Introduction
As of 2015, among the estimated global population of 7.3 billion people, 617.1 million (9%) were 65 years of age and older. In Canada, the older population will increase from 15.3 per cent (as per 2013) to between 23.8 per cent and 27.8 per cent in 2063 (Bohnert, Chagnon, & Dion, Reference Bohnert, Chagnon and Dion2015). With population aging, the number of dementia cases is expected to reach almost 1,000,000 people by 2030 from the approximately 500,000 affected now (McDowell and Canadian Study of Health and Aging Working Group, Reference McDowell1994), resulting in an increase of combined health care system and caregiver costs of more than 60 per cent, from 10.4 to 16.6 billion dollars per year (Public Health Agency of Canada, 2018). With disease progression, people with cognitive impairment (CI), including dementia, eventually require help with most aspects of daily living (Barberger-Gateau et al., Reference Barberger-Gateau, Alioum, Peres, Regnault, Fabrigoule and Nikulin2004; Hsiung et al., Reference Hsiung, Alipour, Jacova, Grand, Gauthier and Black2008). Moreover, CI complicates the management of other medical conditions, necessitating complex medical care and its coordination across multiple sectors of care and community services (Bail & Grealish, Reference Bail and Grealish2016; Walker, Reference Walker2011).
Older adults in acute care settings, while awaiting discharge, commonly experience decline in their overall health and well-being that may stem from complications related to their illness, such as delirium or deconditioning, but that may also stem from hospital-related factors. (Admi, Shadmi, Baruch, & Zisberg, Reference Admi, Shadmi, Baruch and Zisberg2015; Dewing & Dijk, Reference Dewing and Dijk2016; Long, Brown, Ames, & Vincent, Reference Long, Brown, Ames and Vincent2013; Möllers et al., Reference Möllers, Perna, Ihle, Schubert, Bauer and Brenner2019; Watkin, Blanchard, Tookman, & Sampson, Reference Watkin, Blanchard, Tookman and Sampson2012). Delayed discharge (DD), because of decline, can lead to long hospital stays for which, in medical literature, several terms are coined: “alternate level of care (ALC)”, “extended stays”, “long stay”, “inappropriate length of stay”, “overstaying”, and “delayed discharge”, as well as some that are pejorative or express ageism, such as “bed blocking” or “stranded patients”. These terms are used to describe a situation in which a person is deemed medically fit for discharge, no longer in need of the intensity of care provided in an acute care setting, but is not discharged. Here we use “length of stay” (LOS) to denote the total number of days between admission and discharge, including any days that may have been ALC days. “Delayed discharge” is used to describe a length of stay beyond the day on which the patient is judged to be medically fit for discharge, as per the following criteria, which may be all or partially present in the literature: acute care treatments are completed, clinical stability is achieved, procedures for the continuing care are activated, and the family is informed (Bryan, Gage, & Gilbert, Reference Bryan, Gage and Gilbert2006). DD resulting in a lack of acute care beds is an ongoing concern, because it limits access to acute care services and results in inefficient use of hospital resources and poorer quality of care for all patients (Sutherland & Crump, Reference Sutherland and Crump2013).
A recent systematic review of observational studies reports a longer hospitalization for patients with dementia than for patients without a dementia diagnosis (Möllers, Stocker, Wei, Perna, & Brenner, Reference Möllers, Stocker, Wei, Perna and Brenner2019). Because CI and its symptoms are often under-recognized in a clinical setting (Douzenis et al., Reference Douzenis, Michopoulos, Gournellis, Christodoulou, Kalkavoura and Michalopoulou2010; Greco et al., Reference Greco, Cascavilla, Paris, Errico, Orsitto and D’Alessandro2005) and, as mentioned, lead to adverse outcomes for patients and increased costs, it is important to investigate whether there is an association between acute care hospital LOS, including ALC days, and any level of CI. The present scoping review of the literature investigates both this association, and problems with definitions of both LOS or DD and CI.
Methods
This scoping review’s methodology was based on Arksey and O’Malley’s (Reference Arksey and O’Malley2005) iterative approach following five steps: (1) identifying the research question, (2) identifying potentially relevant studies, (3) selecting the relevant studies, (4) charting data, and (5) collating, summarizing, and reporting the results. Researchers must engage with each step in a reflective manner and, if necessary, revisit previous steps to ensure that the literature is comprehensively covered.
