Background
Major depression is known to be related to higher cardiovascular mortality.
However, epidemiological data regarding dispositional optimism in relation
to mortality are scanty.
Objective
To test whether subjects who are optimistic live longer than those who
are pessimistic.
Design
Our analysis formed part of a prospective population-based cohort study
in the Netherlands (Arnhem Elderly Study).
Setting
General community.
Participants
Elderly subjects aged 65 to 85 years (999 men and women) completed the
30-item validated Dutch Scale of Subjective Well-being for Older Persons,
with 5 subscales: health, self-respect, morale, optimism, and contacts. A
total of 941 subjects (466 men and 475 women) had complete dispositional optimism
data, and these subjects were divided into quartiles.
Main Outcome Measure
Number of deaths during the follow-up period.
Results
During the follow-up period of 9.1 years (1991-2001), there were 397
deaths. Compared with subjects with a high level of pessimism, those reporting
a high level of optimism had an age- and sex-adjusted hazard ratio of 0.55
(95% confidence interval, 0.42-0.74; upper vs lower quartile) for all-cause
mortality. For cardiovascular mortality, the hazard ratio was 0.23 (95% confidence
interval, 0.10-0.55) when adjusted for age, sex, chronic disease, education,
smoking, alcohol consumption, history of cardiovascular disease or hypertension,
body mass index, and total cholesterol level. Protective trend relationships
were observed between the level of optimism and all-cause and cardiovascular
mortality (P<.001 and P = .001
for trend, respectively). Interaction with sex (P = .04)
supported a stronger protective effect of optimism in men than women for all-cause
mortality but not for cardiovascular mortality.
Conclusions
Our results provide support for a graded and independent protective
relationship between dispositional optimism and all-cause mortality in old
age. Prevention of cardiovascular mortality accounted for much of the effect.
Many studies have consistently linked depression to an excess risk ofcardiovascular and all-cause mortality,1-7 whereasrelationships with positive aspects of personality have received less attention.The personality trait of optimism for a given individual is relatively stableacross time and has been related to better health outcomes. However, optimismhas been conceptualized in 2 rather different ways; that is, as an explanatory-stylemeasure by Peterson et al8-10 (ie,the general belief that the causes of bad events are not one’s own fault,are temporary, and are confined to the present circumstances rather than attributableto internal, stable, and/or global factors) and as dispositional optimismby Scheier et al11-14 (ie,generalized outcome expectancies that good things rather than bad things willhappen). On the one hand, evidence suggests that explanatory-style optimismhas been associated with better health and lower morbidity and mortality.9,10,15-18 Explanatory-styleoptimism was associated with a lower incidence of coronary heart disease incohort studies.16,18 On the otherhand, dispositional optimism has been linked to medical staff ratings of betterphysical health after surgery for heart transplantation,19 amore rapid recovery from coronary artery bypass surgery,13 anda lower rate of rehospitalization after coronary artery bypass grafting.14 The related score for positive life orientation waslinked to physicians’ and patients’ ratings of good recovery afterhospitalization for myocardial infarction.20 Anotherstudy found that dispositional optimism, assessed by the Life OrientationTest,11 was associated with better cancer survivalamong patients younger than 59 years but not in older patients.21 Becauseexplanatory-style and dispositional optimism do not strongly correlate witheach other,12,22,23 theseconstructs may represent different aspects of well-being. Although hope isnot equivalent to the expectation of a favorable outcome, as dispositionaloptimism is usually defined, hope may be a conceptually related construct.Hopelessness has been associated with an increased risk of fatal and nonfatalcardiovascular disease1,24,25 aswell as cancer25 and the progression of atherosclerosis26 in large cohorts with long follow-up periods. However,the benefits of dispositional optimism in increasing longevity and reducingcardiovascular mortality have not been clearly identified in older people.
