Brunelli 2010.

Study characteristics
General information	Objective  Prediction model compared Journal Annals of Thoracic Surgery Country Italy, Spain Study design Prospective cohort study
Participants	Number of included patients 1696 Surgical specialty Thoracic surgery Age Mean 65 years (SD 11.2 years) Male sex 82% High‐risk surgery Not reported Insulin‐dependent diabetes mellitus Not reported History of ischaemic heart disease 11% History of congestive heart failure Not reported History of cerebrovascular events 4% Elevated creatinine 3.3% 0 RCRI factors Not reported 1 RCRI factor Not reported 2 RCRI factors Not reported 3 or more RCRI factors Not reported
Predictors	Predictor 1: Thoracic RCRI, including serum creatinine, cerebrovascular disease, cardiac ischaemia, pneumonectomy  Objective: prediction model compared Category: prediction model  Scale: categorical  Threshold: not applicable Assay/device: not applicable
Outcome	Outcome category MACE Full outcome definition Acute myocardial infarction (diagnosed by electrocardiogram changes and increased serum troponin level), pulmonary oedema (confirmed by consistent findings at chest X‐ray), ventricular fibrillation or primary cardiac arrest, complete heart block and any cardiac‐related death Prediction horizon In hospital or within 30 days
Analysis	Number of outcomes 57 Handling missing data No information on handling missing data Discrimination reported? Yes Calibration reported? No  Reclassification reported? No
PROBAST: Applicability	Domain 1: Participant selection High Justification: very selective group of patients included Domain 2: Predictors Low Justification: not applicable Domain 3: Outcome Low Justification: not applicable High Justification: only a selected group of patients was included. However, predictors and outcomes were clearly defined and comparable as used in the development study.
Notes	—
Item	Authors' judgement	Support for judgement
Domain 1: Participant selection	Yes	Although only patients undergoing thoracic surgery were included, participant selection was appropriate and the RCRI model can be applied in these patients.
Domain 2: Predictors	Yes	Clear (RCRI) predictor definitions were described.
Domain 3: Outcome	Yes	Clearly defined outcome definitions and appropriate adjudication of outcomes.
Domain 4:  Analysis	No	Low number of events; no information on missing data and calibration.
Overall judgement	No	Appropriate patient selection and clearly defined predictors and outcomes. However, the number of outcomes was low and there was no information on missing data and no calibration was reported.