Prevalence of frailty and associated factors in a community-dwelling older people cohort living in Juiz de Fora , Minas Gerais , Brazil : Fibra-JF Study

Frailty is a state of multisystem physiological vulnerability related to aging and an increased risk of adverse outcomes. This study aimed to evaluate the prevalence and associated factors of frailty in the Fibra-JF Study, Minas Gerais, Brazil. We selected a random sample of 461 individuals aged 65 years or more stratified by territorial unit, gender and age. The frailty syndrome was established by the presence of three or more of five items: a feeling of exhaustion, low handgrip strength, slow gait speed, weight loss, and low caloric expenditure. The mean age was 74.4 (SD ± 6.8) years, 69.6% were women and 71.9% white. The prevalence of frailty was 5.2%; 49.9% was from pre-frail subjects. Advanced age (OR: 6.4; CI 1.76-23.8), impairment of the basic activities of daily living (OR: 5.2, CI 1.1-23.1) and self-perception of poor health (OR: 0.13, CI 0.030.4) were associated with frailty. In this study, a substantial number of individuals was classified as frail, while half of the sample was at risk of progression towards this condition, suggesting that it is urgent to adopt public health measures focused on frailty prevention and reduction of associated adverse health outcomes.


Introduction
The epidemiological and demographic transition and the social, scientific and technological achievements brought about necessary changes in the population composition and the pattern of morbidity and mortality to the world scenario 1 .While the increased life expectancy was gradual in some countries, in Brazil, it hiked from 43 to 74 years in less than a century 2 .With 650 thousand new older people added each year, Brazil will rank sixth in the number of individuals with 60 years or more 3,4 by 2050.This process of demographic transition carries a change in the distribution profile of diseases, with a lower frequency of infectious and contagious diseases, and a higher prevalence of chronic-degenerative diseases.Currently, these represent 74% of all causes of mortality in our country and many of them are causes of dysfunctions and disabilities 5,6 , among which is physiological frailty.
The frailty syndrome is a clinical condition that has been widely studied in the last decades 7 .It is recognized as a geriatric syndrome and is directly associated with undesirable outcomes, increase of cost and mortality [8][9][10] .Age is its main isolated risk factor, and it is estimated that 10 to 25% of the population over 65 years of age is frail 11 .For those over 85, such prevalence may reach 45% 11 .
The term frailty is common, and many are its vernacular meanings.However, in the scientific literature, the term has been used in the past -and to some extent still is -to describe individuals who are dependent on others who are at substantial risk of dependency and other adverse health outcomes, as a characteristic of those with many chronic diseases or complex medical and psycho-social problems 12 .Atypical evidence of diseases, being a candidate for specialized geriatric programs and rapidly aging are other conditions described by the term frailty in the scientific literature. 12owever, over the last two decades, the frailty syndrome has been discussed mainly around two conceptual models and their respective instruments.On the one hand, Rockwood and colleagues 13 define frailty as the set of problems -illnesses, dysfunctions, disabilities -of individuals, and proposes a comprehensive geriatric assessment as a method of identifying patients with this syndrome.On the other hand, considered as an objective manifestation of the interaction of genetic and environmental factors, the "phenotype" -also called physical frailty -proposed by Fried et al. 14 work on the assumption that, from a conceptual viewpoint, frailty is a state of age-related multisystemic physiological vulnerability associated with an increased risk of adverse health outcomes, and propose that it be diagnosed by the presence of three or more of a total of five items: feeling of exhaustion, low handgrip strength, slowing gait speed, weight loss and low caloric expenditure.The presence of one or two items would point to a borderline, intermediate status between the robust and frail individual, named by these authors as pre-frailty.
There is an understanding among researchers in the field that the frailty syndrome is something different than dependency and vulnerabilities of an economic and social nature.However, both conceptual systems identify individuals at risk of adverse health events and high mortality 12 .At least two characteristics of this condition justify this understanding: firstly, the impairment of the functional reserve of the various organic systems associated with chronological aging is relatively well reported in the literature; secondly, frail or pre-frail individuals recover, both spontaneously and through therapeutic interventions, whereas disability conditions are relatively "fixed", with fewer opportunities for recovery 12,[14][15][16] .
This study aimed to evaluate the prevalence and factors associated with frailty in the Fibra-JF study in Minas Gerais, Brazil.

