Age, frailty, disability, institutionalization, multimorbidity or comorbidity. which are the main targets in older adults?

Abstract

Objectives: Age, frailty, disability, institutionalization, multimorbidity or comorbidity are main risk factors for serious health adverse outcomes in older adults. However, the adjusted relevance of each of them in order to determine which characteristics must be of importance for health policies in this population group, has not been established. Design: Concurrent population-based cohort study. Setting: Albacete city, Spain. Participants: 842 participants over age 70 from the FRADEA Study. Measurements: Age, gender, institutionalization, frailty (Fried's criteria), previous disability in basic activities of daily living (BADL) (Barthel index), comorbidity (Charlson index), and multimorbidity (≥ 2 from 14 selected diseases) were recorded in the basal visit. The combined event of mortality or incident disability in BADL was determined in the follow-up visit. The risk of presenting adverse events was determined by Kaplan-Meier analysis and logistic regression adjusted for age, sex, and institutionalization. Results: Mean follow-up 520 days. 63 participants died (7.5%). Among the remaining 779, 191 lost at least one BADL (24.5%). The combined event of mortality or disability was present in 254 participants (30.2%). Age (OR 1.10, 95%CI 1.06–1.14), frailty (OR 3.07, 95%CI 1.63–5.77), disability (OR 2.19, 95%CI 1.43–3.36) and institutionalization (OR 2.73, 95%CI 1.68–4.44) were independently associated with the combined adverse event, but not comorbidity or multimorbidity. In subjects younger than 80, only frailty, disability and institutionalization were risk factors, and in those aged ≥ 80, only age, disability and institutionalization were. Conclusions: Health policies for older adults must take into account mainly frailty and disability in subjects younger than 80 and disability in those older than 80.

Introduction

Healthcare systems must be based on targeted populations at risk for adverse health outcomes (1). In older adults, main adverse health outcomes include mortality, incident disability, institutionalization and hospitalization. Age, disability, frailty, comorbidity or multimorbidity have all been related with adverse outcomes in older adults, but it is not well known neither the relative importance of each one as a predictor of these adverse outcomes, nor changes with age of these associations.

Recently, multimorbidity is being presented as the cornerstone of health policies for older adults because it represents a shift from the traditional single disease paradigm to a more holistic patient centered approach (2, 3, 4, 5, 6). Multimorbidity is defined as the co-occurrence of two or more chronic diseases in a specific period of time, and is related to age (7), with a prevalence estimates of 65-98% in older adults 2, 2, 6. It has been associated with mortality, hospitalization and longer hospital stays, institutionalization, lower quality of life, loss of physical functioning, depression, multiple drug use and higher health care utilization and costs 3, 8.

Frailty has been recently defined as a “A medical syndrome with multiple causes and contributors that is characterized by diminished strength, endurance, and reduced physiologic function that increases an individual's vulnerability for developing increased dependency and/or death” (9), and is a detectable, preventable, and treatable pre-disability state that develops in young older adults and progresses to disability as they grow older 9, 10. Frailty prevalence is also related to age, ranging from 3.2% among 65- to 70-year-olds to 23.1% among people aged 90 years and older 11, 12. Since 2001, frailty has been considered an important predictor of health outcomes in the elderly, such as death, institutionalization, falls, reduced mobility, hospitalization, and increased dependence in basic activities of daily living (BADL), and instrumental activities of daily living (IADL) (11, 13, 14, 15).

Disability is the difficulty of coping with BADL or IADL, and it is usually measured with validated instruments as the Barthel index for BADL or the Lawton index for IADL. It increases with age, and has also been associated with mortality, incident disability, hospitalization, length of hospital stay and institutionalization 15, 17.

Few studies have compared the adjusted association between disability, multimorbidity or comorbidity and adverse health outcomes (18, 19, 20), but none has included frailty in the analysis, or differentiated between youngest (aged 70 to 79) and oldest old (aged 80 or over). In the present study we analyze the longitudinal association between frailty, disability, multimorbidity or comorbidity and mortality or incident disability in a cohort of older adults over age 70 representative of the population in Albacete, Spain, to determine which of these characteristics better identifies high risk older adults.

