This retrospective study aimed to identify factors associated with prognosis in patients who developed postoperative acute kidney injury (AKI) and received continuous renal replacement therapy (CRRT) after cardiac surgery. The authors reviewed cases treated at Fuwai Huazhong Cardiovascular Disease Hospital between January 1, 2018 and December 31, 2020 and performed follow-up for 90 days after surgery.
A total of 113 patients meeting inclusion criteria were enrolled. The cohort had a mean age of 57.7 ± 12.3 years and included 55 males and 58 females. All patients underwent cardiac surgery, developed postoperative AKI, and received CRRT during the index hospitalization. Follow-up extended to 90 days after surgery to capture the study endpoint.
The primary composite endpoint combined dialysis dependence or death at 90 days postoperatively. Patients were categorized into two groups based on whether this endpoint occurred: the successful RRT group (no dialysis dependence or death at 90 days) and the failed RRT group (dialysis dependence or death within 90 days).
Of the 113 patients, 56 were in the successful RRT group and 57 in the failed RRT group. Key univariate differences between groups included:
Preoperative diabetes mellitus was more prevalent in the failed RRT group (29.8% [17/57]) than in the successful group (3.6% [2/56]).
Intraoperative transfusion volumes were lower in the failed RRT group: mean red blood cell transfusion [(474 ± 188) ml] versus [(559 ± 195) ml] in the successful group; platelet transfusion units were also lower [(1.2 ± 0.6) U vs (1.5 ± 0.7) U].
Serum creatinine at discharge was higher in the failed group (230 ± 167 μmol/L) compared with the successful group (129 ± 80 μmol/L).
A substantially larger proportion of patients in the failed RRT group remained dialysis dependent at discharge (80.7% [46/57]) versus the successful group (39.3% [22/56]).
All reported between-group differences above reached statistical significance (P < 0.05) as stated in the source.
A multivariate logistic regression model was used to identify independent predictors of 90-day RRT failure among AKI patients treated with CRRT after cardiac surgery. The model identified four independent risk factors:
Male sex: odds ratio (OR) = 2.930; 95% confidence interval (CI) 1.086–7.904; P = 0.034.
Preoperative diabetes mellitus: OR = 12.033; 95% CI 2.136–67.780; P = 0.005.
Serum creatinine at discharge: OR = 1.006 per μmol/L; 95% CI 1.002–1.010; P = 0.005.
Dialysis dependence at discharge: OR = 5.147; 95% CI 1.762–15.036; P = 0.003.
These variables were retained as independent predictors in the multivariable model reported by the authors.
The authors constructed a risk prediction model incorporating the four independent predictors (male sex, preoperative diabetes mellitus, serum creatinine at discharge, and dialysis dependence at discharge). The model's discriminative ability was assessed using the receiver operating characteristic (ROC) curve; the reported area under the ROC curve (AUC) was 0.850 with a 95% CI of 0.782–0.918 and P < 0.001, indicating good overall discrimination for predicting the composite endpoint of dialysis dependence or death at 90 days.
Within this single-center retrospective cohort of postoperative cardiac surgery patients with AKI managed with CRRT, the study concludes that male sex, preoperative diabetes mellitus, elevated serum creatinine at discharge, and dialysis dependence at discharge are associated with increased risk of dialysis dependence or death at 90 days. The combined model of these factors demonstrated good predictive performance (AUC 0.850) according to the reported data.
Clinically, these findings highlight discharge status (creatinine and dialysis dependence) and baseline factors (sex and diabetes) as important markers for risk stratification in this patient population. Patients with these risk features may warrant closer outpatient monitoring, earlier nephrology follow-up, and targeted interventions to attempt to reduce progression to persistent dialysis dependence or death, though specific interventions were not addressed in the source.
The source reports a retrospective design from a single center and provides no external validation of the predictive model; details about specific inclusion/exclusion criteria, perioperative management, CRRT modality settings, or causes of mortality were not reported in the abstract. The authors declared no conflicts of interest in the article.