Falls among the non-institutionalized elderly in northern Minas Gerais , Brazil : prevalence and associated factors

1 Universidade Estadual de Montes Claros, Faculdade de Medicina, Departamento de Saúde Coletiva. Montes Claros, Minas Gerais, Brasil. 2 Universidade Estadual de Montes Claros, Faculdade de Medicina, Departamento de Clínica Médica. Montes Claros, Minas Gerais, Brasil. 3 Universidade Estadual de Montes Claros, Faculdade de Medicina, Curso de Medicina. Montes Claros, Minas Gerais, Brasil. 4 Universidade Estadual de Montes Claros, Programa de Pós-Graduação em Ciências da Saúde. Montes Claros , Minas Gerais, Brasil.


INTRODUCTION
In recent years, the Brazilian elderly population has increased nearly twice as quickly as the general population, and is the fastest growing population segment in the country.Official records reveal that there are around 20 million people aged over 60 in Brazil at the moment, representing at least 10% of the country's population. 1 This accelerated population aging has resulted in new needs and care provisions. 2 One aspect that stands out in the aging process is that related to physical disabilities and the degree of dependence of the elderly.Although functional disability is not inherent in the aging process, 3 as individuals age, the chance of suffering injuries caused by accidents increases. 4[7][8] Falls in elderly people may indicate the beginning of frailty or acute disease.Besides the previously mentioned risk of death, falls result in constant injuries and fractures, compromise activities of daily living, increase rates of institutionalization, generate a decline in general health and increase fear of falling among the elderly, which, in turn, increases the risk of subsequent falls. 9Falls produce deleterious consequences not only for the elderly but also increase the cost of hospitalization and other health services, and as such represent a major public health problem. 10equate knowledge about the occurrence of falls in the elderly and devoting the necessary attention to the problem can guide more appropriate and effective preventive measures in health facilities.[16] However, there are still variables that have been little explored in Brazilian literature, such as frailty 20 and other comorbidities.There are also no studies on the subject for the northern region of Minas Gerais.This region is characterized as a transition area between the southeast and the northeast of the country, with social indicators that demonstrate major socio-economic needs.Local studies may be representative for other regions of the country.The present study aimed to assess the prevalence of the occurrence of falls and associated factors among the elderly, investigating as yet unexplored variables in a population-based sample of non-institutionalized elderly persons in a regional hub city in the north of Minas Gerais.

METHODS
A cross-sectional and analytical study of noninstitutionalized elderly living in the city of Montes Claros, northern Minas Gerais, was conducted.The city has a population of approximately 400,000 inhabitants and is the main regional urban hub.
Sampling was probabilistic by cluster and performed in two stages.In the first stage, the census tract was used as the sampling unit, with 42 of the 362 urban census tracts in the city randomly selected.In the second stage, the number of households was defined, according to the population density of individuals aged 60 or over.At this stage, more households were allocated in the sectors with the highest number of elderly persons, in order to produce a more representative sample.
The sample calculation considered a total population of 30,790 elderly persons (according to IBGE data), a margin of error of 5%, a confidence level of 95% and a prevalence of 28.5%, as this is the median observed in a systematic review study. 10As cluster sampling was employed, the identified number was multiplied by a correction factor (deff) of two and 10% was added for possible losses.Thus, the minimum number of persons in the sample size calculation of the study was 682.
Elderly persons who were considered by their caregivers/family members to be unable to answer the questions or those who refused to participate in the study were excluded.Elderly persons who were unable to participate in at least three visits on different days and times, even with prior appointments, were considered losses.
Data collection was carried out in the homes of the elderly persons between May and July 2013.Interviewers, who were graduates in nursing, were trained and a pilot study was developed before the collection of data for analysis.The census tracts were visited clockwise from a predetermined point.The data collection instrument was based on similar population-based studies.The pilot study data, from a specially drawn census tract, was not included in the final analysis.
The independent variables studied were: gender (male vs. female), age (younger than 80 years x 80 years or older), self-reported skin color, marital status (married or civil union x without partner), family income (greater than one minimum salary x equal to or lower than one minimum salary), self-perceived health (positive x negative), frailty (defined by the Edmonton Frail Scale, which evaluates nine domains with a score from zero to 17 and defines frailty as a score greater than six), 21 fear of falling (no x yes), presence of chronic comorbidities (hypertension, diabetes mellitus, osteoporosis, arthritis/arthrosis/rheumatism, heart disease, all self-reported -no x yes), polypharmacy (defined as the continuous use of five or more drugs -no x yes), hospitalization in the last year (no x yes), depressive symptoms (defined by the reduced version of the Yesevage Geriatric Depression Scale -GDS 15, wherein a score equal to or greater than six identifies depressive symptoms), 22 functional mobility through the Timed Up and Go test (where a time equal to or greater than 20 seconds is suggestive of more difficult mobility for the elderly) 23 and smoking (no x yes).All the variables were dichotomized for data analysis.
The prevalence of falls in the previous 12 months was taken as the outcome variable.Initially, bivariate associations were established between the record of falls and the variables using the chisquared test.In a second step, variables that were associated to a level of 20% ( p<0.20) were jointly evaluated by logistic regression.At this stage, the odds ratio was calculated (OR) with respective confidence intervals of 95%, to evaluate the magnitude of associations, assuming a significance level of 5% ( p<0.05).Analysis was performed using the Statistical Package for the Social Sciences (SPSS), version 17.0 (SPSS for Windows, Chicago, USA) program.
All the participants were fully informed about the research and gave their consent by signing a Free and Informed Consent Form or by a digital fingerprint record, when unable to sign.The research project was approved by the Research Ethics Committee of the Universidade Estadual de Montes Claros (Montes Claros State University) (Process nº 173397).

