Statistics from Altmetric.com
QUESTION: In patients with severe head injury, or haematoma requiring evacuation, or coma lasting ≥6 hours, what is the reliability of outcome prediction using the Glasgow Head Injury Outcome Prediction Program?
Inception cohort followed up for 6 to 24 months. Calculated predictions were compared with actual outcome.
Department of Clinical Neurosciences, Western General Hospital, Edinburgh, UK.
324 admitted patients (median age 30 y) with head injury.
Assessment of prognostic factors
Using the Glasgow system, algorithms were used to predict outcome at 6 months. Data which can be used for prediction include age, Glasgow coma score (GCS), pupil response, computed tomography findings, temporal changes in GCS, motor response pattern, presence or absence of a lucid interval, type of haematoma requiring evacuation, and eye signs. Predictions were calculated on admission, before evacuation of a haematoma, or 24 hours, 3 days, and 7 days after onset of coma lasting ≥6 hours.
Main outcome measures
Actual patient outcome was assessed using the Glasgow outcome score after a minimum of 6 months up to a maximum of 24 months after the injury. The frequency with which predictions could not be calculated was determined, and calculated predicted outcomes were compared with actual outcome.
Of the 324 patients, 183 were potentially suitable for an admission prediction; 141 for a preoperative prediction, and 180, 145, and 124 suitable at 24 hours, 3, and 7 days, respectively. Overall, predictions could be calculated in 75% of potentially suitable patients on admission; 97% before haematoma evacuation; and 19%, 34%, and 53% of eligible patients at 24 hours, 3 days, and 7 days after the onset of coma, respectively. The use of sedation, paralysis, and intubation was the most common reason for inability to calculate a prediction. 76% of predictions were correct (the prediction with the highest probability corresponded to the actual outcome), 15% were pessimistic (the predicted outcome was worse than the actual outcome), and 10% were optimistic (the predicted outcome was better than the actual outcome). Of 197 patients (267 predictions) whose actual outcome was good recovery or moderate disability, 84% of predictions were correct. For an actual outcome of death or vegetative survival (96 patients, 110 predictions) 84% of predictions were correct, but for severe disability (31 patients, 49 predictions) only 12% were correctly predicted.
Using an independent data set, the utility of the Glasgow Head Injury Outcome Prediction Program compares favourably with other outcome prediction algorithms for patients with head injury.
The director of a frontline military neurosurgical unit faced with a flood of casualties and limited resources will find the predictive power of the algorithm used in this study by Nissen et al of value. By using this algorithm, the limited resources of the unit could be directed to those with the best chance of making a good recovery. But in the more usual medical settings it is less certain, given the high rate of falsely pessimistic predictions, how the predictions of the algorithm should be used to influence management.1, 2
The 18.5% of predictions in which the outcome group (dead or vegetative, severely disabled, or moderately disabled/good recovery) was predicted with a high probability (0.97) were more accurate. Of the 34 confident predictions of death or vegetative state, 28 were correct, 5 ended up severely disabled, and 1 achieved moderate/good recovery. But even these results compare unfavourably with the accurate predictions of poor outcome which have been achieved after anoxic brain injury.3
It was hopeless trying to predict the minority who survive with severe disability. But they are a clinically important group consuming huge resources over a lifetime of disablement. Identifying the factors which result in someone doing worse or better than expected has therefore big potential benefits. For the present, if and when the patient recovers consciousness, the duration of post-traumatic amnesia is the best way of predicting the long term outcome.4
I recommend using a good algorithm, not only to give the best estimate of outcome but also to indicate the range of uncertainty. Otherwise beware of predicting what will happen. Too often I am faced with unhappy relatives, amazed at and unsettled by the overly pessimistic prognosis of the doctors who initially looked after the patient.
Source of funding: Mr Nissen funded by University of Edinburgh.
For correspondence: Mr J J Nissen, Department of Neurosurgery, Newcastle General Hospital, Westgate Road, Newcastle Upon Tyne, NE4 6BE UK. Fax +44 (0)191 256 3169.
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.