Sample selection
This is a cross-sectional study with 65 years old or older individuals residing in the urban area of Juiz de Fora, Minas Gerais, Brazil.
According to the Brazilian national institute of statistics [Instituto Brasileiro de Geografia e Estatística (IBGE)], Juiz de Fora is divided into 16 territorial units, based on the criteria of geographic location and socioeconomic level.In 2001, the city had 500,000 inhabitants, of which 47,379 (11.6%) were aged 60 years or older 2 .In this study, individuals of both genders aged 65 and over were selected from this population by quota sampling criteria.The sample size by gender and age group, for a sampling error of 5%, was 385 individuals.In the first step, five of the 16 units (units 1, 4, 10, 11 and 13) were randomly selected.Next, the neighborhoods were selected in each unit and streets of the selected neighborhoods.The territorial units consist of districts subdivided into census tracts, which consist of streets defined in specific blocks.The proportionality of the older people in each region was preserved in the different districts visited.By adopting the door-to-door procedure, the houses of each street were visited, in a predetermined order, until the target for that location was reached.In total, 461 individuals were interviewed and 427 were analyzed.Thirty-four individuals were excluded from the study because of the following conditions: temporarily bedridden, residents of long-term care institutions, rural residents and people with disabilities -severe sequelae of stroke, immobility, advanced Parkinson's disease, and a score below 14 in the Mini-Mental State Examination (MMSE) (Figure 1).
Participants voluntarily completed an informed consent form and the rules of the National Research Ethics Committee of the Ministry of Health were respected.This work was supported by the National Council for Scientific and Technological Development (CNPq).

Data collection
The interviews were made face to face, at home, between January 2009 and January 2010, after team training.Participants answered a questionnaire consisting of sociodemographic; physical and mental health; health habits; self-perceived health; self-reported comorbidities (heart disease, pulmonary disease, systemic arterial hypertension, diabetes mellitus, cancer, stroke, osteoporosis, osteoarthritis, and depression) variables.Participants were also submitted to functional performance tests 17,18 and cognitive screening 19 .Also, the appendicular skeletal muscle mass was calculated by the Lee formula and those of the first quintile by gender were considered as having low muscle mass 20 .The individuals were evaluated through the five items that make up the scale of Fried et al. 14,21 , and the diagnosis of frailty was established using the frailty criteria proposed by these authors (Chart 1).

Statistical analysis
The distribution of the categorical variables was shown as absolute frequencies and percentages.Continuous variables were described through means and standard deviations.The variables that showed p < 0.2 in the univariate analysis were included in the multivariate logistic regression model by the stepwise forward method.The Hosmer-Lemeshow test was used to analyze the discriminatory capacity of the model.All descriptive statistics were calculated considering a 95% confidence interval, and a level of significance < 0.05 in the multivariate analysis.Data entry and statistical analysis were performed with SPSS version 19.9, IBM software, 2009, Chicago.

Results
The sample consisted of 427 individuals, of whom 69.6% were women, 71.9% were Caucasian, and 48% were married or lived with a partner (Table 1).The mean age was 74.4 years (± 6.8) and the educational level was 5.4 years (± 4.1).The mean income per capita expressed in minimum wages (MW) was 1.6.Of the nine self-reported comorbidities, the most prevalent were systemic arterial hypertension (37.9%), depression (14.5%), osteoporosis (10.5%) and diabetes mellitus (10.5%).Regarding falls, 32.1% of the sample had at least one fall in the previous year.The mean MMSE score was 24.55 (± 3.58).
In the present study the prevalence of frailty was 5.2%; 49.9% were pre-frail and 45% robust individuals.In Table 1, we observed that, among the non-frail, 67% were female, 71% were in the younger age group and 72% were Caucasians.