RESULTS
The study included 683 predominantly female elderly persons, aged between 60 and 98 years, with a mean of 70.9 years (SD±8.08).Most described their skin color as brown (57.1%).There was a predominance of married elderly persons (48.6%).Among the respondents, most reported having up to four years of study, and 26.1% had no schooling.These and other features of the group are presented in Table 1.
continues on next page In relation to health and healthcare, described in Table 2, the majority of respondents said the public system was the most used type of service (68.8%).Health was perceived as good or very good was cited by 291 people (42.6%).About a quarter of respondents (23.6%) reported needing to use five or more drugs of continuously.During the approach, 243 elderly persons (35.6%) were classified with some degree of frailty by the Edmonton Scale and 200 (29.3%) had scores suggestive of depression symptoms according to the Geriatric Depression Scale.For this sample, the prevalence of falls in the previous 12 months prior to the survey was 28.3%, with most individuals suffering a single fall in this period.The main reported cause was slipping/ tripping and most fallers reported needing help to get up (Table 3).Table 4 presents the results of the bivariate analysis between falls and the main variables.For space limitations, not all the associations investigated are shown.Continuation of Table 4 Table 5 shows the variables in the final model, after multivariate analysis.Female gender, a negative self-perception of health, functional mobility impairment (Timed Up and Go test ≥20 seconds), record of hospitalization in the previous 12 months and frailty according to the Edmonton Frail Scale were associated with the occurrence of falls in the previous 12 months.

