OBJECTIVE: To measure interobserver variation in recording injury from case notes and its effect on calculating injury severity scores (ISS) from identical data and predicting probabilities of survival by using the combined trauma and injury severity score (TRISS). DESIGN: Observer variation study using injury severity scoring and subsequent calculation of probability of survival based on combined trauma and injury severity scores. SUBJECTS: 16 patients with a range of injury severity scores, and 15 observers. RESULTS: There was a wide variation in recorded injury severity scores, the probability of two observers agreeing on the score being 0.28 (28%). The probability of any two observers agreeing over which severity band the patient should be in was 0.5 (50%). Observer variation was independent of the training and type of observer. Survival probability (calculated by combined trauma and injury severity scoring methodology from individual observers' scores) varied by over 0.2 in six of the 16 patients and by over 0.5 in three. CONCLUSIONS: There is wide observer variation in injury severity scoring, which highlights a potential fallibility in its use for trauma audit. The use of combined trauma and injury severity scoring for individual prediction of survival is potentially inaccurate except at the extremes of probabilities. The use of the 0.5 survival line on a combined trauma and injury severity score "pre-chart" is statistically and clinically inappropriate.
OBJECTIVE: To measure interobserver variation in recording injury from case notes and its effect on calculating injury severity scores (ISS) from identical data and predicting probabilities of survival by using the combined trauma and injury severity score (TRISS). DESIGN: Observer variation study using injury severity scoring and subsequent calculation of probability of survival based on combined trauma and injury severity scores. SUBJECTS: 16 patients with a range of injury severity scores, and 15 observers. RESULTS: There was a wide variation in recorded injury severity scores, the probability of two observers agreeing on the score being 0.28 (28%). The probability of any two observers agreeing over which severity band the patient should be in was 0.5 (50%). Observer variation was independent of the training and type of observer. Survival probability (calculated by combined trauma and injury severity scoring methodology from individual observers' scores) varied by over 0.2 in six of the 16 patients and by over 0.5 in three. CONCLUSIONS: There is wide observer variation in injury severity scoring, which highlights a potential fallibility in its use for trauma audit. The use of combined trauma and injury severity scoring for individual prediction of survival is potentially inaccurate except at the extremes of probabilities. The use of the 0.5 survival line on a combined trauma and injury severity score "pre-chart" is statistically and clinically inappropriate.
Authors: Leontien M Sturms; Josephine M Hoogeveen; Saskia Le Cessie; Peter E Schenck; Paul V M Pahlplatz; Mike Hogervorst; Gerrolt N Jukema Journal: Langenbecks Arch Surg Date: 2006-05-13 Impact factor: 3.445
Authors: R J Russell; T J Hodgetts; J McLeod; K Starkey; P Mahoney; K Harrison; E Bell Journal: Philos Trans R Soc Lond B Biol Sci Date: 2011-01-27 Impact factor: 6.237
Authors: Cameron S Palmer; Melanie Franklyn; Christine Read-Allsopp; Susan McLellan; Louise E Niggemeyer Journal: Scand J Trauma Resusc Emerg Med Date: 2011-05-08 Impact factor: 2.953