Two claims, and only one of them is graded here
The prognostic claim is that knowing a person's grip strength tells you something useful about their probability of dying or becoming disabled in the following years, beyond what age, sex, smoking status and the conventional risk factors already tell you. That claim is well supported and it is what the B grade above refers to.
The interventional claim is that if a particular person trains and their grip strength rises, their own risk falls correspondingly. That is a claim about causation within an individual, it has not been tested by randomised trial with clinical endpoints, and we grade it C. Our review of VO2 max sets out the same distinction in a literature where it is even more often collapsed.
Grip strength is unusual in that the gap between the two claims is wider than usual, because nobody seriously believes that hand strength itself is protective. Grip is a proxy. What it stands in for is total body muscle strength, and behind that, neuromuscular integrity, nutritional status, physical activity history and the cumulative burden of chronic disease. Training the proxy without changing the thing it proxies for would be expected to do nothing at all.
What is actually being measured
Maximal isometric handgrip force, recorded with a handheld dynamometer, usually as the best of several attempts, usually seated with the elbow at a right angle. The equipment is inexpensive, the test takes well under a minute, it requires no laboratory, and it can be performed on people too frail for any form of exercise testing. That last property is why it has become the standard field measure of muscle strength in older populations.
It has a formal clinical role. European consensus criteria for identifying sarcopenia, the age related loss of muscle mass and function, use low measured grip strength as the primary indicator that prompts further assessment, with muscle quantity used to confirm and physical performance used to grade severity. This is worth knowing because it means grip strength is not only a research curiosity; it is a case-finding tool with published thresholds.
The measurement has real weaknesses. Different dynamometers are not interchangeable. Protocols vary in posture, number of attempts, encouragement given and which hand is used. Grip is affected by hand and wrist conditions, including arthritis and previous injury, which are common in exactly the age groups where the measurement is used most. And normalisation is genuinely unresolved: absolute force, force divided by body mass and force divided by body mass index give different answers about who is weak, and the choice materially changes the association reported.
What the cohort evidence shows, and what it does not
The pattern reported across large cohorts recruited on several continents is consistent in direction. Lower measured grip strength is associated with higher all-cause mortality, with higher cardiovascular mortality, and with later loss of independence, mobility limitation and hospital admission. The relationship is graded rather than a cliff at some cut-point, and it generally persists after adjustment for age, sex, body size, smoking, socioeconomic position and known disease.
Three features earn the grade. The direction replicates in populations with very different diets, health systems and disease profiles, which argues against the finding being an artefact of one setting. The dose-response is orderly. And the association is not obviously explained by any single conventional risk factor, since it survives their inclusion in models.
Three features cap it below A. The first is reverse causation, which is a more serious problem here than for cardiorespiratory fitness. Undiagnosed cancer, heart failure, chronic kidney disease, dementia and depression all reduce grip strength before they are diagnosed. Cohorts address this by excluding deaths in the first years of follow-up, and the association typically survives, but the exclusion windows used are often short relative to the prodromal period of the diseases concerned.
The second is measurement heterogeneity. Pooling studies that used different devices, postures and normalisations produces a summary estimate whose exposure variable is not quite the same thing in each contributing study.
The third is that grip strength is entangled with everything. It is lower in people who are poorer, less active, undernourished, depressed, socially isolated and multimorbid. Statistical adjustment for those factors is only ever as good as their measurement, and several of them are measured crudely.
On the interventional side, what is established is that resistance training increases measured strength in adults of all ages, including in the very old and in people with multiple conditions. What has not been established is that the resulting change in strength produces the survival difference the cohorts describe. Our review of resistance training and mortality covers what the trial evidence on training does and does not reach.
The limitations that constrain what this can tell you
| Limitation | What it affects |
|---|---|
| Reverse causation | Several undiagnosed conditions lower grip strength years before diagnosis, inflating the apparent prognostic effect. |
| Device and protocol variation | Dynamometers and testing positions are not interchangeable, so pooled estimates blend slightly different exposures. |
| Unresolved normalisation | Absolute force, force per kilogram and force per unit body mass index identify different people as weak. |
| Local pathology | Arthritis, tendon injury and neuropathy in the hand reduce grip without indicating anything about systemic health. |
| Confounding by social position | Grip tracks occupation, nutrition, activity and deprivation, all of which independently predict mortality. |
| No randomised survival trial | Nothing tests whether raising strength lowers an individual's own risk. |
The practical reading is narrow and worth stating precisely. A low grip strength in an older adult is a reasonable prompt for a fuller assessment, which is exactly how sarcopenia criteria use it. A single number in a healthy adult in midlife is close to uninterpretable on its own, and a change in that number after a few weeks of training tells you that you have got better at the test.
What would change the grade
The prognostic grade would move to A with cohorts using a standardised protocol and pre-specified normalisation, repeated measurements rather than one, and exclusion windows long enough to make prodromal disease an implausible explanation. Those studies are feasible and some of the necessary data already exist.
The interventional grade would move from C with randomised trials of resistance training in older adults reporting hard endpoints such as fracture, incident disability, admission or death, with the effect tracking measured strength change. Trials in fall and fracture prevention are the most plausible source.