Authors (including presenting author) :
Hung KKC (1)(2), Lai CY (1)(2), Yeung HH (1)(2), Maegele M (3), Leung LY (1), Chang A (4), Chan L (4), Wong J (5), Leung M (6), Wong HT (7), Cheng CH (1)(2), Cheung NK (1)(2), Graham CA (1)(2)
Affiliation :
(1) Accident and Emergency Medicine Academic Unit, Chinese University of Hong Kong, (2) Trauma & Emergency Centre, Prince of Wales Hospital, Hong Kong, (3) Cologne-Merheim Medical Center (CMMC), University Witten-Herdecke (UWH), Germany, (4) Accident and Emergency Department, Queen Elizabeth Hospital, Hong Kong, (5) Accident and Emergency Department, Tuen Mun Hospital, Hong Kong, (6) Accident and Emergency Department, Princess Margaret Hospital, Hong Kong, (7) Accident and Emergency Department, Queen Mary Hospital, Hong Kong
Introduction :
Hong Kong trauma registries have been using TRISS methodology for audit and benchmarking purposes since its establishment. TraumaRegister DGU in Germany devised its own probability of survival model (RISC II) using data from its 900 trauma centres.
Objectives :
The aim is to compare the predictive ability of the probability of survival calculated using TRISS and RISC II for major trauma patients in Hong Kong.
Methodology :
This was a retrospective cohort study using data from all five trauma centres in Hong Kong. Adult major trauma patients (ISS>15) from January 2013 to December 2015 were extracted from the five respective trauma registries. The primary outcome was the area under the ROC curve for TRISS and RISC II using the expected and observed 30-day mortality. Probabilities of survival (Ps) were derived by TRISS with MTOS coefficients and RISC II methodology. The Hosmer-Lemeshow goodness of fit test was used to test for the calibration of the model. Subgroups analyses investigated the performance of TRISS and RISC II for the mechanism of injury and age>80.
Result & Outcome :
1864 patients were recruited. 67.2% was male and the median age was 60. The median ISS was 24, with 40% of patients with ISS> 25. Low fall was the most common mechanism of injury, with head and neck being the most commonly injured body region. The 30-day mortality was 22.4%. The expected mortality was 20.0% using TRISS and 19.7% from RISC II. The AUC was 84.8% (CI 82.7 to 86.9) and HL test 63.2 (p< 0.001) for TRISS. RISC II yielded a superior AUC of 89.6% (CI 88.1 to 91.2) and HL test of 78.9 (p< 0.001). Subgroup analyses showed that both score performed worse for ISS 25 or above (AUC: TRISS 80.4%, RISC II 87.7%), age 80 or above (AUC: TRISS 80.6%, RISC II 82.9%), low falls (< 2m) (AUC: TRISS 81.7%, RISC II 85.5%), and significant head or neck injury (AIS 3 or above) (AUC: TRISS 83.1%, RISC II 87.7%). RISC II was significantly better than TRISS in all subgroups, except in age 80 or above and low falls. RISC II was superior to TRISS in predicting the 30-day mortality for Hong Kong adult trauma patients with ISS >15. These results should be taken note when performing future audit or benchmarking exercises.
