Poor self-rated health predicts the incidence of functional disability in elderly community dwellers in Japan: a prospective cohort study

It has been reported that poor self-rated health (SRH) is an indication of underlying physical and mental abnormalities that are often difficult to detect through other measures of health such as body weight and blood pressure [1]. Previous studies have shown that poor SRH is associated with mortality [2], morbidity (e.g., cardiovascular disease) [3, 4], cancer [5], musculoskeletal disease [6], psychiatric diseases [7], and dementia [8].

The proportions of elderly people over the age of 65 years in populations have recently been increasing [9]. The government and private sector will both need to deal with these situations by addressing new approaches to social security. Japan has achieved the highest longevity ratings in the world (81 years of age for men and 87 years of age for women) [10], and the proportion of people aged 65 years or older is estimated to be 35.6 million (28.1% of the population) [11]. By 2042, it is predicted that the number of elderly people in Japan will reach 38.9 million [12]. There is an ongoing issue about sustaining senior health using social security in order to balance the financial burdens of pension and health care expenditure [13].

In 2000, the Japanese government implemented a long-term care insurance (LTCI) system for elderly individuals [14]. In 2016, the number of individuals using this service was approximately 6.3 million, which was 2.9-times larger than the number of initial certificates in 2000 [15]. Since the number of candidates is increasing, there is an urgent need to take measures of prevention in order to aid disability and intervene for those at high risk for various diseases, which in turn would prevent an increase in the financial burden.

Various endpoints including activities of daily living (ADL), basic ADL, LTCI, and independent criteria were used in previous studies. The studies mentioned above had small numbers of subjects (subjects ranging from 165 to 737) [16‐18] except for one study [19]. Furthermore, the follow-up periods of the surveys conducted in those studies were also relatively short. The follow-up period in one study was only 1 year [18], while the follow-up periods after the baseline surveys were 5 years in the other studies [16, 17, 19]. To the best of our knowledge, there has been no large-scale prospective study on the association between SRH and functional disability in elderly people using LTCI with a long follow-up period (10 years or more).

Previous large population studies in Western countries have shown that elderly people with poor SRH are at a high risk for functional disabilities [20‐23]. Poor SRH has also been shown to be strongly associated with mortality and functional decline in Japanese elderly residents [16‐19]. The results of those studies indicate that SRH can be used for predicting future functional disabilities in Japanese.

Therefore, the aim of this study was to determine whether poor SRH increases the risk of functional disabilities in elderly Japanese people utilizing the certification of LTCI for 10 years.

The research plan was approved by the Ethics Committee of Iwate Medical University Institute Review Board #1 (approval no. H13–33).

The median follow-up period was 10.5 years (inter-quartile range, 10.1–12.4 years). During follow-up (112,490 person-years), 3377 participants were authorized as new recipients of LTCI for functional disability (1245 men and 2132 women, 3.3% per year). The median period to LTCI certification was 7.1 years (inter-quartile range, 4.3–9.3 years).

Baseline characteristics in the SRH groups are shown in Table 1. There was no significant difference in mean age among the SRH groups for both sexes. In men, SRH was worse in participants who were non-employed or retired, single, and never drinker and who had a low frequency of regular exercise, short sleep duration (≤ 6 h), long sleep duration (≥ 9 h), hypertension, and diabetes mellitus. Worse SRH was also related to higher levels of HbA1c and non-HDLC, lower levels of DBP and Hb and lower eGFR. In women, SRH was worse in participants who were having a job, participants who never smoked, and participants who had a low frequency of regular exercise, hypertension, and diabetes mellitus. Furthermore, worse SRH was related to a high level of HbA1c, low levels of TC and non-HDLC, and low eGFR.

