JMIR Publications

ABSTRACT

Background:

Electronic health record (EHR) data is being increasingly used for retrospective observational research through large, robust databases and advanced data extraction tools.

Objective:

We sought to assess the accuracy of vital sign, ventilator, and continuous medication data captured in the EHR in a pediatric intensive care unit (PICU).

Methods:

We conducted a retrospective observational study of children receiving invasive mechanical ventilation in June 2025. Data sources included 1) A bedside clinical researcher, 2) Automated EHR extraction, and 3) A continuous vital sign monitoring system. Vital sign comparisons used the continuous vital sign monitoring system as the gold standard. Ventilator and medication data comparisons used bedside observations as the gold standard. Differences were measured as Means with standard deviation (SD) or Median Differences (MD) with interquartile ranges (IQR).

Results:

We obtained 110 bedside observations from 27 unique patients. All measured vital signs in the EHR were accurate relative to the continuous vital sign monitoring system with mean differences ranging from a low of 0.1% for oxygen saturation to a high of 1.6 breaths per minute for respiratory rate. Most vital signs did have rare outliers such as a diastolic blood pressure difference of 46mmHg, a heart rate difference of 35 beats per minute, and a respiratory rate difference of 18 breaths per minute. Ventilator settings were highly accurate in the EHR with MD of 0.0 and IQR of 0-0. Outliers were less common but included a PEEP difference of 10mmHg, a respiratory rate of 4 breaths per minute, and an FiO2 of 15%. Continuous medication dosing accuracy was variable with an overall low accuracy between 28.0-35.2%.

Conclusions:

EHR data capture in the PICU is accurate for vital signs and ventilator settings, but less accurate for continuous medications.