Cardiometabolic Risk,Socio-Psychological Factors,and Trajectory of Grip Strength Among Older Japanese Adults

Abstract

Objective: To examine the association between cardiometabolic risk (percent body fat [BF%], triglycerides [TG], high-density lipoprotein [HDL]-cholesterol, hemoglobin A1c [HbA1c]), socio-psychological factors (education and self-rated health [SRH]), and trajectories of grip strength (GS) in older adults. Method: Longitudinal 8-year data from 1,381 Japanese adults aged 65 years or above were analyzed using hierarchical linear models, stratified according to gender. Results: GS declined following a linear trajectory. In both genders, higher BF% was associated with weaker GS, but not with the rate of decline. GS trajectory did not correlate with baseline TG, HDL-C, or HbA1c. Cardiometabolic factors mediated educational differences in GS intercept in both genders. In women, better SRH predicted stronger GS. The effect of SRH was robust to adjustments for cardiometabolic risk. Discussion: In older adults, GS and its rate of decline are selectively associated with both cardiometabolic risk and socio-psychological characteristics. Cardiometabolic risk mediates educational disparities in GS but not differences in subjective assessments of health.

Keywords

Japan aging grip strength cardiometabolic risk factors

Introduction

Age-related decline in muscle strength and physical functioning leading to disability creates a considerable burden for many older adults. Knowledge about muscle strength in the elderly is important because the ability to perform normal daily work-related, household, and recreational activities is partially determined by the force-generating capacity of skeletal muscles. Impairments in muscle strength have been associated with falls, decreased mobility, functional dependence, and subsequent disability (Hughes et al., 2001; Stenholm et al., 2009). As such, the evaluation of muscle strength and the identification of factors contributing to the age-related decline in muscle strength in older adults are of utmost clinical and public health interest.

Prior research has shown that hand-grip strength is a valid and reliable estimate of isometric upper-extremity muscle strength, correlates well with strength in other muscle groups (Rantanen, Era, Kauppinen, & Heikkinen, 1994; Rantanen et al., 1998), and therefore has been suggested as an estimate of “overall muscle strength” (Frederiksen et al., 2006). In addition, grip strength (GS) has shown strong predictive associations with other outcomes of interest in the elderly, such as physical functioning and disability (Rantanen et al., 1999; Syddall, Cooper, Martin, Briggs, & Aihie Sayer, 2003), morbidity (Rantanen et al., 1998), and total or cause-specific mortality (Gale, Martyn, Cooper, & Sayer, 2007; Rantanen et al., 2003; Sasaki, Kasagi, Yamada, & Fujita, 2007). Consequently, GS has emerged as a critical outcome measure for research and clinical evaluation in older adults.

Converging evidence documents the association between an individual’s cardiometabolic risk profile and various indicators of muscular strength, including GS (Everson-Rose et al., 2011; Sayer et al., 2007). Cardiometabolic risk refers to a constellation of adiposity-related metabolic abnormalities that increase the risk for cardiovascular disease, such as elevated anthropometric (e.g., body-mass index [BMI] or waist circumference) and body fat composition (e.g., percent body fat [BF%]) indicators, abnormal lipid profiles (e.g., elevated triglycerides [TG] and/or low high-density cholesterol), insulin resistance, and elevated systolic and/or diastolic blood pressure (BP; Janiszewski, Janssen, & Ross, 2007; Klein et al., 2012; Wildman et al., 2008). This prevalent yet treatable constellation of metabolic abnormalities is of clinical and research interest given the growing rates of obesity and type-II diabetes documented in older adults (Flegal, Carroll, Ogden, & Curtin, 2010; Wildman et al., 2008).

Previous studies have shown that in older individuals, abnormalities in cardiometabolic risk are associated with deficits in muscular function, including reduced GS (Atlantis, Martin, Haren, Taylor, & Wittert, 2009; Everson-Rose et al., 2011; Howard et al., 2007; Sayer et al., 2007). However, because these observations have been based primarily on cross-sectional data, which confound intra- and inter-individual differences, important questions remain unanswered. For instance, given that on average GS declines in old age (Hughes et al., 2001; Metter, Conwit, Tobin, & Fozard, 1997), how does cardiometabolic risk affect the level and the rate of change in GS over time?

GS has also been associated with socio-psychological individual characteristics, including various measures of socioeconomic status (SES) and subjective assessments of health status. These studies have shown that older individuals of higher SES, and in particular those with higher educational achievement, have stronger GS and a slower decline in GS over time (Mohd Hairi, Mackenbach, Andersen-Ranberg, & Avendano, 2010; Rautio, Heikkinen, & Ebrahim, 2005; Strand, Cooper, Hardy, Kuh, & Guralnik, 2011; Syddall, Evandrou, Cooper, & Aihie Sayer, 2009). In addition, SRH, a widely validated subjective measure of global health status in older adults (Jylhä, 2009), has shown strong and consistent associations with both individual socioeconomic characteristics (Liang et al., 2010; Molarius et al., 2007) and with the subsequent development of functional limitations and disability (Idler & Kasl, 1995; Idler, Russell, & Davis, 2000; Liang et al., 2007). Notwithstanding one study showing SRH to be positively associated with GS among postmenopausal Finish women (Sirola et al., 2004), research investigating this association in the general population of older adults is sparse. Clarifying how measures of cardiometabolic risk influence the trajectory of GS within the context of heterogeneity in socio-psychological variables (e.g., education and SRH) may be an essential step in understanding how individual socio-psychological characteristics “get under the skin” and modify biological processes that ultimately lead to disease, disability, and mortality.

For this study, we used an integrated framework, including social, psychological, and biological factors (Crimmins & Seeman, 2004) to examine whether cardiometabolic risk factors mediate the association between socio-psychological characteristics and the trajectory of GS (i.e., level and rate of change over time) in a cohort of older Japanese adults. This integrated paradigm posits that socio-psychological attributes influence physiological processes, which in turn predispose individuals to morbidity and mortality. We tested three hypotheses.

Hypothesis 1: Abnormal cardiometabolic risk factors are associated with a lower level and a faster rate of decline in GS over time.

Hypothesis 2: Lower SES and poor SRH are associated with weaker GS and a faster rate of decline in GS over time.

Hypothesis 3: Cardiometabolic risk factors mediate, at least partially, the association between socio-psychological variables (i.e., SES and SRH) and GS trajectory intercept and slope.

Design and Method

Study Design and Sample

Data were collected annually from 2002 to 2009 as part of comprehensive health examinations in the eastern Japanese town of Kusatsu. The study design, recruitment and follow-up procedures, and inclusion/exclusion criteria, have been described in greater detail elsewhere (Fujiwara et al., 2005). Briefly, comprehensive medical examinations were conducted each year for individuals aged 70 years or above to screen for chronic diseases and geriatric conditions. Starting in 2006, individuals aged 65 and above were also enrolled. The sample was supplemented each year with respondents who became eligible. A total of 1,381 individuals (574 men and 807 women) aged 65 years and above participated in the program involving a maximum of eight yearly examinations (2002-2009), with a mean of 3.11 examinations per individual.

