Statistical methods to predict morbidity and mortality: self assessment techniques for burn units

Area of Science:

• Medical Statistics
• Burn Injury Research

Background:

• Retrospective review of 806 adult and pediatric burn patients.
• Need for predictive tools in burn care management and outcome assessment.

Purpose of the Study:

• Develop equations to predict morbidity (length of stay, transfusions, operations).
• Compare survival statistics using probit analysis and LA50s.
• Evaluate discriminant analysis for predicting burn survival.

Main Methods:

• Multiple regression analysis for morbidity prediction.
• Probit analysis for mortality probability and LA50 calculation.
• Discriminant analysis for survival prediction (95.8% accuracy).

Main Results:

• Developed predictive equations for hospital stay, transfusions, and operative procedures.
• Observed significantly improved survival in pediatric, young adult, and older adult groups.
• Discriminant analysis achieved 95.8% accuracy in predicting burn survival.

Conclusions:

• Multiple regression equations serve as valuable tools for prediction, audit, and assessing burn care improvements.
• Improved survival rates suggest the current burn care protocol is effective.
• Discriminant analysis and burn severity scoring aid in identifying high-risk patients.