| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||
1 Department of Geriatrics, Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán, Mexico City, Mexico.
2 Centre de Recherche INSERM, U897, Bordeaux, France.
3 Université Victor Segalen Bordeaux 2, Bordeaux, France.
4 INSERM, U888, Montpellier, France.
5 Montpellier 1 University, Montpellier, France.
6 Department of Geriatrics, Centre de Champmaillot CHU Dijon, France.
Address correspondence to Jean-François Dartigues, MD, PhD, Centre de Recherche INSERM, U897, 146 rue Léo Saignat, 33076 Bordeaux cedex, France. E-mail: jean-francois.dartigues{at}isped.u-bordeaux2.fr
 |
A |
|---|
 Top Abstract MethodsResultsDiscussionReferences |
|---|
Methods. We used data from 6078 persons 65 years old or older participating in the Three-City Study (3C). Frailty was defined as having at least three of the following criteria: weight loss, weakness, exhaustion, slowness, and low activity. Principal outcomes were incident disability, hospitalization, and death. Multiple covariates were used to test the predictive validity of frailty on these outcomes.
Results. Four hundred twenty-six individuals (7%) met frailty criteria. Participants classified as frail were significantly older, more likely to be female, and less educated and reported more chronic diseases, lower income, and poorer self-reported health status in comparison to nonfrail participants. In multivariate analysis, frailty was significantly associated with 4-year incidence of disability in activities of daily living (ADL) and instrumental ADL. However, frailty was marginally associated with incident hospitalization and was not a statistically significant predictor of incident mobility disability or mortality adjusting for potential confounding factors.
Conclusions. Frailty is not specific to a subgroup or region of the world. The construct proposed by Fried and colleagues confirms its predictive validity for adverse-health outcomes, particularly for certain components of disability, thus suggesting that it may be useful in population screening and predicting service needs.
Key Words: Frailty • Community-dwelling • Validity • Prognosis • Elderly
To identify frail individuals, several criteria have been proposed in recent years (3,7–10). According to the criteria used, heterogeneous results regarding frequency have been obtained when applied in clinical practice (11,12). Nevertheless, there is a general agreement that the core feature of frailty is increased vulnerability due to impairments in multiple, inter-related systems resulting in homeostatic reserve disturbance (3,13–16). Multiple impairments are demonstrated by the presence of a combination of several clinical characteristics, and it seems unlikely that a single altered system is sufficient to explain this clinical state (1,17). Recently, a working group proposed a definition that conceptualizes frailty as a clinical syndrome defined as the combination of shrinking, weakness, exhaustion, low walking speed, and low physical activity (3). This conception of frailty implies a biological connection between all its components and is widely used.
To better understand the role of frailty in health outcomes for different subgroups, it is important to examine data from cohort studies across cultures to assess its ability to predict adverse outcomes in different populations. Therefore, the purpose of this report is to describe the characteristics and prognosis of persons classified as "frail" in a large sample of French community-dwelling elderly persons. The main hypothesis is that frail persons defined according to the criteria derived from the study by Fried and colleagues present more adverse outcomes such as the incidence of disability, more frequent hospitalization, and mortality, even after adjustment for potential confounders.
|
METHODS |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
Definition of Frailty
Frailty was defined according to the construct previously validated by Fried and colleagues in the Cardiovascular Health Study (3). All five components from the original phenotype were retained for this study; however, the metrics used to characterize the frailty criteria were slightly different and defined as follows:
Shrinking.--
Recent and unintentional weight loss of 
3 kg was identified and body mass index calculated. Participants who answered "yes" for weight loss or had a body mass index < 21 kg/m2 were considered to be frail for this component. This threshold is used in the Mini-Nutritional Assessment (19) and has been shown to be associated with increased mortality (20). In addition, it was previously associated with adverse outcomes in community-dwelling elderly persons in France (20,21).
Poor endurance and energy.-- As indicated by self-report of exhaustion, identified by two questions from the Center for Epidemiological Studies-Depression scale [CES-D (22)]: "I felt that everything I did was an effort" and "I could not get going." Participants were asked: "How often, in the last week, did you feel this way?" 0 = rarely or none of the time; 1 = some or a little of the time; 2 = a moderate amount of the time; or 3 = most of the time. Participants answering "2" or "3" to either of these questions were considered as frail by exhaustion.
Slowness.-- The slowest quintile of the population was defined at baseline, based on a timed 6-meter walking test, adjusting for gender and height as recommended. The lowest quintile was used to identify participants with slowed walking speed.
Weakness.-- Participants answering "yes" to the following question were categorized as frail for this component: "Do you have difficulty rising from a chair?" Grip strength, which evaluates the muscular power and force that can be generated with the hands, was not available in the 3C Study data set. However, a multidisciplinary expert consensus (nutritionist, neurologist, psychologist, and geriatrician) determined that the question was an adequate "proxy" for weakness. In addition, it was shown that grip strength significantly correlates with muscular power in other muscle groups among elderly persons [elbow flexion, knee extension, trunk extension, and trunk flexion (23)].
