Do Correlates of Handgrip Strength Among the Elderly Ghanaian Population Matter? Empirical Evidence from WHO-SAGE Wave 1 Survey

Background: Recent empirical evidence suggests variations in handgrip strength in the elder years of life depending on multiple parameters, but such evidence is lacking in Ghana. The study determines what multiple factors are associated with handgrip strength among the older population in Ghana. Method: Using data from the World Health organisation Global Ageing and Adult Health Survey (SAGE) Wave 1, four thousand, ve hundred and ninety-six Ghanaians aged 50 and above years were selected through a multistage sampling technique in the country. Results: It was revealed that handgrip strength decreased with age and on the average was lower among females. Older females had a weaker handgrip strength than their male counterparts. Handgrip strength was higher among persons who were not suffering from hypertension or arthritis. Additionally, weight (Model 1, [left hand { (cid:0) = 0.95, p<0.01}, right hand { (cid:0) =0.18, p<0.01}]); Model 2, [left hand { (cid:0) =0.59, p<0.01}, right hand { (cid:0) =0.01, p<0.001}]) and height (Model 1, [left hand { (cid:0) = 0.04, p<0.01}, right hand { (cid:0) =0.45, p<0.01}]); Model 2, [left hand { (cid:0) =0.38, p<0.01}, right hand { (cid:0) =0.03, p<0.001}]) were associated with increased grip strength. Conclusions: Older adults’ physical deterioration is inevitable because of structural and functional musculoskeletal limitations due to ageing. Hence, a timely identi ﬁ cation of those at risk for appropriate intervention to promote their healthy living is necessary.


Introduction
Handgrip strength (HGS) is used as a benchmark for overall body muscle function and a proxy measure of physical health, particularly in older persons (1). It is a simple, non-invasive, reliable, and cost-effective screening technique for assessing muscle functioning (2). The measurement of HGS is also useful as a predictor of health status (3,4), muscular strength, nutritional status, and disability (5). Besides, HGS has been used as a risk stratifying approach for causes of death and recovery time after illness or surgery (6).
Recent empirical evidence suggests variations in HGS of older people depending on multiple parameters such as sociodemographic factors (e.g., age, educational level, income, marital status, sex), anthropometric measurements (height, weight, body mass index (BMI) and muscle circumference), adverse health outcomes (e.g., depression, chronic diseases [e.g., diabetes, stroke and heart diseases]), and negative behavioural indices (e.g., physical inactivity, smoking, alcohol consumption, (7)(8)(9). For example, it has been established that men, generally, have higher HGS than women at all ages, with HGS higher in the right than the left hand in both sexes (10). Furthermore, HGS increases from childhood through adolescence, peaks at age 35 ± 40 years, and declines afterwards. Similarly, poor HGS is predictive of increased mortality from cardiovascular disease and cancer in men after controlling for muscle mass and body mass index (BMI) in some longitudinal studies (11,12). There are also associations between HGS and chronic diseases as well as multi-morbidity in men and women, even when some confounding factors have been adjusted (13). Mostly, HGS is associated with disability in daily lives, walking speed and quality of life (14,15).
Ageing is associated with muscle mass loss which can lead to a decrease in muscle strength (7). Decrease in strength because of ageing is mostly secondary to deterioration in skeletal muscle mass in men and women (16). Evidence suggests a direct relationship between decreasing muscle strength and age-related health complications, including degenerative diseases (3,17,18). Furthermore, maintaining muscle strength, by regular involvement in physical activities, has been regarded as very essential in the reduction in functional limitations of older people (6). Other studies have identi ed an association between depression and muscle strength (12,19,20).
Absolute estimates of the older population showed a sevenfold increase (i.e., from 215,258 in 1960 to 1,643,978 in 2010) in this population group in Ghana (21). Despite the ongoing ageing transition and evidence of the importance of HGS, public health professionals in the country focus more on maternal and child care (22). The health status of the aged can be assessed through a variety of indicators, including HGS, which is cost-effective and requires fewer efforts. Again, as an indicator, HGS supports early detection of diseases; malnutrition; functional disability and frailty, health challenges prevailing in the older years of life. However, to the best of our knowledge, information on HGS and its associated factors in Ghana is relatively lacking. The present study aimed to address this knowledge gap by examining the correlates of grip strength in the elderly population in Ghana. An empirical study on HGS and its associated factors is essential to enhance the health of the aged population and in the planning and implementation of public health initiatives and programs that aim to preserve the muscle strength of older people in the country.