Method

Subjects and Study Setting

Our study presents data from the first and second waves of the FRADEA (Frailty and Dependence in Albacete) Study. The rationale, design, methodology, selection of subjects, and baseline characteristics have been previously described (21). Briefly, 1172 subjects were randomly selected from the population aged 70 years or more from the city of Albacete (n=18,137), of whom 993 (84.7%) agreed to participate. In the second cut, 958 participants agreed to continue (96.5% of the initial cohort).

Study design

Population based concurrent cohort study. The baseline interview was carried out in person by four trained nurses at the Geriatric facilities of the Hospital del Perpetuo Socorro in Albacete between November 2007 and November 2009. The nurses went to the homes of participants who could not come to the site to obtain the required information. The follow-up visit was conducted by telephone by the same four nurses 18 months after the baseline visit.

Study Variables

At the baseline visit, age, sex, institutionalization, previous disability in BADL using the Barthel Index, comorbidity using the Charlson index, multimorbidity and frailty with Fried's criteria were recorded. The Barthel index is an ordinal scale that measures bathing, grooming, dressing, feeding, toilet use, urinary and fecal incontinence, transferring, walking 50 meters and stair use, ranging from 0 (dependence in all BADL) to 100 (independence in all BADL). Baseline BADL disability was considered when the participant required aid to perform the bathing, grooming, dressing, feeding, or toilet use activities from the Barthel Index.

Chronic diseases were identified from the participant's medical records. Diseases were coded according to the ICD-10 and thereafter classified within large homogeneous groups for later analysis. The Charlson Comorbidity Index was used to analyze comorbidity. A Charlson Comorbidity Index rating of 3 or higher represented high comorbidity. Multimorbidity was considered when the participant presented two or more of the following 14 selected chronic diseases: hypertension, dislipemia, diabetes, depression, coronary disease, atrial fibrillation, chronic obstructive pulmonary disease (COPD), dementia, asthma/bronchial hypereactivity, stroke, non skin cancer (except melanoma), heart failure, anemia, and parkinsonism.

During the baseline visit, the frailty criteria proposed by Fried (11) were recorded, with one small modification in the physical activity criteria. Subjects were classified as frail when three or more of the five criteria were present, and pre-frail when one or two were present.

The methodology for application of all the assessment tools used as well as bibliographical references can be found in the original study (21).

Outcome Variables

The main outcome variable was the combined event of mortality or incident disability in BADL. During the telephone interview, death was recorded along with the date it occurred. When this was unknown, the death registry of the Complejo Hospitalario Universitario and that of the Albacete primary care facility were consulted. Incident disability in BADL was defined as presenting lower scores on the Barthel Index in bathing, grooming, dressing, feeding, or toilet use in the follow-up visit compared to the baseline visit. The combined event was considered when mortality or incident BADL disability were present.

Ethical Aspects

Our investigation complies with the standards of the Helsinki declaration concerning investigation with human subjects. The study was approved by the Albacete Health Region Independent Ethics Committee. All participants signed an informed consent form prior to inclusion in the study.

Statistical Analysis

A descriptive analysis of the characteristics of the sample was performed, and chi square and t-Student tests were used to analyze the association between each independent variable and the events recorded. The association between frailty, disability, comorbidity or multimorbidity and the combined event was analyzed with Kaplan-Meier curves with bivariate comparisons using log rank tests. The observations were independent and censoring was non-informative.

The association between frailty, disability, comorbidity or multimorbidity and mortality, incident disability in BADL, or the combined event was analyzed adjusted for the study covariables (age, sex, and institutionalization) using logistic regression models. In model 1 high comorbidity was considered when the Charlson index score was ≥ 3, and in model 2 multimorbidity was considered when 2 or more of the 14 selected diseases were present. Finally, the same associations were determined for the combined event in participants younger or equal/older than 80 years. Interactions between all study covariates and the combined event were analyzed before selecting the final models. In models with only 2 independent variables, interaction between age-frailty, age-institutionalization, sex-frailty, sex-comorbidity, disability-institutionalization and frailty-institutionalization were detected, but in models with all the variables, only the interactions between frailty-age, and disability-institutionalization reached statistical significance. However, the inclusion of these interactions in the final models did not significantly change the adjusted association between covariates and the combined event.

