Hospital outcomes of older people with cognitive impairment: An integrative review

1People with cognitive impairment have higher hospital mortality, a higher incidence of delirium, and longer hospital stays than patients with no cognitive impairment. In addition, intermediate outcomes such as dehydration, reduction in nutritional status, pain, decline in physical and cognitive function, and new infections in hospital may contribute to poorer final hospital outcomes, but have been less well described than in patients with a formal dementia diagnosis.It is important to identify whether older people in hospital have cognitive impairment or an existing diagnosis of dementia, to be aware of their increased susceptibility to adverse events in the hospital environment, and to provide appropriate surveillance for intermediate outcomes to prompt preventative action.Further studies of outcomes for people with cognitive impairment in hospital should consider the care environment, such as ward type, staffing, and episodes of being located outside their designated ward (“outlying”), as these factors may also influence outcomes.Development and use of core outcome sets for people with cognitive impairment is essential to fully understand and describe the patient journey both to evaluate day‐to‐day care, and for use in observational or interventional research.

2People with cognitive impairment have higher hospital mortality, a higher incidence of delirium, and longer hospital stays than patients with no cognitive impairment. In addition, intermediate outcomes such as dehydration, reduction in nutritional status, pain, decline in physical and cognitive function, and new infections in hospital may contribute to poorer final hospital outcomes, but have been less well described than in patients with a formal dementia diagnosis.

3It is important to identify whether older people in hospital have cognitive impairment or an existing diagnosis of dementia, to be aware of their increased susceptibility to adverse events in the hospital environment, and to provide appropriate surveillance for intermediate outcomes to prompt preventative action.

4Further studies of outcomes for people with cognitive impairment in hospital should consider the care environment, such as ward type, staffing, and episodes of being located outside their designated ward (“outlying”), as these factors may also influence outcomes.

5Development and use of core outcome sets for people with cognitive impairment is essential to fully understand and describe the patient journey both to evaluate day‐to‐day care, and for use in observational or interventional research.

6Between 25% and 40% of older people admitted to acute hospitals have been diagnosed with dementia, (eg, Alzheimer's disease, dementia syndrome according to Diagnostic and Statistical Manual of Mental Disorders (DSM) IV, etc.) or have evident cognitive impairment because of undiagnosed dementia or another cause.1, 2 People with dementia occupy approximately 25% of hospital beds in the UK, stay up to 6 times longer than other older patients, and have a greater risk of dying in hospital; however, outcomes for people with any cause of cognitive impairment (CI) are less well described.3, 4 Poor hospital outcomes, eg, death or new discharge to a residential home, may occur following a series of less frequently reported outcomes which patients with CI may be more likely to experience in hospital. These intermediate outcomes may be an appropriate focus of attention to target nursing and other care and treatment, as, to reduce these outcomes, we must first understand how and why these patients deteriorate in hospital and identify the specific risk factors at patient and hospital level. Knowing how day‐to‐day clinical and well‐being outcomes for patients with CI differ from those with no CI during hospitalisation could help us identify specific areas of prevention or care which could improve the journey, and therefore the final outcome, for these patients.

7Dementia is significantly underdiagnosed in the community, and delirium and CI often pass undetected in hospital.5, 6 A full diagnostic assessment for dementia during an acute hospital admission for all older people is neither appropriate nor feasible. However, simple cognition screening tests can be used to detect CI, eg, the Abbreviated Mental Test Score for cognitive function or the Confusion Assessment Method for delirium.7, 8, 9 Studies of acutely hospitalised older people using systematically applied screening tests for CI have highlighted that a significant proportion do not have a dementia diagnosis, but patients with CI experience rates of hospital outcomes, eg, mortality, more similar to those of patients with dementia than patients with no CI.10, 11 Greater understanding of the outcomes of older people with various causes of CI should inform how we can improve care for the whole population at risk. There are currently no published reviews in this area.

