Development of a standardized minimum dataset for including low‐severity trauma patients in trauma registry collections in Australia and Aotearoa New Zealand

Trauma continues to place a burden on individuals, communities and health care systems around the world. To help reduce this burden and improve care, trauma registries in Australia and Aotearoa New Zealand collect standardized data on patients admitted with Injury Severity Scores greater than 12. There is currently no agreed minimum data set for trauma patients with Injury Severity Score less than 13, representing an opportunity to provide more data for quality improvement and injury prevention.


Introduction
Trauma represents a major health problem around the world, including in Australia and Aotearoa New Zealand, and results in enormous loss and suffering among patients and families. [1][2][3] Monitoring the incidence and care of injured patients using data collected in trauma registries has demonstrated benefits in reducing trauma-related morbidity and mortality. 4 However, most trauma registries in Australia and Aotearoa New Zealand collect data on severely injured patients only. This is despite low-severity trauma (irrespective of definition) accounting for a large portion of the societal burden of injury, 5 comprising approximately 90% of all trauma admissions 6 and being responsible for two thirds of total trauma-related hospital expenditure. 7 In an effort to provide a standardized evidence base for informing quality improvement in trauma care and targeting initiatives for reducing the incidence of trauma, representatives from hospital and state-based trauma registries throughout Australia and Aotearoa New Zealand came together to develop the Binational Trauma Minimum Dataset (BNTMDS) in 2007. 8 Since that time, the BNTMDS has been refined and maintained to ensure only fields which are useful, collectible, and relevant to a wide range of patients are included. 8 The BNMTDS is currently used within the combined Australia and Aotearoa New Zealand Trauma Registry (ATR) 9 and the New Zealand Trauma Registry (NZTR). 10 The ATR currently collates data from major trauma service hospitals across the two countries, either directly or via regional registries, for major trauma patients who die or sustain injuries with an Injury Severity Score (ISS) 11 > 12 12 using the 2008 revision of the Abbreviated Injury Scale (AIS). 13 In general, these data for major trauma patients are currently collected manually from patient records by dedicated nursing/data collection staff within the treating facilities. Despite the likely benefits that would result from the inclusion of low-severity trauma data in trauma registry collections, 6,14 the substantial cost of dedicated personnel to process large volumes of low-severity trauma data is a deterrent to progress. The exclusion of low-severity trauma data represents an obvious gap in the information available for evidence-based trauma quality improvement (TQI) and injury prevention initiatives, and raises issues of equity given the aetiology, demography and injury patterns are distinctly different between the two injury severity groups. 7 Recent work has demonstrated the utility of using electronic administrative sources for addressing this gap. Following enhancements to information and digital infrastructure in Western Australia (WA), a pilot project undertaken for WA State Trauma Registry has demonstrated the feasibility of delivering an automated trauma data collection and management solution for non-major trauma. 15 In addition to ensuring low-severity trauma patients are represented in trauma data collections, the use of administrative data sources also minimizes the significant staff resources required for manual data collection and handling. In addition, this approach may also lend itself to the possibility of streamlining the extraction and processing of major trauma data, at least in part. It is believed that a standardized, well-considered, extractable low-severity minimum dataset could provide a focus for facilities and jurisdictions to extend their evidence bases for TQI activities and injury prevention at local, regional, or national levels for minimal cost.
In this paper we aim to: (1) Propose a standardized, low-severity trauma minimum dataset (low-severity trauma MDS), using examples from existing registries where low-severity trauma data have been collected. To maximize the utility and comparability of the MDS, we determined pre-hoc to employ a subset of the wellestablished fields currently endorsed for major trauma (i.e., the BNTMDS 9 ). (2) Determine the potential for collecting these fields via existing electronic administrative sources in Australia and Aotearoa New Zealand.