All data were stored and analyzed using the SPSS 17.0 program.

Results

Participant's mean age was 78.6, with 498 females (59.1%), and 131 (15.6%) institutionalized. 162 participants were frail (19.2%), 469 (55.7%) prefrail and 211 (25.1%) robust. Mean Barthel index for the global sample was 89 and Charlson index 1.2. 235 (27.9%) subjects were considered to have basal disability in any BADL, and 124 (14.7%) to have high comorbidity. The mean count number of the 14 selected diseases for multimorbidity was 2.3, and 580 (68.9%) of participants had multimorbidity. The prevalences for the 14 selected diseases for multimorbidity were hypertension 73.8%, dislipemia 36%, diabetes 23.4%, depression 14.5%, coronary disease 11.8%, atrial fibrillation 11.3%, COPD 10.6%, dementia 9.9%, asthma/bronchial hypereactivity 9.4%, stroke 9%, non skin cancer 7.2%, heart failure 6.2%, anemia 4.6%, and parkinsonism 2.5%.

Mean follow-up was 520 days (SD 164). During the follow-up 63 participants died (7.5%) with a mean time to death of 390 days (SD 230), and of the remaining 779 alive (92.5%), 191 lost one or more BADL (24.5%), 561 (72.0%) did not develop a new disability and in 27 cases (3.5%) data were lacking.

Table 1 shows the differences in the study variables between participants with or without the health outcomes recorded. In bivariate analysis, death was associated with older age, institutionalization, frailty, disability in BADL, high comorbidity, and multimorbidity, but not with sex. However, incident disability in BADL was associated with older age, female sex, institutionalization, frailty, disability in BADL and multimorbidity, but not with high comorbidity, and the combined event was associated with all the study covariables.

Table 1.

Basal characteristics of the sample

		Mortality		Incident disability BADL		Mortality or incident disability BADL
		Yes (n=63)	No (n=779)	Yes (n=191)	No (n=561)	Yes (n=254)	No (N=561)
Age		83.4 (6.1)‡	78.2 (5.8)‡	81.9 6, 6‡	77.1 (5.0)‡	82.3 (6.5)‡	77.1 (4.5)‡
Sex	Male	33 (9.6)	311 (90.4)	58 (19.3)†	243 (80.7)†	91 (27.2)*	243 (72.8)*
	Female	30 (6.0)	468 (93.0)	133 (29.5)†	318 (70.5)†	163 (33.9)*	318 (66.1)*
Institutionalization
	Yes	26 (19.8)‡	105 (80.2)‡	65 (62.5)‡	39 (37.5)‡	91 (70.0)‡	39 (30.0)‡
	No	37 (5.2)‡	674 (94.8)‡	126 (19.4)‡	522 (80.6)‡	163 (23.8)‡	522 (76.2)‡
Barthel index		71 (23)‡	90 (17)‡	80 (21)‡	93 (14)‡	78 (22)‡	93 (14)‡
Charlson index		2.2 (1.8)‡	1.1 (1.4)‡	1.3 (1.4)*	1.1 (1.4)*	1.6 (1.5)‡	1.1 (1.4)‡
Number 14 diseases	2.9 (1.6)†	2.2 (1.4)†	2.4 (1.5)*	2.2 (1.4)*	2.6 (1.6)†	2.2 (1.4)†
Frailty	Robust	3 (1.4)‡	208 (98.6)‡	21 (10.4)‡	181 (89.6)‡	24 (11.7)‡	181 (88.3)‡
	Prefrail	32 (6.8)‡	437 (93.2)‡	108 (26.0)‡	308 (74.0)‡	140 (31.3)‡	308 (68.8)‡
	Frail	28 (17.3)‡	134 (82.7)‡	62 (46.3)‡	72 (53.7)‡	90 (55.6)‡	72 (44.4)‡
Disability BADL
	Yes	46 (19.6)‡	189 (80.4)‡	92 (49.7)‡	93 (50.3)‡	138 (59.7)‡	93 (40.3)‡
	No	17 (2.8)‡	590 (97.2)‡	99 (17.5)‡	468 (82.5)‡	116 (19.9)‡	468 (80.1)‡
High comorbidity
	Yes	18 (14.5)†	106 (85.5)†	32 (30.5)	73 (69.5)	50 (40.7)*	73 (59.3)*
	No	45 (6.3)†	673 (93.7)†	159 (24.6)	488 (75.4)	204 (29.5)*	488 (70.5)*
Multimorbidity
	Yes	49 (8.4)	531 (91.6)	142 (27.7)*	370 (72.3)*	191 (34.0)†	370 (66.0)†
	No	14 (5.3)	248 (94.7)	49 (20.4)*	191 (79.6)*	63 (24.8)†	191 (75.2)†

