References for this Review were identified through searches of PubMed, by use of the search terms “traumatic brain injury” or “head injury” and other appropriate terms, such as “prognosis” and “prognostic models”, up to January, 2010. Papers were also identified from the authors' own files and from references cited in relevant articles. An electronic search of other resources, such as book chapters, was also done. We considered only publications written in English and Dutch. The final
ReviewEarly prognosis in traumatic brain injury: from prophecies to predictions
Introduction
Prognosis is the cornerstone of clinical medicine, because all diagnostic and therapeutic actions aim to improve a patient's prognosis and outcome. Advances in statistical modelling and the availability of large databases have made it possible to consider diagnosis and prognosis in terms of probabilities rather than vague prophecies. Probability estimates can be applied to clinical decision making, research, and assessment of the quality of health care. Such quantitative estimates are of particular relevance to heterogeneous conditions such as traumatic brain injury (TBI).
TBI poses a major public-health problem, with an estimated annual incidence of up to 500 per 100 000 population and more than 200 hospital admissions per 100 000 admissions in Europe each year.1, 2 TBI is heterogeneous in terms of cause, pathology, severity, and prognosis, which poses diagnostic challenges. Furthermore, comparison of results between studies is difficult because case mix and treatments may vary substantially.
Various outcomes can be considered in prediction research. A diagnostic perspective is taken in TBI studies, and involves assessment of the probability of structural brain damage or developing an intracranial haematoma, or is used to underpin recommendations for CT scanning.3, 4, 5 For example, a recent study used a prediction rule to identify a subset of children who had such low risk for intracranial damage that CT scans were unnecessary.6 These types of diagnostic outcomes are particularly relevant for patients with mild TBI. The ability to predict response to treatment would be highly relevant to patients in the intensive-care setting, in whom intracranial pressure is monitored, but such prognostic rules have not yet been developed. For patients with moderate and severe TBI, prediction of clinical outcome is also highly relevant. Typically, most studies have defined clinical outcome as mortality or functional outcome assessed with the Glasgow outcome scale (GOS) as their endpoint.7
In this Review, we focus on the prediction of outcome in terms of mortality and functional outcome in patients with moderate and severe TBI. We aim to describe the basics of prognostic analysis and review the current knowledge about traditional and newly recognised predictors for outcome in TBI. We also discuss prognostic modelling as a novel instrument in medicine, critically review prediction models in TBI, describe the applications for prognostic models in TBI, and provide suggestions for the further development and improvement of prediction research in TBI. We will use the term “outcome” to refer to all endpoints from different studies that use mortality and GOS.
Section snippets
Predictors of outcome
Much research has been done to identify early predictors of mortality and functional outcome, as assessed by the GOS on admission, after moderate or severe TBI. The GOS is usually dichotomised into good recovery and mild disability versus severe disability, vegetative state, and mortality. This is a limitation because we cannot assume that predictors differentiate death from survival as well as they can differentiate good recovery from worse outcomes.
A large body of evidence supports the
Prognostic models
Estimation of prognosis is by definition a multivariable challenge. Predictors should be considered jointly rather than on their own, and can be combined in a multivariable prognostic model to quantify the risk for a particular outcome in individual patients. Combining individual predictors into a model will increase performance and generalisability, and is important because patients could have characteristics that affect the outcome in opposite directions. For example, for a 24-year-old
Clinical practice
Some estimation of prognosis is consciously or subconsciously used by physicians when informing relatives, making treatment decisions, or allocating resources. Estimates derived from large datasets are preferable to the subjective opinion of a physician, whose experience, no matter how vast, can never match the information contained in data from thousands of patients. The Canadian CT rule and the CHIP (CT in head injury patients) prediction rule for CT scanning in mild TBI are examples of how
Conclusions and future directions
Prognostic analysis and modelling have great potential in TBI, both for diagnosis and prognosis. Although some of the gaps in our knowledge have been identified, some issues require further investigation. Validated prognostic models have been based mainly on admission characteristics. Although substantial insight has been gained into the prognostic value of variables obtained during the subsequent clinical course, such variables have not yet been widely included in prognostic models. Further
Search strategy and selection criteria
References (128)
- et al.
The Canadian CT Head Rule for patients with minor head injury
Lancet
(2001) - et al.
Identification of children at very low risk of clinically-important brain injuries after head trauma: a prospective cohort study
Lancet
(2009) - et al.
Assessment of coma and impaired consciousness. A practical scale
Lancet
(1974) - et al.
Outcome prediction in severe head injury: analyses of clinical prognostic factors
J Clin Neurosci
(2001) - et al.
Pedestrians injured by automobiles: relationship of age to injury type and severity
J Am Coll Surg
(2004) - et al.
Talked and deteriorated head injury patients: how many poor outcomes can be avoided?
J Clin Neurosci
(2002) - et al.
Predictors of in-hospital mortality and 6-month functional outcomes in older adults after moderate to severe traumatic brain injury
Injury
(2009) - et al.
The relationship between blood glucose, mean arterial pressure and outcome after severe head injury: an observational study
Injury
(2002) S100B protein: a novel biomarker for the diagnosis of head injury [in French]
Ann Pharm Fr
(2009)- et al.
Predicting outcome after severe traumatic brain injury using the serum S100B biomarker: results using a single (24h) time-point
Resuscitation
(2009)