A contemporary simple risk score for prediction of contrast-associated acute kidney injury after percutaneous coronary intervention: derivation and validation from an observational registry

Abstract

Background: Contrast-associated acute kidney injury can occur after percutaneous coronary intervention (PCI). Prediction of the contrast-associated acute kidney injury risk is important for a tailored prevention and mitigation strategy. We sought to develop a simple risk score to estimate contrast-associated acute kidney injury risk based on a large contemporary PCI cohort.

Methods: Consecutive patients undergoing PCI at a large tertiary care centre between Jan 1, 2012, and Dec 31, 2020, with available creatinine measurements both before and within 48 h after the procedure, were included; only patients on chronic dialysis were excluded. Patients treated between 2012 and 2017 comprised the derivation cohort and those treated between 2018 and 2020 formed the validation cohort. The primary endpoint was contrast-associated acute kidney injury, defined according to the Acute Kidney Injury Network. Independent predictors of contrast-associated acute kidney injury were derived from multivariate logistic regression analysis. Model 1 included only pre-procedural variables, whereas Model 2 also included procedural variables. A weighted integer score based on the effect estimate of each independent variable was used to calculate the final risk score for each patient. The impact of contrast-associated acute kidney injury on 1-year deaths was also evaluated.

Findings: 32 378 PCI procedures were performed and screened for inclusion in the present analysis. After the exclusion of patients without paired creatinine measurements, patients on chronic dialysis, and multiple procedures, 14 616 patients were included in the derivation cohort (mean age 66·2 years, 29·2% female) and 5606 were included in the validation cohort (mean age 67·0 years, 26·4% female). Contrast-associated acute kidney injury occurred in 860 (4·3%) patients. Independent predictors of contrast-associated acute kidney injury included in Model 1 were: clinical presentation, estimated glomerular filtration rate, left ventricular ejection fraction, diabetes, haemoglobin, basal glucose, congestive heart failure, and age. Additional independent predictors in Model 2 were: contrast volume, peri-procedural bleeding, no flow or slow flow post procedure, and complex PCI anatomy. The occurrence of contrast-associated acute kidney injury in the derivation cohort increased gradually from the lowest to the highest of the four risk score groups in both models (2·3% to 34·9% in Model 1, and 2·0% to 38·8% in Model 2). Inclusion of procedural variables in the model only slightly improved the discrimination of the risk score (C-statistic in the derivation cohort: 0·72 for Model 1 and 0·74 for model 2; in the validation cohort: 0·84 for Model 1 and 0·86 for Model 2). The risk of 1-year deaths significantly increased in patients with contrast-associated acute kidney injury (10·2% vs 2·5%; adjusted hazard ratio 1·76, 95% CI 1·31-2·36; p=0·0002), which was mainly due to excess 30-day deaths.

Interpretation: A contemporary simple risk score based on readily available variables from patients undergoing PCI can accurately discriminate the risk of contrast-associated acute kidney injury, the occurrence of which is strongly associated with subsequent death.

Funding: None.