Measures

Hand-grip strength

As recommended (Innes, 2002), hand-grip strength was measured in kilograms (kg), as the average of two maximal trials of the dominant hand with a Smedley-type dynamometer (Yagami Co, Tokyo). A trained nurse administered the test according to protocol.

Cardiometabolic risk

Indicators were selected and the abnormal levels were defined based on the recommendations from the 2009 Joint Interim Statement from multiple scientific associations (Alberti et al., 2009), and included abnormal body fat composition, glucose intolerance, high TG, and low high-density lipoprotein cholesterol (HDL-C).

Because the classification for abdominal obesity varies by country or population (Alberti et al., 2009), we chose BF% as a measure of abnormal body fat composition to extend the generalizability of our findings. BF% was determined by bio-impedance methods with TBF-300 (Tanita Corp., Tokyo, Japan), an instrument with proven reliability in Asian populations (Tong & Fung, 2003). In accordance with prior studies (Ferrucci et al., 1997; Zoico et al., 2004) and because the relationship between body weight and BF% differs among population/ethnic groups, the analyses of associations between BF% and GS were adjusted for measured weight (kg) and height (cm).

Non-fasting blood samples were collected and routine tests of biochemical markers were performed according to standardized methods using a sequential autoanalyzer in a single laboratory over the course of the study. Glucose intolerance was assessed using glycosylated hemoglobin A1c (HbA1c). HbA1c is a better marker of long-term glycemic control compared with fasting glucose levels (which were not available in this study; Weykamp et al., 2008). HbA1c ≥ 6.0% was defined as high/abnormal (Bonora et al., 2011; Selvin, Steffes, Gregg, Brancati, & Coresh, 2011). The thresholds for elevated TG (TG ≥150 mg/dL) and reduced HDL-C (HDL-C ≤ 40 mg/dL for men and ≤ 50 mg/dL for women) were set in accordance with current recommendations (Alberti et al., 2009). Self-reported, physician-prescribed glucose- and lipid-lowering medication use (0 = no medication, 1 = medication use) were included as control variables because they represent proxy indicators for glucose intolerance and dyslipidemia (Alberti et al., 2009). BP status was included as a control variable because of its association with the other cardiometabolic risk factors (as a component of metabolic syndrome); elevated BP was defined as systolic BP ≥130 and/or diastolic BP ≥85 mm Hg. Analyses were performed separately with and without control for elevated blood pressure, and showed similar results (results from BP-controlled analyses shown below).

Socio-psychological and control variables

Education (years of school completed) was used as an indicator of SES (Braveman et al., 2005; Winkleby, Jatulis, Frank, & Fortmann, 1992); as a marker of lifetime SES, education is more stable and less susceptible to reverse causation effects (i.e., lower SES due to poor health) compared with either income or assets. SRH, a subjective global assessment of an individual’s own health status, has been documented as a strong predictor of mortality, decline in functional status and disability, and loss of independent living among older adults (Idler & Benyamini, 1997). SRH was measured using the participants’ ratings of their general health status, on a scale from 1 (excellent) to 4 (poor). Because of previously observed gender differences in levels of GS (Frederiksen et al., 2006), the analyses were performed stratified according to gender. Additional control variables included age (in years, calculated in 2001, regardless of when the respondent was enrolled in the study), current smoking (0 = non-smoker, 1 = smoker) and current alcohol use (0 = non-drinker [“does not drink any alcohol”], 1 = drinker [“drinks alcohol”]).

To account for possible variations over time in the predictors (SRH, BF%, TG, HDL-C, and HbA1c) and control variables (height, weight, smoking, alcohol, BP, and medication use), the respective measures were specified concurrently as time-varying (i.e., to assess intra-individual wave-to-wave changes) and as time-constant covariates (i.e., to assess inter-individual differences at baseline) and included as applicable in the Level 1 or Level 2 model equations.

Statistical Analysis

Model specification

Hierarchical linear models were used to estimate the GS growth trajectory defined by intercept (level at time T_i) and slope (rate of change), while accounting for intra-individual (Level 1) and inter-individual (Level 2) variability in wave-to-wave changes and cross-level interactions of time with the predictors (Raudenbush & Bryk, 2002). The two-level models were specified as follows:

Level 1 (intra-individual changes over time) equation:

Y_{i T} = π_{0 i} + π_{1 i} Time + Σ π_{k i} X_{k i} T + ε_{i} T,

where Y_iT is the GS of individual i at time T. π_0i is the GS status at time T (centered), and π_1i is the GS slope (i.e., rate of change) over time. Time is the distance (in years) of assessment from the baseline, when the respondent was first examined. X_kiT represents the time-varying covariates (e.g., SRH, BF%, TG, HDL-C, and HbA1c) associated with individual i at time T, and π_ki represents the effect of X_k on individual i’s GS. ε_iT is a random error.

Time was centered at its mean to minimize the possibility of multicollinearity when evaluating non-linear time (T) functions. Thus, the intercept for any given model should be interpreted as the GS level at the mean follow-up time. We tested for the possibility of curvilinear components or multiple growth trajectories, as well as single linear change, and selected the most parsimonious and best-fitting model based on the lowest deviance relative to the df (Raudenbush & Bryk, 2002).

Level 2 (inter-individual differences) equation was specified as follows:

π_{p i} = β_{p} 0 + Σ β_{p q} X_{q i} + r_{p i},

where X_qi is the qth time-constant covariate (e.g., education, age at baseline, and baseline values for all time-varying covariates) associated with individual i, and β_pq represents the effect of variable X_q on the pth growth parameter (p; that is, intercept and slope). r_pi is a random effect with a mean of 0.

SRH, BF%, TG, HDL-C, HbA1c, and selected control variables (height, weight, smoking, alcohol use, BP, and medication use) were included both as time-varying covariates to estimate intra-individual longitudinal changes and as time-constant covariates (fixed at the baseline value) to estimate inter-individual baseline differences. Fixed variables (i.e., education, age-at-baseline, and attrition) were included as time-constant covariates to estimate trajectory differences as a function of education, age, and attrition status, respectively. To ensure the correct estimation of intra- and inter-individual differences in GS trajectory coefficients, where appropriate, time-varying covariates were centered at group-mean (i.e., within-individual centering), whereas time-constant covariates were centered at grand mean (i.e., within-sample centering; Raudenbush & Bryk, 2002).

Following the classic mediation-testing approach (Baron & Kenny, 1986), we sequentially estimated the following associations: (a) between cardiometabolic risk indicators and the trajectory of GS, (b) between socio-psychological variables and cardiometabolic risk, and (c) between socio-psychological attributes and changes in GS, while controlling for cardiometabolic risk. The analyses were performed separately for men and women, using an identical sequence of models

Treatment of missing data

To minimize the loss of participants due to item non-response, multiple imputation was undertaken using the NORM software (Schafer & Olsen, 1998). Three imputed data sets were created and hierarchical linear modeling (HLM) analyses were performed using each data set. Parameter estimates and their standard errors were calculated by averaging across the three data sets and adjusting for their variance (Raudenbush & Bryk, 2002).

