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

Self-Rated Current and Future Health Independently Predict Subsequent Mortality in an Aging Population

Divisions of ¹ Epidemiology ² Community Health & Human Development, School of Public Health, University of California, Berkeley.

Address correspondence to Constance Wang, PhD, Division of Epidemiology, University of California at Berkeley, School of Public Health, 140 Warren Hall, MC7360, Berkeley, CA 94720-7360. E-mail: constancew{at}berkeley.edu

	Abstract

Top Abstract Methods Results Discussion References

 
Background. Self-rated current health is an independent, robust predictor of subsequent mortality in older adults. Investigators hypothesize that individuals likely take into account their future health when reporting their current health. However, few have measured and examined self-rated future health in relation to mortality.

Methods. We investigate the effect of three self-rated health measures on 10-year mortality in 2091 men and women in an aging cohort: (i) self-rated current health, (ii) self-rated future health (1 year in the future), and (iii) a combined measure of current and future health. Vital status at follow-up year 10 was the outcome. We used data from SPPARCS (Study of Physical Performance and Age-Related Changes in Sonomans), a population-based, census-sampled, study of the epidemiology of aging, health, and functioning.

Results. Compared to those reporting their future health as better/same, participants reporting their future health as worse or unknown (don't know) experienced elevated 10-year mortality (adjusted rate ratio [RR] = 1.6, 95% confidence interval [CI], 1.2–2.1, p =.01), after adjustment for self-rated current health and other relevant covariates. The combined measure of current and future health also contributed important information. Compared to the referent (the best combination, current health excellent/good and future health better/same), participants reporting the worst combination, fair/poor current health and worse/unknown future health, experienced the highest 10-year mortality in the cohort (adjusted RR = 3.2, 95% CI, 2.2–4.7, p =.00).

Conclusions. Self-rated future health is an independent, robust predictor of mortality. It is as predictive of subsequent mortality in older adults as the standard measure of self-rated current health. Furthermore, a measure that combines self-reports of current health with future health was most useful in the identification of older adults with the highest mortality rates. Thus, the combined measure of current and future health may be most useful in practice, in distinguishing the differential mortality rates among persons reporting fair or poor self-rated current overall health.

Overall health is usually ascertained by asking study participants a very simple question: Overall, how would you rate your health? "Excellent," "Good," "Fair," or "Poor." Those who reported poorer current health have higher rates of mortality and functional decline, compared to respondents who reported excellent or good current health. This association remains robust (effect measure approximately 2.0) after adjustment for objective measures of health (e.g., physician-confirmed chronic conditions and morbidities) and for other covariates likely to explain the association between self-rated current health and subsequent mortality (4,5).

Research has focused on the criteria used by people to assess and report their own current health. People seem to tap into many types of information about themselves when asked to assess their own health. People may consider any combination of the following (3,6–8): (i) family, life, or medical history; (ii) past health status; (iii) capacity to do everyday tasks; (iv) self- and preventive care and adherence to medical or preventive advice; (v) resources including social support and access to care; and (vi) social factors including gender and social positioning.

Results from this research suggest that individuals may have an overall "sense" of their current health that is extraordinarily predictive of their future health, which, in turn, predicts their subsequent survival. Along these lines, we investigate the relationship between 10-year mortality in an aging cohort and three overall health status measures: (i) self-rated current overall health, (ii) self-rated health 1 year in the future, and (iii) a combined measure of current and future health. To our knowledge, there have been no prior studies examining the association between self-assessments of one's own future health and subsequent mortality.

	METHODS

Top Abstract Methods Results Discussion References

 
Population
This study used data collected from a community-based longitudinal study of the epidemiology of aging, health, and functioning: The Study of Physical Performance and Age-Related Changes in Sonomans (SPPARCS). Cohort characteristics by selected baseline variables appear in Table 1. Details of the recruitment and census-sampling procedures have been reported (9). Baseline examination was conducted in 1993 and included home interviews and laboratory assessments of a sample of 2091 residents of Sonoma, California, who were 53 years old or older. Follow-up examinations were conducted every 2 years in three waves, in 1995, 1997, and 1999, respectively. Ongoing vital status updates were implemented, and a full-cohort vital status update was completed on August 15, 2003. Informed written consents were obtained from participants at baseline and at each follow-up. Study protocols were approved by the Committees for the Protection of Human Subjects at the University of California, Berkeley, and the University of California, San Francisco.

