Frailty-associated factors among Brazilian community-dwelling elderly people: longitudinal study

ABSTRACT BACKGROUND: Frailty among elderly people is associated with negative health outcomes. Through gaining better understanding of this syndrome over different time periods, healthcare actions that take predictive factors into consideration may be facilitated. OBJECTIVE: To identify factors associated with frailty syndrome among community-dwelling elderly people over a two-year follow-up. DESIGN AND SETTING: Longitudinal study on elderly people living in Uberaba (MG), Brazil. Methods: Elderly individuals were selected through multiple-stage conglomerate sampling from a national database. Participants were interviewed and evaluated in 2014 and again in 2016. Predictors were considered at the baseline, and frailty categories (frail, pre-frail or non-frail) at the follow-up. Frailty was identified based on the Fried criteria. Associations with socioeconomic factors, health status and physical performance were investigated using multinomial logistic regression. RESULTS: 353 individuals participated in both assessments. The final model showed that age over 80 years was predictive of both pre-frailty and frailty (odds ratio, OR 4.92; 95% confidence interval, CI: 1.57-15.38; OR 8.64; 95% CI: 2.05-36.35, respectively), while dependency regarding basic activities of daily living (OR 3.66; 95% CI: 1.22-11.02) and poor lower-limb physical performance (OR 7.87; 95% CI: 1.97-31.39) predicted frailty. A one-unit increased score for advanced activities of daily living decreased the frailty rate by 15% (OR 0.85; 95% CI: 0.74-0.99). CONCLUSION: Age over 80 years was predictive of pre-frailty and frailty, while dependency in basic activities of daily living and poor physical performance predicted frailty. A one-unit increased score for advanced activities of daily living decreased the frailty rate by 15%.


INTRODUCTION
Frailty among elderly people is considered to be a priority within public health. One reason for this is that presence of this syndrome predicts occurrences of adverse events that threaten the long-term sustainability of healthcare actions and systems. Moreover, frailty presents a negative influence on elderly people's quality of life. 1 Physical frailty is "a medical syndrome with multiple causes and contributing factors" that is characterized by impairment of "strength, endurance and physiological functions", thus leading to "greater individual vulnerability in developing functional dependency and/or death". 2 From an operational point of view, the two measurements of frailty that have been most used (with high validity and reliability) are Fried's frailty phenotype and Rockwood and Mitnitski's frailty index. 3 In a systematic review, frailty was found to be associated with several sociodemographic, physical, biological, lifestyle and psychological factors. 4 Moreover, some risk factors for frailty were identified, such as advanced age, female gender, black race, lower income, lower educational level, cardiovascular diseases, multimorbidity, functional impairment, poor self-rated health, depressive symptoms, cognitive impairment, obesity, undernutrition, smoking and alcohol use. 5 In Brazil, however, the available evidence is only recent and there is a lack of longitudinal studies analyzing the factors that determine frailty. 6,7 In a study on 207 community-dwelling elderly people who were followed up for 12 months, the factors associated with frailty that predicted worsening of frailty status were histories of cancer, urinary incontinence and reduced capacity to perform advanced activities of daily living. 6 Another one-year investigation conducted among 129 elderly people after hospital discharge did not identify any variables that were predictive of change (improvement or worsening) to frailty condition. 7 Feng et al. 5 considered that it was essential to determine the factors associated with frailty when developing interventions to prevent or reduce the frailty-associated burden among community-dwelling elderly people.

OBJECTIVE
Given the low number of studies within the elderly population of Brazil and the need to understand the factors that determine frailty, the aim of this study was to identify frailty-associated factors among community-dwelling elderly people over a two-year follow-up.  8 Population definition was done using a multiple-stage conglomerate sampling process. This process took into consideration the sectors defined by the Brazilian National Household Survey, with information from neighborhoods and streets that was made available by IBGE. Random household selection was conducted to identify elderly people in their homes.

Ethics
The sample for the present study was composed of individuals who met the following inclusion criteria: (a) they participated at presence of diseases that prevented the assessments. In the 2014 baseline assessment, 710 elderly people were interviewed.
In 2016, attempts were made to reach all the elderly people who had participated in the first stage of the survey (n = 710), in their homes. After the eligibility criteria and the losses had been taken into consideration (detailed in Figure 1; other reasons could be insufficient address or incomplete data), 353 elderly people were considered in the present investigation. Thus, these 353 individuals were evaluated both in 2014 and in 2016.
Because of the possibility of reading and comprehension problems, the interviews with the elderly people were conducted faceto-face in their homes. Therefore, interviewers (who were undergraduate and postgraduate students) were selected and trained regarding ethical issues within research and, additionally, they were accompanied by field supervisors (senior researchers).

