All analyses were performed using Stata 11.2 (Stata Corp, College Station, TX). Data are presented as proportions with 95% confidence intervals (95%CIs) and medians with interquartile ranges (IQR). Our primary analyses focused on RSV and HRV, the most commonly detected viruses in children with severe bronchiolitis. For the purposes of this analysis we combined RSV-A with RSV-B since the clinical distinction between the two subtypes of RSV was unremarkable. For analyses, we created a categorical variable that reflected the possible combinations of RSV/HRV status: (1) RSV only infection, (2) HRV only infection, (3) RSV in combination with HRV, (4) RSV in combination with non-HRV pathogens, and (5) HRV in combination with non-RSV pathogens.
We performed univariate analyses using chi-square, and Fisher’s exact test, and Kruskall Wallis test, as appropriate. All P-values were two-tailed, with P<0.05 considered statistically significant. Multivariable logistic regression analyses were conducted to evaluate independent predictors of longer LOS (≥3 days; defined using the median value of 2 days) and other measures of severity: ICU admission and continuous positive airway pressure (CPAP)/intubation. Factors were selected for inclusion in the model if they were found to be associated with the outcome in unadjusted analyses (P<0.20) or were potentially clinically significant. All regression models account for potential clustering by site. To further investigate independent predictors of LOS, a zero-truncated negative binomial model was also used to evaluate the relationship between demographic and clinical factors and LOS in days (continuous outcome). Children who were hospitalized for <1 day were assigned 0.5 days LOS. Results of the zero-truncated negative binomial model are reported as incidence rate ratios (IRRs) with 95%CIs.
We performed univariate analyses using chi-square, and Fisher’s exact test, and Kruskall Wallis test, as appropriate. All P-values were two-tailed, with P<0.05 considered statistically significant. Multivariable logistic regression analyses were conducted to evaluate independent predictors of longer LOS (≥3 days; defined using the median value of 2 days) and other measures of severity: ICU admission and continuous positive airway pressure (CPAP)/intubation. Factors were selected for inclusion in the model if they were found to be associated with the outcome in unadjusted analyses (P<0.20) or were potentially clinically significant. All regression models account for potential clustering by site. To further investigate independent predictors of LOS, a zero-truncated negative binomial model was also used to evaluate the relationship between demographic and clinical factors and LOS in days (continuous outcome). Children who were hospitalized for <1 day were assigned 0.5 days LOS. Results of the zero-truncated negative binomial model are reported as incidence rate ratios (IRRs) with 95%CIs.
