Latest News
November 25, 2015
C4TS lead Professor Karim Brohi discusses the challenges of developing a national trauma registry in the latest edition of the Emergency Medicine Journal. His commentary is reproduced below:
"Data and research sit at the heart of public health models of healthcare systems. It is not an exaggeration to state that no meaningful change is possible without the underlying knowledge that change is first necessary, and subsequent confirmation that change was effective. Only in this way can health systems bootstrap themselves up to effective high quality care. Trauma registries have therefore been at the heart of trauma systems since the concept was first introduced nearly half a century ago.
The Trauma Audit and Research Network (TARN) is now the central trauma registry for England and Wales and is deeply embedded in the structure and functioning of the regional major trauma centres and trauma networks. TARN arguably represents the most integrated, responsive and mission-critical trauma registry in the world. However, being a national trauma registry comes with many challenges, demands and limitations. Providers and users of the system must appreciate and manage these aspects of TARN for maximal effective use of the data and for the development of TARN itself.
Mission creep is the scourge of trauma registries. At their core, trauma registries exist to inform the trauma centre or trauma system's performance improvement programme. However, over time, others see the potential utility of the data and the users and stakeholders expand, each placing slightly different demands on the registry. Thus audit leads to research, research leads to health service delivery, delivery leads to commissioning, and all roads lead to finance. These demands exist at multiple levels—individuals and departments within an institution, the trauma centre itself, the regional network and supra-regional or national programmes. Each user group would like a few more fields added to the registry, not appreciating the additive cost and effort in collecting each single data point. It is conceptually a lot easier to add a field to a registry than to collect the data separately for a defined study, for a defined period—yet additively, year on year, it can lead to unwieldy databases that become difficult to manage and interrogate.
TARN started out as a local audit tool and grew appropriately into a national audit and quality assurance programme. However, it is also now part of multiple research programmes, including its own; it is both a commissioning tool and a financial reimbursement vehicle, and to a certain extent it is a political tool used to lever system change. Each of these has resource demands and requires trade-offs to be made in terms of both overall strategic focus and of daily activity. The TARN management and board have so far worked to carefully manage these competing demands, and need to ensure that their core mission in supporting trauma quality assurance and performance improvement remains at the heart of their activities.
Trauma registries are essentially a continuous prospective cohort of the trauma patient population, capturing demographics, injury details, process measures and outcome data. However, not all injured patients can be captured within a registry. Trade-offs have to be made between the number of patients in the database, the number of data points to be captured on each patient and the intended use of the data. The role of a trauma registry is therefore not the same as an injury surveillance tool, which might capture a small amount of information on all injuries (including cuts and sprains) at a national level. The TARN trauma registry has always existed to examine the care and outcomes of patients who have sustained a potentially life changing event—and whose outcome may be affected by the performance of the institution or system in which they are cared. TARN has therefore focused data collection on severely injured patients—albeit a group that has always been difficult to define.
From the outset, TARN decided that their target patient population would be all significantly injured patients and they would identify these by capturing all patients admitted to hospital for at least 72 h as well as all those who died at any time. In the absence of a method of identifying the severely injured patient on arrival, this was a pragmatic solution to a difficult problem. However, there are issues with using a process measure to define injury severity—not least of which is that, as systems of care improve, patients who previously may have stayed longer than 72 h are now discharged earlier, and are no longer captured in the data. For example, the management of pneumothorax and haemothorax has improved such that now many patients are discharged within the 72 h window whereas previously they may have spent 5–7 days in hospital. The performance improvement in this group may have been missed. Importantly, the baseline population of the registry has now changed, and will likely continue to change as hospital stays for injured patients reduce further.
However, feasible alternatives to this method are not obviously apparent. Including all trauma team activations or all admitted patients would potentially more than double the workload without clear benefit. Including patients retrospectively with a given injury severity score increases the risk of not capturing all patients and potentially misses less anatomically injured patients who may benefit (or otherwise) from a major trauma system. Shifting definitions of what actually constitutes major trauma also makes such decisions difficult, and any change makes subsequent historical comparisons very difficult. TARN, as with many registries, is therefore effectively tied to the initial decision of which patients to include, and must work within these constraints to understand their data and its implications.
The growth in the uptake of TARN has also generated challenges with the volume of data that must be handled and also with its quality and comparability. Initially the few hospitals that contributed to TARN were self-selected as interested in trauma and clinical audit, and therefore were likely to represent units with better processes and outcomes. As subscription for major trauma units became mandatory, less mature trauma centres joined and will likely have reduced the baseline level of care, making retrospective comparisons difficult. Even the techniques of patient identification and data acquisition for the purposes of TARN entry mean that a new hospital joining TARN is likely to initially show skewed performance metrics. These problems are compounded as all trauma units have been required to submit full TARN datasets. With low patient volumes and no funding support, these units tend to submit much lower levels of data completeness and quality than the major trauma centres, making pan-regional assessments of injury care near impossible. With all these caveats in mind, in reality the real power of TARN is only just about to be realised. TARN is the only national registry to which every trauma receiving unit in the country submits data. With a stable set of submitting institutions and improved data quality and completeness, the potential for true national population-based quality assurance and improvement is tangibly within reach.
One area in which TARN has not managed to make much progress over the last quarter of a century is its selection of outcome measures used to evaluate trauma care. TARN, as with all trauma registries worldwide, remains at its heart a mortality audit. Undoubtedly mortality is important, and remains a strong global indicator of quality of care. TARN has almost exclusively presented mortality through its risk-adjusted Probability of Survival model. While the baseline methodology for this is undoubtedly sound, TARN has not been strong at publishing its detailed methodologies or validations as it has updated and tested its models. Furthermore, while risk-adjusted outputs enable between-unit comparisons allowing for case mix variations, their outputs are not easy to interpret for a performance improvement programme, especially in the absence of the baseline inputs and the associated crude mortality rates. There is also a clear desire to move beyond mortality to assess health outcomes among survivors. However, efforts to produce models that predict outcomes such as length of stay have been thwarted by complexity and non-clinical aspects of these outcomes. Conversely, outcomes such as quality of life and patient-reported measures require orders of magnitude more effort and resource to capture meaningful data. While efforts to effectively incorporate other outcomes into TARN are progressing as part of research and commissioning enterprises, in reality mortality is likely to remain TARN's core outcome benchmark for the foreseeable future.
None of these issues is unique to England and Wales, nor to TARN. Delivering a useful responsive trauma registry for a single hospital is challenging. Delivering one for a nation—for clinicians, managers, commissioners, financiers and for the public—is near impossible. Yet, in 25 years, TARN has moved from a small academic enterprise of the interested to a national clinical quality assurance tool, a health services research engine and a benchmarking tool for commissioners. At all times TARN has had to weave a careful line between too few data and too much data; responsiveness and data quality; desires and resources; and perfection and pragmatism. Throughout they have always proceeded with a spirit of inclusiveness, integrity, good humour and the fundamental desire to improve the care and outcomes of trauma patients. Without TARN there would be no national trauma system, and without TARN there can be no assurance of the continued quality and improvement of national trauma care."
