This Article Abstract Full Text (PDF) Services Download to citation manager

Intertwining Courses of Functional Status and Subjective Health Among Older Japanese

¹ School of Public Health and ² Institute of Gerontology, University of Michigan, Ann Arbor.
³ School of Public Health, University at Albany, State University of New York, Rensselaer, New York.
⁴ Tokyo Metropolitan Institute of Gerontology, Japan.

Address correspondence to Jersey Liang, Department of Health Management and Policy, University of Michigan School of Public Health, 1420 Washington Heights, Ann Arbor, MI 48109-2029. E-mail: jliang{at}umich.edu

	Abstract

TOP Abstract Methods Results Discussion References

 
Objectives. We sought to depict how trajectories of functional status are related to the average changes in self-rated health and its underlying trajectories.

Methods. Data came from a five-wave panel study of a national sample of 2,200 Japanese older adults between 1987 and 1999. We employed hierarchical linear models and multinomial logistic regression to depict the interrelationships among patterns of temporal change in functional status and self-rated health.

Results. Trajectories of functional status were associated with the average age-related changes in subjective health (i.e., linear and nonlinear slopes). Furthermore, there were significant correlations between the courses of functional health and those of self-rated health. Finally, recovery from poor self-rated health was characterized by having poor health and functional ability at baseline.

Discussion. Researchers can generalize prior observations of the association between functional status and subjective health at one or more points in time to their long-term trajectories. These findings provide further insights into understanding the dynamics between two key dimensions of health among older adults in Japan.

THE linkage between functional status and subjective health has been a focus of intense research in gerontology (e.g., Ferraro, Farmer, & Wybraniec, 1997; Idler, Hudson, & Leventhal, 1999). Current research has incorporated an increasingly dynamic perspective by examining health transitions and trajectories. Whereas a transition refers to change between two points in time, a trajectory is a pathway charted by a series of transitions that can be linear or nonlinear. The trend of deteriorating health across age groups conceals increasing heterogeneity (Dannefer, 2003). Although some older persons are very sick and disabled, others are free of illness and disability (Rowe & Kahn, 1987). This suggests the need to examine not only the average health change with age in a given population but also its underlying heterogeneity.

The distinction between the average change and its underlying heterogeneity corresponds well with that between the variable-centered approach and the person-centered approach. The former focuses on the parameters of the average change (i.e., intercept and slopes) as the dependent variables, whereas the latter concentrates on subgroup heterogeneity that reflects qualitatively distinct trajectories as the dependent variable. As complementary perspectives, the variable-centered approach yields information on factors that apply to everyone in general, and the person-centered approach identifies effects specific to individuals experiencing a given trajectory (Magnusson, 2003; Schulenberg, Wadsworth, O'Mally, Bachman, & Johnston, 1996).

Recent studies have documented health changes in later life on average and in multiple trajectories. Maddox and Clark (1992) used polynomial models to fit the average changes in functional impairment by using data from the Longitudinal Retirement History Study. Based on 8,804 observations of 1,515 male veterans, Aldwin and colleagues reported that physical symptoms increase with age linearly on average, but interpersonal differences can be characterized by seven distinct trajectories (Aldwin, Spiro, Levenson, & Cupertino, 2001). Finally, using data from 634 men between 1940 and 1986, Clipp and associates identified five health trajectories including (a) linear decline, (b), decline at end of life, (c) decline and improvement, (d) constant good health, and (e) constant poor health (Clipp, Pavalko, & Elder, 1992).

With regard to subjective health, researchers have found stability over time (Idler, 1993), improvements with age (Ferraro, 1980), and declines with age in a nonlinear pattern (Hoeymans, Feskens, Van den Bos, & Kromhout, 1997; Pinquart, 2001). In addition, recent research on older Japanese has shown that subjective health worsens slightly between ages 60 and 85 but improves marginally after age 85. More important, underlying this pattern of average change are four distinct trajectories including constant good health, early onset of worsening health, late onset of worsening health, and recovery from poor health (Liang et al., 2005).

Given that both functional status and subjective health change with age along multiple trajectories, a full understanding of the linkages between them demands an examination of how their trajectories interrelate dynamically. At the present time, researchers know almost nothing about these linkages in later life. The importance of this research question stems from a major tenet in adult developmental psychology and the sociology of age stratification that suggests that interpersonal variability in intrapersonal plasticity is a cornerstone for the understanding of adaptation in later life (Baltes & Baltes, 1990; Riley, 1987). Individuals differ significantly in the ways in which they age, which can be generally characterized as normal, optimal, or pathological (Baltes & Baltes, 1990; Rowe & Kahn, 1987). These heterogeneities may converge as well as diverge across the life course (Dannefer, 2003; House, Lantz, & Herd, 2005). To understand how intraindividual plasticities vary across individuals, it is critical to analyze how multiple dimensions of aging interface dynamically with one another. More specifically, because health development is a lifelong adaptive process, an essential research issue is how multiple trajectories of health interrelate across the life course (Halfon & Hochstein, 2002).

