Temporal change in the association between life satisfaction and functional decline with gender differences: an age-specific prospective cohort study

Background Although life satisfaction (LS) has been shown to predict mortality, research studying the relationship between LS and functional decline is scarce. This study examined the association between LS and functional decline across four time points in young-old Japanese adults. Methods We analysed 1,899 community-dwelling 65-year-olds in this age-specific cohort study conducted between 2000 and 2005. The Life Satisfaction Index K was used to evaluate LS and was classified into quartiles. Functional decline was determined using the Japanese Long-Term Care Insurance (LTCI) system: 1) mild disability; 2) severe disability; 3) all-cause mortality; 4) mild or severe disability; 5) severe disability or death; 6) mild or severe disability, or death. Adjusted hazard ratios (HR) with 95% confidence intervals (CI) were calculated using the Cox proportional hazard model. The analyses were conducted in the 8th, 10th, 12th, and 14th years to assess the effect of LS on functional decline across time points. Results The impact of LS gradually weakened over time. In the 8th year (aged 72–73), a higher LS was associated with a lower risk of mild or severe disability among the women participants (adjusted HR [95% CI] = 0.30 [0.11–0.81]). However, the effect disappeared gradually (adjusted HR [95% CI] = 0.55 [0.27–1.14]) in the 10th year (aged 74–75), 0.72 (0.41–1.26) in the 12th year (aged 76–77), and 0.68 (0.41–1.14) in the 14th year (aged 78–79). This trend continued in severe disability or death (adjusted HR [95% CI] = 0.24 [0.06–0.70], 0.31 [0.11–0.76], 0.57 [0.28–1.14], and 0.60 [0.32–1.12]) and mild or severe disability, or death (adjusted HR [95% CI] = 0.30 [0.14–0.68], 0.46 [0.24–0.87], 0.67 [0.41–1.10], and 0.65 [0.42–1.02]) in the 8th, 10th, 12th, and 14th years, respectively. No statistically significant association was found among men at any time points or in any classification of outcomes. Conclusions Higher LS scores in 65-year-old women were associated with a lower risk for functional decline in any combination of mild disability, severe disability, or death. Additionally, the effect of LS was observed to weaken over time. Trial registration This is not an intervention survey and does not require registration.


Background
Globally, the rate of ageing has increased [1]. Japan has the highest rate of ageing worldwide, with 28.4% of older adults aged ²65 in 2019 [2]. However, it has the second highest average and healthy life expectancies globally [3]. Because of the difference between their average and healthy life expectancies, older Japanese experience functional disabilities for an average of 8.84 years in men and 12.35 years in women [2]. Functional decline has been found to be caused by cardiopulmonary diseases, neurological conditions, diabetes mellitus, cancer, obesity, dementia, affective disorders, ophthalmologic and auditory disorders, and fractures [4]. Additionally, social and financial support and physical environment have been shown affect functional decline [4].
Positive psychology, initially investigating the behavioural actions leading to well-being for all ages, has received attention in the field of psychology for several decades [5]. Meanwhile, well-being has been considered an essential factor for health-related outcomes in the ageing population [6]. Previous studies have used various tools to evaluate older people's subjective well-being. These metrics include life satisfaction (LS), hedonic well-being, and eudemonic well-being [7] and have been linked to mortality among the geriatric population [8][9][10][11]. LS represents the evaluation of their life spans from the past to the present; hedonic well-being represents current feelings, including positive and negative ones; eudemonic well-being represents individuals' perspectives ranging from the present to the future [5]. Many studies have demonstrated an association between functional decline and both hedonic [12][13][14] and eudemonic well-being [15][16][17][18]; however, studies regarding the association between functional decline and LS are limited. Although ageing does not always worsen one's view of life [19], older people may experience the ageing process through events such as retirement, deterioration of health, and the loss of a spouse [20], which can affect LS. Previous studies have reported that older people show a decrease in LS over time [21][22][23]. However, the subsequent impact of LS on functional decline was unknown. Since LS was described as a variable indicating an evaluation of life mainly in the past, its effect may gradually weaken. However, to the best of our knowledge, there are no studies concerning the chronological effects of LS and its functional decline among older adults. Furthermore, women are more likely to have a disability than men. Few studies have examined the impact of LS on disability according to gender, although social determinants such as education and economic status may be responsible for this difference [24].
Regarding the effect of LS on functional decline, a cross-sectional study from Poland reported the association between LS and the prevalence of activities of daily living (ADL) and/or instrumental activities of daily living (IADL) disabilities in older inhabitants [25]. Furthermore, a prospective cohort study demonstrated that poor LS was among the strongest risk factors for declining IADL in older Norwegians [26]. Previously, we reported that low LS for 64-year-old women was a significant risk factor for the decline of the Tokyo Metropolitan Institute of Gerontology index of competence (TMIG-IC) after six years [27]. The ADL mentioned above, the IADL or TMIG-IC, merely demonstrated a single aspect of functional decline. Regarding the effect of LS on mortality, a previous study reported Finnish men aged 18-64 with high LS had a significantly lower mortality rate than women aged 18-64 with high LS [28]. The multinational World Health Organization (WHO) monitoring trends and determinants in cardiovascular disease (MONICA) project demonstrated that men aged 25-74 with higher LS have a substantial longterm survival benefit than women [29]. Further, a Korean study showed a lower hazard ratio in reducing mortality among adults 55 years of age or older [30] with high LS. These results suggest that LS affects mortality independent of age, gender, or nationality.
This study aimed to comprehensively evaluate the association between LS and functional decline using the cer-tification of the long-term care insurance (LTCI) system across four time points to investigate a potential chronological effect of LS.

