characteristics to produce important prognostic information in patients with chronic HF, systolic dysfunction, and elevated HR. This may help better calibrate management to individual patient risk.