Multiple reports over the past 20 years have criticised the care delivered to major trauma patients in England. The setting up of regional trauma networks is well underway to address these deficiencies. Outcomes currently collected in major trauma are restricted to mortality rates and process measures; however, many patients are left with long-term disabilities and reduced quality of life (QoL), which are not measured at present. Trauma systems need to develop outcome measures that capture this important information to judge the quality of care and influence the development of services. This review aimed to identify the current practice of international trauma registries with regard to outcome measurement and review the literature surrounding these measures. An internet-based search of international trauma registry websites including data dictionaries and annual reports was carried out as well as a literature review to determine what outcome measures are routinely collected by trauma registries. All registries were found to look at mortality and process measures, but only one registry routinely collected morbidity outcome measures for both adults and children. There is a considerable amount of disability and reduced QoL in survivors of major trauma that lasts for years following the injury. At present only the Victorian State Trauma Registry collects outcome measures that evaluate the extent of this. English regional trauma networks and the Trauma Audit and Research Network have the opportunity to learn from this experience and routinely start to collect measures of outcome that can drive service improvement and reduce patient morbidity.
- Outcome assessment (health care)
- treatment outcome
- wounds and injuries
Statistics from Altmetric.com
The setting up of regional trauma networks is well underway in England, with four networks already running in London, and more networks due to come online throughout the nation between April 2011 and April 2012. One of the recommendations of the recent National Audit Office report into major trauma care in England1 was the development of outcome measures for major trauma patients to assess the quality of care provided to these patients and combat the inequities of the current system. Addressing this issue is difficult and is a problem not just in England, but internationally.
Historically, much of the literature into major trauma refers to outcomes and mortality synonymously. There is a growing recognition that as trauma systems mature and more patients' lives are saved, the focus has to shift from amenable mortality to improving long-term patient morbidity. This need is highlighted by the fact that following major trauma one study found full recovery in only 30% at 5 years,2 and 42% were unable to return to work in another.3
With the anticipated and long overdue changes in major trauma care beginning to take shape, we have reviewed the literature and current practice of international trauma registries regarding the use of outcome measures for major trauma patients. As all English regional trauma networks will need to monitor and report their outcomes, the development of evidence-based outcome measures for major trauma patients is a high priority. The importance of outcome measurement has been recognised at a national level. The NHS outcomes framework has been designed and will be used to hold the NHS commissioning board accountable for the outcomes of the services it commissions.4 This framework highlights the importance of helping people to recover from injury and the need for improvement in recording outcomes in trauma; however, to date no robust indicator for major trauma has been identified.
The main internet-based trauma registry search aimed to identify the current practice of trauma registries in North America, Europe and Australasia where trauma systems are well established.
A literature search using PubMed and MEDLINE was also used to identify any other countries with existing trauma registries and any outcome measures they might collect.
The internet search initially used Google and search terms including variations on ‘trauma’, ‘network’, ‘registry’, ‘audit’ and the addition of country/region or state/province names to narrow down search results. Countries/regions selected for the search included Australia, New Zealand, USA, Canada, UK, Germany, The Netherlands and Scandinavia. States/provinces within these countries were also searched for to obtain regional/local trauma registry details. Websites of professional associations were also searched to ensure that all registries with a web presence were discovered. Data dictionaries were examined for any outcome measures that were collected routinely or were optional data fields.
From the literature search it was evident that several countries either have trauma registries or are in the development stages of establishing them. It was difficult to ascertain precisely which outcome measures were recorded from the literature alone, for this reason the internet-based search was used predominantly. Countries such as Hong Kong have a trauma registry that collects some outcome data;5 most, however, still collect information on processes, procedure or injury type such as Pakistan6 and the United Arab Emirates,7 or are in development, such as Greece.8
The internet-based trauma registry search found six national trauma registries and 12 state or provincial trauma registries. The full results for adult outcome measures are provided in table 1. It is clear from the search that trauma registries collect risk-adjusted mortality rates and large amounts of quality indicators such as length of stay, but most do not collect functional or quality of life (QoL) outcome measures. The Glasgow outcome scale (GOS) recorded at discharge is included in the data dictionaries of several trauma registries as it was included in the Utstein trauma template,9 but it is not routinely collected. Only the Victorian State Trauma Registry (VSTR) collects outcome measurement data routinely.
Selecting an outcome measure
As already mentioned, the nascent English trauma networks must develop outcome measures by which they can monitor the recovery of their patients and performance of their systems, but there is no obvious path forward.
There is no shortage of potential outcome measures in trauma, with one review10 identifying 27 different measurement tools used to identify outcome in trauma patients. Outcome measures need to be valid, reliable and responsive to change. Once these characteristics have been shown, the choice of instrument is down to the practicality of implementation. Outcome following major trauma is not an easy-to-measure variable and instruments cannot be compared against a gold standard (criterion validity), instead construct or face and content validity are required to be assessed and these require a cumulative pattern of evidence.
