Is grip strength a good marker of physical performance among community-dwelling older people?

Abstract

Introduction

There is increasing interest in physical performance as it relates to both the current and future health of older people. It is often characterised using the Short Physical Performance Battery including assessment of gait speed, chair rises and standing balance. However this battery of tests may not be feasible in all clinical settings and simpler measures may be required. As muscle strength is central to physical performance, we explored whether grip strength could be used as a marker of the Short Physical Performance Battery.

Objective

To examine associations between grip strength and components of the Short Physical Performance Battery in older community dwelling men and women.

Methods

Grip strength measurement and the Short Physical Performance Battery were completed in 349 men and 280 women aged 63–73 years taking part in the Hertfordshire Cohort Study (HCS). Relationships between grip strength and physical performance (6m timed-up-and-go [TUG], 3m walk, chair rises and standing balance times) were analysed using linear and logistic regression, without and with adjustment for age, anthropometry, lifestyle factors and co-morbidities.

Results

Among men, a kilo increase in grip strength was associated with a 0.07s (second) decrease in 6m TUG, a 0.02s decrease in 3m walk time, and a 1% decrease in chair rises time (p<0.001 for all). Among women, a kilo increase in grip strength was associated with a 0.13s decrease in 6m TUG, a 0.03s decrease in 3m walk time, and a 1% decrease in chair rises time (p<0.001). Higher grip strength was associated with better balance among men (p=0.01) but not women (p=0.57). Adjustment for age, anthropometry, lifestyle and co-morbidities did not alter these results.

Conclusions

Grip strength is a good marker of physical performance in this age group and may be more feasible than completing a short physical performance battery in some clinical settings.

Introduction

There is increasing interest in physical performance as it relates to both the current and future health of older people (1). Physical performance is often characterised using the Short Physical Performance Battery (SPPB) which includes assessment of gait speed, chair rises and standing balance (2) and which has been effectively utilised in a range of research settings. Studenski (3) and colleagues have also suggested that the SPPB was suitable for use in clinical settings in the United States but older people admitted acutely to hospital in the United Kingdom characteristically have mobility impairment as well as multiple co-morbidities (4). Use of the SPPB in this context could potentially be very challenging and simpler measures of physical performance may be required. As muscle strength is central to physical performance, we explored whether grip strength could be used as a marker of the Short Physical Performance Battery.

The relationship between muscle strength and physical performance has been studied previously. However, not all studies have considered grip strength as their marker of muscle strength, or have measured physical performance directly, and few have considered whether the association between muscle strength and physical performance is linear or curvilinear. Sirola et al studied 1,166 post-menopausal women who participated in the Kuopio Osteoporosis Risk Factor and Prevention (OSTPRE) study and demonstrated that the women who had better performance on balance and squat tests had greater grip and quadriceps strength (as measured using strain gauge dynamometers) (5). Davis et al studied 705 community dwelling older Japanese women resident in Hawaii (mean age 74) and demonstrated associations between greater quadriceps, triceps and grip strength and improved physical performance across a panel of measures comprising walking speed, the get up and go test, chair stands, functional reach, and hand and foot reaction times (6). Visser et al demonstrated that low leg muscle strength was associated with poorer performance on a repeated chair-stands test among men and women, aged 70 to 79 years, who participated in the Health, Aging and Body Composition Study (7). Baker et al have shown that performance of a leg cycle ergometry test is influenced by a muscular contribution from the upper body and by upper body strength(8). Stenholm studied 2,208 subjects aged 55 years and older and demonstrated associations between reduced grip strength and walking limitation (difficulty walking 0.5km or maximum walking speed <1.2m/s) (9). Hughes et al studied 485 men and women aged sixty years and older and showed that lower grip strength at baseline was associated with decline in a timed manual performance test across a two year follow-up period (10).

