The results of this study showed that grip strength decreased at a similar slope in each age group as the age group increased. In contrast, total SPPB score, an indicator of lower limb muscle strength, showed a greater drop between ages 80 and 90 than between ages 70 and 80. This suggests a rapid decline in physical function, especially in the lower limbs, during the transition from age 80 to age 90, the age of the oldest old.
Grip strength indicates a decline in total muscle strength, which gradually declines with aging. On the other hand, SPPB scores include not only muscle strength of the lower limbs, but also the state of functional aspects such as gait and balance in the figures. Therefore, it is possible that the SPPB is also influenced by the frequency of daily activities (standing up, moving), etc. The rapid decline in lower limb function at age 90 is assumed to be due to much less outdoor mobility, such as fewer outings and participation in organizations.
The present results versus previous study findings
Table 6 presents a comparison of the findings from earlier large-scale cohort studies of physical functioning in elderly populations. The participants in the BFC80+ [(39)], the Leiden 85-Plus [(40)], the Tokyo Metropolitan Institute of Gerontology-Longitudinal Interdisciplinary Study on Aging (TMIG-LISA) 6 cohort study [(15)], and the Newcastle 85-Plus Study [(41)] were 85 to 89 years old; the BFC80+ included 85-year-olds; and the others included minimum 85 years up to over age 90. The participants in the 90+ Study [(16)], the validity 90+ Study [(17)], the Danish 1905 cohort survey [(18)], and Genetics of Healthy Aging (GEHA) [(42)] studies were age 90 or older. The GEHA physical function assessment study was conducted in Italy with a large number of subjects (more than 1,000), similar to the Danish 1905 cohort survey. Only the validity 90+ Study targeted 90-year-olds; the others had a mixture of subjects aged 90 years and older. The TMIG-LISA 6 cohort study had similar participation rates for men and women [(15)], but the overall sample size was smaller than the other studies at 116 participants. The other studies had more than 1.5 times more women than men.
In this study, grip strength decreased as age increased for both men and women, which was consistent with earlier findings among the elderly in Japan [(43)]. In previous studies conducted in Europe, the average grip strength of men and women in the Newcastle 85-Plus Study [(41)] (mean age 90) were 21.05kg and 11.5kg respectively, compared with our findings of 24.1kg men and 14.4kg women, and the average in the Leiden 85-Plus Study [(40)] (mean age 89) were 25.6kg men and 16.4kg women. That is, we had higher value than in the UK study and lower than in the Dutch Leiden study, and these variations reflect the overall inconsistency in research findings on the oldest old age 90 and over.
Usual gait speed
For usual gait speed per second in this study, the rates for men (0.8 m/s) and women (0.72) were higher than the overall gait speeds in the Leiden 85-Plus (all: 0.52 m/s) [(40)] and BFC80+ [(39)] (age 85; M: 0.7 m/s, W: 0.5 m/s), which used respondents at a younger average age than the subjects in the present study. The usual gait speeds we found were faster than those in the 90 + Study [(16)] (ages 90–94; M: 0.66, W: 0.52), which used older participants than ours, and those in the Danish 1905 cohort survey (ages 92–93; M: 0.64, W: 0.52), whose participants were also older. In contrast, the present gait speeds were slower than the speeds in the TMIG-LISA 6 cohort study [(44)] (M: 1.11 m/s, W: 0.92 m/s), which was conducted in Japan on the elderly over age 85.
Usual and maximal gait speed
The Leiden 85-Plus (age 89) and BFC80+ (age 85), both study cohorts younger than the SONIC’s, measured maximum gait speed, while the other studies (90 + Study, Danish 1905) measured usual gait speed. Because maximum gait speed is generally faster than usual speed, the fact that our findings for usual gait speed were faster than the maximum gait speeds in other studies indicates a faster gait speed in the oldest of Japanese individuals than the speed of the oldest old in the West.
The speeds in this study for the five-times sit-and-stand test (M: 15.9 and W: 16.3 × the upper limit of quartile 3 compared with the 90 + Study) were faster than those in the validity 90 + Study (M: 18.0, W: 20.0) for the same age group and those in the 90 + Study (M: 16.1, W: 16.7). In addition, the differences between men and women in the results of this study were small, and there was no statistically significant gender difference in the results in the 90-year-old reference value (3rd normal quintile). In addition, women in the SONIC study were faster in the 70-year-old age cohort.
