Introduction

Sarcopenia is a reliable predictor of disability, frailty, and mortality, especially in the oldest-old. Sarcopenia is an age-related change in body composition, beginning in middle age (Cherin et al., 2014) where muscle mass decreases and fat mass increases. For most practicioners, body mass index (BMI) is their primary assessment of body composition; however, BMI does not discriminate between muscle and adipose tissue and does not directly assess regional adiposity. Sarcopenia is best appreciated by measuring lean soft tissue, body fat, and bone density. Dual-energy X-ray absorptiometry (DEXA) is currently considered the gold standard (Blake & Fogelman, 2010) for measuring these separate components and a superior index of sarcopenia. DEXA works by comparing and interpolating the absorption of varying energy levels of X-rays as they pass through the body to appreciate lean, fat and bone masses. Because of the high costs associated with purchasing and maintaining DEXA machines to measure sarcopenia, they are not readily available in low and middle-income countries (LMIC). Thus, multiethnic and cross-cultural validation of an alternate, low cost and reliable measure of body composition would accelerate clinical research and practice in developing nations, where population aging will have a significant impact on the world-wide burden for chronic age-related disease.

While girth indices such as waist-to-hip and waist-to-height (Flegal et al., 2009) explain malnutrition (Briend et al., 1989) and frailty (Landi et al., 2013, 2014), they may be inaccurate for predicting percent body fat (%BF). Other methods used to measure muscle mass include bioelectrical impedance, computed tomography, magnetic resonance imaging, urinary excretion of creatinine, anthropometric assessments, and neutron activation assessments (Morley, 2008). Body composition can also be measured with low-cost, easy-to-administer and objective measures of muscle strength like the short physical performance battery (Guralnik et al., 1994), gait speed (Buchner et al., 1996), timed get-up-and-go test (Podsiadlo & Richardson, 1991), and the stair climb power test (Bean et al., 2007).

In the current cross-cultural research we investigated if handgrip strength (HGS) was associated with DEXA equivalently in healthy aging Costa Ricans (CR) versus healthy aging Kansans (KS) (Vidoni et al., 2015) and if HGS added predictive value to BMI to predict DEXA values. We chose this widely accepted measure of upper body strength (indexed by dynamometer) because it shares a similar profile of association with disability (Newman et al., 2003) and mortality (Y. H. Kim et al., 2016; Sobestiansky et al., 2019) as DEXA. Furthermore, HGS has been associated with cognitive decline (McGrath et al., 2019; Rogers & Jarrott, 2008) and may also be a low-cost index to help stratify older adults at risk to lose their independence. HGS may prove to be an excellent and low-cost index of muscle mass that will add precision to traditional measures of body composition. Further, we want to test if the association between HGS and DEXA is not influenced by well-documented differences in height and body weight distribution across ethnic groups. In this context, ethnic-specific heterogeneity of risk factors for sarcopenia, disability, and frailty calls for culturally sensitive and reliable prediction of DEXA values using low-cost alternatives like HGS. Thus, HGS is a variable of great interest in this multiethnic comparison of predictors of health and disability in older adults, as it has been used primarily in the US and EU with majority white populations.

The primary purpose of this paper is to describe the efficacy of HGS as a predictive tool for health outcomes in both European Americans and Central Americans by comparing the relationship between HGS and measurements of body composition and upper body strength using the DEXA across race and gender. We hypothesize the HGS predicts DEXA values equally well in Meso-Americans as in Euro-Americans.

Methods

Participants

Seventy-eight CR participants (26 men and 52 women) were recruited from the Epidemiology and Development of Alzheimer's Disease Project (EDAD), the Costa Rican Gerontological Association (AGECO), and the Integral Program for Older Adults (PIAM) from the University of Costa Rica (UCR). These participants were given a measurement appointment at the Human Movement Sciences Research Center at the UCR in San José, CR (CIMOHU). In the US, a sample of 100 European Americans from KS were recruited (35 men and 65 women) and were given a measurement appointment at the Alzheimer’s Disease Center at the University of Kansas. The Scientific Ethics Committee at the UCR and the Institutional Review Board at the University of Kansas approved their respective research protocols.

Instruments and procedures

Participants underwent body height (cm), mass (kg), and body composition assessment using standard protocols (American College of Sports Medicine, 2010, 2014; Nana et al., 2015). Body height was measured to the nearest 0.1 cm using a stadiometer. Individuals stood still with their heads in the Frankfort horizontal plane, barefoot, feet together, and the back surfaces of the calcaneus, pelvic, pectoral girdles and occipital regions in contact with the wall. Body mass was measured on a digital platform scale where individuals remained in light clothing, barefoot, feet positioned in the center of the platform, and arms next to their bodies.