Low physical activity.-- A single response was used to estimate physical activity (24). Individuals who denied doing daily leisure activities such as walking or gardening and/or denied doing some sport activity per week were categorized as physically inactive. Those who reported doing them were considered to be physically active.
As proposed by Fried and colleagues, the participants were considered to be "frail" if they had three or more frailty components among the five criteria; they were considered "prefrail" or "intermediate" if they fulfilled one or two frailty criteria, and "nonfrail" if none (3).
Outcomes
Three measurements of disability were investigated as outcomes: mobility, instrumental activities of daily living (IADL), and basic activities of daily living (ADL). Mobility was assessed by the Guttman's health scale (25): doing heavy housework, walking a half mile, and going up the stairs. For the IADL, participants reported their ability to perform eight IADLs based on the Lawton and Brody scale: using the telephone, having responsibility for one's own medication, managing money, being able to transport oneself, shopping, grooming, doing housework, and doing laundry [the last three were only asked of women (26)]. For the ADL, participants were asked if they needed help for any task from the Katz ADL scale [bathing, dressing, transferring from bed to chair, toileting, and feeding (27)]. For each domain of disability, if participants indicated that they were unable to perform one or more activities without help, they were considered as having mobility, IADL, or ADL disability (25–27). The 4-year incidence of disability was established only among participants without prevalent disability in the same domain at baseline.
Four-year incident hospitalization was considered when the participants declared it either at the first follow-up (2 years) or subsequent follow-up interview (4 years). Cause and time of death were obtained from interviews with family or from medical records at both follow-ups, and treated as cumulative 4-year mortality.
Covariates
Sociodemographic variables included age, sex, marital status, educational level, living alone, and monthly income. Participants were asked whether they had a physician's diagnosis of cardiac failure, myocardial infarction, angina pectoris, chronic obstructive pulmonary disease, fractures during the two preceding years (femoral or vertebral), cancer diagnosis, or arthrosis. Participants were considered as hypertensive if self-reported or systolic blood pressure was 160 mmHg or diastolic blood pressure was 95 mmHg or if they were on antihypertensive medications. Participants were considered as diabetics if self-reported or having high glucose level ( 7.0 mmol/L) or they were on hypoglycemic treatment (oral diabetic medications or insulin). The presence of each of these diseases was summed up in a score ranging from 0 to 9, where a higher score indicates more chronic disease. Self-reported health was also recorded and treated as a categorical variable (good, regular, or poor).
Depressive symptoms were assessed using the CES-D [20-item version (22,28)]. For the multivariate analyses, the two questions used for the frailty definition were excluded from the total CES-D score. Depressive symptoms were used as a continuous variable, and a higher score represents a worse mood.
The Mini-Mental State Examination [MMSE (29)] was used to assess global cognitive function (0–30 points; higher score indicates better cognitive status).
Smoking status (nonsmoker, former smoker, or current smoker) and usual alcohol intake (nondrinker, former drinker, or current drinker) were self-reported.
Plasma cholesterol total levels were used as continuous variable.
Sample
For the present research, only participants from two cities were considered, because in Montpellier, the timed walking test was not administered. Moreover, of the 7188 participants interviewed at baseline in Bordeaux and Dijon, those with conditions that could be a consequence of a single disease and not of generalized frailty as already proposed were excluded (3). In contrast, participants whose frailty status could not be determined (missing data) were also excluded (Figure 1). As expected, those excluded were significantly older (78.4 vs 74.1 years), more depressed (mean CES-D score 13.1 vs 8.1), and more likely to be disabled for mobility (80.6% vs 44.9%), IADL (29.3% vs 8.1%), and ADL (1.8% vs 0.4%). Data for 2354 (38.7%) men and 3724 (61.3%) women, who had complete clinical and functional data at baseline, were finally included in the statistical analyses. Four-year incidence outcomes were computed as the sum of information concerning to the 2- and 4-year follow-ups.
|
|
RESULTS |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
85 years), the differences between the sexes were statistically significant for the three domains of disability (p <.001). Fifty-four percent of participants reported to be completely autonomous for the three domains of disability evaluated.
|
|
The unadjusted results showed that, in comparison to the nonfrail subgroup, the prefrail and frail subgroups had significantly higher risks of incident disability for mobility and IADL However, for incident ADL disability, there were significant differences between frail and nonfrail participants, but not between prefrail and nonfrail participants (Table 3). Multivariate logistic regression analyses showed that, after adjusting for sociodemographic and health covariates, there were significant differences between the prefrail and nonfrail subgroups, but not between frail and nonfrail participants for the incidence of disability for mobility. For incident ADL disability, there were significant differences between frail and nonfrail, but not between prefrail and nonfrail participants, whereas for incident IADL disability, the relationship between prefrail and nonfrail, and frail and nonfrail subgroups remained significant (Table 3).
