Dive Brief:
- Tailoring infection control programs to a hospital’s patient demographic and caseload can help to identify high-risk patients and improve outcomes, a new study in Infection Control & Hospital Epidemiology suggests.
- The researchers used EHRs from nearly 257,000 admissions at Massachusetts General Hospital and the University of Michigan Hospitals to build facility-specific machine learning models that predict daily risk of Clostridium difficile. Data extracted included patient demographics, admission details, patient history and daily hospitalization details.
- The approach offers an alternative to one-size-fits-all infection programs, the researchers say.
Dive Insight:
Using the EHR data, models at both hospitals achieved 95% confidence levels using a standard value curve. While they did share some predictive factors, many of the leading predictors were unique to each facility.
“These hospital-specific models allow for earlier and more accurate identification of high-risk patients and better targeting of infection prevention strategies,” according to the report.
Hospitals are in a constant struggle to control costs as expenses increase and payers push providers to shift more patients to lower-acuity outpatient settings. One area to target is hospital-acquired infections, which can increase inpatient stays and require treatment in more costly ICUs.
According to a recent Premier analysis, one of the key opportunities to reduce ICU care lies in infectious and parasitic diseases linked to operating room procedures and associated complications and comorbidities.
Roughly one in 25 patients has at least one hospital-acquired infection at any given time, according to the Centers for Disease Control and Prevention.
Currently, though, the focus has been on developing risk prediction models for different infection agents that can be applied across healthcare organizations. The EHR/machine learning approach could help hospitals prevent certain infections by understanding the particular vulnerabilities of their patient populations.
To reduce HAIs, hospitals are adopting a range of new technologies — from ultraviolet light and real-time locating systems that monitor employee hand washing to rapid diagnostics that can identify a specific bacterium within 24 hours.
Hospitals are also using data mining to tackle the problem. Such programs link to a hospital’s data feed to identify patients with multidrug-resistant organisms or other infections so they can be isolated. They can also be used to target trends or associations of different infections within a facility, leading to faster notification of outbreaks and transmissions.