To test the hypothesis whether subjects who are optimistic live longerthan those who are pessimistic, we performed a prospective analysis in elderlymen and women who were participants in the Arnhem Elderly Study. Dispositionaloptimism was assessed as a bipolar construct with questions about whethera participant thinks in a positive way, sees the future as meaningful andfulfilling, has a desire to achieve new goals, and has a sense of happinessand joy.27 To determine whether the trait ofoptimism was independently associated with known predictors of mortality,we adjusted for cardiovascular risk factors and sociodemographic characteristics.Potential confounders included dietary factors, smoking habits, obesity, andphysical activity, which are related to an individual's emotional state,28-33 andalcohol dependence, which is associated with a lack of optimism.30 Additionally,low socioeconomic status (eg, poverty) or a recent loss in the social networkmay diminish optimistic feelings and increase the risk of premature death.We also explored whether this potential relationship was modified by sex becausea lower level of explanatory-style optimism has been associated with prematuredeath, especially in men.10 However, sex differencesfor the effects of dispositional optimism received little attention.
The Arnhem Elderly Study is a population-based cohort study that startedin 1991-1992. We evaluated a random sample (stratified for age and sex) of1793 independently living men and women aged 65 to 85 years in the city ofArnhem, the Netherlands. Of these individuals, 49 were excluded because theywere institutionalized, had moved elsewhere, or had died. In addition, 732people (42% of the eligible subjects) refused participation for various reasons.A total of 1012 noninstitutionalized subjects (56%) agreed to be interviewed,and 685 agreed to undergo a physical examination and venipuncture.
The study design and population characteristics have been describedelsewhere.34 Subjects participating in thestudy were more likely to be men (52% vs 44%; P = .01)and of younger age (73.6 years vs 76.1 years; P<.001)than nonparticipants and subjects who were only interviewed. Other characteristics,including lifestyle factors and self-perceived health, did not significantlydiffer between these groups. All subjects provided written informed consent.The study was approved by the Ethical Committee of Wageningen University (Wageningen,the Netherlands).
Questionnaire on well-being and the optimism subscale
Subjective well-being was assessed by the Dutch Scale of SubjectiveWell-being for Older Persons (SSWO) developed by Groningen University (Groningen,the Netherlands).27,35,36 TheSSWO was constructed with 93 items from 5 existing scales. With factor analyses,this number was reduced to 30 items. The SSWO sum score is an indicator ofhow the elderly individual experiences subjective general well-being. Fivesubscales were identified through factor analysis: health (5 items; Cronbach α = .87),self-respect (7 items; α = .73), morale (6 items; α = .77),optimism (7 items; α = .76), and contacts (5 items; α = .65),accounting for 48.1% of variance.
The questionnaire included 3-point scales. Several scores for negativeitems had to be reversed so that all items were coded from 0 to 2, with higherscores indicating greater well-being. For each subscale and the sum score,a mean item score was calculated and multiplied by 10, resulting in rangesfrom 0 to 20. Validity was previously assessed by testing against objectivemeasures of well-being (eg, physical activity, mobility, use of health care,and activities of daily living). The test-retest reliability coefficient was0.85 for the total SSWO score and 0.76 for the optimism subscore.27 Validating the SSWO with the Hopkins Symptom Checklist37 in elderly subjects yielded correlation coefficientsranging from –0.50 to –0.70 (P<.01for all).27
Participants were ranked according to their SSWO subscores and dividedinto quartiles. Quartiles categorized a person as being a pessimist (ie, quartile1 = lowest score) or an optimist (ie, quartile 4 = highestscore), defined relative to one another. The 7 questions of the optimism subscaleconsisted of the following: “I often feel that life is full of promises,”“I still have positive expectations concerning my future,” “Thereare many moments of happiness in my life,” “I do not make anymore future plans,” “Happy laughter often occurs,” “Istill have many goals to strive for,” and “Most of the time Iam in good spirits.” Participants were asked to fill out the questionnaireat home and return it. Subscales that contained any blank items were excludedfrom the analyses.