Possible causes for value differences
When comparing the results of different studies on evaluating physical functions, it is important to use the same measurement methods. The differences between our results and those of previous studies might be attributed to the differences in measurement methods. Here we discuss possible causes for differences between studies in the evaluated PPMs and propose research directions for future studies focusing on the oldest old.
Grip strength varies depending on standing versus sitting, upper limb position, and forearm posture, so appropriate comparisons require similar measurement methods. In this study, we used the measurement method recommended by the American Society of Hand Therapists [(45)], whereby participants were required to sit, rotate their shoulders inward to a neutral position, bend their elbows to 90°, place their forearms in a neutral position, and dorsiflex their wrists between 0° and 30°. In populations other than solely older adults, grip strength in Japan is often measured in a standing position with the elbow extended, and there are few measurement data on older adults in a sitting position [(46)].
In examinations of differences in grip strength by posture, researchers found greater grip strength in the standing position than in people who were seated [(45)]. In previous studies, some researchers measured grip while standing, and some did not mention the measurement posture. These differences in measurement methods could have affected the results of comparing studies.
It is often difficult for oldest old people over age 90 to hold a standing position. In fact, about 20% of the 90-year-old participants in the SONIC study needed some kind of assistance to stand up and to hold the standing position, and it was difficult to measure their ability to stand up from a chair. Not only the oldest old but also those who cannot hold a standing position are expected to have weaker grip (muscle) strength than those who can hold a standing position without holding anything. Therefore, a participant’s ability to hold the standing position is a prerequisite for measuring grip strength, which can cause selection bias; studies might show high findings because the authors selected people with high muscle strength. Similarly, some investigators measured grip strength in an unstable state in which the subject held the balance with one hand while standing. In the Leiden 85-Plus, where the grip strength was greater than in the present study, 14.8% of the subjects could not be measured, and it is possible that those results were influenced by selection bias because people who could not stably hold the standing position were excluded from the measurement.
In short, the effect of measurement position on 90-year-olds is significant due to their physical vulnerability, and therefore, when comparing PPMs, it is important to consider the measurement method (such as measurement position and frequency) more strictly than is generally necessary with older adult. The oldest old often have difficulty maintaining a standing position and are at greater risk of falling, which suggests that it is safer and more stable to measure grip strength of the oldest old in a sitting posture.
Gait is a method of assessing physical function that many researchers have used in studies of older adults. Although many standardized data on gait speed have been published, the most common one is usual gait speed, which is highly sensitive in predicting ADL disability in people over 75 years old [(46)]. For this study, we measured gait following Guralnik’s original SPPB gait measurement method [(28)]. That is, we used a static start method in which we measured normal gait speed from the starting line within a frame of 8 ft. (2.44 m) without an acceleration period. In many cases, 90-year-olds are unable to participate in surveys outside their homes due to their declining physical functioning. We adopted a gait distance of 2.44 m in this study because we considered that oldest old citizens who were in institutions or who could not come to the survey site could measure their gaits at the places where they lived. Since the advent of the COVID-19 pandemic in 2020, group surveys conducted in large venues have been severely restricted to prevent infection, and such restrictions are expected to continue in the future. Because the SPPB can be measured at home or in a facility with limited space, we considered it appropriate for measuring physical function in 90-year-olds even during the pandemic.
The gait distance was 3 m in the Danish 1905 cohort survey and the BFC80 + and 4 m in the 90 + Study, and we believe that the researchers used these short distances in part because of the ease of measurement in facilities and homes. As the age of the elderly increases, the maximum gait speed slows down, and the effect of the acceleration period becomes relatively small [(47)]. Thus, we consider that lower limb function in the oldest old can be appropriately assessed even by walking a short distance.
Researchers in Japan commonly use dynamic start with an acceleration period before the measurement distance, whereas Western researchers commonly use a static start without an acceleration period. The dynamic start is faster because it reduces the influence of the slow acceleration period. Because studies on Japanese elderly people including the TMIG-LISA 6 cohort study used dynamic start [(44)], it is not possible to accurately compare Japanese findings with findings from overseas studies. However, with the present study, we used the static start method, and thus, we consider that our findings are comparable with those from Western studies. This is the first study to report walking speed by gender in Japanese subjects aged 90 years who were not certified as needing long-term care using an internationally standardized measurement method.
Chair stand and chair height
It is possible that the height of the chair had an advantage for Japanese women, who are shorter than those in Europe and the United States and than men. One of the factors that affects standing behavior is the height of the chair used for measurement [(48)]. In the SONIC study, a Japanese standard chair height of 40 cm was used for both men and women, and because women are shorter than men, a chair of the same height would make it relatively easier for women to stand up and harder for men. This could be why there was little difference between men and women on the five-times sit-and-stand test.
