Decline in independence after three years and its association with dietary patterns and frailty factors in community-dwelling older people: an analysis by sex and age stage

Support based on sex and age is required to implement longevity strategies. This study claried the association between declining independence after three years and the dietary patterns and frailty factors of community-dwelling independent older people. We analyzed sex and age stages for people between 65 and 75 years (earlier-stage) and people aged 75 years or above (latter-stage), respectively. In a longitudinal study of 25 Japanese prefectures from 2013 to 2016, 3,693 respondents completed baseline and follow-up questionnaires. We analyzed 2,250 participants (1,294 men), after excluding individuals younger than 65 years (n = 510), deceased (n = 35), with missing data (n = 866), or disabled (n = 32). Independence was evaluated based on Instrumental Activities of Daily Living (IADL) scores (maximum = 12). Disability was dened as scoring lower than 9. Dietary patterns were derived from a principal component analysis of 7 food groups. Frailty factors showing a signicant relationship with baseline IADL scores were selected. Multivariate logistic regression analysis revealed an association between baseline factors and declining independence after three years.

females alone showed that a high-protein (total and animal protein) diet is related to the prevention of frailty onset, independent of the intake of total antioxidant capacity [8].
The multifactorial support of nutrition and diet besides addressing other frailty contributors in older people may be effective in preventing frailty. An intervention focused on physical exercise and nutritional status [12] or a combination of physical, nutritional, and cognitive was most effective for older people with pre-and post-frailty [13]. However, few studies on the multifactorial relationship between dietary patterns and frailty in older people have shown the effects of multifactorial interventions, such as physical activity, cognitive function, and community involvement. Understanding the effective targeting of support by sex and age in older people may provide useful evidence.
Therefore, we conducted a nationwide, longitudinal survey of community-dwelling older people in Japan to assess respondents' instrumental activities of daily living (IADL) three years from baseline and evaluated changes in the degree of their independence. This study aimed to comprehensively clarify the association among a decline in independence over three years, dietary patterns, and frailty factors in community-dwelling independent older people, as analyzed by the sex and age stage of two groups: people between 65 and 75 years (earlier-stage older people) and people aged 75 years or above (latterstage older people).

Study participants
This longitudinal survey, which was planned and conducted in collaboration with the Foundation of Social Development for Senior Citizens (FSDSC, Tokyo) and Tokyo Metropolitan University, involved 25 prefectures in Japan. In 2013, a baseline survey of 9,508 residents, who had taken part in healthy longevity events carried out by FSDSC, was conducted (response rate: 45.7%) wherein a questionnaire was sent by the FSDSC's staff to the participants personally or by mail. A follow-up survey using the same questionnaire was conducted three years later and involved 3,990 respondents from the baseline survey who consented to cooperate in the second and subsequent surveys (response rate: 92.6%). A previous paper describes the details of this survey [14].
From the 3,693 valid respondents in both surveys, we excluded surveys of those younger than 65 years (n = 510), deceased (n = 35), or with missing data (n = 866). Their independence was evaluated using the instrumental activities of daily living score (IADL score) [15], which ranged between 1 and 12 points. Disability was operationally de ned as participants with an IADL score lower than 9. After excluding 32 individuals, we analyzed the responses of 2,250 people (men: 1,320, women: 962) who were identi ed as independent.
Evaluation of independence and classi cation of study participants categories: 18.5 < BMI, 18.5 ≤ BMI < 25, 25 ≤ BMI < 30, and BMI ≥ 30. (4) For exercise frequency, we used the index of "exercise frequency with enjoyment and ful llment" [14] that we developed for a previous study for the same participants. Besides the high frequency of exercise for older people, doing so with enjoyment and ful llment is strongly related to maintaining independence three years from baseline [14]. There were two questions; the rst asked "how much do you exercise or play sports." We classi ed each respondent as "exercises three or more times a week" or "exercises two or less times a week." Next, we asked the participants about their enjoyment and ful llment. If they responded that they enjoyed their exercise, we reclassi ed them as "exercises three times or more per week with enjoyment and ful llment" (or not) and "exercises twice or less per week with enjoyment and ful llment" (or not). Finally, four categories were created for the exercise index. (4) We also inquired about fall fractures that had occurred over the previous year. (5) We asked the participants to self-report their cigarette smoking status (current, former, never). (6) The respondents' social participation was tallied through the frequency of their community and volunteer activity engagement. The above items were used in the analysis as multidimensional factors related to frailty.