Other studies in the literature have assessed physical performance indirectly by considering self-reported markers of customary physical activity or difficulties in activities of daily living. Glenmark et al studied 55 men and 26 women at 16 and 27 years of age and found a positive correlation between hand grip strength and physical activity during leisure time among women at both ages but no association among men at either age (11). Bassey et al studied longitudinal changes in grip strength and physical activity across an eight year follow-up period among 350 men and women aged 65 years and older and found that reduced self-reported use of the handgrip muscles was associated with greater reductions in hand grip strength (12). Martin et al found that higher muscle strength and physical performance were associated with higher levels of physical activity among older women but not men but did not examine the associations between grip strength and physical performance (13).

The relationship between muscle strength and mass is linear (14). The relationship between muscle mass and physical performance is curvilinear, whereby losing muscle mass to a certain point will not affect physical performance, but once a threshold is reached, physical performance declines (15, 16). The relationship between muscle strength and physical performance may therefore be of the same nature as the relationship between physical performance and muscle mass and grip strength may be a good marker of physical performance.

The objective of this study was to examine the nature of the relationship between grip strength and directly measured markers of physical performance among older community dwelling men and women who participated in the Hertfordshire Cohort Study (HCS)(17).

Methods

Study population

The HCS has been described in detail previously (17). In brief, in 1998, a total of 3,822 men and 3,284 women born in Hertfordshire between 1931 and 1939 and still living in the county were traced with the aid of the NHS central registry in Southport and confirmed as currently registered with a general practitioner in Hertfordshire. Permission to contact 3,126 (82%) men and 2,973 (91%) women was obtained from their general practitioners. 1,684 (54%) men and 1,541 (52%) women agreed to take part in a home interview where trained nurses collected information including self-reported walking speed, smoking history, alcohol intake, social class, medical history (including the Rose chest pain questionnaire, existing diagnosis of diabetes, history of cerebrovascular disease, symptoms of bronchitis as well as falls history) and self-assessed quality of life using the SF-36 questionnaire. 1,579 (94%) of these men, and 1,418 (92%) of these women subsequently attended a clinic for a number of investigations. Height was measured to the nearest 0.1 cm using a Harpenden pocket stadiometer (Chasmors Ltd, London, UK) and weight to the nearest 0.1 kg on a seca floor scale (Chasmors Ltd). Grip strength was measured to the nearest 1kg three times on each side, alternating between right and left hands, for 1,572 (99.6%) of the men and 1,415 (99.8%) of the women using a Jamar handgrip dynamometer (Promedics, Blackburn, UK). Participants were given standardised encouragement to squeeze the dynamometer as hard as possible (18).

Assessment of co-morbidity was as follows. Participants who had not reported an existing diagnosis of diabetes attended the morning clinics after fasting overnight and completed a 2 hour oral glucose tolerance test (OGTT) using 75 g anhydrous glucose; diabetes mellitus and impaired glucose tolerance were classified according to WHO criteria. Blood pressure was recorded as the mean of three measurements taken with a Dinamap Model 8101 (GE Medical Systems, Slough, UK) after the subject had been seated for 5 minutes. An ECG was also performed, and graded for ischaemic changes, according to the Minnesota protocol. Clinical examination was used to assess presence of hand osteoarthritis.

Physical performance was tested at clinic in the West Hertfordshire phase of the fieldwork (349 men and 280 women) using an adaptation of Guralnik’s (2) short physical performance battery (SPPB) of tests. This consisted of a 6 metre timed-up-and-go (TUG), a 3 metre walk from a standing start, 5 chair rises, and a one-legged balance test (flamingo stand) for up to 30 seconds. Standard instructions and encouragement were given for each part of the test. Participants were allowed to use walking aids if necessary, and could stop at any point during a test if they felt unable to complete it. Intra- and inter-observer studies were carried out at regular intervals during the fieldwork to ensure comparability of measurements within and between observers. The study received ethical approval from the Hertfordshire and Bedfordshire Local Research Ethics Committee, and all subjects gave written informed consent.

