Exploring Inpatient Medication Patterns: A Big Data and Multilevel Approach.

Purpose: Exploratory study to examine inpatient medication administration patterns.

Methods: Data from multiple sources were utilized for this study. The outcome was time difference between medication schedule and administration. A 3-level hierarchical linear regression approach, both unadjusted and adjusted, was considered for this study where medication administration events are nested within patients nested within nurses or units. Intraclass correlation coefficients (ICCs) were calculated and compared.

Results: On average, medications were delayed by 12 (SD, 48.8) minutes. From the full model, patient ICCs decreased when "unit" replaced "nurse" as the 3rd level (0.541 vs 0.444). Patients who spoke Spanish had a significant 2.3- to 4.2-minute delay in medication administration. Certified nurses significantly give medications earlier compared with noncertified nurses by 1.6 minutes.

Discussion: Optimal medication administration is a multifactorial concern with nurses playing a role. Nursing leaders should also consider patient demographics and unit conditions, such as culture, for medication administration optimization.