All data are means (standard deviation) or number of participants (percentage). BADL : Basic activities of daily living. ‡ p<0.001, †p<0.01, * p<0.05.

Figure 1 presents the unadjusted probability of not presenting the combined event (mortality or incident disability in BADL) for frailty (panel A), disability in BADL (panel B), high comorbidity (panel C), and multimorbidity (panel D) using Kaplan-Meier analysis. Mean combined event-free time was 939 ± 43 days for the non-frail, 694 ±15 among the pre-frail, and 619 ± 20 for frail subjects(Log-rank χ2 57.1, p<0.001), 614 ± 15 days among those with previous disability and 820 ± 25 days in those without (Log-rank χ2 60.9, p<0.001), 654 ± 30 days for those with high comorbidity, and 753 ± 18 among those without (Log-rank χ2 7.8, p=0.005), and 692 ± 15 days for those with multimorbidity and 816 ± 33 among those without (Log-rank χ2 10.0, p=0.002).

Figure 1.

Combined event analysis depending on frailty status (panel A), disability in BADL (panel B), high comorbidity (panel C), or multimorbidity (panel D). Kaplan-Meier analysis

Table 2 shows adjusted models of association between age, frailty, disability, institutionalization, comorbidity, multimorbidity and the three events recorded. Comorbidity and multimorbidity were not associated with any of the events recorded, or with the combined event either. However, frailty was associated with the combined event and with incident disability in the three adjusted models, and with mortality in models 1 and 3. Age, institutionalization and previous disability were also associated with all the events in all the models analyzed.

Table 2.

Models of association between age, sex, institutionalization, frailty, disability, comorbidity, and the three events recorded