Table 2 shows the hazard ratios of functional disability among the SRH groups. The percentages of participants with functional disability increased with worsening of SRH (percentage in men: good SRH group, 25.9%; rather good SRH group, 28.4%; neither good nor poor SRH group, 29.6%; poor SRH group, 42.2%; percentages in women: good SRH group, 27.6%; rather good SRH group, 30.7%; neither good nor poor SRH group, 35.1%; poor SRH group, 47.7%). HRs of functional disability were found to be significantly higher in the neither good nor poor SRH group, and poor SRH group for men (HR [95% CI] in men: neither good nor poor SRH group = 1.21 [1.02, 1.45], poor SRH group = 2.07 [1.69, 2.54]) (Additional file 1). In women, HRs were significantly higher in three SRH groups (HR [95% CI]: rather good SRH group = 1.15 [1.03, 1.28], neither good nor poor SRH group = 1.39 [1.21, 1.59], and poor SRH group = 2.13 [1.83, 2.48]). After adjustments for covariates were made, these differences disappeared in the neither good nor poor SRH group for men but remained statistically significant in the other groups in Model 2. There were significant linear trends between the SRH groups and functional disability for both sexes (p < 0.001). In men, HRs for functional disability were found to be significantly higher in participants who had a high HbA1c level, who were non-employed or retired, who had a single marital status, and who had a long sleep duration (≥ 9 h), while HRs for functional disability were found to be significantly lower in participants with a high Hb level and high eGFR and who were drinkers 1–4 days per week. In women, HRs for functional disability were found to be significantly higher in participants who had a high level of HbA1c, who were non-employed or retired, who had a single marital status and who had a low frequency of physical exercise (< 1 h per week), while HRs for functional disability were found to be significantly lower in participants with a high eGFR. Figure 3 showes the cumulative survival curve of functional disability-free rates in the SRH groups. The differences in cumulative probability of functional disability-free rates among the SRH groups were significant (p < 0.001). Utilizing time-dependent Cox regression analysis with incorporation of the interim CV diseases, the HR for functional disability showed significant increases for men in the poor SRH group and for women in three SRH groups (Additional file 2).

The HRs for the risk of functional disability were further compared in the SRH groups with and without adjustments for objective indicators (Additional file 3). The HRs for the risk of functional disability remained significantly high in the poor SRH group in both models. After excluding sub-clinical cases, a significant association was found only in the poor SRH group for men, but significant associations remained in all three SRH groups for women (Table 3). In analysis of severe functional disability, the HRs for men were attenuated in all three SRH groups and a significant association was found for the poor SRH group. The HRs for functional disability for women were attenuated, but significant associations remained in the worst group. The linear trends show significant associations for both sexes in these analyses.

In the present study, SRH in the baseline survey was found a strong predictor of functional disability by utilizing certification of LTCI in a community of Japanese elderly dwellers. A linear relationship was found between the incidence of functional disability and SRH. In addition, poorer SRH group was significantly higher risk for functional disability in analysis excluding participants who were functionally disabled within the initial 2 years of the survey. Furthermore, the association between SRH and functional disability remained significant in analysis of the endpoint for severe cases of functional disability.

Most of the previous studies on the relationship between poor SRH and functional disability were conducted in Western countries [20‐23, 33]. The results of our study are in line with the results of studies in Western countries. Several studies have shown a relationship between SRH and the development of functional disability in Japan. Haga et al. demonstrated an association between SRH and ADL that were rated on the basis of five items (e.g., walking and dressing) [34]. Hirosaki et al. revealed an association between SRH and functional ability that was evaluated on the basis of basic ADL using seven items in community-dwelling Japanese elderly adults [17]. Compared with our study, these studies had a smaller sample size (124 in Haga’s study, and 654 in Hirosaki’s study). Additionally, the endpoints in those studies (ADL and basic ADL) were different from the endpoint in this study although the focus in both of those studies had similarities with the focus of the current study in terms of elucidating the association of SRH with functional disability. In the present study, functional disability as the endpoint was defined according to the levels of the abovementioned LTCI items (from support levels 1 and 2 and care need levels 1 to 5). People in support level 1 were defined as “those with limited ability to perform instrumental activities of daily living but remaining independent in basic activities of daily living.” [25]. This endpoint is compatible with the endpoint of Lee’s study and Hirai’s study in which functional disability was defined as disability including IADL limitations. Lee investigated the extent to which elderly people’s self-assessments of general health including SRH predicted functional decline in the US Longitudinal Study of Aging (LSOA) [21]. The odds ratios of functional decline were significantly higher in those who responded good, fair, or poor for SRH than in those who responded excellent (odds ratios [95% CI]: good SRH group = 1.43 [1.16, 1.76] and fair/poor SRH group = 1.56 [1.20, 2.03]). Although analysis with stratification by sex was not performed in Lee’s study, the classification of SRH (which was divided into four categories) was compatible with that in our study, and the magnitude of risk for functional disability was similar to that of our results. Hirai et al. examined risk factors for certification of LTCI in community dwelling elderly [19]. SRH was significantly related to increased risk of onset of LTCI in persons in support level 1 or more (HR [95% CI]: men in the poor SRH group = 2.79 [2.25, 3.47]; women, in the poor SRH group = 2.28 [1.91, 2.73]). Although the endpoint in Hirai’s study was exactly the same as that in our study, the classification of SRH (which was divided into two categories [good vs. poor]) was not compatible with that in our study. Furthermore, the HRs of risk for functional disability were higher in Hirai’s study than in our study possibly due to dichotomizing rather poor and poor as a poor SRH group. Our results suggest that poor SRH could be a predictor of functional disability even with some IADL limitations. Since SRH has also been recommended as a tool for disease risk screening [1], our results will be important role for estimation of the risk of functional disability in a general population.