We aim to describe how age-related changes in functional status couple with those of subjective health by using data from a national sample of older Japanese between 1987 and 1999. As most empirical research on health and aging is based on data derived from Western societies, observations from non-Western societies are useful in cross-validating and interpreting current findings. In this regard, Japan provides a useful context for research, particularly in contrast with the United States. Similar in economic development, both nations are experiencing rapid population aging. However, with the highest life expectancy at birth in the world, Japanese outlive Americans by 4 to 6 years (World Health Organization, 2006). Nevertheless, findings concerning functional status and self-rated health among older Japanese often parallel those among older Americans (Liang et al., 2002; Schoeni et al., 2006).

In the following we offer several hypotheses. In these hypotheses the key independent variables were the trajectories of functional status, whereas the dependent variables were changes in self-rated health, both in terms of average change and multiple trajectories. According to our recent research, functional limitations increase with age as a quadratic function. This age-related change, in turn, consists of three trajectories: (a) minimal functional decrement, (b) early onset of functional decline, and (c) late onset of functional decline (Liang et al., 2003). Approximated by a cubic function, subjective health on average becomes only marginally worse between ages 60 and 85, whereas it appears to improve slightly after age 85. Underlying this nonlinear pattern of change are four trajectories: (a) constant good health, (b) early onset of worsening health, (c) late onset of worsening health, and (d) recovery from poor self-assessed health (Liang et al., 2005).

There is substantial evidence that functional status and subjective health are positively correlated cross-sectionally and prospectively (Ferraro et al., 1997; Idler, 1993; Pinquart, 2001). Extrapolating from these findings and following a variable-centered approach, we predicted that distinct trajectories of functional status would be significantly associated with the intercept as well as the linear and nonlinear rates of change of subjective health within the sample as a whole. Specifically, we offer the following two hypotheses:

H₁: In comparison to those with minimal functional decrement, older Japanese who experience early onset of functional decline are likely to start with poorer self-rated health and to suffer a greater rate of decline with age.

H₂: Subjective health will correlate similarly with the trajectory of late-onset functional decrement but at a lower initial level and with smaller rates of changes than those associated with early onset of functional decline.

In addition, we hypothesized that changes in functional status and self-rated health would be influenced by baseline demographic, socioeconomic, and health characteristics as well. In this research, we focused on the effects of functional health on changes in self-rated health. However, theoretically, reciprocal causal linkages are likely to exist. For instance, subjective health trajectories may well influence changes in functional status (Idler & Kasl, 1995). We discuss the implications of our assumption later.

On the basis of a person-centered approach and the positive correlations between functional status and subjective health (Ferraro et al., 1997; Idler, 1993; Pinquart, 2001), we offer two additional hypotheses:

H₃: There is a convergence between the trajectories of functional status and subjective health. In particular, individuals who experience an early onset of functional impairment are more likely to experience an early onset of perceived health decline. Similarly, late onset of functional decrement should predict late onset of worsening health.

H₄: Relative to individuals with minimal functional decrement, those who experience either an early or a late onset of functional decline are more likely to experience recovery from poor perceived health.

A brief rationale for Hypothesis 4 is in order. There is no parallel in the trajectories of functional health that corresponds to recovery or improving subjective health among older Japanese (Liang et al., 2005). This appears contrary to recent observations of recovery from disability (e.g., Hardy, Dubin, Holford, & Gill, 2005). However, findings of functional recovery have been largely based upon health transitions over a relatively short interval of time (ranging from 1 month to 3 years), and hence it could be a short-term phenomenon. In the longer term (i.e., 10 years or more), functional status tends to decline (Aldwin et al., 2001; Clipp et al., 1992; Liang et al., 2005).

