Developing a clinical decision support framework for integrating predictive models into routine nursing practices in home health care for patients with heart failure.

The proposed CDS tool was not implemented or tested with clinicians, patients, or outcomes; the paper describes design, mapping, and a workflow. Data came from one large non-profit home health agency in the Northeastern United States, so other settings may differ. The records were from patients admitted between January 1, 2015 and December 31, 2017, which may not match current coding, documentation, or home health practice. Missed visits were not adjusted for visits that may have been missed because the patient was already hospitalized or in the emergency department. Several important predictors came from free-text clinical notes and were not interpretable or mappable to FHIR, which limits real-time EHR use. Traditional vital signs were not among the top 20 predictors, but the authors suggest this may reflect infrequent home health measurement rather than lack of clinical importance.

Funded by the Agency for Healthcare Research and Quality. Some authors were supported by National Institute for Nursing Research training grants. All authors report no conflicts of interest relevant to this article.

Do not treat this as evidence that the proposed alert reduces hospitalizations or emergency visits. The paper describes a model and workflow, but the alert was not tested with clinicians or patients.