Statistical methods

The best of the six grip strength measurements was used to characterise muscle strength. The time taken to complete five chair rises was loge transformed to a normal distribution for analysis. The standing-balance time variable was highly bimodal, with participants either losing balance very quickly or remaining balanced for the full 30 seconds. This was therefore coded to a binary variable with a ‘good balance’ group (>5 seconds) contrasted with a ‘poor balance’ group <5 seconds). Only two men and four women used a walking aid to complete the timed up and go, 3m walk or chair rises tests and only two men failed to complete all five chair rises; the times obtained for the tests by these individuals were excluded from the relevant analyses.

Height and weight were highly correlated (r = 0.45, P<0.001 for men; r=0.28, P<0.001 for women). A standardized residual of weight adjusted for height was therefore coded as a marker of weight which could be included simultaneously in a regression model with height without concern about multicollinearity problems.

Participant characteristics were described using means and standard deviations, medians and inter-quartile ranges and frequency and percentage distributions. The relationships between grip strength and physical performance were explored as follows. Firstly, the unadjusted associations between grip strength and each component of the physical performance battery in turn were explored using linear regression (for the 6m timed up and go, 3m walk and loge transformed chair rises time) and logistic regression (for standing balance). Unadjusted linear, quadratic and inverse models for grip strength as a predictor of each component of the physical performance battery were fitted to examine whether there was evidence that a curvilinear model provided a better explanation of the relationship between grip strength and physical performance than a simple linear model. Secondly, the associations between grip strength and physical performance were conducted with adjustment for the potential confounding influences of age, height, weight-for-height, lifestyle factors (smoking, alcohol intake, current social class) and co-morbidities.

Grip strength was principally analysed as a continuously distributed variable, but was also classified into sex-specific quartiles for presentational purposes only. All analyses were conducted for men and women separately using the Stata 10 statistical software package (19).

Results

Summary characteristics of the study participants are shown in table 1. The mean age at clinic was 67.8 for men and 68.1 for women. Men had higher average grip strength (44.3kg) than women (26.0kg) (p<0.0001) and obtained slightly faster times than women for the 6m timed up and go(10.6s vs 11.2s, p=0.0001), 3m walk (3.3s vs 3.5s, p=0.0001) and chair rises tests (15.2s vs 16.7s, p<0.0001). A lower proportion of men than women lost their balance in less than five seconds (17.3% vs 24.3%, p=0.03).

Table 1.

Summary characteristics of study participants

Mean (SD)	Men (n=349)	Women (n=280)
Age (years)	67.8 (2.5)	68.1 (2.5)
Height (cm)	174.1 (6.6)	160.9 (6.0)
Weight (kg)	82.1 (12.5)	71.8 (13.5)
Grip strength (kg)	44.3 (7.6)	26.0 (5.8)
Short physical performance battery
6m timed up and go (secs)	10.6 (1.9)	11.2 (2.4)
3m walk (secs)	3.3 (0.5)	3.5 (0.7)
Time to complete 5 chair rises (secs)a	15.2 (1.2)	16.7 (1.3)
Balance lost in <5secs (N, %)b	58 (17.3)	66 (24.3)
Lifestyle factors, N(%)
Ex-smoker	186 (53.3)	69 (24.7)
Current smoker	40 (11.5)	25 (9.0)
Moderate or higher weekly alcohol intakec	286 (82.2)	150 (53.6)
Non-manual social class	202 (58.1)	169 (60.4)
Co-morbiditiesd N(%)
Diabetes	62 (17.8)	38 (13.6)
Hypertension	152 (43.6)	123 (44.1)
Ischaemic heart disease	63 (18.3)	29 (10.4)
Cerebrovascular disease	26 (7.5)	7 (2.5)
Bronchitis	14 (4.0)	11 (3.9)
Hand osteoarthritis	118 (34.5)	185 (66.6)