Identifying the Research Question
In our first approach to studying LOS or DD among older persons, we posed several questions. Which patients experience longer LOS or DD? Which age group is more prevalent among extended LOS or DD patients? Which diagnosis is more prevalent among them? Why are these patients’ discharges delayed? When submitted to a literature search strategy, these questions generated a large number of references. Further discussion led us to refine this research question: what evidence exists in the literature of an association between CI and the LOS or DD?
Identifying Relevant Studies
Six bibliographic databases (Ovid Allied and Complementary Medicine [AMED], HealthSTAR, MEDLINE®, PsycINFO, Cumulative Index to Nursing and Allied Health Literature [CINAHL], and AgeLine) were searched for relevant publications between January 1997 and November 1, 2019. Considering the changes in health systems and the progress in the detection of CI, we decided to include studies published in roughly the last 20 years. The search strategy comprised a combination of free and controlled vocabulary including: length of stay, delayed discharge, alternate level of care, bed block, aged, and middle-aged (see Supplementary Material 1). “Middle-aged” was included to consider patients with early CI. Only publications in English, French, or German were considered. All identified references were imported into an EndNote library for record management, including the removal of duplicate copies.
Selecting Relevant Studies
The screening process is summarized in the Results section. Using Covidence, two reviewers independently examined the titles and abstracts first, and then the full text of the remaining references according to Population Intervention Comparison Outputs Study (PICOS) criteria (Figure 1). Subsequently, the reviewers included studies reporting on CI and durations of LOS or data on DD. Discrepancies between reviewers were resolved through discussion, or with the opinion of a third reviewer (E.K.) until consensus was obtained. Reference lists of relevant review articles and of included studies were screened for additional references. We did not screen grey literature.
Figure 1. Population Intervention Comparison Outputs Study (PICOS) criteria for study selection
Charting Data
For data extraction, we used the descriptive-analytical method within the narrative tradition described by Arksey and O’Malley (Reference Arksey and O’Malley2005). This method involved applying a common analytical framework to all primary research reports, with data being filed in an Excel database of included studies. The data charting form comprised study references; study characteristics; measures of CI, LOS, or DD; and observed associations between CI and LOS or DD.
Results
Collating, Summarizing, and Reporting the Results
The database search yielded a total of 6,716 references (Figure 2). After removal of duplicates, the titles and abstracts of 4,373 references were screened and 447 references were retained for full-text screening. We retained 58 articles for further analysis after reading the text of all articles and applying the exclusion criteria. Although the search strategies included the term “middle-aged,” the included studies dealt mostly with older adults.
Figure 2. Flow diagram of the scoping review
The characteristics of the included studies are summarized in Figure 3. The subject has gained increased interest since 2014. Most of the studies adopted a quantitative approach (prospective or retrospective).
Figure 3. Characteristics of the included studies
Assessment of LOS or DD
Although LOS was reported in most studies, different approaches were used to assess whether delayed discharge occurred. In 14 studies, the definition of DD was based on the clinical judgement of a multidisciplinary team, including a doctor or a nurse, for acute care patients who no longer required a hospital-based level of care and whose needs could be met in another setting (Bo et al., Reference Bo, Fonte, Pivaro, Bonetto, Comi and Giorgis2016; Canadian Institute for Health Information, 2009; Challis, Hughes, Xie, & Jolley, Reference Challis, Hughes, Xie and Jolley2014; Chin, Sahadevan, Tan, Ho, & Choo, Reference Chin, Sahadevan, Tan, Ho and Choo2001; Costa & Hirdes, Reference Costa and Hirdes2010; Costa, Poss, Peirce, & Hirdes, Reference Costa, Poss, Peirce and Hirdes2012; Johansen & Fines, Reference Johansen and Fines2012; Landeiro, Leal, & Gray, Reference Landeiro, Leal and Gray2016; Lenzi et al., Reference Lenzi, Mongardi, Rucci, Di Ruscio, Vizioli and Randazzo2014; Little, Hirdes, & Daniel, Reference