key: cord-0269212-legjd3fj
authors: Whitby, J.; Nitchingham, A.; Caplan, G.; Davis, D.; Tsui, A.
title: Persistent delirium in older hospital patients: an updated systematic review and meta-analysis
date: 2022-01-21
journal: nan
DOI: 10.1101/2022.01.20.22269044
sha: 7b0790f8745db1ff19b0aa7c1fa7f65076e2e1bf
doc_id: 269212
cord_uid: legjd3fj

Introduction Persistent delirium is recognised as a substantial problem, but there are few insights into which patient groups might be particularly affected. 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. Methods We adopted an identical approach to a previous systematic review, only including studies using a recognised diagnostic framework for ascertaining delirium at follow-up (persistent delirium). We applied risk of bias assessments based on Standards of Reporting of Neurological Disorders criteria. Estimates were pooled across studies using random-effects meta-analysis, and we estimated associations with age and follow-up duration using meta-regression. Results We identified 13 new cohorts, which we added to 8 from the previous systematic review (21 relevant studies reporting persistent delirium over 37 time points). Studies were mainly at low risk of bias. Pooled delirium prevalence estimates at discharge were 40% (95% CI 24% to 54%, 10 studies). Meta-regression showed variation in prevalence of persistent delirium over time (0.6% per week, 95% CI -1.2 to -0.1, p=0.02). Older study sample age was associated with higher prevalence of persistent delirium (8.8% per SD age, 95% CI 1.1% to 17%). The rate of improvement was the same regardless of age, but the overall burden was higher with older age such that 44% (95% CI 11% to 76%) of 95-year-olds would be expected to have persistent delirium at 12 months. Conclusions This systematic review emphasises the key importance of delirium as a persistent and extensive problem, particularly in the oldest-old. Addressing persistent delirium will require a whole-system, integrated approach in order to detect, follow-up and implement opportunities for recovery across all healthcare settings.

Delirium is widely recognised as having a substantial impact on healthcare for older people, with several clinical guidelines serving to address its detection, prevention and management in hospitals. 4, 5 However, many questions remain over the natural history of delirium, particularly its relationship with dementia. 6 For example, it is not clear if underlying dementia leads to a more severe or prolonged delirium. 7 We know that delirium is associated with future dementia progression. [1] [2] [3] Yet whether this occurs subclinically over months and years, or persistent delirium merges into worsened dementia is not understood.

Persistent delirium has been of interest since a systematic review from 2009 suggested one in five cases were still evident six months after discharge. 8 Subsequently, the number of older people presenting for urgent and emergency care has increased. 9 There has also been a consistent trend for more acute presentations of people living with dementia; around half are admitted within the first 12 months of diagnosis. 10 In light of these changes, we set out to update the systematic review to provide current estimates for the epidemiology of persistent delirium.

We followed the 2020 PRISMA guidance. 11 We used the same criteria applied in the previous systematic review: (i) study population of at least 20 hospital patients; (ii) patients aged ≥ 50 years; (iii) prospective study with follow-up of at least one week; (iv) acceptable definition of delirium at enrolment. We excluded studies investigating delirium in critical care and in the context of terminal illness or palliative care. Our only modification was to require studies to use a recognised diagnostic framework for ascertaining delirium at follow-up (persistent delirium); studies from the original systematic review meeting this criterion were carried over into the current analysis. Given this was an update of a previous systematic review, we did not devise a de novo protocol for PROSPERO.

The primary outcome was proportion of patients with delirium at follow-up, where the denominator was the number of participants who had delirium at inception. We considered any definition based on the Diagnostic and Statistical Manual (DSM), International Classification of Disease, the Confusion Assessment Method (CAM) or the Delirium Index to be sufficiently detailed to ascertain delirium reliably.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 21, 2022. 

Covidence (www.covidence.org, Veritas Health Innovation Ltd.) was used to manage the abstract and full-text screening, assessing risk of bias and data extraction. Three researchers independently reviewed titles and abstracts (J.W., A.N., A.T.) to determine the eligibility. Conflicts were resolved by discussion and consensus. The same reviewers extracted data using a pro forma.

We determined risk of bias based on the framework outlined by the Standards of Reporting of Neurological Disorders (STROND) criteria. 12, 13 We considered bias arising (i) from specific patient settings, e.g., general medical patients, cohort with intracerebral haemorrhage; (ii) from sample selection, e.g., convenience, consecutive; (iii) from sample criteria, e.g., excluding patients unable to consent, assessed as being too sick; (iv) from reference standard used, e.g., DSM, CAM; (v) from expertise in applying the reference standard, e.g. routine data, dedicated researcher. We rated studies high, low or unclear risk of bias in each of the five domains. Quality of evidence was assessed using the GRADE framework, which considers risk of bias, inconsistency, indirectness, imprecision and publication bias.

