Simplified Estimates of the Probability of Death After Burn Injuries: Extending and Updating the Baux Score

TLDR

The Baux score, summing age and percent body burn, has long been used to predict mortality after burn trauma, but advances in burn care have made its predictions too pessimistic and it ignores inhalation injury. The authors revised the Baux score to add a 17‑year equivalent for inhalation injury and recalibrated it with a single‑term logistic regression model trained on 39,888 patients from the national burn repository, then compared it to a more complex logistic regression model using the same data. The revised Baux score, expressed as Age + Percent Burn + 17 × (Inhalation Injury), performed well for mental calculation and precise mortality prediction, yet a modern statistical model achieved better discrimination and calibration. Clinicians familiar with the original Baux score may welcome this updated, simpler version.

Abstract

Background: Generations of clinicians have used the Baux score, defined as the sum of age in years and percent body burn, to predict percent mortality after trauma, but advances in burn care have rendered the predictions of this score too pessimistic. Additionally, this score does not include the effects of inhalation injury. Methods: We revised the Baux score to include inhalation injury and recalibrated its predictions using a single-term logistic regression model developed using data on 39,888 burned patients provided by the national burn repository. We compared this revised Baux score to a more complex logistic regression model derived from the same data set and predictors. Results: A preliminary logistic regression model showed that age and percent burn contribute almost equally to mortality and further that the presence of inhalation injury added the equivalent of 17 years (or 17% burn). These observations suggested a revised Baux Score: Age + Percent Burn + 17 * (Inhalation Injury, 1 = yes, 0 = no) A logistic model based on the Revised Baux Score performed well, but a more complex model obtained using modern statistical model building tools had better discrimination and calibration. Conclusions: Our proposed revised Baux score is simple enough for mental calculation, and its inverse logit transformation (provided with a calculator or nomogram) can provide precise predictions of mortality. Better predictions can be obtained using our more complex statistical model. Burn surgeons and nurses accustomed to using the original Baux score may welcome an updated version.