Discussion
Knowing the factors associated with lower gait speed in quartiles allows us to propose actions that target each modifiable risk criterion in ageing, and reference values from healthy individuals are important for comparison to other samples and populations with different characteristics and limitations. This is the first study in a developing country with special focus on the social determinants of health showing that poor socioeconomic conditions, together with modifiable factors, play an important role in gait speed. Being older, illiterate, having difficulty in one or more instrumental activities of daily living, the presence of cardiovascular disease and being sedentary are independent factors associated with lower walking speed among the elderly.
Our results were similar to those found worldwide [
4,
32]; and nationally [
30,
31] they indicate that gait speed decreased with older age. Our older adults, however, were significantly slower than foreign populations [
34,
35] and their gait speed was slower than the overall fast gait speed of participants who were 70 and older with mobility limitations living in community [
36]. Bohannon (2011) found that for healthy women and men aged 70–79, the usual gait speed was 1.13 m/s and 1.26 m/s, respectively, and for those aged 80–89, the values were 0.94 and 0.97 m/s respectively [
36], both higher values compared to our result. The gait speed of older adults in our study were similar to the elderly aged 80–89 living in community in Dublin(Ireland), 30 percent of whom needed more assistance to walk and longer TUG, 14.2 s (versus 5.6) compared to the Brazilian sample showed in this study [
37].
Older adults aged 75 or over observed in our study were 3.56 (OR) more likely to be slower compared to younger subjects, and causal factors have been widely cited in the literature such as the loss of alpha motor neurons after the seventh decade [
34], the loss of type II fibres [
35] and muscle mass, with more rapid decline after age 65 [
36,
37] and the interposition of fat in muscle decreases muscle contraction [
10].
Interestingly, comparing our results with those obtained from another sample of Brazilian elderly subjects –
FIBRA network study (Frailty among Brazilian Older Adults), our average gait speed value was slower than the average of (1.11 m/s) that study. The
FIBRA study included subjects from different Brazilian cities with different Human Development Indexes, at an increased average age of 71.4 [
34,
35,
37,
38]. Besides that, the percentage of illiterate older adults was smaller and the methodology was different from our research. Unlike the
FIBRA study, which used a convenience sample, ours used a weigthing sample in which a weight is attributed to each individual, which indeed makes it a representative sample of the city of Sao Paulo. Other studies included volunteer subjects [
33,
32] or only women [
34].
Maybe these differences can explain why independent associations between gait speed and educational level or income were observed only in our study. Another relevant point: our data indicate a population of elderly patients living in São Paulo with chronic diseases that affect mobility and result in low physical activity [
38]. According to international literature, our older adults are deemed as frail and [
39] at high risk of poor outcome [
10] and poor survival [
8].
Despite gait assessment being a quick, safe, inexpensive and highly reliable measure, methodology can vary widely making it difficult to compare studies. Similar methods can differ in walking length, and different populations weaken the comparisons.
Although in Studenski’s research (2011) the vast majority of the sample comprised white men and women, similar characteristics observed in our study, the mean gait speed in older adults was 0.92 m/s [
8], a higher value compared to our study despite 45% of all participants being older than 85. Watson [
6], analyzing data from well-functioning sub-cohort (USA), found mean gait speed of 1.20 m/s and the mean gait speed of the first quartile slower than 1.05 m/s; a higher value comparing to our results although mean age of sample was higher (75.2 years) Although the subjects in Watson’s studies were older, and 68.2% of them were sedentary, they were faster. A factor that may have influenced his study more favourably is that his sample comprised greater numbers of men and black individuals; moreover, those subjects had more education than the ones in our sample. However, similarly to our results, the participants in the lower quartile of gait speed were more likely to be older, sedentary, have less education and have more chronic health conditions.