Specific difficulties with assessing outcomes other than mortality are recognised in trauma registries.11 Major trauma may include a wide range of organ and body region injuries, from brain to bladder, which can be challenging for any unified measure. When such measures are derived in a research environment the cost of collecting such data routinely in a registry may be inhibitory and logistically challenging, especially when phone or face-to-face interview time is required. A further barrier to the cost effectiveness of collecting such outcomes is the absence of a risk-adjustment model for long-term disability measures in general trauma patients. This currently limits the ability of the outcome measures to drive performance improvement. Introducing them into the performance management of trauma systems, without risk adjustment, could lead to perverse incentives and ‘cherry picking’ of patients.
In addition, trauma patients are generally younger and can easily be lost to follow-up. They may be injured away from their resident locality, and they tend to be more mobile post-injury in terms of their place of work and home. There may also be a tension between the needs of the registry, the funding organisation and the patient in terms of what data should be collected; for example, patients may be most interested in pain, disability and function, whereas the health authorities may be more concerned with the efficacy of care and population-level measures of return to work and social dependency.
Outcome measures were classified in to five levels by Wilson and Cleary:12 (1) biological and physical; (2) symptom status; (3) functional status; (4) general health perceptions; (5) overall QoL. There was no ascribed hierarchy of importance within these. For major trauma registry outcomes, measuring two of these levels has been advocated.13 Functional status and QoL would result in the most comprehensive description of patient outcomes. However, although functional status can be measured for all patients (by proxy if necessary), QoL is a measure that can only be collected directly from the patient and so may not be obtainable from a significant subset with serious brain injuries.
Functional outcome measures
Although many functional outcome measures have been proposed for use in major trauma, few have been collected on a large scale, and only recently are some registries moving towards collecting data on functional status. Two measures in the literature are of interest: the functional independence measure (FIM) and Glasgow outcome score.
FIM and modified FIM
The FIM was developed for use in the inpatient rehabilitation setting to measure function on admission and discharge for patients with all diagnoses. Its items only relate to basic function and some activities of daily living. It does not measure any activity or participation elements such as social interaction, return to work or study, or recreational activities. It has gained a wide usage in the measurement of rehabilitation outcomes, for example, the UK Specialist Rehabilitation Outcomes Collaborative use it to monitor outcomes in specialist neurorehabilitation (L. Turner-Stokes, 2010, personal communication). A number of studies have used the FIM to monitor functional limitations at different time points following trauma,14 15 and have found it reliable and valid for trauma. However, there are significant barriers to its routine application as it requires training and regular re-accreditation; there are also significant time implications as it takes approximately 20 minutes to complete.
The American College of Surgeons has implemented a short modified version of the FIM as part of the core dataset for the national trauma databank to measure functional status at discharge. Although attractive as a measure because of its brevity, it has not been found reliable or consistent when compared with FIM;16 it has also not been established as reliable or valid as a measurement tool for long-term outcomes.3
GOS and GOS-E
The GOS is the most widely used functional outcome following brain injury. It allocates patients into five broad categories of outcome using a structured interview, and has been found to be highly reliable for scoring and valid for these groups. It has also been found to function as a general measure of disability, but it has been criticised for not recognising subtle changes in the upper levels of function. This has been addressed by the Glasgow outcome scale extended (GOS-E), which that subdivides the upper three categories of the scale.17 Despite advocation of the use of the GOS as an outcome tool for all trauma patients irrespective of head injury status,18 there is limited evidence for its use in this setting.
A recent validation study from Victoria, Australia,19 compared the use of different functional outcome measures following major trauma including both brain-injured and non-brain-injured patients. It found that the GOS-E was the most responsive measure and had the lowest ceiling effect when compared with the GOS, FIM and modified FIM.
Measuring patient-reported QoL is of increasing interest to healthcare services and clinicians. Functional measures alone may not explain patient difficulties in return to work or social participation, and often QoL measures do not correlate with physical injury severity; there may be important differences between social groups and genders; for example, women are at higher risk of poorer QoL outcomes after major trauma than men.20
Short form 36 and 12
Short form 36 (SF-36) was developed as a multipurpose survey of general health status and has been validated as being responsive to changes in health status over time among unintentionally injured adults.21 It has been recommended by a consensus conference as a generic QoL tool for all trauma patients.18 Further studies have looked at the performance of the SF-36 in different injury domains such as multiple trauma with head injury22 and spinal cord injury.23
SF-36 scores at 3 months in adult patients who have sustained trauma remain significantly below population means,24 and studies have shown that although SF-36 scores progressively move towards the norms in all subscales over a 6–12-month period, they are still significantly worse at 12 months25 than preinjury. This deficit persists for at least 5 years post-injury.26 Work status, general health and satisfaction correlated with mental health status independent of the severity of the injury or the physical recovery achieved emphasising the importance of collecting QoL measures.
The EQ-5D evaluates peoples' current health status along five dimensions (mobility, self-care, participation in usual activities, pain/discomfort, anxiety/depression). A systematic review of the use of the EQ-5D in injuries found 44 studies,29 but only one looked at the long-term functional outcomes after major trauma. This30 and a further study31 from the same research group found that major trauma patients had EQ-5D scores significantly below population means at more than 12 months post-injury. A third of these patients were unable to work despite 77% of survivors having a maximum functional score (GOS of 5) again emphasising the importance of the use of both functional and health-related QoL scores.
