Comparison of the predictive value of APACHE II, SOFA, SAPS II, GCS and GCS-P scores for in-hospital mortality in critically ill patients after craniotomy: a retrospective cohort study in a Chinese tertiary hospital

Objective The purpose of this study was to evaluate the accuracy of different scoring systems, including the Acute Physiology and Chronic Health Evaluation (APACHE) II, the Simplified Acute Physiology Score (SAPS) II, the Sequential Organ Failure Assessment (SOFA) score, the Glasgow Coma Scale (GCS) and the Glasgow Coma Scale-Pupils (GCS-P), in predicting in-hospital mortality for critically ill patients after craniotomy. Design This was a single-centre retrospective diagnostic study. Setting The study was conducted in three intensive care units (70 beds) of a teaching hospital. Participants Adult patients who underwent craniotomy and stayed in the ICU for more than 24 hours were included. Pregnant or lactating women and patients enrolled in other clinical studies were excluded. Outcome measures Data on demographics, clinical characteristics and scoring systems (APACHE II, SOFA, SAPS II, GCS and GCS-P) were collected. Receiver operating characteristic (ROC) curves were used to assess the predictive accuracy of each scoring system. Results Among 1717 patients included, the in-hospital mortality rate was 6.8%. SAPS II (AUC = 0.822) and APACHE II (AUC = 0.819) demonstrated the highest predictive accuracy, followed by GCS-P (AUC = 0.812), GCS (AUC = 0.803) and SOFA (AUC = 0.762). SAPS II and APACHE II significantly outperformed SOFA, while no significant differences were observed among SAPS II, APACHE II, GCS and GCS-P. For patients with supratentorial lesions, APACHE II, SAPS II, GCS and GCS-P showed similar predictive accuracy, all superior to SOFA. For infratentorial lesions, SOFA outperformed GCS and GCS-P. Among patients with cerebrovascular diseases, SOFA had the lowest predictive accuracy, while GCS-P outperformed GCS. Conclusions APACHE II, SAPS II, GCS and GCS-P demonstrated comparable predictive accuracy for in-hospital mortality in critically ill post-craniotomy patients, with SOFA being less effective. Trial registration number NCT06762184 .