Data
The data used for the study were obtained from the 2007-2008 World Health Organisation (WHO) Global Ageing and Adult Health, (SAGE) Wave 1 for Ghana. SAGE is a longitudinal study which collects data on adults aged 50 years and older, together with a comparable sample of adults aged 18-49 years, from nationally representative samples in six countries, Ghana inclusive. The survey was performed according to legal guidelines for carrying out population-based research in the country. SAGE Wave 1 recruited adults 18 years ad above using a multistage cluster sampling technique.
The households became the clusters within which persons 18 years and above were selected. Together, a total of 5,110 respondents were involved in the survey. A detailed description of the SAGE survey, including data and its collection procedures, are provided on the project website.

Variables
The outcome variable for the study was handgrip strength for the right and left hands among the elderly, de ned as the proportion of people aged 50 years and above who were either right-handed, left-handed or ambidextrous in Ghana. Eighteen variables were selected as predictor variables. These were sociodemographic (i.e., age, sex, marital status, and level of education), anthropometric (i.e., weight, height, hand dominance), economic (i.e., employment, wealth quintile), and health and behavioural variables (i.e., hypertension, diabetes, stroke, arthritis, fatigue, depression, physically active, tobacco use, and alcohol consumption). The selection of these variables was per previous studies and achievement of numerically stable and adaptable models (23). While age, height, weight, handgrip (left and right), were numeric variables, sex, employment, hypertension, diabetes, arthritis, tobacco use, alcohol consumption, and employment were dichotomous variables. Wealth quintile, hand dominance, fatigue, level of education and marital status were polychotomous.
Further, age was transformed into a categorical variable to re ect the signi cant stages of ageing and health (24). Marital status was assigned a different code with married as the initial status and never married as the nal status. Respondents who were age 50 and above years and those who had a record of grip strength became the respondents for this study. Therefore the sample size of the study became 4596. Body mass index (BMI) was estimated using weight and height (in metres). As a measure of body fat, it was introduced to assess the in uence of body-fat on health in the older years of life (25). Hence, this newly generated variable was added to those listed above. Missing values for the variables were imputed using the predictive mean matching multiple imputation approach. It helped to reduce the effects of missing data in all computations (26).

Data Analyses
The variables of interest were summarised to provide a clearer understanding of the distribution among males and females. The distribution was complemented with a test of variation in the outcome. Following this, a stepwise regression model was tted to choose the best predictor variables from the variables-a backward elimination method was used for this purpose (27). The variables which generated the parsimonious model were included in the multivariate analysis. Linear regression (ordinary least squares-OLS) models were tted for the analysis. Thus, multiple linear regression models were built to ascertain how each of the predictors contributed to the prediction of the mean value of the response variable (28). Three groups of linear models were tted for the study. The rst model used a standard linear regression. The second model included transformed variables in the form of logarithmic and interaction effects. The third model was a robust linear regression. Each of these models was repeated for the grip estimates of the left hand and right hand.
Model diagnostics were performed to explore problems that may compromise the regression analysis and determine whether certain assumptions appear reasonable (29). Autocorrelation was tested using the Durban-Watson test. Shapiro-Wilk test was used to assess the assumption of normality. The Breusch-Pagan test was used to assess the assumption of homogeneity of variance. Since the spread varied, the assumption of constant variance was violated. The presence of multicollinearity in the model was assessed using a variance in ation factor (VIF). The base for the VIF assessment was ve and above. The model diagnostics' outcomes that did not appear reasonable were resolved using the fourth approach, tting of robust models (30). These robust regressions were modelled for the standard linear regressions as well as the model with transformed variables. All analyses were weighted and performed using R programme. Statistical signi cance tests were based on a two-sided probability set at p < 0.05.