As a major advantage, multilevel models can include every participant in the estimation, regardless of how many observations one contributed to the data set. With reference to attrition, multilevel models are predicated on the assumption of missing-at-random, namely, that the probability of missing depends only on the observed data for either the covariates or the outcome variables, hence permitting valid inference (Raudenbush & Bryk, 2002).

We also addressed the potential for selection bias due to non-ignorable missing data resulting from participants’ death and other non-random causes of dropout (Little & Rubin, 2002). A binary indicator for overall attrition (1 = dropped out, 0 = completed study) was included as an inter-individual level covariate to control for selective mortality and dropout. This approach, previously used in other studies of trajectories of health in older adults (Liang et al., 2008; Quiñones, Liang, Bennett, Xu, & Ye, 2011), is similar to pattern-mixture modeling, in which participants are classified into different groups, based on patterns of missing data, and estimates are obtained by averaging over the identified patterns (Hedeker & Gibbons, 2006).

Finally, the statistical significance level was set at p < .05 (two-tailed). All analyses were performed using the HLM 6.6 software (Scientific Software International, Lincolnwood, IL).

Results

Sample descriptive characteristics at baseline, for the full sample and separately by gender, are shown in Table 1. Although similar in education and self-assessments of own health (SRH), men and women differ in both GS and cardiometabolic risk profiles (Table 1). As expected, GS was substantially higher among men (31.8) compared with women (18.9). In terms of cardiometabolic risk indicators, women had higher BF% (28.7 % vs. 20.2 % in men) and higher rates of abnormal HDL-C (17.1% vs. 8.4% in men), whereas men had higher rates of elevated TG (44.3% vs. 36.9% in women) and high HbA1c (17.1% vs. 11.4% in women).

Table 1.

Sample Descriptive Characteristics at Baseline.

	Men (n = 574)	Women (n = 807)	Total (N = 1,381)
	M (SD) or %	M (SD) or %	M (SD) or %
Baseline characteristics
Grip strength (kg)	31.8 (7.1)	18.9 (5.4)	24.3 (8.9)
Age (years, in 2001)	71.9 (5.5)	72.6 (6.3)	72.3 (5.9)
Education (years)	9.9 (2.7)	9.0 (2.4)	9.4 (2.6)
Self-rated health (1-4: excellent-poor)	2.0 (0.7)	2.1 (0.7)	2.0 (0.7)
Height (cm)	161.4 (6.2)	148.0 (6.2)	153.6 (9.1)
Weight (kg)	60.3 (9.3)	51.0 (8.3)	54.9 (9.9)
Percent body fat (%)	20.2 (5.4)	28.7 (6.4)	25.2 (7.3)
Triglycerides (0 = normal, 1 = high)	44.3%	36.9%	39.9%
HDL-C (0 = normal, 1 = low)	8.4%	17.1%	13.5%
HbA1c (0 = normal, 1 if ≥6%)	17.1%	11.4%	13.8%
Elevated BP (0 = no, 1 = yes)	68.1%	63.8%	65.6%
Cholesterol medication (0 = no, 1 = yes)	5.4%	17.4%	12.4%
Diabetes medication (0 = no, 1 = yes)	8.0%	4.6%	6.0%
Alcohol use (0 = no, 1 = yes)	59.6%	24.0%	38.8%
Smoker (0 = no, 1 = yes)	29.0%	10.9%	18.4%
Attrition^a (0 = no; 1 = yes)	53.7%	55.6%	54.8%

Note. Triglycerides: abnormal if ≥150 mg/dL; HDL-C (high-density lipoprotein cholesterol): abnormal if ≤40 mg/dL for men and ≤50 mg/dL for women; HbA1c (glycosylated hemoglobin A1c): abnormal if ≥6.0%; BP (blood pressure): elevated if systolic BP ≥130 and/or diastolic BP ≥85 mm Hg.

Attrition denotes participants who either died or permanently left the study.

The unconditional time-only model (M₀, Table 2) showed that in men GS followed a trajectory with a higher GS intercept (b = 31.06 vs. 18.36, p < .001 for both) but a faster rate of linear decline (b = −0.48 vs. −0.30, p < .001 for both) compared with women. We also tested for non-linear changes in GS over time; the quadratic and cubic time-slope coefficients were non-significant in all models (i.e., no significant acceleration or deceleration observed in the rate-of-decline over time) and were not included in subsequent models (not shown, available upon request). Results from the sequentially adjusted linear models are shown below, separately for men and women.

Table 2.

GS Trajectory Estimates for Men and Women: Hierarchical Linear Modeling Results (2002-2009).

	Men (n = 574)								Women (n = 807)
	Model M₀		Model M₁		Model M₂		Model M₃		Model M₀		Model M₁		Model M₂		Model M₃
Fixed effects	coefficient	p	coefficient	p	coefficient	p	coefficient	p	coefficient	p	coefficient	p	coefficient	p	coefficient	p
GS intercept
Intercept	31.06	.00	30.95	.00	31	.00	31.07	.00	18.36	0	18.26	.00	18.36	.00	18.37	.00
Age (years, in 2001)					−0.54	.00	−0.41	.00					−0.35	.00	−0.25	.00
Education (years)					0.22	.015	.00	.975					0.18	.008	0.04	.482
Attrition (yes vs. no)			−2.54	.00	−1.28	.015	−1.29	.008			−1.46	.00	−0.60	.073	−0.55	.073
Percent body fat			−0.03	.772			−0.12	.008			−0.05	.27			−0.10	.034
TG (high vs. normal)			−0.05	.932			0.06	.91			0.3	.317			0.2	.491
HbA1C (high vs. normal)			−0.61	.383			0.02	.973			−0.91	.145			−0.82	.186
HDL-C (low vs. normal)			2	.022			1.51	.067			−0.59	.156			−0.45	.253
SRH					−0.41	.257	−0.29	.377					−0.74	.01	−0.56	.029
GS slope
Intercept	−0.48	.00	−0.49	.00	−0.50	.00	−0.41	.00	−0.30	.00	−0.35	.00	−0.30	.00	−0.27	.001
Age (years, in 2001)					−0.02	.133	−0.01	.416					−0.02	.133	−0.01	.405
Education (years)					.00	.953	−0.01	.672					0.02	.953	0.01	.497
Attrition (yes vs. no)			−0.32	.017	−0.26	.062	−0.24	.083			−0.19	.029	−0.20	.062	−0.19	.032
Percent body fat			0.03	.226			0.03	.194			−0.00	.937			−0.00	.904
TG (high vs. normal)			−0.03	.818			−0.02	.91			−0.03	.748			−0.01	.886
HbA1C (high vs. normal)			0.06	.755			0.04	.798			0.33	.181			0.36	.137
HDL-C (low vs. normal)			−0.11	.68			−0.13	.619			−0.01	.904			−0.05	.639
SRH					0.1	.25	0.13	.182					0.13	.25	0.11	.127
Time-varying measures
Percent body fat			0.08	.079			0.09	.054			−0.13	.163			−0.11	.196
TG (high vs. normal)			0.16	.489			0.17	.473			−0.18	.392			−0.17	.41
HbA1C (high vs. normal)			−0.86	.085			−1.01	.041			0.13	.743			0.09	.816
HDL-C (low vs. normal)			−0.24	.617			−0.24	.619			−0.72	.013			−0.74	.011
Random effects	Variance	p	Variance	p	Variance	p	Variance	p	Variance	p	Variance	p	Variance	p	Variance	p
Intercept	38.6	.00	26.11	.00	28.2	.00	22.56	.00	18.89	.00	11.68	.00	13.15	0	10.12	.00
Time (t)	0.27	.00	0.3	.00	0.22	.00	0.28	.00	0.06	>.500	0.04	>.500	0.05	>.500	0.03	>.500
Level 1 E	8.25		8.1		8.22		8.07		9.34		9.13		9.23		9.11

