was performed for nominal or ordinal variables between groups. A multivariable
logistic regression model, with backward elimination, was
constructed in order to find independent risk factors associated
with BPI and CF. P values less than 0.05 were considered
significant.