Preoperative estimated glomerular filtration rate to predict cardiac events in major noncardiac surgery: a secondary analysis of two large international studies

Figures

• 

  Fig 1. Participant flow diagram. POISE-2, PeriOperative Ischemic Evaluation 2; VISION, Vascular events In noncardiac Surgery patIents cOhort evaluatioN.

• 

  Fig 2. Relationship between age, sex, and preoperative kidney function. eGFR, estimated glomerular filtration rate; POISE-2, PeriOperative Ischemic Evaluation 2; VISION, Vascular events In noncardiac Surgery patIents cOhort evaluatioN.

• 

  Fig 3. Age- and sex-specific relationship between preoperative kidney function and perioperative cardiac events. Results from regression models for the composite of MINS, nonfatal cardiac arrest, or death owing to a cardiac cause occurring within 30 days after nonurgent noncardiac surgery in 35,815 patients in VISION and 9219 patients in POISE-2, pooled across 20 imputed datasets. The top panels show the relative risk across eGFR in female and male patients aged 50, 65 and 80 yr in modified Poisson regression. The reference eGFR was 90 ml min−1 1.73 m−2. eGFR was associated with the composite outcome (P<0.001) in a nonlinear fashion (Pnonlinearity<0.001) and differed by age (Pinteraction<0.001) and modestly by sex (Pinteraction=0.02). In POISE-2, the association was not significantly nonlinear (Pnonlinearity=0.33) and differed by age (Pinteraction=0.008) but not sex (Pinteraction=0.79). The bottom panels show absolute risk predictions for hypothetical patients undergoing hip surgery who had a preoperative haemoglobin of 13.2 g L−1 and no other comorbidities, predicted from logistic regression models. eGFR, estimated glomerular filtration rate; MINS, myocardial injury after noncardiac…

• 

  Fig 4. Predictor importance for the composite of MINS, nonfatal cardiac arrest, or death owing to a cardiac cause after nonurgent noncardiac surgery. Percentage of total predictive information, gain in C-statistic, and gain in mean net benefit contributed by individual predictors from logistic regression models predicting a composite of MINS, nonfatal cardiac arrest, or death owing to a cardiac cause occurring within 30 days after nonurgent noncardiac surgery in patients without missing data in VISION (n=33 171) and POISE-2 (n=8893). The contributions from all surgical variables are summed in a single ‘Type of Surgery’ item for clarity. Point estimates (medians) and 95% confidence intervals are derived from 1000 cluster-based bootstrap samples. ADLs, activities of daily living; AF, atrial fibrillation; CAD, coronary artery disease; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; eGFR, estimated glomerular filtration rate; MINS, myocardial injury after noncardiac surgery; PAD, peripheral arterial disease; POISE-2, PeriOperative Ischemic Evaluation 2; TIA, transient ischaemic attack; VISION, Vascular events In noncardiac Surgery patIents cOhort evaluatioN.