Results
Descriptive statistics on socio-demographic and anthropometric characteristics Data referent to all 4596 respondents is presented in Table 1. Generally, more males (N = 2308) than females (N = 2288) were included in the study. Majority of the respondents were aged 60-89 years (males = 50.4%, females = 54.6%) with a statistical difference between males and females. Many male respondents were married (83.0%), whereas many females were widowed (46.5%). While many males had secondary education (45.9%), quite a proportion of females had no primary education (38.3%). There were more right-handed respondents than others, although the right-hand female respondents (94.2%) were more than males (92.8%). Alcohol consumption and tobacco use were relatively lower among females (45.9% and 7.7%) than among males (69.5% and 42.6%), but no statistical difference between males and females was recorded for alcohol consumption. Many females and males reported not having diabetes, even though more females (4.4%) compared males (3.2%) had diabetes. More females (64.7%) than males (64.3%) reported no fatigue. There was, however, no statistical difference between both sexes. There were differences in the distribution of wealth quintile for males and females, but no statistically signi cant difference was observed between males and females.  A Stepwise regression analysis on the association between predictor variables and grip strength The relationship between grip strength and correlates such as age, sex, weight, BMI, diabetes and physically active were evaluated, and are presented in Table 3. There were differences in the relationship outcomes between these variables and grip strength of a hand type.   Findings also show an age-dependent decline in HGS for both genders across handedness (i.e., left and right hand) from the 6th) decade (i.e., 60+) and beyond, but the pattern of decline in HGS with advancing age was more pronounced (i.e., lower) among females. HGS was higher among persons who were not hypertensive and suffered arthritis. Among the three anthropometric measures, increased HGS was signi cantly associated with weight and height. These ndings are indications that HGS is in uenced by multifaceted factors which were con rmed through the regression analyses.
The aged have their bodies' manifest age-related degenerative variations in the musculoskeletal, vascular, and nervous systems regardless of gender. These degenerative changes usually cause the decline of hand function in older adults and subsequently in uence their hand structure such as joints, muscle, tendon, bone, nerve and receptors, blood supply, skin, and ngernails (31). Consistent with previous studies, the decrease of muscle strength across handedness because of ageing cannot be ruled out regarding the HGS, and ageing association noted in the current study. Two primary reasons account for HGS loss during ageing. First, there is a decrease in muscle mass because of functional loss of motor units (32), resulting in an incomplete re-innervation of abandoned muscle bres by surviving motor neurons (33) and selective atrophy of fast-twitch muscle bres in the hand (34). The second is the reduced capacity to adequately stimulate muscles in the hand during ageing (32). Other mechanisms include joint stiffness, reduction in muscle coordination, decreasing physical activity, decreasing hormone levels, and chronic diseases that come with advancing age (35)(36)(37). However, sex variations in HGS mechanisms underlying ageing are not well understood (38). According to Kim et al. (9); Lenardt et al. (38), HGS in women degenerates intensely from 55 ± 59 years onwards. These authors note that sex hormones are vital in explaining this sex-related variance. It has been proven that muscle strength is similar in men and premenopausal women, although there is a sudden deterioration in muscle strength from menopause thereafter (39). Additionally, sex differences in HGS may be explained by hereditary sex variations in muscle mass (40,41).
The nding that HGS was higher among persons free from hypertension or arthritis suggests that low grip strength is associated with increased susceptibility in persons who might develop cardiovascular disease (42) and some degenerative diseases (e.g., osteoarthritis). Loss of muscular strength might likely be part of the causal link to cardiovascular and other degenerative diseases. The current study did not, however, directly predict the inverse association between HGS and incident hypertension or arthritis and it possible that unmeasured parameters like nutritional status (43), endothelial dysfunction and autonomic imbalance (44,45), as well as arterial stiffness, might mediate the relationship between HGS and cardiovascular events (42,46). Future studies are required to provide a clearer understanding of whether increased HGS directly reduces the risk of incident cardiovascular disease and other degenerative diseases.
Similar to other research, signi cant associations were identi ed between increased HGS with height and weight in older adults in the current study (47)(48)(49). Regarding height, Stulp et al. (50) showed that taller people have absolute superior strength compared to shorter individuals. Absolute strength is associated with cross-sectional muscle area. It is related to one's body surface area or the square of an individual's body height and a factor that is more closely linked to lean mass (i.e., muscles, bone and non-fat tissues), a characteristic associated with muscle strength (51). According to Chandrasekaran et al. (52), persons with greater heights have longer arms, with greater arm leverage for force generation, resulting in an adequate amount of force. For weight, older people who have low body weight have low muscle mass and hence weaker physical strength, a biomarker of poor HGS. Lower muscle mass could also be associated with undernutrition or chronic disuse often connected to older age (6).

