Association of Self-Rated Health in Midlife With Mortality and Old Age Frailty: A 26-Year Follow-Up of Initially Healthy Men

Abstract

Self-rated health (SRH or “self-perceived health”) is considered to represent a global perception of one’s current state of health, which is not necessarily identical with objective health status ( 1–7 ). SRH is usually defined by asking individuals to evaluate their health status on a three- to five-point scale (with response options ranging from excellent to very poor) or to compare their health status with their age peers. As a person may be intuitively aware of pathologic processes far before they become measurable, SRH could be a valuable instrument for identifying persons at risk ( 1 , 5 ). A wide body of literature has risen around SRH with attempts to conceptualize it ( 7 ). Dimensions of SRH are numerous, but physical ones are probably very important ( 2 ). SRH has been associated with follow-up mortality in many relatively short-term studies ( 3 , 6 , 8 ), but decades-long studies are scarce ( 4 , 5 , 9 ).

Frailty is a geriatric syndrome characterized by excess vulnerability to stress, loss of resiliency, and multisystem decline, leading to a risk of disability, immobility, and death ( 10–12 ). Although it is a frequent and clinically important condition among older people, frailty still lacks a universal definition, and there are both phenotypic and frailty-index-based definitions currently in use ( 11 ). Studies on long-term predictors of frailty are scarce, but we have reported that in healthy men midlife clinical characteristics, such as obesity and cardiovascular disease (CVD) risk factors ( 13 ), as well as lower levels of leisure-time physical activity ( 14 ) were associated with the development of phenotypic frailty 26 years later. Because these characteristics are well known to associate with mortality, we hypothesized that SRH in midlife would be independently associated with frailty as well. To the best of our knowledge, this association has not been tested in long-term studies.

In the Helsinki Businessmen Study (HBS) ( 13–15 ), SRH was assessed in midlife, and mortality follow-up extends to old age when the presence of frailty also was determined using simplified, self-reported phenotypic criteria. The purpose of the present study was to investigate the long-term relationships between SRH, mortality, and frailty in the HBS cohort.

Participants and Methods

Study Population

Details of the HBS have been described in detail previously (refs. 13–15 ; full bibliography at http://www.gernet.fi/artikkelit/1463/helsinki-businessmen-study-hbs-summary-and-bibliography ). In brief, 3,490 men, mostly business-executives born between 1919 and 1934, participated in volunteer health check-ups during the 1960s and early 1970s organized by Finnish Institute of Occupational Health. At that time, they received health education to diminish their CVD risk. In 1974, men were evaluated by questionnaires and clinical and laboratory examinations to find participants in a CVD primary prevention trial. At that time, 63 men were dead, 563 men were found to be unsuitable for a primary prevention trial due to clinical diseases or medications, and 1,064 refused to participate or did not respond. The volunteering 1,815 men had a mean age of 47 years and were found to be professionally active and clinically healthy (without signs and symptoms of clinically meaningful chronic diseases including severe hypertension, drug-treated diabetes, CVD, cancer, metabolic, renal, and psychiatric diseases, and without regular medications). However, the majority of them had various untreated CVD risk factors.

Because preliminary analyses showed that participation in the primary prevention trial during the 1970s did not affect the present analyses 26 years later, all 1,815 men were included in the analysis in order to improve statistical power. Data of SRH were available of 1,753 healthy men (96.6%), and they form the population of the present study. The follow-up of the HBS has been approved by the ethical committee of the Department of Medicine, Helsinki University Central Hospital, and the study is registered as ClinicalTrials.gov identifier: NCT02526082.

Baseline Risk Factors

In 1974, the following CVD risk factors were assessed: body mass index (BMI) was calculated from measured weight and height as weight (kilograms) divided by height (meters) squared. Additionally, participants were asked to recall their weight at 25 years of age. Weight status was defined as normal weight (BMI < 25kg/m ² ), overweight (≥25 to <30kg/m ² ), and obese (≥30kg/m ² ). Systolic and diastolic blood pressure (mmHg) was registered after a 10-minute rest with the participant in a sitting position and using a mercury sphygmomanometer. Fasting serum, total cholesterol and triglyceride, and fasting blood glucose levels were measured using standard methods. Smoking (cigarettes per day) and alcohol consumption (grams per week) were assessed with questionnaires. Glucose tolerance was measured as blood glucose after 1 hour of a glucose load (1g per kg) administered orally. In 1974, the composite risk score for coronary artery disease (CAD) was calculated according to the contemporary Keys’ risk equation ( 16 ). It takes into account age, systolic blood pressure, cholesterol, smoking, and BMI and estimates the 5-year risk (%) of CAD among European men aged 40–59 years. CAD or CVD risk could not be assessed according to the Framingham risk equation because serum high-density lipoprotein cholesterol was not available.

Assessment of SRH

The way SRH is inquired has varied in studies. In our study in 1974, global SRH was assessed by asking “What do you think about your present state of health; is it ‘very good’, ‘fairly good’, ‘average’, ‘fairly poor’ or ‘very poor’.” Because there were very few men perceiving their status as “very poor,” they were combined with “fairly poor” as “poor.” This wording of SRH was similar to that used in the Whitehall II study ( 17 ). Perhaps the most widely used wording of SRH in the United States (and also used in SF-36/RAND-36) is excellent, very good, good, fair, or poor ( 18 ). In our study, this version was used in 2000 as part of the RAND-36 questionnaire (see below).

