Frailty in an Older Inpatient Population

Abstract

Objective: To evaluate the impact of frailty, measured using the Canadian Study of Health and Aging Clinical Frailty Scale, on outcomes of older people hospitalized with acute illness. Method: Consecutive patients were randomly allocated to a model development sample or a model validation sample. Multivariate analyses were used to model in-hospital mortality, new nursing home placement, and length of stay. Variables selected in the development samples were tested in the validation samples. Results: The mean age of all 2,125 patients was 82.9 years. Most (93.6%) were admitted through the emergency department. Frailty predicted in-hospital mortality (odds ratio [OR] = 2.97 [2.11, 4.17]), new nursing home placement (OR = 1.60 [1.14, 2.24]), and length of hospital stay (hazard ratio = 0.87 [0.81, 0.93]). Discussion: Frailty is a strong predictor of adverse outcomes in older people hospitalized with acute illness. An increased awareness of its impact may alert clinicians to screen for frailty.

Keywords

outcomes frailty elderly inpatients length of stay death

Introduction

Frailty describes a condition characterized by decreased physiological reserve and a decreased resistance to stressors, leading to an increased vulnerability to adverse outcomes (Fried et al., 2001). Such outcomes include falls, admission to hospital and increased length of hospital stay, post-operative complications, disability, institutionalization, and mortality (Fairhall et al., 2011; Fried et al., 2001; Heppenstall, Wilkinson, Hanger, & Keeling, 2009; Khandelwal et al., 2012; McMillan & Hubbard, 2012).

As the population ages, the numbers in the oldest age groups will increase the fastest (Wright, 2010). Because frailty prevalence rises with increasing age (McMillan & Hubbard, 2012), its consequences will place a mounting burden on older people, their carers, and the health care system (Fairhall et al., 2011; Smith & Lindley, 2009).

Several conceptual models have been proposed to explain frailty and a large number of studies have sought to assess and measure frailty. The two models most commonly cited are those of Fried (Fried et al., 2001) and Rockwood and Mitnitski (Mitnitski, Mogilner, & Rockwood, 2001; Rockwood et al., 2005). Fried’s model comprises five items: unintentional weight loss, self-reported exhaustion, muscle weakness (grip strength), slow walking speed, and low physical activity. Rockwood and Mitnitski’s model is based on the accumulation of deficits, with the degree of frailty increasing as one accumulates more deficits.

The numbers of studies that have sought to assess and measure frailty have been captured in several recent systematic reviews (Borges & Menezes, 2011; Forti et al., 2012; Sternberg, Wershof Schwartz, Karunananthan, Bergman, & Mark Clarfield, 2011; Theou, Brothers, Mitnitski, & Rockwood, 2013). Although many of these studies report similar outcomes of frailty, there is little consensus on how frailty should be defined and diagnosed. Most of the larger studies have been conducted in community settings. Fewer studies have explored the impact of frailty in older people hospitalized with acute illness.

From a clinical perspective, understanding frailty is important because its effect will generate greater complexity in treatment choices, care planning, and costs of care. Several studies and systematic reviews have shown that improvements can be made to levels of frailty through targeted, multidisciplinary interventions (Binder et al., 2002; Cadore, Rodriguez-Manas, Sinclair, & Izquierdo, 2013; Cameron et al., 2013). A better understanding of frailty may also increase the likelihood of finding practical interventions to reduce its negative outcomes (Fairhall et al., 2011; McMillan & Hubbard, 2012).

To monitor and improve the care of older people, we maintain a comprehensive database of all patients admitted under our geriatric medicine service. To include a measure of frailty within this database, we reviewed the literature on 24 frailty scales. Although none was perfect in all regards, we chose the Canadian Study of Health and Aging Clinical Frailty Scale (CSHA-CFS; Rockwood et al., 2005). In acute care settings, the CSHA-CFS has several advantages over other scales. It encompasses a broad assessment of frailty based on clinical judgment and does not rely on direct measurement of specific items (e.g., grip strength and walking speed), specialized equipment, or extra staff. It does not involve a lengthy questionnaire, which is often difficult for busy clinicians to complete. The CSHA-CFS has also been validated and used in a number of studies of frailty (Chan, Tsou, Chen, & Chen, 2010; Conroy & Dowsing, 2013; Martocchia et al., 2013; Masud et al., 2013; Matusik et al., 2012; Rockwood et al., 2005).

The aim of this study was to evaluate the impact of frailty, measured using the CSHA-CFS, on in-hospital mortality, new nursing home placement, and length of hospital stay.

Method

Study Setting and Participants

The study was undertaken at Liverpool Hospital, a tertiary referral hospital in south-western Sydney, Australia. Participants were 2,125 consecutive patients admitted under the care of eight geriatricians between August 1, 2010 and January 31, 2014. Most were admitted based on geriatric targeting criteria that included delirium, deconditioning, functional impairment, gait abnormality and falls, multiple medical diagnoses, and psychosocial problems. The Human Research Ethics Committee of the area health service provided ethics approval for the study.