It is also possible that the differences in chair heights used between the studies in Europe, the United States, and Japan could have affected the rise times in this study in addition to the differences. However, we could not verify this proposition because researchers on the previous studies did not include the heights of the chairs used. Therefore, accurately comparing rising data internationally requires noting and considering differences in the chair heights used.
Differences in lifestyles
Research has established that differences in walking speed are attributable to differences in lifestyle. In the case of this study, we assumed that the differences in walking speed from findings in other studies were attributable to that people in Japan who tend to live a tatami mat lifestyle; that is, they regularly stand up and sit down each day, which strengthens their lower limbs on a daily basis [(49)]. In Japan, many elderly people continue to live on tatami mats until they require nursing care, when it is common for them to change to chairs or beds. It is necessary to further investigate whether similar cultural factors and lifestyle features affect ADLs and other measurements of functioning in the oldest old in different countries.
Standing itself as a screening criterion
In this study, the proportions of participants who failed or were unable to perform the five-times sit-and-stand test was the largest among the tests of physical performance assessment: 1.6% in the age 70 cohort, 4.7% in age 80, and about 20% in age 90. In other words, the percentage of 90-year-olds who were unable to perform the test five times was much higher than that of other age cohort.
Furthermore, compared between PPMs, the proportion of 90-year-olds who could not complete the chair stand test failure (19.7%) was larger than those for grip strength (2.1%) and normal gait speed (6.7%) among the other tests. In addition, 20.9% of the subjects took longer than 20 s to perform the test (Table 1). This means that at the age of 90, the test of standing up from a chair proves to be remarkably difficult.
In the Danish cohort survey of nonagenarians, 61% of men and 50% of women were able to stand up without using their hands [(18)]; in other words, 39% of men and half of women could not stand up without using their hands. Similarly, in the 90 + Study, 32% of all participants failed to complete the five-times sit-and-stand test. These results indicate significantly higher proportions of people who have difficulty standing up without using their hands in the oldest old (90 years and older) than in younger old age groups.
Guralnik originally gave an SPPB score of 0 for failures to complete the five-times sit-stand and a score of 1 to the lowest 25% of those who completed the test [(28)]. Based on the results of this study and of previous studies, it is possible that about one quarter of the oldest old are unable to rise from a chair without holding on to something, which corresponds to the lower approximately 25% of the total oldest old population to the criterion for 1 point on the original SPPB. Therefore, the five-times sit-to-stand test could be a screening test for lower limb muscle strength. Indeed, the AWGS, in its 2019 criteria for the diagnosis of sarcopenia, allows general practices and facilities to use the test if they do not have dual-energy X-ray absorption or bioelectrical impedance analysis [(14)].
Researchers on a study of elderly people in Japan used a test called the Frail 10-Second Chair Stand Test to measure the number of times a person could stand up in 10 s without using their hands in a similar way to the SPPB and found that it was independently associated with quadriceps strength and TUG test score, which represents dynamic balance [(50)]. These findings suggest that the chair stand test can be used to assess overall lower extremity muscle strength even when used alone in the oldest old over 90 years of age (in Japan).
Study strengths and limitations
The main limitation in this study was selection bias. The subjects in this study were 90-year-olds who were not certified as needing long-term care who were otherwise eligible to participate, but according to the 2015 census, approximately 50% of Japanese aged 90 and older are certified as needing long-term care in an institution . Therefore, we believe we can assume that the participants in this study were relatively healthy community-dwelling 90-year-olds, and the distribution of physical functions across all 90-year-olds can be estimated by adding assessments that include those certified as requiring long-term care. In addition, as mentioned in the discussion, we grouped the three populations from different participation years, so it is necessary to consider the rejuvenation phenomenon separately among these populations. Finally, due to the cross-sectional design of this study, it was not possible to make causal inferences about age or sex differences in PPMs.
The main strength of this study is that we had access to a large sample size of the oldest old over 90 years of age one country in Asia, where there are still few oldest old people; generally, there are fewer oldest old men than women, but rates in this study were similar: 48.8% for men and 51.2% for women. Furthermore, the results of this study can be easily compared with other studies because we assessed physical function using internationally standardized measurement methods. At the present stage, there are no better representative data for Asia’s oldest old, including in Japan, and this is why we believe our results can be used as the best current reference values for the oldest old people in Asian countries.