Self-rated Health as a Predictor of Active Life in the Community Elderly

The present study aims at showing the association of self-rated health with a long active life based on a 7-year longitudinal study of the community elderly. Baseline survey was carried out at home by interviewers in 1983. Ten hundred and twenty-four subjects who were independent in activity of daily living (ADL) at baseline, were followed-up in 1990. Those who were completely independent for every item of ADL at the follow-up were regarded as being in active life. Inactive life was defined as status changed into dependence in any item of ADL, admission to institution or deceased, at the follow-up. In 1990, 62.4% of men and 60.4% of women were defined to be in active life. Percentage with active life was the highest in the good self-rated health, followed by fair and poor in both sexes. A significantly direct relationship of the self-rated health to the active life was observed after controling for age, education, spouse, drinking, smoking, physician visits, history of stroke using a logistic model. The present study shows that self-rated health is a useful predictor of active life. J Epidemiol, 1995; 5 : 11-15.

Previous studies on self-rated health have demonstarated that subjective health assessment could be a valid and cost effective substitute for objective health assessment1-2). It is also known to be closely related to quality of life as assessed by life satisfaction and morale 3). Therefore, selfrated health may provide a useful measurement of successful aging in the elderly.
Most people wish to live a long active life, and fear being dependent in later life. Studies concerning predictors for mortality or physical disability have often been done among the community elderly4-8) . However, factors associated with extending active life, which means complete independence in activities of daily living, has not been thoroughly investigated. Identification of factors influencing active life is important for improving the quality of life for the elderly.
The present study aims at showing the association of self-rated health with a long active life based on a 7-year longitudinal study of the community elderly.

SUBJECTS AND METHODS
The study area, the town of Yuwa in Akita Prefecture, is a rural community located in the northern part of Japan. In 1983 it had a population of about 9,000, of whom 13.9 percent were elderly. Subjects were recruited from elderly adults living at home, aged 65 years and over at the baseline survey in 1983. All residents (n= 1,235) of that age group were regarded as eligible.
The survey was carried out at home by trained interviewers. Eleven hundred and forty two people (92.5%) participated in the baseline survey. Out of the 1,142 Table 1. Subjects of the present studỳ Subjects refused or moved out were excluded from the analysis subjects, 1,024 were completely independent in activities of daily living (ADL) at baseline survey. Those subjects were followed-up for the present study in 1990, 7 years later. The interview was conducted in the same way as in the baseline. Six hundred and ninety nine (68.3%) responded to the follow-up survey at home. Prior to the follow-up survey in 1990, 233 (22.7%) had died. Ninetytwo (9.0%) did not respond to the survey because of admission to hospital or residential care, moved out, or refused to be interviewed (Table 1). ADL was rated by 5 items : walking, eating, toileting, bathing, and dressing, which were classified into independent, needed some help, and dependent. Subjects were recoded into two groups : active life and inactive life. Those who were completely independent for every item of ADL at the follow-up were regarded as being in active life. Inactive life meant status changed from living in the community and being fully independent in ADL at the baseline to any of the following : dependent in any item of ADL at the follow-up, admission to hospital or residential care, or deceased. The definition of active or inactive was based on that of Katz et al.'). Forty-five subjects who had moved out of the town or refused to be interviewed were excluded from the present analysis. Active life was scored 1, and inactive life 0.
In order to assess their self-rated health, subjects were asked "how would you rate your present health ?". They were investigated according to 4 categories, good, fair, poor, and very poor. However, poor and very poor were combined in the analysis because the frequency of very poor was too low, to be analysed separately.
The relationship of self-rated health to active life was tested using a logistic model. Seven control variables investigated at the baseline survey, age, education, spouse, drinking, smoking, physician visit, and history of stroke were entered into the model with self-rated health. Selfrated health and some control variables varied markedly between men and women, and analysis was therefore conducted separately for men and women. Table 2 shows percentages for independence in each category of ADL at the follow-up survey. Subjects maintaining independence for each item of ADL were found to be from 93.9% to 96.6% in men, and from 88.7% to 97.5% in women. There was a significant sex difference for independence in walking. Percentage for independence of eating, toileting, bathing, and dressing did not differ between sexes.