SD=standard deviation; secs=seconds; N=number; %=percentage; m=metre. a. Variable was loge transformed. Gometric means and standard deviations therefore presented. b. Number and percentage of people losing balance within the first 5 seconds. c. Defined as weekly alcohol consumption of >11 units for men and >8 units for women. d. Comorbidities defined as follows: Diabetes was identified from previously known cases or those newly diagnosed from the OGTT. Hypertension was defined as a measured systolic pressure greater than or equal to 160mmHg or diastolic pressure greater than or equal to 100mmHg or taking an anti-hypertensive medication. Participants were defined as having ischemic heart disease (IHD) if major-q waves were present on the ECG, or if they had typical angina according to the Rose chest pain questionnaire or had previously undergone a coronary artery bypass graft or angioplasty. A history of cerebrovasular disease was identified if the participant reported having ever been diagnosed by a doctor with a stroke or transient ischaemic attack. Bronchitis was defined by a self-report of a productive cough on most days for more than three months of the year and hand osteoarthritis was identified by presence of Heberden’s or Bouchard’s nodes, or squaring at the thumb base, upon clinical examination. Missing data: Two, four, four, sixteen and thirteen men had missing data for grip strength, 6m up and go, 3m walk, chair rises, and balance time respectively. One, four, four, eleven and eight women had missing data for grip strength, 6m up and go, 3m walk, chair rises, and balance time respectively.

The associations between grip strength and physical performance are shown in table 2 for men and women, for models including linear, quadratic or inverse terms for grip strength (to assess curvilinear relationships), and with and without adjustment for age, anthropometry, lifestyle and comorbidities. The principal unadjusted linear models showed that grip strength was significantly and inversely associated with 6m timed up and go, 3m walk and chair rises times among both men and women (p<0.001 for all unadjusted associations). Among men, a kilo increase in grip strength was associated with a 0.07 second decrease in 6m timed up and go, a 0.02 second decrease in 3m walk time, and a 1% decrease in chair rises time. Among women, a kilo increase in grip strength was associated with a 0.13 second decrease in 6m timed up and go, a 0.03 second decrease in 3m walk time, and a 1% decrease in chair rises time. Higher grip strength was associated with decreased odds of losing balance in the first five seconds among men (odds ratio 0.95, p=0.01) but not among women (odds ratio 1.02, p=0.57). Adjustment for age, anthropometry, lifestyle and co-morbidities did not substantively alter these results.

Table 2.

Associations between grip strength and physical performance

	betaa	MEN	(95%CI)	p	betaa	WOMEN (95%CI)	p
6m timed up and go (secs)
Unadjusted
Grip strength	−0.07		(−0.10,-0.05)	<0.001	−0.13	(−0.18,-0.08)	<0.001
Quadratic of grip strengthb				0.33			0.04
Inverse of grip strengthc				0.14			0.19
Adjusted
Grip strength	−0.06		(−0.08,-0.03)	<0.001	−0.13	(−0.18,-0.08)	<0.001
Quadratic of grip strengthb				0.96			0.07
Inverse of grip strengthc				0.64			0.26
3m walk (secs)
Unadjusted
Grip strength	−0.02		(−0.03,-0.01)	<0.001	−0.03	(−0.04,-0.02)	<0.001
Quadratic of grip strengthb				0.63			0.05
Inverse of grip strengthc				0.35			0.11
Adjusted
Grip strength	−0.02		(−0.03,-0.01)	<0.001	−0.03	(−0.05,-0.02)	<0.001
Quadratic of grip strengthb				0.83			0.09
Inverse of grip strengthc				0.77			0.13
Time to complete 5 chair rises (secs)
Unadjusted
Grip strength	0.99		(0.99,1.00)	<0.001	0.99	(0.98,0.99)	<0.001
Quadratic of grip strengthb				0.11			0.28
Inverse of grip strengthc				0.10			0.69
Adjusted
Grip strength	0.99		(0.99,1.00)	<0.001	0.98	(0.98,0.99)	<0.001
Quadratic of grip strengthb				0.15			0.61
Inverse of grip strengthc				0.21			0.29
Balance lost in <5 secs
Unadjusted
Grip strength	0.95		(0.92,0.99)	0.01	1.02	(0.97,1.07)	0.57
Quadratic of grip strengthb				0.60			0.56
Inverse of grip strengthc				0.87			0.63
Adjusted
Grip strength	0.95		(0.91,1.00)	0.05	1.01	(0.95,1.07)	0.70
Quadratic of grip strengthb				0.38			0.85
Inverse of grip strengthc				0.46			0.51

