Fall prediction using decision tree analysis in acute care units.
Shuntaro Tamura, Makoto Kobayashi, Yasuyuki Saito and 2 others
PMID 33281287WHAT IT FOUND
A decision tree model predicted inpatient falls with moderate accuracy (AUC 0.7919).
Patients with no fall history and no need for transfer assistance had no falls. For those needing transfer help, a Berg Balance Scale score of 51 or higher indicated low fall risk.
Key findings
01The decision tree model achieved an area under the curve of 0.7919 for predicting falls.
02Patients with no history of falling and no requirement for assistance with transfers experienced no falls during the study.
03Among patients requiring transfer assistance but with no fall history, those with a Berg Balance Scale score of 51 or higher did not fall, whereas those scoring below 51 had a higher fall rate.
STILL TO COME
How it was doneWhat they foundWhat it means for PTsWhat it means for OTs
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 decision tree analysis included only 86 patients, which is a small sample size for developing a predictive model, potentially limiting its generalizability. The study was conducted at a single institution using a specific hospital fall assessment sheet, so the results may not apply to facilities using different risk assessment tools. The decision tree cohort was heavily weighted toward neurosurgical patients and excluded those taking narcotics, meaning the model may not be accurate for these common acute care populations. The model's predictive accuracy was not verified in a separate, independent population, so it remains an internal validation.
Declared interests
The authors declared no conflicts of interest, and the research did not receive specific funding from any agency.
The easy way to misread this
Do not apply the Berg Balance Scale cutoff of 51 to all patients. This threshold was derived from a small subgroup of acute care patients who required transfer assistance but had no fall history, and the model was not validated in a separate population.