Predicting Pressure Injury in Critical Care Patients: A Machine-Learning Model.
Jenny Alderden, Ginette Alyce Pepper, Andrew Wilson and 5 others
PMID 30385537WHAT IT FOUND
Routine ICU electronic records predicted pressure-injury risk; body mass index, surgery time, hemoglobin, creatinine and age ranked highest.
Key findings
01The final sample included 6376 ICU patient records, and pressure injuries of stage 2 or greater occurred in 257 patients (4.0%), while stage 1 or greater occurred in 516 patients (8.1%).
02The random forest model had an area under the curve of 0.79 for both stage 1 and greater and stage 2 and greater pressure-injury outcomes.
03For stage 1 and greater pressure injuries, the most important variables were body mass index, hemoglobin level, creatinine level, time required for surgery, and age; for stage 2 and greater, the same variables appeared in a different order.
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 single-center and used retrospective electronic health record data. The model was not validated in an independent clinical sample. Some relevant nursing skin assessments and treatment-related data, such as surfaces and repositioning schedules, were unavailable. The random forest output ranks variable importance but does not establish that any variable caused pressure injury.
Declared interests
The article is listed as supported by NIH extramural funding; no author conflict-of-interest declaration appears in the supplied text.
The easy way to misread this
Do not treat this model as a ready-to-use bedside risk tool. It was built from retrospective records at a single hospital, lacked some nursing assessment data, and has not been tested in an independent sample.