8This review aims to summarise existing available evidence about the outcomes of older patients with cognitive impairment admitted to hospital, specifically to establish which outcomes have been investigated, the additional risk of outcomes in people with CI, and factors that may influence outcomes.

9Integrative review methodology enables inclusion of a broad range of study designs and nonresearch literature, eg, audits and theoretical perspectives.12, 13 The method summarises findings with mixed narrative and tabular presentation, identifies common themes in study results, and highlights inconsistencies, without numerical synthesis.

10MEDLINE, Cumulative Index to Nursing and Allied Health Literature, PsycINFO and EMBASE, AgeInfo, and the Cochrane Library were searched. Terms used (eg, medical subject headings) to describe the population included (1) demographic group: “Aged, hospitalised”, “aged hospital patient”, aged, geriatric, senior; (2) clinical group: “cognition disorders”, dementia, “Alzheimer's disease”, “cognitive impairment”, “cogniti* impair*”, “cognitive defect” “delirium/dementia/amnestic, cognitive disorders”, “frontotemporal dementia”, “dementia vascular”, “dementia, multi‐infarction”, “Lewy body dementia”, “dementia, senile”; and (3) health service use group: “hospital admission”, hospital*. (See supplementary material). Additional evidence was retrieved by reviewing reference lists, forward citation searches, and searching websites of organisations focussing on the care of older people, eg, Age UK, British Geriatrics Society, Royal College of Nursing, Alzheimer's Society, and Alzheimer's UK.

11Studies included were those which investigated (i) outcomes of older people with CI with a hospital admission as a main purpose of the study, or (ii) the contribution of CI to an outcome of interest related to hospitalisation, including other disease outcomes, surgical, or medical treatments, or (iii) outcomes of people with CI in intensive care units during hospital admission, where the outcomes occurred during hospitalisation or at discharge. The search was limited to articles published in the last 20 years (since 1997) as these will reflect contemporary service provision, care practices, and up‐to‐date methods of detecting dementia/CI. Studies which reported on outcomes of Emergency Department visits only, elective surgical patients, patients with delirium with no evidence of prior CI, and those taking place within specialised psychogeriatric units were excluded.

12Titles and abstracts were screened for review aims. Full texts were obtained for potentially relevant articles and screened against eligibility criteria. Screening and data extraction was undertaken by a single reviewer, and decisions checked with a second reviewer in case of uncertainty. The relevance of all included studies was verified by 3 reviewers. As one of the purposes of this review was to understand which outcomes are being measured for this population in hospital, no formal quality assessment was performed to maintain inclusivity. Methodological issues, eg, the potential for bias, are indicated in text or tables where appropriate.

13One thousand sixty‐two records were identified from database searches, reference lists, and website searches. Following review of abstracts and full papers against eligibility criteria, 104 articles were included in the review (Figure 1). The median number of participants was 498 (range 4‐3 000 000), mostly of people aged ≥50. Participant cohorts included general inpatients, specific conditions, eg, heart failure or fractures, or with specific clinical interventions, eg, catheterisation. CI was defined in several ways, eg, of dementia diagnosis, cognitive spectrum disorder (delirium, dementia, or Abbreviated Mental Test <8), or other assessments, eg, Short Blessed Test.

14The articles encompass a range of methodologies, eg, observational studies comparing patients with/without CI, studies in which cognitive status or dementia were evaluated as risk factors for specific outcomes, and qualitative studies and audits. A variety of outcomes were explored, not only in patients with dementia compared to those without but also in patients with measurable CI regardless of diagnosis. Associations between CI and outcomes were assessed using a variety of covariates, reflecting the study context and data sources available. Articles with more than 1 outcome are presented in the appropriate tables.