Recovery from poor self-rated health differs significantly from improvement in self-rated health in general in that the former stems from poor baseline health conditions, whereas the latter can originate from good as well as poor baseline health. There is evidence that greater morbidity and poor functional status at baseline lead to either early death or recovery from poor self-rated health. This is because, relative to the other three trajectories of self-rated health, persons who recovered from poor health had worse health conditions at baseline. In fact, they were very similar to those who died during the period of follow-up (Liang et al., 2005). Therefore, it is possible that poor functional health trajectories (i.e., early- and late-onset decline in functional health relative to minimal functional decrement) are associated with recovery from poor perceived health. We discuss further the attributes of recovery from poor health later in this article.

One could explain health optimism (i.e., recovery from poor health) of those with poor baseline health by psychological adaptation and social comparison (Hoeymans et al., 1997; Idler, 1993; Pinquart, 2001). Given sufficient time, an individual may adapt to poor physical health and integrate it into daily functioning, hence leading to improved self-rated health (Idler et al., 1999). This is also consistent with the hypothesis of response shift in that changes in self-assessed health or quality of life may result from changes in internal standards, values, or conceptualizations (Sprangers & Schwartz, 1999).

	METHODS

TOP Abstract Methods Results Discussion References

 
Sample and Data
Data came from a five-wave panel study of older adults in Japan. The baseline survey involved a national sample of 2,200 Japanese aged 60 and older in 1987. With a response rate of 69%, the sample was representative of the total elderly population in Japan (Jay, Liang, Liu, & Sugisawa, 1993). After the baseline survey, study organizers followed up with all participants, except those who were deceased, once every 3 years (i.e., in 1990, 1993, 1996, and 1999). Proxy interviews were obtained for those unable to complete the survey themselves. Between Waves 2 and 5, the proportion of self-respondents ranged from 62% to 76%, whereas proxy interviews varied from 7% to 13% and death accounted for 8% to 14%.

Measures
The dependent variable, self-rated health, was assessed with three indicators: (a) a 5-point rating of physical health, (b) a 3-point scale of health comparisons to other people one's age, and (c) a 5-point scale of overall satisfaction with one's health. With health compared to others rescaled to reflect a 5-point scale, we summed the three items to form a composite of perceived ill health (range = 3–15) with Cronbach's alphas ranging from.82 to.86 across five waves. Functional status was assessed with six items measuring difficulty with basic or instrumental activities of daily living (i.e., bathing; climbing stairs; walking a half mile; using the phone; shopping; and traveling by bus, train, or subway). Each item was scored with a 5-point scale (0–4), resulting in a composite ranging from 0 to 24.

With reference to other health conditions at baseline, cognitive impairment was a count of incorrect responses to Pfeiffer's (1975) Short Portable Mental Status Questionnaire (range = 0–9). Depressive symptoms were measured by a composite of seven items ( =.807; range = 7–21) selected from the Center for Epidemiologic Studies–Depression scale (Radloff, 1977). This brief form taps the same symptom dimensions as does the original Center for Epidemiologic Studies–Depression scale with similar reliability (Kohout, Berkman, Evans, & Cornoni-Huntley, 1993). Based on a checklist of conditions, we generated an index of serious conditions (range = 0–4) by including diabetes, heart disease, hypertension, and stroke. We grouped the remaining conditions as chronic conditions, which consisted of a count of 13 types of conditions, including, for example, arthritis, eye disease, asthma, liver disease, Parkinson's disease, and backache (Ferraro & Farmer, 1999). We did not include cancer in the list of serious illnesses because it is highly threatening and Japanese physicians often do not disclose this diagnosis to their patients (Long & Long, 1982). We coded all mental and physical health covariates as well as the dependent variable (self-rated health status) with a higher score to reflect greater morbidity, impairment, or poor health.

To control for population heterogeneity, we included gender (female = 1), marital status (married = 1), education (number of years), employment (1 = working), and baseline health conditions in the multivariate analyses. Table 1 presents the descriptive statistics of all of the baseline covariates and the relative proportions of respondents assigned to various trajectories for functional status and subjective health.

where Y_iA is self-rated health for individual i at age A, _0i is the intercept of subjective health for individual i, _1i is the rate of change (slope) in self-rated health for individual i across age, and _iA represents random error in self-rated health for individual i at age A. For the sample as a whole, we considered both linear and nonlinear changes in health. We centered age at its mean across all five surveys in order to minimize the possibility of multicollinearity when we evaluated nonlinear functions of age-related changes. To identify trajectories of functional and self-rated health, we used K-means cluster analysis to create groups of individuals with similar growth parameters from Equation 1 (i.e., _0i + _1i).