Weight loss
For verification of weight loss, those with self-reported unintentional loss of more than 4.5 kg in the year were considered positive.At follow-up, the body mass index (BMI) was measured and when the corresponding result for the previous year for weight was ≥ 0.05 by the formula below, without report of intentional loss, it was also considered positive for weight loss.(Weight in the previous year -Current Weight / Weight in the previous year).

Muscle strength
A manual dynamometer was used in the dominant upper limb to measure muscle strength, requiring the participant to exercise the highest possible force three times.Those in the first quintile, after adjusting the result for gender and body mass index, were considered positive for the item muscle weakness.

Feeling of exhaustion
Two questions of the Center for Epidemiological Study Center Scale (CES-D) were used to verify the feeling of exhaustion, namely, items seven and 20."I feel tired in everything I do" and "I cannot continue this way".Those who answered "yes" to any of the two questions were considered positive for the feeling of exhaustion question.

Slowing of gait score
The time spent to walk a 4.6m course in a straight line was evaluated to verify slowing of gait score and, after adjustment for height and gender, those of the first quintile were considered positive for the slowing of gait score.

Decreased physical activity
The Minnesota Leisure Time Activities (MLTA) questionnaire was used to verify the decreased physical activity and the weekly caloric expenditure was calculated for each activity performed by the individual.Those of the first quintile were considered positive for decreased physical activity.
Among the frail, females were predominant (68%) and 77% were older than 75 years.They also had lower schooling (68%), lower income (64%) and a substantial number of comorbidities (77%).In the item self-perceived health, 77% of frails reported that their health was poor/very poor and also had a worse score in the MMSE (mean of 21 points).
Table 2 shows that in the univariate analysis of advanced age, ADL impairment, negative self-perceived health, poor performance in MMSE, the presence of comorbidities and falls had a significant association with the dependent variable.In the multivariate analysis, only advanced age (OR: 6.4; 95% CI 1.76-23.8),impairment in ADL (OR: 5.2; 95% CI 1.1-23.1)and negative self-perceived health (OR: 0.13; CI 0.03-0.4)were shown to be associated with frailty.
Of the five items that make up the CHS frailty scale, Table 3 shows that slowing gait was the most affected item for the frails (95%).For the pre-frail, affected items were weight loss (57.7%), followed by low caloric expenditure (42.3%).