Study population
The New Integrated Suburban Seniority Investigation (NISSIN) Project is an ongoing age-specific prospective cohort study [31]. The baseline investigation was conducted from 1996 through 2005 in Nisshin City, near Nagoya, Japan. It comprised 3,073 community-dwelling residents who would be 65 years old in the target year (1,548 men and 1,525 women, with an overall response rate of 43.9%). Of the 3,073 participants, two were excluded because they relocated before this cohort commenced. We excluded those who joined in 1996-1999 because they did not meet the inclusion criteria; this study focused on the LTCI system certification launched on the 1st April 2000 (n = 1,105). One participant was excluded due to the certification of long-term care before the beginning of this cohort. Additionally, we excluded the participants who had missing data regarding LS (n = 66). The final sample included 1,899 participants (977 men and 922 women), whose data were used for the analyses (Fig. 1).
This study was approved by the ethics committees of the Nagoya University Graduate School of Medicine, the National Center for Geriatrics and Gerontology of Japan, the Aichi Medical University School of Medicine, and the Hokkaido University Graduate School of Medicine. All the participants provided informed consent, with oral informed consent provided through an opt-out approach until 2001 after which written consent was provided through an opt-in approach.

Life satisfaction
LS was evaluated using Life Satisfaction Index K (LSIK) at baseline. It is a self-administered questionnaire comprising nine items under three sections: 'satisfaction with the whole life', 'evaluation of ageing', and 'psychological stability' [32]. A single point was assigned for each selec- Excluded: Participants at the baseline survey from 1996 to 1999 (n=1,105) Participants who were certificated before the start of this cohort (n=1) Excluded: Participants who missed Life Satisfaction scores (n=66) Excluded: Participants who relocated before the start of this cohort (n=2) n=3,071 tion of a positive option and zero points were assigned for selecting a negative option. The highest possible total score was nine, with greater total scores indicative of greater degrees of LS.

Covariates
The covariates obtained through the self-administered questionnaire included years of participation; lifestyle variables including smoking status (never, former, current smokers), alcohol consumption (occasional, current drinkers), exercise habit (seldom, <1/week, ²1/week); sociodemographic factors, including educational attainment (high school or lower, junior college or higher), marital status (married, other), and social activity (<10, ²10 points). Social activities were defined as activities that required contact with society and were composed of four major facets measured using 21 questions related to professional, social, learning, and personal activities [33]. The highest possible total score was 21; higher scores indicated a higher degree of social activity. As the distribution was skewed, social activity was divided into two groups based on the median. Comorbidity variables included selfreported hypertension, diabetes mellitus, dyslipidaemia, cardiovascular disease, cerebrovascular disease, and cancer. TMIG-IC, a validated questionnaire ranging from 0-13 points, was used to access functional capacity [34]. Stress was measured by asking, 'Do you feel stressed?' which could be answered using three alternatives: always, sometimes, and seldom.

