Back

DECODE the quality of dying in the intensive care unit

Dougan, S.; Duke, G. J.; Hunter, S.; Ha, S. N.; Johnson, N. J.; Lehane, D.; Saw, J. T.; Teo, A.; Zou, S.; Barrett, J.

2025-07-15 palliative medicine
10.1101/2025.07.13.25331477 medRxiv
Show abstract

ObjectiveExisting national quality of care processes do not assess quality of care for the dying patient. We developed the Documentation and Evaluation of Care of Dying Equation (DECoDE) score, a 21-point unweighted binary questionnaire. for this purpose. DesignRetrospective clinical audit by clinical staff of medical records. SettingTwo metropolitan adult intensive care units (ICU) in Victoria. Participants157 randomly selected records of adults who died in ICU between 1 Jan 2015 - 30 June 2017. Main outcomesDECoDE audit score, patient characteristics, and therapeutics delivered in the last 24hrs of life. ResultsOver 30-months there were 5,194 ICU admissions and 539 (10.4%) deaths - 157 (43.9%) were randomly selected for audit. 122 (82.2%) were receiving curative treatment within the 24-hours prior to death. 130 (83%) deaths were expected and occurred 1.5 (IQR =1-4) days after agreement to withdraw treatment. Median DE-CoDE score was 11 (IQR =9-13) out of a maximum score of 21. Factors associated with a lower DECoDE score included shorter length of stay, unexpected death, and non-oncologic diagnosis. Other patient factors including age, severity of illness (APACHE-IIIJ) score =102 [IQR =80-128]), and organ donation (7%) were not significant. ConclusionThe DECoDE audit tool may be useful for screening of care provided to dying patients in ICU. Not all patients require a high score for high quality care but a low score may indicate poor quality of care, warranting further review.

Matching journals

The top 2 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.