Study Protocol

We collected data on patient demographics (age, gender, country of birth, whether English-speaking, admission, and discharge domicile), referral source, medical diagnoses, length of stay (LOS), in-hospital mortality, and premorbid frailty (defined as the level of frailty present 1 month before the admission to the hospital). All patients were seen by a multidisciplinary team, comprising allied health, nursing, and medical staff. Although the geriatrician addressed premorbid disease symptoms and determined the level of frailty using the CSHA-CFS (Table 1; Rockwood et al., 2005), this determination was predominantly based on premorbid data collected and documented by allied health staff.

Table 1.

Canadian Study of Health and Aging Clinical Frailty Scale.

Category	Category description
1	Very fit (energetic, motivated, exercise regularly)
2	Well (no active disease, less fit than Category 1)
3	Well, with treated comorbid disease
4	Apparently vulnerable (not frankly dependent, disease symptoms, slowed up)
5	Mildly frail (limited dependence in IADL)
6	Moderately frail (help needed with ADL and IADL)
7	Severely frail (completely dependent in ADL, or terminally ill)

Note. IADL = instrumental activities of daily living; ADL = activities of daily living.

Based on Version 5.1 of the Australian Refined Diagnosis Related Groups (AR-DRG) classification system, medical diagnoses were grouped into 177 categories. These categories were chosen because they reflect the diagnoses most relevant to geriatric medicine, based on diagnoses recorded in a clinical database maintained by our service since 2001 and in consensus with other geriatricians at the hospital. We coded up to 15 diagnoses per patient (those affecting physical, social, or psychological function, or needing medication changes, investigations, or increased monitoring to treat symptoms and guide management). Overall comorbidity (comorbid disease) was defined as the total number of diagnoses (range = 1-15). Although some diagnoses are more appropriately termed problems or syndromes, for consistency we refer to all as diagnoses.

Statistical Analyses

Multivariate logistic regression was used to model the probabilities of in-hospital mortality and new nursing home placement. Multivariate Cox proportional hazards regression was used to model LOS. Cox regression models use the hazard ratio (HR) to estimate the effect of a variable on the time to an event (discharge from the hospital). A HR greater than 1 indicates an increased probability of discharge at any point of time (and hence a shorter LOS). Conversely, a HR less than 1 indicates a longer LOS. Using random numbers, patients were allocated to either a model development sample or a model validation sample. Independent, plausible predictors (p < .05) from our data set were included in the final development models, together with the study variable (CSHA-CFS) and risk factors in the literature (dementia, delirium, and age; Campbell, Seymour, Primrose, & ACME plus Project, 2004; Campbell et al., 2005; Marengoni, Aguero-Torres, Timpini, Cossi, & Fratiglioni, 2008; Saravay et al., 2004; Shen, Lu, & Li, 2012; van Zyl & Seitz, 2006). Based on the chi-square values, the plausible predictors were removed one by one, until only significant ones remained. We measured the degree of confounding exerted by a variable by calculating the change in the value of the parameter estimate for the study factor (CSHA-CFS) when the variable was excluded from the analysis. A ≥20% change in the parameter estimate was needed for a potential confounder to be retained in the model. Variables selected for use in the model development samples were then tested in the validation samples. Between-groups comparisons were tested using t tests for continuous, normally distributed variables, Wilcoxon rank–sum tests for ordinal variables, and Fisher’s exact tests for dichotomous variables. The linear correlation between variables was measured with Spearman’s rank-order correlation coefficient. Mantel–Haenszel trend analysis was used to evaluate trends. SAS software (Version 9.2, SAS Institute, Inc., Cary, North Carolina) was used for all analyses.

Results

Characteristics of Study Participants

There were no clinically meaningful differences between the development and validation samples in any of the characteristics (Table 2). The diagnoses shown in Table 2 were those tested in the multivariate models (Tables 3 -5). Most of the study participants were admitted through the emergency department and were acutely unwell.

Table 2.

Characteristics of Study Participants by Random Sample.

