Variables relative to significant periods associated with the injury:
(1) initial (at time of injury): age, sex, burn area as percent body surface area, third degree burn as percent body surface area, presence of inhalation injury
(2) transport: time to starting IV, fluid volume received during the first 24 hours after injury
(3) on admission: admission packed cell volume, base deficit, serum osmolarity
(4) during hospital course: presence of sepsis, serum creatinine > 2 mg/dL, platelet count < 20,000 per µL, requirement for inotropic support, dependency on mechanical ventilation
Patient selection: children with total burn area > 80% of body surface area and third degree burn area > 70% body surface area
Variables used in equations:
(1) age in years (range not specified, but included teenagers, so will take up to 18 years)
(2) percent body surface area burned (using all burns; some scores refer to area involved by second and third degree burns only)
(3) number of hours before IV started
(4) need for inotropic support
(5) presence of sepsis
(6) platelet count < 20,000 per µL
(7) days on mechanical ventilation during first 4 weeks after injury
X based on variables from all time periods but excluding ventilatory support =
= (-14.30) – (0.18 * (age in years)) + (0.15 * (percent body surface area burned)) + (0.38 * (number of hours before IV started)) + (0.31 is inotropic support required) + (1.96 if septic) + (1.87 if platelet count < 20000 per µL)
probability of death =
= EXP(X) / (1 + EXP(X))
Y based on variables from all time periods with ventilatory support =
= (-3.59) – (0.21 * (age in years)) + (3.25 if septic) + (4.85 * (percent days requiring ventilatory support in first 28 days as decimal fraction))
where:
• percent days requiring ventilatory support = (days requiring ventilation) / MIN(28, days hospitalized) * 100%
probability of death =
= EXP(Y) / (1 + EXP(Y))
Performance:
• The scores using variables from all time periods performed better than scores using only injury, transport or admission data. The ROC curve plots (Figure 2, page 991) show a significantly higher area under the curve for the equations using variables from all time periods.
• When these scores are compared to other burn scores (Baux, O'Keefe, Ryan, Smith, Saffle, Tobiasen ABSI) these scores were better able to predict outcome. Some of the improved performance may be related to a limited population (children with catastrophic burns).