To examine the linkages between changes in functional status and subjective health, we first applied a variable-centered approach. In particular, we included dummy variables representing functional trajectories as predictors in the Level 2 (or person-level) equation in the multilevel analysis. This is represented in the following model for each of the growth curve parameters:

Here, X_qi is a vector of covariates (e.g., gender, education, trajectories of functional status) associated with individual i, and β_pq represents the effect of X_qi on the pth growth parameter (_pi). r_pi is a random effect with a mean of 0.

Next, we applied a person-centered approach by using multinomial logistic regression analyses. Specifically, we treated trajectories of subjective health as the dependent variables. We coded courses of changes in functional status as dummy variables and entered them as predictors in conjunction with baseline characteristics. None of the Hausman statistics (Hausman & McFadden, 1984) associated with these analyses were statistically significant; hence, the assumption of independence of irrelevant alternatives was satisfied.

Missing Items and Attrition
To minimize the loss of participants due to missing items and attrition, we undertook multiple imputation. There were few missing items at baseline. For proxy interviews and unit nonresponse at a given wave, we imputed missing values by using baseline data and repeated observations throughout the period of follow-up. We imputed three complete data sets with the NORM software developed by Schafer (1997), and we ran analyses on each of these three data sets. We derived parameter estimates and their standard errors by averaging across three imputations and by adjusting for the variance.

Mortality
To assess bias due to selective mortality, we first undertook a competing risk analysis to delineate the similarities and differences between those who died before 1993 (n = 355) and the survivors for whom sufficient data were available for charting the trajectories of self-rated health. Second, following an approach suggested by Heckman (1979; for a more accessible description, see Berk, 1983), we created a measure of predicted probability of mortality between 1987 and 1999 for each respondent by adjusting for baseline demographic and socioeconomic attributes and health conditions. We subsequently included this measure as a covariate in our multivariate analyses to control for the bias due to selective mortality.

	RESULTS

TOP Abstract Methods Results Discussion References

 
Variable-Centered Approach
In Table 2, the dependent variables are the intercept and linear, quadratic, and cubic slopes that describe the average change in self-rated health with age. To assess the gross effects of functional trajectories, we first included them without controlling for any baseline attributes (Model 1). We ascertained their net effects by adjusting for baseline sociodemographic and health attributes (Model 2).

View larger version (9K): [in this window] [in a new window]   Figure 1. Age-related changes in self-rated health by trajectory of functional status

In contrast to minimal functional decrement, late-onset functional decrement (effective age = 70–90) was associated with increasingly worsening health approximating a cubic function (b_{linear slope} = 1.340, p <.001; b_{quadratic slope} =.867, p <.01; b_{cubic slope} = –.709, p <.001, Model 2). Between ages 70 and 75, those with late-onset functional decline were very close to those with minimal functional decrement in their subjective health (Figure 1). The gap in subjective health between these two groups started to increase between ages 75 and 85, and it then diminished after age 85. Indeed, the self-rated health of these two groups converged before age 90, with older Japanese who had experienced late-onset functional decline having slightly better subjective health than those who had minimal functional decrement.

In addition to functional health trajectories, the average change in self-rated health was correlated with baseline health conditions. In particular, poor health at baseline led to a greater level of perceived ill health at the mean age. For instance, functional impairment at baseline was associated with worse self-rated health at the mean age (b_intercept =.203, p <.001, Model 2). In contrast, chronic conditions and functional impairment at baseline were associated with smaller linear changes in self-rated health.

To control for possible selection bias due to the exclusion of all or partial observations of those who died between 1987 and 1999, we included predicted risk of dying as a covariate. Predicted mortality was associated with the rate of change in subjective health in distinctive fashions (Table 2). In particular, those with greater predicted risk of dying had an elevated level of worsening health at the mean age (b_intercept =.683, p <.01, Model 2) but a reduced quadratic age slope (b_{quadratic slope} = –.221, p <.05, Model 2).

One can illustrate these effects with graphs of age-related self-rated health at various levels of predicted risk of dying (i.e., mean and one standard deviation below and above the mean; graphs not shown). In particular, before age 85, levels of self-rated health corresponded well with those of predicted risk of dying. Thus, individuals with a high risk of dying had worse self-rated health. For those with a high risk of dying, self-rated health got worse over time in a rather linear fashion. In contrast, for those at lower risks of dying, perceived health was better overall but worsened in an accelerated fashion starting at about age 70. This led to a convergence of self-rated health at age 85. In fact, the subjective health of those with a high risk of dying was slightly better than that of those with a lower risk of dying after age 85. This finding parallels the crossover in health disparities observed in the United States, commonly attributed to selective survival or reduced impact of key risk factors (Beckett, 2000; Johnson, 2000).