Other variables
We also analyzed variables related to demographic and socioeconomic status (SES), including age, sex, household status (living alone), economic satisfaction, and salaried job. We used economic satisfaction as an alternative SES variable for annual income, because subjective economic status for older people is more closely related to psychological health than annual income [18].

Analysis
The baseline characteristics were compared between the high and low independence groups after three years, using the Mann-Whitney U test for the ordinal scaled scores (including dietary intake frequency).
Chi-squared tests were used for all the other variables. We performed principal component analysis to identify dietary patterns, using the dietary intake frequency scores for seven types of food groups.
Multivariate logistic regression analysis was used to evaluate the association among low independence three years after baseline, dietary pattern, and frailty factors. A signi cant single relation with low independence three years after baseline was found, with no correlation coe cient of 0.5 or more between the indices.
All statistical analyses were performed using SPSS Statistics 24.0 (IBM). The statistical signi cance level was set at p < 0.05.

Results
Relationship between baseline-related factors and independence three years later The study participants were 1,294 men (57.5%) and 956 women (42.5%), a total of 2,250 independent older people. The average IADL score three years after baseline was 11.5 ± 1.1, which was signi cantly lower than the baseline average score of 11.6 ± 0.74. We classi ed the participants with above-average scores three years after baseline as having high independence (n = 1,565, 69.6%) and those with below-average scores as having low independence (n = 685, 30.4%; Table 1). The proportion of low independence respondents three years after baseline was 37.5% for men, 20.9% for women, 28.3% for earlier-stage older people, and 35.2% for latter-stage older people. Of the low independence participants, 17.9% had no change three years from baseline, whereas 12.5% of the high independence group changed to low independence. Of the high independence group, 60.9% maintained their status three years later, and 8.7% of the low independence group changed to high independence. The ordinal scaled scores were examined by Mann-Whitney U test othets were by chi-square tests. * p<0.05, ** p<0.01, *** p<0.001 Table 1 shows the relationship between the baseline values and the independence scores three years later, with a comparison between the high and low independence group results. The analysis was performed at each age stage. The main item with which only earlier-stage older people showed a signi cant relationship was smoking; the percentage of those who had never smoked in the high independence group three years after baseline was the highest at 77.8%, whereas those who smoked and became low independence was 44.4%, the highest among the low independence group. An analysis of the BMI categories by sex revealed that the earlier-stage older men showed a signi cant relationship between the proportion at 18.5 ≤ BMI < 25 and 25 ≤ BMI < 30 who became high independence three years later, at 64.5% and 62.8%, respectively. Of those who were at 18.5 < BMI and BMI ≥ 30, the percentages of low independence were 64.3% and 71.4%, respectively, which were the highest among the low independence group (p = 0.034, results not shown).
Having salaried employment was the main item in which only the latter-stage older people showed a signi cant relationship. The percentage of salaried employees who became high independence three years later was the highest at 72.2%, whereas those who were employed without a salary who became low independence reached 36.5%, the highest among the low independence respondents. The results of the exercise frequency of the latter-stage older people showed that the percentage of those who exercised three times or more per week with enjoyment and ful llment and were identi ed as having high independence three years after baseline was the highest at 69.1%; those who exercised two or less times a week without enjoyment and ful llment and had low independence was 45.0%, the highest among the low independence respondents.
Relationship between dietary intake frequency and independence three years later Table 2 shows the results of the relationship between baseline dietary intake frequency and independence three years later, with a comparison between the high and low independence groups. Overall, the high independence group was signi cantly higher in the intake frequency of each food group as compared to the low independence group. However, the percentage of those who ate meat every day and became low independence three years later was high among the low independence group (overall result: 32.7%), and it did not differ signi cantly from those who ate meat 1-2 days a week (32.7%) or did not eat meat (32.8%). The ordinal scaled scores were examined by Mann-Whitney U test . * p<0.05, ** p<0.01, *** p<0.001