	Mortality	incident disability	Combined event	Combined event OR (95%CI)
	OR (95%CI)	OR (95%CI)	OR (95%CI)	Age < 80	Age ≥ 80
		Model 1
Age	1.07 (1.02-1.12) †	1.09 (1.06-1.13) ‡	1.10 (1.06-1.14) ‡	1.06 (0.96-1.16)	1.13 (1.05-1.21) †
Female sex	0.34 (0.19.0.62)‡	1.16 (0.77-1.74)	0.90 (0.62-1.31)	1.04 (0.60-1.79)	0.81 (0.48-1.36)
Institutionalization	1.58 (0.83-3.03)	2.69 (1.59-4.55) ‡	2.73 (1.68-4.44) ‡	3.69 (1.33-10.23) *	2.29 (1.31-3.98) †
Frailty status
Prefrail	3.16 (0.90-11.04)	1.93 (1.12-3.31) *	2.14 (1.28-3.57) ‡	3.64 (1.70-7.78) †	1.05 (0.49-2.23)
Frail	4.18 (1.09-16.04)*	2.51 (1.27-4.93) †	3.07 (1.63-5.77) ‡	6.87 (2.58-18.31) ‡	1.33 (0.56-3.19)
Disability in BADL	4.24 (2.07-8.71)‡	1.71 (1.07-2.75) *	2.19 (1.43-3.36) ‡	1.91 (0.96-3.78)	2.42 (1.38-4.23) †
High comorbidity	1.45 (0.74-2.84)	1.21 (0.72-2.03)	1.28 (0.80-2.06)	1.62 (0.82-3.21)	0.97 (0.51-1.85)
		Model 2
Age	1.07 (1.02-1.12) †	1.09 (1.06-1.13) ‡	1.10 (1.06-1.13) ‡	1.05 (0.96-1.16)	1.13 (1.05-1.21) †
Female sex	0.31 (0.18.0.56)‡	1.13 (0.75-1.69)	0.87 (0.60-1.25)	0.94 (0.55-1.60)	0.82 (0.49-1.36)
Institutionalization	1.53 (0.80-2.94)	2.66 (1.57-4.49) ‡	2.69 (1.66-4.37) ‡	3.64 (1.14-10.07)*	2.34 (1.34-4.08) †
Frailty status
Prefrail	3.26 (0.93-11.40)	1.93 (1.12-3.32) *	2.15 (1.29-3.59) ‡	3.60 (1.68-7.71) †	1.07 (0.50-2.29)
Frail	4.38 (1.14-16.87)*	2.55 (1.30-5.02) †	3.15 (1.67-5.92) ‡	6.88 (2.59-18.27) ‡	1.37 (0.57-3.29)
Disability in BADL	4.53 (2.22-9.23)‡	1.76 (1.10-2.81)*	2.26 (1.48-3.45) ‡	1.98 (1.01-3.91) *	2.45 (1.41-4.24) †
Multimorbidity 14	1.09 (0.55-2.14)	0.88 (0.68-1.57)	1.06 (0.72-1.57)	1.50 (0.83-2.70)	0.74 (0.42-1.28)

Model 1: High comorbidity is considered as a Charlson index score ≥ 3. Model 2: Multimorbidity is considered when 2 or more of the 14 selected diseases are present. Disability in BADL is considered when lower scores on the Barthel Index in bathing, grooming, dressing, feeding, or toilet use are present in the follow-up visit compared to the baseline visit. OR: Odds Ratio. 95%CI: 95% Confidence interval. BADL: Basic activities of daily living. ‡ p<0.001, †p<0.01, * p<0.05.

Moreover, table 2 presents adjusted models of association between age, frailty, disability, institutionalization, comorbidity, multimorbidity and the combined event, both in participants younger than 80 years and in those aged 80 or over. While frailty, previous disability and institutionalization are independently associated with the combined event in the younger participants in all the three models, only age, disability and institutionalization are in the oldest ones. Neither comorbidity nor multimorbidity were associated with the combined event.

Discussion

The primary conclusion of our study is that age, institutionalization, frailty and disability are associated with mortality and incident disability in BADL in Spanish subjects over age 70, but multimorbidity and comorbidity are not. Moreover, in subjects younger than 80 years, only institutionalization, frailty and disability are associated with the combined adverse event, while in those older than 80, only age, institutionalization, and disability are.

It is a topic of great interest to determine the individual characteristics that identify subpopulations of older adults at high risk of heath adverse outcomes, in order to implement health policies specifically adapted for this age group, the principal user of the healthcare systems. In the last years, there has been a movement towards the implementation of health services for older adults based only on comorbidity or multimorbidity (2, 3, 4), although the absence of longitudinal studies supporting this rationale. Geriatric Medicine has always stated that health services for older adults must be based not only on chronic diseases, but also on age and function, due to the clear association between age and disability with serious adverse outcomes 22, 23. Recently, the identification of frailty defined as a phenotype (11) or as a deficit accumulation, a pre-disability state in the continuum of the disablement process (22), has also emerged as a clear risk factor for adverse outcomes in older adults (11, 12, 13, 14, 15). Our data are consistent with this finding, especially in the youngest old.

