The Utility of Nursing Notes Among Medicare Patients With Heart Failure to Predict 30-Day Rehospitalization: A Pilot Study.
Youjeong Kang, Maxim Topaz, Sandra B Dunbar and 2 others
PMID 34935742WHAT IT FOUND
Nursing notes predicted 30-day heart failure rehospitalization better than physician discharge summaries in this pilot study.
The nursing notes model achieved an accuracy score of 0.85, compared to 0.74 for discharge summaries, suggesting nurses' narrative documentation holds valuable prognostic information.
Key findings
01The best model using nursing notes achieved an area under the curve (AUC) of 0.85 and an F1 score of 0.80.
02The best model using physician discharge summaries achieved an AUC of 0.74 and an F1 score of 0.61.
03The study concluded that nursing notes provided superior input for predicting rehospitalization compared to discharge summaries.
STILL TO COME
How it was doneWhat they foundWhat it means for RNs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study was a pilot with a relatively small sample of 500 patients. Data came from a single institution, so results may not apply to other hospitals. The machine learning models used (neural networks) do not explain which specific words or phrases in the notes drove the predictions, making it hard to understand the clinical reasoning behind the model's output. The study did not identify the specific factors in nursing notes associated with rehospitalization risk, only that the notes as a whole were more predictive.
Declared interests
The study was supported by non-U.S. government funding. No other conflicts of interest were declared in the provided text.
The easy way to misread this
Do not assume that nursing notes are currently ready for use in clinical decision-making tools. This was a pilot study using complex machine learning models that cannot explain their predictions. The high accuracy scores (AUC 0.85) have not been validated in larger, multi-center studies, and the specific content in the notes that drives this predictive power remains unknown.