Accuracy of Fall Assessments in Long-Term Care Residents
Author Information
Author(s): Graham Laura, Li Yongmei, Dave Chintan, Odden Michelle
Primary Institution: VA Palo Alto Health Care System
Hypothesis
The Minimum Data Set (MDS) underreports falls compared to a rule-based NLP algorithm.
Conclusion
The study found that the MDS significantly undercounts falls among long-term care residents compared to the NLP algorithm.
Supporting Evidence
- The NLP algorithm identified 154,165 falls among 22,331 residents, while the MDS identified only 16,356 falls among 8,859 residents.
- The sensitivity of the MDS was found to be 0.30, indicating a high rate of missed falls.
Takeaway
This study shows that the way we check for falls in nursing homes might miss a lot of them, so we need better ways to find out how often they happen.
Methodology
Developed and validated a rule-based NLP algorithm to identify falls and compared it to MDS assessments using a national cohort of patients.
Limitations
The study may not generalize to all long-term care settings outside the VA system.
Participant Demographics
Patients receiving care at 114 VA long-term care facilities.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website