15An integrative review summarising 24 papers on patient and carer experience concluded that people with dementia are stigmatised in hospitals, and acute care needs and tasks are prioritised over personalised care14 (Table 1). The UK National Audit of Dementia Care found that 17% of comments about patient care (collected via a carer questionnaire) described care negatively, and 9% expressed that patient did not receive care appropriate to their needs.15 Surveys estimate that around 60% of people with dementia are not treated with dignity or understanding whilst hospitalised, and the majority are frightened by the hospital environment.3 Reporting of negative experiences has been observed to follow a model, the “cycle of discontent”, in which poor communication and relationship building between staff and patients/carers lead to expectations not being met, subsequent cycles of identification of poor care and challenge to staff, further deterioration in the relationship, and ultimately reporting of poor experiences.16 It has been observed that there are many missed opportunities in hospitals to provide person‐centred care and enable a person with dementia to sustain personhood.17 No studies were found that discussed experiences of older patients with any cause of CI.

16The prevalence of behavioural and psychological symptoms of dementia (BPSD) symptoms in people with dementia in hospital rises during admission, likely because of unmet needs and distress, and a higher overall Behavioural Pathology in Alzheimer Disease Scale (BEHAVE‐AD score) (incorporating BPSD) associated with increased mortality.18 Behavioural and psychological symptoms of dementia have been identified as a frequent cause of complications in an Alzheimer Special Acute Care inpatient unit, with agitation and aggressiveness representing 60% of BPSD events.19 A qualitative study identified disruption in routine, eg, admission to hospital, triggering negative changes in behaviour as the person with dementia attempts to gain control over an unfamiliar environment.20

17Older people with dementia are more likely to have a low Mini‐Nutritional Assessment (MNA) score and laboratory indices indicating malnutrition at hospital admission, with overall MNA score and subscore related to dietary habits (MNA‐3) significant predictors of death in hospital.21 Of admitted patients who are already undernourished, those with CI are less likely to meet their required energy and protein intake, achieving <50% of total energy expenditure requirements.22 Organisational factors may contribute to decline in nutritional status through lack of availability of adequate nutrition. An audit revealed that only 76% of staff considered people with dementia had their nutritional needs met “always or most of the time”, and <75% of staff said that they could obtain snacks between meals for patients with dementia, who were unable to eat full meals at regular times.15 Fluid intake is also a key indicator of fundamental care in hospital. Assessment of renal conservation of water in older patients showed concentrated urine in 16% of patients, more commonly in patients with confusion and/or dementia, and was related to higher 30‐day mortality.23

18A meta‐analysis identified functional decline (measured by activities of daily living (ADL), instrumental ADLs (IADL), Barthel index (BI), mobility, functional independence measure (FIM), or Rankin scale) in hospitalised adults aged ≥65 is independently associated with CI or a dementia diagnosis.24 Further cognitive decline during hospitalisation is associated with an increased risk of functional decline, defined as a loss of ability to perform 1 or more ADLs without help between admission and discharge.25

19The prevalence of delirium in general hospital patients is around 20%, and approximately half these patients have pre‐existing dementia.6 Although patients with dementia are more likely to have delirium at admission, dementia increases the likelihood of new‐onset delirium (or “delirium superimposed on dementia” (DSD)) during hospitalisation.26, 27, 28, 29 Regardless of a dementia diagnosis, lower cognitive scores are associated with increased occurrence of delirium in hospital, and symptoms of greater severity, eg, disordered attention, orientation, thought organisation, and memory.27, 30, 31, 32, 33 Cognitive impairment and dementia are predictive of delirium occurring prior to or following surgery for fractures of the hip or proximal femur.34, 35, 36, 37 Hospital outcomes including mortality, institutionalisation, and length of stay for patients with delirium are worse with pre‐existing dementia.38, 39, 40, 41 Dementia was associated with an increased risk of least 1 episode of delirium during the first 3 days of admission in adults aged ≥65, and increased the odds of unanticipated ICU admission or in‐hospital death.42

