Tools to study round-the-clock hospital medicine

Marc Ruben 7/8/2019

This repository describes the workflow (with code) for using electronic medical record (EMR) data to study treatment as a function of time-of-day.

Background

We analyzed the daily distribution of ~500K doses of 12 separate drugs in ~1.5K inpatients at a major children’s hospital in the U.S. Treatment orders and first-doses administered were time-of-day-dependent (figure below). These 24 h rhythms were generally consistent across drugs, diagnoses, and hospital units. https://www.biorxiv.org/content/10.1101/617944v1

Steps below describe how to reproduce these types of analyses.

Part 1. Extract EMR data from Epic

Schematic (below) of how EMR data in Epic moves from Chronicles, to Clarity, to CCHMC (Children’s Hospital Medical Center) divisional data mart. While different institutions maintain Epic data differently, the key data elements (ie. patient variables of interest) required for this study are provided in the data folder of this repository. This file also contains all of the data necessary to run Parts 2 and 3 (below).

Part 2. Evaluate 24 h patterns in TREATMENT

Characterize the daily distribution of treatment orders and first-doses administered. Code provided. Requires R.

Part 3. Evaluate 24 h patterns in RESPONSE

Characterize clinical responses as a function of time of day. Code provided. Requires R.