|
The unadjusted regression analyses showed that prefrail and frail statuses were significantly associated with the incidence of hospitalization (Table 4). Multivariate logistic regression analyses showed that, after adjusting for all covariates mentioned above, including disability for mobility, IADL, and ADL at baseline, there were significant differences between frail and nonfrail, but not between prefrail and nonfrail subgroups associated to incident hospitalization; however, the overall association was only marginally significant.
|
The unadjusted Cox proportional hazards model showed that, in comparison to being nonfrail, being frail at baseline significantly increased the risk of cumulative death at 4 years, whereas there were no significant differences between prefrail and nonfrail participants (Table 5). After adjusting for sociodemographic and health covariates (including disability for mobility, IADL, and ADL at baseline), frailty was no longer a statistically significant predictor of death (Table 5 and Figure 2).
|
|
|
DISCUSSION |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
The main limitation of this study may be the use of slightly different measures to define frailty criteria [because the measures originally used by Fried and colleagues (3) were not available in the 3C Study]. Using incomplete scales in multivariate analyses may be a limit also. In particular, with the exclusion of two items from the CES-D, the modified score slightly differs from the validated one. This score was computed to minimize the colinearity between frailty and mood. In addition, exclusion of participants with missing frailty scores (around 6.5% from the original sample) could induce a selection bias and affected the results, and a lack of power could explain the lack of relationship between all subgroups of frailty with the adverse outcomes in the adjusted analyses.
Nonetheless, despite these limits, the prevalence of frailty in elderly French persons (around 7%) was similar to that reported in other studies carried out in community-dwelling white people. Indeed, persons classified as frail were more likely to be older and female or to have more health problems. Besides the association with comorbidity and disability, adverse social conditions such as living alone or low income were also more frequent among frail and prefrail people. However, it is necessary to insist that frailty overlapped, but is not synonymous with, comorbidity and disability. Thus, as shown, not all frail participants were disabled at baseline, and not all who had a 4-year incident disability were frail at baseline. These findings support the hypothesis that frailty often precedes disability and that they are distinct entities. However, even if in this study the strength of the association between frailty and incident IADL and ADL disability was as previously reported, the phenomenon was not a statistically significant predictor of incident mobility disability after adjusting for multiple covariates. The exclusion of individuals with a prior mobility disability led to the exclusion of the most vulnerable persons, reducing substantially the risk for incident mobility disability.
In contrast, the amplitude of the confidence intervals in the significant strength of association between frailty and ADL disability in multivariate analyses may be explained by the low number of participants affected by this disability over time, possibly because of the general good health of participants at baseline or the relatively short period of follow-up (4 years).
Using data from the Women's Health and Aging Studies I and II, Bandeen-Roche and colleagues analyzed the number of categories or "classes" of frailty that are necessary to better capture its heterogeneity (two-class model: nonfrail and frail; three-class model: nonfrail, prefrail, and frail). The results showed that the two-category model is the most relevant. Considering our results showing that only one of the categories of frailty was associated with the incidence of mobility and ADL disability, and hospitalization, our study also suggests that the two-class model performs better than the three-class model. As it has been proposed, frailty may represent one extreme of a health continuum, and the inconsistency of an "intermediate condition" in predicting middle-term adverse outcomes could be explained by the longer duration of this status before "true frailty" and its consequences manifest.
In addition to physical aspects, other domains have to be considered to define frailty (33,34). Among the age-related conditions that could potentially be included, cognitive impairment is a good candidate. In this study, frail participants showed worse performance on the MMSE in comparison to prefrail and nonfrail subgroups. Although previously reported (35,36), the relationship between frailty and cognitive decline is largely debated. Both could share etiologic mechanisms, including chronic inflammation (37).
Fried's definition of frailty proves to be reproducible and relevant to predict different adverse outcomes through different populations showing its predictive validity. The use of a standardized phenotype will lead to a comparison between different populations and will possibly serve to identify etiological factors, components, or other correlates of frailty. This approach may be an acceptable option and awaits studies that consider the frailty concept as their principal objective of research. Nevertheless, important advances have occurred in the field with the proposal of methodological guidelines to include frail people in future research (38).
Despite the limits previously mentioned, this study has several strengths. Previous research in frailty has been conducted in France (39,40). However, this is the first one that uses a definition of frailty widely acknowledged to identify the affected individuals and to report its characteristics and prognosis. In addition, the study was conducted in a large population-based sample and had a prospective design.
Exploration of other possible dominions of frailty is necessary. Understanding the medical, biological, and environmental factors that contribute to the phenomenon of frailty is the goal of current research in the field (38). Elderly persons who are frail would benefit from complex, multidisciplinary care compared with usual care (41,42), which explains why efforts must be directed to detect this clinical state before irreversible disability or other adverse outcomes appear. Prospective research is required to ascertain whether intervention programs targeting frail persons may delay or reverse disability and loss of autonomy.
|
A |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
This paper was presented, in part, as a poster at the 60th Annual Scientific Meeting of The Gerontological Society of America in San Francisco, California (November 16–20, 2007).
|
F |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
Received July 12, 2007
Accepted January 21, 2008
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
