The effect of baseline cognition and delirium on long-term cognitive impairment and mortality: a prospective population-based study

Alex Tsui, Samuel D. Searle, Helen Bowden, Katrin Hoffmann, Joanne Hornby, Arley Goslett, Maryse Weston-Clarke, Lee Hamill Howes, Rebecca Street, Rachel Perera, Kayvon Taee, Christoph Kustermann, Petronella Chitalu, Benjamin Razavi, Francesco Magni, Devajit Das, Sung Kim, Nish Chaturvedi, Elizabeth L. Sampson, Kenneth RockwoodColm Cunningham, E. Wesley Ely, Sarah J. Richardson, Carol Brayne, Graciela Muniz Terrera, Zoë Tieges, Alasdair MacLullich, Daniel Davis*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

5 Citations (Scopus)
30 Downloads (Pure)


Background: There is an unmet public health need to understand better the relationship between baseline cognitive function, the occurrence and severity of delirium, and subsequent cognitive decline. Our aim was to quantify the relationship between baseline cognition and delirium and follow-up cognitive impairment.

Methods: We did a prospective longitudinal study in a stable representative community sample of adults aged 70 years or older who were registered with a Camden-based general practitioner in the London Borough of Camden (London, UK). Participants were recruited by invitation letters from general practice lists or by direct recruitment of patients from memory clinics or patients recently discharged from secondary care. We quantified baseline cognitive function with the modified Telephone Interview for Cognitive Status. In patients who were admitted to hospital, we undertook daily assessments of delirium using the Memorial Delirium Assessment Scale (MDAS). We estimated the association of pre-admission baseline cognitive function with delirium prevalence, severity, and duration. We assessed subsequent cognitive function 2 years after baseline recruitment using the Telephone Interview for Cognitive Status. Regression models were adjusted by age, sex, education, illness severity, and frailty.

Findings: We recruited 1510 participants (median age 77 [IQR 73-82], 57% women) between March, 2017, and October, 2018. 209 participants were admitted to hospital across 371 episodes (1999 person-days of assessment). Better baseline cognition was associated with a lower risk of delirium (odds ratio 0·63, 95% CI 0·45 to 0·89) and with less severe delirium (-1·6 MDAS point, 95% CI -2·6 to -0·7). Individuals with high baseline cognition (baseline Z score +2·0 SD) had demonstrable decline even without delirium (follow-up Z score +1·2 SD). However, those with a high delirium burden had an even larger absolute decline of 2·2 SD in Z score (follow-up Z score -0·2). Once individuals had more than 2 days of moderate delirium, the rates of death over 2 years were similar regardless of baseline cognition; a better baseline cognition no longer conferred any mortality benefit.

Interpretation: A higher baseline cognitive function is associated with a good prognosis with regard to likelihood and severity of delirium. However, those with a high baseline cognition and with delirium had the highest degree of cognitive decline, a change similar to the decline observed in individuals with a high amyloid burden in other cohorts. Older people with a healthy baseline cognitive function who develop delirium stand to lose the most after delirium. This group could benefit from targeted cognitive rehabilitation interventions after delirium.

Original languageEnglish
Pages (from-to)e232-e241
Number of pages10
JournalThe Lancet Healthy Longevity
Issue number4
Early online date15 Mar 2022
Publication statusPublished - Apr 2022


  • cognitive impairment
  • mortality
  • delirium

ASJC Scopus subject areas

  • Geriatrics and Gerontology
  • Health(social science)
  • Psychiatry and Mental health
  • Family Practice


Dive into the research topics of 'The effect of baseline cognition and delirium on long-term cognitive impairment and mortality: a prospective population-based study'. Together they form a unique fingerprint.

Cite this