Development of a Predictive Score for Mortality at 3 and 12 Months After Discharge From an Acute Geriatric Unit as a Trigger for Advanced Care Planning

J Gerontol A Biol Sci Med Sci. 2022 Aug 12;77(8):1665-1672. doi: 10.1093/gerona/glab217.

Abstract

Background: There is a need for a mortality score that can be used to trigger advanced care planning among older patients discharged from acute geriatric units (AGUs).

Objective: We developed a prognostic score for 3- and 12-month mortality after discharge from an AGU, based on a comprehensive geriatric assessment, in-hospital events, and the exclusion of patients already receiving palliative care.

Methods: Devenir Après la Médecine Aigue Gériatrique (DAMAGE) is a French multicenter, prospective, cohort study. The broad inclusion criteria ensured that the cohort is representative of patients treated in an AGU. The DAMAGE participants underwent a comprehensive geriatric assessment, a daily clinical checkup, and follow-up visits 3 and 12 months after discharge. Multivariable logistic regression models were used to develop a prognostic score for the derivation and validation subsets.

Results: A total of 3 509 patients were assessed and 3 112 were included. The patient population was very old and frail or dependant, with a high proportion of deaths at 3 months (n = 455, 14.8%) and at 12 months (n = 1 014, 33%). The score predicted an individual risk of mortality ranging from 1% to 80% at 3 months and between 5% and 93% at 12 months, with an area under the receiving operator characteristic curve in the validation cohort of 0.728 at 3 months and 0.733 at 12 months.

Conclusions: Our score predicted a broad range of risks of death after discharge from the AGU. Having this information at the time of hospital discharge might trigger a discussion on advanced care planning and end-of-life care with very old, frail patients. Clinical Trials Registration Number: NCT02949635.

Keywords: Acute geriatric unit; Advanced care planning; Discharge; Mortality; Prognostic score.

Publication types

  • Multicenter Study
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Aged
  • Cohort Studies
  • Geriatric Assessment*
  • Humans
  • Patient Discharge*
  • Precipitating Factors
  • Prospective Studies

Associated data

  • ClinicalTrials.gov/NCT02949635