20Events occurring during hospitalisation, eg, urinary tract infections (UTI), pneumonia, or gastroenteritis (hospital‐acquired infections (HAI)), pressure ulcers (PU), adverse drug reactions (ADR) falls, and fractures impair recovery by reducing mobility, functional ability, and nutritional status, increase care required, and extend hospitalisation. Cognitive impairment or dementia leads to an increased risk of falls in hospital,43, 44 including recurrent falls45 and falls related to impulsive behaviour.46 In addition, factors identified in >75% of falls in patients with dementia included being in hospital at night, acute disease or symptoms of disease, and/or acute drug side effects.47 Falls may result in fractures, which delay recovery and lengthen hospitalisation. Occurrence of fractures in patients with dementia is associated with hypnotic medicines, specifically short‐acting benzodiazepine hypnotics, ultrashort‐acting nonbenzodiazepine hypnotics, hydroxyzine, risperidone, and perospirone.48 Both medical and surgical inpatients with dementia are at higher risk of 4 common complications, UTIs, PUs, pneumonia, and delirium, and medical patients are also at increased risk from sepsis and “failure to rescue”.49 Pressure ulcers are also more common in patients with CI.27 Cognitive impairment was shown to be the most significant risk factor for developing urinary and faecal incontinence,43 with 36% and 2% new incontinence at discharge respectively.50

21Polypharmacy (≥5 drugs/day) and dependence for at least 1 ADL were related to occurrence of at least 1 ADR in inpatients with dementia.51 Cognitive impairment in older people is associated with increased HAIs, ADRs, and length of stay ≥7 days52 Adverse drug reactions may be less frequently reported in patients with CI, because of reduced ability to recognise and communicate side effects, leading to unsafe care.53 However, older patients with CI may be less likely to use inappropriate medication (as per Beers criteria), thus reducing ADR reporting in this group.54 A study exploring the relationship of adverse clinical events (ie, any acute clinical problem that occurs newly during hospitalisation) and mortality in patients with dementia showed at least 1 adverse clinical event (eg, electrolyte disorders, hypertensive crisis, fractures, or infections) increased the risk of death 10‐fold.55 Mild/moderate CI was associated with adverse events defined as “incidents” (eg, following an unintended “accident” in hospital such as a slip or trip, medication error, or staff miscommunication), but not subsequent mortality.56

22Inpatients with dementia have a higher risk of acute organ dysfunction and severe sepsis, particularly patients with comorbidities such as chronic obstructive pulmonary disease (COPD).57, 58 Inpatients with COPD and dementia were less likely to be receiving treatment for COPD and to have their lung function assessed, suggesting that undertreatment could contribute to poorer outcomes.59

23Pressures on hospital beds lead to older people not always being placed in the most suitable location for their care: known as “outlying” or “boarding”. These patients may be moved around the hospital several times until they reach their “home ward”. Of patients under an Older Person Evaluation Review and Assessment team, who were more likely to be boarding than general medicine patients, those with pre‐existing CI were more likely to be moved 3 or times during their hospital admission (Table 2).60 In a further study, boarding patients with dementia and/or delirium had higher mortality within 48 hours of admission.61 Although hospital organisational factors result in night‐time bed moves, these were deemed avoidable by 50% of staff surveyed in an audit, and considered detrimental to patient experience.15

24Pain may indicate a new infection, injury, or worsening in condition. The prevalence of pain amongst inpatients with CI is estimated at 39 and is associated with increases in the BEHAVE‐AD score, and increased aggression, phobia, and anxiety.62 Dementia reduces a patient's ability to describe pain characteristics and changes, thus delaying diagnosis of infections or overtreating with analgesics like opioids, contributing to complications, eg, delirium, bowel problems, and lengthened stay.63 There is no current evidence as to whether patients with CI experience more pain during hospitalisation, probably because of difficulties in assessment.