DISCUSSION
The prevalence of falls in the previous 12 months reported by the elderly persons in the present study was very close to that revealed by a Brazilian study which sampled 100 municipalities in 23 states. 11Other Brazilian studies have reported a higher prevalence of falls among the elderly, although these dealt with institutionalized populations. 13,14In a review of the topic, the authors observed studies where the prevalence ranged from 15.9% to 56.3%, although it should be noted that this broad range included samples limited to the institutionalized population and larger studies with a population-based analysis. 10ong the sociodemographic variables, only the female gender was associated with falls.6,18 There are no specific explanations for this finding, but other authors have highlighted characteristics specific to women such as reduced muscle strength and muscle mass and a higher prevalence of degenerative chronic diseases. 11,24It is worth noting that one study found that the female gender is also associated with recurrent falls. 17 for the variable of hospitalization in the year preceding the survey, which was also present in the final model, the risk of falls was greater among elderly persons who had been hospitalized at least once.It was not possible in the present study to verify whether these hospitalizations were directly related to falls or not.However, it can be inferred that this apparent association may be either because a fall is a cause for hospitalization, or as a serious indicator of morbidity and frailty.In this case, immobility, the medications used or the underlying disease, when present, could predispose the individual to falls.Although the association between falls and hospitalizations is common, 24,25 the association between records of hospitalization in the previous year and the occurrence of falls was observed in one Brazilian study, which addressed the occurrence of falls in institutionalized elderly persons. 5This finding should be further investigated in future research.
A negative perception of health (classified here as "regular", "poor" or "very poor") was also revealed to be associated with a greater chance of falls.5][16] It is possible that the indicator is closely related to difficulties with balance. 14Selfrated health is considered an important health indicator, and is widely used in research and health surveys due to its easy applicability and low cost, compared to other more complex methods.It is a subjective health measure, but reflects the presence of diseases and their impact on the quality of life of individuals. 26In Brazil, the development of further research in this area is desirable, according to a recent review on the subject. 27e impairment of functional mobility as measured by the Timed Up and Go test was also associated with the occurrence of falls in the present study.This result has also been recorded in other studies. 14,28The test is simple to perform and can quickly identify disorders in balance and mobility.It is therefore an essential aspect of geriatric evaluation. 23It is imperative, therefore, that this test is routinely implemented in assessments of older people, in order to identify situations of risk and anticipate specific advice and care.
The greater incidence of falls in elder persons with some degree of frailty has been little investigated in Brazilian literature.Only one study conducted in São Paulo recorded a positive association between the variables, with the authors highlighting the possibility of a two-way relationship -in other words, that falls can make elderly persons frail and frailty can cause elderly persons to fall. 20Literature from outside Brazil shows this association more frequently, although there is no standardization of studies on how to assess frailty. 29,30In the present study, analysis of the studied population by the Edmonton Frail Scale confirmed the association between falls and frailty.This scale is a culturallyadapted instrument validated for the Portuguese language which evaluates nine domains (cognition, health, functional independence, social support, medication, nutrition, mood, urinary continence and functional performance), and is considered a robust instrument due to its multidimensional character. 21,31her variables identified in literature as associated with the occurrence of falls in the elderly were not identified in this study.Some studies investigated morbidities, for example, which revealed an association in bivariate analysis but did not remain in the final model.An association with osteoporosis, arthritis/arthrosis/rheumatism and falls was identified in another study. 17It is natural to assume that such conditions lead to uncomfortable situations, with increased sensitivity and stiffness, among other symptoms, contributing to decreased physical ability, affecting mobility and leading to falls.However, these authors did not evaluate functional mobility or frailty, aspects which may include the aforementioned morbidities, in joint multivariate analysis.
The identification of fear of falling as a variable associated with the occurrence of falls in some studies 9,32 was also not maintained in the final model of the present study.It should be noted, however, that such studies involved a small sample, a fact that may explain the difference observed.Still, we must emphasize that fear of falling is a common profile among the elderly and can make older people feel less confident in performing their daily activities, increasing immobility and postural instability. 33e present study did not find a relationship between the presence of depressive symptoms/ depression and the occurrence of falls, unlike other Brazilian 13,17 and non-Brazilian 34,35 studies.This is an important relationship and deserves greater attention from researchers and practitioners.There is not always enough information in studies about the prior exclusion of elderly patients using antidepressant medication, an aspect that can affect comparisons between studies.
This study has some significant limitations, such as the fact that the data was obtained from the responses of respondents.Moreover, the consequences of the falls were not included, which could further contribute to defining the impact on health of the elderly in the region, justifying more emphatically the need for intervention and more effective prevention measures.Environmental aspects related to falls, such as the characteristics of the site of the fall or the circumstances associated with the event, were not investigated.Also, although there were few losses and refusals relating to participation, the fact that some elderly persons were not at home restricts the more general interpretation of the data.However, it should be emphasized that the study featured a large sample, allowing knowledge to be obtained of a health issue for the elderly that had not been studied in the region.

CONCLUSION
The study found that falls are a common event among elderly persons living in the community in the north of Minas Gerais.Knowledge of associated variables can contribute to the implementation of more effective preventive measures, as such variables are at least partly subject to intervention.Changes in these factors may decrease the risk of falling and its consequences.To achieve this, comprehensive and multidisciplinary care of the elderly is required.

Table 2 .
Characteristics related to health care and health conditions among elderly persons in Montes Claros, Minas Gerais, 2013.

Table 3 .
Characteristics linked to falls and postural instability of elderly persons in Montes Claros, Minas Gerais, 2013*.
* Refers only to number of elderly persons who reported falling in last 12 months (n=194)

Table 4 .
Bivariate analysis for variables associated with falls in previous 12 months among elderly persons in Montes Claros, Minas Gerais, 2013.
* The categories regular/poor/very poor were considered as a negative self-perception of health.** Incomplete information was excluded (patients that did not know about disease)

Table 5 .
Variables associated with the occurrence of falls among elderly persons in Montes Claros, Minas Gerais, 2013.