We extracted summary statistics for the prevalence of persistent delirium at each time point and calculated their standard errors (sqrt [p (1-p) / n)]. We assumed methodological heterogeneity across studies, accounting for this using DerSimonian-Laird random-effects models. 14 Statistical heterogeneity was assessed with the I² statistic. Meta-regression was used to estimate the relationship between persistent delirium and follow-up time, mean age, and dementia prevalence in each study sample. We plotted marginal predictions for any parameter associated with persistent delirium. To assess publication bias, we plotted the estimated proportion of delirium occurrence against the standard error of that estimate and . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269044 doi: medRxiv preprint inspecting the degree of asymmetry. We used Stata 16.1 (StataCorp, Texas) for all analyses.

We identified 6474 articles, screening 5556 after removal of duplicates (see PRISMA diagram). We assessed 115 full-text articles for eligibility. We identified 13 new cohorts, which we added to eight from the previous systematic review, giving a total of 21 relevant studies reporting persistent delirium over 37 time points (Table 1) . [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] Most studies were in medical patients and they were all from high-income settings and casemix by ethnicity was not generally reported. Samples ranged in size (n=23 to n=278) and duration of follow-up (up to 12 months). Average age and proportion with dementia were reported as 82.3 years and 43% respectively, though both varied substantially between different study cohorts.

In the main, studies were at low risk of bias insofar as most followed a sample representative of the target setting (e.g., acute hospital care) and used consistent outcome ascertainment instruments (GRADE certainty rating for risk of bias = 'low') ( Figure 2 ). Around half excluded patients unable to give consent (and did not report procedures to allow proxy consent) and/or excluded participants too sick to assess for delirium. Case ascertainment was based on a consensus definition (e.g. DSM) in 6 studies; the rest used instruments such as CAM.

Given each has different degrees of restrictiveness in their definition, 35 the GRADE certainty rating for indirectness was 'moderate'.

Pooled delirium prevalence estimates at discharge were 40% (95% CI 24% to 54%, 10 studies) (Figure 4 ). Although there were few studies with data beyond 6 months, each time point reported considerable persistent cases (pooled estimates ranging from 20% to 59%).

There was substantial statistical heterogeneity at all time points (all subgroup I² >89%) (GRADE certainty rating for imprecision = moderate). Funnel plots of prevalence versus the standard error of the estimate did not demonstrate any asymmetry, leading to a GRADE certainty rating for publication bias of 'high'.

Meta-regression showed variation in prevalence of persistent delirium over time (0.6% per week, 95% CI -1.2 to -0.1, p=0.02) ( Table 2 ). This monotonic decrease in prevalence over time led to a GRADE certainty rating for inconsistency as 'low'. Older study sample age was associated with higher prevalence of persistent delirium (8.8% per SD age, 95% CI 1.1% to 17%). Persistent delirium did not vary with the proportion of study sample with dementia, though only 14 studies reported this. Figure 4 shows the predicted meta-regression margins . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269044 doi: medRxiv preprint for three age groups over time. The rate of improvement was the same regardless of age, but the overall burden was higher with older age such that 44% (95% CI 11% to 76%) of 95year-olds would be expected to have persistent delirium at 12 months. The overall GRADE certainty for this estimate is 'moderate'.

We showed that persistent delirium remains a substantial problem well beyond the acute phase. This persistence does not appear to vary with baseline dementia, albeit using a study-level ecological measure of dementia. A clearer association was evident with increasing age. Taken together, these findings suggest a considerable need to focus delirium recovery for both individuals and health services.

Our data should be considered in light of a number of limitations. Other than duration, the variables we could use for meta-regression relied on study-level data on mean sample age and dementia prevalence. This limits the interpretation due to residual confounding for quantities such as frailty. Most studies had significant attrition due to mortality and loss to follow up, though this might be expected to underestimate the true prevalence of persistent delirium. All studies were performed in high-income countries, making it difficult to generalise our findings to other settings. Many studies only included delirium detected in the first 24-48 hours of admission, which could have led to under-ascertainment. There is likely to be a degree of overlap between the reports of delirium at discharge and some of the earlier follow-up time points (2-weeks and 4-weeks) because the former category encompassed a variable duration. Some publications only reported persistent delirium at discharge, so it is impossible to know how long these patients remained as cases. Similarly, it is unclear how long delirium might have been present prior to transfer for individuals admitted to post-acute care. We also restricted our search to studies published in English. Finally, we detected significant heterogeneity between studies, though we attempted to limit this by defining our outcome measure as precisely as possible.