Characteristic	Development (n = 1,055)	Validation (n = 1,070)
Age (M ± SD)	83.0 ± 7.9	82.7 ± 7.5
Male gender (%)	40.8	39.9
Born in CALD country (%)	54.7	53.8
English-speaking (%)	68.1	67.5
Admission domicile (%)
Nursing home	18.1	20.6
Hostel	8.7	10.1
Other^a	73.2	69.3
Referral source to geriatrician (%)
Emergency department	93.4	93.8
Consultation and transfer of care	5.3	5.2
Direct admission from community	0.9	0.8
Other^b	0.4	0.2
Diagnoses (%)
Dementia	40.9	43.7
Delirium	39.6	42.2
BPSD	8.0	7.6
Secondary delusional disorder^c	2.3	2.5
Acute renal failure	22.8	21.8
Septic shock	5.3	5.4
Respiratory infection	25.4	27.4
Urinary tract infection	19.8	21.0
Type 2 respiratory failure	4.5	4.3
Stroke	7.1	6.0
Seizure disorder	2.3	2.8
Syncope	6.7	6.9
Dysphagia	12.6	11.2
Malnutrition	10.1	8.5
Gastrointestinal malignancy	2.9	2.6
Lung malignancy	2.1	1.7
Urine retention	7.8	9.4
Arthritis (degenerative and crystal)	14.7	16.0
Deconditioning	24.6	26.8
Total no. medical diagnoses (median, IQR)^d	7, 5-10	7, 5-10
CSHA-CFS category (%)
1	0.7	0.6
2	1.4	1.0
3	3.5	3.0
4	5.8	5.4
5	26.8	25.4
6	38.4	42.3
7	23.4	22.3

Note. CALD = culturally and linguistically diverse; BPSD = behavioral and psychological symptoms of dementia; IQR = interquartile range; CSHA-CFS = Canadian Study of Health and Aging Clinical Frailty Scale.

Private home, including independent living unit, boarding house, and caravan.

Other hospital and outpatient department.

Mental disorder marked by delusions, due to an organic brain disorder (e.g., dementia).

Diagnoses affecting patients’ current physical, social, or psychological function, or needing medication changes, investigations, or increased monitoring to treat symptoms and guide management.

Table 3.

Logistic Regression Models for In-hospital Mortality.

Variable	PE	SE	p value	OR (CI 95%)
Development sample^a
CSHA-CFS	0.89	0.16	<.0001	2.44 [1.77, 3.36]
Dementia	−0.46	0.28	.10	0.63 [0.36, 1.09]
Delirium	−0.10	0.27	.70	0.90 [0.53, 1.53]
Age	0.04	0.02	.03	1.04 [1.00, 1.07]
Male gender	0.85	0.26	.001	2.35 [1.41, 3.93]
BPSD	−2.95	1.12	.008	0.05 [0.01, 0.47]
Respiratory infection	1.03	0.26	<.0001	2.80 [1.68, 4.65]
Septic shock	1.34	0.36	.0002	3.83 [1.89, 7.78]
Acute renal failure	0.83	0.26	.001	2.30 [1.38, 3.85]
Type 2 respiratory failure	2.06	0.39	<.0001	7.87 [3.67, 16.85]
Stroke	0.81	0.38	.04	2.24 [1.06, 4.75]
Malnutrition	0.66	0.33	.047	1.93 [1.01, 3.68]
Validation sample^a
CSHA-CFS	1.09	0.17	<.0001	2.97 [2.11, 4.17]
Dementia	−0.04	0.26	.89	0.97 [0.59, 1.59]
Delirium	−0.16	0.25	.52	0.85 [0.53, 1.38]
Age	−0.001	0.02	.95	1.00 [0.97, 1.03]
Male gender	0.42	0.24	.08	1.52 [0.96, 2.42]
BPSD	−2.91	1.06	.006	0.06 [0.01, 0.44]
Respiratory infection	0.74	0.24	.002	2.10 [1.32, 3.33]
Septic shock	1.46	0.36	<.0001	4.29 [2.14, 8.60]
Acute renal failure	0.93	0.25	.0001	2.54 [1.57, 4.12]
Type 2 respiratory failure	1.39	0.39	.0003	4.03 [1.89, 8.58]
Stroke	−0.02	0.49	.96	0.98 [0.38, 2.54]
Malnutrition	0.08	0.33	.80	1.09 [0.57, 2.06]

Note. PE = parameter estimate; SE = standard error; OR = odds ratio; CI = confidence interval; CSHA-CFS = Canadian Study of Health and Aging Clinical Frailty Scale; BPSD = behavioral and psychological symptoms of dementia.

Dementia and delirium were forced into model.

Table 4.

Logistic Regression Models for New Nursing Home Placement.

Variable	PE	SE	p value	OR (CI 95%)
Development sample^a
CSHA-CFS	0.73	0.20	.0003	2.08 [1.40, 3.10]
Dementia	0.60	0.31	.05	1.83 [1.00, 3.37]
Delirium	0.08	0.31	.80	1.08 [0.59, 1.97]
Age	0.03	0.02	.08	1.03 [1.00, 1.07]
Urine retention	0.96	0.38	.01	2.60 [1.23, 5.47]
Deconditioning	1.08	0.30	.0003	2.93 [1.64, 5.23]
Seizure disorder	1.30	0.58	.03	3.68 [1.18, 11.51]
Validation sample^a
CSHA-CFS	0.47	0.17	.006	1.60 [1.14, 2.24]
Dementia	0.72	0.28	.01	2.06 [1.19, 3.55]
Delirium	−0.22	0.27	.42	0.81 [0.48, 1.37]
Age	0.002	0.02	.91	1.00 [0.97, 1.04]
Urine retention	1.19	0.32	.0002	3.30 [1.77, 6.13]
Deconditioning	0.94	0.27	.0004	2.57 [1.53, 4.32]
Seizure disorder	−0.07	0.77	.92	0.93 [0.21, 4.16]

Note. PE = parameter estimate; SE = standard error; OR = odds ratio; CI = confidence interval; CSHA-CFS = Canadian Study of Health and Aging Clinical Frailty Scale.