Outcomes
The outcomes measured were functional decline, defined LTCI certification, and mortality. The participants were followed until relocation, received LTCI certification or they died from all causes. LTCI certification dates, deaths, and relocations were identified using the resident registry by the public health nurse in Nisshin City year-round.
Japanese residents aged ²65 obtained long-term care certifications by applying to municipalities and receiving approval from the long-term care board that such care was needed based on their screening results [27]. The evaluation covered seven levels according to the physical or mental disorders, namely, support levels 1-2 and care levels 1-5. The higher the level of care, the more severe the functional status. At support level 1, the patient is independent with ADL with some assistance required to perform IADL; care level 5 is described as the patient requiring complete assistance to carry out ADL and being unable to function without it. Detailed information regarding the LTCI system has been described in previous studies [35][36][37]. Based on previous research, we divided the outcome variables into six groups: 1) mild disability (support levels or care level 1); 2) severe disability (care levels 2-5); 3) death from allcause mortality; 4) mild or severe disability (support levels or care levels); 5) severe disability or death (care levels 2-5 or death); and 6) mild or severe disability or death (support or care levels or death). We considered care levels 2-5 to have severe functional declines because participants with care levels 1-5 had lower ADL abilities than those with support levels 1-2. For consistency over the follow-up period, the former care level 1 was divided into support level 2 and care level 1 in 2008 [36].

Statistical analysis
LS was divided into four groups according to its distribution: 0-3, 4-5, 6, and 7-9 points. Person-years were calculated from baseline to the date of relocation, LTCI certification, death from all causes, or the end of each time point. Cox proportional hazard regression models were used to estimate the hazard ratio (HR) and the confidence interval (CI) of functional decline and mortality was set at 95%. As gender differences were found in the incidence of functional decline, we also analysed the data stratified by gender [38,39]. The analyses were adjusted for year of participation, smoking status, alcohol consumption, exercise habits, educational attainment, marital status, social activity, comorbidity variables, functional capacity, and stress. The analyses were conducted for four time points in the 8 th , 10 th , 12 th , and 14 th years from each baseline year to evaluate the impact of LS over time (Fig. 2). LS performed a test of trends as continuous variables. All the pvalues evaluated were two-sided, and values less than 0.05 were considered statistically significant. Data analysis was performed using JMP Pro 14.0.0 (SAS Institute Inc., United States of America).

Results
Tables 1 and 2 show the distributions of the baseline characteristics based on gender. Men and women with higher LS were more likely to participate in social activities, have a higher functional capacity, and experience less stress; the men had a higher level of education and a lower prevalence of hypertension and diabetes mellitus; married women exercised more frequently than married men. Tables 3 and 4 show the relationship between LS, longterm care certification, and death stratified by gender. The events gradually increased among men and women at each point of the 8 th , 10 th , 12 th , and 14 th years. The women in the higher LS groups were prone to have lower risks of LTCI certification in all the categories than those in the lowest LS groups in all four time points. The second highest LS group was significantly associated with a lower risk of mild or severe disability compared to the lowest LS group. However, this trend showed a yearly U-shape. Across the

