Abstract

Background
Due to several limitations of the Trauma and Injury Severity Score (TRISS), the ICD-9CM-based Injury Severity Score (ICISS) has recently been proposed as an alternative method. Few limitations are known about the ICISS because of it’s recent application. The purpose of this study was to examine the limitations of the ICD-10-based ICISS and the ICISS full model by using a case analysis. Objects & Method: Seven years of data on 595 patients who suffered from serious injury were collected prospectively. We compared the final outcomes of the patients with the probabilities of survival (Ps) from the ICISS full model. The cutoff value of Ps from the ICISS full model that we decided for patient’s survival was 0.5. Expired patients with high values of the Ps were defined as“ unexpected deaths”; conversely, patient with low values of the Ps that survived were defined as “unexpected survivals”. We only reviewed the medical records in “unexpected” cases, searching for the causes of the mismatch. The known limitations of the ICISS (and the full model) were categorized and then assigned to each case. Throughout the process, we were able to identify limitations of the ICISS (and the full model) not known till now. Results: Among the 595 patients, 212 patients (35.6%) expired. The average score of the ICISS full model was 0.622. Limitations of the ICISS (and the full model) are as follows: First, various levels of severity could not be distinguished with the ICISS because they are represented as a single code. Secondly, some diagnoses couldn’t be represented in current ICD-10 code. Thirdly, the ICD code couldn’t distinguish a unilateral injury from a bilateral injury, distorting the injury severity. Fourthly, the survival risk ratio (SRR) database may be unreliable in cases of low-incidence injuries and in cases of patients who died before a definite diagnosis. Because the structure of the ICISS full model resembles the TRISS, the limitations are similar, and those are a dichomotous age portion and consideration of only the ‘initial’revised trauma score (RTS). Though it is said that the ICISS has the potential to reflect non-trauma illnesses, preexisting medical conditions and post-injury medical complications cannot be included in the ICISS model. Conclusion: We suggest the followings for improving the ICISS (and the full model). First, limitations shared with the TRISS should be corrected. Second, so that pre-existing conditions can be considered, some pre-existing conditions should be included in the ICISS as constituents. Third, a SRR database for non-trauma illnesses should be established so that post-injury medical complications can be considered. Some ICD-10 codes not established until now should be newly made. The exsiting ICD-10 code should be modified to include various severities and lateralities. Fourth, several SRR databases, corresponding to the levels of the trauma care of the medical institution, should be established, and multicenter trials on SRR databases should be conducted in the future. Fifth, autopsies should be routine in cases where definite diagnoses were not made.

• Keywords: Trauma score, ICISS, Limitation, Improvement