Our results revealed that gait speed increases with the highest level of education (OR 3.20), similar to that found in Brunner’s study (2009) [
40]. Years of schooling are used as a proxy for social status and, thus, health condition. Most of the elderly individuals in Brazil live in poor conditions, particularly in São Paulo, a city with great economic and social contrasts (almost 40% of the elderly subjects were illiterate). Although education alone does not ensure the end of social discrimination, it is part of the formation of a more egalitarian society and a critical factor in reducing socioeconomic disparities. Despite the positive correlation between education and income, education is considered a major factor in overcoming income inequality. Educational level is a protective factor and prevents poor outcomes in health. Individuals with more education are more likely to obtain financial resources, seek medical advice and detect diseases earlier; therefore, they have better self-reported health, get better health treatment and better understand the importance of prevention, such as doing physical activity and, thereby, decrease their chance of comorbidity. It is a fact that prevalence of chronic disease may also be influenced by an individual’s access to health services, by their socioeconomic condition, and self-reported health status [
41]. Self-reported health is recognized internationally as an indicator of health status, and may justify a positive association with gait speed in bivariate analyses.
Although a positive correlation between handgrip and gait speed was observed, the handgrip association did not remain significant to the first quartile of gait speed in the final model, probably due to the fact that the subjects were already very committed in walking speed.
Regarding the TUG, our study found a mean of 12.9 s, a higher value than the results found in an meta-analysis (9.4 s) [
39]. Although we did not explore the mechanisms underlying the changes observed in this study, the negative correlation found between TUG and gait speed shows the importance of evaluation of gait and balance. Gait is dependent on postural control and the integration of various systems, such as proprioceptive, visual and vestibular, their sensory input, integration in the Central Nervous System (CNS) and, depending on effective motor response [
39,
40] and gait assessment, it is a form of prevention against disability and motor decline [
16].
Another population-based study revealed that each increment of one standard deviation in the usual gait speed was associated with a reduced likelihood of disability from 26 to 44% [
41]. Similarly, our current findings revealed 33.7% of subjects disabled in one or more instrumental activities with 2.74(OR) to be on the first quartile of gait speed.
The results presented here show in bivariate analysis that the presence of cognitive impairment was significantly associated with gait speed. The same results were found in other studies, reinforcing the notion that cognition influences gait speed [
6,
42,
43]. The significant association between gait and cognition maybe can be explained by the influence of cognitive aspects and mood on the maintenance of functional capacity, and by the need of physical and intellectual integrity to remain autonomous and independent. Although it is discussed whether cognitive decline is a predisposing or precipitating factor in the decline of gait speed [
44], our data seems to indicate that the decline in physical function is secondary to cognition. Perhaps most of the older adults in São Paulo cannot afford cognitive rehabilitation services.
Regarding depression, the present results show that depression levels have a positive correlation to gait speed, which agrees with what Mossey and colleagues presented (2000) [
45]. Adopting a healthier lifestyle is an important part of treating depression, e.g. doing physical activities on a regular basis. Previously published systematic reviews and meta-analysis concluded that exercise reduces depressive symptoms among patients with a chronic disease [
46,
47]. Research has also shown that depressed patients are less fit and have diminished physical work capacity [
48], which in turn may contribute to other physical health problems. Depression in Brazil is underdiagnosed, probably because its diagnosis is often hampered by the presence of comorbidities, the difficulty of the healthcare teams to recognize it and the lack of mental health care in the primary health care system. Studies show that between 50 - 60% of the cases of depression are not detected or adequately treated [
49]. Furthermore, depending on the intensity of the depressive symptoms, it becomes impossible to motivate the subject to do physical activity.
In the final model, those who considered being active showed to be significantly associated with higher walking speed. One of the important ways to prevent the insidious loss of bone and muscle strength is to stay active. When an individual loses muscle strength, walking becomes less frequent and slower, as one becomes physically unconditioned. Consequently, the individual becomes more sensitive to fatigue and, thereby, increases inactivity. Once this vicious cycle is triggered, it ends up compromising initially instrumental activities and, subsequently, the basic ones if nothing is done to halt the cycle. Physical inactivity is an important risk to cardiovascular disease. It was shown, however, that it is preventable, up to 80%, by eliminating shared risk factors such as physical inactivity [
50]. Our data revealed an important association between cardiovascular diseases with the lowest quartile of gait speed in the final model. This reinforces the fact that the usual speed of gait is related to aerobic capacity showing association with functional reserve [
51]. Walking imposes demands on the nervous, cardiovascular, pulmonary, musculoskeletal and hematologic systems, as they require more oxygen to contract the muscles. These systems work synergistically – if one of them does not work well, it can impair gait speed [
8].