RESULTS
Distribution of active life and inactive life at the followup survey is shown in Table 3. By our definition, 62.4% of men and 60.4% of women were defined to be in active life at the follow-up. Although men were more likely to have active life than women, the sex difference was not significant. There was also little difference in active life by sex in both age groups, 65 to 74, and 75 and over. Table 4 shows distribution of self-rated health at the baseline. Data of self-rated health in 15 subjects were missing at the baseline. The good category was observed in 12.9% of men and 10.8% of women in all age groups. On the other hand, the poor category was 12.7% for men and 20.7% for women in all age groups. Men were more likely to report better health than women (p<0.01). This sex difference was also significant in the age group of 65 to 74 years (p<0.01). Figure 1 shows the percentage with active life according to level of self-rated health at the baseline. Percentage with active life was the highest in the good category,  fifteen subjects were excluded because of missing in self-rated health.
p<0.01, chi-square between sexes in age groups, 65 to 74 and all. Next, an investigation of influences on active life according to self-rated health while controlling for other confounding factors was examined by multivariate logistic regression. Table 5 shows regression coefficient(*), relative odds, and 95% confidence interval. Self-rated health was transformed into two dummy variables : one dummy variable scored 0 for poor and I for good, while the other dummy variables scored 0 for poor and 1 for fair. Even controlling for other confounding variables, self-rated health displayed a significant correlation with active life in both men and women. Those with good self-rated health were more likely to have active life in 3.57 times of men and in 4.11 times of women than thoes with poor self-rated health. Those with fair self-rated health were likely to have active life in 2.36 times of men and 2.23 times of women than those with poor self-rated health. In addition to self-rated health in men, age (75 and over), current smoking, and physician visits revealed significantly inverse correlations to active life. Further, current drinking in- dicated a significantly direct correlation in men. In terms of variables other than self-rated health for women, only age (75 and over) was of significance.

DISCUSSION
There has been considerable research into factors of longevity, however epidemiological research into factors pertainning to a long active life for the elderly is in its infancy. Katz et al.10) proposed the concept "active life expectancy", meaning "the duration of functional wellbeing", as an index for measuring group levels of health. For the elderly, maintaining functional well-being is essential for quality of life. The aim of the present study was to clarify, by means of a seven year longitudinal study, factors contributing to long active life of the elderly. Based on the definition of active life by Katz et al., our study investigated living a long active life by maintaining complete independence for each item of basic DDL at home.
There has hitherto been much research by longitudinal surveys into factors influencing decline of physical functioning, however, almost all of this research was based on follow-up data obtained from living respondents7-9). Since the purpose of the present study was to investigate factors for "a long active life", deceased persons were also included in the follow-up data. The primary object of the present study was to clarify the degree of contribution of self-rated health to maintaining active life. It is wellknown that self-rated health presents a good reflection of objective health1,11). Self-rated health is further recognized as a useful predictor of mortality12,13). The present authors, too, have reported that even controlling for objective health status by follow-up survey of elderly subjects after a seven year duration, self-rated health had a significant correlation to motality6). However, research into whether self-rated health is useful as a predictor of long active life for the elderly is just beginning to be undertaken.
The present paper is based on a longitudinal survey of elderly subjects over a seven-year period, in which we investigated the degree of correlation between self-rated health at the time of baseline interview and subsequent active life. Results for both men and women showed that, even controlling for the effect of age, education, spouse, drinking, smoking, physician visits, and history of stroke, self-rated health was an important predictor for active life.
Ross and Havens 14) demonstrated the significance of self-rated health as a predictor of successful aging, based on their longitudinal study undertaken between 1971-1981 of 3573 subjects aged from 65 to 84 years. Their research indicated that self-rated health at baseline interview could be an even stronger predictor of successful aging when other confounding factors were controlled. While their definition of successful aging is different from our definition of active life in that their physical function represents a higher level than our ADL and also includes remaining mentally alert, their report supports the results of the present study.
Self-rated health is reported to be not only a useful indicater of objective health, but closely related to life satisfaction15-1 ) , subjective well-being3), and social activity"). These factors latent in self-rated health may contribute to the significance of self-rated health as a predictor of a long active life for the elderly.
The present study investigated the effects on active life of such health behavior as drinking and smoking with the relation of self-rated health, and found that, among men, current drinking had a positive effect, while current smoking had a negative effect. Other studies with concepts similar to the present study's active life, Pinsky's et al.18) "good function" , LaCroix's et al.19) "maintaining mobility", Guralinik's et al 20) "healthy aging", have investigated the effects of drinking and smoking, too. Their results were in accordance with those of the present survey for smoking. However, in the cases of drinking, other surveys adopted the quantitative assessment, while our survey investigated whether or not subjects habitually drank. Though the approaches were different, LaCroix et al. and Guralinik et al. found that moderate drinking maintains higher mobility and acts more positively on healthy aging than total abstinence, which generally supports the results of the present study.
A strong positive correlation has been reported between level of education as socio-economic status and active life expectancy21) or maintaining good function"). However, this correlation was not clearly shown in the present study. This is due to the region where the survey was undertaken ; there were few individual differences in educational background, and thus this factor did not reflect differences in socio-economic status. As stated earlier, the present study found self-rated health to be useful as a predictor of active life, however, the study was limited in that biological variables considered to effect aging were not involved in the variables investigated. Further research will be necessary to determine if self-rated health is significant as a predictor of long active life when the effects of biological measures are examined.