The value of trauma patients' centralization: an analysis of a regional Italian Trauma System performance with TMPM-ICD-9.

BACKGROUND: In recent years, many studies showed that the Trauma Mortality Probability Model (TMPM-ICD-9) had better calibration compared to other ICD-9-based models and to the ones based to the Abbreviated Injury Scale (AIS). The study aims to assess the validity of TMPM-ICD-9 in predicting injury severity in an Italian region and, through this model, to assess the performances of the Trauma Systems SIAT Romagna.
METHODS: Administrative data of trauma patients admitted in the Trauma System of SIAT Romagna, in Northern Italy, from 2014 to 2018 were obtained. The XISS, an indirect indicator of Injury Severity Score (ISS) and the TMPM-POD (Probability of Death) were calculated from ICD-9-CM codes. Only patients with XISS > 15 were included. Student t-test, Mann-Whitney test and Chi-square test were used for univariate analyses, while logistic regression for multivariate analyses.
RESULTS: 3907 trauma patients with XISS > 15 were included. The Hub hospital (HUB) received 47.1% of these patients. Patients treated in HUB had higher TMPM-POD than in SPOKE + PST (mean TMPM-POD ± SD: HUB 0.093 ± 0.091, SPOKE + PST 0.082 ± 0.90, p < 0.027), but only age and sex were significant risk factors for centralization at multivariate analyses. Higher age (73.1 ± 21.2 vs 66.9 ± 21.2, p < 0.001), higher XISS (16(9) vs 16(4), p < 0.001) and higher TMPM-POD (0.15 ± 0.14 vs 0.08 ± 0.08, p < 0.001) resulted significant risk factors for mortality at multivariate analysis. Lower age, higher XISS and lower Trauma Centers (TC) level were significant risk factors for splenectomy at multivariate analysis. The splenectomy rate was 1.3% in HUB and of 2.2% in SPOKE + PST (Risk Ratio = 0.4, p = 0.002).
CONCLUSIONS: Present analysis proved the validity of TMPM-ICD-9 in predicting mortality of trauma patients in an Italian region. Furthermore, the usefulness of data extracted from an administrative database to assess the performance of a TS and the importance of an adequate centralization process have emerged. Even with a higher TMPM-POD and with the same mortality rate, HUB showed a higher spleen salvage rate compared to SPOKE + PST. However, thanks to this model, an improvable centralization process in SIAT Romagna was found in the study period. Probably, an enhanced centralization would have improved the spleen salvage rate, which is an important quality indicator in the evaluation of the performance of the TS.