Premium
A risk score for predicting 30‐day mortality in heart failure patients undergoing non‐cardiac surgery
Author(s) -
Andersson Charlotte,
Gislason Gunnar H.,
Hlatky Mark A.,
Søndergaard Kathrine Bach,
Pallisgaard Jannik,
Smith J. Gustav,
Vasan Ramachandran S.,
Larson Martin G.,
Jensen Per Føge,
Køber Lars,
TorpPedersen Christian
Publication year - 2014
Publication title -
european journal of heart failure
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 5.149
H-Index - 133
eISSN - 1879-0844
pISSN - 1388-9842
DOI - 10.1002/ejhf.182
Subject(s) - medicine , framingham risk score , heart failure , body mass index , underweight , logistic regression , cardiac surgery , valvular heart disease , overweight , cardiology , surgery , disease
Background Heart failure is an established risk factor for poor outcomes in patients undergoing non‐cardiac surgery, yet risk stratification remains a clinical challenge. We developed an index for 30‐day mortality risk prediction in this particular group. Methods and results All individuals with heart failure undergoing non‐cardiac surgery between October 23 2004 and October 31 2011 were included from Danish administrative registers ( n = 16 827). In total, 1787 (10.6%) died within 30 days. In a simple risk score based on the variables from the revised cardiac risk index, plus age, gender, acute surgery, and body mass index category the following variables predicted mortality (points): male gender (1), age 56–65 years (2), age 66–75 years (4), age 76–85 years (5), or age >85 years (7), being underweight (4), normal weight (3), or overweight (1), undergoing acute surgery (5), undergoing high‐risk procedures (intra‐thoracic, intra‐abdominal, or suprainguinal aortic) (3), having renal disease (1), cerebrovascular disease (1), and use of insulin (1). The c‐statistic was 0.79 and calibration was good. Mortality risk ranged from <2% for a score <5 to >50% for a score ≥20. Internal validation by bootstrapping (1000 re‐samples) provided c‐statistic of 0.79. A more complex risk score based on stepwise logistic regression including 24 variables at P < 0.05 performed only slightly better, c‐statistic = 0.81, but was limited in use by its complexity. Conclusions For patients with heart failure, this simple index can accurately identify those at low risk for perioperative mortality.