Notes. GS = grip strength; SRH = self-rated health (range 1 = excellent to 4 = poor); TG = triglycerides: abnormal (high) if ≥150 mg/dL; HDL-C = high-density lipoprotein cholesterol (abnormal if ≤40 mg/dL for men and ≤50 mg/dL for women); HbA1c = glycosylated hemoglobin A1c (abnormal if ≥6.0%).

Models M₁ and M₃ controlled for weight, height, cholesterol- and glucose-lowering medication, BP (blood pressure): elevated if systolic BP ≥130 and/or diastolic BP ≥85 mm Hg), alcohol use, and smoking status.

Intercept for each fixed effect represents the estimate when all other variables are held constant as appropriate at 0 (binary variables) or at sample mean (continuous variables); GS slope represents the estimated change per year throughout the period of observation; random effects represent the estimated residual (unexplained) variance in GS slope, intercept, and wave-to-wave (Level 1) variation, respectively.

Attrition denotes participants who either died or permanently left the study.

Cardiometabolic Risk Factors and GS Trajectory (Model M₁ in Table 2)

Abnormal baseline HDL-C was associated a higher GS intercept only among men (b = 2.00, p = .022), whereas wave-to-wave changes in HDL-C status (from normal to abnormal) were associated with a wave-to-wave decline in GS only among women (b = −0.72, p = .013). In both genders, high BF%, elevated TG, and high HbA1c measured at baseline were not correlated with either the intercept or the rate-of-change in GS, and wave-to-wave changes in these indicators were not correlated with variations over time in GS.

Education, Self-Rated Health, and GS Trajectory (Model M₂ in Table 2)

Participants (both men and women) with higher levels of education had stronger GS (intercept: b = 0.22, p = .015 for men; b = 0.18, p = .008 for women), but did not differ in the GS rate-of-decline from participants with lower education. Worse (higher) SRH at baseline was associated with weaker GS only in women (intercept: b = −0.74, p = .010; Figure 1). SRH was not significantly associated with the GS rate of decline or with intra-individual wave-to-wave variations in GS in either sub-group.

Figure 1.

Differences in trajectories of grip strength according to SRH (2002-2009).

Model M₂ also showed pronounced age differences in GS intercepts, but not in the rate of decline over time for both men and women (Figure 2). Older participants had weaker GS (intercept: b = −0.54, p < .001 for men; b = −0.35, p < .001 for women) and similar rates of decline relative to their younger counterparts.

Figure 2.

Differences in grip strength trajectories according to gender and age (2002-2009).

Socio-Psychological Variables, Cardiometabolic Risk Factors, and Trajectory of GS (Model M₃ in Table 2)

To assess whether the effects of education and baseline SRH on the GS trajectory intercept and slope are mediated by the cardiometabolic risk indicators, we first examined the correlations between these indicators at baseline, separately for men and women (Table 3). Education was negatively correlated with BF% (r = −.72, p < .05) in women and directly correlated with elevated TG (r = .105, p < .01) in men, but was not correlated with abnormal HDL-C or HbA1c. In contrast, poor (higher) SRH was correlated with reduced HDL-C (r = .101, p < .01) and elevated HbA1c (r = .075, p < .05) in women only; SRH was not correlated with BF% or elevated TG in either gender group. All four indicators of cardiometabolic risk (i.e., BF%, TG, HDL-C, and HbA1c) were significantly correlated among themselves, with the exception of the two-way correlation between HDL-C and HbA1c in men only.

Table 3.

Correlations Among Socio-Psychological and Metabolic Covariates.

	Education	SRH	Percent body fat	High TG	Low HDL-C
Men (n = 574)
SRH	−.078
Percent body fat	.046	−.068
High TG	.105*	−.009	.325**
Low HDL-C	.047	.018	.146**	.201**
High HbA1c	.053	.041	.167**	.063*	−.053
Women (n = 807)
SRH	−.041
Percent body fat	−.072*	−.021
High TG	.064	−.004	.224**
Low HDL-C	.002	.101**	.132**	.340**
High HbA1c	−.051	.075*	.156**	.140**	.109**

Note. SRH = self-rated health (range: 1 = excellent to 4 = poor); TG = triglycerides (abnormal [high] if ≥150 mg/dL; HDL-C = high-density lipoprotein cholesterol (abnormal if ≤40 mg/dL for men and ≤ 50 mg/dL for women); HbA1c = glycosylated hemoglobin A1c (abnormal if ≥6.0%).

p < .05. **p < .01 (2-tailed).

The fully adjusted model (Model M₃, Table 2), which included both the socio-psychological variables and the cardiometabolic indicators, showed the following: First, the addition of cardiometabolic indicators to the previous model (M₂) rendered the association between education and GS intercept non-significant in both men and women, suggesting that educational differences in GS intercept were fully mediated by cardiometabolic risk indicators. Second, the effects of SRH on GS intercept observed in women persisted, though they were slightly attenuated (intercept: b = −0.56, p = .029 in M₃ vs. b = −0.74, p = .010 in M₂) after adjustment for baseline and time-varying cardiometabolic indicators. Hence, cardiometabolic risk indicators do not appear to mediate the association between SRH and the level of GS in women. Third, after adjustment for age, education, and SRH (Model M₃ vs. Model M₁ in Table 2), BF% became significantly associated with a lower GS intercept in both men and women (b = −0.12, p = .008 in men; b = −0.10, p = .034 in women; Figure 3), indicating that inter-individual differences in education, age, or SRH are likely to confound the relationship between lower GS levels and higher adiposity (body fat levels) in both genders.

Figure 3.

Differences in trajectories of grip strength according to gender and percent body fat (2002-2009).

