In this study, 22.8% elderly women were frail. A previous Japanese study  which used Fried’s criteria reported a prevalence of frailty among Japanese of 2.7%. However, this study excluded subjects with a dependent (10.0%), and we assume that the actual proportion of frailty in all subjects was actually higher than reported. Other studies [26, 27] reported a prevalence of frail subjects of 18.5% and 10.6%, but these used different criteria which were not confirmed to be consistent with Fried’s criteria.
We found that a higher intake of total protein was associated with a lower prevalence of frailty among elderly Japanese women. This association was also observed for both animal and plant protein. Although the intakes of eight selected amino acids were also inversely associated with frailty, the association of total protein was stronger than those of any of these amino acids individually. To our knowledge, this is the first study to examine the association of intake of not only protein but also amino acids with frailty.
To date, three studies have examined the association of protein intake with frailty. Bartali et al. reported that low intake of total protein was significantly associated with frailty in a cross-sectional study . Beasley et al. assessed the effect of protein source (animal protein) or protein quality (essential amino acids) on frailty in a large-scale cohort study, and found that high intake of protein was associated with a lower risk of frailty, and that these associations were independent of animal protein and essential amino acid intake . In contrast, Bollwein et al. showed that amount of protein intake was not associated with frailty, but the distribution (in the morning, at noon, and in the evening) was significantly associated with frailty .
Our study showed that the intake of total protein was inversely associated with frailty in a large cohort of elderly Japanese women. Although the intakes of animal and plant protein and individual amino acids were also associated with frailty, these associations were less marked than that of total protein. These results might indicate that the source of protein or kind of amino acid included in dietary protein might not be particularly important in preventing frailty. Many amino acids are assumed to suppress age-related muscle loss by regulating muscle protein synthesis and overcoming oxidative stress [7, 12], and the amino acids in the present study might prevent frailty in a cooperative manner. The strong association of total protein with frailty might be due to an additive effect of these amino acids. Meanwhile, previous study showed that vitamin E, vitamin C, and folate were also associated with frailty . Given that many food sources of plant protein are rich in these antioxidants, the effect of plant protein on frailty observed in the present study may have been caused by these antioxidant nutrients included in plant food rather than amino acids.
Japanese government recommends a total protein intake of 50 g/d for women aged ≥70 years . In the present study, subjects who consumed about ≥70 g/d protein had a significantly lower risk of frailty. A previous review study indicated that a protein intake level which meets the nutritional requirements of all healthy individuals does not protect the aged from sarcopenic muscle loss . Although we cannot adequately discuss the appropriate amount of protein intake in this study due to the limited validity of BDHQ, the amount of protein required to prevent frailty might be higher than the recommendation. Distribution of protein intake to meals (in the morning, at noon, and in the evening) may relate to frailty besides the total amount of protein intakes . One recent study supported this hypothesis . Further studies are warranted to clarity both quantitative and qualitative value of protein intake against frailty.
The major strength of our study is that we could examine the relation of protein and amino acid intakes with frailty in a large number of elderly women using multicenter epidemiologic data. Subjects lived over a wide geographical range of Japan and had various dietary and lifestyle habits. Additionally, the dietary questionnaire used has been validated [15, 16], albeit that the validity of the intakes of individual amino acids has not been published. Pearson’s correlation coefficients of the amino acids used in the present study were 0.31-0.37 (unpublished observations, H. Suga, ), which is comparable to the value for protein (0.35) in the previous validation study .
However, several limitations also warrant mention. First, we defined frailty using the score of the physical functioning scale of the SF-36 as a surrogate for walking speed and grip strength, as proposed by Woods and colleagues . Although we were unable to use the most common frailty criteria, developed by Fried and colleagues , Woods et al. showed that the physical functioning scale dichotomized at the 25th percentile was strongly associated with poor walking speed and moderately associated with poor grip strength, and maintained that their definition predicted outcomes as well as Fried’s definition . These results may indicate the appropriateness of the criteria we used. Also, the definition of low physical activity was dependent on the distribution of activity level among study subjects. We therefore examined the association of protein and amino acid intake with frailty using different cut-off points for low physical activity. Classification of low physical activity by lowest tertile or lowest quintile produced similar results to those in Table 3 (data not shown). These results suggest that the lowest quartile was an acceptable cut-off point of low physical activity. Second, although we were unable to include the intake of dietary supplements in calculating protein intake, the use of supplements containing mainly protein or amino acid is uncommon in Japanese adult women (1.5%) , and any influence of supplements on protein intake may be low. Third, almost all subjects of the present study were grandmothers of selected dietetic students, not a random sample of Japanese elderly women. Not all Japanese adolescents enter college or university (enrollment ratio = 56% ) and the grandmothers of students who do might accordingly have a relatively high social and economic status. Thus, our results cannot be readily extrapolated to the general Japanese elderly population. Additionally, the response rate of the grandmothers (33.2%) was low compared to that of the students (70.3%). This might have resulted in a degree of selection bias. The grandmothers who did not participate in the study might have had unhealthy dietary habits and were frail, and this might have attenuated the results. Fourth, although we attempted to adjust for a wide range of potential confounding variables, we were unable to rule out residual confoundings. Finally, the present study was conducted under a cross-sectional design, which prevents the investigation of a causal effect of protein and amino acid intake on frailty. However, we tried to minimize the effect of reverse causality by excluding subjects assumed to be under restricted protein intake (chronic kidney disease) and who had disability (Parkinson’s disease or those who were unable to walk), and also by calculating ORs adjusted for the history of chronic disease.