Little, Hirdes and Daniel2015; Mayo, Wood-Dauphinee, Gayton, & Scott, Reference Mayo, Wood-Dauphinee, Gayton and Scott1997; McCloskey, Jarrett, Stewart, & Nicholson, Reference McCloskey, Jarrett, Stewart and Nicholson2014; Tucker, Hargreaves, Wilberforce, Brand, & Challis, Reference Tucker, Hargreaves, Wilberforce, Brand and Challis2016; Walker, Morris, & Frood, Reference Walker, Morris and Frood2009). The authors of six studies used fixed values such as: a LOS greater than 30 days (Barba et al., Reference Barba, Marco, Canora, Plaza, Juncos and Hinojosa2015; Kozyrskyi, De Coster, & St John, Reference Kozyrskyi, De Coster and St John2002; Lang et al., Reference Lang, Heitz, Hedelin, Drame, Jovenin and Ankri2006), LOS greater than 10 days (Dent & Perez-Zepeda, Reference Dent and Perez-Zepeda2015; McAlister & van Walraven, Reference McAlister and van Walraven2019), and, in one study LOS was categorized as low (< 8 days), intermediate (8–13 days), and high (> 13 days) (Beauchet et al., Reference Beauchet, Launay, de Decker, Fantino, Kabeshova and Annweiler2013). Finally, in eight articles, the actual LOS was compared with a relative value such as a hospital stay exceeding the diagnosis-related group LOS (Lang et al., Reference Lang, Heitz, Hedelin, Drame, Jovenin and Ankri2006; McCusker, Cole, Dendukuri, & Belzile, Reference McCusker, Cole, Dendukuri and Belzile2003), the Healthcare Resource Groups (HRG)-predicted LOS by physical and cognitive function score (Carpenter, Bobby, Kulinskaya, & Seymour, Reference Carpenter, Bobby, Kulinskaya and Seymour2007), not being discharged more than 24 hours after last time deemed clinically fit in a medical note (Moore, Hartley, & Romero-Ortuno, Reference Moore, Hartley and Romero-Ortuno2018), being two standard deviations above the mean (Foer, Ornstein, Soriano, Kathuria, & Dunn, Reference Foer, Ornstein, Soriano, Kathuria and Dunn2012), being in the 75th percentile of the LOS distribution (Antonelli Incalzi et al., Reference Antonelli Incalzi, Pedone, Onder, Pahor and Carbonin2001), or being in the 90th percentile (Brousseau et al., Reference Brousseau, Dent, Hubbard, Melady, Émond and Mercier2019) greater than the fifth quintile limit (Lisk et al., Reference Lisk, Uddin, Parbhoo, Yeong, Fluck and Sharma2019). Unfortunately, in the remaining studies, the criteria used for these assessments were not described in detail. It is of note that LOS and DD are not mutually exclusive: the total length of stay may include the number of days considered as delayed.
Association between Cognitive Impairment and LOS or DD
Thirty-three studies indicated a positive (n = 31) or a neutral (n = 2) (Antonelli Incalzi et al., Reference Antonelli Incalzi, Pedone, Onder, Pahor and Carbonin2001; Dinkel & Lebok, Reference Dinkel and Lebok1997) association between CI and LOS or DD, which are summarized in Table 1. Also, seven other studies, among those with mixed results, found an association with some aspects of CI (Table 2), and 14 others showed an indirect association (Table 3), mainly with scores that include an assessment of CI.
Table 1. Summary of studies reporting an association between cognitive status and hospital length of stay, by country
Note. 3MS = Modified Mini Mental State Examination; 6-CIT = 6-item Cognitive Impairment Test; AD8 = Eight-item Informant Interview; ALC= alternate level of care; AMTS = Abbreviated Mental Test Score; AMT = Abbreviated Mental Test; CI = cognitive impairment; 95% CI = 95% confidence Interval; COPD = chronic obstructive pulmonary disease; DD = delayed discharge; DRG = diagnosis-related group; DSM = Diagnostic and Statistical Manual of Mental Disorders; ICD = International Classification of Diseases; IQR = interquartile range; IRR = incidence rate ratios; LOS = length of stay; LTC = long-term care; MCI = mild cognitive impairment; MMSE = Mini Mental State Examination; MoCA = Montreal Cognitive Assessment; NH = nursing home; NINCDS-ADRDA = National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association; OR = odds ratio; OMHRS = Ontario Mental Health Reporting System; RAI: Resident Assessment Instrument-Mental Health; SD = standard deviation.
Table 2. Summary of studies presenting mixed results or a negative association between CI and LOS or DD, by country
Note. 95% CI = 95% confidence interval; AMTS = Abbreviated Mental Test Score; CI = cognitive impairment; DD = delayed discharge; HRG = Healthcare Resource Group; ICD = International Classification of Diseases; IQR = interquartile range; LOS = length of stay; LTC = longer-term care; MMSE = Mini Mental State Examination; OR = odds ratio; SD = standard deviation.