Potential for Bias Due to Selective Attrition

The intercept coefficients for attrition were consistently significant in all the models for men, but were no longer significant for women in the fully adjusted model, indicating that men (but not women) who died or dropped out of the study had lower GS levels (intercept: b = −1.29, p = .008 in M₃) compared with those who stayed in the study. In addition, women who died or dropped out of the study had a significantly faster rate of decline in GS (b = −0.19, p = .032 in M₃) compared with women who stayed in the study; conversely, men had similar rates of decline regardless of their attrition status. This result suggests that the GS trajectory intercept estimates for men and slope estimates for women would have been biased if attrition was not taken into account and confirms the need to control for non-random sources of missing values in longitudinal studies of health in older adults.

Discussion

Distinct bodies of literature have documented socio-demographic disparities in upper-extremity muscle strength (Guralnik, Butterworth, Wadsworth, & Kuh, 2006; Kuh et al., 2006) and cross-sectional associations between weaker hand-grip strength and abnormal cardiometabolic functioning (Aoyama et al., 2011; Yang et al., 2012). To our knowledge, this is the first study to investigate the longitudinal, time-varying relationship between cardiometabolic biomarkers and trajectories of upper-extremity strength from an integrated biopsychosocial perspective in a large sample of older men and women.

The inverse association between upper-extremity strength and BF% and the positive association with body weight have been previously described (Kuh, Bassey, Butterworth, Hardy, & Wadsworth, 2005; Newman et al., 2003) and have been attributed to a more favorable proportion of lean-to-fat mass in older individuals with higher weight and/or lower percentage of body fat. Our study investigates further the association between body composition and the rate of decline in GS over time and shows that higher BF% is associated with weaker GS in both men and women, but is not correlated with the decline in GS after adjustments for age, education, SRH, and the other metabolic abnormalities investigated here, because of their strong correlation with adiposity levels.

An interesting finding in this context is the lack of an association between the constellation of cardiometabolic abnormalities commonly linked with excess weight (i.e., elevated TG, high HbA1c as evidence of insulin resistance, low HDL-C) and the level or rate of decline in upper-extremity strength over time in both men and women. This result suggests that additional biological pathways linking obesity, and in particular excess body fat, to lower muscle strength in older adults need to be explored, possibly including inflammation, nutritional deficiencies, age-related neural and vascular signaling changes, and poor muscle quality due to intra-muscular fat infiltration (Rolland et al., 2008; Stenholm et al., 2011). In light of prior research showing an association between metabolic syndrome, a composite of adiposity-linked biomarkers largely overlapping with the indicators investigated in this study, and muscle strength (Atlantis et al., 2009), our negative finding raises the possibility that some but not all the components of metabolic syndrome are associated with muscle function in older adults. This hypothesis is further supported by the significant correlations among cardiometabolic indicators found in this study (Table 3). Additional research is required to investigate the possibility that one indicator, in particular body fat mass, acts as a proxy for other adiposity-related biomarkers and drives the association between metabolic functioning and muscle strength in older adults.

Prior reports on the association between educational attainment and upper-extremity strength have been inconsistent. Although some studies found no association (Mohd Hairi et al., 2010; Rautio et al., 2005), others reported significantly weaker muscle strength in individuals with lower educational attainment (Rantanen et al., 1994). In our analyses, educational differences in the level of GS in both men and women became non-significant with the inclusion of cardiometabolic indicators (in particular BF% in conjunction with body weight) in the models, thus supporting the role of adiposity or adiposity-related biological processes as mediators between educational attainment and muscle strength. Growing evidence also supports the biological plausibility of such a mediating effect; the metabolic impairments considered here have shown a graded relationship with educational level (Lee, Jung, Park, Rhee, & Kim, 2005; Seeman et al., 2008; Silventoinen, Pankow, Jousilahti, Hu, & Tuomilehto, 2005) and were, either individually or in combination, important determinants of muscle strength in older adults (Miyatake et al., 2007; Sayer et al., 2007).

SRH assessments form as a result of psychological processing of bodily sensations and feelings, which in turn may be a reflection of pre-clinical or clinically manifest biological processes leading to morbidity and mortality. Hence, SRH is a sensitive barometer of physiological health, potentially reflecting pre-clinical morbidity prior to the diagnosis of a disease (Jylhä, 2009). As such, SRH can add an important dimension to our understanding of the processes influencing the trajectory of objectively measured functional capabilities in older adults. Although SRH is a subjective assessment of health status, it has shown a graded relationship with multiple metabolic and inflammatory biomarkers, including total and HDL cholesterol, albumin, white blood cell count, C-reactive protein, hemoglobin, and creatinine (Dowd & Zajacova, 2010; Jylhä, Volpato, & Guralnik, 2006). Research on the relationship between SRH and muscle strength, in particular upper-extremity strength, is extremely limited. Our study shows that women, but not men, who rate their overall health as poor at baseline have weaker GS, but do not differ in the rate of decline in GS from those with better self-assessments of health. Unlike educational differences, the SRH differences in GS level in women are not mediated by metabolic functioning. This finding is consistent with a vast literature documenting that SRH is a powerful independent predictor of mortality, disability, and healthcare utilization (Idler et al., 2000). Our research provides evidence that, in women, SRH exerts a similar independent effect on GS, an objective measurement of physical functioning, and suggests that the interplay between muscle strength, biological mediators, and psychological or subjective assessments of health vary according to gender.

In our study, variations in SRH over time were not correlated with concurrent changes in upper-extremity strength. We are unable to fully explain this finding, except by noting that recent evidence indicates that older people adapt to their worsening health conditions and tend to report their health as similar or even better with increasing age, despite an increase in the number and severity of chronic diseases and declines in functional status (Leinonen, Heikkinen, & Jylhä, 2001). In addition, individual psychological resources, such as self-mastery, self-esteem, and ability to handle stress, contribute to better health ratings even among older adults with chronic conditions and disability (Cott, Gignac, & Badley, 1999). Further research is needed to understand how changes in muscle strength and function translate into objective or perceived functional impairments, possibly differentially in women and men, and affect the older adults’ assessments of their general health condition.

Finally, this research sheds light on gender differences in trajectories of upper-extremity strength in old age. In all the models (M₀ to M₃), women had a GS roughly 40% weaker than men (intercept: 18.4 vs. 31.1 in men; M₃, Table 2), but a slower rate of decline in GS (slope: −0.27 vs. −0.41 in men) relative to men, indicating that despite a substantial female disadvantage in initial upper-extremity strength, women appeared to better preserve this function over time. In additional analyses (not shown, available upon request), we tested the same sequence of models on the full (non-stratified) sample and included gender as a time-constant (level 2) covariate. This allowed us to directly estimate the gender coefficient corresponding to the difference in intercept and slope between men and women. These additional analyses showed that although heterogeneity in socio-psychological factors and cardiometabolic biomarkers did not account for the difference in the level of GS between men and women, they fully explained the difference in the rate of decline in strength over time. Consistent with prior observations (Frederiksen et al., 2006; Kuh et al., 2006), these findings suggest that optimal control of cardiometabolic risk may delay the decline in GS and potentially upper-body muscular strength in both genders, but will likely not reduce the female disadvantage in upper-body strength in late life. The mechanisms underlying gender differences in upper-extremity strength have been summarized elsewhere (Cooper et al., 2011). Our analyses addressed several of these factors (e.g., differences in adiposity levels, body weight, health risk behaviors) but others, such as differences in hormonal exposures, neurological and cardiovascular fitness, or inflammatory status, were not fully investigated. Future studies are needed to clarify how these factors affect the gender differences in trajectories of upper-extremity muscle strength.

