August 11, 2021
Read for 2 minutes
Source / information
Disclosure:
Jones does not report any relevant financial information. Please refer to the study for all relevant financial information from the other authors.
ADD SUBJECT TO EMAIL ALERTS
Receive an email when new articles are published on
Please enter your email address to receive an email when new articles are published on . “data-action =” subscribe “> subscribe
We could not process your request. Please try again later. If this problem persists, please contact [email protected].
Back to Healio
Computerized versions of the Pneumonia Severity Index accurately identified low-risk mortality patients with community-acquired pneumonia, according to data published in the Annals of the American Thoracic Society.
“Recent advances in electronic health records and the modeling of clinical predictions may make consistent severity assessment more practical, and recent guidelines call for the assessment of computational approaches,” Dr. Barbara E. Jones, a physician in the Department of Pulmonary and Intensive Care Medicine, Veterans Affairs Salt Lake City Health Care System at the University of Utah, Salt Lake City, and colleagues wrote. “Computer-aided risk assessments can reduce the effort and involve more complexity than those that are designed for manual calculations.”
Source: Adobe Stock.
Researchers evaluated 297,498 adults with community-acquired pneumonia (mean age 68 years; 95% men) who presented to emergency rooms at 117 Veterans Affairs medical centers from January 2006 to December 2016. The researchers compared a computerized Pneumonia Severity Index score, which included all variables except confusion and pleural effusion score, with 10 novel models that considered additional patient factors.
The primary endpoint was all-cause mortality within 30 days of the first emergency room visit. Secondary endpoints included hospitalization within 24 hours of initial emergency room visit and 7-day secondary hospitalization within 7 days of discharge from the emergency room.
The median pneumonia severity index score was 86. Sixty percent of the patients were hospitalized. The 30-day overall mortality was 6.6%. In addition, of 179,682 originally treated and discharged patients, 7% returned with a secondary hospitalization within 7 days.
Seven percent of the 297,498 encounters resulted in death within 30 days of the first visit to the emergency room.
The performance of the models increased as the complexity of the models and the number of variables increased. The area under the curve (AUC) predicting 30 day mortality was 0.77 for the Pneumonia Severity Index computerized model. High performance was also observed in models limited by age, gender, and physiological variables.
With a Pneumonia Severity Index Score threshold of 970 and a mortality risk cutoff of <2.7%, the computerized Pneumonia Severity Index Score identified 31% of all patients who were at lower risk for all-cause mortality while using the improved decision tree algorithm machine learning the Pneumonia Severity Index model using the Extreme Gradient Boosting algorithm with age, six and 26 other physiological factors identified 53% of all patients as lower risk and the same enhanced machine learning model with all 69 variables identified 56% of all patients as a low risk, according to the results. The models found similar mortality rates, hospital stays, and 7-day secondary hospital stays.
“As our health systems evolve, so too will the way we conceptualize the severity of illness. The advantage of the original [pneumonia severity index] is that it is known well-
validated heuristic that provides consistency that can reduce inequalities and unjustified variation despite increasing information and demands on clinicians, ”the researchers write. “The downside is that it’s both tedious and oversimplified: clinicians consider many more features when assessing disease severity, and the amount of clinical data that computers can process has increased.”
ADD SUBJECT TO EMAIL ALERTS
Receive an email when new articles are published on
Please enter your email address to receive an email when new articles are published on . “data-action =” subscribe “> subscribe
We could not process your request. Please try again later. If this problem persists, please contact [email protected].
Back to Healio