In both testing sites, dual-energy X-ray absorptiometry (DEXA; Lunar Prodigy (GE Medical Systems, Madison, WI, USA) was used to determine regional and total body composition variables of fat mass, lean mass, and bone material density (BMD). Scans were performed according to the laboratory standard protocol following safety precautions and quality control according to international standards (Lewiecki et al., 2016). The appendicular lean soft tissue (ALST) in kilograms was considered equivalent to the sum of total lean soft tissue in both right and left arms and legs. Intermuscular adipose tissue-free skeletal muscle mass (IMAT-F SMM) was determined by the equation developed from magnetic resonance imaging scans (Kim et al., 2004) as follows: 1.19 (ALST)—1.65 ± 1.46 kg. In addition, body fat mass index (BFMI; total fat mass/height2), lean tissue mass index (LTMI; total lean mass/height2), and ALST index (ALSTI; ALST/height2) were calculated.

Handgrip strength (HGS) was measured with a hand-held dynamometer. The KS sample used a JAMAR hydraulic dynamometer (Patterson Medical, Warrenville, IL, USA) and the CR sample an electronic CAMRY, model EH-101 dynamometer (CAMRY, City of Industry, CA, USA). HGS was measured in a seated position with the elbow flexed at 90°, forearm in neutral position, and wrist between 0° and 30° dorsiflexion and between 0° and 15° ulnar deviation. Participants were encouraged to squeeze the dynamometer as hard as they could and to maintain it that way for 3–5 s. HGS was measured three times on the dominant hand. The final score was the average of the three attempts, which has been reported to provide the highest test–retest reliability (Haidar et al., 2004; Shiratori et al., 2014). HGS values were transformed to z-scores (z-score = (raw score – mean) / standard deviation) for statistical analysis given that two different dynamometer brands were used during data collection (Amaral et al., 2012).

Statistical analysis

Statistical analysis was performed with the IBM-SPSS Statistics, version 22 (IBM Corporation, Armonk, NY, USA). Descriptive statistics are presented as mean and standard deviation (M ± SD), unless otherwise noted. Inferential analysis was performed by 2 × 2 (sample by gender) ANOVA on anthropometric, body composition and strength variables. Post-hoc analysis was completed using Tukey comparisons. Because of an observed difference between mean age in the CR and KS groups, a single factor ANOVA was conducted to assess the between group differences. Because body composition and strength change across a lifespan, age was included as a covariate in all analyses to equate the CR and KS groups on all outcomes. Adjusted Pearson correlations were then computed between HGS and anthropometric and body composition variables controlling for age and BMI.

Results

Descriptive statistics were collected to characterize the two racial groups by body composition, and gender (Appendix Table 1). In general, the KS sample was older than the CR sample (72.84 ± 5.59 vs. 68.91 ± 4.79 yr.; p ≤ 0.001). The CR sample was shorter than the KS sample (158.63 ± 8.77 vs. 167.39 ± 9.72 cm; p ≤ 0.001), and women were shorter than men (158.54 ± 7.21 vs. 172.99 ± 8.39 cm; p ≤ 0.001). A significant interaction between samples and genders in body weight was found (p = 0.046). Post hoc analysis showed that within KS and CR samples, men were heavier than women (84.27 ± 16.15 vs. 69.05 ± 11.61 kg; p < 0.05). In addition, between samples, KS participants had a higher body weight than CR participants (78.60 ± 15.28 vs. 68.84 ± 13.17 kg; p < 0.05). No significant interactions (p = 0.293) or main effects between samples (p = 0.222) and genders (p = 0.511) were found on BMI.

In general, KS older adults showed higher total adiposity (29.59 ± 8.83 vs. 25.59 ± 8.40 kg; p = 0.001), regional arms adiposity (2.62 ± 0.93 vs. 2.30 ± 0.82 kg; p = 0.003), legs adiposity (9.62 ± 3.63 vs. 7.96 ± 3.22 kg; p = 0.001), and trunk adiposity (29.59 ± 8.83 vs. 25.59 ± 8.40 kg; p = 0.001) than CR older adults. Within genders, women showed higher regional arms (2.62 ± 0.79 vs. 2.20 ± 1.01 kg; p = 0.001) and legs (9.88 ± 3.25 vs. 7.02 ± 3.33 kg; p ≤ 0.001) adiposity than men. However, no significant interaction was found between samples and genders on fat arm mass (p = 0.058), fat leg mass (p = 0.814), or total fat mass (p = 0.173). A significant interaction was found between samples and genders on fat trunk mass (p = 0.042). Follow-up analysis indicated that within men, fat trunk mass was lower in the CR older adult than in the KS older adult (p = 0.004). Fat trunk mass was similar between KS and CR women (p = 0.774). In addition, men had lower %BF than women (31.24 ± 7.02 vs. 41.28 ± 5.87%; p ≤ 0.001), but no significant interaction was found between samples and genders on %BF (p = 0.182). In this study, regardless of the sample, women had higher BFMI than men (11.3 ± 3.2 vs. 8.9 ± 3.2 kg/m2; p ≤ 0.001). There was also no significant interaction found between samples and genders on BFMI (P = 0.264).