Dementia, delirium, and age were forced into model.

Table 5.

Cox Proportional Hazard Regression Models for LOS in the Hospital.

Variable	PE	SE	p value	HR (CI 95%)
Development sample^a,b
CSHA-CFS	−0.13	0.03	.0001	0.88 [0.82, 0.94]
Dementia	−0.03	0.08	.67	0.97 [0.83, 1.13]
Delirium	−0.29	0.08	.0001	0.75 [0.64, 0.87]
Admitted from nursing home	0.44	0.10	<.0001	1.55 [1.27, 1.89]
Comorbidity^c	−0.06	0.01	<.0001	0.94 [0.92, 0.96]
BPSD	−0.30	0.13	.02	0.74 [0.58, 0.95]
Secondary delusional disorder	−0.53	0.23	.02	0.59 [0.38, 0.92]
Type 2 respiratory failure	−0.45	0.21	.03	0.64 [0.42, 0.96]
Lung malignancy	−0.56	0.25	.03	0.57 [0.35, 0.94]
Dysphagia	−0.26	0.11	.02	0.77 [0.62, 0.97]
Malnutrition	−0.27	0.13	.04	0.77 [0.60, 0.98]
Deconditioning	−0.48	0.08	<.0001	0.62 [0.53, 0.72]
Validation sample^a,d
CSHA-CFS	−0.14	0.04	.0001	0.87 [0.81, 0.93]
Dementia	−0.05	0.08	.51	0.95 [0.82, 1.11]
Delirium	−0.16	0.07	.03	0.85 [0.74, 0.98]
Admitted from nursing home	0.62	0.10	<.0001	1.85 [1.52, 2.26]
Comorbidity^c	−0.09	0.01	<.0001	0.92 [0.90, 0.94]
BPSD	0.01	0.13	.92	1.01 [0.78, 1.31]
Secondary delusional disorder	−0.26	0.20	.19	0.77 [0.52, 1.15]
Type 2 respiratory failure	−0.14	0.19	.47	0.87 [0.60, 1.27]
Lung malignancy	0.10	0.25	.69	1.11 [0.67, 1.82]
Dysphagia	−0.27	0.12	.03	0.76 [0.60, 0.97]
Malnutrition	−0.47	0.13	.0005	0.63 [0.48, 0.81]
Deconditioning	−0.38	0.08	<.0001	0.68 [0.59, 0.79]

Note. LOS = length of stay; PE = parameter estimate; SE = standard error; HR = hazard ratio; CI = confidence interval; CSHA-CFS = Canadian Study of Health and Aging Clinical Frailty Scale; BPSD = behavioral and psychological symptoms of dementia.

Dementia and delirium were forced into model.

98 of 1,023 (9.6%) patients with complete data for variables in the development sample were censored due to in-hospital death.

Number of medical diagnoses (range = 1-15).

108 of 1,026 (10.5%) patients with complete data for variables in the validation sample were censored due to in-hospital death.

In-Hospital Mortality

The overall in-hospital mortality was 10.0%. The rates were 9.6% and 10.4% in the development and validation samples, respectively. Patients who died were more likely to be within the higher CSHA-CFS categories (χ² = 207, p < .0001; Mantel–Haenszel χ² for trend = 112, p < .0001). The death rates for Categories 5, 6, and 7 (comprising the majority of patients) were 3.4%, 6.5%, and 27.4%, respectively. Of the Category 7 deaths, 60.9% were admitted from nursing homes and 35.2% from private homes.

Table 3 shows the final multivariate logistic regression models for in-hospital mortality. Significant predictors in both samples were frailty (CSHA-CFS), respiratory infection, septic shock, acute renal failure, Type 2 respiratory failure, and presence of behavioral and psychological symptoms of dementia (BPSD). Patients with BPSD were less likely to die. Although not confirmed in the validation sample, age, male gender, stroke, and malnutrition were significant predictors in the development sample. Neither dementia nor delirium independently predicted in-hospital mortality. Overall comorbidity was neither a univariate predictor nor a confounder.

New Nursing Home Placement

A minority of patients were newly placed in nursing homes directly from the hospital (overall sample 6.4%, development 5.9%, validation 6.8%). Most were in Categories 6 (placement rate 9.2%) and 7 (placement rate 9.0%).