The strength of this study was that we showed an association between SRH and functional disability in a large-scale prospective study in elderly community dwellers in East Asia over 10 years. Additional analyses such as analysis with exclusion of early cases and severe cases in endpoints improved the validity of our study. In addition, we showed an between SRH and functional disability even when adjustments were made for association objective indicators.

The results obtained by using the two models indicated that the association between SRH and functional disability was explained to some extent by confounding factors such as clinical indicators, social factors, and lifestyle, but there might have been residual confounders such as household income and the existence of mental disorders that were not incorporated in the models. Unfavorable social risk factors (non-employed or retired and single marital status) and unhealthy objective indicators (low Hb level, low eGFR, and high HbA1c level) contributed to the higher risk of functional disability. These results are consistent with the results of previous studies [35‐37]. People with unfavorable social status and unhealthy status might conceive their health as being poor, which consequently lead to be functional disability.

CV diseases are well-known risks for future functional disability [38‐41]. The main reasons for receiving LTCI were cerebrovascular disease (stroke) (21.5%) followed by dementia (15.3%) [42]. Several studies have demonstrated that poor SRH is a consistent predictor of the incidence of several type of CV diseases such as stroke, coronary artery disease, and heart failure [4, 43, 44]. Based on these facts, we initially hypothesized that poor SRH related to functional disability was mediated by physical dysfunctions caused by CV diseases. However, HRs of the incident risk for functional disability did not change in time-dependent Cox regression analyses adjusted for interim CV diseases. Another possibility for the higher incidence of functional disability in people with poor SRH is the influence of dementia. Poor SRH was shown to be a risk for incident dementia [8], and decline in cognitive function was shown to be associated with LTCI after an 18-month observation using a cognitive performance scale [45]. Lin et al. reported that people with dementia who lived alone had a higher risk of care needs than did those without dementia [46]. Although we were not able to evaluate the causes of functional disability due to a lack of information about cognitive dysfunction, we speculate that dementia relates to functional disability in elderly adults with poor SRH.

Previous studies have shown that poor SRH is associated with objective indicators such as serum cholesterol and HbA1c levels [47, 48], and studies have shown that cardiovascular risk factors are associated with functional disability [49, 50]. In most previous studies, only subjective indicators such as self-reported information or interview-based information was examined even though these subjective items cannot be avoided by a recall bias [17‐19]. Jylhä et al. mentioned the importance of investigating SRH adjusted for biological or clinical variables [51]. Therefore, adjustment for these clinical indicators gives light to new findings in the relationship between SRH and functional disability. In the analysis in the present study, after adjusting for clinically verified health indicators (such as results of anthropometrical examination and blood tests), poor SRH remained independently associated with functional disability. Even though there may still be residual confounding, e.g., inflammatory biomarkers, that does not allow for full adjustment, we could show a significant association between SRH and functional disability independent of objective indicators.

Though the mechanisms underlying the predictive ability of SRH for functional disability were not revealed in this study, there are some possible explanations of this pathway from poor SRH to functional disability. It is believed that the incidence of functional disability in the present study was affected by physical impairment and also cognitive dysfunction [42]. In pathophysiology, there is an association between SRH and subcortical dysfunction, which are often linked to vascular risk factors [52]. Gray matter atrophy and cortical thinning, as a consequence of microvascular diseases, cause neuronal damage [53]. Neuronal damage might lead to cognitive dysfunction by expressing social vulnerability [52]. In addition, recent studies have shown that SRH is related to several inflammatory biomarkers including C-reactive protein and cytokines [30, 54]. High IL-6 plasma levels are associated with functional impairment in elderly individuals with vascular dementia [55]. A previous study showed that albuminuria predicted future functional disability [29]. Given these findings, poor SRH might be an alternative index of functional disability with physical impairment and cognitive dysfunction caused by asymptomatic brain infarction or endothelial dysfunction.

Poorer SRH was significantly related to higher hazard ratios for future functional disability in both sexes among the Japanese community-dwelling elderly participants. Although it is difficult to accurately predict future functional disability using a single question, SRH might be useful for predicting later functional disability in elderly people as a simple, low-cost questionnaire, thus allowing for identification of high-risk individuals for future functional disability at an early stage.