Discussion
Herein, we found that high LS was associated with a lower risk of functional decline based on the LTCI system; however, the impact of LS weakened gradually over time among women while no such association was reported among men. This was the first study of an age-specific cohort to demonstrate LS's chronological impact on classified functional decline among the young and old populations stratified by gender. Although our previous study investigated LS and functional decline in gender stratification, its results were limited due to a single cut-off point for functional decline [27]. It did not follow the association between functional decline and the chronological impact of LS. The findings showed that LS could be an indicator expressing the risk of functional decline regardless of mild disability or severe disability for women; furthermore, the impact of LS may be gradually weakened among women. Some previous studies have explored the relationship between social support, economic status, health, and LS [40,41]. LS reflects many aspects of people's lives and vice versa. High LS may improve their lives and prevent older people from worsened health conditions and dying.
Among women, the effect of LS on functional decline diminished over time. LS is known to reflect an individual's well-being from the past to the present [5]. The effect of showing well-being may weaken over time as it gets further from the baseline. Furthermore, older adults find it difficult to have the opportunities and resources to live positively because of their decreasing life span. They feel less motivated moving forward while focusing on maintaining health and preventing losses [22]. In a 15-year longitudinal study, more decline in social participation over time and less perceived support at the end of life indicated a more severe terminal decline in LS among older people [21]. Consequently, the effect of LS may decrease the impact of preventing functional decline over time.
We found that the second highest level of LS is likely lower the subsequent risk of functional decline and subsequent death in women by 70% in the 8 th year. Women tend to live longer on average and have a lower risk of suffering from fatal diseases than men; conversely, older women suffer from higher rates of functional disability than men [42]. One possible mechanism that prevents functional decline in women with high LS is gender difference in social roles. The global concept of gender roles is based on the division of labour in the family [43]. Traditionally, women perform more housework, cooking, and nurturing than men. Therefore, women can prepare meals to maintain their well-being, leading to healthier diets than men [44]. Research has positively linked LS with good nutrition [45]. We observed differences in our participants' food intake based on gender; although women consumed more vegetables and fruits than men, men ate more rice than women (data not shown). Previous studies reported that when older adults actively support their families, such as by frequently caring for grandchildren, their LS is positively influenced [46]. These household responsibilities are considered the primary preventive factors for maintaining LS by women in Japan [47], thus leading to this result.
Moreover, women may use social networks well. In our study, participants with higher LS indicated higher social activities. Higher LS led older people to participate in social activities, subsequently increasing their physical and mental activities, which are protective factors for disability [48]. Having many friends and engaging in many social activities enhanced women's LS; however, men were more satisfied with life when living with their spouses [49].
Despite not being stratified by gender, our results were consistent with those reported in previous studies that demonstrated a protective effect of LS on functional decline in Norwegian adults aged 60-69 [26] and Polish adults aged 71-80 [25]. Our age-specific cohort research design avoided the wide age range and showed the benefits of LS for subsequent functional decline in young and old women. Furthermore, the tools employed in previous studies, TMIG-IC [27], ADL [25,26], and IADL [26], can evaluate older people's functional ability. In our study, by contrast, we defined the Japanese LTCI system and assessed it comprehensively using classified information that covered physical or mental disorders (support levels 1-2 and care levels 1-5). We divided the data into six groups and analysed it to determine the association between LS and functional decline and to follow four time points to show the chronological effect of LS. Higher LS scores in women were associated with a lower risk for functional decline in any combination of mild disability, severe disability, or death, with the effect of LS weakening over time across the four time points.
In this sophisticated society, our study provides evidence for establishing a healthy ageing community to decrease the risk of functional decline besides the traditional risk factors by maintaining a higher LS. From a health behaviour perspective, lower LS is likely to impact its chronological effect and further lead to functional decline or mortality. LS is positively associated with health behaviours-such as not smoking, engaging in physical exercise, using sun protection, eating fruits, and limiting fat intake-known as protective factors for well-being [50].
This study's strengths included the following: first, we targeted participants who attained 65 years old at the beginning, (i.e., the onset of old age), as the incident rate of disability has risen exponentially with age, and followed up on this population. These results allowed us to eliminate the confounding factor of age. Second, our study enabled the examination of LS by assessing functional decline Environmental Health and Preventive Medicine (2023) 28:42 in detail according to the classification of the Japanese LTCI, which in previous studies was measured using only a single criterion.
Despite the strengths, there were some limitations associated with our study. First, the statistical power may not be sufficient due to the relatively small sample size, low functional decline, and death incident rate. Nisshin, the city where this research was conducted, is located in the Aichi prefecture, which has a higher healthy life expectancy than other prefectures [51]. Thus, this sample may have resulted in lower incidence of functional decline. Second, the LTCI system provided in Japan is a unique healthcare service; therefore, we cannot directly generalise this study's results to other parts of the world. There are specific steps involved in the registration of functional decline by LTCI: apply, examine, and record [52]. This process must be carried out first by the older adults, their families, or care managers. However, because of the traditional Japanese concept of older adult care, which states that older people should be tended to by their families [53][54][55]. Unreported cases, those who are cared for by their families but not reported to the public office, may cause underestimation. Third, we could not follow relocated people, which were recorded as lost to follow-up. Although the number of relocated people was small (n = 142, 7.5%), we found that the relocated people had significantly lower LS scores than the participants. It could lead to an underestimation of the study's results.

Conclusions
In conclusion, our findings suggested that LS might be an indicator measuring the risk of functional decline, especially during severe levels or death. At the same time, the impact of LS may decrease chronologically in Japanese young and old women. We believe the findings in this study provide new insights and useful parameters for older adults living in communities in Japan for future social medicine, hygiene, and preventive medicine.