Table 4 shows the final multivariate logistic regression models for new nursing home placement. Significant predictors in both patient samples were frailty (CSHA-CFS), urine retention, deconditioning, and dementia (although p = .05 in the development sample). Seizure disorder was a significant predictor in the development sample. Although overall comorbidity was a univariate predictor, it was not a multivariate predictor or a confounder.

LOS

The median LOS in the hospital among all 2,125 patients was 10 days (interquartile range [IQR] = 6-17 days). The LOS in the model development (median = 10, IQR = 5-18) and model validation (median = 10, IQR = 6-17) samples were similar.

Table 5 shows the final multivariate Cox proportional hazards models. Although in many patients secondary delusional disorder (mental disorder marked by delusions, due to an organic brain disorder) and BPSD may be manifestations of the same disease process (i.e., dementia), they were poorly correlated (Spearman correlation coefficients of 0.12 and 0.13 in development and validation samples, respectively). Hence, both were included in the models shown in Table 5. Significant predictors of LOS in both patient samples were frailty (CSHA-CFS), delirium, dysphagia, malnutrition, deconditioning, comorbid disease, and nursing home residence. Patients residing in nursing homes had shorter LOS, whereas others stayed longer. Although not confirmed in the validation sample, type 2 respiratory failure, lung malignancy, secondary delusional disorder, and BPSD were significant predictors in the development sample.

Most patients who were newly placed in a nursing home were in Categories 6 and 7. Some of the effects of frailty (CSHA-CFS) on LOS may be mediated through delays in finding nursing home beds for those needing placement. Compared with other patients, those needing placement had longer LOS (21 days vs. 9 days, p < .0001, total sample).

Nursing Home Residents

Patients admitted from nursing homes comprised 19.4% of our total sample. Compared with other patients, they were older (M age 84.9 years vs. 82.4 years, p < .0001) and were more likely to be within the higher CSHA-CFS categories (χ² = 586, p < .0001; Mantel–Haenszel χ² for trend = 365, p < .0001). Nursing home residents had higher rates of dementia (65.0% vs. 36.9%, p < .0001), delirium (56.4% vs. 37.3%, p < .0001), and in-hospital mortality (22.7% vs. 6.9%, p < .0001). Nursing home residents who died had shorter LOS (median 4 days vs. 10 days, p < .0001).

Discussion

Our large cohort study of acutely unwell older patients shows significant relationships between premorbid frailty, measured using the CSHA-CFS, and several adverse outcomes, including in-hospital mortality, new nursing home placement, and length of hospital stay. These relationships were consistent in both study samples and were independent of established risk factors and appropriately selected diagnoses. Patients within the higher CSHA-CFS categories had longer LOS and were more likely to die in the hospital. Those from private homes and hostels were more likely to be discharged to a nursing home directly from the hospital.

Screening for frailty is important because frailty and its impact will increase in proportion to the aging of the population. This increase will become more and more exponential given the disproportional growth in the numbers of people in the oldest age groups (Wright, 2010). The consequences of frailty will place an enormous burden on older people, their carers, and the health care system (Fairhall et al., 2011; Smith & Lindley, 2009).

Although frailty was initially described and validated in geriatric medicine, frailty and its adverse consequences are being considered by clinicians from a range of clinical specialties. In elective surgery patients and those with hip fracture, frailty predicted 30-day surgical complications (Makary et al., 2010; Revenig et al., 2013), length of hospital stay (Krishnan et al., 2014; Makary et al., 2010), need for institutional care (Makary et al., 2010; Robinson et al., 2011), and 30-day mortality (Krishnan et al., 2014). In critically ill patients, including those with burns, frailty predicted major adverse events (Bagshaw et al., 2014), in-hospital mortality (Masud et al., 2013), 1-year mortality (Bagshaw et al., 2014), and post-discharge functional dependency (Bagshaw et al., 2014). In patients with coronary artery disease, frailty predicted functional dependency and a decline in health-related quality of life after 1 year (Freiheit et al., 2010). After kidney transplantation, frailty predicted early hospital readmission (McAdams-DeMarco et al., 2013). Our study adds to the growing body of evidence. This increasing awareness should alert clinicians to the need for a methodical approach to screening for frailty.

Systematic reviews and studies show that improvements in the levels of frailty can be made through targeted, multidisciplinary interventions. In a recent systematic review of exercise interventions, Cadore et al. (2013) found that multicomponent programs that consist of strength, endurance, and balance training improved physical function in frail older people and reduced the rate of falls. In a randomized trial of community-dwelling frail older people, Cameron et al. (2013) identified specific frailty components in each participant and prescribed individually tailored interventions. The frailty components targeted were those from the Cardiovascular Health Study reported by Fried et al. (2001). At 12 months, intervention participants were significantly less frail (p = .02) and had higher scores on the Short Physical Performance Battery (p < .001), in which higher scores indicate better physical function (Cameron et al., 2013).

