The current study provides for the first time the age group- and sex-stratified values of HGS and cutoff points for muscle weakness based on a nationally representative sample of the community-dwelling Brazilian population aged 50 years and over. We found that HGS decreased with age in both sexes and men had higher HGS values than women in all age groups. Furthermore, the sociodemographic and anthropometric variables associated with muscle weakness slightly differed between the sexes. Among men, muscle weakness was positively associated with low schooling (0–4 years), and negatively associated with overweight and obesity. Among women, muscle weakness was positively associated with low schooling (0–4 years) and low and middle monthly household income per capita and negatively associated with obesity.
Considering the increase in life expectancy and the progressive growth of the older Brazilian population, it is expected that an expressive number of people will present a loss of muscle strength in the coming years. In this sense, our study contributes to the current body of literature providing up-to-date and reliable reference values of HGS. Furthermore, as muscle weakness defined by low HGS is used for the diagnosis of sarcopenia  and physical frailty , our results may be directly or indirectly useful for the research and management of these two geriatric syndromes among older adults in Brazil. Evidence shows that HGS mediates the relationship between muscle mass and frailty  and is a more useful single marker of frailty than chronological age alone .
As expected, our results showed that advancing age negatively affects HGS, which is in line with previous reports from different countries [7, 23,24,25,26]. Potential underlying mechanisms for this finding have been proposed in the literature. The aging process itself is associated with functional and structural modifications in multiple physiological systems, including musculoskeletal, nervous, vascular, and endocrine. These modifications lead to muscle strength decline due to reduction in muscle mass and impairment in muscle activation pattern (e.g. activation of agonist muscles decreases and coactivation of antagonist muscles increases) , as well as affect hand structures, such as joints, tendons, nerves, receptors, and blood supply [24, 26]. In addition, age-related chronic low-grade inflammation, characterized by increased levels of circulating cytokines (e.g. interleukin-6 and tumor necrosis factor-alpha), plays an important role in the loss of muscle strength that accompanies aging  by activating or blocking signaling pathways related to proteolysis and protein synthesis.
Accurate comparisons of our results with data from other five studies with nationally representative samples are hampered by methodological heterogeneity between studies [7,8,9,10,11]. In general, these differences refer to the number of measurements, hand examined (right and/or left, dominant and/or non-dominant), and use of the highest or mean value of the measurements. In addition, two studies considered height in addition to the age group in determining the HGS reference values for men and women [8, 9].
In the Brazilian scenario, five studies provided reference values of HGS for older adults [12, 13, 29,30,31]. No study used a nationally representative sample and four of these studies were carried out with individuals from only one municipality [13, 29,30,31]. Again, great difficulty was observed in comparing our results with those described in these studies because of methodological differences. Some authors presented the HGS values for each sex stratifying the age group in decades [13, 29] and other authors in 5-year intervals different from the present study . There were also different categorizations of the oldest age group: ≥ 70 ; ≥ 76 ; ≥ 80 years . It is also worth noting that all Brazilian studies differed in terms of how to obtain the HGS variable considering factors such as the number of measurements, hand examined, and use of the highest or mean value of the measurements. These methodological differences reinforce the urgent need for a consensus for standardization on how to measure HGS in order to improve data comparison in the future.
In recent years, a growing number of studies have determined cutoff points for muscle weakness (i.e. low HGS) for older populations [7, 13, 32,33,34]. For instance, Lera et al. established cutoff points for muscle weakness in a sample of 5250 Chileans aged 60 years and over based on the 25th percentile (≤ 27 kg for men and ≤ 15 kg for women) . In a study conducted in 1317 participants aged 60 years and over from China, Dong et al. determined cutoff points for muscle weakness associated with mobility limitation (walking speed < 0.8 m/s), with HGS values < 32.45 kg and < 18.20 kg for men and women, respectively . Another different method for determining HGS cutoff values is calculating the mean minus 2.5 standard deviations of young healthy population . Similarly to our study, Ramírez-Vélez et al., using nationally representative data from the Survey on Health, Well-Being and Aging in Latin America and the Caribbean (SABE), defined cutoff points for muscle weakness specific for age group and sex for Colombian older adults aged 60 years and over . All cutoff points in the present study were higher when compared to those described for the older Colombian population. Potential reasons for this divergence between studies include heterogeneous methods for calculating the cutoff points and differences in participants’ genetic and clinical characteristics.
Our study demonstrated that low schooling was significantly associated with increased odds of muscle weakness in both sexes. Such association was also observed in previous national and international studies [35,36,37]. It is well known that individuals with higher educational attainment have more access to health services and health knowledge, adopt healthier lifestyle habits in terms of nutrition and physical activity, experience lower rates of unemployment, and earn a higher income, all of which might affect the overall health and also muscle strength . In this sense, it is recommended that education and adequate information on healthy eating and physical exercise should be offered to older adults with a low educational level in order to prevent, postpone, or reverse muscle weakness.