95%CI=95% confidence interval; secs=seconds; p=p-value. Adjusted models considered the associations between grip strength (and where relevant also its quadratic or its inverse, see footnotes b and c) and each physical performance measure in turn after adjustment for age, height, weight for height, smoking, alcohol, social class, diabetes, hypertension, ischaemic heart disease, cerebrovascular disease, bronchitis and hand osteoarthritis. a. The beta values present the regression coefficients for the association between grip strength and each of the physical performance measures. For the 6m timed up and go (TUG) and the 3m walk, the coefficients represent the estimated average change in TUG or 3m walk times in seconds, per kilo increase in grip strength. For chair rises time, which was loge transformed for analysis, the coefficients represent the proportional change in chair rises time per kilo increase in grip strength (a proportional change of 1 indicates no change, a proportional change of e.g. 0.98 represents a 2% decrease, and a 1.02 proportional change a 2% increase). For the balance test, the beta is the odds ratio for losing balance in less than five seconds per kilo increase in grip strength. An odds ratio of 1 represents no change, an odds ratio greater than one implies increased odds of losing balance with higher grip strength, and an odds ratio less than one implies decreased odds of losing balance (i.e. better balance) with higher grip strength. The R2 statistics for the proportions of variance in the PP measures which were explained by grip strength were as follows. Men: timed up and go 8.2%; 3m walk 10.0%; chair rises time 5.6%; standing balance pseudo R2 2.1%. Women: timed up and go 9.5%; 3m walk 7.1%; chair rises time 8.3%; standing balance pseudo R2 0.1%. b. In addition to a regression model including a linear term for grip strength, we also fit a model including linear and quadratic terms for grip strength to test for any curvilinear relationship between grip strength and each physical performance measure. The p-values for the quadratic grip strength term from the relevant models are presented only. c. In addition to a regression model including a linear term for grip strength, we also fit a model including linear and inverse terms for grip strength to test for any curvilinear relationship between grip strength and each physical performance measure of an inverse nature (more dramatically curvilinear than quadratic). The p-values for the inverse grip strength term from the relevant models are presented only.

Table 2 shows that there was little evidence for a curvilinear relationship (of either a quadratic or inverse nature) between grip strength and physical performance across the range of measurements observed in this study, for men or women, with or without adjustment for age, anthropometry, lifestyle and comorbidities.

Figure 1 illustrates the associations between grip strength and physical performance and confirms the absence of any convincing evidence for curvilinear relationships between grip strength and physical performance in these data.

Figure 1.

Physical performance according to gender and quintiles of grip strength

Discussion

This study has identified statistically significant linear associations between greater grip strength and improved performance in 6m timed up and go, 3m walk, and chair rises tests among men and women. These results were all robust to adjustment for age, anthropometry, lifestyle factors and major co-morbidities. Among men, higher grip strength was also associated with improved performance on standing balance but there was no association among women. We found no convincing evidence for curvilinear relationships between grip strength and physical performance among this population of young-old men and women. Our findings demonstrate that grip strength is a good marker of physical performance in this age group and we suggest that measurement of grip strength may be more feasible than completion of a short physical performance battery in some clinical settings.