Although studies of longer duration in community-dwelling people show that frailty can be improved through targeted, multidisciplinary interventions, more work is needed in older people hospitalized with acute illness. Although comprehensive geriatric assessment (CGA) may offset some of the complications of acute hospital care (Ellis, Whitehead, O’Neill, Langhorne, & Robinson, 2011), many older patients deteriorate despite this intervention, with up to 45% developing functional decline (Hoogerduijn et al., 2012; Zisberg et al., 2011). Approaches additional to CGA appear necessary. Relatively simple interventions to maintain in-hospital mobility may be effective in reducing functional decline and associated frailty (Zisberg et al., 2011). Following treatment of acute illnesses, higher intensity interventions in subacute settings may be effective. A better understanding of frailty may increase the likelihood of finding new and practical interventions.

A systematic approach to screening will also highlight patients at the extreme end of frailty, many of whom have a high risk of imminent death. Early identification of such patients should activate a brake on automatically choosing aggressive treatments, and should prompt clinicians to undertake discussions on palliation and future care options with the patient and family (Iqbal, Denvir, & Gunn, 2013).

Although frailty appears to be a valid construct, there is little consensus on how it should be defined. Operational definitions may be rules-based (Fried et al., 2001) or based on the accumulation of deficits (Mitnitski et al., 2001). While both are time-consuming and unlikely to be widely used in busy clinical settings, rules-based definitions might not apply to an individual case (Rockwood et al., 2005). The third class of operational definitions, such as those used by the CSHA-CFS, are subjective and rely on clinical judgment. Although the CSHA-CFS is portable and easy to administer, it requires well-trained clinicians to interpret the results of the history and the clinical examination. Because the CSHA-CFS mixes items such as comorbidity, cognitive impairment, and disability (Rockwood et al., 2005), it appears particularly suited to services encompassing multidisciplinary CGA.

The literature comparing hospital LOS of older residents of nursing homes with those residing in community settings shows mixed results (Godden & Pollock, 2001; Ingarfield et al., 2009). Our nursing home residents had shorter lengths of stay, whereas others stayed longer. Nursing homes provide a structured and familiar environment that allows many patients to be discharged early, often before they recover fully. Although functional and cognitive impairments should not exclude nursing home residents from multidisciplinary rehabilitation, those with severe disability are unlikely to benefit. The high prevalence of severe cognitive and functional impairments, together with earlier and higher in-hospital mortality, explained the shorter LOS of nursing home residents. We are uncertain why patients with BPSD had reduced mortality. Increased vocalization and physical agitation may improve pulmonary ventilation and reduce respiratory infection. Patient-to-patient cross infection may be reduced by the necessity of nursing many patients with BPSD in single rooms. More frequent review and a unit policy strongly discouraging sedation may also reduce mortality.

Our study has several limitations. First, we treated CSHA-CFS, an ordinal scale measure, as an interval scale measure. Quantitative output from our multivariate analyses should therefore be interpreted cautiously. However, we believe our qualitative conclusions to be valid. Second, the population of older patients is much larger and more diverse than our study population. Consequently, our results cannot be extrapolated to all older inpatients, particularly those managed by general physicians in other settings. Similarly, our findings cannot be generalized beyond older people. Third, we did not measure inter-rater reliability of the CSHA-CFS. Although overall a subjective measure, Categories 5 to 7 (comprising the majority of patients) are mostly objective. Finally, our selection of predictor variables in the final regression models may seem subjective. With the exception of well established and consistent risk factors in the literature, selection of other predictors for consideration in regression models is always somewhat subjective. Although the literature should be used as a guide, inconsistencies in the literature are common (Campbell et al., 2004). Our use of two random samples, one to develop the models (development sample) and the other to measure their performances (validation sample), helps to address this issue. Strong predictors in both samples, such as the CSHA-CFS, are likely to be true predictors rather than statistical anomalies.

In summary, we found strong relationships between frailty, measured using the CSHA-CFS, and adverse outcomes in older people hospitalized with acute illness. An increased awareness may help to alert clinicians to screen for frailty. A better understanding of frailty may lead to new and practical interventions to reduce its prevalence and impact.

Footnotes

Acknowledgements

The authors would like to thank the following geriatricians for their contributions to data collection and data entry (alphabetical order by surname): Leemin Chan, David Conforti, Rinaldo Gonzales, Tabitha Hartwell, Angela Khoo, Florence Loh, and Nalini Thayaparan.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Bagshaw

S. M.

Stelfox

H. T.

McDermid

R. C.

Rolfson

D. B.

Tsuyuki

R. T.

Baig

. . . Majumdar

S. R.

(2014). Association between frailty and short- and long-term outcomes among critically ill patients: A multicentre prospective cohort study. Canadian Medical Association Journal, 186, E95-E102.