Person-Centered Approach
Trajectories of functional status were significantly correlated with their counterparts in subjective health (Table 3, Model 1). The probability of experiencing early-onset worsening perceived health was more than 4 times greater for those with early-onset functional decrement (relative risk ratio [RRR] = 4.103, p <.001, Model 1) than for those with minimal functional decline. Similarly, those with late onset of functional decline were more than 3 times (RRR = 3.364, p <.001, Model 1) more likely than those with minimal functional decrement to experience late-onset worsening subjective health.

Consistent with Hypothesis 4, those with early or late onset of functional decline were also more likely to experience improvement in perceived health. This hypothesis was supported when baseline attributes were not controlled (Model 1, Table 3), in that those with early- and late-onset functional decline were approximately twice as likely as those with minimal functional decrement to experience recovery from poor health. With baseline attributes adjusted, only early-onset functional decline remained significantly associated with recovery from poor health (RRR = 1.896, p <.05, Model 2), whereas the effect of late-onset functional decline was no longer significant.

In addition to functional health trajectories, baseline health attributes were associated with the trajectory of recovery from poor health. For example, those with more serious conditions at baseline were more than twice as likely as those with fewer serious conditions to be in the recovery from poor self-rated health subgroup (RRR = 2.225, p <.001, Model 2). To some, these findings may be counterintuitive. However, we can explain them by the fact that individuals who recovered from poor self-rated health tended to have poor health conditions to begin with.

This was demonstrated by a competing risk analysis contrasting the respondents in the self-rated health trajectories in conjunction with those who died between 1987 and 1993. According to Table 4, poor baseline health conditions (i.e., cognitive impairment, Center for Epidemiologic Studies–Depression, serious conditions, chronic conditions, and functional impairment) were significant predictors of the trajectory of recovery from poor self-rated health relative to the other three trajectories. More important, those who recovered from poor self-rated health shared many of the baseline poor health predictors with those who died between 1987 and 1993, who presumably would have had poorer health in comparison with the survivors. However, individuals who experienced early or late onset of worsening health differed from those with constant good health only in terms of age and diseases at baseline (Table 4).

	DISCUSSION

TOP Abstract Methods Results Discussion References

 
There is emerging gerontological research on how trajectories of two or more variables interrelate with each other. For instance, McDonough and Berglund (2003) reported that although increasing income slowed the worsening of subjective health, it did not erase the health effects of earlier poverty experiences. Other investigators have employed latent growth curve models (George & Lynch, 2003; Kelley-Moore & Ferraro, 2004; Taylor & Lynch, 2004). Furthermore, these investigators have focused exclusively on average changes. To the best of our knowledge, no prior studies have analyzed the linkages between the trajectories of functional status and subjective health.

Variable-centered and person-centered approaches yield new insights regarding health dynamics across life course. On the basis of the variable-centered approach, early-onset functional decline, relative to minimal functional decrement, is significantly correlated with an accelerated worsening of subjective health. In contrast, individuals with late-onset functional decrement enjoy better perceived health in later life. The experience of the latter individuals resembles the changes in self-rated health of those with minimal functional decrement between ages 70 and 75 but then starts to diverge between ages 75 and 85 (Table 2). Consequently, it is important to incorporate changes on both sides of the equation when analyzing the relationships between functional status and self-rated health.

According to the person-centered approach, evidence is equivocal regarding the hypothesized convergence between the trajectories of functional status and those of self-rated health (Table 3). Early-onset functional decline is associated with early onset and late onset of worsening health, whereas late-onset functional decline is correlated with neither. However, there appears to be a gradient in the effect of early-onset functional decrement in that its impact on early onset of worsening health (i.e., RRR = 3.639) is substantially greater than that on late-onset functional decline (RRR = 2.000).

Relative to minimal functional decrement, early onset of functional decline leads to a greater probability of recovery from poor self-rated health, but late onset of functional decline does not (Table 3). This is consistent with our finding that improvement from poor self-rated health is associated with poor health at baseline (Table 4). In contrast, individuals with late-onset functional decline do not differ from those with minimal functional decrement in baseline health. Thus, recovery from perceived poor health is largely confined to individuals with poor health conditions at baseline and a greater risk of dying.

According to recent research, an important mediator of this adaptation process to a life-threatening or chronic disease is response shift, which is defined as changes in the meaning of one's self-evaluation of health or quality of life resulting from changes in internal standards, values, or conceptualization (Rapkin & Schwartz, 2004; Sprangers & Schwartz, 1999). Integrating key components of response shift such as coping processes (e.g., social comparison, goal reordering) and appraisal (i.e., internal standards, values, and conceptualization) into future research would allow a better understanding of the linkages between trajectories of functional status and those of self-rated health.