Vital Status Updates
Assessment of vital status was conducted in several ways. Standardized active surveillance was used in combination with verification procedures. Study staff conducted a daily review of obituary records in the local newspaper and a monthly review of Sonoma County death certificates. Copies of the death certificates were obtained and were reviewed and verified by physician consultants. The exact dates (month, day, and year) of deaths were recorded.

Covariates
Information on covariates was obtained as part of the baseline examination. Covariates were selected based on established correlates and confounders of self-rated health and mortality. Covariates covered a wide range of domains thought to account for the association between self-rated health and subsequent mortality in the aging population, including: (i) sociodemographic and household characteristics, (ii) health practices, (iii) social engagement, (iv) clinical indicators of health status, and (v) standardized measures of health and functioning.

Sociodemographic and household characteristics include age (53–97), gender, marital status (married, divorced, separated, widowed, never married), race-ethnicity (white, black, other), birthplace (United States–born, foreign born), education (< 12 years, 12 years, some college, college graduate, > college), employment (working, not working, retired), annual household income (in U.S. dollars, < $10,000; $10,000–$14,999; $15,000–$19,999; $20,000–$24,999; $25,000–$29,999; $30,000+), and number of people residing in the household (1, 2, 3, 4+).

Health practices include smoking status (current, former, never), alcohol consumption (yes/no), vitamin use (yes/no), and physical activity (standard metabolic equivalent [MET] in standardized ordinal scale, < 4, 4, 4–6, > 6 METs). MET is the ratio of the metabolic rate for a specific activity divided by the basal metabolic rate.

Social engagement is the number of social contacts (0–1, 2–4, 5–10, 11+) and civic engagement (number of community-based organizations in which the participant is engaged in volunteer work: 0, 1, 2, 3+).

Self-reports of clinical indicators of health are the number of prescription medications (0, 1, 2, 3, 4+), number of over-the-counter medications (0, 1, 2, 3, 4+), number of symptoms in the past month (0–4, 5–9, 10–14, 15+), and number of diagnosed chronic conditions and morbidities (0, 1, 2, 3, 4+).

Health and functioning were determined by level of difficulty in the completion of generic functional tasks based on questions from Rosow and Breslau (11) and Nagi (12). Cognitive functioning impairment (yes/no) was determined based on the overall level of cognitive functioning from the Folstein Mini-Mental State Examination–Folstein (13). Depression (yes/no) was classified based on the standardized cut point of 16 on the Center for Epidemiologic Studies Depression (CES-D) scale (14).

Statistical Analysis
Crude 10-year mortality rates were estimated by dividing the number of deaths by the total cohort follow-up time. Age-specific mortality rates were estimated by four categories (ages 53–59, 60–69, 70–79, 80–97). Age-specific mortality rates were also presented by the three overall health assessments (Table 2). Mortality rates and corresponding 95% confidence intervals (CI) were presented per 1000 person-years. The lower bounds of the 95% CIs were truncated at 0 given that rates are, by definition, > 0.

This analysis is a standard fixed cohort analysis, as vital status was obtained for all cohort members who completed baseline protocols. Person-time (in years) was calculated based on the baseline interview date to the date of death prior to August 15, 2003, or August 15, 2003.

Cox proportional hazards models were used to estimate adjusted mortality rate ratios (RRs) and two-sided 95% CIs. All analyses were conducted using Stata, version 9.0 (College Station, TX) (15). Cox regression models with increasing numbers of covariates were used to evaluate the residual effect on mortality of the three health measures. Beginning with unadjusted 10-year mortality RRs for each health measure, we built models that were adjusted for progressively increasing number of covariates. Covariates were added by variable domains in blocks.

