Data are presented as either column frequency (percentage) or mean (standard deviation(SD)), as appropriate. Overall p-values were computed using chi-square for categorical variables and ANOVA or Kruskal-Wallis for continuous variables.
* Missing data were imputed based on age- and hospital-specific mean values.
a 2017/2018 Cluster 1 vs 2017/2018 Cluster 2, p<0.05
b 2017/2018 Cluster 1 vs 2017/2018 Cluster 3, p<0.05
c 2017/2018 Cluster 2 vs 2017/2018 Cluster 3, p<0.05