Dietary patterns
Principal component analysis identi ed three components that explained 59.4% of the total variance ( Table 3). The rst component was characterized by a dietary variety that showed that all food groups had strong relations with a high frequency of eating throughout the week. It was described as "dietary diversity." The second pattern was a negative relationship for vegetable, fruit, and dairy products, which was called "low fruit, vegetable, and dairy product frequency." The third pattern was a negative relationship with soy products and blue sh, and a strong positive relationship with a high frequency of meat dishes during the week (principal component loading: 0.660); it was called "high meat frequency." Comprehensive analysis of low independence three years from baseline   Logistic regression analysis was conducted. OR odds ratio, CI con dence interval, ref refference, * p<0.05, ** p<0.01, *** p<0.001 Model 2,363 The analysis by sex and age group revealed that earlier-stage older men showed a signi cantly high OR for 18.5 ≤ BMI < 25 (OR = 0.25, 95% CI: 0.08-0.82) and 25 ≤ BMI < 30 (OR = 0.28, 95% CI: 0.08-0.93). The OR for having a BMI < 18.5 or a BMI ≥ 30 did not differ signi cantly from one another (OR 0.97, 95% CI 0.12-7.61). The OR for community and volunteer activity engagement (OR = 0.44, 95% CI: 0.33-0.60) was found to be signi cantly low.
The results also indicated that latter-stage older women showed a signi cantly low OR for no fall fractures in the previous year (OR = 0.43, 95% CI: 0.21-0.88) and subjective health (OR = 0.54, 95% CI: 0.30-0.99). The OR for high meat frequency in the latter-stage older women was signi cantly higher in all results, regardless of sex or age (OR = 1.59, 95% CI: 1.17-2.15).

Discussion
The present study aimed to clarify how multidimensional factors, especially dietary patterns, affect a decline in independence after three years. This study was a nationwide longitudinal survey of communitydwelling independent older people in Japan. Our results revealed commonalities and distinctions by sex and age stage. Dietary patterns were categorized as dietary diversity, high meat frequency, and low fruit, vegetable, and dairy product frequency.

Evaluation of low independence after three years
In this study, low independence was de ned as a below-average IADL score of the survey respondents. The mean IADL score after three years was signi cantly lower than that at baseline. Of the percentage of low independence respondents three years later (30.4%), 12.5% changed from high to low independence. This result was higher than the percentage of those who changed from low to high independence (8.7%). However, the changes in identi cation from low to high independence suggest that it is possible to maintain or improve independence at this stage. Effective health care support based on the needs of older people is important.

Relationship between dietary patterns and low independence three years later
In our multivariate analysis, dietary diversity appeared to prevent low independence after three years.
Consuming various types of foods leads to a varied nutritional intake, which is related to maintaining independence in old age. Additionally, there are many opportunities for a well-balanced combination of dishes [19], which is associated with frailty prevention [7].
In a study in the United States [10], older people with a high score for three healthy dietary patterns had a higher ratio of monounsaturated fatty acids, such as olive oil, than saturated fatty acids, such as meat, a high intake of vegetables, moderate alcohol, and a low intake of processed and unprocessed meat and salt. According to a report from Taiwan [20], the dietary pattern related to frailty prevention constituted vegetables and fruits with high antioxidant capacity, teas, sh rich in n-3 fatty acids, and other highprotein foods, such as seafood and dairy products, and whole grains. Common to these reports is a healthy dietary pattern with a low intake of meats and saturated fatty acids.
In this study, high meat frequency had a signi cantly high OR for low independence after three years in all analyses, except for older men. The results of this study support the ndings of previous studies that revealed the relationship between frailty and diet in older people [7,10,11,20].
Diets with a high intake of animal protein, such as meats rich in saturated fatty acids, and a low intake of blue sh, which is rich in n-3 fatty acids, and soy-based foods with vegetable protein, may promote a decline in independence. Meat may play a key role in the relationship between in ammation and low independence in older people. The Dietary In ammatory Index (DII) [21] is an index created by Shivappa and colleagues that scores 45 types of foods and nutrients based on their in ammatory properties according to a review of 1,943 research papers. The DII scores have been validated with in ammatory markers, such as blood C-reactive protein (CRP), which is an indicator of chronic in ammation [21]. Many studies have reported that older people on a high pro-in ammatory diet with high DII scores are associated with disability or death [22]. Those on a moderately pro-in ammatory diet were associated with frailty and were independent of obesity [23].
The DII score of saturated fatty acids, which are abundant in meat, is 0.373. This is the highest possible score and is evaluated as the highest pro-in ammatory nutrient [21]. Conversely, the n-3 fatty acids contained in blue sh are − 0.436 [21]. Using the National Health and Nutrition Survey Data of 2,572 Japanese adults, a positive association was found between DII scores and CRP, and the participants in the group with a low DII score consumed a diet rich in vegetables, fruits, seafood, and beans, and their intake of meat and cereals was low [24]. The higher the DII score, the higher the likelihood of observing high meat intake in Japanese adults. The results of the present study showed that the frequent intake of meat in older people, especially older women, was related to a decline in independence after three years, independent of multidimensional frailty and adjusted factors. These results may be explained by the proin ammatory nature of saturated fatty acids in meat.
In recent years, high protein intake by older people has been recommended as a support for frailty prevention. Meat from mammals, compared to seafood and soy-based foods, has better micronutrients against aging, such as the arachidonic acid of an n-6 fatty acid [9], and zinc, which helps prevent cognitive decline. The recommendation of eating moderate amounts of meat along with other variety of food is necessary for older people and should be included in nutrition education.