25End‐of‐life patients with dementia have fewer referrals to palliative care and have less prescribed palliative medicines, although no differences were found in 1 study comparing patients with terminal dementia to terminal heart failure.64, 65, 66 Whereas invasive interventions were equally utilised in 1 study, arterial blood gas measurement and catheterisation were more frequent for patients with dementia, and central line placement less used in another study.64, 65 Drug withdrawal rates in hospitalised end‐of‐life patients with dementia were higher than for patients with COPD or heart failure.67 In patients with terminal dementia, only 46% had adequate symptom control, with 13.5% experiencing uncontrolled pain and 51.5% dyspnoeic.66 In an evaluation of suffering at end of life in patients with dementia using the Mini Suffering State Examination scale, which includes psychological distress, spiritual concerns, and physical pain, only 7% of patients died with the lowest level of suffering, with the majority experiencing significant suffering, highlighting insufficient assessment and palliative treatment.68

26Catheterisation could indicate deterioration in a person with CI in hospital, a sign of poor care (if inappropriately performed), or reduction in the ability of staff to provide effective care. The presence of CI was related to inappropriate catheterisation in older patients, with “convenience of care” cited in 50% of cases, and led to a greater decline in ADLs during admission.69

27Of 11 studies comparing mortality in general inpatients with/without dementia, 8 concluded that patients with dementia have an increased risk of death, with estimates varying from adjusted odds ratio (aOR) 1.09 [1.03‐1.16] to aOR 2.1 [1.0‐4.5] (Table 3).5, 55, 70, 71, 72, 73, 74, 75, 76, 77 This difference is greater in people >65 years with dementia as compared to older patients (aOR 1.93 [1.55‐2.41]).71 Inpatients with COPD and dementia have a higher mortality risk.58 Moderate and severe CI was associated with mortality after ICU admission, even adjusting for acuity scores (Acute Physiology and Chronic Health Evaluation II.78 A large cohort demonstrated significant differences in mortality for patients with CI but no diagnosis of dementia as compared to patients with no CI (11.8% vs 9.0%), and a further study showed a difference between “all‐cause” CI and no CI (13.6% vs 9.0%).10, 11 The presence of CI, regardless of dementia, may independently predict in‐hospital mortality, with the highest risk in patients with severe CI.5, 79

28Studies which have not shown a difference in mortality between people with/without dementia include a stratified analysis by occurrence of delirium, and 1 study excluding patients with sensorial deficits, communication problems, or severe acute illness, ie, a higher mortality risk.28, 29, 41, 80 A systematic review concluded that although cognitive function was a predictor of in‐hospital mortality in 6 of 12 studies assessed, assessments of physical function and nutrition were also important in older patients.81 In patients aged ≥80, functional status and comorbidities were predictive of poor outcomes, whereas dementia or other CI was not.82, 83 Studies exploring the contribution of CI to mortality have been adjusted for a range of covariates, eg, functional/nutritional assessments, comorbidities, and laboratory indicators, which influence estimates of effect.

29Contradictory findings regarding the contribution of dementia to mortality in patients presenting to hospital with acute myocardial infarction (AMI) could relate to variation in care provision, as patients with dementia report less chest pain and wait longer for treatment, have fewer transfers to intensive or coronary care units, and less frequent provision of invasive interventions.84, 85, 86 Dementia was not found to be associated with hospital mortality in patients with stroke, or those with an ICU admission.87, 88

30In most studies, CI or dementia increased length of hospital stay (LOS).10, 11, 71, 73, 77, 89, 90, 91, 92, 93, 94, 95, 96, 97 (Table 4) Patients with DSD had longer mean LOS than those with dementia or delirium alone.11 Concurrent dementia extends stays in older patients with hip fracture98 and haemorrhagic peptic ulcer disease.99 However, similar LOS was described in 1 article, and comorbidities found more predictive of longer hospital stays in another study.76, 82 Discharge after the patient is “medically fit”, because of delays in discharge planning or difficulties in organising residential care, contribute to longer LOS in people with CI,90, 100 in addition to mental and behavioural manifestations, falls, or hospital‐acquired complications.45, 101, 102 LOS was longer in patients with Parkinsonism‐related dementia or vascular dementia than Alzheimer's, and patients with concurrent diabetes mellitus, pneumonia, and fall‐related hip fracture had more hospital stays of >14 days.103