Follow-Up

Mortality was comprehensively followed up using national central registers (Population Information System). About 425 men out of 1,815 had died by the year 2000. Their frailty status (see below) at death was not known. The survivors ( n = 1,390, mean age 73 years) were sent a mailed questionnaire 26 years later. The questionnaire was resent once for nonrespondents, and in all 1,207 (86.8%) men responded. The questionnaire included items about anthropometric measures, living conditions, medication, and lifestyle factors (eg, alcohol consumption, smoking). History or presence of the following conditions were asked: hypertension, memory disturbances, stroke, coronary heart disease, heart failure, chronic pulmonary disease, claudication, diabetes, cancer, musculoskeletal disorder, psychiatric disease, any other long-term condition, or trauma. In addition, the Finnish version of the RAND-36-Item Health Survey 1.0 (practically identical to SF-36 and validated in the Finnish population) was embedded into the questionnaire ( 19 ). From the responses, a comorbidity index was calculated, taking into account the number and severity of comorbid conditions ( 20 ).

Frailty Assessment

In this study, phenotypic frailty status at follow-up in 2000 was defined by simplified self-reported criteria as a modification of the method initially described in the Cardiovascular Health Study and included shrinking, subjective exhaustion, physical inactivity, slow walking speed, and physical weakness ( 10 ). Because walking speed was not measured in our study in 2000, we used four criteria as follows: (i) shrinking was defined as weight loss of ≥5% from midlife or having current BMI < 21kg/m ² ; (ii) evaluation of physical weakness was based on self-reported difficulty in carrying or lifting a grocery bag (an item in the physical functioning scale of RAND-36); (iii) assessment of exhaustion was based on reported low energy most of the time during the preceding 4 weeks (an item in the vitality scale of RAND-36); and (iv) evaluation of physical inactivity was based on the question “Do you exercise regularly weekly?” The answer “No” was taken to denote physical inactivity. The participant was classified to be frail or prefrail if (iii)–(iv) or (i)–(ii) of the aforementioned criteria were met, respectively, and not frail if zero criteria were present. This frailty definition prospectively predicted important clinical end points (walking speed, mobility disability, and mortality) in our cohort ( 15 ).

Statistical Analysis

T -tests, nonparametric tests, and analyses of covariance were used where appropriate to compare continuous variables (mean with standard error [ SE ], logarithmic transformation where appropriate) across baseline SRH status. Chi-square and trend tests were used to compare proportions. Logistic regression was used to assess the relationship between SRH and mortality during follow-up and multinominal logistic regression (with forward stepping) was used to assess the relationship between midlife SRH and old age prefrailty and frailty. The largest SRH group (“Fairly good”) was taken as reference category. The results are presented as odds ratios (OR) with their 95% confidence intervals (CI). We report results as (i) unadjusted, (ii) adjusted for age, (iii) adjusted for the log Keys’ equation in midlife (includes age), and (iv) log Keys’ equation plus comorbidity in old age. The last one may actually be an over adjustment because comorbidities contribute to frailty. In statistical analyses, two-sided p values <.05 were taken as significant. The statistical software NCSS (version 2004, www.ncss.com , Kaysville, UT) was used for the statistical analyses.

Results

In 1974, when all men were clinically healthy without chronic diseases or regular medications, 124 (7.1%), 862 (49.2%), 706 (40.3%), and 61 (3.5%) rated their health as very good, fairly good, average, and poor, respectively. Table 1 shows clinical and laboratory characteristics according to SRH status in 1974. Poorer SRH was associated with higher BMI, systolic and diastolic blood pressure, triglycerides, and blood glucose, as well as with smoking.

During the follow-up through 2000 (at the mean age of 73 years), 410 men out of 1,753 (23.4%) had died. Of the deceased men, 32 (7.8%) rated their health very good in midlife, 183 (44.6%) fairly good, 174 (42.4%) average, and 21 (5.1%) poor. With fairly good SRH as reference and fully adjusted (log Keys’ equation reflecting combined CAD risk in midlife), SRH was related to mortality in a J-shaped fashion: OR: 1.65, 95% CI: 1.04–2.62 for very good; OR: 1.17, 95% CI: 0.91–1.50 for average; and OR: 2.15, 95% CI: 1.19–3.88 for poor SRH ( Table 2 ). In 2000, SRH and comorbidity status could be assessed in 1,180 (84.9% of survivors) and 1,184 men (85.2%), respectively. Among survivors, SRH in 1974 and 2000 were significantly associated (data not shown). Age-adjusted comorbidity index was 0.82 ( SE 0.15), 1.28 (0.05), 1.50 (0.06), and 1.84 (0.2) among men with very good, fairly good, average and poor midlife SRH, respectively ( p < .001). Frailty status could be assessed in 1,088 men (81.0% of survivors with SRH in 1974) and of those 434 (39.9%) were classified as not frail, 552 (50.7%) prefrail, and 102 (9.4%) frail. Among men without frailty assessment, the distribution of SRH in 1974 resembled that of men found to be prefrail in 2000, but as a whole, there was no significant difference in 1974 SRH between men with and without frailty assessment in 2000 ( p = 0.29).

