Statistical analysis
Demographic and clinical characteristics were compared between the
treatment groups in a bivariate manner, using Wilcoxon rank-sum tests
for continuous variables, due to data skewness, and Chi-Square tests for
categorical data, using Fisher’s Exact when expected cell counts were
small.
The primary outcome was hospitalization. If a hospitalization occurred
within 28 days of an outpatient treatment episode, it was counted as a
failure of that treatment. Generalized Estimating Equations (GEE) models
were used to determine the association between treatment (AWC alone
versus any antibiotics) and hospitalization. Unadjusted and adjusted
models were both performed, with the covariates in the multivariable
adjusted model being chosen based on their clinical relevance:
hospitalization within the past three months of a treatment episode, any
prior respiratory culture positive for Pseudomonas aeruginosa ,
indication for tracheostomy (prematurity, neuromuscular disease and
airway obstruction) and time since tracheostomy placement.
All analytic assumptions were verified and analyses were performed using
SAS v9.4 (SAS Institute, Cary, NC). This study was approved by Indiana
University Institutional Review Board. All data was stored on a REDCap
database.