This study also identified monthly household income per capita as a determinant of muscle weakness among older women in Brazil. Women with a low and middle income had 1.78 and 1.32-fold increased odds of muscle weakness, respectively, when compared to women with high income. A possible explanation for the observed association is that income affects individuals’ health outcomes in several ways, including by means of the availability of material resources and health services . For example, limited financial resources might reduce access to adequate diet and nutrients (e.g. protein intake) and rehabilitation services, especially among the poorest individuals. By contrast, a previous large-population survey conducted in Korean participants aged 65 years and over found no significant association between income and muscle weakness in both sexes . The authors argued that the income may not fully represent resources available at old age, mainly after retirement, and suggested that wealth could be a better measure of financial resources . Similarly, research conducted in 27,351 participants aged 50 years and over from the Survey of Health, Ageing and Retirement in Europe (SHARE), involving 11 countries, showed that wealth has a greater impact on HGS than income . Thus, further studies to examine the relationship between income or wealth and muscle weakness, particularly in developing countries, are warranted to draw conclusions that are more robust on this issue.
Another remarkable result of the present study is related to BMI. Overweight and obesity had a protective effect for muscle weakness when compared to BMI’s normal range (eutrophy). This observation is in agreement with a recent cross-sectional study conducted in 3342 participants aged 60 years and over from the Irish Longitudinal Study on Ageing (TILDA) that found an inverse relationship of probable sarcopenia (defined as HGS < 27 kg for men and HGS < 16 kg for women) with both overweight and obesity . Recently, researchers also observed that when obesity accompanies probable sarcopenia, it might be less detrimental in terms of frailty and physical performance, compared to probable sarcopenia without increased adiposity . In addition, the literature reports a U-shaped relationship between BMI and all-cause mortality in older adults, with the BMI ranges representing overweight and class I obesity being advantageous for survival . According to Turusheva et al. , obese individuals have greater muscle mass and more type IIb muscle fibers (fast-twitch, glycolytic), which can lead to higher HGS. On the other hand, a Finnish longitudinal study demonstrated that being overweight or obese at baseline predicted greater HGS decline over a 22-year follow-up . The researchers concluded that a lengthened duration of obesity can lead to muscle strength decline by means of inflammation and insulin resistance, which have catabolic effects on muscles . Other studies found no significant association between HGS and BMI in older men and women in the multivariate analyses [24, 25]. Therefore, the relationship between these variables calls for further examination.
Unlike several previous studies conducted with healthy cohorts, we chose to keep individuals with chronic diseases in our analyses. It is noteworthy that identifying a healthy cohort (i.e. separating pathological from physiological age-related changes) is a hard task  and removing all participants with chronic diseases could generate a small and highly atypical sample . In addition, the exclusion of participants with chronic diseases could reduce statistical power and affect the representativeness of our sample, resulting in a significant bias in the results.
The main strength of this work includes data analysis of a large sample derived from a nationwide-based study with rigorous sampling plans, data-collection procedures, and quality-control practices, which enhances the generalizability of our findings and strengthens the statistical reliability of the results. Another advantage refers to the use of a dynamometer, which made it possible to objectively and reliably evaluate muscle strength. It is also worth mentioning that, in the present study, we chose to consider the age group in determining the cutoff points for muscle weakness for both sexes, due to the widely known impact of age on HGS. The adoption of a single cutoff point for each sex, regardless of age, could lead to misleading findings, as a small number of participants in the younger age groups would be classified as having muscle weakness and a high proportion of participants in the oldest age groups would be classified as weak .
On the other hand, our findings should be interpreted in light of some limitations. First, our HGS data are cross-sectional and are likely to underestimate individual decline. Thus, curves and percentiles generated should not be used to monitor the trajectory of an individual’s muscle strength over time. There is also a possibility of our findings being influenced by differences in the birth cohort. Second, the exclusion of a higher proportion of older participants tends to underestimate the force of associations. Third, the sample of participants aged 85 and over was small, which prevented stratification and assessment of HGS among nonagenarians. Fourth, the associations of muscle weakness with important anthropometric variables, such as upper arm circumference, arm length, calf circumference, and hand size were not tested, because these variables were not collected in the ELSI-Brazil study. Finally, the method used to determine the cutoff points for muscle weakness was based on the 20th percentile of HGS according to sex and age group. The optimum methods to define cutoff points for muscle weakness include anchoring these cutoff points to any meaningful outcome (e.g., mobility limitation or disability) or determining mean minus 2.5 standard deviations of young and healthy individuals of a given population. Thus, additional research should be carried out to determine these cutoff points for the older Brazilian population using the aforementioned methodological approaches.