Linchuang shenzangbing zazhi (Jun 2024)

Construction of a predictive model for acute kidney injury after acute ischemic stroke

  • Jing Xue,
  • Da-wei Chen,
  • Xin Wan

DOI
https://doi.org/10.3969/j.issn.1671-2390.2024.06.006
Journal volume & issue
Vol. 24, no. 6
pp. 475 – 483

Abstract

Read online

Objective To explore the risk factors of acute kidney injury （AKI） through a predictive model in patients with acute ischemic stroke （AIS）. Methods From January 1, 2022 to December 31, 2023, a total of 708 AIS hospitalized patients were retrospectively recruited as training set. The least absolute shrinkage and selection operator regression model was utilized for optimizing the selection of clinical profiles. Multivariate Logistic regression analysis was performed for constructing a Nomogram prediction model. Receiver operating characteristic curve, area under curve （AUC）, calibration curve and decision curve analyses were utilized for evaluating the predictive value of the model. Bootstrap repeated sampling method was employed for internal validation. And 97 AIS inpatients from June 1, 2022 to May 31, 2023 at Stroke Center of Affiliated Sir Run Run Hospital, Nanjing Medical University were utilized for external validation. Results Concurrent acute respiratory failure （OR = 3.104, 95%CI:1.276-7.759, P = 0.013）, elevated levels of blood urea nitrogen （OR = 1.099, 95%CI:1.012-1.215, P = 0.042）, D-dimer （OR = 1.027, 95%CI:1.003-1.554, P = 0.046） and monocyte count （OR = 2.229, 95%CI:1.119-4.941, P = 0.044）, dosing of antibiotics （OR = 3.770, 95%CI:1.608-9.549, P = 0.003） and diuretics （OR = 2.681, 95%CI:1.550-4.709, P＜0.001）, mechanical ventilation （OR = 4.616, 95%CI:2.101-10.283, P＜0.001） and mannitol （OR = 2.552, 95%CI:1.457-4.470, P = 0.001） were independent risk factors for AKI in AIS patients . The prediction model was constructed with the above 8 variables. AUC of the model was 0.877（95%CI:0.844-0.910） could reach 0.875（95%CI:0.844-0.911） and 0.798 （95%CI:0.679-0.917） in internal and external validations respectively. Calibration curve showed that the model was well-calibrated and decision curve analysis indicated that the model had some clinical practicability. Conclusion This study has constructed a novel risk prediction model for AKI after AIS. It is both convenient and effective.

Keywords