Statistical Analysis
The median and interquartile range were computed for continuous
parameters, and counts and percentage for categorical variables. Group
comparisons were carried out using the Mann–Whitney test and the χ²
test as appropriate. Independent predictors of AHR were identified by
multivariate logistic regression modeling entering parameters with P
< .05 in a univariate analysis in a block. Odds ratios with
their 95% confidence intervals and regression coefficients were
computed. In order to create a weighted multivariate prediction score,
points were assigned to each of the elected predictors based on the
respective regression coefficient multiplied by ten and rounded up or
down to the next integer number. An individual score value was
calculated for each child by summing the weighted points of all the
predictors. For each predictor, as well as for the multivariate score,
sensitivity, specificity, a positive predictive value (PPV), and a
negative predictive value (NPV) were determined at respective cutoff
values, and receiver operating characteristics (ROC) curves were
constructed. The SPSS software package 25 (IBM Armonk, New York, USA)
was used for all calculations. P values < .05 were considered
significant.