Persistent delirium in older hospital patients: an updated systematic review and meta-analysis

Abstract

Introduction: Delirium is associated with future dementia progression. Yet whether this occurs subclinically over months and years, or persistent delirium merges into worsened dementia is not understood. Our objective was to estimate the prevalence of persistent delirium and understand variation in its duration.

Methods: We adopted an identical search strategy to a previous systematic review, only including studies using a recognised diagnostic framework for ascertaining delirium at follow-up (persistent delirium). Studies included hospitalised older patients outside critical and palliative care settings. We searched MEDLINE, EMBASE, PsycINFO and the Cochrane Database of Systematic Reviews on 11th January 2022. We applied risk of bias assessments based on Standards of Reporting of Neurological Disorders criteria and assessed strength of recommendations using the grading of recommendation, assessment, development and evaluation (GRADE) approach. Estimates were pooled across studies using random-effects meta-analysis, and we estimated associations with follow-up duration using robust error meta-regression.

Results: We identified 13 new cohorts, which we added to 10 from the previous systematic review (23 relevant studies, with 39 reports of persistent delirium at 7 time-points in 3186 individuals admitted to hospital care (mean age 82 years and 41% dementia prevalence). Studies were mainly at moderate risk of bias. Pooled delirium prevalence estimates at discharge were 36% (95% CI 22% to 51%, 13 studies). Robust error meta-regression did not show variation in prevalence of persistent delirium over time (-1.6% per month, 95% CI -4.8 to 1.6, p=0.08). Margins estimates for this model indicate a prevalence of persistent delirium of 16% (95% CI 6% to 25%) at 12 months.

Conclusions: This systematic review emphasises the importance of delirium as a persistent and extensive problem (GRADE certainty = moderate), raising questions on chronic delirium as a clinical entity and how it might evolve into dementia. Addressing persistent delirium will require a whole-system, integrated approach to detect, follow-up and implement opportunities for recovery across all healthcare settings.

Keywords: meta-regression; older people; persistent delirium; systematic review.

Figure 1. PRISMA flow diagram detailing study selection process.

* One study included in original systematic review could not be directly retrieved, but it was possible to use data secondary reports. Reasons for exclusion at full

Figure 2. Forest plot showing pooled estimates of persistent delirium, stratified by study follow-up period.

NOTE: Weights are from random-effects model

Figure 3. Robust error meta-regression showing estimated prevalence of persistent delirium over time.

Solid line shows fitted meta-regression, with 95% CI (dashed lines). Circles are study estimates, proportional to sample size (inverse variance weighting).