Variation in batch ordering of imaging tests in the emergency department and the impact on care delivery
This study examines heterogeneity in physician batch ordering practices and their impact on patient outcomes and resource utilization in emergency departments (ED). We analyze how different diagnostic test ordering strategies - batch ordering (multiple tests at once) versus sequential ordering - affect key healthcare metrics using comprehensive EMR data from Mayo Clinic of Arizona ED.
- Source: Mayo Clinic of Arizona Emergency Department
- Period: October 6, 2018 - December 31, 2019
- Sample: 43,299 patient encounters from 50,836 total visits
- Assignment: Random physician allocation to patients
-
Length of Stay (LOS)
- Physicians with batch tendency 1SD above average associated with 4.5% increase in ED LOS (p < 0.001)
-
72-Hour Returns
- 14.8% decrease (0.2 percentage points) in 72-hour return with admission for high-batch physicians
- Suggests more comprehensive initial evaluations
-
Resource Utilization
- Additional 8 imaging tests per 100 patient encounters for high-batch physicians
- Indicates potential over-ordering in batch approach
- Retrospective analysis using comprehensive EMR data
- Multivariable linear regression controlling for various covariates
- Balance testing for random assignment
- Subgroup analysis across different clinical conditions
The study demonstrates significant associations between physician ordering strategies and ED outcomes:
- Trade-off between longer initial visits and reduced return rates
- Potential over-utilization of imaging resources with batch ordering
- Need for optimized guidelines for ED test ordering practices
The analysis code in this repository allows for:
- Data processing and cleaning
- Statistical analysis and model estimation
- Generation of figures and tables
- Robustness checks and sensitivity analyses
For questions about this research or repository:
- Author: Jacob Jameson
- Email: jacobjameson@g.harvard.edu
Jacob Jameson. (2024). jacobjameson/batch-vs-sequence: HSR Final (v1.0.0). Zenodo. https://doi.org/10.5281/zenodo.14029338