Homogram prediction modeling of acute kidney injury after heart valve replacement

Abstract

Objective To construct and validate a nomogram prediction model for acute kidney injury （AKI） after heart valve replacement （HVR）. Methods A prospective cohort study was conducted by selecting 150 patients with heart valve disease after HVR at Third Hospital of Hebei Medical University from January 1, 2020 to March 31, 2022 as a training population. The postoperative incidence of AKI within 7 days was statistically analyzed. According to the occurrence of AKI, they were assigned into two groups of AKI and non-AKI. The relevant clinical data of two groups were compared and the influencing factors of AKI after HVR examined by Logistic regression. And a Nomogram prediction model was constructed on the basis of various influencing factors. Another 150 patients with cardiac valve disease after HVR were selected as a validation population during the same period according to the 1∶1 matching principle. The value and clinical utility of this model in predicting AKI after HVR were evaluated through receiver operating characteristic （ROC） curve and decision curve analysis （DCA） for external validation. Results The postoperative incidence of AKI within 7 days post-HVR was 34.67%（52/150）. In AKI group, proportion of male, age, proportion of diabetes mellitus （DM）, hypertension, renal dysfunction, history of cardiac surgery, red blood cell distribution width （RDW）, serum homocysteine （Hcy） and serum cystatin C （Cys C） were 67.31%, （61.23 ± 6.35） year, 26.92%, 48.08%, 34.62%, 38.46%, （0.71 ± 0.22）%, （16.10 ± 4.28） μmol/L and （2.03 ± 0.57） mg/L. They were higher than （41.84%）, （54.26 ± 5.19） year, 10.20%, 24.49%, 9.18%, 13.27%, （0.39 ± 0.12）%, （13.65 ± 3.71） μmol/L and （1.02 ± 0.45） mg/L in non-AKI group. Preoperative nutritional risk index （NRI） was lower than non-AKI group [（82.34 ± 9.20） vs （113.57 ± 10.58）]. Extracorporeal circulation time and aortic occlusion time were [（98.12 ± 18.69） vs （84.19 ± 15.37） min] and [（84.27 ± 16.73） vs （73.58 ± 15.11） min]. The differences were statistically significant （P＜0.05）. Gender （OR = 5.810）, age （OR = 5.380）, DM （OR = 5.113）, hypertension （OR = 5.625）, renal dysfunction （OR = 7.825）, history of cardiac surgery （OR = 6.935）, preoperative NRI （OR = 7.630）, RDW （OR = 6.495）, serum Hcy （OR = 3.244）, Cys C （OR = 6.819）, cardiopulmonary bypass time （OR = 3.604） and aortic occlusion time （OR = 3.374） were all influencing factors for AKI post-HVR （P＜0.05）. A nomogram prediction model was employed for predicting the post-HVR occurrence of AKI. External validation indicated that its area under the curve （AUC） was 0.864（95%CI: 0.815-0.901） with a predictive sensitivity and specificity of 86.01% and 85.36% respectively. And DCA showed that the model demonstrated an excellent clinical efficacy in predicting the post-HVR occurrence of AKI. Conclusion The influencing factors for AKI post-HVR include gender, age, DM, hypertension, renal dysfunction, history of cardiac surgery, preoperative NRI, RDW, serum levels of Hcy and Cys C, extracorporeal circulation time and aortic occlusion time. A nomogram prediction model based upon the above factors may assist in clinical screening of high-risk AKI patients.

Keywords

• acute kidney injury
• influencing factors
• prediction