E1469. Frequency Analysis of Emergency Radiology Requests
  1. Azar Dagher; Universsity of Maryland School of Medicine
  2. Thomas Ptak; Universsity of Maryland School of Medicine
We investigate the periodicity of hourly, daily, weekly, and monthly incoming radiology requests at a Level I trauma center and emergency room using frequency domain analysis. Studies used to describe, and models used to forecast, ER activity are overwhelmingly skewed towards analyses in the time domain [1-6]. The relationship between the frequency and time domains is studied with Fourier analysis in order to make data more palatable in one form or the other [7]. We present such a frequency analysis using a large multi-year time series database of radiology requests for ER and incoming trauma.

Materials and Methods:
This was an institutional review board(IRB)–exempt and HIPAA-compliant retrospective review of the occurrence of 393275 emergent radiology study requests between January 1, 2013 and September 30, 2017 organized into hourly, daily, weekly, and monthly time series at a level I trauma center. Fourier frequency analysis of each time series was performed to identify the major peaks or cycles.

At least three periodicities or cycles for each of the time series were identified. The top three harmonics from all time series were identified as 1/year, 0.4/year, and 1.7/year.

Frequency analysis of emergent incoming radiology study requests shows that the time series are not random but have multiple discrete periodicities. The cyclical nature of radiology requests in the ER or trauma setting could have value in short-term staffing and longer-term hiring needs as well as resource management.