Practical implications
Ageing has a degenerative impact on hand function, including HGS. Therefore, HGS is a marker of the frailty phenotype in older adults (37). By entering the sample values of the predictor variables (i.e., age, sex, height, weight, BMI) into the regression models, speci c estimates of Ghanaian older adults' left and right grip strength were provided. These estimates might be useful for clinicians or researchers who use normative values within their working software to provide more accurate predictions of normative strength scores for speci c applications in different populations (53). The study ndings suggest that assessment of HGS in elderly Ghanaians who are free from cardiovascular illness (e.g., hypertension) and degenerative disease (e.g., arthritis) could also be a risk biomarker for the incidence of these diseases across the Ghanaian aged population and perhaps other similar geographical locations. Future studies through longitudinal designs are required to examine how older people with low HGS or muscle strength may be assisted to improve their structural and functional musculoskeletal representation. To promote healthy ageing, helping elderly Ghanaian people to maintain physical function at old age is vital. On the bases of current ndings, interventions to improve the HGS through regular physical activity (e.g., Activities of Daily Living, [ADL]), appropriate nutritional intake and other health care programmes for the aged are required (54). Sub-group level physical activity and elderly care programmes could prioritize women who were identi ed as the most affected with lower HGS compared to their male counterparts.

Strengths and limitations
To best of our knowledge, this is the rst population-based study to describe the correlates and pattern of HGS loss across age and gender in a large representative cohort of Ghanaian people. The large sample size recruited from the general population and the use of rigorous analysis through the regressions models for the measurement variables give generalizability and credibility to the results as well as ndings. These ndings could be used as baseline data for context-speci c interventions and more research. However, the cross-sectional age pro les of HGS across the sample may either under or overestimate individual-speci c declines.
Further research is required through longitudinal designs using multiple momentary ecological assessments to give more accurate predictions of the magnitude of grip strength loss with age. There are other factors (e.g., arm length, grip force, grip speed, grip endurance, hand dominance, muscle mass) that are connected to HGS and were not investigated in the current study. Therefore, current ndings should be noted with caution.

Conclusions
HGS is a vital marker of future health-compromising outcomes (e.g., reduced physical function-gait-related problems [impaired mobility], cardiovascular disease [hypertension]). It is very essential for the maintenance of healthy ageing and independence as well as a signi cant prerequisite for proper hand function. The present study demonstrated that sociodemographic (i.e., age, sex) and anthropometric (i.e., height, weight) correlate with HGS among the elderly population in Ghana. Older adults' physical deterioration is inevitable because of structural and functional musculoskeletal limitations due to ageing. Hence, a timely identi cation of those at risk for appropriate intervention to promote their healthy living is necessary. The reported mean values for HGS across age and gender with different height and weight could offer clinicians with baseline information when assessing the elderly. These reference values would also provide useful information for the general public to observe the HGS function of this vulnerable target group for early assistance.