Some limitations of this study should be noted. First, consistent with the harmonized definition of metabolic syndrome and its components (Alberti et al., 2009), cardiometabolic indicators were included as dichotomous variables (normal vs. abnormal). The current recommended cut-off values for the metabolic impairments considered here are not population-specific. However, Asians display a higher proportion of body fat compared with people of European descent with similar BMI, as well as a higher predisposition to insulin resistance at a lesser degree of obesity (Yoon et al., 2006). As such, the lack of an association between selected cardiometabolic impairments and the trajectory of GS may be a measurement artifact. To evaluate this hypothesis, we performed additional analyses with the cardiometabolic indicators coded as either continuous or categorical (quartiles) variables and obtained similar results (not shown, available upon request). Nevertheless, the possibility that other metabolic indicators (e.g., adiponectin or leptin) may be involved in processes underlying the association between excess adiposity and muscle strength in the Japanese population remains to be investigated. Second, because our participants were 65 years of age or older and relatively healthy, the generalizability of these results to younger and/or less healthy populations remains to be established. Third, because of data constraints, the trajectories of GS were modeled as a function of time rather than age (Alwin, Hofer, & McCammon, 2006). Although we controlled for age-at-baseline differences to minimize the confounding between age and cohort, the results should not be interpreted to reflect the “true” effect of aging on upper-body strength, but the effect of time in this specific cohort. Lastly, our analyses were not adjusted for differences in intensity or frequency of physical activity, important correlates of muscle strength, because this information has not been routinely collected.

In conclusion, this study provides new insights on how cardiometabolic risk and socio-psychological characteristics jointly influence the decline in hand-grip strength in older Japanese, by showing that the educational gradient in GS trajectories is mediated by differences in body composition, whereas SRH is a robust independent predictor of upper-extremity strength in women. In addition, our results support the utility of integrated models, which include socio-psychological individual characteristics in combination with appropriate measures of biological status, when analyzing the natural course of muscle strength and physical performance in older adults.

Footnotes

Acknowledgements

The authors thank the funding institutions for their generous support.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by Grants T32-AG027708 and T32-AG019134 (to A.B.) and R01-AG154124 and R01-AG028116 (to J.L.) from the National Institute on Aging at the National Institutes of Health. The Japanese Ministry of Health, Labor and Welfare Longevity Foundation, the Tokyo Metropolitan Institute of Gerontology, and the Michigan Claude D. Pepper Older Americans Independence Center (P60-AG08808) provided additional support (to J.L.).

References

Alberti

Eckel

R. H.

Grundy

S. M.

Zimmet

P. Z.

Cleeman

J. I.

Donato

K. A.

. . . Smith

S. C.

(2009). Harmonizing the metabolic syndrome: A joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation, 120, 1640-1645.

Alwin

D. F.

Hofer

S. M.

McCammon

R. J.

(2006). Modeling the effects of time: Integrating demographic and developmental perspectives. In Binstock

R. H.

George

L. K.

(Eds.), Handbook of aging and the social sciences (6th ed., pp. 20-38). San Diego, CA: Elsevier.

Aoyama

Asaka

Ishijima

Kawano

Cao

Z. B.

Sakamoto

. . .Higuchi

(2011). Association between muscular strength and metabolic risk in Japanese women, but not in men. Journal of Physiological Anthropology, 30(4), 133-139.

Atlantis

Martin

S. A.

Haren

M. T.

Taylor

A. W.

Wittert

G. A.

(2009). Inverse associations between muscle mass, strength, and the metabolic syndrome. Metabolism, 58, 1013-1022.

Baron

R. M.

Kenny

D. A.

(1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182.

Bonora

Kiechl

Mayr

Zoppini

Targher

Bonadonna

R. C.

Willeit

(2011). High-normal HbA1c is a strong predictor of type 2 diabetes in the general population. Diabetes Care, 34, 1038-1040.

Braveman

P. A.

Cubbin

Egerter

Chideya

Marchi

K. S.

Metzler

Posner

(2005). Socioeconomic status in health research. Journal of American Medical Association, 294, 2879-2888.

Cooper

Hardy

Sayer

A. A.

Ben-Shlomo

Birnie

Cooper

. . . Kuh

(2011). Age and gender differences in physical capability levels from mid-life onwards: The harmonisation and meta-analysis of data from eight UK cohort studies. PLoS ONE, 6, e27899.

Cott

C. A.

Gignac

M. A.

Badley

E. M.

(1999). Determinants of self-rated health for Canadians with chronic disease and disability. Journal of Epidemiology & Community Health, 53, 731-736.

10.

Crimmins

E. M.

Seeman

T. E.

(2004). Integrating biology into the study of health disparities. Population and Development Review, 30, 89-107.

11.

Dowd

J. B.

Zajacova

(2010). Does self-rated health mean the same thing across socioeconomic groups? Evidence from biomarker data. Annals of Epidemiology, 20, 743-749.

12.

Everson-Rose

S. A.

Paudel

Taylor

B. C.

Dam

Cawthon

P. M.

LeBlanc

. . .Ensrud

K. E.

(2011). Metabolic syndrome and physical performance in elderly men: The Osteoporotic Fractures in Men Study. Journal of American Geriatrics Society, 59, 1376-1384.

13.

Ferrucci

Guralnik

J. M.

Buchner

Kasper

Lamb

S. E.

Simonsick

E. M.

. . . Fried

L. P.

(1997). Departures from linearity in the relationship between measures of muscular strength and physical performance of the lower extremities: The Women’s Health and Aging Study. Journals of Gerontology: Medical Sciences, 52, 275-285.

14.

Flegal

K. M.

Carroll

M. D.

Ogden

C. L.

Curtin

L. R.

(2010). Prevalence and trends in obesity among US adults, 1999-2008. Journal of American Medical Association, 303, 235-241.

15.

Frederiksen

Hjelmborg

Mortensen

Mcgue

Vaupel

J. W.

Christensen

(2006). Age trajectories of grip strength: Cross-sectional and longitudinal data among 8,342 Danes aged 46 to 102. Annals of Epidemiology, 16, 554-562.

16.

Fujiwara

Chaves

P. H. M.

Takahashi

Amano

Yoshida

Kumagai

. . . Shinkai

(2005). Arterial pulse wave velocity as a marker of poor cognitive function in an elderly community-dwelling population. Journals of Gerontology: Medical Sciences, 60, 607-612.