Comprehensive analysis of low independence after three years
A single intervention with nutrition for frailty may have little effect. There may be stronger results from an intervention that combines physical activity, nutrition [12], and cognitive function [13]. However, the effects of multiple interventions focusing on dietary patterns have not been fully clari ed. For older women, who have a longer life expectancy than men and a large difference between healthy life expectancy or not, it is critical to standardize effective support. To that end, it is important to clarify relevant frailty factors comprehensively by sex and age stage and accumulate evidence to create an effective integrated support program. The current study is among the rst to evaluate the multifactorial association among a decline in independence, dietary patterns, and frailty factors in earlier-and latterstage older people, individually.
The results of this study showed that an appropriate BMI was associated with the prevention of a decline in independence in earlier-stage older men. Earlier-stage older people were determined between 65 and 75 years old, suggesting the necessity of continuing measures, such as weight management, to prevent chronic lifestyle diseases and aggravation. Additionally, the OR of obesity with a BMI of 30 or more in latter-stage older people in the present study was signi cantly lower, at 0.09, and the effect on a decline in independence was smaller than the thinness of a BMI of less than 18.5. However, in a study of Japanese participants, obesity in latter-stage older people was signi cantly higher with an OR for the loss of independence of 1.4 [2]. These results suggest that latter-stage older people, both lean and obese, should be considered in measures to support frailty.
Our results of the latter-stage older people suggest that subjective health and exercise frequency with enjoyment and ful llment may be priority support items for preventing a decline in independence after three years. Furthermore, social participation, such as community and volunteer activity engagement, could contribute to its prevention for both earlier-and latter-stage older people. The vulnerability of mental and psychological factors in older people is one of the multifaceted problems associated with frailty. Particularly, this study suggests that consideration of mental and psychological factors must be included in the support offered to latter-stage older people.
A cross-sectional study of community-dwelling older people in Hong Kong reported that a Mediterranean diet, one's living space, and social participation opportunities were associated with risk reduction for frailty [25]. This result suggests that dietary lifestyle in older people may be determined by their lifestyle. A decrease in food intake in independent older people was signi cantly related to a decrease in the outing frequency of less than once a week with an OR of 2.0 [26]. Therefore, it is necessary to provide environmental support to facilitate older people's exibility and ability to walk around their residence. Likewise, older people may go out more.
Hoshi [27] clari ed the causal structure of multiple factors associated with extending a healthy life expectancy. The results of the analysis model show that there is an indirect relationship between socioeconomic status (SES) and healthy life expectancy, mediated by the effects of environmental status (green living environment), and mental, physical, and social health. In this study, multivariate analysis did not show any signi cant correlation with SES-related satisfaction. However, the results on the subjective economic status of the earlier-stage older people revealed that the percentage was highest for "satis ed" (76.2%) in the high independence group, whereas "not very satis ed" (31.9%) was the highest in the low independence group.
Therefore, effective support for older people living in communities needs to account for SES. Creating a multifactorial program for an integrated support, designed by sex and age stage, might be more effective.
Moreover, the use of resources other than specialists has been reported [28]. For example, utilizing earlier-stage older people as leaders in volunteer activities in a frailty prevention program for the latter group may present a model of support in the future.
We should consider several limitations of this study. First, we evaluated the respondents' independence using the validated TMIG-IC, which asked for subjective information on the participants' ability to engage in IADL. Converting the analysis outcomes into a numerical index, such as for healthy life expectancy, which is calculated by a period of not needing nursing or support [29], and identifying various factors related to a reduction in healthy life expectancy will strengthen the evidence found in this study. More studies are needed to address these issues.
Next, the results were based on the frequency of food group consumption. Although the BMI related to total intake was added to the analysis in the multivariate-adjusted model, it was not the result calculated based on quantitative intake. Future studies should further examine the intake of foods and nutrients and perform quantitative veri cation.

Conclusions
Our results suggest that the priority factors of effective and integrated support for maintaining independence after three years are associated with the prevention of diseases and aggravation. It mainly includes weight management for earlier-stage older men, mental and psychological factors related to subjective health and enjoyment and ful llment, especially for latter-stage older people, and low consumption of meat for latter-stage older women.