Methods: We conducted a retrospective cohort study in a diverse network of 31 pediatric primary care practices. Patients 3 months to 18 years of age with CAP diagnosed between July 1, 2009 and June 30, 2013 were identified using ICD-9-CM codes. Clinical and provider data were abstracted electronically and confirmed with manual chart review. Multivariable logistic regression models including children prescribed amoxicillin or macrolides measured patient and provider level predictors of prescribing.
Results: Of 10927 children who received antibiotics for pneumonia, 4462 (40.8%) received amoxicillin and 4604 (42.1%) received macrolides (range 17 – 65% across practices). Patient age > 5 years (aOR, 5.37; 95% CI, 4.87, 5.92; p<0.01), antibiotic receipt in prior 3 months (aOR, 1.67; 95% CI, 1.49, 1.89; p<0.01), and suburban practice (aOR, 1.81; 95% CI, 1.52, 2.15; p<0.01) were independently associated with macrolide prescribing. Fever (aOR, 0.61; 95% CI, 0.55, 0.67; p<0.01), black race (aOR, 0.59; 95% CI, 0.51, 0.67; p<0.01), and treatment by a clinician who had practiced for > 10 years (aOR, 0.74; 95% CI, 0.67, 0.83; p<0.01) decreased the odds of macrolide prescribing. These relationships were not confounded by individual practice.
Conclusion: Antibiotic choice for CAP varied widely across practices. Macrolide prescribing was associated with relevant clinical features, such as patient age and fever. However, factors unlikely to be related to the microbiologic etiology of CAP such as race, practice location, and clinician years in practice were also associated with macrolide prescribing. Understanding these drivers of off-guideline prescribing will inform targeted antimicrobial stewardship initiatives.
T. Zaoutis, Merck: Investigator , Consulting fee and Research grant
Cubist: Investigator , Research support
M. Bryan, None
K. Feemster, None