Binder

E. F.

Schechtman

K. B.

Ehsani

A. A.

Steger-May

Brown

Sinacore

D. R.

. . . Holloszy

J. O.

(2002). Effects of exercise training on frailty in community-dwelling older adults: Results of a randomized, controlled trial. Journal of the American Geriatrics Society, 50, 1921-1928.

Borges

L. L.

Menezes

R. L.

(2011). Definitions and markers of frailty: A systematic review of literature. Reviews in Clinical Gerontology, 21, 67-77.

Cadore

E. L.

Rodriguez-Manas

Sinclair

Izquierdo

(2013). Effects of different exercise interventions on risk of falls, gait ability, and balance in physically frail older adults: A systematic review. Rejuvenation Research, 16, 105-114.

Cameron

I. D.

Fairhall

Langron

Lockwood

Monaghan

Aggar

. . . Kurrle

S. E.

(2013). A multifactorial interdisciplinary intervention reduces frailty in older people: Randomized trial. BMC Medicine, 11, Article 65.

Campbell

S. E.

Seymour

D. G.

Primrose

W. R.

Lynch

J. E.

Dunstan

Espallargues

, . . . The Acmeplus Project. (2005). A multi-centre European study of factors affecting the discharge destination of older people admitted to hospital: Analysis of in-hospital data from the ACMEplus project. Age and Ageing, 34, 467-475.

Campbell

S. E.

Seymour

D. G.

Primrose

W. R.

, & ACME plus Project. (2004). A systematic literature review of factors affecting outcome in older medical patients admitted to hospital. Age and Ageing, 33, 110-115.

Chan

D. C.

Tsou

H. H.

Chen

C. Y.

Chen

C. Y.

(2010). Validation of the Chinese-Canadian Study of Health and Aging Clinical Frailty Scale (CSHA-CFS) telephone version. Archives of Gerontology and Geriatrics, 50(3), e74-80.

Conroy

Dowsing

(2013). The ability of frailty to predict outcomes in older people attending an acute medical unit. Acute Medicine, 12, 74-76.

10.

Ellis

Whitehead

M. A.

O’Neill

Langhorne

Robinson

(2011). Comprehensive geriatric assessment for older adults admitted to hospital. Cochrane Database of Systematic Reviews, Issue 7, Article CD006211.

11.

Fairhall

Langron

Sherrington

Lord

S. R.

Kurrle

S. E.

Lockwood

. . . Cameron

I. D.

(2011). Treating frailty—A practical guide. BMC Medicine, 9, Article 83.

12.

Forti

Rietti

Pisacane

Olivelli

Maltoni

Ravaglia

(2012). A comparison of frailty indexes for prediction of adverse health outcomes in an elderly cohort. Archives of Gerontology and Geriatrics, 54, 16-20.

13.

Freiheit

E. A.

Hogan

D. B.

Eliasziw

Meekes

M. F.

Ghali

W. A.

Partlo

L. A.

Maxwell

C. J.

(2010). Development of a frailty index for patients with coronary artery disease. Journal of the American Geriatrics Society, 58, 1526-1531.

14.

Fried

L. P.

Tangen

C. M.

Walston

Newman

A. B.

Hirsch

Gottdiener

. . . Cardiovascular Health Study Collaborative Research Group. (2001). Frailty in older adults: Evidence for a phenotype. Journals of Gerontology: Series A-Biological Sciences & Medical Sciences, 56, M146-M156.

15.

Godden

Pollock

A. M.

(2001). The use of acute hospital services by elderly residents of nursing and residential care homes. Health and Social Care in the Community, 9, 367-374.

16.

Heppenstall

C. P.

Wilkinson

T. J.

Hanger

H. C.

Keeling

(2009). Frailty: Dominos or deliberation? New Zealand Medical Journal, 122(1299), 42-53.

17.

Hoogerduijn

J. G.

Buurman

B. M.

Korevaar

J. C.

Grobbee

D. E.

de Rooij

S. E.

Schuurmans

M. J.

(2012). The prediction of functional decline in older hospitalised patients. Age and Ageing, 41, 381-387.

18.

Ingarfield

S. L.

Finn

J. C.

Jacobs

I. G.

Gibson

N. P.

Holman

C. D.

Jelinek

G. A.

Flicker

(2009). Use of emergency departments by older people from residential care: A population based study. Age and Ageing, 38, 314-318.

19.

Iqbal

Denvir

Gunn

(2013). Frailty assessment in elderly people. The Lancet, 381, 1985-1986.

20.

Khandelwal

Goel

Kumar

Gulati

Narang

Dey

A. B.

(2012). Frailty is associated with longer hospital stay and increased mortality in hospitalized older patients. Journal of Nutrition, Health & Aging, 16, 732-735.

