Jordan T. Banks, Margaret Rosenfeld, Lloyd Mancl, Donald L.

Medication use is important to collect accurately in medically complex patients in both clinical and research settings. We assessed patient-level agreement for medication use between self-reported survey and electronic health record (EHR) for children with cystic fibrosis (CF). Our retrospective cross-sectional study focused on children with CF ages 6-20 years from Seattle Children's Hospital in Washington state, USA (N=85). A self- or parent-reported survey included questions on current use of specific medications and antibiotic use in the past two months. We compared survey data with data abstracted from the individual's EHR and derived Cohen's Kappa statistics to estimate the level of agreement between the two methods. Self-reported medication use was generally higher in the survey than in the EHR. The level of agreement ranged from slight for probiotics (74.1% agreement; 95% confidence interval [CI]: 64.6% to 83.6%; kappa: 0.07), pancreatic enzymes (80% agreement; 95% CI: 71.3% to 88.7%; kappa: 0.12), and vitamin D (55.3% agreement; 95% CI: 44.5% to 66.1%; kappa: 0.20) to moderate for chronic azithromycin (80% agreement; 95% CI: 7.13% to 88.7%; kappa: 0.50), proton pump inhibitors (76.5% agreement; 95% CI: 67.3% to 85.7%; kappa: 0.46), and oral antibiotics (70.6% agreement; 95% CI: 60.7% to 80.5%; kappa: 0.42). There is considerable heterogeneity in level of agreement in medication use between self-reported survey and EHR data for children with CF. Standardized approaches are needed to improve the accuracy of medication data collected in clinical practice and research.