Elsevier

Social Science & Medicine

Volume 56, Issue 2, January 2003, Pages 203-217
Social Science & Medicine

Two views of self-rated general health status

https://doi.org/10.1016/S0277-9536(02)00020-5Get rights and content

Abstract

Global self-evaluations of health have proven to be sensitive predictors of morbidity and mortality. Yet researchers have only a limited understanding of how these self-evaluations are reached. This research compares two interpretations of self-rated health, as reflecting either a spontaneous assessment of one's health status and related practices, or an aspect of one's enduring self-concept. Using longitudinal data from successive waves of the National Population Health Survey in Canada (Statistics Canada, 1994–95, 1996–97, NPHS public use microdata documentation. Ottawa, Ontario: Statistics Canada; n=7505), our analysis tests a model of change in self-rated health as predicted by respondents’ baseline physical and mental health symptoms, social support, leisure physical activity, smoking, body mass index, and 2-yr changes in these characteristics. As in past research, self-rated health was sensitive to improvement or decline in these predictors. Much of the explained variance, however, was unique to respondents’ self-rated health 2 yr earlier. Moreover, the effect of several predictors on respondents’ self-rated health varied according to whether respondents intended to improve specific health-related behaviours in the future. These findings suggest that self-rated health is not only a spontaneous assessment of one's health status and related practices; like a self-concept, self-rated health may be regulated by efforts to achieve one's relatively important health-related goals.

Introduction

People's perceptions of their health can change in response to a variety of events. Consider an individual who receives a diagnosis of a previously unsuspected chronic condition, or who starts a new exercise program, or who relocates to an environment where unfamiliar social norms of smoking or alcohol consumption prevail. A person in any of these circumstances would likely feel the need to re-evaluate his or her health, and a variety of adaptive responses might depend upon the outcome of this self-evaluation. The outcome itself, however, is difficult to anticipate. On one hand, insofar as self-ratings of general health are based upon feedback about one's state of wellness or illness, they should be sensitive to and directly reflect the impact of informational, behavioural, or normative changes. On the other hand, self-ratings of general health may be driven less by bodily or environmental feedback than by an individual's prior beliefs about him- or herself as a healthy or unhealthy person. A variety of psychological mechanisms would then serve to protect these personal explanatory beliefs from change, even in response to highly discrepant feedback (see Banaji & Prentice, 1994, for a review).

The purpose of this research is to compare these alternative interpretations, which we have termed the spontaneous assessment and enduring self-concept views of self-rated general health status (hereafter referred to as self-rated health). The critical distinction between these views lies in the construct that each takes self-rated health to represent: health status, or the self-concept of health, respectively. Although these constructs are interrelated, they lead to different predictions about the stability and conditional nature of change in self-rated health. In particular, if self-rated health measures health status directly, then it should change strictly as a linear function of change in other variables that are closely associated with health status, which collectively should predict self-rated health to a great extent. Conversely, if self-rated health taps individuals’ self-concept of health, then it should change as a non-linear function of change in these other variables, depending on circumstances that would favour stability or change in the self-concept. On this view, self-rated health should evince significant stability over time, independent of observed changes in health status during the same period. To provide a suitably broad empirical test of these predictions, we present a secondary analysis of longitudinal data from two successive waves of the Canadian National Population Health Survey (NPHS, World Health Organization (1994–95), Statistics Canada (1996-97)), in which respondents’ self-rated health, self-reported physical and mental health status, personal health practices, and social support were repeatedly measured, among a nationally representative adult sample.

Lay definitions of health take a wide range of factors into account. Beyond the contribution of physical, mental, and social well-being, reflected in the World Health Organization's holistic definition of health (World Health Organization, 1958), research has demonstrated the importance of functional ability (Barsky, Cleary, & Klerman, 1992; Johnson & Wolinsky, 1993), lifestyle and preventive health practices (Krause & Jay, 1994; Ross & Bird, 1994), and socio-cultural constructions of particular health risks (Barsky, 1988). Cross-sectional studies, in which many of these factors have been measured at the same time, have typically shown that self-rated health is most closely related to the experience of physical symptoms (Bailis, Segall, Mahon, Chipperfield, & Dunn, 2001; Garrity, Somes, & Marx, 1978). Thus, one might expect improvement or decline in self-rated health to reflect primarily ongoing changes in the amount of distress or limitation associated with physical symptoms.