Table 3. Summary of studies presenting the relationship between cognition-related geriatric manifestations/scales and LOS or DD, by country
Note. 95% CI = 95% confidence interval; ALC = alternate level of care; CAM = Confusion Assessment Method; DD = delayed discharge; DRG = diagnosis-related group; HAC-OP = Hospital-Associated Complications of Older People; ICD = International Classification of Diseases; IQCODE = Informant Questionnaire On Cognitive Decline in the Elderly; LOS = length of stay; MMSE = Mini-Mental State Examination; OR = odds ratio; SD = standard deviation; SPMSQ = Short Portable Mental Status Questionnaire.
Among the 33 studies, some presented only observations without assessing associations. When the prevalence of CI among DD patients was reported (Antonelli Incalzi et al., Reference Antonelli Incalzi, Pedone, Onder, Pahor and Carbonin2001; Canadian Institute for Health Information, 2009; Costa & Hirdes, Reference Costa and Hirdes2010; Costa et al., Reference Costa, Poss, Peirce and Hirdes2012; Little et al., Reference Little, Hirdes and Daniel2015; McCloskey et al., Reference McCloskey, Jarrett, Stewart and Nicholson2014; Walker et al., Reference Walker, Morris and Frood2009), it varied from 4.6 per cent to 63 per cent, illustrating the differences in population samples, in CI definitions (see Table 1, 5th column), and also in DD assessment (see Table 1, 6th column). Two studies showed that patients with a DD had poorer cognition (Costa & Hirdes, Reference Costa and Hirdes2010; Moore et al., Reference Moore, Hartley and Romero-Ortuno2018).
According to 20 studies, patients with CI had a longer LOS or more frequent DD than those without (Beauchet et al., Reference Beauchet, Launay, de Decker, Fantino, Kabeshova and Annweiler2013; Briggs et al., Reference Briggs, Coary, Collins, Coughlan, O’Neill and Kennelly2016, Reference Briggs, Dyer, Nabeel, Collins, Doherty and Coughlan2017; Canadian Institute for Health Information, 2009; Connolly & O’Shea, Reference Connolly and O’Shea2015; Cullum, Metcalfe, Todd, & Brayne, Reference Cullum, Metcalfe, Todd and Brayne2008; Draper, Karmel, Gibson, Peut, & Anderson, Reference Draper, Karmel, Gibson, Peut and Anderson2011; Guijarro et al., Reference Guijarro, San Roman, Gomez-Huelgas, Villalobos, Martin and Guil2010; King, Jones, & Brand, Reference King, Jones and Brand2006; Kozyrskyi et al., Reference Kozyrskyi, De Coster and St John2002; Langsetmo et al., Reference Langsetmo, Katz, Cawthon, Cayley and Taylor2019; Lisk et al., Reference Lisk, Uddin, Parbhoo, Yeong, Fluck and Sharma2019; Lyketsos, Sheppard, & Rabins, Reference Lyketsos, Sheppard and Rabins2000; Möllers, Stocker, et al., Reference Möllers, Perna, Ihle, Schubert, Bauer and Brenner2019; Moore et al., Reference Moore, Hartley and Romero-Ortuno2018; Motzek, Werblow, Tesch, Marquardt, & Schmidt, Reference Motzek, Werblow, Tesch, Marquardt and Schmidt2018; Power et al., Reference Power, Duffy, Bates, Healy, Gleeson and Lawlor2017; Tucker et al., Reference Tucker, Hargreaves, Wilberforce, Brand and Challis2016; Walker et al., Reference Walker, Morris and Frood2009; Wolf, Rhein, Geschke, & Fellgiebel, Reference Wolf, Rhein, Geschke and Fellgiebel2019) whereas Dinkel and Lebok (Reference Dinkel and Lebok1997) found that among the patients who were discharged from the hospital, a dementia diagnosis had only a minor effect on LOS, although they stressed that hospital mortality was double in cases of dementia.
Some of the aforementioned studies (Möllers, Stocker, et al., Reference Möllers, Perna, Ihle, Schubert, Bauer and Brenner2019; Power et al., Reference Power, Duffy, Bates, Healy, Gleeson and Lawlor2017) and a further seven using more sophisticated statistical analyses, such as adjusted regression models, showed a significant association of CI with longer LOS (Barba et al., Reference Barba, Marco, Canora, Plaza, Juncos and Hinojosa2015; King et al., Reference King, Jones and Brand2006; Lang et al., Reference Lang, Heitz, Hedelin, Drame, Jovenin and Ankri2006; Lenzi et al., Reference Lenzi, Mongardi, Rucci, Di Ruscio, Vizioli and Randazzo2014; Loren Guerrero & Gascon Catalan, Reference Loren Guerrero and Gascon Catalan2011; Tropea, LoGiudice, Liew, Gorelik, & Brand, Reference Tropea, LoGiudice, Liew, Gorelik and Brand2017; Zhu et al., Reference Zhu, Cosentino, Ornstein, Gu, Andrews and Stern2015). A Canadian population study reported that dementia and delirium, whether as primary or co-morbid conditions, were the neurocognitive diagnoses that accounted for the largest number of acute care hospital days (Johansen & Fines, Reference Johansen and Fines2012). Consistent with this, Saravay et al. (Reference Saravay, Kaplowitz, Kurek, Zeman, Pollack and Novik2004) showed that admission scores on rating scales of CI, delirium, and dementia predicted the emergence of mental and behavioural manifestations of delirium and dementia in hospital and greater LOS. Mental and behavioural manifestations seemed to be the proximate causes of greater LOS.