Demographic, behavioral, and biological factors
Standardized data collection was performed by trained interviewers atbaseline. Physical activity was scored continuously according to a validatedquestionnaire on household activities, sports, and other leisure time activities.38 Dichotomous variables were created for sex (1 = men;2 = women), marital status (1 = living together as a marriedor unmarried couple; 2 = otherwise), physical disability (basedon the Activities of Daily Living Scale39;1 = having some or great problems with 1 or more of 22 activities;2 = otherwise), alcohol (1 = ≤1 alcoholic consumptionper day; 2 = otherwise), education (1=high school or university;2 = otherwise), and socioeconomic status (1 = housewives,unskilled and skilled workers, and lower employees; 2 = small-businessowners, employees, and higher professions). For married or widowed women,socioeconomic status was classified according to that of the husband. Smokingstatus was coded as current, former, or never. Respondents were asked if theyhad chronic or acute health conditions that might affect longevity (eg, cardiovasculardisease [ie, a history of heart disease or stroke], diabetes, or chronic obstructivepulmonary disease), which were represented by binary variables. In addition,the variable of chronic disease coded for the total number of chronic disordersand illnesses of respondents (0, 1, 2, 3, 4, or 5 or more from a list of 24;eg, chronic venous leg ulcers, chronic gastric disease, chronic low back pain,rheumatic disease, cancer, thyroid disease, and deafness). Subjects were consideredto be receiving cardiovascular medication if they used angiotensin-convertingenzyme inhibitors, anticoagulants, lipid-reducing agents, and/or salicylatesduring the 3 months prior to the interview. The semiquantitative food frequencyquestionnaire was based on the validated Euronut-SENECA [Survey in Europeon Nutrition and the Elderly, A Concerted Action] dietary history questionnaire40 and the Dutch Food Consumption Survey.41 Fatintake was calculated with the Dutch nutrient database.42
Body mass index was calculated by dividing weight in kilograms (to thenearest 0.5 kg with the subject dressed but not wearing shoes) by height inmeters squared (to the nearest 0.5 cm). Systolic and diastolic blood pressureswere measured twice in supine position with a random-zero sphygmomanometer(Hawksley Technology, Lancing, England), and the mean was used. Hypertensionwas defined as a blood pressure reading of 160/95 mm Hg or higher or the useof antihypertensive medication.
A single nonfasting blood sample was obtained in 641 subjects. Venipuncturewas performed between 8:00 AM and 5:30 PM usingcitrate collection tubes, and time of blood sampling was recorded. Sampleswere stored at –80°C. Plasminogen activator inhibitor type 1, themain inhibitor of fibrinolysis and a potential risk factor for cardiovasculardisease, was evaluated using the Chromolize kit (Biopool, Umeå, Sweden)and adjusted for time of blood sampling.43 Totalserum cholesterol level was determined enzymatically (cholesterol oxidase–phenolaminophenazone peroxidase reaction; Boehringer Mannheim, Mannheim, Germany),and high-density lipoprotein (HDL) and low-density lipoprotein (LDL) cholesterollevels were measured directly (Dimension HDL; Dade Behring Marburg, Marburg,Germany; and N-geneous LDL; Genzyme Diagnostics, Cambridge, Mass). We assessedC-reactive protein using a highly sensitive enzyme-linked immunosorbent assayprocedure. Several blood parameters could not be assessed for 31 subjects.
Municipal registries provided data on mortality and migration untilFebruary 2001. One person was lost to follow-up owing to emigration (and wascensored); in the remaining 1011 subjects, follow-up for all-cause mortalitywas complete.
Data regarding cause-specific mortality were obtained for the 641 subjectsfrom whom blood was sampled. These data were obtained from general physiciansby means of a standard form. Some subjects (n = 39) gave no permissionfor the collection of follow-up data, and for other subjects the general physiciancould not be traced (n = 31), did not cooperate (n = 39),or did not provide proper data (n = 10). Follow-up for cause-specificmortality was complete for 518 subjects (81%; 268 men and 250 women) to codefor cardiovascular death according to the InternationalClassification of Diseases, 10th Revision (ICD-10) (codes I00-I96). Characteristics of subjects who received follow-upwere similar to those of subjects who did not, except for a lower total cholesterollevel (232 mg/dL [6.0 mmol/L] vs 243 mg/dL [6.3 mmol/L]; P = .02).