Outcome measures in children
There are differences in the ways in which children and adults report on their health, and these must be taken into account when selecting outcome measures in this population. It can be difficult to assess whether change is due to the underlying condition or because of normal child development. The age of the child will create varying degrees of cognition and attention span so the appropriate language must be used for their age. The design of the questionnaire must also be considered in terms of length, graphics and so on. The child needs to be seen in a variety of social contexts including family, peers, school and the community, and the impact of their injury and subsequent morbidity accounted for in each. There is also the contentious issue of who completes the questionnaire in order to give the most accurate account of the child's health status with minimal bias.32
Kings outcome score of childhood head injury
The Kings outcome score of childhood head injury (KOSCHI) is a functional outcome measure; it is a paediatric adaptation of the adult GOS. It expands upon the GOS to ensure there is more variability at the less severe end of the spectrum of disability.33 The score can be carried out prospectively or retrospectively from medical notes, and its compatibility with the GOS allows for follow-up to continue into adult life. The authors initially designed the score for those between 2 and 16 years of age. There has been some concern over interobserver reliability, which can be largely allayed by the provision of training.33
The Peds-QL is a QoL assessment tool for use in children from 2 to 18 years of age. It involves a generic section and 30 disease-specific components. Health is measured across four domains: physical; emotional; social and school function and normative data are available. The Peds-QL can be self-reported or reported by parents and is quick to complete, taking approximately 5 minutes.34 The tool has not been evaluated in any depth in the paediatric trauma population.32
Current practice in trauma registries
Despite a consensus on measuring outcomes other than mortality there has been little movement internationally towards routinely collecting outcome data. A search of the websites of national and provincial/state trauma registries for Australasia, Europe and North America has revealed that only one registry, the VSTR, has implemented routine follow-up of all adult major trauma patients. Several registries including Vancouver coastal and Washington state have included functional outcome data as pilot projects or in their minimum datasets, but no others declare routine collection.
The VSTR collect outcome measures data by telephone questionnaire at 6, 12 and 24 months after injury,35 and use a variety of outcome measures including health-related QoL (SF-12 and EQ-5D), functional (GOS-E), pain and work disability. This has been implemented only for adult patients; however, the VSTR have started to follow-up paediatric major trauma cases routinely across the population using outcome measures that have been validated for children, both functional (KOSCHI) and QoL measures (Peds-QL).36
In Germany, the polytrauma outcome chart (POLO chart) has been developed. This takes into account the fact that patients sustaining multiple injuries will have sequelae for many years afterwards. It is a modular instrument to assess health-related QoL, which measures many trauma-related aspects of QoL and includes the GOS, EUROQOL, SF-36 and trauma outcome profile. This tool is still undergoing validation but may prove to be a useful tool for trauma patients.37
The use of outcome measures as a quality metric in trauma systems is an attractive one; however, there are difficulties with this. As discussed earlier, no QoL or functional outcome measures have had risk-adjustment models developed for general trauma patients. In the absence of this, these outcome measures will be limited in their ability to be used for system performance management because of the possibility of perverse incentives. These might include rehabilitation units choosing to treat less seriously injured patients preferentially as they have a better chance of functional recovery.
Studies have shown that the majority of major trauma survivors exhibit considerable levels of disability and health-related QoL impairment even at 1 year post-injury,38 with outcomes improving up to 24 months post-injury.39 However, outcomes often do not return to baseline and vary dependent on location. There is a need to collect these data to identify those who are not reaching their potential and focus improvement on those elements of the trauma care system. The ideal, logistically and clinically, would be a measure that, when collected at inpatient hospital discharge, would give an accurate prediction of the patient's future outcome. Unfortunately, current functional measures such as the FIM and GOS at discharge do not reliably predict longer-term outcomes including return to work at 6 months.3 Instituting some form of long-term reassessment therefore seem unavoidable.
Although many studies have looked at measuring outcomes following major trauma, they have generally used instruments that have not been developed specifically for the seriously injured patient, and there is little published regarding the validity, reliability and responsiveness of these instruments for this population. The work of the VSTR has started to change this and their methodology is currently being implemented in Singapore, and a number of US and Canadian sites have expressed interest (B. Gabbe, 2011, personal communication). The possibility of international comparisons through a consensus on outcomes is attractive, but more needs to be done in this area as the VSTR protocol may not have a wider validity.
The financial implications and cost effectiveness of routinely collecting these data may yet prove confounding. A pilot is planned for the NHS in England to establish whether a similar protocol over the 24 months following trauma is valid for measuring outcomes in the English major trauma population. If successful, such outcome measure collection could be implemented across England, which raises the possibility of international comparisons between trauma systems.
The authors acknowledge the help of Associate Professor Belinda Gabbe of Monash University, Australia and Professor Lynne Turner-Stokes of King's College, University of London, UK in the preparation of this article by providing details of their work relating to outcome measure collection in the fields of trauma and rehabilitation.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.