Skip to main content

Table 4 The “enhanced” risk model of AKI derived through logistic regression with backward stepwise variable selection using 500 bootstrap samples

From: Prediction of acute kidney injury risk after cardiac surgery: using a hybrid machine learning algorithm

Characteristics AORs, 95% CI β coefficient Risk scores
Surgery type
CABG Ref Ref 0
Single Valve 1.02 (0.80–1.28)  − 0.0809  − 2
Combined CABG/valves 1.52 (1.24–1.87) 0.2509 5
CARE score
1 Ref Ref 0
2 1.94 (1.60–2.34)  − 0.2059  − 4
3 2.83 (2.13–3.75) 0.1737 3
4 5.84 (3.93–8.68) 0.8988 18
Atrial fibrillation
No Ref Ref 0
Yes 1.48 (1.23–1.78) 0.1959 4
NYHA functional class
0 Ref Ref 0
1 0.84 (0.65–1.10)  − 0.2797  − 6
2 1.04 (0.85–1.28)  − 0.0673  − 1
3 1.31 (1.07–2.08) 0.1610 3
4 1.50 (1.08–2.08) 0.2951 6
Hypertension
No Ref Ref 0
Yes 1.50 (1.25–1.79) 0.2019 4
Anemia
No Ref Ref 0
Yes 1.70 (1.46–1.98) 0.2652 5
Preoperative intra-aortic balloon pump therapy
No Ref Ref 0
Yes 3.28 (1.84–5.82) 0.5938 12
Redo sternotomy
No Ref Ref 0
Yes 1.45 (1.13–1.87) 0.1857 4
Smoking status
Never Ref Ref 0
Current 1.33 (1.07–1.66) 0.1185 2
Former 1.24 (1.06–1.46) 0.0497 1
BMI 1.03 (1.01–1.04) 0.0280  
 ≤ 18    − 0.308  − 6
19 ≤ BMI ≤ 24   Ref 0
25 ≤ BMI ≤ 29   0.154 3
 ≥ 30   0.490 10
  1. CARE score, Cardiac Anesthesia Risk Evaluation (CARE) mortality risk score; NYHA class, New York Heart Association Classification; BMI, body mass index