Like all scientific endeavors, our research could be improved in several respects. First, although we focused on the effects of trajectories of functional status on those of self-rated health, reciprocal linkages are likely to exist. Our prior analysis partially addressed these in that baseline self-rated health was shown to influence not only on the initial level of functional status but also on the linear and quadratic rates of change (Liang et al., 2003). Nevertheless, reciprocal linkages between trajectories of two variables is a complicated and often intractable issue. In this research, because changes in both functional status and self-rated health were charted over the same period of time, the causal order is ambiguous. As a result, a strong assumption is required to infer causal effects. This limitation applies to almost all current studies involving hierarchical linear models or latent growth curve models. McArdle, Ferrer-Caja, Hamagami, and Woodcock (2002) have recently proposed an approach involving latent difference scores that may partially address this issue.

Second, the life course is multidimensional, as it unfolds in distinct but related domains of life (e.g., work and family) within the context of biological and psychological maturation and decline (Mayer, 2003). In this regard, functional status and self-assessed health represent two key dimensions of health and well-being that include disease avoidance, sustained engagement in life, and maintenance of cognitive and physical function (Rowe & Kahn, 1987). Further explication of the dynamic linkages across the trajectories of all of these dimensions is an important task for future research. Researchers could readily extend the approach in the present research to such analyses.

Third, investigators need to connect health trajectories in later life with those in early stages of the life course (Ben-Shlomo & Kuh, 2002). This formulation enables the researcher to evaluate not only early life course exposure with later disease, but possible pathways with potential intermediaries or confounding factors. As advantages accumulate and compound, such research also offers new opportunities to examine the processes that generate health inequalities (Mirowsky, Ross, & Reynolds, 2000). That is, adverse social factors that affect infant health and development may be important in leading to health disparities in adulthood and later life.

Finally, can our findings from older Japanese be generalized to other populations? There is some evidence that the linkages between self-rated health and key determinants such as socioeconomic status, gender, social relations, and functional status vary across nations (Jylha, Guralnik, Ferrucci, Jokela, & Heikkinen, 1998; Kunst, Geurts, & van den Berg, 1995). Accordingly, further replications of our analyses with data derived from other societies are necessary. Investigators should regard our research as a prelude for a more systemic understanding of how health trajectories vary across different cultures.

	Acknowledgments

 
This research was supported by Grant R01-AG154124 (Jersey Liang, principal investigator) from the National Institute on Aging. Additional support was provided by the Japanese Ministry of Health, Labor and Welfare Longevity Foundation, the Tokyo Metropolitan Institute of Gerontology, and the Michigan Claude D. Pepper Older Americans Independence Center (Grant P60-AG08808). Jersey Liang planned the study, supervised the data analysis, interpreted the findings, and wrote and revised the article. Benjamin A. Shaw helped in planning the study and played a significant role in data analysis and writing and revising the article. Joan M. Bennett managed the data, undertook missing data imputation, performed statistical analyses, and contributed to writing and revising the article. Neal Krause contributed to writing and revising the article for important intellectual content. Erika Kobayashi, Taro Fukaya, and Yoko Sugihara were responsible for the collection and maintenance of the longitudinal data. In addition, they assisted in interpreting the findings from a Japanese perspective and contributed to writing and revising the article.

	Footnotes

 
Decision Editor: Kenneth F. Ferraro, PhD

Received for publication May 23, 2006. Accepted for publication May 3, 2007.

	References

TOP Abstract Methods Results Discussion References

 

