Prior to enrollment, a power analysis was carried out to determine the
number of subjects needed to demonstrate differences in systolic function from
healthy controls. Based on a normal EF% of 60±20%, a
hypothesized effect size of 15%, a type-I error rate (α) of
0.05, and power (1-β) of 0.8, it was determined that 39 burn subjects
would need to be enrolled to detect a statistically significant differences in
EF%. All analyses were carried out with R 3.3 for Windows (Vienna,
Austria) or Graphpad Prism 7.00 for Windows (La Jolla, CA). Student’s
t-test and one-way ANOVA were used to compare continuous outcomes. Standard
univariate and multivariate least-squares regression models were fit to
continuous responses. As necessary, predictors and responses were transformed to
allow for better fitting of the model assumptions. For categorical outcomes,
logistic regression models were fit; inference was based on comparisons of
deviances among hierarchically fit models. Multi-variable logistic regression
models were fit and assessed using standard generalized linear model functions
in R. All data are reported as mean ± SD unless otherwise noted. For all
analyses, statistical significance was reported with p < 0.05.
number of subjects needed to demonstrate differences in systolic function from
healthy controls. Based on a normal EF% of 60±20%, a
hypothesized effect size of 15%, a type-I error rate (α) of
0.05, and power (1-β) of 0.8, it was determined that 39 burn subjects
would need to be enrolled to detect a statistically significant differences in
EF%. All analyses were carried out with R 3.3 for Windows (Vienna,
Austria) or Graphpad Prism 7.00 for Windows (La Jolla, CA). Student’s
t-test and one-way ANOVA were used to compare continuous outcomes. Standard
univariate and multivariate least-squares regression models were fit to
continuous responses. As necessary, predictors and responses were transformed to
allow for better fitting of the model assumptions. For categorical outcomes,
logistic regression models were fit; inference was based on comparisons of
deviances among hierarchically fit models. Multi-variable logistic regression
models were fit and assessed using standard generalized linear model functions
in R. All data are reported as mean ± SD unless otherwise noted. For all
analyses, statistical significance was reported with p < 0.05.