We have found evidence of associations between all four measures of physical capability investigated (grip strength, walking speed, chair rises, and standing balance) and all cause mortality. People in community dwelling populations who perform less well in these tests were consistently found to be at higher risk of death. The estimates from meta-analyses for grip strength, walking speed, and chair rises show a dose-response relation. With the exception of grip strength, studies have been done exclusively in older populations, and most have relatively short follow-up, so whether similar associations would be found at younger ages or after participants have been followed-up for longer than 10 years is unclear. For grip strength, we found evidence of an association even in populations with an average age at baseline of less than 60 years, although the association weakened with increasing length of follow-up.
Explanation of findings
Several possible explanations exist, which are not necessarily exclusive, for finding associations between objective measures of physical capability and mortality in community dwelling populations. Firstly, these findings could be explained by confounding (for example, by socioeconomic position or levels of physical activity), as effect estimates have only been adjusted for age, sex, and body size. These factors were considered to be the most likely confounders yet did not explain the associations of the physical capability measures with mortality. Apart from these, no standard set of multiple adjustments across papers existed and we thought that requesting further adjustments would have affected the response from study authors and led to inconsistencies in adjustments across studies.
Secondly, these measures of physical capability could be markers of disease and general health status. Some of the community dwelling populations included in this review consisted of people with diseases or comorbidities that were not considered severe enough to warrant exclusion from the study but that may have affected both their physical performance and mortality risk. This could definitely apply to walking speed, chair rises, and standing balance, for which studies have been done only in older populations with shorter follow-up. However, this seems less likely to fully explain associations between grip strength and mortality, as these were also found in studies with follow-up over 20 years, in younger populations in which the prevalence of sub-clinical disease and existing comorbidities would be lower, and in studies that by the nature of their design (for example, recruitment of men from the active workforce) excluded people with health problems.25
A related possibility is that underlying ageing processes led to poorer performance and a higher probability of chronic disease and death. Walking speed, chair rising, and standing balance require strength, balance, and motor control; walking speed and chair rising also require muscle power and speed and adequate cardiorespiratory function; standing balance requires mental concentration. These functions decline with age, may co-vary, and contribute to the risk of frailty. The progressive dysregulation of homoeostatic equilibrium across multiple systems may be the biological basis of frailty; common pathways proposed include endocrine dysfunction, inflammation, oxidative stress, and disequilibrium between the sympathetic and parasympathetic systems.55
Functional status in later life reflects the peak achieved during growth and development, as well as the rate of decline. Thus, the relation between these measures of physical capability, even when measured at younger ages, and mortality could also reflect initial differences in development that affect both. Evidence that early growth, cognitive and motor development, and childhood social environment are associated with adult physical capability, chronic disease, and mortality support this possibility.56 57 58 59 60 61 62
Many of the estimates of I2
calculated in these meta-analyses would be judged to be high.49
We have presented summary, overall estimates from meta-analyses despite this, as most effect estimates from individual studies were in the same direction and doing meta-analyses has been argued to still be appropriate.63 64
Furthermore, the value of I2
depends on both the within study and the between study variance,48 65
and as many of the studies included in this review have precise estimates (figs 3 and 4), either because the study population was large or a high proportion of participants died, the values of I2
will have been affected by this. However, these summary estimates, which are an average of estimates across populations with different characteristics, should be cited with caution, and the reasons for finding these levels of heterogeneity should be explored.
Insufficient variation often existed in the characteristics we proposed a priori as possible sources of heterogeneity to allow us to examine their role in explaining heterogeneity fully—for example, four of the five studies of walking speed and all five studies of chair rise time included in meta-analyses were in populations aged over 70. However, meta-regression analyses of comparisons of quarters of grip strength suggested that associations with mortality were weaker in those studies with longer follow-up, even after adjustment for age. Comorbidities at the time of assessment of capability, which would increase short term mortality, may be more likely to explain the association between grip strength and mortality in studies with shorter follow-up. With increasing length of follow-up, the proportional hazards assumption within a study may be violated.
In addition to the characteristics investigated, many others vary between studies and could result in heterogeneity. These include differences in exclusion criteria, the instruments used, the main causes of death, levels of underlying comorbidity, and ethnic diversity.
This review has highlighted the paucity of studies that have measured physical capability in younger populations with subsequent follow-up for mortality. This situation is expected to change; these measures are being introduced in studies of younger populations as overall markers of functioning at the multi-system level, rather than as markers of severity and stage of specific chronic diseases.66
However, investigation of associations with mortality in studies with measurement of capability at younger ages will obviously need lengthy follow-up. Research is also needed to examine the associations between changes in capability with age and mortality, as a steep decline in physical capability may be a better predictor of mortality than is the absolute level at a single point in time. In addition, associations between these measures and cause specific mortality and other health outcomes may help to elucidate the pathways underlying the associations with all cause mortality, although few studies identified reported on these. Elucidating the underlying biological pathways that link poorer capability to mortality will inform the development of effective interventions.
We chose to examine the relation between each individual measure of physical capability, by using a standardised exposure measurement (comparisons of quarters), and mortality. The rationale behind this is that a variety of composite scores exist that are derived by using these measurements in combination, but whether results with such scores are driven by one measure or whether they each make a similar contribution is unclear. Although our findings suggest that all four measures of physical capability assessed are associated with all cause mortality, the relative paucity of data for walking speed, chair rises, and standing balance makes us cautious about drawing conclusions on their relative strengths. As these measures of physical capability are highly correlated with each other, more studies are needed that consider the value of each additional test once the findings for one test are known.14
For clinical practice, investigating whether a derived composite score representing overall lower or upper body function, such as the short physical performance battery score43
or one of the frailty indices, may be a stronger predictor of mortality than any of the individual measures are by themselves would be of interest.
The associations found between measures of physical capability and mortality seem to operate across the whole range of ability, with no apparent threshold effect. Therefore, if these measures were to be used as screening tools, clinicians and researchers would need to identify thresholds with caution and recognise that differences in the most appropriate place to set these may exist, depending on the characteristics of the population to be screened. Ultimately, randomised controlled trials will be needed to determine whether interventions aimed at improving physical capability are effective at improving capability and as a consequence are effective at reducing morbidity and mortality.
Strengths and limitations
The main strengths of this systematic review are its inclusion of several measures of physical capability and its inclusion of as many relevant studies as possible by making contact with study authors. By following a strict protocol, testing a priori hypotheses, and including unpublished results, we have minimised bias.
The study also has some limitations, although none of these affects our conclusions. Results from each study are based on the assumption of proportional hazards, which may not hold in all studies. The meta-regression analyses were likely to be underpowered, as were the formal tests of publication bias, especially for walking speed and chair rises for which we had less than the recommended number of data points. That the funnel plots and formal statistical tests produced no clear evidence of publication bias should thus be interpreted with caution. However, our success in obtaining unpublished results should limit publication bias.
This review shows the value of objective measures of physical capability as predictors of subsequent mortality in older community dwelling populations. Grip strength measured at younger ages also predicted mortality, but whether walking speed, chair rise time, and standing balance performance are associated with mortality in younger populations remains to be seen.
What is already known on this topic
- Growing evidence indicates that simple objective measures of physical capability may be useful markers of future as well as current health
- Interest is increasing in these tests and their potential use as simple screening tools
What this study adds
- Despite heterogeneity between studies, consistent evidence shows associations between grip strength, walking speed, chair rise, and standing balance performance and all cause mortality in older community dwelling populations
- The inverse association between grip strength and all cause mortality was also seen in younger populations
- These measures may provide useful tools for identifying older people at higher risk of death