Primary Predictor and Outcome Measures
The primary predictor was diagnosis of any type of ARI at the time of respiratory culture collection, as defined using International Classification of Diseases, Ninth Revision (ICD-9) and Tenth Revision (ICD-10) diagnostic codes placed by clinicians during the encounter in which the respiratory culture was obtained. Diagnostic codes consistent with conservatively-defined acute bacterial, viral, or nonspecific infection of the trachea or lower respiratory tract (e.g., pneumonia, tracheitis, ventilator-associated pneumonia; Appendix Table 1 ) were identified from review of the Clinical Classification Software-Respiratory Group diagnoses (Agency for Healthcare Research and Quality, Rockville, MD) and selected by group consensus between authors. A secondary predictor was evaluated for the subgroup of children with bacterial-specific ARI diagnoses (bARI) at the time of respiratory culture collection. bARI was defined using previously-identified ICD-9 codes12 and corresponding ICD-10 codes (e.g., bacterial pneumonia, acute bronchitis due to Streptococcus , acute tracheitis; Appendix Table 2 ).
The primary outcome was respiratory culture organism isolation (any isolation and specific organism isolation) in the first respiratory culture obtained in each encounter. Cultures with no speciated organisms or identification of only “oropharyngeal flora” were categorized as “negative”. The CCHMC Microbiology Laboratory performs semi-quantification of species for TA cultures and full quantification for BAL cultures. The Microbiology Lab does not have specimen rejection criteria. The Lab defines oropharyngeal flora broadly, and categorizes such species as Haemophilus , S. pneumoniae , and M. catarrhalis as oropharyngeal flora when isolated in small numbers in the presence of other oropharyngeal flora (Appendix Figure 1 ).