Method
Research Design and Setting
This was an observational study using data from the Queensland Trauma Registry (QTR). QTR was established in 1998 and collected data on seriously injured people in the state of Queensland, Australia. Queensland is situated in the north-east of Australia and has a population of approximately 4.5 million people. In 2005, 14 regional and tertiary public hospitals participated in the QTR, increasing to 20 hospitals in 2009. These hospitals are estimated to account for more than 90% of seriously injured patients admitted to Queensland public hospitals.
With some exclusions (eg, iatrogenic injuries and pathological fractures), patients were included on the QTR if they were directly admitted to, or transferred for admission to, a participating hospital for 24 h or more for the acute treatment of injury and were coded with an ICD-10-AM (the Australian modification of the International Statistical Classification of Diseases) code indicating trauma to single or multiple body regions. Patients who died after ED presentation (regardless of length of admission) also were included on the registry but patients who died before reaching hospital were not included. Cases were identified for potential inclusion on the QTR via a standard system report generated by the Emergency Department Information System which was cross-matched with hospital coded morbidity data to ensure all eligible injury patients were identified. Data were abstracted from the medical record for all eligible patients and manually entered on the database by QTR nurses trained and accredited in specialised injury coding, including the Abbreviated Injury Scale. Further details on data capture, collection and quality assurance methodologies for the QTR are available elsewhere.
For this study, all patients aged 15 years and older who met the QTR inclusion criteria between 1 January 2005 and 31 December 2009 with an Injury Severity Score (ISS) greater than 15 were included. Ethics approval for this study was granted by Queensland Health and The University of Queensland.
Methods of Measurement
The ATS is used across Australasia to assess the urgency of treating a patient according to the following question: 'This patient should wait for medical care not longer than…?'. The instrument consists of a 5-point ordinal scale ranging from 1 to 5 (Table 1). ATS category 1 is assigned to patients with the highest clinical urgency necessitating immediate treatment.
To compare the triage of patients across different age groups, the ATS score assigned to the patient at the first ED of presentation was used. Thus, for patients who were transferred from a regional referring hospital to a second hospital for definitive care, the ATS score from their regional referring hospital was used in this study.
In addition to the first ATS score, the following information was extracted from the QTR database for each case to enable adjustments in multivariable analyses: demographic details, injury characteristics, injury severity and acute care factors. Demographic details included age and sex, and injury characteristics included cause of injury (transport crash, animal-related, fall, striking or other), nature of main injury (fracture, injury to nerve/vessel/muscle/tendon, injury to internal organ, intracranial injury or other), intent of injury (accidental, assault or other) and injury type (blunt or penetrating).
Injury severity was assessed using the ISS and the Revised Trauma Score (RTS). Acute care factors included level of definitive care hospital (tertiary, large regional, small regional), mode of arrival to definitive care hospital (ambulance (fixed wing), ambulance (helicopter), ambulance (road), own transport or other) and transfer from another hospital (yes/no).
Data Analysis
Data were analysed using IBM SPSS Statistics V.19 (SPSS Inc, Chicago, Illinois, USA). Descriptive statistics were used to compare the injury characteristics and triage categories across various age groups. Patients with missing data were excluded from the analyses. Logistic regression analyses were performed to determine whether age was a predictor of triage category after controlling for sex, ISS, injury characteristics and acute care factors. These potential confounding variables were selected for the model because of their clinical relevance in terms of determining triage category. All these variables displayed significant univariate associations with triage category. ATS was dichotomised as ATS 1 or 2 (Emergency) versus ATS 3–5 (Urgent) for the purpose of analysis. This breakdown was chosen because the triage guidelines indicate that major trauma should be categorised as ATS 1 or 2 and this categorisation reflects the difference between patients requiring emergency rather than urgent treatment. Age was divided into four categories; 15–34, 35–54, 55–74 and ≥75 years. ISS was entered as a dichotomous variable (ISS 16–24 vs ≥25). The cut-off of 25 was chosen because it was clinically meaningful in terms of increased mortality risk and provided approximately equivalent numbers in each stratum. RTS also was entered as a dichotomous variable owing to the non-linear effect of RTS on ATS and the high proportion of patients with an RTS of 7.8408. The categorisation reflected normal (7.8408) versus abnormal (<7.8408) physiology. All categorisations were determined a priori. A number of variables in the final logistic regression model assessed similar constructs (injury characteristics and injury severity). Therefore, multicollinearity was assessed using the collinearity diagnostics in SPSS, which found that in all instances, the tolerance was above 0.3 and the variance inflation factor below 3.5 (values less than 0.1 and greater than 10 respectively indicate issues with multicollinearity). Therefore, the collinearity in the data was deemed acceptable.