21.

Krishnan

Beck

Havelock

Eeles

Hubbard

R. E.

Johansen

(2014). Predicting outcome after hip fracture: Using a frailty index to integrate comprehensive geriatric assessment results. Age and Ageing, 43, 122-126.

22.

Makary

M. A.

Segev

D. L.

Pronovost

P. J.

Syin

Bandeen-Roche

Patel

. . . Fried

L. P.

(2010). Frailty as a predictor of surgical outcomes in older patients. Journal of the American College of Surgeons, 210, 901-908.

23.

Marengoni

Aguero-Torres

Timpini

Cossi

Fratiglioni

(2008). Rehabilitation and nursing home admission after hospitalization in acute geriatric patients. Journal of the American Medical Directors Association, 9, 265-270.

24.

Martocchia

Frugoni

Indiano

Tafaro

Comite

Amici

. . . Falaschi

(2013). Screening of frailty in elderly patients with disability by the means of Marigliano–Cacciafesta Polypathology Scale (MCPS) and Canadian Study of Health and Aging (CSHA) scales. Archives of Gerontology and Geriatrics, 56, 339-342.

25.

Masud

Norton

Smailes

Shelley

Philp

Dziewulski

(2013). The use of a frailty scoring system for burns in the elderly. Burns, 39, 30-36.

26.

Matusik

Tomaszewski

Chmielowska

Nowak

Parnicka

. . . Grodzicki

(2012). Severe frailty and cognitive impairment are related to higher mortality in 12-month follow-up of nursing home residents. Archives of Gerontology and Geriatrics, 55, 22-24.

27.

McAdams-DeMarco

M. A.

Law

Salter

M. L.

Chow

Grams

Walston

Segev

D. L.

(2013). Frailty and early hospital readmission after kidney transplantation. American Journal of Transplantation, 13, 2091-2095.

28.

McMillan

G. J.

Hubbard

R. E.

(2012). Frailty in older inpatients: What physicians need to know. Quarterly Journal of Medicine, 105, 1059-1065.

29.

Mitnitski

A. B.

Mogilner

A. J.

Rockwood

(2001). Accumulation of deficits as a proxy measure of aging. The Scientific World Journal, 1, 323-336.

30.

Revenig

L. M.

Canter

D. J.

Taylor

M. D.

Tai

Sweeney

J. F.

Sarmiento

J. M.

. . . Ogan

(2013). Too frail for surgery? Initial results of a large multidisciplinary prospective study examining preoperative variables predictive of poor surgical outcomes. Journal of the American College of Surgeons, 217, 665-670.

31.

Robinson

T. N.

Wallace

J. I.

D. S.

Wiktor

Pointer

L. F.

Pfister

S. M.

. . . Moss

(2011). Accumulated frailty characteristics predict postoperative discharge institutionalization in the geriatric patient. Journal of the American College of Surgeons, 213, 37-42.

32.

Rockwood

Song

MacKnight

Bergman

Hogan

D. B.

McDowell

Mitnitski

(2005). A global clinical measure of fitness and frailty in elderly people. Canadian Medical Association Journal, 173, 489-495.

33.

Saravay

S. M.

Kaplowitz

Kurek

Zeman

Pollack

Novik

. . . Hoffman

(2004). How do delirium and dementia increase length of stay of elderly general medical inpatients? Psychosomatics, 45, 235-242.

34.

Shen

H. N.

C. L.

C. Y.

(2012). Dementia increases the risks of acute organ dysfunction, severe sepsis, and mortality in hospitalized older patients: A national population-based study. PLoS ONE, 7(8), Article e42751.

35.

Smith

J. L.

Lindley

R. I.

(2009). The assessment of frailty in older people in acute care. Australasian Journal on Ageing, 28, 182-188.

36.

Sternberg

S. A.

Wershof Schwartz

Karunananthan

Bergman

Mark Clarfield

(2011). The identification of frailty: A systematic literature review. Journal of the American Geriatrics Society, 59, 2129-2138.

37.

Theou

Brothers

T. D.

Mitnitski

Rockwood

(2013). Operationalization of frailty using eight commonly used scales and comparison of their ability to predict all-cause mortality. Journal of the American Geriatrics Society, 61, 1537-1551.

38.

van Zyl

L. T.

Seitz

D. P.

(2006). Delirium concisely: Condition is associated with increased morbidity, mortality, and length of hospitalization. Geriatrics, 61(3), 18-21.

39.

Wright

(2010). 2008-based national population projections for the United Kingdom and constituent countries. Population Trends, 139, 91-114.

40.

Zisberg

Shadmi

Sinoff

Gur-Yaish

Srulovici

Admi

(2011). Low mobility during hospitalization and functional decline in older adults. Journal of the American Geriatrics Society, 59, 266-273.