A Prediction Model for Acute Kidney Injury After Pericardiectomy: An Observational Study

OBJECTIVES: Acute kidney injury is a common complication after pericardiectomy for constrictive pericarditis, which predisposes patients to worse outcomes and high medical costs. We aimed to investigate potential risk factors and consequences and establish a prediction model. METHODS: We selected patients with constrictive pericarditis receiving isolated pericardiectomy from January 2013 to January 2021. Patients receiving concomittant surgery or repeat percardiectomy, as well as end-stage of renal disease were excluded. Acute kidney injury was diagnosed and classified according to the KDIGO criteria. Clinical features were compared between patients with and without postoperative acute kidney injury. A prediction model was established based on multivariable regression analysis. RESULTS: Among two hundred and eleven patients, ninety-five (45.0%) developed postoperative acute kidney injury, with fourty-three (45.3%), twenty-eight (29.5%), and twenty-four (25.3%) in mild, moderate and severe stages, respectively. Twenty-nine (13.7%) patients received hemofiltration. Nine (4.3%) patients died perioperatively and were all in the acute kidney injury (9.5%) group. Eleven (5.2%) patients were considered to have chronic renal dysfunction states at the 6-month postoperative follow-up, and eight (72.7%) of them experienced moderate to severe stages of postoperative acute kidney injury. Univariable analysis showed that patients with acute kidney injury were older (difference 8 years, P < 0.001); had higher body mass index (difference 1.68 kg·m(−2), P = 0.002); rates of smoking (OR = 2, P = 0.020), hypertension (OR = 2.83, P = 0.004), and renal dysfunction (OR = 3.58, P = 0.002); higher central venous pressure (difference 3 cm H(2)O, P < 0.001); and lower cardiac index (difference −0.23 L·min(−1)·m(−2), P < 0.001) than patients without acute kidney injury. Multivariable regression analysis showed that advanced age (OR 1.03, P = 0.003), high body mass index (OR 1.10, P = 0.024), preoperative atrial arrhythmia (OR 3.12, P = 0.041), renal dysfunction (OR 2.70 P = 0.043), high central venous pressure (OR 1.12, P = 0.002), and low cardiac index (OR 0.36, P = 0.009) were associated with a high risk of postoperative acute kidney injury. A nomogram was established based on the regression results. The model showed good model fitness (Hosmer-Lemeshow test P = 0.881), with an area under the curve value of 0.78 (95% CI: 0.71, 0.84, P < 0.001). CONCLUSION: The prediction model may help with early recognition, management, and reduction of acute kidney injury after pericardiectomy.