The Development of a Probabilistic Dose–Response for a Burn Injury Model. (1st March 2017)

• Record Type: Journal Article
• Authors: Iyoho, Anthony; Ng, Laurel; Chan, Philemon
• Is Part Of: Military medicine. Volume 182(2017)Supplement 1
• Page Start: 202
• Page End: 209
• Publication Date: 2017-03-01
• Subjects: Surgery, Military -- Societies, etc; Medicine, Military -- Societies, etc; Medicine, Military -- Periodicals; Surgery, Military -- Periodicals; Medicine, Military; Surgery, Military; Military Medicine -- Periodicals; Periodicals; Electronic journals; 616.98023
• DOI: 10.7205/MILMED-D-16-00235
• Languages: English
• ISSNs: 0026-4075

Abstract

ABSTRACT: Objective: The objective was to augment a burn injury model, BURNSIM, with probabilistic dose–response risk curves. Methods: To develop the dose–response, we drew on a considerable amount of historical porcine burn injury data collected by U.S. Army Aeromedical Research Laboratory in the 1970s. The experimental parameters of each usable data point served as inputs to BURNSIM to calculate the burn damage integral (i.e., the internal dose) for 4 severities (mild, intermediate, deep second- and third-degree burns). The binary probability response was constructed and logistic regression was applied to generate the respective dose–response. Historic data collected at the University of Rochester in the 1950s were used for validation. Results: Four dose–response curves were generated, ranging from mild to third degree, with tight 95% confidence bands for mild to deep second degree, and slightly wider bands for third degree. Parametric sensitivity analysis revealed that epidermal and whole skin thicknesses, skin temperature, and blood flow rate have a large effect on predicted outcomes. Conclusions: Addition of dose–response curves provides a critical augmentation to BURNSIM to improve operational risk assessments of burn hazard. Future recommendations for BURNSIM include the use of body location- and gender-specific parameters with coupling to a thermoregulatory model.