Few studies to date have examined changes in self-rated health over time; still fewer have assessed sociodemographic, psychosocial, and behavioural determinants of health at the same intervals (Rodin & McAvay, 1992). This small body of research nonetheless highlights the importance of chronic disease and disability as determinants of change in self-rated health. Goldstein, Siegel, and Boyer (1984) examined health perceptions in a probability sample of Los Angeles County residents over a 1-yr period. Approximately two-thirds of this sample showed no change in self-rated health over the year, suggesting along with earlier investigations that self-rated health is a relatively stable construct (Maddox, 1964). Among the remaining one-third, however, variations in chronic illness and disability predicted change in self-rated health, whereas those in acute conditions did not. Gold, Franks, and Erickson (1996) compared measures of functional ability and self-rated health as predictors of morbidity and mortality in a national probability sample of Americans, over a period spanning 5 yr. Both measures contributed independently to the prediction of health outcomes in this study; however, respondents with initially low scores for functional ability also showed greater risk of a decline in self-rated health 5 yr later.

It is noteworthy for the present investigation that a number of researchers have found support for the reverse causal direction in longitudinal studies: that initially low self-rated health predicts poorer functional ability (Crimmins & Saito, 1993; Ferraro, Farmer, & Wybraniec, 1997; Rodin & McAvay, 1992; Strawbridge, Camacho, Cohen, & Kaplan, 1993) and more physician visits and hospitalization (Menec & Chipperfield, 2001). These studies have concentrated primarily on elderly respondents. An exception is the analysis by Ferraro et al. (1997) of data from a nationally representative sample of American adults, initially ranging in age from 25–74, who were followed for 15 yr. Using structural equation modelling, these investigators demonstrated that more negative self-assessments of health were associated with greater subsequent disability, which contributed in turn to a cycle of health decline. Controlling for incident morbidity did not account for this effect of self-assessments on functional health. Thus, it appears that the relationship of self-rated health to chronic illness and disability is one of mutual influence.

The picture to emerge from this body of research is somewhat consistent with both the spontaneous assessment and enduring self-concept interpretations of self-rated health. On one hand, self-rated health is clearly responsive to observable indicators of illness, as one would expect for an indicator of health status. On the other hand, negative self-rated health bears a unique and perhaps self-fulfilling relationship to future reports of illness and functional limitation, as one would expect for an indicator of the self-concept of health.

The intended contribution of this research is to gain insight into the process of judging health by specifying how the spontaneous assessment and enduring self-concept views lead to differential predictions of change in self-rated health, and by evaluating those predictions empirically. We note at the outset that these objectives have conceptual and methodological problems associated with them. Conceptually, health status and the self-concept of health are likely to agree with and indeed influence each other; thus, it will frequently be difficult to distinguish between them as possible sources of self-rated health. Methodologically, we are constrained by the lack of a comprehensive measure of respondents’ self-concept of health. This prevents us from using a modelling analytic approach, which would allow an empirical comparison of the two views in terms of goodness-of-fit criteria. Nevertheless, we contend that it is possible to distinguish logically and empirically between the spontaneous assessment and enduring self-concept views of self-rated health. There are strong theoretical grounds, to be reviewed shortly, for assuming that behavioural intentions to improve one's health may buffer the influence of several variables on the self-concept of health, but not on health status. Therefore, evidence that buffering does or does not occur in relation to self-rated health should indicate to which of these constructs it belongs. In this paper, we test two empirical predictions: if self-rated general health reflects a self-concept of health, then (1) temporal stability of self-rated health will not depend on other health status measures, personal health practices, or social support, and (2) behavioural intentions will have a moderating effect on self-rated health.

Fig. 1 illustrates the logic of our specific predictions and is intended to guide our further elaboration of the two views. The figure is not meant to serve as a comprehensive theoretical model. We have deliberately omitted possible relations among constructs that are irrelevant to our objective of distinguishing between the two views of self-rated health. Furthermore, for heuristic purposes, we have illustrated social support and personal health practices as single constructs, though they are multi-dimensional (Sarason, Pierce, & Sarason, 1990; Segall & Chappell, 2000). The aim of Fig. 1, rather, is to provide a compact representation of the ways in which particular forms of social support, personal health practices, and behavioural intentions, all of which were measured in the NPHS, are theoretically related to both health status and the self-concept of health.

