Survival time after marked reduction in oral intake in terminally ill noncancer patients: A retrospective study

'How long do you think?' Unresponsive dying patients in a specialist palliative care service: A consecutive cohort study

BACKGROUND

Predicting length of time to death once the person is unresponsive and deemed to be dying remains uncertain. Knowing approximately how many hours or days dying loved ones have left is crucial for families and clinicians to guide decision-making and plan end-of-life care.

AIM

To determine the length of time between becoming unresponsive and death, and whether age, gender, diagnosis or location-of-care predicted length of time to death.

DESIGN

Retrospective cohort study. Time from allocation of an Australia-modified Karnofsky Performance Status (AKPS) 10 to death was analysed using descriptive narrative. Interval-censored survival analysis was used to determine the duration of patient's final phase of life, taking into account variation across age, gender, diagnosis and location of death.

SETTING/PARTICIPANTS

A total of 786 patients, 18 years of age or over, who received specialist palliative care: as hospice in-patients, in the community and in aged care homes, between January 1st and October 31st, 2022.

RESULTS

The time to death after a change to AKPS 10 is 2 days (n = 382; mean = 2.1; median = 1). Having adjusted for age, cancer, gender, the standard deviation of AKPS for the 7-day period prior to death, the likelihood of death within 2 days is 47%, with 84% of patients dying within 4 days.

CONCLUSION

This study provides valuable new knowledge to support clinicians' confidence when responding to the 'how long' question and can inform decision-making at end-of-life. Further research using the AKPS could provide greater certainty for answering 'how long' questions across the illness trajectory.

Prediction Models for Impending Death Using Physical Signs and Vital Signs in Noncancer Patients: A Prospective Longitudinal Observational Study

Background: Accurate information on the prognosis in the last days of life is essential for providing better end-of-life care; however, few studies have examined the signs of impending death (SID) or developed short-term prediction models in noncancer patients.

Objective: To investigate the prevalence and onset of SID and to develop models that predict death within 7 days, 72 hours, and 24 hours in noncancer patients.

Design: This is a prospective longitudinal observational study.

Setting/Subjects: Subjects were noncancer patients admitted to a hospital in Japan between 2019 and 2020.

Measurements: We investigated 11 physical signs and vital signs every 12 hours until death after confirming a reduced daily oral intake to less than a few mouthfuls.

Results: We analyzed data from 50 noncancer patients. The prediction model “pulselessness of the radial artery OR respiration of mandibular movement OR the shock Index (SI) >1.0” predicted death within 7 days with an accuracy of 83.9%, whereas the models developed to predict death within 72 and 24 hours had an accuracy of 65.0% or less. The median onset of all signs was within 3 days of death. The frequencies of decreased response to verbal stimuli and decreased response to visual stimuli were 76.0% and 74.0%, respectively.

Conclusions: The prediction model using physical signs and SI predicted death within 7 days in noncancer patients with high accuracy. The prediction of death within 72 and 24 hours in noncancer patients requires investigation of physical signs not examined in this study.