Abstract
Background Sequential organ failure assessment (SOFA) score is widely used in many cardiac surgical
intensive care units (ICUs). Its derivatives (mean and maximum values) are known to
be more accurate than the original daily values of SOFA itself. However, they were
designed for research purposes and could be calculated only after ICU discharge. We
aimed to develop a reliable derivative that can be easily calculated daily (Daily-Mean-SOFA)
for aiding daily-decision-making and resource allocation.
Methods All consecutive adult cardiac surgical patients from our ICU between January 1, 2007
and December 31, 2008 were included. We obtained Initial-SOFA (on day 1), the Original-Daily-SOFA
value from the 1st to the 6th postoperative day, Max-SOFA (highest SOFA value during
the whole ICU-stay), Mean-SOFA (sum of all daily SOFA values/the length of ICU-stay),
and the new “Daily-Mean-SOFA” from day 2 to 6 (sum of SOFA from day 1 until day-n/n). We compared their accuracies at predicting ICU mortality using calibration and
discrimination statistics.
Results Total 2801 patients were included. The newly developed “Daily-Mean-SOFA” was significantly
more accurate than the corresponding SOFA value of the same day in correctly predicting
survival and mortality in the whole study population (OCC: 94.1 to 95.0%) and in accurately
identifying the individual patient’s risk of mortality (AUC: 0.859 to 0.904). It was
better than all other derivatives except the Mean-SOFA which was superior to it (OCC:
96.3%; AUC: 0.913).
Conclusions The Daily-Mean-SOFA is a reliable derivative for daily risk stratification in cardiac
ICUs. Due to its accuracy and daily availability, it may be used for risk-directed
therapy in cardiac ICUs.
Keywords
outcomes (includes mortality morbidity) - intensive care - statistics