17.

Gale

C. R.

Martyn

C. N.

Cooper

Sayer

A. A.

(2007). Grip strength, body composition, and mortality. International Journal of Epidemiology, 36, 228-235.

18.

Guralnik

J. M.

Butterworth

Wadsworth

M. E.

Kuh

(2006). Childhood socioeconomic status predicts physical functioning a half century later. Journals of Gerontology: Medical Sciences, 61, 694-701.

19.

Hedeker

D. R.

Gibbons

R. D.

(2006). Longitudinal data analysis. Hoboken, NJ: John Wiley.

20.

Howard

Ferrucci

Sun

Fried

L. P.

Walston

Varadhan

. . . Semba

R. D.

(2007). Oxidative protein damage is associated with poor grip strength among older women living in the community. Journal of Applied Physiology, 103, 17-20.

21.

Hughes

V. A.

Frontera

W. R.

Wood

Evans

W. J.

Dallal

G. E.

Roubenoff

Fiatarone Singh

M. A.

(2001). Longitudinal muscle strength changes in older adults. Journals of Gerontology: Medical Sciences, 56, 209-217.

22.

Idler

E. L.

Benyamini

(1997). Self-rated health and mortality: A review of twenty-seven community studies. Journal of Health and Social Behavior, 38, 21-37.

23.

Idler

E. L.

Kasl

S. V.

(1995). Self-ratings of health: Do they also predict change in functional ability? Journals of Gerontology: Psychological Sciences, 50, 344-353.

24.

Idler

E. L.

Russell

L. B.

Davis

(2000). Survival, functional limitations, and self-rated health in the NHANES I epidemiologic follow-up study. American Journal of Epidemiology, 152, 874-883.

25.

Innes

(2002). Handgrip strength testing: A review of the literature. Australian Occupational Therapy Journal, 46, 120-140.

26.

Janiszewski

P. M.

Janssen

Ross

(2007). Does waist circumference predict diabetes and cardiovascular disease beyond commonly evaluated cardiometabolic risk factors? Diabetes Care, 30, 3105-3109.

27.

Jylhä

(2009). What is self-rated health and why does it predict mortality? Towards a unified conceptual model. Social Science & Medicine, 69, 307-316.

28.

Jylhä

Volpato

Guralnik

J. M.

(2006). Self-rated health showed a graded association with frequently used biomarkers in a large population sample. Journal of Clinical Epidemiology, 59, 465-471.

29.

Klein

Allison

D. B.

Heymsfield

S. B.

Kelley

D. E.

Leibel

R. L.

Nonas

Kahn

(2012). Waist circumference and cardiometabolic risk: A consensus statement from Shaping America’s Health: Association for Weight Management and Obesity Prevention; NAASO, the Obesity Society; the American Society for Nutrition; and the American Diabetes Association. Obesity, 15, 1061-1067.

30.

Kuh

Bassey

E. J.

Butterworth

Hardy

Wadsworth

M. E. J.

(2005). Grip strength, postural control, and functional leg power in a representative cohort of British men and women: Associations with physical activity, health status, and socioeconomic conditions. Journals of Gerontology: Medical Sciences, 60, 224-231.

31.

Kuh

Hardy

Butterworth

Okell

Wadsworth

Cooper

Aihie Sayer

(2006). Developmental origins of midlife grip strength: Findings from a birth cohort study. Journals of Gerontology: Medical Sciences, 61, 702-706.

32.

Lee

W. Y.

Jung

C. H.

Park

J. S.

Rhee

E. J.

Kim

S. W.

(2005). Effects of smoking, alcohol, exercise, education, and family history on the metabolic syndrome as defined by the ATP III. Diabetes Research and Clinical Practice, 67(1), 70-77.

33.

Leinonen

Heikkinen

Jylhä

(2001). Predictors of decline in self-assessments of health among older people—A 5-year longitudinal study. Social Science & Medicine, 52, 1329-1341.

34.

Liang

Bennett

J. M.

Shaw

B. A.

Quinones

A. R.

Ofstedal

M. B.

(2008). Gender differences in functional status in middle and older age: Are there any age variations? Journals of Gerontology: Social Sciences, 63, 282-292.

35.

Liang

Quinones

A. R.

Bennett

J. M.

Shaw

B. A.

Ofstedal

M. B.

(2010). Evolving self-rated health in middle and old age: How does it differ across black, Hispanic, and white Americans? Journal of Aging and Health, 22, 3-26.

36.

Liang

Shaw

B. A.

Bennett

J. M.

Krause

Kobayashi

Fukaya

Sugihara

(2007). Intertwining courses of functional status and subjective health among older Japanese. Journals of Gerontology: Social Sciences, 62, 340-348.

37.

Little

R. J. A.

Rubin

D. B.

(2002). Statistical analysis with missing data (2nd ed.). Hoboken, NJ: Wiley.

38.

Metter

E. J.

Conwit

Tobin

Fozard

J. L.

(1997). Age-associated loss of power and strength in the upper extremities in women and men. Journals of Gerontology: Medical Sciences, 52A, 267-276.

39.

Miyatake

Wada

Saito

Nishikawa

Matsumoto

Miyachi

Numata

(2007). Comparison of muscle strength between Japanese men with and without metabolic syndrome. Acta Medica Okayama, 61, 99-102.

40.

Mohd Hairi

Mackenbach

J. P.

Andersen-Ranberg

Avendano

(2010). Does socio-economic status predict grip strength in older Europeans? Results from the SHARE study in non-institutionalised men and women aged 50+. Journal of Epidemiology & Community Health, 64, 829-837.

41.

Molarius

Berglund

Eriksson

Lambe

Nordström

Eriksson

H. G.

Feldman

(2007). Socioeconomic conditions, lifestyle factors, and self-rated health among men and women in Sweden. European Journal of Public Health, 17, 125-133.

42.

Newman

A. B.

Haggerty

C. L.

Goodpaster

Harris

Kritchevsky

Nevitt

Visser

(2003). Strength and muscle quality in a well-functioning cohort of older adults: The Health, Aging and Body Composition Study. Journal of American Geriatrics Society, 51, 323-330.

43.

Quiñones

A. R.

Liang

Bennett

J. M.

(2011). How does the trajectory of multimorbidity vary across black, white, and Mexican Americans in middle and old age? Journals of Gerontology: Social Sciences, 66, 739-749.

44.

Rantanen

Era

Kauppinen

Heikkinen

(1994). Maximal isometric muscle strength and socioeconomic status, health, and physical activity in 75-year-old persons. Journal of Aging and Physical Activity, 2, 206-220.

45.

Rantanen

Guralnik

J. M.

Foley

Masaki

Leveille

Curb

J. D.

White

(1999). Midlife hand grip strength as a predictor of old age disability. Journal of American Medical Association, 281, 558-560.

46.

