Differences in mortality predictions between Injury Severity Score triplets: a significant flaw
Related Experiment Videos
Robert Russell1, Elizabeth Halcomb, Erica Caldwell
1Royal Army Medical Corps, Ministry of Defence Hospital Unit (Peterborough), Peterborough, United Kingdom. robrussell@doctors.org.uk
Background:
This study investigated the validity of similar Injury Severity Scores (ISS) generated by different Abbreviated Injury Scale triplets.
Methods:
A cohort of trauma patients admitted to a single major trauma service between 1995 and 2002 was studied retrospectively. Mortality rates were compared for groups with identical ISS scores but different triplets.
Results:
For 2,223 of the 5,946 trauma patients studied, 12 ISS scores were generated by two different Abbreviated Injury Scale triplets. Six of these ISS totals showed significant differences in mortality depending on the triplet source. One of the most striking was ISS 25 (triplet 5,0,0), with a mortality of 20.6%, as compared with 0% for triplet 4,3,0 (p = 0.005). The other statistically significant mortality differences for ISS totals were ISS 27-28.6% (5,1,1) versus 7.4% (3,3,3) (p = 0.05); ISS 29-30.3% (5,2,0) versus 4.6% (4,3,2) (p = 0.002); ISS 33-50% (4,4,1) versus 6.7% (5,2,2) (p = 0.034); ISS 34-45.2% (5,3,0) versus 4.3% (4,3,3) (p = 0.0009); and ISS 41-60% (5,4,0) versus 11.1% (4,4,3) (p = 0.05).
Conclusions:
The mortality rates are significantly different between pairs of triplets that generate the same ISS total. Caution must be used in the interpretation of outcomes from ISS values generated by different triplets.