Patient Risk Factor Profiles Associated With the Timing of Goals-of-Care Consultation Before Death: A Classification and Regression Tree Analysis.
Lauren T Starr, Connie M Ulrich, Paul Junker and 3 others
PMID 32602349WHAT IT FOUND
Among hospitalized patients who had a palliative goals-of-care consultation before death, most had it within 14 days of death, median 13 days.
A model identified close-to-death consultations with 54.7% accuracy, but not early or moderately timed ones.
Key findings
01Of 1,141 patients who died after a goals-of-care palliative consultation, 54% received it within 14 days of death, 26% received it 15-60 days before death, and 21% received it more than 60 days before death.
02The classification model identified goals-of-care palliative consultation close to death with 54.7% accuracy overall and 88.2% sensitivity for that group, but sensitivity was 1.7% for early consultation and 27.4% for moderately timed consultation.
03ICU patients with extreme illness severity who self-identified as Hispanic, Other, or Unknown had an 85% probability of receiving a goals-of-care palliative consultation close to death, while patients with less than extreme severity and no ICU care had a 24% probability.
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
Single academic medical center with high acuity and an established palliative care team, so results may not apply to other hospitals or community settings. Observational design; it cannot show that any factor caused late consultation, and the authors say reverse causation is unknown. Prediction accuracy was modest, and the model performed poorly for early and moderately timed consultations, so it is not a reliable tool for identifying patients who need earlier conversations. Many relevant social, religious, cultural, family, language, and health literacy factors were not measured; Medicaid status was only a proxy for income. Timing of some variables relative to the consultation was unclear, because the first consultation may have occurred before ICU admission or severity ratings. Patients whose death was not recorded in the system were not included, and the analysis could not compare deaths inside versus outside the health system.
The easy way to misread this
Do not treat the risk profiles as a validated bedside tool for deciding who needs palliative care. The model had 54.7% accuracy and poor sensitivity for early and moderately timed consultations, and the study is observational, so it does not show that these factors cause late consultation.