In this study, men showed more lean mass tissue than women (Appendix Table 1). Compared to women, men had higher lean arm mass (6.34 ± 1.60 vs. 3.72 ± 0.59 kg; p ≤ 0.001) and higher lean trunk mass (27.08 ± 9.05 vs. 18.95 ± 2.48 kg; p ≤ 0.001). Additionally, women had lower LTMI (15.0 ± 1.7 vs. 18.0 ± 1.7 kg/m2; p ≤ 0.001) and ALSTI (6.2 ± 0.8 vs. 8.1 ± 1.0 kg/m2; p ≤ 0.001) than men. No significant interactions were found between samples and genders on lean arm mass (p = 0.251), lean trunk mass (p = 0.593), LTMI (p = 0.527), or ALSTI (p = 0.668). However, regardless of gender, KS participants showed higher lean arm mass than CR participants (4.85 ± 1.51 vs. 4.34 ± 1.75 kg; P = 0.002). Also, regardless of gender, lean leg mass (p = 0.017), total lean mass (p = 0.030), and IMAT-F SMM (p = 0.025) was lower in the CR group than in the KS group. Follow-up analysis indicated that within men, lean leg mass (p ≤ 0.001), total lean mass (p ≤ 0.001), and IMAT-F SMM (p ≤ 0.001) was lower in the CR participants than in the KS participants, and within women, lean leg mass (p = 0.095), total lean mass (p = 0.174), and IMAT-F SMM (p = 0.180) was similar. Within the KS sample and within the CR sample, men had higher lean leg mass, total lean mass, and IMAT-F SMM than women (p ≤ 0.001 for all).

In general, KS older adults showed higher total BMD (1.16 ± 0.11 vs. 1.08 ± 0.11 g/cm2; p ≤ 0.001) than CR older adults. Additionally, regional head (2.23 ± 0.27 vs. 2.07 ± 0.30 g/cm2; p = 0.001), arms (0.83 ± 0.12 vs. 0.78 ± 0.09 g/cm2; p ≤ 0.001), legs (1.26 ± 0.18 vs. 1.14 ± 0.16 g/cm2; p ≤ 0.001), ribs (0.69 ± 0.10 vs. 0.64 ± 0.07 g/cm2; p ≤ 0.001), pelvis (1.11 ± 0.13 vs. 1.04 ± 0.12 g/cm2; p ≤ 0.001), and spine (1.08 ± 0.18 vs. 1.04 ± 0.14 g/cm2; p = 0.034) BMD were higher in KS older adults than CR older adults. For all BMD measures, men showed higher BMD than women. No significant interactions were found between samples and genders for all BMD measures (Appendix Table 1).

No significant interaction was found between samples and genders on HGS z-scores (p = 0.843). Men had higher HGS than women (1.01 ± 0.87 vs. -0.55 ± 0.48 z-score; p ≤ 0.001). Significant correlations were obtained between HGS and age (r = -0.18, 95% CI [-0.03, -0.32], p ≤ 0.05). HGS was also significantly associated (p < 0.001 for all) with body height, body weight, fat trunk, head BMD, legs BMD, ribs BMD, pelvis BMD, spine BMD, total BMD, lean arms mass, lean legs mass, lean mass trunk, total lean mass, and %BF, BFMI, LTMI, and ALSTI. The correlations between HGS and anthropometric and body composition variables were different between men and women from CR and KS (Appendix Table 2). HGS was significantly predictive of age in CR men (r = -0.27, 95% CI [-0.01, -0.43], p ≤ 0.05), CR women (r = -0.47, 95% CI [-0.23, -0.66], p ≤ 0.001), KS men (r = -0.44, 95% CI [-0.12, -0.67], p ≤ 0.01) and KS women (r = -0.28, 95% CI [-0.04, -0.49]).

When controlled for age, the only significant correlation that extended across both samples and genders was between HGS and lean arms mass (CR men r adj = 0.36, CR women r adj = 0.48, KS men r adj = 0.64, KS women r adj = 0.42; p ≤ 0.05 for all). As expected, BMI was heavily correlated with a majority of anthropometric and body composition variables across samples and gender groups (Appendix Table 2). When controlled for BMI, lean arms mass (CR men r adj = 0.44, CR women r adj = 0.46, KS men r adj = 0.71, KS women r adj = 0.48; p ≤ 0.05 for all) and total lean mass (CR men r adj = 0.42, CR women r adj = 0.39, KS men r adj = 0.50, KS women r adj = 0.56; p ≤ 0.05 for all) were significantly correlated with HGS across samples and gender (Appendix Table 2 and Fig. 1).