	RESULTS

Top Abstract Methods Results Discussion References

 
A total of 2091 cohort members contributed 17,424 at-risk person-years over 10 follow-up years. Each cohort member contributed, on average, 8.3 person-years (standard deviation 2.4 person-years; median 9.2 person-years; minimum 0.22 and maximum 10.3 person-years). During this period, 557 cohort members died. The crude 10-year mortality rate of the cohort was 32/1000 person-years (95% CI, 29/1000–31/1000). The mortality rate increased with increasing age (Table 2, age-specific mortality rates). In each 10-year age category, the age-specific mortality rates were lowest among participants who reported excellent/good current health, and among those who reported better/same future health, as well as among those whose combined current and future health was: current health excellent/good and future health better/same. The mortality rates increased as the corresponding ratings of current health, future health, and combined current and future health were progressively worse (Table 2). No gender differences were observed in reports of current or future health. The effect on mortality of each of the three health measures did not differ by age category, although the estimates fluctuated across categories of responses, due to small sample size in the younger ages. There was no correlation between the self-rated current and future health measure (Spearman = 0.003, p =.89).

Self-reported current health was independently associated with 10-year mortality (Table 3). Compared to participants who reported their current health as excellent/good (after adjustment for age, gender, and future health), those who reported their current health as fair/poor had elevated 10-year mortality (RR fair/poor = 2.0, 95% CI, 1.7–2.5; p =.00). The mortality RR remained elevated in the full model (Model 7) for participants who reported their current health as fair/poor (RR fair/poor = 1.8, 95% CI, 1.4–2.3; p =.01), compared to those who reported their current health as excellent/good.

Although assessments of both overall current and future health independently predicted subsequent mortality, the largest effect estimate was observed for the combined measure of current and future health. Compared to the referent (current health excellent/good and future health better/same), participants who reported their current health as excellent/good and their future health as worse/unknown had a greater mortality rate than those who reported current health as excellent/good and future health as better/same (Table 3, RR ranging from 1.6 to 2.0, compared to RR ranging from 1.2 to 1.5, respectively, depending on the number of covariates included in the model). Participants who reported the worst of the combinations (fair/poor current health and worse/unknown future health, compared to the referent), have the highest 10-year mortality RR in the cohort (Table 3, RR = 3.2, 95% CI, 2.2–4.7; p =.00) after adjustment for all covariates (Model 7).

	DISCUSSION

Top Abstract Methods Results Discussion References

 
Our findings provide new information about overall assessments of health and subsequent mortality in an aging population. We observed that a simple assessment of one's own future health (1 year in the future) is independently predictive of mortality over a 10-year period. This relationship remains robust after adjustment for self-rated current overall health and other relevant covariates. We observed that the magnitude of the effect of self-rated future health on subsequent mortality is approximately the same as the effect on mortality of self-rated current health. Furthermore, the use of a combined measure of current and future health may be more useful in practice. For example, adding a question about self-rated future health in either an epidemiologic study or as part of a clinical health interview provides practitioners an opportunity to distinguish among people who report their current health as either fair or poor. Among older people who reported their current health as fair/poor, those who also reported that their future health would be better/the same next year had a lower mortality rate (adjusted RR = 1.7, compared to those reporting best of the combinations) than those who reported that their future health would be worse/unknown (adjusted RR = 3.2, compared to those reporting the best of the combinations).

