TABLE 4.

Sensitivity Analyses: Reduction in Selection of Third-generation Cephalosporins for Empirical Treatment of Children With CAP After Release of National Guidelines

Original and Sensitivity Analyses	Absolute Difference, % (95% CI)
Original estimation	−12.4 (−19.8 to −5.1)
Estimated from last month of preintervention	−10.6 (−16.3 to −5.0)
Restricted to independently confirmed radiographic pneumoniaa	−16.2 (−25.7 to −6.6)
Adjusted for proportion <5 yrs old	−13.6 (−27.0 to −0.2)
Adjusted for total no. of hospitalizations	−13.7 (−21.8 to −5.7)
Adjusted for proportion with comorbidity	−12.6 (−20.4 to −4.8)
Adjusted for pleural effusion	−12.6 (−20.3 to −4.9)
Adjusted for median length of stay	−11.2 (−22.2 to −2.5)
Adjusted for direct admission to intensive care	−12.7 (−21.8 to −3.7)

Absolute difference was calculated by comparing the observed antibiotic prescribing patterns by the end of the study period (estimated from the postguideline regression model) with the expected antibiotic prescribing by the end of the study period had the guideline not been released (estimated from the preguideline trend).

^aAll children enrolled in the EPIC study had a chest radiograph consistent with pneumonia, although 10% lacked independent radiographic confirmation by a study radiologist.