Statistical analysis
All statistical analyses, apart from the ROC power calculation, were
performed using SPSS Statistics version 25 [14]. Continuous
variables were compared for statistical significance to an alpha of
<0.05 using Mann Whitney U tests. Categorical data were
compared for statistical significance to an alpha of <0.05
using chi-square tests. Prior to binary logistic regression analysis
variables were screened for collinearity with all values showing an
acceptable level of tolerance (VIF <10). Univariate and
simultaneous multivariate binary logistic regressions were carried out
to compare the components of the BOOST and LACE models. Model accuracy
was determined by calculating the area under the receiver operating
characteristic curve. Youden’s index was calculated for each point in
the BOOST score and LACE index, with the maximum value representing the
ideal cut-off point for screening. Chi-squared testing to alpha
<0.05 was used to determine significance in chance of
readmission for high and low risk BOOST and LACE patients.