PURPOSE: As global initiatives increase patient access to surgical treatments, there is a need to define optimal levels of perioperative care. Our aim was to describe the relationship between the provision and use of critical care resources and postoperative mortality. METHODS: Planned analysis of data collected during an international 7-day cohort study of adults undergoing elective in-patient surgery. We used risk-adjusted mixed-effects logistic regression models to evaluate the association between admission to critical care immediately after surgery and in-hospital mortality. We evaluated hospital-level associations between mortality and critical care admission immediately after surgery, critical care admission to treat life-threatening complications, and hospital provision of critical care beds. We evaluated the effect of national income using interaction tests. RESULTS: 44,814 patients from 474 hospitals in 27 countries were available for analysis. Death was more frequent amongst patients admitted directly to critical care after surgery (critical care: 103/4317 patients [2%], standard ward: 99/39,566 patients [0.3%]; adjusted OR 3.01 [2.10-5.21]; p < 0.001). This association may differ with national income (high income countries OR 2.50 vs. low and middle income countries OR 4.68; p = 0.07). At hospital level, there was no association between mortality and critical care admission directly after surgery (p = 0.26), critical care admission to treat complications (p = 0.33), or provision of critical care beds (p = 0.70). Findings of the hospital-level analyses were not affected by national income status. A sensitivity analysis including only high-risk patients yielded similar findings. CONCLUSIONS: We did not identify any survival benefit from critical care admission following surgery.
Intensive care med
971 - 979
Critical care/utilisation, Postoperative care/methods, Postoperative care/statistics and numerical data, Surgical procedures, operative/mortality, Adult, Elective Surgical Procedures, Female, Hospitalization, Humans, Intensive Care Units, Length of Stay, Logistic Models, Male, Middle Aged, Perioperative Care, Postoperative Period, Prospective Studies