The association between CI and LOS was mixed or negative in eight studies (Table 2). In one of these studies, patients with “medium” cognitive impairment (Cognitive Performance Scale [CPS] score 1–3) had a LOS greater than the one predicted by their physical and cognitive function scores upon admission, whereas those with low or high CPS scores did not (Carpenter et al., Reference Carpenter, Bobby, Kulinskaya and Seymour2007). Challis et al. (Reference Challis, Hughes, Xie and Jolley2014) found that CI was significantly associated with DD, but no longer a significant predictor of DD after discharge variables were considered. Fogg, Meridith, Bridges, Gould, and Griffith (Reference Fogg, Meridith, Bridges, Gould and Griffith2017) reported that patients with “cognitive impairment/no dementia” had significantly longer LOS than patients without CI or patients with a diagnosis of dementia. In two studies, univariate analyses indicated a relationship between CI and LOS, but this association was lost in multivariate analysis in favour of co-morbidity (Zekry et al., Reference Zekry, Herrmann, Grandjean, Vitale, De Pinho and Michel2009), or in favour of a decline in basic activities of daily living capacity and need for skilled nursing (Chin et al., Reference Chin, Sahadevan, Tan, Ho and Choo2001). Another study by Bo et al. (Reference Bo, Fonte, Pivaro, Bonetto, Comi and Giorgis2016) reported that greater CI was independently associated with prolonged LOS among patients admitted from home; however, for patients admitted from intermediate or long-term care facilities, it was lower CI that was independently associated with prolonged LOS. In a study comparing patients discharged from psychiatric hospitals in Turkey with similar patients in Ireland, the presence of delirium or dementia was not associated with LOS, but the proportion of such patients in the sample was low: 13.6 per cent in Turkey and 25.5 per cent in Ireland; the only factor significantly associated with LOS was living alone (Carpar et al., Reference Carpar, McCarthy, Adamis, Donmezler, Cesur and Fistikci2018). In the last mixed-results study, a trend towards longer stay was seen for dementia patients, which was only significant in the subgroup who lived with a family member (11.61 vs. 7.55 days, p = 0.002) (Ahern et al., Reference Ahern, Cronin, Woods, Brady, O’Regan and Trawley2019). Authors also found that people with dementia admitted for a surgical procedure had a shorter stay than people without dementia (2.4 vs. 5.87 days, p = 0.001). Three additional studies (Table 2) identified an inverse association between cognitive impairment and LOS. In one study, dementia was less prevalent in the long stay than in the non-long stay group (Foer et al., Reference Foer, Ornstein, Soriano, Kathuria and Dunn2012). In another, authors reported that dementia was associated with a significant decrease in the likelihood of DD (Landeiro et al., Reference Landeiro, Leal and Gray2016). In a third study, an inverse association was found between dementia and an increased LOS for electively admitted patients, which the authors (Vetrano et al. Reference Vetrano, Landi, De Buyser, Carfi, Zuccala and Petrovic2014) hypothesised was the result of a significantly higher mortality rate in demented patients, reducing their LOS.
