Print “PDF version

What’s New in the Patient Safety World

March 2022

Predicting C. diff Infection in Just 6 Hours?

 

 

What if you could accurately predict in a timely fashion which of your hospitalized patients were likely to develop C. diff (Clostridiodes difficile) infection? You could institute prophylactic measures, make the diagnosis earlier, and implement infection control measures in a more timely fashion.

 

A new study looked at the ability of machine learning algorithms to predict which hospitalized patients will become infected with C. diff (Panchavati  2022). The study used electronic health record data from several hospitals to develop 3 machine learning algorithms and then tested the algorithms using data 1,149,088 inpatient encounters with 7,107 CDI (C. diff infection) encounters at other hospitals. They found that machine learning algorithms can predict future CDI in hospitalized patients using just the first 6 hours of inpatient data!

 

You’ll have to read the publication itself for details of each of the 3 machine learning algorithms and all the incorporated data elements. Age was one of the most important features in generating predictions in each of the 3 machine learning algorithms. Other data elements that were important predictors were sodium, BMI, white blood cell count, bilirubin, heart rate, diastolic blood pressure, active medication treatment with antibiotics or proton pump inhibitors (PPI’s).

 

The size and breadth of underlying data was a major strength of the study. The data came from EMR’s at over 700 US hospitals. The development dataset contained 13,664,840 inpatient encounters with 80,046 CDI encounters and the external validation dataset used data on 1,149,088 inpatient encounters with 7,107 CDI (C. diff infection) encounters at other hospitals. And fact that all the data was easily gleaned from the EMR makes potential use of these machine learning algorithms very attractive.

 

While we’ve always been able to use many of the above risk factors to identify patients at risk of developing CDI, the accuracy of these machine learning algorithms has the potential to significantly impact our approach to CDI.

 

 

Some of our prior columns on C. diff infections:

·       August 2021               Updated Guidelines on C. diff

·       October 2021              HAI’s Increase During COVID-19 Pandemic

 

References:

 

 

Panchavati S, Zelin NS, Garikipati A, et al. A comparative analysis of machine learning approaches to predict C. difficile infection in hospitalized patients. AJIC 2022; Published online: January 19, 2022

https://www.ajicjournal.org/article/S0196-6553(21)00757-4/fulltext

 

 

 

 

Print “PDF version

 

 

 

 

 


 

http://www.patientsafetysolutions.com/

 

Home

 

Tip of the Week Archive

 

What’s New in the Patient Safety World Archive