Methods
A binational, multidisciplinary expert group with broad practical experience in areas such as Trauma Quality Improvement (TQI), the development of the BNTMDS, low-severity trauma data collection and utilization, and the use of administrative trauma data was assembled under the auspices of the Australia and Aotearoa New Zealand College of Surgeons Trauma Quality Improvement Sub-Committee (RACS TQI-SC). Members were selected by the committee based on their depth and breadth of experience in designing and using trauma databases for research and quality improvement activities. These included ATR, former Queensland Trauma Registry, NSW Trauma Registry, Western Australian State Trauma Registry, Midland Trauma Registry and New Zealand Major Trauma Registry. Members were involved in research through the Jamieson Trauma Institute, Te Manawa Taki Trauma Research Centre, Institute of Trauma and Injury Management (ITIM), National Trauma Research Institute (NTRI). Minimum datasets for major and low-severity trauma from both countries were compared, and fields were scrutinized. Alignment with administrative health databases, included those capturing data from emergency departments, ambulance services, retrieval services, and hospital admitted patients at the local, state, and national level were considered. Fields were considered eligible for administrative extractions if they could be obtained directly from, or derived from, existing data sources (either at a local hospital level, or through linkage with statistical health data collections held centrally). Informal evaluation of existing health databases across the Australian states and Aotearoa New Zealand was conducted to determine the potential for capturing the proposed low-severity trauma MDS using existing electronic health databases. The term 'low-severity' refers to a patient's injury severity score. This sets a clear threshold for inclusion into a low-severity databases and avoids the potentially negative inference for patients from the term 'non-major', that their injuries are not important or result in minimal impact on their well-being. In addition, the term low severity is a generic term for potential use in registries and other databases outside of Australia and Aotearoa New Zealand. The trauma registries reviewed for their capture of low severity trauma patients and alignments with the BNTMDS included the Te Manawa Taki (formerly Midland) Trauma Registry (TMTTR), 16 the Western Australia (WA) State Trauma Registry, 17 the Royal Children's Hospital Melbourne (RCH) Trauma Registry 18 and the Queensland Trauma Registry (QTR). 6 The TMTTR, WA Trauma Registry and RCH Trauma Registry represent regional, State, and single site registries, respectively, and currently include all trauma patients, regardless of severity, admitted to hospitals included in the registry, or who die in their emergency department, following injury. For the WA Trauma Registry, this admission must be for 24 hours or more to be included. Funding for the State-based Queensland Trauma Registry ceased in 2012, and as such, the historical perspective of the QTR is considered in this project. Prior to funding cessation, the QTR included all trauma patients admitted to participating hospitals for 24 h or more, or died in their emergency department, following injury. These trauma registries form a convenience sample; there may be other trauma registries where low severity trauma patients are also specifically included.
The criteria for identification of suitable fields to include in the low-severity trauma MDS were: (1) Included in the BNTMDS (2) Associated with high levels of completeness and accuracy within the reviewed trauma registries (3) Used routinely for TQI and/or injury prevention activities in low-severity trauma patients within the reviewed trauma registries (4) Preferably able to be extracted or derived from administrative health data (either at a local hospital level, or through linkage with statistical data collections held centrally), thus reducing manual data entry (5) Inclusive of time-stamping for process evaluation There were several reasons for excluding fields from consideration. This included fields which were: (1) Not routinely used in TQI or injury prevention activities.
(2) Related to pre-hospital clinical management and patient physiology (due to a frequent lack of integration with ambulance service data). (3) Related to in-hospital physiology, given these data are not routinely available to extract electronically from administrative systems The TMTTR was used as the reference point for initial lowseverity field selection and field descriptions. The fields in the TMTTR were firstly mapped to a corresponding BNMTDS field, where applicable. The WA Trauma Registry, RCH Trauma Registry and QTR datasets were then collated and matched. Matched fields were assessed against the above inclusion and exclusion criteria to determine their value in TQI or injury prevention. Any variations in the assessments of field validity for inclusion or exclusion were discussed in detail until consensus was reached. This process continued until all members of the group were satisfied with the appropriateness of the final list of fields.