In updating the original systematic review with 15 years' worth of new studies, we have extended those findings to demonstrate the highest prevalence of persistent delirium with older age. While older age is likely to be a proxy for dementia and frailty, the gradient suggests that delirium may never resolve in the oldest-old. More broadly, these data are consistent with terminal decline in cognition observed longitudinal studies. 36 Higher education is associated with delayed onset, though a faster trajectory once terminal decline begins. 37 Though previous cohorts have not considered delirium to be a driver, the Delirium . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269044 doi: medRxiv preprint and Population Health Informatics Cohort has that shown the largest post-delirium decline in cognition occurs in those with higher baseline cognition. 38 Dedicated prospective studies are needed to fully capture the influence of evolving dementia reciprocally affecting persistent delirium.

The clinical implications of our findings indicate an urgent need to develop delirium recovery and cognitive rehabilitation services. These are almost non-existent, certainly out of proportion with the potential demand, though multicomponent interventions show promise. 39 Recognising the opportunities for the emerging field of interface acute geriatrics would be an important starting point, 40 and continued community-based management of delirium is likely to have an impact. 41 This systematic review emphasises the key importance of delirium as a persistent and extensive problem, particularly in the oldest-old. Addressing persistent delirium will require a whole-system, integrated approach in order to detect, follow-up and implement opportunities for recovery across all healthcare settings.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269044 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269044 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269044 doi: medRxiv preprint

Delirium is a strong risk factor for dementia in the oldest-old: a population-based cohort study. Brain

Delirium and Long-term Cognitive Trajectory Among Persons With Dementia

Recurrent delirium over 12 months predicts dementia: results of the Delirium and Cognitive Impact in Dementia (DECIDE) study. Age Ageing

The Scottish Intercollegiate Guidelines Network: risk reduction and management of delirium

Nat Rev Dis Primers

The epidemiology of delirium: challenges and opportunities for population studies

Delirium superimposed on dementia

Persistent delirium in older hospital patients: a systematic review of frequency and prognosis

Emergency department use by older adults: a literature review on trends, appropriateness, and consequences of unmet health care needs

Hospitalisation of people with dementia: evidence from English electronic health records from 2008 to 2016

Explanation and Elaboration of the Standards of Reporting of Neurological Disorders Checklist: A Guideline for the Reporting of Incidence and Prevalence Studies in Neuroepidemiology

Development of the Standards of Reporting of Neurological Disorders (STROND) checklist: A guideline for the reporting of incidence and prevalence studies in neuroepidemiology

Meta-analysis in clinical trials

The occurrence and persistence of symptoms among elderly hospitalized patients

The natural history of delirium in older hospitalized patients: a syndrome of heterogeneity

Delirium is independently associated with poor functional recovery after hip fracture

Characteristics associated with delirium persistence among newly admitted post-acute facility patients

Older adults discharged from the hospital with delirium: 1-year outcomes

The course of delirium in acute stroke

Delirium in long-term care setting: indicator to severe morbidity

Frequency, risk factors, and prognosis of prolonged delirium in elderly patients after hip fracture surgery

Different functional outcomes in patients with delirium and subsyndromal delirium one month after hospital discharge

The neuropsychological sequelae of delirium in elderly patients with hip fracture three months after hospital discharge

Partial and No Recovery from Delirium in Older Hospitalized Adults: Frequency and Baseline Risk Factors

Undiagnosed longterm cognitive impairment in acutely hospitalised older medical patients with delirium: a prospective cohort study

Delirium among elderly patients admitted to a post-acute care facility and 3-months outcome

Quantifying the Severity of a Delirium Episode Throughout Hospitalization: the Combined Importance of Intensity and Duration

Partial and no recovery from delirium after hospital discharge predict increased adverse events

Impact of Delirium on Outcomes After Intracerebral Hemorrhage. Stroke

The consistent burden in published estimates of delirium occurrence in medical inpatients over four decades: a systematic review and metaanalysis study

Education associated with a delayed onset of terminal decline

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

Interventions to support recovery following an episode of delirium: A realist synthesis

Follow-up services for delirium after COVID-19-where now? Age Ageing

Does home treatment affect delirium? A randomised controlled trial of rehabilitation of elderly and care at home or usual treatment (The REACH-OUT trial)