As mentioned previously, self-rated current health is one of the most robust predictors of subsequent mortality. There is ongoing speculation about why the assessment of one's current health is such a robust predictor of subsequent health, functional decline, and mortality (3–5). It may be that people predict their own future health as part of a comprehensive assessment. However, if it were only that, then we would expect that a single assessment of overall health, either current or future, but not both, would be independently associated with subsequent mortality. In other words, the residual effect on mortality of one measure, either the current or future health, would subsume the effect of the other measure, when both variables are in the model. Furthermore, if assessments of one's current and future health are based on overlapping underlying constructs, then the expected magnitude of the combined current and future health measures would have been approximately 1.6–1.8, similar to the effect of either measure mutually adjusted for the other. However, we observed a near-additive effect on 10-year mortality of the combined measure of current and future health for the worst combination (RR = 3.2, 95% CI, 2.2–4.7; p =.00), compared to the best combination. This observation suggests that all three health assessments are important in predicting subsequent mortality in an aging population: Self-rated current health and future health are equally important, and the combined measure of current and future health is useful in distinguishing among older adults who might experience the highest mortality rate.

Self-assessments of future health in older adults may be based on expectations of the future, for example, based on the type, severity, and treatment course for specific chronic conditions. This may include an appreciation of one's own health or disease trajectory to date. If the person perceives that she or he is declining, she or he may simply expect that the downward trajectory will continue. These expectations for the future may be based in part on having previously witnessed the course of friends and family members with similar conditions. In contrast, assessments of one's future health may be based on general feelings of optimism or pessimism about the future (16). In this case, subsequent mortality might not only be related to some form of individual self-monitoring of overall health and well-being but might also be related to one's general world view (17). For example, common psychosocial variables that have been shown to be associated with subsequent risk of death, such as "self-efficacy" (18), a "sense of control" (18,19), a "sense of destiny" (19), a "sense of coherence" (20), and "dispositional optimism" (21) all seem to imply an expectation about the future (22,23). Self-efficacy (18) implies a level of self-confidence in one's ability to achieve a particular goal. Sense of control or destiny (19) also implies the capacity to exert one's control over future events. A sense of coherence, as defined by Antonovsky (20), is "a global orientation that expresses the extent to which one has a pervasive, enduring though dynamic feeling of confidence that one's internal and external environments are predictable and that there is a high probability that things will work out as well as can reasonably be expected" (emphasis added). Finally, measures of "dispositional optimism" are intended to represent a general optimism about the future. People are asked to indicate whether they agree or disagree with statements such as "I often feel that life is full of promise" and "I still have positive expectations concerning my future" (22,24). Further conceptual work is necessary to examine whether self-ratings of future health overlaps with these underlying constructs.

Summary
Assessments of one's future health add an important and robust component of health and well-being that is independently associated with subsequent mortality in older adults. Furthermore, an overall health measure that combines self-rated current and future health might be most useful and valuable in practice settings in the identification of older adults with the highest mortality rate. The issue of whether reports of current health and of future health are based on similar or completely overlapping underlying constructs, as well as whether self-rated future health taps into such underlying constructs as "optimism" or "control of destiny," warrants further exploration.

	Acknowledgments

Top Abstract Methods Results Discussion References

 
Dr. Wang thanks the Robert Wood Johnson Health & Society Scholars Program for its financial support. This research was also supported in part by a grant (AG09389) from the National Institutes of Health, National Institute on Aging.

We thank Professor S. Leonard Syme for helpful comments.

Both authors contributed equally to the study concept and design, data analysis design, and the interpretation of results. C. Wang carried out the data analysis. Both authors drafted and edited the manuscript and contributed equally to the development of the intellectual content. Both authors had full access to the data from the SPPARCS study and take full responsibility for the integrity of the data and the accuracy of the data analysis.

	Footnotes

Top Abstract Methods Results Discussion References

 
Decision Editor: Luigi Ferrucci, MD, PhD

Received December 7, 2006

Accepted March 30, 2007

	References

Top Abstract Methods Results Discussion References

 