Discussion
Although circumscribing non-exclusively biological realms, instruments such as the Edmonton Frail Scale 22,23 and the Tilburg Frailty Indicator 24,25 have been shown as alternatives to identify frailty in Brazil.However, the proposal suggested by the frailty scale of Fried et al. 14 and its criteria is one of the most widely used methods of assessing frailty and prediction of adverse health events worldwide 23,[25][26][27] .
The following discussion will focus on three of these aspects: the first, the relationship between the prevalence values and the cut-off points of the items that underpin the scale used in this study; the second concerns the relationship between the conceptual and operational definitions of frailty; finally, we will discuss the state of the art of intervention on risk factors, particularly interested in the case of pre-frail individuals.
In Brazil, the prevalence of frailty in community-dwelling people, using the criteria proposed by Fried et.al. 14 ranged from 5.2% to 17.1% [28][29][30] .Among the plausible explanations for such variability is the way in which the normality criteria were defined for each item.
In the work of Sousa et al. 30 , in a sample of 391 older persons in the city of Santa Cruz, Rio Grande do Norte, Brazil, the prevalence of frailty was 17.1%.The authors suggest that such prevalence would be associated with the low socioeconomic status of this population.However, a methodological aspect should be mentioned as a central element to elucidate this high prevalence: handgrip strength and gait speed.In this work, the absolute values proposed by Fried et al. 14 were used, indeed suitable for the American population, but controversial when applied to different realities 31,32 .
In Brazil, other studies used the same CHS criteria to define cut-off points for these variables -the handgrip strength and gait velocity quintile, rather than the absolute values proposed by Fried  et al. 14 .This strategy generated much lower cutoff points, finding lower prevalence and closer to those also evidenced in international studies, both in high income and middle-income countries 28,33,34 .Therefore, there is enough evidence to recommend reference values for gait speed and handgrip strength adapted to the Brazilian population.
Although it is recognized as an operational framework for the identification of the frail individual, the CHS frailty criteria still must be widely discussed as to its applicability 32,[35][36][37] .On the one hand, concerning the operational issue, in a systematic review, Theou et al. 38 have found at least 262 different ways in which, in the last 15 years, the five CHS criteria have been clustered to establish a frailty evaluation instrument.On the other hand, regarding the conceptual definition, frailty is interpreted as a multisystemic syndrome associated with chronological aging.Therefore, changes in the items of this scale are supposed to be associated with changes in multisystemic physiological parameters, although, according to some authors, muscle mass can be an essential element in the development of frailty, either as a component of the initial stages or as an outcome undesirable by itself 35,39 .The items underlying the mentioned scale -weight loss, reduced caloric expenditure, declining handgrip strength, slower gait speed and feeling of exhaustion -suggest that this unique system, namely, the muscular system, is involved, without essential components of this pathophysiological model, such as the neurohumoral and immunological systems, are explicitly included 11 .
Finally, identifying individuals at risk for undesirable outcomes is an essential concern in the field of aging, and in the case of the frailty syndrome, intervention on these factors is an essential part of the healthcare of pre-frail individuals 40 .For example, Theou et al. 41 carried out a systematic review of the effectiveness of the exercise to manage frailty.Although they did not find uniformity in the type of exercise, their work evidenced that such intervention promoted a positive impact on the management of the syndrome.Tarazona-Santabalbina et al. 42 , in turn, identified that exercises with multiple components (upper limbs, lower limbs and trunk) were able to reduce the degree of frailty and improve functional capacity, gait speed, as well as cognitive, emotional and social aspects of the participants.With the objective of clarifying concerns regarding the effectiveness of a set of multidisciplinary, pharmacological and non-pharmacological procedures, Apóstolo et.al. 43 proposed a systematic review of preventive measures for the pre-frail individual.Papers that evaluate psychosocial, nutritional and physical stimulus interventions for the prefrail and frail will be included in the analysis of these authors.
Although evidence of rehabilitation of these individuals is still based on a small number of clinical trials, the search for intervention methods is a fundamental activity to establish effective measures to prevent frailty.The potential health benefits to the older population can be estimated taking into account that almost 50% of the sample of this study consists of pre-frail individuals, theoretically those who would most benefit from preventive activities.
This study has limitations due to its cross-sectional design and, consequently, the possibility of reverse causality associated with its results.However, because it is a population-based work with a significant number of individuals from the older people population of Juiz de Fora, Minas Gerais, their findings may also be extended to populations of medium-sized Brazilian cities.

Conclusion
The prevalence of frailty in this study was 5.2% and was associated with advanced age, ADL impairment, and poorer self-perception of health.

Figure 1 .
Figure 1.Sample selection by gender and age group of community-dwelling individuals aged 65 years and over living in the city of Juiz de Fora, Minas Gerais, Brazil.Fibra-JF.

Chart 1 .
Frailty scale proposed by the Cardiovascular Health Study: case definition.

Table 1 .
Sociodemographic, functional and health conditions of community-dwelling individuals over 65 years of age.Juiz de Fora, Brazil.Fibra-JF (n=427).

Table 3 .
Frequency distribution of frailty items by CHS in community-dwelling older people in Juiz de Fora (MG), Brazil.Fibra-JF (n = 427).