Rantanen

Masaki

Foley

Izmirlian

White

Guralnik

J. M.

(1998). Grip strength changes over 27 years in Japanese-American men. Journal of Applied Physiology, 85, 2047-2053.

47.

Rantanen

Volpato

Luigi Ferrucci

Eino Heikkinen

Fried

L. P.

Guralnik

J. M.

(2003). Handgrip strength and cause-specific and total mortality in older disabled women: Exploring the mechanism. Journal of American Geriatrics Society, 51, 636-641.

48.

Raudenbush

S. W.

Bryk

A. S.

(2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Thousand Oaks, CA: Sage.

49.

Rautio

Heikkinen

Ebrahim

(2005). Socio-economic position and its relationship to physical capacity among elderly people living in Jyväskylä, Finland: Five- and ten-year follow-up studies. Social Science & Medicine, 60, 2405-2416.

50.

Rolland

Czerwinski

Abellan Van Kan

Morley

J. E.

Cesari

Onder

. . .Vellas

(2008). Sarcopenia: Its assessment, etiology, pathogenesis, consequences and future perspectives. Journal of Nutrition, Health & Aging, 12, 433-450.

51.

Sasaki

Kasagi

Yamada

Fujita

(2007). Grip strength predicts cause-specific mortality in middle-aged and elderly persons. American Journal of Medicine, 120, 337-342.

52.

Sayer

A. A.

Syddall

H. E.

Dennison

E. M.

Martin

H. J.

Phillips

D. I. W.

Cooper

Byrne

C. D.

(2007). Grip strength and the metabolic syndrome: Findings from the Hertfordshire Cohort Study. QJM, 100, 707-713.

53.

Schafer

J. L.

Olsen

M. K.

(1998). Multiple imputation for multivariate missing-data problems: A data analyst’s perspective. Multivariate Behavioral Research, 33, 545-571.

54.

Seeman

Merkin

S. S.

Crimmins

Koretz

Charette

Karlamangla

(2008). Education, income and ethnic differences in cumulative biological risk profiles in a national sample of US adults: NHANES III (1988–1994). Social Science & Medicine, 66, 72-87.

55.

Selvin

Steffes

M. W.

Gregg

Brancati

F. L.

Coresh

(2011). Performance of A1C for the classification and prediction of diabetes. Diabetes Care, 34, 84-89.

56.

Silventoinen

Pankow

Jousilahti

Tuomilehto

(2005). Educational inequalities in the metabolic syndrome and coronary heart disease among middle-aged men and women. International Journal of Epidemiology, 34, 327-334.

57.

Sirola

Rikkonen

Kröger

Honkanen

Tuppurainen

Airaksinen

Saarikoski

(2004). Factors related to postmenopausal muscle performance: A cross-sectional population-based study. European Journal of Applied Physiology, 93, 102-107.

58.

Stenholm

Alley

Bandinelli

Griswold

Koskinen

Rantanen

. . . Ferrucci

(2009). The effect of obesity combined with low muscle strength on decline in mobility in older persons: Results from the InCHIANTI study. International Journal of Obesity, 33, 635-644.

59.

Stenholm

Sallinen

Koster

Rantanen

Sainio

Heliövaara

Koskinen

(2011). Association between obesity history and hand grip strength in older adults—Exploring the roles of inflammation and insulin resistance as mediating factors. Journals of Gerontology: Medical Sciences, 66, 341-348.

60.

Strand

B. H.

Cooper

Hardy

Kuh

Guralnik

(2011). Lifelong socioeconomic position and physical performance in midlife: Results from the British 1946 birth cohort. European Journal of Epidemiology, 26, 475-483.

61.

Syddall

Cooper

Martin

Briggs

Aihie Sayer

(2003). Is grip strength a useful single marker of frailty? Age and Ageing, 32, 650-656.

62.

Syddall

Evandrou

Cooper

Aihie Sayer

(2009). Social inequalities in grip strength, physical function, and falls among community dwelling older men and women. Journal of Aging and Health, 21, 913-939.

63.

Tong

Fung

(2003). The congruence of two different height-weight ratios in estimating obesity prevalence: A study with Hong Kong children age 6 to 12. Journal of Exercise Science and Fitness, 1, 133-135.

64.

Weykamp

John

W. G.

Mosca

Hoshino

Little

Jeppsson

J. O.

. . . Siebelder

(2008). The IFCC reference measurement system for HbA1c: A 6-year progress report. Clinical Chemistry, 54, 240-248.

65.

Wildman

R. P.

Muntner

Reynolds

McGinn

A. P.

Rajpathak

Wylie-Rosett

Sowers

M. R.

(2008). The obese without cardiometabolic risk factor clustering and the normal weight with cardiometabolic risk factor clustering: Prevalence and correlates of 2 phenotypes among the US population (NHANES 1999-2004). Archives of Internal Medicine, 168, 1617-1624.

66.

Winkleby

M. A.

Jatulis

D. E.

Frank

Fortmann

S. P.

(1992). Socioeconomic status and health: How education, income, and occupation contribute to risk factors for cardiovascular disease. American Journal of Public Health, 82, 816-820.

67.

Yang

E. J.

Lim

Kim

K. W.

Jang

H. C.

Paik

(2012). Association between muscle strength and metabolic syndrome in older Korean men and women: The Korean Longitudinal Study on Health and Aging. Metabolism, 61, 317-324.

68.

Yoon

K. H.

Lee

J. H.

Kim

J. W.

Cho

J. H.

Choi

Y. H.

S. H.

. . . Son

H. Y.

(2006). Epidemic obesity and type 2 diabetes in Asia. Lancet, 368, 1681-1688.

69.

Zoico

Di Francesco

Guralnik

J. M.

Mazzali

Bortolani

Guariento

. . . Zamboni

(2004). Physical disability and muscular strength in relation to obesity and different body composition indexes in a sample of healthy elderly women. International Journal of Obesity, 28, 234-241.

Cardiometabolic Risk,Socio-Psychological Factors,and Trajectory of Grip Strength Among Older Japanese Adults

Abstract

Keywords

Introduction

Design and Method

Study Design and Sample

Measures

Hand-grip strength

Cardiometabolic risk

Socio-psychological and control variables

Statistical Analysis

Model specification

Treatment of missing data

Results

Cardiometabolic Risk Factors and GS Trajectory (Model M1 in Table 2)

Education, Self-Rated Health, and GS Trajectory (Model M2 in Table 2)

Socio-Psychological Variables, Cardiometabolic Risk Factors, and Trajectory of GS (Model M3 in Table 2)

Potential for Bias Due to Selective Attrition

Discussion

Footnotes

Acknowledgements

Declaration of Conflicting Interests

Funding

References

Cardiometabolic Risk Factors and GS Trajectory (Model M₁ in Table 2)

Education, Self-Rated Health, and GS Trajectory (Model M₂ in Table 2)

Socio-Psychological Variables, Cardiometabolic Risk Factors, and Trajectory of GS (Model M₃ in Table 2)