Factors associated with the place of death of persons with advanced dementia: A systematic review of international literature with meta-analysis.

Abstract

Background: Many individuals with advanced dementia die in hospital, despite preferring home death. Existing evidence of factors affecting their place of death is inconsistent. To inform policies/practices for meeting needs/preferences, systematically establishing the evidence is pertinent, particularly given the exponential rise in advanced dementia prevalence.

Aim: To identify factors influencing where people with advanced dementia die.

Design and data sources: This systematic review with meta-analysis was registered on PROSPERO (CRD42022366722). Medline, CINAHL, PsycINFO, SocINDEX and a grey literature database, Overton, were searched on 21/12/2022, supplemented by hand-searching/citation tracking. Papers reporting quantitative data on factors associated with place of death in advanced dementia were included and appraised using QualSyst. Data were analysed using random effects with the certainty of evidence determined using the GRADE criteria.

Results: Thirty-three papers involving >5 million individuals (mean age = 89.2 years) were included. Long-term care setting deaths were relatively common but hospice deaths were rarer. Marriage's association with home death underscores social networks' importance, while younger age's and male gender's associations with hospital death demonstrate patients' and families' interdependency. Pneumonia/COPD's opposing effects on hospital deaths with cancer/functional impairment highlight the challenges of advanced dementia care. Unlike hospital/nursing home bed availability's lack of effect, capitated funding (fixed-amount-per-patient-per-period) decreased hospital death likelihood.

Conclusion: This comprehensive review of place of death determinants highlight the profound challenges of advanced dementia end-of-life care. Given that bed capacity did not affect place of death, a capitation-based, integrated palliative care model would appear more likely to meet patients' needs in a resource-constrained environment.