Results
Comparison of the reviewed registries, and the alignment to the BNMTDS, is shown in Table S1. There were 35 data fields recommended by the study group for inclusion in a low-severity trauma MDS; all of which were collected by the TMTTR. The WA State Trauma Registry currently collects 31 of the 35, the RCH currently collects 33 of the 35, and QTR collected 31 of the 35 proposed data fields. Table S1 also shows whether the data field can likely be extracted from electronic administrative data systems.
Overall, there were only two fields that are not currently included in administrative systems: Abbreviated Injury Scale (AIS) codes and Injury Severity Score (ISS). Existing trauma registries utilize manual AIS coding and derived ISS. AIS coding is time-and staffintensive, and hence is likely to be unfeasible for the purposes of low-severity trauma collection.
However, other injury severity classification systems which use International Classification of Disease (ICD)-coded diagnostic information have been developed and may be considered as an alternative. [19][20][21] Tools that use direct conversion between ICD-10 and AIS have also been developed, 22,23 although their utility is under review. [24][25][26] Two of the proposed fields; triage category (as recorded using the Australasian Triage Scale) and the total cost of inpatient treatment are not currently active in the BNTMDS. However, both are obtainable from electronic administrative sources (either at a local hospital level, or through linkage with statistical health data collections held centrally), and are believed to be relevant to process improvement and hospital burden estimates in low-severity trauma patients.

Discussion
This study has defined a practical MDS for low-severity trauma for use within Australia and Aotearoa New Zealand. It may also be applicable to other trauma systems worldwide. It is believed that collection of data using this template should be achievable within existing administrative resources (whether at a local hospital level, or through linkage with statistical health data collections held centrally within health departments) in Australia and Aotearoa New Zealand. This opens opportunities for health authorities or facilities wishing to address issues related to low-severity trauma patients to extract and utilize these data in a cost-effective manner.
Critical requirements of an administrative data extraction method are that the data are complete and accurate, and that data can be obtained without incurring significant costs in staff time or other resources. A trial extraction of patient data using this newlydeveloped tool was outside the scope of this study, but would be an essential exercise prior to widespread adoptions outside the existing systems that collect and use low-severity trauma data across Australia and Aotearoa New Zealand.
The evaluation of existing state and national health databases showed that the low-severity trauma MDS fields were contained within these systems. However, it was beyond the scope of this study to determine the completeness and accuracy of data, or constraints to data supplies, that would impact on the attainment process.
Whilst administrative data extraction may seem parsimonious, manual data collection by clinical staff may provide significant other benefits to patients that may not be possible with administrative-only data extraction. These benefits include information validation and other quality improvement opportunities provided by direct contact with both patients and the health care professionals involved in their care. These benefits should be carefully considered when choosing between administrative and manual data collection when the options for manual collection for both major and low-severity trauma data exists.
Although electronic medical records (EMR) are now in widespread use across Australia and Aotearoa New Zealand 27 and elsewhere 28 for capturing patient information, challenges may arise with the secondary use of these data. A systematic review by Edmonson and Reimer 29 found that the four most common limitations of EMRs in the literature were data quality issues (91.7%), data preprocessing challenges (53.3%), privacy concerns (18.3%) and potential limitations on generalizability (21.7%). When considering administrative data extractions, these potential limitations must be considered. In addition, the purpose of trauma registry and administrative data collections differ, and hence must also be understood and acknowledged.

Conclusion
This paper has proposed a practical MDS for low-severity trauma and determined the potential for using administrative collection methods to extract the MDS from existing health databases in Australian and Aotearoa New Zealand. Questions remain regarding the quality of administrative data and cost-to-benefit comparison between manual and administrative data collection that will ultimately be considered by jurisdictions and facilities that are considering the collection of low-severity MDS. The breadth and depth of available electronic data in an increasingly digital healthcare system as well as the technological advancements in data linkage in recent times offer significant potential to substantially increase the sources of rich clinical data able to be linked routinely to inform trauma care quality improvement activities and to supplement trauma registry data to provide a more comprehensive understanding of the burden of trauma in Australia and Aotearoa New Zealand.