First, exploratory factor analysis was performed for the 30 questionsof the SSWO questionnaire (with orthogonal varimax rotations and Kaiser-Meyer-Olkinand Bartlett tests44). Baseline characteristicswere reported as number (percentage) or (geometric) means (± SDor with 95% confidence interval [CI]). The positively skewed distributionsof plasma total, LDL, and HDL cholesterol, C-reactive protein, and plasminogenactivator inhibitor type 1 were log transformed. Comparisons between men andwomen and between quartiles for the SSWO scores were analyzed using Mann-Whitney,χ2, or independent t tests as appropriate.Linear trends from the lowest to the highest quartile were tested. The Kaplan-Meiermethod was used to examine crude all-cause and cardiovascular mortality (Figure 1).
Second, we analyzed optimism in relation to all-cause mortality in 941subjects who completed all of the optimism questions. Cox proportional hazardsanalysis was used to adjust for potential confounders (eg, sex, age, and sociodemographicand cardiovascular covariates). We explored the associations of quartilesof the SSWO total score and its subscales with mortality, using the firstquartile as the reference category. To examine whether the relationship differedbetween men and women, tests for interaction were performed by entering cross-productterms for sex and optimism into the Cox model (for the highest vs lowest quartile).The analyses were repeated stratified for sex. In post-hoc analyses, we testedeach of the 7 questions of the optimism subscale for the highest discriminatingpower by dichotomizing the answer categories.
Third, we analyzed optimism in relation to cardiovascular mortalityusing Cox proportional hazards analysis, similar to the analyses of all-causemortality. P<.05 was considered statisticallysignificant using a 2-tailed test. The software used was SPSS version 10.0(SPSS Inc, Chicago, Ill).
The SSWO was filled in by 999 (98.7%) of 1012 subjects. Of those subjects,893 completed all 30 questions and 941 (93.0%) completed the 7 questions ofthe optimism subscale. The main reasons for not completing the SSWO questionnairewere difficulties with questions and forgetfulness. The 71 nonresponding subjectsas compared with the 941 subjects who did complete the questions on optimismdid not differ in sex, body mass index, marital status, socioeconomic status,physical activity, or physical disability; however, they were slightly older(mean age, 75.7 vs 74.5 years; P = .09)and reported slightly more chronic disease (P = .07).Factor analysis confirmed that the 30 SSWO questions yielded 5 reliable subscales(Kaiser-Meyer-Olkin measure, 0.89; Bartlett test, P<.001).Following varimax rotation, the 7 items of the optimism subscale were loadedinto 1 cluster similar to the original SSWO questionnaire (item loading onthis cluster ranged from 0.28 to 0.77), although 4 of the 7 items also loadedinto the contacts subscale.
The 941 subjects (466 men [49.5%] and 475 women) who completed the 7optimism questions had a mean ± SD age of 74.5 years. Themedian ± SD optimism subscore was 12.9 ± 4.8(range, 0-20.0); this score was similar for men and women (P = .77) and was divided into quartiles (quartile 1, 0-8.6;quartile 2, 10.0-12.9; quartile 3, 14.3-15.7; and quartile 4, 17.1-20.0).The medical history included hypertension in 215 subjects (22.8%), cardiovasculardisease in 41 subjects (4.4%), diabetes mellitus in 54 subjects (5.7%), andchronic obstructive pulmonary disease in 34 subjects (3.6%). The mean ± SDbody mass index was 25.9 ± 3.7, and the mean ± SDdiastolic and systolic blood pressures were 81.4 ± 10.9 and150.1 ± 20.9 mm Hg, respectively. On average, men were younger(mean age, 73.9 years vs 75.0 years), less obese (mean body mass index, 25.4vs 26.4), more physically active, less often living alone, of higher socioeconomicstatus, more often smokers or former smokers, more often consuming 2 or morealcoholic drinks per day, and less often experiencing diabetes mellitus, hypertension,or chronic disease in general than women (P<.005for all). Only 2 subjects reported a depressive disorder and 1 a nonspecifiedpsychiatric disorder.
