gms | German Medical Science

Objective: The cranial Abbreviated Injury Score (AIS) is coded by results of initial computed tomography (CT). The AIS is influential to the therapy and part of the score for the overall prognosis: Injury Severity Score. MRI may also be used for coding of cranial injuries but this is only seldom performed. We examined the impact of MRI on the prognostic potency of different scores: Injury Severity Score (ISS), New Injury Severity Score (NISS), Trauma and Injury Severity Score (TRISS) and Revised Injury Severity Classification (RISC)

Methods: We identified 63 patients with a cranial AIS ≥ 1 out of our trauma database from a period of four years. All patients had cranial MRI and were rated by the Glasgow Outcome Scale (GOS). All values for AIS, ISS, NISS, TRISS, RISC and GOS were statistically used as ordinal variables. For statistical hypothesis testing, the Fisher-Exact-Test was used for dichotomous variables. Mann-Whitney-U-Test and Wilcoxon Tests were also used for specific samples. The level of significance was set as p≤0.05. The Receiver-Operating-Characteristic (ROC) accompanied by associated Area Under the Curve (AUC) served as cut-off independent instrument to quantify the predictive potency. The correlation-coefficient after Spearman (r_S) was calculated for correlation between the scores and the GOS. Sensitivity, specifity and accuracy (correct-classification rate) were also calculated for each Score.

Results: We found favourable outcome (GOS 4-5) in 31 patients and unfavourable outcome (GOS 1-3) in 31 patients. The prediction of outcome was poor by CT data: ISS (ROC 0.553; r_s = -0.14; p = 0.465), NISS (ROC 0.643; r_s = -0.34; p = 0.051), TRISS (ROC 0.731; r_s = 0.37; p = 0.002), RISC (ROC 0.781; r_s = 0.55; p < 0.001). 35% of the patients had brain stem injury in MR. However, those injuries were detected by CT only in 5%. The prognostic potency of the AIS could be improved after each brain stem lesion was coded by the score of 5: ISS (ROC 0.614; r_s = -0.25; p = 0.118), NISS (ROC 0.686; r_s = -0.42; p = 0.011), TRISS (ROC 0.768; r_s = 0.45; p < 0,001), RISC (ROC 0.831; r_s = 0.64; p < 0,001). The predictive potency could furthermore be improved after adjusting the AIS coding by matching the score to the specific mortality risks of brain stem injuries: ISS (ROC 0.639; r_s = -0.32, p = 0,058), NISS (ROC 0.752; r_s = -0.56; p = 0.001), TRISS (ROC 0.831; r_s = 0.61; p < 0.001), RISC (ROC 0.837; r_s = 0.69; p < 0.001). The RISC Score had the best predictive potency in all of the three calculating variants.

Conclusion: The best predictive potency was achieved by matching the AIS coding to the predictive impact of brain stem injuries. MRI data are preferable to CT data to improve the prognostic potency of trauma-scores after brain injury. The RISC Score should be used. Nevertheless, calculating efforts are increased due to this measure but not using MRI data, if they are available, means missing the possibility to improve the prognostic potency.