Outcome Prediction in Pediatric Submersion Victims

 

Population studied:

(1) 194 children from 5 months to 18 years (median age 2.6 years)

(2) experienced submersion in non-icy waters

(3) 131 were neurologically normal at time of discharge

(4) 10 had some degree of neurologic impairment at the time of discharge

(5) 38 died and 15 were vegetative at the time of discharge

 

Outcomes - one of the following:

(1) unfavorable: persistent vegetative state or death

(2) favorable: all other outcomes

 

Step 1: Separate children into comatose and noncomatose groups.

 

Children Not Comatose on Admission

 

All children who were alert on admission were normal at hospital discharge.

 

Most children who were lethargic or confused on admission were normal at hospital discharge. In the study series:

(1) 2 of 31 in this group were not normal at discharge.

(2) 1had a mild neurologic impairment, and

(3) 1 was in a vegetative state.

 

Thus, if a child is not comatose on admission, then a favorable outcome is predicted.

 

Children Comatose on Admission

 

Parameters associated with outcome in the logistic regression model:

(1) pupillary reflex: variable most strongly associated with unfavorable outcome

(2) serum glucose

(3) sex: male gender unfavorable

Parameter	Finding	Value
pupil reflex	present	0
	absent	1
sex	female	0
	male	1

 

X =

= 6.38 - (4.23 * (pupil reflex)) - (0.01 * (blood glucose in mg/dL)) - (2.3 * (sex))

 

probability of unfavorable outcome for comatose patients =

= 1 / (1 + EXP(X))

 

Probability of Unfavorable Outcome Calculated from the Logistic Regression Model	Sensitivity	Specificity
>= 0.1	100%	31%
>= 0.3	100%	50%
>= 0.5	89%	75%
>= 0.7	87%	81%
>= 0.9	81%	100%
>= 0.95	66%	100%

(Table 3, page 316)