There were also 14 studies pertaining to common frailty scores or frailty-related conditions of older adults, which are highly related to CI (Table 3). A comparison of four indices (Frailty Index [FI], Score Hospitalier d’Evaluation du Risque de Perte d’Autonomie [SHERPA], Hospital Admission Risk Profile [HARP], and Acute Physiology and Chronic Health Evaluation [APACHE] II) predicting adverse outcomes in hospitalised older Mexican adults showed that those indices that include cognitive scores correlated with the LOS (Dent & Perez-Zepeda, Reference Dent and Perez-Zepeda2015). Two other studies showed an association of CI with a version of the FI (Basile et al., Reference Basile, Catalano, Mndraffino, Maltese, Alibrandi and Cianco2019; Brousseau et al., Reference Brousseau, Dent, Hubbard, Melady, Émond and Mercier2019). The Hospital Frailty Risk Score was used in a third study with similar conclusions. (McAlister & van Walraven, Reference McAlister and van Walraven2019) Two other studies investigated the value of prognostication scores and found an association with the Charlson Comorbidity Index (CCI) (Bahrmann et al., Reference Bahrmann, Bluner, Christ, Bertsch, Sieber and Katus2019) and the Multidimensional Prognostic Index (MPI) (Pilotto et al., Reference Pilotto, Veronese, Daragiati, Cruz-Jentoff, Polidori and Mettace-Raso2019). In another study, frailty and delirium, but not dementia, were predictors of increased LOS (Basic & Shanley, Reference Basic and Shanley2015). In three studies, the impact of delirium was specifically investigated. In the first, an adjusted analysis found that patients with prevalent or incident subsyndromal delirium had longer LOS (Cole, McCusker, Dendukuri, & Han, Reference Cole, McCusker, Dendukuri and Han2003), whereas the second reported an association between prevalent delirium and LOS, but not with incident delirium (McCusker et al., Reference McCusker, Cole, Dendukuri and Belzile2003). Also, five geriatric syndromes including delirium were introduced in a model (Hospital-Associated Complications of Older People [HAC-OP]), showing an association of any of them with LOS and a graded correlation between a greater number of them and still longer LOS (Mudge et al., Reference Mudge, McRe, Hubbard, Peel, Lim and Barnett2019). “Confusion” was also related to a longer LOS (Jasinarachchi et al., Reference Jasinarachchi, Ibrahim, Keegan, Mathialagan, McGourty and Phillips2009; Mayo et al., Reference Mayo, Wood-Dauphinee, Gayton and Scott1997), as was agitation (Cots, Chiarello, Perez, Gracia, & Becerra, Reference Cots, Chiarello, Perez, Gracia and Becerra2016). In Mizrahi, Arad, and Adunsky’s (Reference Mizrahi, Arad and Adunsky2016) study, pre-stroke patients with dementia had a shorter LOS than those without dementia; however, these authors found that the presence of lower gains on a scale measuring functional independence during the course of rehabilitation was frequently associated with earlier discharge.
Discussion
Extended LOS in acute care hospital settings and DD are major concerns to health care systems (Bryan et al., Reference Bryan, Gage and Gilbert2006; Canadian Institute for Health Information, 2012) and the well-being of hospitalized older persons (Allen, Reference Allen2016; Hirsch, Sommers, Olsen, Mullen, & Winograd, Reference Hirsch, Sommers, Olsen, Mullen and Winograd1990). This scoping review examined the impact of cognitive impairment on acute care hospital LOS. A positive association between cognitive impairment and LOS or DD was reported to some extent by 52 of the 58 identified studies, even if the definitions and ways to assess CI, as well as LOS or DD, varied among studies. The data we gathered indicate that the magnitude and direction of this association, with regard to the fact that both positive and inverse relationships were found, is influenced by CI or by factors related to its assessment, such as its prevalence in the studied samples (Carpenter et al., Reference Carpenter, Bobby, Kulinskaya and Seymour2007) or the severity of the impairment. It is of note that LOS or DD are also influenced by other factors inherent in the patient, for example, the causes of hospitalization and co-morbidities (Fick, Steis, Waller, & Inouye, Reference Fick, Steis, Waller and Inouye2013). Such factors, including admission diagnoses, could therefore be confounding the association between CI and LOS or DD (Ahern et al., Reference Ahern, Cronin, Woods, Brady, O’Regan and Trawley2019) by leading to a specific health care trajectory. Conversely, the self-management of some health issues may require greater cognitive function and could play a role in delayed discharge in less-severely impaired patients who desire to return home but may need additional services not readily available (Costa & Hirdes, Reference Costa and Hirdes2010; Costa et al., Reference Costa, Poss, Peirce and Hirdes2012). Additionally, LOS or DD is influenced by factors related to the life/health care environment, namely: the initial living arrangement of a patient (e.g., being admitted from home vs. from a care facility) (Ahern et al., Reference Ahern, Cronin, Woods, Brady, O’Regan and Trawley2019; Bo et al., Reference Bo, Fonte, Pivaro, Bonetto, Comi and Giorgis2016), the discharge destination, the availability of relevant services (Afilalo et al., Reference Afilalo, Soucy, Xue, Colacone, Jourdenais and Boivin2015; Amador, Reyes-Ortiz, Reed, & Lehman, Reference Amador, Reyes-Ortiz, Reed and Lehman2007; Canadian Institute for Health Information, 2017; Costa & Hirdes, Reference Costa and Hirdes2010; Costa et al., Reference Costa, Poss, Peirce and Hirdes2012) and other co-variates (Challis et al., Reference Challis, Hughes, Xie and Jolley2014; Chin et al., Reference Chin, Sahadevan, Tan, Ho and Choo2001); hence, a cognitively impaired patient already living in a nursing home or having access to the relevant home care services at the time of admission might be discharged more rapidly.