1. Mossey JM, Shapiro E. Self-rated health: a predictor of mortality among the elderly. Am J Public Health. 1982;72:800-808.[Abstract/Free Full Text]
2. Kaplan GA, Camacho T. Perceived health and mortality: a nine-year follow-up of the Human Population Laboratory Cohort. Am J Epidemiol. 1983;117:292-304.[Abstract/Free Full Text]
3. Idler EL, Benyamini Y. Self-rated health and mortality: a review of twenty-seven community studies. J Health Soc Behav. 1997;38:21-37.[Medline]
4. Ider EL, Russell LB, Davis D. Survival, functional limitations, and self-rated health in the NHANES I epidemiologic follow-up study, 1992. Am J Epidemiol. 2000;152:874-883.[Abstract/Free Full Text]
5. Lee Y. The predictive value of self assessed general, physical, and mental health on functional decline and mortality in older adults. J Epidemiol Community Health. 2000;54:123-129.[Abstract/Free Full Text]
6. Benyamini Y, Idler EL, Leventhal H, Leventhal EA. Positive affect and function as influences of self-assessments of health: expanding our view beyond illness and disability. J Gerontol Psychol Sci. 2000;55B:P106-P16.
7. Jylha M, Volpato S, Guralnik JM. Self-rated health showed a graded association with frequently used biomarkers in a large population sample. J Clin Epidemiol. 2006;59:465-471.[Medline]
8. Lyyra TM, Heikkinen E, Lyyra AL, Jylha M. Self-rated health and mortality: could clinical and performance-based measures of health and functioning explain the association? Arch Gerontol Geriatr. 2006;42:277-288.[Medline]
9. Satariano WA, Smith J, Swanson A, Tager IB. A census-based design for the recruitment of a community sample of older adults: efficacy and costs. Ann Epidemiol. 1998;8:278-282.[Medline]
10. Rothman KJ, Greenland S. Modern Epidemiology, 2nd ed. Philadelphia: Lippincott-Raven Publishers; 1998.
11. Rosow I, Breslau N. A Guttman health scale for the aged. J Gerontol. 1969;21:556-559.
12. Nagi SZ. An epidemiology of disability among adults in the United States. Millbank Mem Fund Q Health Soc. 1976;54:439-468.
13. Folstein MF, Folstein SE, McHugh PR. "Mini-mental state": a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12:189-198.[Medline]
14. Radloff LS. The CES scale: a self-report depression scale for research in the general population. Appl Psych Meas. 1977;1:385-401.
15. StataCorp. Stata statistical software: release 8.0. College Station, TX: StataCorp LP; 2003.
16. Smith N, Young A, Lee C. Optimism, health related-hardiness and well-being among older Australian women. J Health Psychol. 2004;9:741-753.[Abstract/Free Full Text]
17. Brissette I, Leventhal H, Leventhal EA. Observer ratings of health and sickness: can other people tell us anything about our health that we don't already know? Health Psychol. 2003;22:471-478.[Medline]
18. Bandura A. Self-efficacy: the exercise of control. New York: W. H. Freeman; 1997.
19. Syme SL. Control and health: an epidemiological perspective. In: Rodin J, Schooler C, Schaie KW, eds. Self-Directedness: Cause and Effects Throughout the Life Course. Hillsdale, NJ: Lawrence Erlbaum; 1990:213–229.
20. Antonovsky A. Health, stress, and coping. San Francisco, CA: Jossey-Bass; 1979:184.
21. Scheier MF, Carver CS. Optimism, coping and health: assessment and implications of generalized outcome expectancies. Health Psychol. 1985;4:219-247.[Medline]
22. Penninx BW, van Tilburg T, Kriegsman DM, Deeg DJ, Boeke AJ, van Eijk JT. Effects of social support and personal coping resources on mortality in older age: the Longitudinal Aging Study Amsterdam. Am J Epidemiol. 1997;146:510-519.[Abstract/Free Full Text]
23. Poppius E, Tenkanen L, Hakama M, Kalimo R, Pitkänen T. The sense of coherence, occupation and all-cause mortality in the Helsinki Heart Study. Eur J Epidemiol. 2003;18:389-393.[Medline]
24. Giltay EJ, Geleijnse MJ, Zitman FG, Hoekstra T, Schouten EG. Dispositional optimism and all-cause and cardiovascular mortality in a prospective cohort of elderly Dutch men and women. Arch Gen Psychiatry. 2004;61:1126-1135.[Abstract/Free Full Text]

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