Outcome prediction of different groups of patients using a modified scoring system

  Abstract   Background: In this study we aimed to examine the discrimination and calibration of   a severity characterization of trauma (ASCOT) in our setting to determine whether its usage   is appropriate to predict outcome of our trauma patients.   Methods: This study was conducted in three hospitals. All patients admitted in studied   hospitals were divided randomly into two equal subgroups. In each group, new coefficients   for ASCOT were derived from the first subgroup of patients. Then the newly developed   model was validated in the second subgroup and the measures of discrimination   and calibration were calculated.   Results: 78% of our patients were male and 27% were children. The study mortality   rate was 6%, and 20% of patients had penetrating trauma. The average age of our patients   was 28 ± 19 (Mean ± Standard Deviation).The area under ROC for ASCOTwas 0.96 and   Hosmer-Lemeshow goodness of fit p value was 0.81.   Conclusion: In spite of many differences of the trauma care systems between our   country and western countries, current survival probability models can be used in our   country after customization and development of new coefficients derived from regional   databases.