• Aiken, L. S., West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage.
• Aldwin, C. M., Spiro, A., III, Levenson, M. R., Cupertino, A. P. (2001). Longitudinal findings from the Normative Aging Study: III. Personality, individual health trajectories, and mortality. Psychology and Aging, 16,450-465.[Medline]
• Baltes, P. B., Baltes, M. M. (1990). Psychological perspectives on successful aging: The model of selective optimization with compensation. In P. B. Baltes & M. M. Baltes (Eds.), Successful aging: Perspectives from the behavioral sciences (pp. 1–34). New York: Cambridge University Press.
• Beckett, M. (2000). Converging health inequalities in later life—an artifact of mortality selection? Journal of Health and Social Behavior, 41,106-119.[Medline]
• Ben-Shlomo, Y., Kuh, D. (2002). A life course approach to chronic disease epidemiology: conceptual models, empirical challenges and interdisciplinary perspectives. International Journal of Epidemiology, 31,285-293.[Free Full Text]
• Berk, R. A. (1983). An introduction to sample selection bias. American Sociological Review, 48,386-398.
• Clipp, E. C., Pavalko, E. K., Elder, G. H., Jr.. (1992). Trajectories of health: In concept and empirical pattern. Behavior, Health, and Aging, 2,159-179.
• Dannefer, D. (2003). Cumulative advantage/disadvantage and the life course: Cross-fertilizing age and social science theory. Journal of Gerontology: Social Sciences, 58B,S327-S337.[Abstract/Free Full Text]
• Ferraro, K. (1980). Self-ratings of health among the old and the old-old. Journal of Health and Social Behavior, 21,377-383.[Medline]
• Ferraro, K. F., Farmer, M. M. (1999). Utility of health data from social surveys: Is there a gold standard for measuring morbidity? American Sociological Review, 64,303-315.
• Ferraro, K. F., Farmer, M. M., Wybraniec, J. A. (1997). Health trajectories: Long-term dynamics among black and white adults. Journal of Health and Social Behavior, 38,38-54.[Medline]
• George, L. K., Lynch, S. M. (2003). Race differences in depressive symptoms: A dynamic perspective on stress exposure and vulnerability. Journal of Health and Social Behavior, 44,353-369.[Medline]
• Halfon, N., Hochstein, M. (2002). Life course health development: An integrated framework for developing health, policy, and research. Milbank Quarterly, 80,433-479.[Medline]
• Hardy, S. E., Dubin, J. A., Holford, T. R., Gill, T. M. (2005). Transitions between states of disability and independence among older persons. American Journal of Epidemiology, 161,575-584.[Abstract/Free Full Text]
• Hausman, J., McFadden, D. (1984). Specification tests for the multinomial logit model. Econometrica, 52,1219-1240.
• Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica, 47,153-161.
• Hoeymans, N., Feskens, E. J. M., Van den Bos, G. A. M., Kromhout, D. (1997). Age, time, and cohort effects on functional status and self-rated health in elderly men. American Journal of Public Health, 87,1620-1625.[Abstract/Free Full Text]
• House, J. S., Lantz, P. M., Herd, P. (2005). Continuity and change in the social stratification of aging and health over the life course: Evidence from a nationally representative longitudinal study from 1986 to 2001/2002 (Americans' Changing Lives Study). Journal of Gerontology: Social Sciences, 60B,(Special Issue II), 15-26.[Medline]
• Idler, E. L. (1993). Age differences in self assessments of health: Age changes, cohort differences, or survivorship? Journal of Gerontology: Social Sciences, 48,S289-S300.
• Idler, E. L., Hudson, S. V., Leventhal, H. (1999). The meaning of self-ratings of health. Research on Aging, 21,458-476.[Abstract/Free Full Text]
• Idler, E. L., Kasl, S. V. (1995). Self-ratings of health: Do they also predict change in functional ability? Journal of Gerontology: Social Sciences, 50B,S344-S353.[Abstract]
• Jay, G., Liang, J., Liu, X., Sugisawa, H. (1993). Patterns of nonresponse in a national survey of the Japanese elderly. Journal of Gerontology: Social Sciences, 48,S143-S152.
• Johnson, N. E. (2000). The racial crossover in comorbidity, disability, and mortality. Demography, 37,267-283.[Medline]
• Jylha, M., Guralnik, J. M., Ferrucci, L., Jokela, J., Heikkinen, E. (1998). Is self-rated health comparable across cultures and genders? Journal of Gerontology: Social Sciences, 53B,S144-S152.[Abstract]
• Kelley-Moore, J. A., Ferraro, K. F. (2004). The black/white disability gap: Persistent inequality in later life? Journal of Gerontology: Social Sciences, 59B,S34-S43.[Abstract/Free Full Text]