31Excess costs relating to increased LOS for patients with dementia exceeded £80 million, and dementia estimated to increase the average cost of an admission 3‐fold (UK figures, 2011).77, 96 Cognitive impairment (dementia or delirium coded during admission) increased costs of hospital stay by 51% in Australia, and 39% for dementia alone.91, 95 In Ireland, dementia adds 246 908 hospital days per annum, costing €199 million.92 Dementia was associated with increased treatment costs of $1171 for endoscopic haemostasis of hemorrhagic peptic ulcer.99 Patients with dementia experiencing complications accounted for 10.4% of hospital episodes, and 22% of extra costs.102 Large numbers of patients with dementia die in hospital, where costs for end‐of‐life care can be 6 times higher than hospice/home care,104 although appropriate management, eg, palliative care consultations, reduces pharmacy costs through prescribing changes.105

32Patients with CI are frequently discharged to nursing/residential homes.10, 77, 82, 89, 106 Dementia predicts of institutionalisation (odds ratio 2.14 [1.24‐3.70]), although less so in ambulatory care sensitive conditions.71, 82, 107, 108 However, in stroke patients, no difference in discharge disposition was found between patients with/without dementia.87 Contributors to nursing home admissions in people with dementia include poor, uncoordinated hospital care, noncognitive symptoms of dementia (eg, depression, agitation, and delusions), and aggression as part of BPSD.93, 109, 110 Discharge planning should include considering the patient's wishes and using multidisciplinary‐informed standards for discharge from hospital to a care home, although in an audit, consent to a change in residence was not recorded in >30% of patients, nor evidence of “best interests” decision making where patients lacked capacity.15, 106, 111 Fifty‐four per cent of carers' comments regarding discharge/care transfer said that discharge was unsafe and poorly planned, which may lead to readmissions because of lack of available support in the discharge location.

33It appears that the presence of cognitive impairment (particularly dementia) in older hospitalised patients influences a variety of clinical and health service outcomes. This is replicated globally, within different health care systems and patient populations. Although most studies focus on patients with diagnosed dementia rather than all‐cause CI, an increased risk of poor outcomes, eg, in‐hospital mortality, delirium, longer LOS, and institutionalisation at discharge was common. Higher mortality rates may partly reflect lack of available suitable care at end of life, lack of end‐of‐life care plans, eg, “do‐not‐hospitalise” advance directives, or unnecessary transfers from nursing homes.70, 112, 113, 114, 115 Delays in organising appropriate discharge contribute to lengthened hospital stays, highlighting that administrative management and linked services required by these patients may impact on final hospital outcome, as more days in hospital may lead to deconditioning, and policy changes to health and social care infrastructure have unforeseen impacts.116

34Patients with CI are at increased risk of new infections in hospital, decline in functional and nutritional status, behavioural symptoms, and incontinence. These may be considered “intermediate” outcomes, precipitating patient deterioration, for example, CI was associated with mortality only in patients who had at least 1 adverse event in hospital, and dementia associated with mortality only if delirium had occurred.41, 79 Such adverse clinical events could indicate a “failure to maintain” patients' basic health needs, leading to further deterioration.117 A better understanding of how CI precipitates these events, and what can be done to prevent, detect, and reduce their risk, would enable development of better care models and improved patient outcomes. The multifactorial nature of these events requires a multilevel approach at 7 levels of care—patient, task, staff, team, environment, organisation, and institution—to make improvements, and outcomes for hospital dementia care should reflect changes at each of these levels.118 Maintaining clinical and functional status of patients may impact on postdischarge outcomes, eg, mortality, short‐term readmissions, institutionalisation within a year, and continued functional decline.119, 120, 121, 122 A focus on fundamentals of care, eg, ensuring nutrition, hydration, skin care and mobilisation of patients, and psychological care, may improve intermediate outcomes and reduce in‐hospital and postdischarge decline.