Dependent variable
The presence of frailty syndrome was investigated using the five items that Fried et al. described as components of the frailty phenotype. 10 These were the following: (1) Presence of non-intentional weight loss, as assessed through the question "In the past year, did you lose 4.5 kg without intention?"; (2) Muscle strength loss verified based on handgrip strength, using a manual hydraulic dynamometer; the mean value from three measurements was obtained and the cutoff points proposed by Fried et al. 10 were used; (3) Self-reported exhaustion and/or fatigue, as measured through two questions: "Did you feel that you had to make an effort to take care of your habitual tasks?" and "Were you unable to move forward with your things?"; (4) Presentation of Activity Questionnaire (IPAQ). Elderly people presenting three or more of these items were classified as frail; those with one or two of these items were classified as pre-frail; and those with none of these items were considered to be robust or non-frail. 10 A detailed description of the components can be accessed in previous publications. [10][11][12]

Statistical analysis
Statistical analysis was done using the absolute and percentage frequency distribution for categorical variables and central trend (mean) and dispersion (standard deviation) measurements for quantitative variables. Univariate and multivariate analyses were done using logistic multinomial regression analysis, in order to investigate associations between the exploratory variables and the dependent variable (frailty status). Thus, the exploratory variables (predictors) were obtained from the baseline (2014) and the frailty status (frail, pre-frail or non-frail) was obtained from the follow-up assessment (2016). The variables of interest were chosen in accordance with the criterion established (P < 0.20) and were included in the multivariate regression model. Predictors associated with pre-frailty and frailty were identified using odds ratios, through multinomial logistic regression, considering a significance level of 5% (P < 0.05) and a 95% confidence interval (CI). The data were analyzed using the Statistical Package for the Social Sciences (SPSS), version 21.0.

RESULTS
In 2014, the majority of the 353 elderly people who were interviewed were women, in the age range of 60-69 years, and were living with a companion. Table 1 presents the distribution of the socioeconomic variables according to frailty status at the baseline.   Table 3).

DISCUSSION
The present study identified frailty predictors over a two-year follow-up period. These included advanced age, dependency relating to BADL and poor physical performance. On the other hand, ability to perform AADL provided a protective effect.
The results indicated that advanced age (80 years or over) was an independent predictor for both pre-frailty and frailty. Other investigations have also found that age was a frailty marker, 18-20 including two systematic reviews. 4,5 Age is an important indicator of the association between frailty categories and mortality. 21 A systematic review indicated that the numbers of pre-frail and frail elderly people become greater at advanced ages, which suggests that frailty is a progressive condition and, hence, that it may appear more frequently among elderly people older than 80 years. 5 Moreover, Fulop et al. 22 discussed the existence of common but non-identical pathways of frailty and aging; they suggested that the characteristics of frailty syndrome were more accentuated than those of regular ageing. Thus, all individuals older than 70 years would need to be screened for frailty syndrome, in order to improve the management of individuals with this condition. 2,23 The association of BADL dependency as a frailty predictor seen in the present study is divergent from the findings of other Brazilian studies. 6,7 Nevertheless, an investigation in Italy, with a 4.4-year follow-up, found that worsening of the condition presented by non-frail individuals was associated with dependency  in relation to activities of daily living. 18 Furthermore, according to Fried et al., 24 functional incapacity may cause difficulty in accessing healthcare services or actions from healthcare professionals, which would lead to increases in unrecognized and unaddressed healthcare needs. 24 Thus, implementation of monitoring actions and control over functional incapacity factors are strategies not only for maintaining functional capacity among elderly people, [25][26] but also for prevention of consequent conditions of frailty.
The present study found that an increase of one unit in the AADL score may have a protective effect against occurrences of frailty. These results are corroborated by an investigation among Brazilian elderly people that identified that the chance that frailty would worsen within 12 months was smaller (20%) when the elderly individual was categorized as "still doing" an AADL. 6 AADLs are complex activities involving social interaction, such as work or participation in community groups, meetings, cultural events, trips and other activities. 15 Hence, they represent integrity of physical function, social function and performance in social roles. 27 In addition, they are predictors of frailty. 28 Therefore, elderly people with active social networks are likely to be less frail than those with less social engagement. 29 Moreover, social participation and factors such as security, strong social cohesion and neighborhood belongingness 29 are protective and provide balance in community frailty levels. 30 Another frailty predictor is poor physical performance (4-6 points), as assessed using the SPPB. An Italian study with a mean follow-up period of 4.4 years found that poor physical performance (score lower than 8 points) was significantly associated with increased risk of becoming frail and with worsening frailty status. 18 Previous cross-sectional studies identified the feasibility of using the SPPB to detect frailty among elderly people (score lower than 9 points), 31 including detection of early signs of frailty before occurrence of slow walking speed among very old people (score of 8 points). 32 Cesari et al. 33,34 highlighted that the SPPB identified elderly people with greater vulnerability to stressors and elevated risk of negative health-related events, which are matters related to frailty syndrome. Therefore, these findings may explain the results from the present study.
The SPPB provides a simple measurement of physical performance that is easy to carry out, without any need for special equipment or extensive training for evaluators. 32 Furthermore, it is one of the clinical tools most used for identifying frailty. 35 Additionally, it provides a viable and objective definition for the complex concept of frailty, both in clinical practice and in research. 33,34 Among the limitations of the present study, there were considerable losses of follow-up. A further limitation was that absence of cognitive decline was considered to be an inclusion criterion in the present study, given that presence of cognitive decline could have interfered with comprehension of the variables analyzed (especially considering the self-reported nature of some of the data).
Moreover, it needs to be acknowledged that a relationship between frailty and cognitive decline exists.
In the light of the results from the present study and the fact that frailty is a highly prevalent syndrome in aging populations, 1 it is imperative to identify and manage this condition properly. 23 In this regard, knowledge of frailty-associated factors and the complexity of their determinants aids construction of early preventive and intervention actions. 5,12

CONCLUSION
Being 80 years of age or older was a predictor for conditions of pre-frailty and frailty, while dependency in basic activities of daily living and poor physical performance were predictive of frailty. An increase of one unit in the score for advanced activities of daily living decreased the rate of occurrence of the condition of frailty among these elderly people by 15%.