Development and internal validation of a prediction model for intravenous contrast‐associated acute kidney injury

Concerns about contrast-associated acute kidney injury (CA-AKI) may lead to individuals not undergoing necessary contrast-enhanced imaging. Effective risk stratification plays a crucial role in patient management. To develop a prediction model combining the effects of multiple risk factors to predict risk of CA-AKI for individuals undergoing intravenous contrast-enhanced computed tomography (CECT) imaging. In this retrospective cohort study, 906 inpatient and outpatient adults who underwent CECT imaging in two tertiary centers, St. Luke's Medical Center-Quezon City and St. Luke's Medical Center-Global City, located in Manila, Philippines, were included. Multivariate logistic regression was used to identify independent predictors for CA-AKI. A p-value ≤ 0.05 was considered significant. Based on the odds ratio, five parameters were identified and included in the model. The incidence of CA-AKI was 10.38%. A significant association was found between CA-AKI with in-hospital mortality (odds ratio [OR]: 4.23, confidence interval [CI]: 2.59-6.90, p = 0.001) and AKI requiring renal replacement therapy (OR: 3.57, CI: 1.94-6.59, p = 0.001) among admitted patients. Multivariate analysis included reduced estimated glomerular filtration rate (OR = 17.13, p = 0.005), acute heart failure (OR = 9.21, p = 0.006), hypotension (OR = 5.15, p = 0.011), anemia (OR = 4.34, p = 0.004), and use of nephrotoxic antibiotics (OR = 5.82, p = 0.009) in the final prediction model. The prediction model had fair predictive power (area under the curve = 0.737) and good calibration (Hosmer-Lemeshow p-value = 0.418). This prediction model, incorporating clinical and laboratory parameters, provides a practical tool for estimating CA-AKI risk with fair discrimination and good internal calibration. It may support informed decision-making regarding the risks and benefits of intravenous CECT. External validation is recommended.