The spontaneous assessment perspective characterizes self-rated health as a summary evaluation of a person's transitory standing with respect to multiple distinct components of health status. The principal reason for adopting this perspective is that people's self-appraisals of health often correspond well with both physician ratings of their health (LaRue, Bank, Jarvik, & Hetland, 1979; Maddox, 1964; Maddox & Douglass, 1973) and with longevity (Idler & Benyamini, 1997; Idler & Kasl, 1991; Idler, Kasl, & Lemke, 1990; Kaplan & Camacho, 1983; Menec, Chipperfield, & Perry, 1999; Mossey & Shapiro, 1982). Idler and Benyamini (1997) particularly emphasize the superior inclusiveness and predictive validity of self-rated health, compared to any and all of the health status indicators used as covariates in previous studies. This body of research argues persuasively that people can and do synthesize a great deal of information about themselves in the single global assessment of their health as excellent, very good, good, fair, or poor.

Most research in medical sociology has implicitly adopted the spontaneous assessment perspective in modelling the relation of self-rated health to the physical, mental, social, and functional components of health status (Fylkesnes & Førde (1991), Fylkesnes & Førde (1992); Johnson & Wolinsky, 1993; Segovia, Bartlett, & Edwards, 1989). Changes in any component of health status, according to this view, should be reflected to some degree on each occasion in which self-rated health is assessed. Thus, a key implication of this view, which contrasts with the enduring self-concept view, is that the degree to which self-rated health at one time predicts that at a later time should be due to changes in other established measures of health status. Some evidence suggests that social support and personal health practices also contribute to global self-evaluations of health, possibly because of their directly felt or anticipated effect on health status (Krause & Jay, 1994; Ross & Bird, 1994). It follows that changes in self-rated health should reflect changes in personal health practices or social support, according to the spontaneous assessment view. The NPHS included measures that allowed us to test these predictions empirically. They are illustrated by the lower horizontal elements of Fig. 1.

These predictions are not altered by the assumption that the self-concept of health also influences health status, as shown in Fig. 1. The essential point of the spontaneous assessment view, rather, is that self-rated health does not reflect this influence any more or less than other measures of health status do. Thus, it remains the case that, holding measurement error aside, changes in self-rated health should be largely explained by changes in these other measures.

Another key point of contrast with the enduring self-concept view involves the indirect role of one's health-improvement goals, or behavioural intentions, in determining self-rated health. According to much research on the theory of planned behaviour (Ajzen, 1985), behavioural intentions increase the likelihood that a given behaviour will be performed. The vertical arrow linking behavioural intentions to personal health practices in Fig. 1 illustrates this influence. Intentions are theorized to represent an individual's motivation to perform behaviour. Thus, they may indirectly affect an individual's self-rated health by influencing his or her performance of health-related behaviours. According to the spontaneous assessment view, however, one's self-rated health should equally reflect a change in one's performance of health-related behaviours, whether or not this change was encouraged by a health-improvement goal.

The enduring self-concept view characterizes self-rated health as an indication of a person's established beliefs about his or her health. The fact that people experience continuity in their health provides the most basic reason for adopting the enduring self-concept view. More pointedly, however, there are many instances in which people's self-rated health does not reflect more objective indicators of their health status (Borawski, Kinney, & Kahana, 1996; Chipperfield, 1993; Litva & Eyles, 1994; Wilcox, Kasl, & Idler, 1996). For example, Wilcox and colleagues (1996) followed self-ratings of general health over time among a subsample of elderly participants in an epidemiological survey, who subsequently survived hospitalization for a major medical event (e.g., stroke, myocardial infarction, or hip fracture). Despite the seriousness and salience of these events, half of the participants showed no change and approximately a quarter showed better self-rated health, from before to 6 weeks following hospitalization.

Chipperfield (1993) conducted a prospective analysis of the health consequences associated with incongruities in self-rated and objective health status, within a sample of elderly Canadians. Respondents were classified as health optimists or pessimists based on their level of self-rated health, given the number of chronic conditions from which they suffered. In other words, respondents’ subjective views of their health were deemed overly positive or negative for the level of illness they showed on selected objective indicators. Those who overestimated their health were significantly more likely to be living up to 12 yr later. This research suggests that self-rated health involves an active construal or interpretation of chronic symptoms, which may be subject to bias in the interests of maintaining one's self-concept as a healthy or unhealthy person. Moreover, in the case of overly positive individuals, this bias may be adaptive for survival.