With regard to our research question, the reviewed literature tends to show an association between CI at large, including delirium and dementia, and the length of hospital stay or the risk of a DD, although definitions of the latter may vary. However, few studies articulated the reasons for this association, and the results should be interpreted with caution because numerous factors may be confounding this association, such as delirium (Kozyrskyi et al., Reference Kozyrskyi, De Coster and St John2002; McCusker et al., Reference McCusker, Cole, Dendukuri and Belzile2003). Also, McCloskey et al. suggested that hospitals may actually be contributing to the functional decline experienced by patients as they await discharge, leading to an increase in the level of care needed when they return to the community (McCloskey et al., Reference McCloskey, Jarrett, Stewart and Nicholson2014). In addition, literature suggests that patients who experienced most of their decline before being admitted to the hospital may have a better recovery than those who mostly deteriorated in the hospital (Gagliardi et al., Reference Gagliardi, Corsonello, Di Rosa, Fabbietti, Cherubini and Mercante2018), a development that may contribute to DD in an important way. Costa et al. suggested that CI limits the choice of accessible community-based institutions, and that therefore some patients who might theoretically benefit from community care are nevertheless admitted to nursing homes (Costa & Hirdes, Reference Costa and Hirdes2010; Costa et al., Reference Costa, Poss, Peirce and Hirdes2012). Hence, those with the least potential to return to community care may remain longer in the acute care hospital awaiting a suitable setting. The contrary has also been described in aforementioned studies, which could be explained by the fact that some of the more impaired patients, or patients already living in long-term care, have a poorer recovery potential, higher mortality, or less uncertainty about their discharge destination, and, therefore, shorter LOS or DD. Three Canadian studies also suggested that the potential need for continuing care, or the lack of support in the community, could explain the association between cognitive impairment and LOS (Johansen & Fines, Reference Johansen and Fines2012; Mayo et al., Reference Mayo, Wood-Dauphinee, Gayton and Scott1997; Walker et al., Reference Walker, Morris and Frood2009). In that context, it has been suggested to increase the number of beds in long-term care. However, although a lack of suitable housing and a shortage of community support were the reasons most commonly cited by treatment teams as barriers to discharge (Gigantesco et al., Reference Gigantesco, de Girolamo, Santone, Miglio and Picardi2009), changes in bed supply may at best have mixed effects on LOS or DD, because in some contexts, for example in Canada, the population 75 years of age and older is growing faster than the number of long-term care beds could (De Coster, Bruce, & Kozyrskyi, Reference De Coster, Bruce and Kozyrskyi2005). Although a discharge plan tailored to the individual patient may bring a small reduction in hospital LOS, reduce the risk of readmission, and increase satisfaction with health care for patients and professionals, there is little evidence that such discharge planning reduces the costs of health service delivery (Goncalves-Bradley, Lannin, Clemson, Cameron, & Shepperd, Reference Goncalves-Bradley, Lannin, Clemson, Cameron and Shepperd2016). Finally, the idea that patients’ experiences in hospitals can be parcelled into neat categories, and resources varied accordingly, may be misleading. A recent commentary described many vulnerable patients as being “rehabbed to death” (Flint, David, & Smith, Reference Flint, David and Smith2019). In this scenario, a patient who remained diminished after an acute illness, but did not meet certain administrative criteria for discharge, would be “transferred to rehabilitation”. Such transfers may suggest that adverse outcomes, such as mental or behavioural symptoms of cognitive impairment, are in part caused by the lack of attention to medical issues that will not be addressed in the new (rehabilitation) facility. Canadian experiences with patients labelled as “subacute” suggest an uncomfortable similarity here (Elbourne, Hominick, Mallery, & Rockwood, Reference Elbourne, Hominick, Mallery and Rockwood2013). This approach also lets hospitals off the hook for their role in nosocomial dependency, especially reduced mobility. Likewise, the close relationship between a longer LOS or DD and common frailty syndromes associated with CI, for example, delirium and functional decline, supports the need for proactive cognitive/frailty screening so that hospitals can improve their care of people at risk for these common, costly, and unintended adverse outcome of current care practices (Muscedere et al., Reference Muscedere, Andrew, Bagshaw, Estabrooks, Hogan and Holroyd-Leduc2016) .