Discussion

Body composition measures differed between older adults from the US and CR. HGS was a dominant predictor of age and the upper body strength biomarker, lean arm mass, in both samples. Men had higher HGS than women and KS women’s biomarkers of upper body strength showed the greatest relationship to HGS. From the strong correlations between HGS and measures of body composition across sample and gender, we expected a similarly robust predictive relationship when controlled for age. However, only lean arm mass was significantly related to HGS across cultural groups and gender. ALSTI showed the potential of a subtle effect, but the relationship was unclear or lost due to its addition of lean leg mass, indicating the effect to lie with lean arm mass alone. This weakened relationship suggests that age confounded for some of the original associations seen with HGS and the biomarkers of upper body strength. These findings converge with previously published reports from international cohorts, suggesting that lean arm mass may be an indicator of overall health in aging across racial groups that have characteristically varying body composition measures.

Significant race and gender differences in adiposity were observed in the present study, a previously reported finding (Silva et al., 2010). However, the absence of differences in BMI between men and women from CR and KS supports the discussion regarding the appropriateness of this marker as a phenotypic proxy of adiposity across populations differing in race and ethnicity (Heymsfield et al., 2016; Natale & Rajagopalan, 2014). Based on the findings of this study, BMI may be a questionable measure of adiposity across racial groups.

To find that women are shorter than men in both samples, and that Caucasians are taller and heavier than Mexican and Central Americans follows established patterns since childhood, where approximately 20% of variation in body height is attributed to, but not limited, to environmental variation (e.g., nutrition, diseases) (Clark et al., 2016; Dodds et al., 2016; Heymsfield et al., 2016; Natale & Rajagopalan, 2014; Silventoinen, 2003). To further contextualize the variation in body composition and HGS across racial and ethnic groups, a thorough literature review across international cohorts is presented in Appendix B. A larger cross-cultural comparison enhances the understanding of the pragmatic application of HGS as a translation of previous methods for multiethnic populations.

Limitations

The primary limitation experienced was the wide difference between the mean ages of the two sample groups. Age was found to have a robust effect on the relationship seen between HGS and the measures of body composition and biomarkers of upper body strength. This limitation was mitigated by controlling for age in our analyses. Another limitation in the make-up of the samples was the low number of participants. This may have impacted the power of the relationships between HGS and biomarkers of upper body strength. In the future, multiethnic comparisons should be made between large sample sizes of demographically comparable groups to amplify all subtle effects. Never-the-less these data indicate that comparing BMD between different groups also requires the development of ethnic-specific reference data. Most DEXA equipment usually reports European American reference values, which have been shown to misclassify other populations (e.g., Chinese) (Lo et al., 2016). In this study, we used the same DEXA equipment brand, software (i.e., reference standards) and calibration process; therefore, we reduced the variation between DEXA equipment brands influencing BMD scores and classification (Schousboe et al., 2014).

A source of potential error or variance also came from measuring HGS with two different dynamometer brands. Strong associations have been reported between different dynamometer brands and the JAMAR dynamometer, considered a benchmark in hand dynamometry (r > 0.77); however, the agreement between HGS was poor (Guerra & Amaral, 2009). We corrected potential bias in this measure as best as possible by z-transforming raw scores for analysis.

Conclusion

Results of this study show that there are significant physiological differences between older US Euro-American and Meso-American adults. A comparison across nations highlights the potential of HGS to serve as an effective, universal predictor of upper body strength, and the corresponding health outcomes in older adults. Biological and environmental factors may affect aspects of physical health like muscle function, body mass, and bone mineral density, which could play a role in explaining variation in cardiovascular disease, hypertension, diabetes mellitus, asthma, and mortality across racial groups (Dodds et al., 2016; Leong et al., 2016; Silverman, 2015; Stenholm et al., 2012). Furthermore, the observed ethnic-specific heterogeneity on biologic factors and physical-related performance creates a need for culturally diverse prevention programs for older adults. While the gold standard of body composition measurement is currently DEXA, this methodology is expensive to acquire and maintain, limiting the accuracy of reference values used to predict health outcomes in low and middle income countries. HGS is a reliable, low-cost alternate to DEXA for the measurement of proxies of age-related disability across cultures. The results of this study justify further investigation into developing HGS as a standardized tool for healthcare providers for a global population.