This article is more than 5 years old. Information may no longer be current.

Algorithm using EMR data may better predict mortality in polytrauma patients

Add topic to email alerts

Receive an email when new articles are posted on

Please provide your email address to receive an email when new articles are posted on .

Subscribe

Added to email alerts

You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

Back to Healio

We were unable to process your request. Please try again later. If you continue to have this issue please contact customerservice@slackinc.com.

Back to Healio

Ryan W. Fairchild

DENVER — At the Orthopaedic Trauma Association Annual Meeting, a presenter said an algorithm that uses electronic medical record data may offer clinicians a clearer description of mortality risk in polytrauma and assist them in decision-making during early hospitalization.

“The polytrauma early mortality model ... stratifies patients into just two groups: high and low risk,” Ryan W. Fairchild, MD, said during his presentation. “High-risk patients have a one in three risk of mortality in the first 2 days. In low-risk patients, they have a 99% plus chance of survival in the first 48 hours.”

At a level 1 trauma center, Fairchild and colleagues developed the polytrauma early mortality model, a machine-learning algorithm that predicts 48-hour mortality during the first 72 hours after hospitalization using EMR data. Every 12 hours, the model updates and evolves with a patient’s physiologic response to trauma and ongoing resuscitation. From 2009 to 2014, the model was trained on 4,567 hospitalized polytrauma patient encounters. From 2015 to 2016, the model was tested on 484 encounters.

Results showed that of the 56 12-hour time intervals within 48 hours of mortality, the model predicted 52 12-hour time intervals. The sensitivity was 92.8% and specificity was 92.2%. The positive predictive value was 31.7%. Survival was predicted 1,342 times, and the model was incorrect four times. This yielded a 99.7% negative predictive value. The positive and negative likelihood ratio was 12 and 0.08, respectively. During the first 72 hours of hospitalization, the model performance was stable.

“The advantage of our system is its entirely EMR based, its automated, there’s no scoring system and no calculations to be done by hand,” he said. “Importantly, it evolves with patients during their early hospital stay.” – by Monica Jaramillo

Reference:

Fairchild RW, et al. Abstract 93. Presented at: Orthopedic Trauma Association Annual Meeting; Sept. 25-28, 2019; Denver.

Disclosure: Fairchild reports no relevant financial disclosures.