It’s the quality of data captured, not quantity, that matters when applying predictive analytics to the delivery of care, finds a new study from the University of Texas Southwestern.
The UT researchers analyzed data from about 33,000 patient admissions to test models predicting the likelihood of 30-day readmissions. They expected to see that adding more detailed clinical data across an entire hospitalization would allow them to better identify patients most at risk for readmission.
What the researchers discovered surprised them. Using clinical data across the entire length of stay compared to only using data from the patient’s first 24 hours in the hospital was only marginally better at predicting readmission.
The authors suggest that the non-clinical factors, including patient health literacy, socioeconomic challenges and behavioral health, have far more of an impact than previously believed on which patients are more likely to return within the 30-day window.
In a press release on the study, author Dr. Ethan A. Halm, said:
More ‘big data’ alone did not make much of a difference. Better models for predicting readmissions will require ‘better data’ on things like psychosocial and behavioral factors that are not currently captured in electronic health records.
Learn more about Polestar, the collaborative behavioral care platform, and how it can help you better identify and manage the behavioral health risks that can complicate care.