For both sexes, higher optimism subscores were associated with youngerage, less chronic disease, a higher health subscale score, a higher totalactivity score, and less physical disability (P<.001for all). In men, higher optimism subscores were also associated with livingtogether vs alone (P = .005), a higherlevel of education (P = .004), more vitaminuse (P = .02), less use of primary healthcare (P = .02), and less often currentlysmoking (P = .03). In women, higher optimismsubscores were also associated with living together vs alone (P<.001), a higher level of education (P<.001),no medical history of diabetes (P = .02),less use of primary health care (P = .003),less often currently smoking (P = .03),and more often consuming either no alcohol (P = .01)or a moderate amount of alcohol (ie, 1-2 consumptions per day; P = .005). Optimism was unrelated to dietary variables, bodymass index, blood pressure, or any serum or plasma parameter.
Optimism and all-cause mortality
During the mean ± SD follow-up period of 9.1 ± 0.1years, 397 (42%) of 941 subjects died (48.8% of men vs 35.9% of women; P<.001). Using January 1996 through December 2000 asa reference period, sex-specific data, and the same age distribution as thecohort that survived until January 1, 1996, the mortality rates of our cohortand the Dutch population45 were comparable:all-cause mortality was 37.1% and 27.9% for men and women participating inour cohort, as compared with 37.8% and 30.2%, respectively, among the wholeDutch population.
The SSWO total score was associated with a decreased risk of all-causemortality (P<.001 for trend) (Table 1). For the SSWO subscores, only the health and optimism subscoreswere associated with a decreased risk of all-cause mortality (P<.001 for trend) (Table 1).The crude death rates were 56.5%, 45.1%, 38.2%, and 30.4% for quartiles 1to 4 of the optimism subscale (ranging from a high level of pessimism to ahigh level of optimism), respectively, in men and women combined (P<.001 for trend). Compared with subjects with a high level of pessimism(quartile 1), those reporting a high level of optimism (quartile 4) had anage- and sex-adjusted hazard ratio of 0.55 (95% CI, 0.42-0.74). When adjustedfor age, sex, smoking, alcohol consumption, education, total activity score,socioeconomic status, and marital status, this hazard ratio was 0.71 (95%CI, 0.52-0.97; P = .02 for trend).
As expected, women had lower all-cause mortality than men (hazard ratio,0.51; 95% CI, 0.41-0.64). The Kaplan-Meier curves suggested modification ofthe relationship between optimism and mortality by sex; that is, the protectiveeffect of optimism seemed stronger in men than in women (Figure 1). An interaction term between sex and optimism was addedto the multivariate model with sex and age and did show a statistically significantinteraction (P = .04), supporting a strongerprotective effect of optimism in men as compared with women. This interactionwas of borderline statistical significance (P = .07)after adjustment for age, sex, chronic disease, smoking, alcohol consumption,education, total activity score, socioeconomic status, and marital status.
Figure 1 illustrates survivalrates among quartiles of the optimism subscale. In men, death rates were reducedby 1.0%, 14.0%, and 62.9% across the second to fourth quartiles (using thelower quartile as the reference) at 6 years of follow-up. In women, correspondingdeath rates were 8.6%, 22.7%, and 34.8% at 6 years, respectively. Figure 2 shows the hazard ratios according toage groups and optimism quartiles in men and women separately and suggeststhat the level of optimism is inversely associated with mortality rate, independentof age. Table 2 and Table 3 show the unadjusted and adjusted hazard ratios for all-causemortality in men and women, respectively. In both sexes, adjusting for thesubjective health score attenuated the relationship more so than adjustmentfor chronic disease. After adjustment for prognostic (sociodemographic) factorsassociated with optimism, the hazard ratio in men (but not in women) was statisticallysignificant (men: 0.58; 95% CI, 0.37-0.91 [Table2]; women: 0.80; 95% CI, 0.51-1.25 [Table 3]).
In post-hoc analyses, the questions “I do not make any more futureplans” and “I still have many goals to strive for” allowedfor a highest significant discrimination in men (hazard ratio, 0.55 [95% CI,0.42-0.73] and 0.53 [95% CI, 0.39-0.71], respectively). In women, the question“There are many moments of happiness in my life” showed the highestdiscriminating power (hazard ratio, 0.70 [95% CI, 0.50-0.99]).
