Predictive value of the Seattle Heart Failure Model in patients undergoing left ventricular assist device placement

Abstract

Background: Left ventricular assist devices (LVADs) are increasingly used in advanced heart failure patients. Despite proven efficacy, optimal timing of LVAD implantation is not well defined.

Methods: Patients receiving an LVAD were prospectively recorded. Laboratory and clinical data were extracted and used to calculate the predicted survival with medical therapy using the Seattle Heart Failure Model (SHFM). This was compared with observed survival, hospital length of stay and timeliness of discharge.

Results: We identified 104 patients. Survival with an LVAD vs SHFM predicted survival was 69% vs 11% at 1 year, corresponding to a hazard ratio of 0.17 (p < 0.0001). SHFM-estimated 1-year survival with medical therapy increased from 4% in 1997 to 2004 to 25% in 2007-2008 (p < 0.0001). Subgroup analysis of higher vs lower risk LVAD patients showed observed 1-year survival of 83% vs 57% (p = 0.04). The lower risk group had a shorter length of stay (46 vs 75 days, p = 0.03), along with higher rates of discharge prior to transplant (88% vs 61%, p = 0.01) and discharge within 60 days of LVAD placement (77% vs 52%, p = 0.03).

Conclusions: The SHFM allows prediction of important features of a patient's hospital course post-operatively, including length of stay and 1-year survival. Given evidence of improved survival and shorter hospital stay in lower risk patients, earlier LVAD placement based on a prediction model like the SHFM should be considered in advanced heart failure patients. The SHFM may have utility as a virtual control arm for single-arm LVAD trials.