Research in social psychology has identified a variety of buffering mechanisms that routinely serve to protect the self-concept from change (Banaji & Prentice, 1994). For example, people seek feedback that confirms their existing self-views, even when negative (Swann, 1987). People further avoid feedback that may be discrepant with their self-views, by several means: active striving in self-relevant domains (Cantor & Kihlstrom, 1987); selective interaction and comparison with others (Taylor & Lobel, 1989; Tesser & Campbell, 1984); selective attention to, and biased interpretation of the feedback these others provide (Greenwald, 1980; Markus, 1977). It is interesting in this light to consider the finding by Menec et al. (1999) that older adults’ self-rated health, assessed in 1991, predicted their use of primary-control or active coping strategies (e.g., persistence, exerting more effort) to deal with age-related challenges 4 yr later. Were these active coping strategies motivated by older adults’ desire to maintain their self-concept of positive health? Correlational analysis cannot provide a definitive answer, but Menec's study controlled for and effectively eliminated a number of plausible alternatives to this causal scenario, including respondents’ initial age, morbidity, or functional disability. Active striving and the use of cognitive buffering mechanisms ensure a degree of consistency to the self-concept that is broadly adaptive for human functioning (Antonovsky, 1987; Ryff & Singer, 1998; Schulz & Heckhausen, 1996; Steele, 1988). Thus, if self-rated health behaves like a self-concept, an empirical sign will be that this measure demonstrates significant stability over time, independent of changes in respondents’ self-reported physical or mental health status. As noted earlier, this finding would counter the spontaneous assessment view.

The indirect role of health-improvement goals provides a still more definitive point of contrast. This complex role, illustrated by the rightward, vertical elements of Fig. 1, involves not only one's performance but also the meaning one assigns to performance of health-related behaviours. Previous research on the stability of the self-concept does not deny that people are motivated by images of what they would like to become, or to avoid becoming. These images serve as goals that motivate behaviour and provide a context for self-evaluation (Hooker & Kaus, 1994; Lavallee & Campbell, 1995; Markus & Nurius, 1986; Ryff, 1991). For example, the impact of successfully maintaining weight loss over 1 yr on an individual's self-esteem is apt to be greater for one who strives to fulfill a more slender self-image, than for one who lacks a self-image in this domain. We propose that if self-rated health reflects these self-strivings, then it should be affected by performance of the corresponding health behaviour, particularly among those who currently report this striving in terms of the intention to improve. By the same token, other potential determinants such as physical or mental health status should have a smaller influence on the self-rated health of individuals with an intention to improve, since their attention may be focused narrowly on the relevant behaviour.

The NPHS included measures that we felt were amenable to interpretation as evidence of respondents’ goals for health behaviours. These measures consist of respondents’ freely stated intentions to make specific behavioural changes to improve their health in the coming year. According to the enduring self-concept view, self-rated health should reflect changes in the identified health-related behaviour to a greater extent, and other indicators of health status to a lesser extent, when these intentions are present than when they are lacking.

In summary, we sought to answer two specific research questions representing the theoretical contrasts between the spontaneous assessment and enduring self-concept views:

  • (1)

    Does self-rated health exhibit significant stability over time after controlling for changes in respondents’ self-reported physical and mental health status, personal health practices, and social support, during the same period of time? The results would support the spontaneous assessment view if respondents’ self-rated health at baseline failed to significantly predict their self-rated health 2 yr later, after changes in their self-reported physical and mental health status, personal health practices, and social support were controlled. An additional requirement would be that these changes do significantly predict later self-rated health. By contrast, the results would support the enduring self-concept view if respondents’ self-rated health at baseline still significantly predicted their self-rated health 2 yr later, after the intervening changes were controlled.

  • (2)

    Does the relation between changes in personal health practices and self-rated health depend on whether or not respondents intend to improve their health by means of these behaviours? The results would support the spontaneous assessment view if this relationship did not vary according to whether respondents expressed a health-improvement goal in the relevant behavioural domain. The results would support the enduring self-concept view if this relationship proved stronger among respondents who expressed a relevant goal than among those who did not, and if the former respondents showed correspondingly less sensitivity to changes in other indicators of their health status, compared to the remainder of respondents.

