Revised Cardiac Risk Index (RCRI) for preoperative risk

Clinical background

The Revised Cardiac Risk Index (RCRI) was constructed to separate, before elective non-cardiac surgery, patients whose cardiac risk is so low that further preoperative investigation adds nothing from those whose risk is increased and in whom further measures may be warranted. The tool replaced earlier, more complex indices and quickly came to dominate because it rests on six variables all available from the history and basic laboratory tests, without requiring imaging or stress testing.

The decision the tool serves is in practice twofold: whether the patient can be operated on without specific cardiac investigation, and whether perioperative management needs to be intensified with respect to monitoring, beta blockade or other optimisation. The RCRI does not answer which specific measure should be taken, but it provides a structure for when the question should be asked.

Calculating the Revised Cardiac Risk Index

The RCRI is the sum of six binary risk markers, each scoring 0 or 1:

$$\text{RCRI} = \text{High-risk surgery} + \text{Ischaemic heart disease} + \text{Heart failure} + \text{Cerebrovascular disease} + \text{Insulin-treated diabetes} + \text{Creatinine} > 2{.}0,\text{mg/dL}$$

The variables are defined as:

• High-risk surgery: intraperitoneal, intrathoracic or suprainguinal vascular surgery
• A history of ischaemic heart disease: previous myocardial infarction, exertional angina, previous PCI or coronary artery surgery
• A history of heart failure: previous decompensation or known left ventricular dysfunction
• A history of cerebrovascular disease: stroke or TIA
• Insulin-treated diabetes mellitus
• A preoperative serum creatinine above 2.0 mg/dL

The score translates into four risk classes: 0 points = class I, 1 point = class II, 2 points = class III, 3 or more points = class IV.

The derivation study was carried out at Brigham and Women's Hospital, an American tertiary-care university clinic [1]. In total 4,315 patients aged ≥50 years undergoing elective major non-cardiac surgery were included. The cohort was divided into a derivation group of 2,893 patients and a prospective validation group of 1,422 patients. The primary outcome was a major cardiac complication during the hospital stay, defined as myocardial infarction, heart failure with pulmonary oedema, ventricular tachycardia or cardiac arrest, and cardiac death. In the derivation cohort there were 56 major cardiac complications (2%). The six variables were identified by multivariable logistic regression and formed the final model.

The complication rates in the validation cohort were 0.4%, 0.9%, 7% and 11% for classes I to IV, and these are the figures the calculator is based on. In the derivation cohort the corresponding figures were 0.5%, 1.3%, 4% and 9% [1].

Interpretation in practice

The main value of the RCRI lies in its ability to identify low-risk patients. A patient in class I has an absolute risk below 1% and normally needs no specific cardiac investigation beyond a basic preoperative assessment. This is what the instrument does best, and it is the situation in which the score has the greatest clinical value.

For classes III and IV the score is not sufficient as the sole basis for a decision. Here it should rather be seen as a signal to proceed with additional assessment. Adding preoperative biomarkers, particularly NT-proBNP and troponin, has been shown to improve discrimination considerably (see Validation and performance).

Validation and performance

A Cochrane systematic review from 2021 identified 107 articles in which the RCRI had been validated externally, compared with other models or supplemented with biomarkers [2]. The review found that no other prediction model performed better than the RCRI for predicting major adverse cardiac events (MACE) as a composite outcome. The RCRI was, however, outperformed by ACS-NSQIP-MICA for predicting myocardial infarction and cardiac arrest separately (median delta c-statistic 0.11) and by ACS-NSQIP-SRS for predicting all-cause mortality (median delta c-statistic 0.15) [2]. These differences therefore concern specific components, not the composite outcome for which the RCRI was originally designed.

Discrimination improved when biomarkers were added to the RCRI. Adding NT-proBNP gave a median delta c-statistic of 0.08, troponin 0.14 and the combination of both 0.12 [2]. BNP and NT-proBNP alone, without the RCRI, showed in some studies even higher discrimination than the RCRI for MACE (median delta c-statistic 0.15 and 0.12 respectively in favour of the biomarker) [2]. Calibration was, however, rarely reported, which makes it difficult to judge whether the absolute risk percentages hold in new populations.