Limitations
We used a transparent approach to complete this scoping review. Six scientific databases were searched, but we excluded the grey literature. Although identified references were independently screened by two reviewers, the search strategy might have missed relevant studies. Although associations between a patient’s physical status, including CI, and LOS or DD, can be understood across provincial and national boundaries, the practical interpretation of these results is largely regional and based on where each study was made. The management and availability of care resources specific to each region’s health care setting may greatly influence a patient’s trajectory to, and beyond, acute care. As such, the present scoping review is not aimed at making precise regionally based recommendations for reducing LOS or DD for cognitively impaired patients. There are also numerous possible confounders of the association between CI and LOS or DD, which have been described, and there is no consensus as to the definitions of CI or DD, which mandates care in the interpretation of the results and the inference of a causal link.
Conclusion
Although many factors may influence LOS or DD, we focused on CI, and the studies included in this scoping review point to an association between the two. Although CI among older in-patients has been estimated to vary between 15 per cent and 42 per cent in different countries (Jackson et al., Reference Jackson, Gladman, Harwood, MacLullich, Sampson and Sheehan2017), McCloskey et al. (Reference McCloskey, Jarrett, Stewart and Nicholson2014) found 63.6 per cent of a sample of seniors admitted for acute hospital care in New Brunswick, Canada to be affected by dementia, which is obviously higher. But, despite this, dementia tends to be under-recognized in acute care hospitals (Agarwal, Kazim, Xu, Borson, & Taffet, Reference Agarwal, Kazim, Xu, Borson and Taffet2016; Crowther, Bennett, & Holmes, Reference Crowther, Bennett and Holmes2017; Douzenis et al., Reference Douzenis, Michopoulos, Gournellis, Christodoulou, Kalkavoura and Michalopoulou2010; Greco et al., Reference Greco, Cascavilla, Paris, Errico, Orsitto and D’Alessandro2005; Jackson, MacLullich, Gladman, Lord, & Sheehan, Reference Jackson, MacLullich, Gladman, Lord and Sheehan2016).
Hospitalization can have negative impacts not only on people with CI, but also on their family members and the hospital staff (Hirsch et al., Reference Hirsch, Sommers, Olsen, Mullen and Winograd1990; Swinkels & Mitchell, Reference Swinkels and Mitchell2008). Furthermore, the risk of functional decline is by far the greatest risk associated with the hospitalization of older adults (Admi et al., Reference Admi, Shadmi, Baruch and Zisberg2015), and it may be aggravated by longer stays. It is therefore essential that future studies aim to clarify the mechanisms underlying the relationship between cognitive impairment and LOS or DD in order to come up with solutions to this public health challenge. Practically speaking, a systematic screening for CI in older in-patients using a more consensual definition, and a standardization of DD assessment, could help to elicit these mechanisms, also allowing for the identification of other factors bearing on LOS. Recent data tend to show that such screening is feasible, allows to better target interventions at this vulnerable group (Kurrle et al., Reference Kurrle, Bateman, Cumming, Pang, Patterson and Temple2019), and has been tried with tools such as the Mini-Cog© (Geschke, Weyer-Elberich, Mueller, Binder, & Fellgiebel, Reference Geschke, Weyer-Elberich, Mueller, Binder and Fellgiebel2019) and the Frailty Index (Basile et al., Reference Basile, Catalano, Mndraffino, Maltese, Alibrandi and Cianco2019). Other data tend to show that providing a trained geriatric nurse to coordinate intra-hospital transitions (Bristol, Reference Bristol2019), facilitating the patient’s attending an adult day-care service routinely before that person will need a hospitalization, or establishing dedicated geriatric intervention units implementing “personal engagement specialists”; that is, a type of nursing assistant offering one-on-one care (Sinvani et al., Reference Sinvani, Warner-Cohen, Strunk, Halbert, Harisingani and Mulvany2018), could lead to better outcomes in patients with CI by respectively reducing delirium incidence, need for hospitalization, emergency visits, LOS, and mortality. These means to mitigate or reduce DD may become important for the development of efficient and practical interventions, to inform future policy and research and to limit its negative impacts on these particularly vulnerable seniors.