Section snippets

Sample

Statistics Canada administered the NPHS to a national probability sample of household residents in all provinces, excluding those living on Indian Reservations, Canadian Forces bases, in the Yukon and Northwest Territories, or in long-term care institutions. Computer-assisted interviews were conducted in person, initially during 1994–95 (hereafter 1994) and again during 1996–97 (hereafter 1996), with individuals aged 12 and older who were selected at random within households. In the initial

Incidence of change in self-rated health

The incidence of change in self-rated health was established by cross tabulating respondents’ 1994 and 1996 self-rated health. The cross-sectional distributions of self-rated health scores in each wave of the NPHS were nearly identical. Yet self-rated health status changed over the period from 1994–1996 for nearly half the sample (48%), of whom 24% reported better health, and 24% reported worse health. The correlation of 1994 with 1996 self-rated health was r=0.57, p<0.001, suggesting moderate

Discussion

The purpose of this research was to compare the utility of the spontaneous assessment and enduring self-concept perspectives for interpreting the meaning of self-rated health. Consistent with the spontaneous assessment view, people's evaluations of their general health are dynamic, changing within a 2-yr period for approximately half the adult population. Change in self-rated health is clearly not random but rather coincides with changes in self-reported physical and mental health, and to a

Acknowledgements

This research was supported by a grant to all three authors from Health Canada, National Health Research and Development Program Grant No. 6607-1794-NPHS. The research and analysis are based on data from Statistics Canada, and the opinions expressed do not represent the views of Statistics Canada. We gratefully acknowledge comments on an earlier draft of this manuscript by Verena Menec and Jacquie Vorauer.

References (60)

  • A.J. Barsky

    The paradox of health

    New England Journal of Medicine

    (1988)
  • J. Berthelot et al.

    The calculation of health-adjusted life expectancy for a Canadian province using a multi-attribute utility functionA first attempt

  • E.A. Borawski et al.

    Meaning of older adults’ health appraisalsCongruence with health status and determinant of mortality

    Journal of Gerontology: Social Sciences

    (1996)
  • N. Cantor et al.

    Personality and social intelligence

    (1987)
  • J.G. Chipperfield

    Incongruence between health perceptions and health problemsImplications for survival among seniors

    Journal of Aging and Health

    (1993)
  • R. Crawford

    Healthism and the medicalization of everyday life

    International Journal of Health Services

    (1980)
  • E.M. Crimmins et al.

    Getting better and getting worseTransitions in functional status among older Americans

    Journal of Aging and Health

    (1993)
  • K.F. Ferraro et al.

    Health trajectoriesLong term dynamics among black and white adults

    Journal of Health and Social Behaviour

    (1997)
  • T.F. Garrity et al.

    Factors influencing self-assessment of health

    Social Science & Medicine

    (1978)
  • M. Gold et al.

    Assessing the health of a nationThe predictive validity of a preference-based measure and self-rated health

    Medical Care

    (1996)
  • M.S. Goldstein et al.

    Predicting changes in perceived health status

    American Journal of Public Health

    (1984)
  • J.M. Gottman et al.

    The analysis of changeIssues, fallacies, and new ideas

    Journal of Consulting and Clinical Psychology

    (1993)
  • A.G. Greenwald

    The totalitarian egoFabrication and revision of personal history

    American Psychologist

    (1980)
  • K. Hooker et al.

    Health-related possible selves in young and middle adulthood

    Psychology and Aging

    (1994)
  • E.L. Idler et al.

    Self-rated health and mortalityA review of twenty-seven community studies

    Journal of Health and Social Behaviour

    (1997)
  • E.L. Idler et al.

    The meanings of self-ratings of healthA qualitative and quantitative approach

    Research on Aging

    (1999)
  • E.L. Idler et al.

    Health perceptions and survivalDo global evaluations of health status really predict mortality?

    Journal of Gerontology: Social Sciences

    (1991)
  • E.L. Idler et al.

    Self-evaluated health and mortality among the elderly in New Haven, Connecticut, and Iowa and Washington Counties, Iowa, 1982–1986

    American Journal of Epidemiology

    (1990)
  • R.J. Johnson et al.

    The structure of health status among older adultsDisease, disability, functional limitation, and perceived health

    Journal of Health and Social Behaviour

    (1993)
  • G.A. Kaplan et al.

    Perceived health and mortalityA nine-year follow-up of the human population laboratory cohort

    American Journal of Epidemiology

    (1983)
  • Cited by (0)

    View full text