35The variety of covariates used for adjustment in the articles and different approaches used to account for the same underlying characteristics (eg, individual diagnostic groups vs Charlson comorbidity score) may explain variability in study conclusions. For example, functional scores were more significant in predicting mortality than dementia in older patients, but few studies investigating the relationship between CI and mortality adjusted for patient function, suggesting residual confounding. The current trend for including frailty assessments in acute hospital care will provide key information, although it will become difficult to disentangle the relative contributions of frailty and CI, as CI comprises part of commonly used frailty assessments. The majority of studies explored associations between patient characteristics at the beginning of hospital admission with a binary outcome during hospitalisation or outcomes at discharge, not accounting for time‐varying covariates, eg, staffing levels and changes in illness acuity or function. Availability of longitudinal data representing day‐to‐day care, or outcomes reflecting care processes, is essential to understand more about modifiable risk factors contributing to poor outcomes.

36Staffing levels, knowledge, and skills are a barrier to provision of best‐practice care for people with CI in hospital.15, 123 However, studies in this review neither included detailed descriptions of staffing levels and skill mix, staff continuity, training and knowledge, and the general hospital environment, nor took account of these in analyses. Outcomes of value in capturing aspects of care, eg, patient experience, may require specific questionnaires or assessments, and are not commonly available. For example, the person‐centred care of older people with CI in acute care scale (POPAC) measures nursing staff best‐practice care processes to identify CI and employment of nursing interventions to meet associated needs, and could be useful in evaluating routine care and service developments such as training, as well as an outcome in research.124

37No single study included a wide range of care, clinical, and well‐being outcomes. Given the role of intermediate outcomes in influencing catastrophic events such as mortality, a core outcome set for CI focussed on hospital care is required. This could be used to standardise outcomes for interventional and observational studies, improving comparability of studies, and in routine care to improve care quality and enable evaluation of care innovations. Dementia care audits provide a good starting place to develop outcome sets, as they focus on fundamental care that should be in place to prevent negative outcomes. Examples include delirium screening, mobility assessment, nutritional status, pressure ulcers, pain, continence, and functioning,15 plus access to services, eg, liaison psychiatry, speech and language, occupational therapy, social work, and continence services, which indicate holistic care.100 Assessments used in long‐term institutions such as the quality of life in late‐stage dementia scale125 could be useful, as the hospital environment can negatively influence health outcomes, eg, functional independence and quality of life, through a range of processes.126

38Because of the diffuse questions addressed and limited resources, a single reviewer took decisions on study exclusion and data extraction, involving other reviewers in case of ambiguity. Conclusions would be altered substantively only if a number of large‐scale studies had been accidentally omitted, which seems unlikely. Trial registers were not searched for ongoing studies in this area. Non‐English language articles were not included because of translation restrictions. The majority of findings indicate a relationship between CI and outcomes. Although selective publication of significant results is possible, there would have to be several large unpublished studies to substantially change the overview of findings.

39Whilst it is important to understand the impact of CI on mortality, length of stay, and institutionalisation, improvement of care for these patients requires insight into the precipitating factors for intermediate outcomes, eg, infections, dehydration, and functional decline, during hospitalisation. Extrinsic factors, eg, staffing and environment, need exploration. Core outcome sets which reflect intermediate outcomes in hospital could be developed and used for clinical trials and quality improvements.

40The research was supported by the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research and Care (CLAHRC) Wessex at Southampton NHS Hospitals Foundation Trust and Portsmouth Hospitals NHS Trust. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health and Social Care.