• Kohout, F. J., Berkman, L. F., Evans, D. A., Cornoni-Huntley, J. (1993). Two shorter forms of the CES-D depression symptoms index. Journal of Aging and Health, 5,179-193.[Abstract/Free Full Text]
• Kunst, A. E., Geurts, J. J., ven den Berg, J. (1995). International variation in socioeconomic inequalities in self-reported health. Journal of Epidemiology and Community Health, 49,117-123.[Abstract/Free Full Text]
• Liang, J., Bennett, J., Krause, N., Kobayashi, E., Kim, H., Brown, J. W., et al. (2002). Old age mortality in Japan: Does the socioeconomic gradient interact with gender and age? Journal of Gerontology: Social Sciences, 57B,S294-S307.[Abstract/Free Full Text]
• Liang, J., Shaw, B. A., Krause, N., Bennett, J. M., Blaum, C., Kobayashi, E., et al. (2003). Changes in functional status among older adults in Japan: Successful and usual aging. Psychology and Aging, 18,684-695.[Medline]
• Liang, J., Shaw, B. A., Krause, N., Bennett, J. M., Kobayashi, E., Fukaya, T., et al. (2005). How does self-assessed health change with age? A study of older adults in Japan. Journal of Gerontology: Social Sciences, 60B,S224-S232.[Abstract/Free Full Text]
• Long, S. O., Long, B. D. (1982). Curable cancers and fatal ulcers: Attitudes toward cancer in Japan. Social Science & Medicine, 16,2101-2108.
• Maddox, G. L., Clark, D. O. (1992). Trajectories of functional impairment in later life. Journal of Health and Social Behavior, 33,114-125.[Medline]
• Magnusson, D. (2003). The person approach: Concepts, measurement models, and research strategy. New Directions for Child and Adolescent Development, 101,3-23.[Medline]
• Mayer, K. U. (2003). The sociology of the life course and lifespan psychology: Diverging or converging pathways? In U. M. Staudinger & U. Lindenberger (Eds.), Understanding human development: Dialogues with lifespan psychology (pp. 463–481). London: Kluwer Academic.
• McArdle, J. J., Ferrer-Caja, E., Hamagami, F., Woodcock, R. W. (2002). Comparative longitudinal structural analysis of the growth and decline of multiple intellectual abilities over the life span. Developmental Psychology, 38,115-142.[Medline]
• McDonough, P., Berglund, P. (2003). Histories of poverty and self-rated health trajectories. Journal of Health and Social Behavior, 44,198-214.[Medline]
• Mirowsky, J., Ross, C., Reynolds, J. (2000). Links between social status and health status. In C. E. Bird, P. Conrad, & A. M. Fremont (Eds.), Handbook of medical sociology (5th ed., pp. 47–67). Upper Saddle River, NJ: Prentice Hall.
• Pedhazur, E. J. (1982). Multiple regression in behavioral research (2nd ed.). New York: Holt, Rinehart, & Winston.
• Pfeiffer, E. (1975). A short portable mental status questionnaire for the assessment of organic brain deficit in elderly patients. Journal of the American Geriatrics Society, 23,433-441.[Medline]
• Pinquart, M. (2001). Correlates of subjective health in older adults: A meta-analysis. Psychology and Aging, 16,414-426.[Medline]
• Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1,385-401.[Abstract]
• Rapkin, B. D., Schwartz, C. E. (2004). Toward a theoretical model of quality-of-life appraisal: Implications of findings from studies of response shift. Health and Quality of Life Outcomes, 2,(14), 14-27.
• Raudenbush, S. W., Bryk, A. S. (2002). Hierarchical linear models (2nd ed.). Thousand Oaks, CA: Sage.
• Riley, M. W. (1987). On the significance of age in sociology. American Sociological Review, 52,1-14.
• Rowe, J. W., Kahn, R. L. (1987, July 10). Human aging: Usual and successful. Science, 237,143-149.
• Schafer, J. L. (1997). Analysis of incomplete multivariate data. London: Chapman & Hall.
• Schoeni, R., Liang, J., Bennett, J., Kobayashi, E., Fukaya, T., Sugihara, Y. (2006). Trends in old-age functioning and disability in Japan: 1993–2002. Population Studies, 60,(1), 39-53.[Medline]
• Schulenberg, J., Wadsworth, K. N., O'Mally, P. M., Bachman, J. G., Johnston, L. D. (1996). Adolescent risk factors for being drinking during the transition to young adulthood: Variable- and pattern-centered approaches to change. Developmental Psychology, 32,659-674.
• Sprangers, M. A., Schwartz, C. E. (1999). Integrating response shift into health-related quality of life research: A theoretical model. Social Science & Medicine, 48,1507-1515.
• Taylor, M. G., Lynch, S. M. (2004). Trajectories of impairment, social support, and depressive symptoms in later life. Journal of Gerontology: Social Sciences, 59B,S238-S246.[Abstract/Free Full Text]
• World Health Organization. (2006). Countries. Retrieved May 19, 2006, from http://www.who.int/country/en/.

This Article Abstract Full Text (PDF) Services Download to citation manager