The major finding of this study was that higher GS and HGS and less time to perform CST were associated cross-sectionally with independent ageing in very old men. Higher GS at baseline was also associated with independent ageing five years later. Further, severe sarcopenia, only present in 2%, was associated with non-independent ageing in the cross-sectional analyses. However, there was no association between SMI or the other sarcopenia categories, as defined according to the EWGSOP1 and 2 criteria [5, 8], and independent ageing.
Regarding functional variables, both GS and HGS are well-established markers of biological ageing so the findings are not surprising [30,31,32,33,34]. However, studies of the oldest old with outcomes similar to ours are scarce [16, 18]. In the Honolulu Heart Program, high GS and HGS at mean age 76 was not associated with healthy ageing at age 85 . Healthy ageing was defined as being free from major diseases, no physical limitation and with normal cognitive test performance . However, in the cross-sectional Helsinki Businessmen Study, higher GS was correlated with active and healthy ageing in octogenarian men . In contrast to the present study, the criteria for active and healthy ageing were self-reported and included “feeling happy”, absence of major diseases, functional or cognitive impairment . Lower HGS was also correlated with the combination of physical and cognitive impairment in a cross-sectional study in participants with a mean age of 68 years . To the best of our knowledge, there are, so far, no studies of the association between CST and combined cognitive and physical function.
In the ULSAM, higher GS was correlated with well-preserved cognition at baseline, which is in line with other studies on very old subjects [13, 35,36,37], and confirmed by a meta-analysis . One explanation of this relationship can be white matter lesions or neurodegeneration in brain areas involved in both gait and cognitive functions . However, we could not confirm the results from other longitudinal reports showing an association between high GS and lower risk of cognitive decline [34, 35, 39]. One reason for this may be loss to follow-up, i.e. men with incident cognitive impairment declined to participate in the MMSE at follow-up. HGS was not associated with cognition in the present study and previous results are inconsistent. Two different cross-sectional studies on old women showed opposite results [13, 40]. In the Leiden 85-plus Study, low HGS was cross-sectionally correlated with a lower MMSE and also predicted an accelerated decline in MMSE-score four years later . Furthermore, a review of longitudinal studies showed an association between low HGS and decline in cognition . In a US study in nonagenarians slower CST was cross-sectionally associated with dementia , while no association was seen between slower CST at baseline and dementia 2.6 years later . Walking and chair-rising are complex activities that depend on more than just strength. Thus, GS and CST might be more sensitive than HGS to cognitive function and also to skeletomuscular disorders, such as arthritis. Further studies are needed on CST, which is a functional measure that is less investigated than GS and HGS. In the present study, higher GS was associated with independency in personal care five years later, which is in line with previous reports [32, 43,44,45]. Other longitudinal reports, but not our study, have also shown an association between lower GS and immobility [32, 44]. High HGS was cross-sectionally associated with independency in outdoor walking in ULSAM, which is in line with a previous review . Although other reports have shown a relationship between low HGS and subsequent dependence in personal ADL the associations were not significant in our study [31, 41, 43]. In an American study, low GS and HGS were associated with institutionalisation six years later, independently of chronic conditions . This is in contrast to our current findings in the ULSAM cohort, which show associations between low HGS, but not low GS, and institutionalisation five years later, also independently of comorbidity. One explanation of the different findings may be the high mean GS in the present population. CST was not associated with any of the single components of independent ageing. No other studies of the association between CST and institutionalization or disability in ADL and mobility in the oldest old have been found.
Regarding muscle mass, similar to our study the cross-sectional study by Tolea et al. found no association between low muscle mass measured by bio-electrical impedance analysis (BIA) and the combination of physical and cognitive impairment . The MEDIS study reported that higher SMI was cross-sectionally associated with successful ageing . However, that study included both genders, with an age range of 65–100 years. Furthermore, SMI was equation-based and the successful ageing index added together 10 components including education, BMI, physical activity and social participation, among others. With this heterogeneity in outcomes it is difficult to compare the results. No longitudinal study with a similar outcome was found. Previous studies on subjects with mean age ≥ 75 years have not shown any association between SMI and dementia or severe cognitive impairment [13, 36, 46]. Furthermore, SMI was not correlated with mobility in older women . However, low muscle mass at mean age 77 was associated with ADL disability and institutionalisation five to seven years later for Australian community-dwelling men .
