Performance of SOFA, qSOFA and SIRS to predict septic shock after percutaneous nephrolithotomy

The new clinical criteria termed SOFA and qSOFA were demonstrated to be more accurate than SIRS in screening patients at high risk of sepsis. We aim to evaluate the ability of SOFA, qSOFA and SIRS to predict septic shock after PCNL.

Consecutive patients undergoing PCNL were included to assess the performance of SOFA, qSOFA and SIRS in predicting septic shock, the AUC of ROC curve and decision curve analysis were used, and the optimal cutoff values and their achieving time were calculated.

Of the 431 included patients, 12 (2.7%) cases developed septic shock. Compared with non-septic shock patients, patients with septic shock were more likely to be female, have positive history of urine culture and higher urine leukocyte count, and show increased postoperative serum creatinine, PCT and decreased leukocyte. The optimal cutoff of SOFA, qSOFA and SIRS was > 2, > 0 and > 1, respectively. All of the 12 patients with verified septic shock met SOFA and SIRS criteria, while only 11 cases met qSOFA criterion. SOFA had the identical highest sensitivity (100%) and greater specificity (87% vs. 81%) than SIRS. qSOFA had higher specificity (92%) than both SOFA and SIRS at the expense of lower sensitivity (92%). The AUC of SOFA (0.973) to predict septic shock was greater than that of qSOFA (0.928) and SIRS (0.935). When combined with SIRS, SOFA outperformed qSOFA for discrimination of septic shock (AUC 0.987 vs. 0.978). Decision curve analysis indicated SOFA was clearly superior to both qSOFA and SIRS with a higher net benefit and net reduction in intervention. The qSOFA achieved the best time-based predictive efficiency, with the shortest median time to meet its cutoff, followed by SOFA and SIRS.

The performance of SOFA in predicting septic shock after PCNL was slightly greater than qSOFA and SIRS. The comprehensive application of various criteria is recommended to assist early detection of septic shock following PCNL.