Our findings are consistent with those of Davis (6), Baker (8), Stenholm (9) and Hughes (10) who have all described associations between grip strength and directly assessed measures of physical performance. Our findings are also consistent with those of Glenmark (11), Bassey (12) and Martin (13) who have identified associations between grip strength and indirect measures of physical performance such as selfreported markers of customary physical activity or difficulties in activities of daily living. However, the absence of association between grip strength and balance among women in the current study was surprising and at odds with results from the Kuopio OSTPRE study of sixty year old post-menopausal women (5); different study protocols for the measurement of grip strength and standing balance may have contributed to this discrepancy.

This study had some limitations. Firstly, multiple comparisons have been made which raises the possibility of false positive results. However, the magnitudes of effect and highly statistically significant associations between grip strength and 6m timed up and go, 3m walk and chair rises times and grip strength in this study, among men and women, suggests that the associations have not arisen purely due to chance. Replication studies could address this issue further. Secondly, residual confounding by unmeasured co-morbidities and lifestyle factors could have mediated the relationship between grip strength and physical performance. However, this seems unlikely given that the associations between grip strength and physical performance in this study were robust to adjustment for major co-morbidities and a range of lifestyle factors. Thirdly, we were only able to study men and women who were born and still resident in Hertfordshire and willing to participate in our study; this raises some concerns about the generalisability of our results. Although some evidence for a healthy-participant effect is evident in HCS, we have nonetheless previously shown that the Hertfordshire Cohort Study participants are broadly comparable with the wider population of England and Wales (17) which gives confidence that our results can be reasonably generalised to the wider population of young-old people in England and Wales. Further, selection effects would only lead to serious bias in our results if the relationships between grip strength and physical performance were significantly different among men and women who participated in our study and among the wider population of older men and women in England and Wales; this seems unlikely. Fourthly, no data were available on muscle quality rather than quantity as indicated by grip strength; this would be a useful inclusion in any future studies of this cohort. Finally, our sample comprised young-old men and women who, by definition of their ability to attend our clinics, were ageing relatively healthily and maintaining physical performance well on average. This may explain the absence of evidence for curvilinear relationships between grip strength and physical performance in our study. Future studies of older individuals in whom muscle mass may have declined closer to the threshold at which loss of muscle mass impacts on physical performance would be fascinating and would perhaps have greater ability to identify curvilinear associations between muscle mass and physical performance than the current study.

Our study also had many strengths. Firstly, the Hertfordshire Cohort Study participants are a well characterised group of community dwelling young-old men and women who have been shown to be broadly comparable with the wider population of men and women in England and Wales (17). Secondly, we were able to adjust our results for a range of major chronic co-morbidities and lifestyle factors. Thirdly, our physical performance measures were obtained from direct tests which were conducted by highly trained research nurses who measured according to strict protocols and completed inter- and intra-observer variation studies for all of the measures during the fieldwork.

In summary, physical performance relates to both the current and future health of older people (1) and is often characterised using the Short Physical Performance Battery (2). However, older people admitted acutely to hospital in the United Kingdom characteristically have mobility impairment as well as multiple co-morbidities (4); use of the SPPB may not be feasible in this context. A clinical application of our findings would be that the SPPB may in some clinical settings be replaced by the simpler, easier grip strength test. In order to validate these findings, replication studies should be carried out in varied groups of older people.

Key points

• •Physical performance relates to both the current and future health of older people and is often characterised using the Short Physical Performance Battery (SPPB). However this may not be feasible in all clinical settings.
• •As muscle strength is central to physical performance, we explored whether grip strength could be used as a marker of the SPPB.
• •In the Hertfordshire Cohort Study, men and women (aged 63-73 years) with lower grip strength were significantly more likely to have poorer physical performance, even after allowing for age, anthropometry, lifestyle and known comorbidities.
• •Grip strength is a marker of physical performance in this age group and may be more feasible than completing a short physical performance battery in some clinical settings.

Conflict of Interest

The authors have no conflict of interest to declare.