Muscle strength is lost more rapidly with ageing than muscle mass  and this can be one reason for finding an association between muscle function, but not muscle mass, and independent ageing in ULSAM. Furthermore, not just muscle size but also changes in muscle quality, i.e. architecture and biochemistry, may require attention . However, muscle quality, including the amount of myosteatosis, cannot be evaluated by DXA.
When we related the composite sarcopenia diagnoses to our outcomes we noticed that severe sarcopenia, but not confirmed or probable sarcopenia, was cross-sectionally associated with loss of independent ageing and needing assistance when outdoor walking. However, severe sarcopenia was only present in 2% of the participants. To the knowledge of the authors, only two studies using the EWGSOP2 definition have, so far, been published [9, 10]. These studies investigated the prevalence of sarcopenia and its association with mortality and risk factors in populations with a mean age around 75 years. Sarcopenia, as defined by the EWGSOP1 definition, was not associated with independent ageing in the present study. The only similar study found was the cross-sectional study by Tolea who showed an association between sarcopenia and combined impairment in cognitive and physical performance . In one cross-sectional study sarcopenia was associated with cognitive impairment , while no association was found in another study . The participants in both studies had a mean age > 80 years. However, one meta-analysis also including younger subjects found a correlation between sarcopenia and cognitive impairment . Another meta-analysis also found an association between sarcopenia and subsequent functional decline . Finally, sarcopenia at mean age 77 years was associated with ADL disability and institutionalisation five to seven years later for community-dwelling Australian men . However, comparisons may be slightly hampered due to somewhat different definitions used.
The prevalence of sarcopenia differs by definition criteria but also by age and gender . When using the EWGSOP1 definition the prevalence in ULSAM, as well as for men in the Newcastle 85+ Study, was actually the same, 21% . . In Italian octogenarian men the prevalence was 17% , while 13% of Belgian octogenarian men had sarcopenia . These three studies also used the EWGSOP1 definition , but measured muscle mass with BIA, and not DXA. Using the updated definition on confirmed sarcopenia decreased the prevalence from 21 to 20% in the present study. In the only two published studies, so far, using the up-dated definition, the prevalence was below 10%, which most likely is explained by the lower mean age in these populations [9, 10].
Regarding strengths and limitations, this cohort consisted of men of similar age and ethnic background, which limits its generalizability. On the other hand this design makes it possible to exclude age, gender and ethnicity as confounders. The study population was skewed toward a healthier population than the background population, i.e. only 5% had dementia at baseline, which is lower than expected in this age group. This further reduces the generalizability. However, this might lead to underestimation of the associations observed, rather than overestimation. Misclassification is a possibility as some of the components for independent ageing were self-reported, but this potential introduction of bias was reduced by reviewing the medical records for consistency with the self-reported information. Participants who lacked information on MMSE at follow-up might also have been wrongly classified as independently aged. Although the associations were adjusted, other possible confounders, such as depression , might not have been taken into consideration. Finally, it is not possible to draw any conclusions about the direction of the associations between exposure and outcome in the cross-sectional analysis. However, in the longitudinal analysis of participants with independent ageing at baseline, a high GS or HGS predicted independent ageing five years later.
GS was measured with a dynamic start giving a high mean GS compared to other studies in the same age group . Also, the effect of HGS might be affected as both hands, and not only the dominant, were measured and the highest value was used. This might lead to a false and excessively low prevalence of sarcopenia but also an underestimation of the associations with sarcopenia, GS and HGS.
In conclusion, muscle function, but not muscle mass, was associated with independent ageing in very old men. When using the updated EWGSOP2 definition, severe, but not probable or confirmed sarcopenia, was associated with loss of independent ageing. Otherwise, none of the present definitions of sarcopenia according to EWGSOP were associated with independent ageing. Measuring GS, HGS and CST is easy and possible to do in the clinical setting. Those assessments may identify individuals who can benefit from interventions that might at least postpone the loss of independence. However, further studies are needed to identify these possible interventions.