Other studies showed that regular exercise significantly improved physical fitness (aerobic capacity), walking capacity and cardiovascular dimensions [
52].
Considering the relevance of this problem, Matsudo and colleagues interviewed 2001 individuals aged 14–77 in 29 cities within the state of Sao Paulo and showed that the levels of physical activity did not differ among age ranges. There were similarities between the genders, but people from metropolitan regions and the poorer ones were less active [
53]. Probably the modern world with electronic novelties encourages a sedentary life style.
Another study conducted in Santos (a beach city in Brazil) recruited healthy elders of both genders as volunteers, who also led a sedentary life style. Their gait speed was 1.34 m/s among men, and 1.27 m/s among women. The values of gait speed found were significantly lower than those foreign benchmarks (p < 0.05) [
32,
54] but higher than our findings in São Paulo, probably because of different habits and socioeconomic backgrounds. It is important to mention that although our data has been adjusted for height, the average height of the Brazilian elderly population is shorter compared to populations of a similar age range from developed countries.
Regarding cerebrovascular disease and the lack of association with lower walking speed in the final model, it could be explained by the exclusion of all individuals with motor sequels, which could have influenced gait measures. COPD is a systemic disease that affects beyond the respiratory, cardiovascular and muscular systems. Among the muscle changes are loss of muscle mass, loss of efficiency to carry out protein synthesis, decrease in type I fibres [
55,
56]. There are several factors that can cause these changes in the muscular system such as chronic hypoxemia, prolonged usage of high doses of corticosteroids, nutritional changes, the response to systemic inflammation and even physical deconditioning [
56]. The inability to exercise and the ventilatory limitations increase deconditioning, which ends up compromising their functionality [
55]. These factors may explain the significant association between lower gait and COPD found in the bivariate analyses. However, this association was not found in the multivariate analyses. Although 54.4% of the elderly subjects reported two or more chronic diseases, 50.9% related their health as very good, which reinforces that health is no longer measured by the presence or absence of diseases, but by the degree of preservation of one’s functional capacity and independence. Such result was also evidenced by the high prevalence of elderly people without disability in basic and instrumental activities. What is at stake in old age is autonomy, the ability of the elderly to remain socially integrated and, for all purposes, health [
57]. In our study, 46% of the elderly aged 75 years or over were in the first quartile, which could be related to the prevalence of more chronic diseases and worse handgrip strength; they had lower educational levels and were inactive, similar to other studies [
6,
9].
Study limitations
Despite the advantages of quickness and costs of the cross-sectional study, it presents limitations since it does not allow one to identify causality, whether the factors identified as associated to lower gait speed came before or after it, since expositions and outcomes are collected at the same moment. This may also explain the lack of significance between race and lower gait speed found in this study. In addition, the presence of chronic diseases, health condition, disability and physical activity were assessed by means of self -reporting, which may result in over or under-estimation of prevalence. However, participants report only those conditions diagnosed by a physician. As to physical activity, IPAQ was validated in a sample of the Brazilian population. Older adults who used assistive devices to walk, or those with severe neurological conditions, were excluded, which might limit the external validity of the study. Our current results show that poor socioeconomic conditions present in developing countries influence lower walking speed such as education, and they may be particularly related to some modifiable factors such as impairment of IADL, physical inactivity and cardiovascular disease.
Authors’ contributions
TAB and YAD worked on the conception and design, data analysis and interpretation and drafting of this article, performed a critical review of the manuscript and approved the final version to be published. DPN worked on data analysis and interpretation, performed a critical review of the manuscript and approved the final version to be published. MSN and ASR performed a critical review of the manuscript and approved the final version to be published. MLL and EAJ worked on the interpretation of the data, performed a critical review of the manuscript and approved the final version to be published.