Study On Trauma Severity Score Method Based On ICD

Trauma score, as an important tool for injury treatment supporting and a key branch of trauma research, is more and more concerned by people in trauma field world wide. More and more researches on trauma score methods have been carried out and the methods began to be applied in more and more jobs in trauma field. In China, researches on trauma score were carried out more recently than those in western countries and the practical utilizations of them were much limited. In recent years, however, trauma score has been concerned by more people and more efforts have been made for their development. Trauma diseases, whose incidences are among those highest ones, are main death causes in China. Therefore, it is much important and valuable for improving the whole health condition of China to make more efforts on trauma score researches.The qualities of trauma score researches carried out in China are not so high. Most researches on trauma score are only related to testing and validating the methods directly imported from foreign countries for Chinese patients and few of them are related to developing new methods. In western countries, there often is a well developed social supporting system for traumapatients, which can become the key and basic factor for trauma score methods development. Furthermore, most methods of trauma score involve the calculating of some certain parameters based on certain populations. Thus those trauma score methods directly imported from western countries can not be easily or directly used for Chinese patients. In brief, trauma score methods commonly used world wide are not necessarily suitable for Chinese patients and are much hard for us to master and practice. These all should be the main bottle-necks for trauma score development and the important causes of the limitation of trauma score utilizations in China.In this research, we intend to establish a trauma score method which can be easily mastered, considerably effective and comparatively cheap in practice and thus to explore a new approach for trauma score researches such that help to build the basis for advancing trauma score studies. Inspired by jobs done by MacKenzie et al, we made a scheme of trauma score method based on ICD codes which can be used to instantly evaluate the injury severities for trauma patients with regularly collected information such as diagnoses and some certain vital data. In recent years, a conversion table from ICD codes to AIS values has been composed by our department. It can be good base for this research.In this research, the ICD-AIS conversion table was rebuilt to the second version with new content added into it. The new content added into conversion table is about the reliabilities of the mapping relations between ICD codes and AIS values. In the second version of conversion table we gave quantitative indices indicating the reliabilities of all correspondence relations, which is named as RLB and can be used to weight the AIS values obtained through conversion table in calculating NISS values, and the principles for using theindices were made. In the primary evaluation of the conversion table, we found it is effective in predicting the outcomes of trauma patients. In the results of ROC analysis, the AUC was up to 0.85. It showed that the second version of conversion table is practically useful in evaluating the injury severities of trauma patients.In the designing of new trauma score method, we followed the way of trauma scoring in TRISS method. In the structure of the new trauma score method, anatomic score, physiological score and the age factor are combined together with certain weights to attain a certain integrated score indicating the overall status of the trauma patient. Among the three parts of the integrated score, the anatomic score will be acted by NISS value with the AIS values collected using the ICD-AIS conversion table, the physiological score will be acted by PSI value, which is attained using PSI method developed by ourselves following the way of trauma score of PHI method, and the age factor will take the same form as that in TRISS method. Because of the lack of trauma data of Chinese patients corresponding to the MTOS in U.S., we could not directly obtain the weights for the three kinds of trauma scores in equation for the integrated score. As an alternative way, we primarily specified the weights manually according to the relative magnitudes of them for blunt injury in TRISS method.We calculated the trauma scores for two groups of patients collected from two hospitals, i.e. cerebral trauma patients and facial trauma patients, using the new method designed in this research, namely CISS method, with the aid of computers. CISS values and RLB weighted CISS values were obtained simultaneously for all trauma patients involved in the research. Results of linear correlation analysis and rank correlation analysis showed that CISSvalues and RLB weighted CISS values have strong correlations with the NISS values obtained by professional trauma score raters. This indicated that the CISS method with AIS values obtained through conversion table is consistent with NISS method with AIS values obtained by professional raters. Through ROC analysis, we have evaluated the effectiveness of trauma outcome prediction of two kinds of trauma scores and compared them using AUC comparisons. From the results, we found that two kinds of trauma scores are all effective in predicting trauma outcomes. For cerebral trauma patients, the AUCs of two kinds of trauma scores are all greater than 0.9 and there are no statistically significant differences between them (P>0.05). For facial trauma patients, however, two kinds of trauma scores are not so effective to some extent, for which the AUCs are between 0.67 and 0.70 and there are also no differences between them. Even though, we should believe that the reason for the unwanted results is likely the shortage of information in the data of facial trauma patients because of the similar performances of the NISS values obtained by professional raters.Through comparison between the performances of CISS values and RLB weighted CISS values, we found that this two scores are highly correlated, i.e. the correlation coefficient is near to 1 (P<0.0001), and the AUCs for them are nearly equal. This showed that the effect of weighting by RLB is minor. From the results for two groups of trauma patients, however, we found that the RLB weighted CISS values always have better performances than those of original CISS values even though the differences between them are extremely minor. This showed that the weighting of RLB exactly improves the performances of CISS method. Therefore, in the structure of CISS method, the weighting process by RLB should be retained and be further discussed in futureresearches.The new trauma score method based on ICD codes, which combined together the NISS value with AIS values obtained through ICD-AIS conversion table, PSI value calculated using regular data of trauma patient and the age factor, is more effective in grading the overall status of trauma patient and thus has more practical values. In this research, the weights for three items of the equation of CISS value are specified manually because of the lack of trauma data corresponding to MTOS of U.S. This would make the CISS value a little uncertain. Thus there would be some certain spaces for CISS method to improve its performance. In further researches, we will collect enough trauma data to objectively estimate the weights for the equation of CISS value and continuously make revisions to it every time when the data are updated.At present, the 10th edition of ICD codes has been published by WHO and the composition of the 6-digit version of them are finished in China. A newer version of AIS is also under consideration. The base of the newly proposed trauma score method, i.e. ICD-AIS conversion table, will be continuously updated each time when either of the two codes systems is updated. We will compose a conversion table mapping the ICD-9 codes to the ICD-10 codes soon thus make it possible to use the CISS method for patients with their diagnoses coded using 10_th ICD codes. Though there will be many revisions or updates in the contents of ICD-AIS conversion table, the principle of CISS method will almost keep constant and it won't increase the difficulties in using CISS method.In brief, the new trauma score method, namely CISS method, is effective in grading the injury severity for trauma patients and has a good performance in predicting trauma outcomes. All information needed in this method can be...