Ong et al developed an equation for predicting neurologic outcome in children after head injury using logistic regression. This can help identify children who are likely to have a poor outcome and who may require more aggressive management. The authors are from Kuala Lumpur in Malaysia.
Patient selection: children < 15 years of age with severe head injury
Poor outcome – one of the following:
(1) severe disability
(2) persistent vegetative state
(3) death
Parameters:
(1) Glasgow Coma Scale (GCS) at 24 hours after injury, from 3 to 15
(2) hypoxia (PaO2 < 65 mm Hg)
(3) subarachnoid hemorrhage, either diffuse or focal
(4) diffuse axonal injury (small eccentrically placed hemorrhages in the corticomedullary junction, deep white matter, corpus collosum ganglionic region or brainstem
(5) brain swelling (compression or obliteration of the mesencephalic cisterns with non-enlarged ventricles)
|
Parameter
|
Finding
|
Points
|
|
Glasgow Coma Scale
|
|
(GC score)
|
|
hypoxia
|
present
|
-1
|
|
|
absent
|
1
|
|
subarachnoid hemorrhage
|
present
|
-1
|
|
|
absent
|
1
|
|
diffuse axonal injury
|
present
|
-1
|
|
|
absent
|
1
|
|
brain swelling
|
present
|
-1
|
|
|
absent
|
1
|
X =
= 2.7837 – (0.2925 * (Glasgow Coma Scale)) – (1.0012 * (points for hypoxia)) – (1.0139 * (points for subarachnoid hemorrhage)) – (1.2138 * (points for diffuse axonal damage)) – (1.1011 * (points for brain swelling))
According to the paper:
probability of poor outcome =
= EXP(-X)
where:
• On page 288, the equation is "-log odds (poor outcome)". I am assuming LOG refers to the natural logarithm.
If this equation is tried at the extremes of values, bogus values are returned. For example, a GCS of 3, PaO2 of 60, with all 3 CT findings gives a probability of poor outcome as 0.002.
If the typical equation is used, then reasonable values are returned:
probability of poor outcome =
= 1 / (1 + EXP((-1) * X))
Observations:
(1) GCS alone unable to predict a poor outcome
(2) The presence of hypoxia and GCS < 8 was associated with poor outcome.
(3) The presence of the 3 CT scan findings plus hypoxia increased risk of poor outcome irrespective of GCS.
(4) The presence of the 3 CT scan findings without hypoxia was modified if the GCS was > 12, else there was a poor prognosis.