Perhaps the most important reason that could explain the exclusion of comorbidity or multimorbidity from the models is that older adults with multimorbidity are heterogeneous in terms of illness severity, frailty, functional status, mental status, geriatric syndroms, prognosis, personal priorities, and risk of adverse events even when diagnosed within the same pattern of condition (5). Thus, it is not the disease but the underlying disability, biological vulnerability or frailty, age-related conditions, time to reach the adverse event conditioned by age, and model of care what will determine the risk of adverse events. Sanitary models and treatment options in older adults need more flexible approaches taking into account their particular characteristics, and not only the presence and number of medical diagnosis. For example, heart failure and diabetes should be clearly treated in a different way in a subject of 50 years than in another of 85, although they have the same multimorbidity or illness severity. Age, frailty, disability, and geriatric syndromes must be the hallmarks that modulate the needs of care in older adults (24).

In our sample, 15.6% of participants were institutionalized. It could be argued that the lack of association between comorbidity and multimorbidity with the combined event could be determined by high comorbidity in this subgroup of participants. However, institutionalized participants did not have higher frequency of increased comorbidity than the non-institutionalized ones (16.8%-14.3%). Nevertheless, the combined event was more prevalent in the first ones (70.0%-23.8%). Furthermore, there was no interaction between institutionalization and high comorbidity in logistic models. These findings support the conclusion that age, frailty and disability, and not only comorbidity or multimorbidity, must be main targets for older adults policies.

Although frailty and disability are well defined, a great challenge lies in the definition of multimorbidity and comorbidity. There is no consensus yet as to which health conditions or diseases should be considered, assessed, summarized and weighted in order to arrive at some overall measure of burden of illness. In our study we used two different constructs to minimize de biases due to this challenging definition. It has been described that considering 4 to 7 diagnoses would lead to an underestimation of the prevalence of multimorbidity, while the inclusion of 12 or more diagnoses does not yield much variation (2). It is also not clear which specific diseases should be included, but a list of the most prevalent chronic diseases with a high impact or burden in a given population would be a good compromise. We selected 14 diseases with high prevalence and important impact on mortality and disability. We excluded osteoarthritis, osteoporosis and sensory impairment from the most common diseases due to low recording in the medical history when symptoms are mild, which could lead to underestimation.

Another important bias in research related to multimorbidity is the under-diagnosis due to data collection. In our study chronic diseases were collected directly from the participant's medical records by the same investigator. To confirm the validity of our data, we compared the observed study prevalences with those of the 2006 National Spanish Survey of Health and Diseases, showing similar values (25).

We cannot forget that frailty, disability and multimorbidity, although different concepts, are clearly interrelated one to each other (12), and that the association between comorbidity and disability is known from long time ago (20). The possibility of interactions among diseases to produce frailty has been described (26). Nevertheless, it is well known that the physiopathology of frailty is based on several physiologic impairments, dysregulations and inefficiencies, low grade chronic inflammation, oxidative stress, hormone age-related changes, sarcopenia, genetic susceptibility and changes in the energetic balance, rather than on the presence of single or multiple diseases 9, 2, 28.

Our study has different practice implications. The first one is that health services and health policies for older adults should take into account age, frailty and disability, and not only comorbidity or multimorbidity. The second is that in older adults younger than 80, frailty prevention, detection, and treatment should be the main target for health policies, while in those older than 80, disability should be the target, probably because at that age frailty has already triggered detected or undetected disability, and the contribution of disability to adverse events exceeds that of frailty. Physical frailty can potentially be prevented or treated with specific approaches, such as exercise, protein-calorie supplementation, vitamin D, and reduction of polypharmacy (9). Thus, the identification of prefrail and frail older adults must be a clear target for screening in adults older than 70 years (9), in order to implement primary and secondary prevention to decrease incident disability.

There is a need for future studies in this field, implementing multidimensional health service models based not only on multimorbidity, but also on frailty and disability prevention and treatment, as the Gerontopôle one 29, 30. Age could help to stratify interventions, giving more emphasis to frailty in younger ages and to disability in more advanced ones.