Article Text

PDF

An aid to predicting prognosis in patients with non-traumatic coma at one day
  1. J Overell,
  2. I Bone,
  3. G N Fuller
  1. Dr GN Fuller, Department of Neurology, Gloucester Royal Hospital Great Western Road, Gloucester GL1 3NN, UK geraint{at}Fullerg.demon.co.uk

Statistics from Altmetric.com

Neurologists are often asked to give a prognosis of a patient in coma. Table 1 is designed to help provide guidance. The data are extracted from a series of 500 patients with coma from causes other than trauma and overdose.1 Patients with a Glasgow coma scale of less than 2:4:2 (eye opening; motor; verbal) for more than six hours were included. The data are presented to allow discrimination of patients with the poor outcomes at one year (death, vegetative state, and severe disability) from those with moderate or good outcomes as defined by the Glasgow outcome scale2 on the basis of the clinical signs one day after admission.

Table 1

 Data from Levyet al 1 about patients with non-traumatic coma of more than six hours duration (excluding drug induced coma), showing the relation between physical signs at 24 hours and their long term outcome

The raw data indicating the number of patients with each outcome in each neurologic response category at one day are given on the left of the table. The sensitivity describes how good the test is at detecting those with poor outcome at one year. Thespecificity describes how good the test is at correctly excluding those individuals with a good prognosis at one year. The positive predictive valuedescribes the probability that the outcome will be bad if the patient has the poorer (first) neurological outcome. Thenegative predictive value describes the probability of a good outcome if an individual has the better (second) response.

The likelihood ratio (shown in bold type) is a useful way to use these probabilities clinically. The likelihood ratio of a positive test indicates the effect of the worse neurological state while the likelihood ratio of a negative test indicates the effect of the better state on outcome.

Using the likelihood ratio nomogram3 shown in fig 1A the effect of a negative or positive test can be appreciated. From the population of patients admitted in coma examined at one day 79% have a poor outcome (305/385), so the pre-test probability for a poor outcome is 79%. The likelihood ratio indicates the effect of the different elements of clinical examination. If, for example, their best motor response is flexion or worse (likelihood ratio of positive test is 3.66), then using a straight edge on the nomogram the post-test probability of a poor outcome is about 95% (fig 1B, line A). If, however their best motor response is withdrawing or better (likelihood ratio of negative test is 0.4) the post-test probability of a poor outcome falls to about 60% (fig 1B, line B).

Figure 1

 The likelihood ratio nomogram is a tool to show the effect of a test result on the pre-test probability. Reproduced from Sackett et al3 with permission of the publisher.

As the signs are not independent of one another the prognosis should be taken from the most informative sign rather than from combining the signs.

References

View Abstract

Request permissions

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.

Linked Articles