Background: The TRISS methodology, in combination with coefficients derived from the Major Trauma Outcome Study (MTOS), is the most widely used outcome prediction model for the care of trauma patients. Utilizing the M-statistic, different populations of trauma patients can be compared with the population originally enrolled in the MTOS.
Purpose: We hypothesized that databases outside of North-America would not be well matched to the MTOS study and thus the TRISS methodology would not accurately predict outcome in these different populations.
Methods: All trauma studies utilizing TRISS methodology that were published between 1990-2003 were reviewed and M-statistics calculated based on the population described in the study. The populations were grouped by the following geographic locations: Europe, Asia/Africa and North-America.
Results: The median M-statistic for Europe was 0.65, compared with 0.88 for Asian/African databases, and 0.90 for North-American studies. There was a significant difference between European and North-American studies (p < 0.05).
Conclusion: The trauma populations described in European studies differ significantly from the MTOS with respect to injury severity match, indicating the need for the development of regional trauma databases and modified TRISS coefficients based on the geographic location of the injured population included.