A new model to predict adverse events in spine surgery: 4 key points

Spine

Researchers in the department of neurosurgery at Stanford (Calif.) University School of Medicine developed and evaluated predictive models for adverse events associated with spine surgery.

They used machine learning to develop the model to predict adverse events 30 days postoperatively and published their results in The Spine Journal. The researchers used a least absolute shrinkage and election operator regularization method and logistic regression approach with the data to examine adverse event risks, according to the study abstract.

Data from 345,510 patients entered into the Truven MarketScan and MarketScan Medicaid Database were included, as well as 760,724 Medicare and Medicaid beneficiaries.

Here are the results:

1. Cardiac dysfunction was the most common individual adverse event among CMS beneficiaries, with 10.6 percent of the patient population experiencing cardiac dysfunction. The second most common adverse event was pulmonary complications, with 4.7 percent of patients experiencing it.

2. The prediction accuracy, or area under the curve, was 0.7 for all adverse events. The model for pulmonary complication risk prediction was the most accurate, at 0.76.

3. Researchers also found Medicaid status was an important factor in predicting the adverse events with beneficiaries increasing their odds for adverse events 20 percent to 60 percent.

4. The article authors recommend clinicians use the model during patient counseling and treatment discussion, and it could be considered for accurate risk adjustment.

Copyright © 2024 Becker's Healthcare. All Rights Reserved. Privacy Policy. Cookie Policy. Linking and Reprinting Policy.

 

Featured Webinars

Featured Podcast

Featured Whitepapers