Midlife SRH was related to both to prefrailty and frailty in old age. With fairly good SRH as reference and fully adjusted (combined CAD risk in midlife plus comorbidity in old age), average SRH in midlife was associated with a 1.5-fold increased risk of prefrailty (OR: 1.52, 95% CI: 1.14–2.04), and poor SRH with a 3.6-fold risk of prefrailty (OR: 3.56, 95% CI: 1.16–10.9), and 8.4-fold increased risk of frailty (OR: 8.38, 95% CI: 2.32–30.3) ( Table 2 ). The relationship between poor SRH and frailty was significant irrespective of adjustments, but it is of note that the analysis was based on only 31 individuals. As compared to fairly good SRH, the point estimates of very good SRH for prefrailty and frailty were below unity, but they were not statistically significant ( Table 2 ).

Discussion

Our results indicate that poorer SRH in clinically healthy midlife was related to phenotypic prefrailty and frailty status in old age. Although this association was independent of midlife CVD risk factor status, as well as of comorbidities in old age, the relationship can also be explained by unmeasured health attributes such as depression or subclinical disease at baseline. The results nevertheless suggest that the development of frailty is a long-term pathophysiological process. In contrast to frailty, the association between midlife SRH and subsequent mortality was not linear in our study.

Frailty and SRH have been associated in cross-sectional studies ( 21 ), but this is the first observation—to the best of our knowledge—of a long-term association between SRH and frailty.

There are several mechanisms that could mediate the long-term relationship. Inflammation and psychoneuroendocrine dysregulation seem to be involved in the development of frailty and sarcopenia ( 22–24 ), and low-grade inflammation may also be associated with poor SRH ( 25 , 26 ).

Depression is related to inflammation, immunology ( 22 , 27 , 28 ), and SRH ( 1 , 29–32 ), but the association between depression and frailty is complex ( 27 , 28 , 33–35 ). In our study, depression was not assessed at baseline, but the association between SRH in midlife and frailty in old age prevailed after adjustment for the mental health scale (reflecting depression) of RAND-36 in old age (data not shown). Also, other psychosocial processes, including individual constitution (optimism, pessimism), mood (anxiety), resources (coping, social support, meaning), and various stressors may affect vulnerability to illness via biopsychosocial model ( 21 , 36 ) and be associated with SRH ( 36 , 37 ). However, the effect of, for example, social support on frailty may be minor ( 36 , 38 , 39 ). Finally, one link between SRH and frailty could be the feeling of physical fitness. In a subcohort of the HBS, lower physical activity in midlife was strongly associated with frailty in old age ( 14 ), but in this subgroup, a significant association between SRH and frailty prevailed after midlife physical activity was adjusted for (T. Strandberg, MD, unpublished data, 2015).

Our study has not only several strengths but also limitations. Obvious strengths include the exceptionally long follow-up from the clinically healthy midlife up to old age, and the homogenous population in which socioeconomic confounders have been minimized. On the other hand, homogeneity is also a limitation, and generalization to other groups, for example women, should be done cautiously. The men were obviously not free of every possible disease (cf, Methods section) in 1974, but they were nevertheless considered suitable for a primary prevention trial. Frailty status was not formally assessed in 1974, but all men were professionally active at that time, and they did not have clinical diseases including CVD or diabetes, or regular medications. We did not evaluate the participants clinically in 2000, and self-report on prevalent diseases in 2000 was used to assess comorbidity. Also, the definition of frailty was based on questionnaire data, specifically from the RAND-36 health survey. However, there is no strict consensus for defining frailty, and various definitions may be used, also based on simplified, self-reported questionnaire data ( 11 ). Our criteria recognized about 10% of older men as frail, a figure quite similar to large epidemiological surveys using Fried criteria ( 12 , 40 ). But most importantly, the frailty criteria used in this study have been shown to be associated with future development of slower walking speed and disability and powerfully predict mortality ( 15 ). We believe that our definition can be taken to satisfactorily reflect phenotypic frailty. We could not assess the frailty status of those who died before 2000, and because frailty and mortality share common risk factors (eg, cardiovascular [ 24 ]), the long-term association between SRH and frailty may be underestimated, that is, drive the association between SRH and frailty toward null.

In conclusion, SRH in midlife is an important risk marker of phenotypic frailty in old age. Mechanisms behind this association clearly call for further research in order to better understand the long-term origins and better methods to predict and prevent frailty.

Funding

This work was supported by the Jahnsson Foundation, the University Central Hospitals of Oulu and Helsinki (EVO funding), the Konung Gustaf V:s och Drottning Victorias Frimurarestiftelse, a grant from the Academy of Finland (138730), and Paulon Säätiö. The funding sources had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.

Conflict of interest

The authors have no direct conflicts of interest related to this manuscript.

References

Author notes