Variation in Order Set Usage Across Two Pediatric ICUs

Introduction: Order sets are a form of passive decision support linked to improved efficiency and patient outcomes. However, little is known about their implementation and standardization across pediatric intensive care units (PICUs). We measured usage variation across hospitals based on key workflow factors such as provider type and timing of order placement. We hypothesize that order set usage will vary significantly by site, provider type, and order timing.


Methods: We performed a retrospective study across two academic PICUs, including all admissions from 1/2016 to 6/2023. Data was extracted, anonymized, and aggregated. For each order, we recorded type, category (e.g., nursing, medication, lab), and source (order set or not). Admission orders were defined as those placed within two hours of PICU entry. Analyses were limited to orders from residents, fellows, attendings, and advanced practice providers (APPs). Comparisons used chi-squared tests.


Results: Our cohort included 7.3 million orders across 407,162 ICU patient-days, with variation by site (A: 83%, B: 17%) and timing (15% admission, 85% non-admission). Order placement provider was significantly associated with site (p <.001), with physicians placing most orders at both institutions. Order set usage varied significantly across all tested variables. Usage was higher at site A than site B (40% vs. 22%, p< 0.001) and during admission vs. non-admission (67% vs. 28%, p< 0.001). Fellows and APPs had the highest order set usage (49% and 46%), followed by residents (34%) and attending physicians (25%). At both sites, all provider groups were significantly associated with timing, with higher usage during admission (p < 0.001). At both sites, common types included medication orders, followed by nursing and laboratory, a trend consistent across provider types. Outside admission time, site A usage included more nursing and medication orders compared to labs (43%, 31%, 9%) while site B showed a more balanced distribution (36%, 25%, and 34%, p < .0001).


Conclusions: We demonstrated significant differences in order set usage across sites. These findings suggest that usage is shaped by institutional culture, clinical context, and provider role. Understanding how order sets are adopted in practice is critical to optimizing and standardizing clinical workflows.