In an external validation in 870 patients aged ≥45 years undergoing non-cardiac vascular surgery, with systematic perioperative measurement of high-sensitivity troponin T, the c-statistic for the RCRI was 0.60 (95% confidence interval 0.54 to 0.65) [3]. In this population, in which the complication rate was 8.7%, the RCRI systematically underestimated the predicted risk (P<0.001 on calibration belts) [3]. The study used modern diagnosis of myocardial infarction according to the Third Universal Definition with troponin measurement, which captures more complications than the methods in use when the RCRI was derived in 1999. This may partly explain why the model underestimates risk in modern cohorts.

A Swedish multicentre study at seven hospitals included 1,291 patients aged ≥50 years undergoing elective major non-cardiac surgery [4]. The study showed that a perioperative rise in high-sensitivity troponin T of ≥14 ng/L above baseline identified acute perioperative myocardial injury and gave a net benefit over the RCRI for predicting death and major adverse cardiovascular and cerebrovascular events within 30 days [4]. The study underlines that the RCRI alone does not capture the proportion of patients who develop myocardial injury without having a raised preoperative score.

Limitations

The RCRI was derived and validated in patients undergoing elective major non-cardiac surgery at a single American university clinic. It does not apply to emergency surgery, in which patients are at high risk whatever the score, and it does not apply to minor procedures under local anaesthesia or to day-case procedures.

Some specific pitfalls:

• Vascular surgery: the RCRI performs less well in vascular surgical cohorts than in the mixed original population [3]. Vascular patients have a higher baseline risk and the model tends to underestimate it.
• The renal variable: the threshold of a creatinine >2.0 mg/dL is a fixed cut-off that takes no account of age, sex or body mass. The estimated glomerular filtration rate (eGFR) is today considered a more accurate measure, and a VISION study planned to update the renal component to an eGFR calculated with the CKD-EPI equation [5]. The calculator nonetheless still uses the original creatinine threshold.
• Modern diagnosis of myocardial infarction: when the RCRI was developed in 1999, perioperative myocardial infarction was defined by CK-MB and the ECG. With high-sensitivity troponin assays and the Third Universal Definition, more and smaller complications are identified, so that the risks predicted by the RCRI systematically underestimate in modern cohorts [3].
• A static assessment: the RCRI measures risk factors at one point in time. A patient whose score is low today may be at increased risk at operation if their condition changes, for example with new heart failure or worsening renal function.
• No biomarkers: the RCRI includes neither natriuretic peptides nor troponin, even though these have been shown to add substantial prognostic information beyond the score [2, 4].

Comparison with perioperative troponin surveillance

The Swedish multicentre study by Chew et al. [4] showed that perioperative troponin surveillance identified patients with myocardial injury that the RCRI missed, and that the biomarker gave a net benefit over the score. This argues that Swedish practice in high-risk surgery, particularly for patients in RCRI classes III and IV, should include perioperative troponin measurement as a complement to the preoperative risk score. The 2022 ESC guidelines generally recommend considering biomarkers in high-risk patients, but the guideline's detailed position on the RCRI could not be established from the available material [6].

References

1. Lee TH et al. Derivation and prospective validation of a simple index for prediction of cardiac risk of major noncardiac surgery. Circulation. 1999. PMID: 10477528
2. Vernooij LM et al. The comparative and added prognostic value of biomarkers to the Revised Cardiac Risk Index for preoperative prediction of major adverse cardiac events and all-cause mortality in patients who undergo noncardiac surgery. Cochrane Database Syst Rev. 2021. PMID: 34931303
3. Fronczek J et al. External validation of the Revised Cardiac Risk Index and National Surgical Quality Improvement Program Myocardial Infarction and Cardiac Arrest calculator in noncardiac vascular surgery. Br J Anaesth. 2019. PMID: 31256916
4. Chew MS et al. Identification of myocardial injury using perioperative troponin surveillance in major noncardiac surgery and net benefit over the Revised Cardiac Risk Index. Br J Anaesth. 2022. PMID: 34857357
5. Roshanov PS et al. External validation of the Revised Cardiac Risk Index and update of its renal variable to predict 30-day risk of major cardiac complications after non-cardiac surgery: rationale and plan for analyses of the VISION study. BMJ Open. 2017. PMID: 28069624
6. Halvorsen S et al. 2022 ESC Guidelines on cardiovascular assessment and management of patients undergoing non-cardiac surgery. Eur Heart J. 2022. PMID: 36017553