Optimism and cardiovascular mortality
In the subgroup of 494 subjects (259 men and 235 women) who receivedfollow-up for cause-specific mortality, 110 men and 69 women died (42% ofmen vs 29% of women; P = .002). A totalof 41 deaths in men and 25 deaths in women were due to cardiovascular disease(37% and 36%, respectively; P>.99 for sex difference).There were 29 incident myocardial infarctions (15 fatal), 48 transient ischemicattacks, 57 strokes (38 fatal), 11 fatal events due to other cardiovascularcauses (ICD-10 codes I00-I99), and 113 fatal eventsof noncardiovascular (35 malignancies) or unknown cause. One man committedsuicide at age 79 years.
The SSWO total score was not significantly associated with a decreasedrisk of cardiovascular mortality (Table 1).For the SSWO subscores, only the optimism subscore was associated with a decreasedrisk of cardiovascular mortality (P = .001for trend) (Table 1). Compared withsubjects with a high level of pessimism, those reporting a high level of optimismhad a multivariate-adjusted hazard ratio of 0.27 (95% CI, 0.12-0.57). Whenadjusted for age, sex, chronic disease, education, smoking, alcohol consumption,history of cardiovascular disease or hypertension, body mass index, and totalcholesterol level, the hazard ratio was 0.23 (95% CI, 0.10-0.55).
Table 2 and Table 3 show the unadjusted and adjusted hazard ratios for cardiovascularmortality in men and women, respectively. The protective effect of dispositionaloptimism was somewhat stronger in men than in women, and the age- and sex-adjustedinteraction term with sex was of borderline statistical significance (P = .09) but not after multivariate adjustment(P = .26). After multivariate adjustmentfor cardiovascular risk factors or a combination of sociodemographic and cardiovascularrisk factors, there was still a protective effect of dispositional optimismagainst cardiovascular mortality in both men and women.
We found that dispositional optimism was predictive of lower all-causeand cardiovascular mortality in elderly subjects. Prevention of cardiovascularmortality accounted for much of the effect on all-cause mortality. Trend relationshipswere found between the level of dispositional optimism and mortality, andoptimism remained a significant predictor of mortality even after adjustmentfor many potential confounders. In addition, we found that the beneficialeffect of dispositional optimism on mortality was significantly stronger inmen than in women; this was also apparent in multivariate analyses. Therefore,our findings indicate that dispositional optimism is an independent determinantand that men predisposed to optimism experience a higher survival benefitthan women predisposed to optimism.
Our findings are consistent with those from other prospective studies,1,13,14,19,20,24-26 yetcohort studies of a possible association between dispositional optimism andmortality are rare.21 Four cohort studies (1from Appels and Mulder,24 1 from Anda et al,1 and 2 from Everson et al25,26)found an adverse effect of hopelessness on the progression of disease, morbidity,and mortality. Notably, questions about having future plans and the desireto achieve new goals provided the largest discriminatory power among men inour study. Other previous studies used an explanatory-style measure of optimism.9,10,15-18 Thesestudies also showed that optimists report better physical and mental health9,17 and experience lower rates of fatalmyocardial infarction and coronary death16,18 aswell as better survival.10,15 The2 constructs of optimism are not strongly correlated,12,22,23 sothese may represent different aspects of the tendency to respond in a positivelytoned manner to a variety of stimuli.22 Theadvantage of our study is that it considerably strengthens the relationshipbetween dispositional optimism and lower (cardiovascular) mortality.
The mechanism underlying the link between optimism and mortality remainsunclear, and there are several possible explanations. First, optimism wasrelated to a higher level of physical activity, moderate alcohol use in women,and less smoking.31-33 However,in our study the association persisted even after adjustment for these potentialconfounding health behaviors. We also found that subjects are more optimisticwhen they live with a spouse or have a higher educational level, yet in ourstudy these factors did not substantially modify the relationship betweenoptimism and mortality. Furthermore, optimistic subjects did not have healthierdiets, as indicated by similar intakes of total and saturated fat. Finally,no associations were found between optimism and body mass index, blood pressure,or plasma and serum cardiovascular risk markers.
Second, optimism is associated with better health in general. It seemsplausible that clinical or subclinical disease increases pessimism, resultingin a spurious association between optimism and mortality due to reverse causality.33 In our study, subjects who perceived their healthas bad or deteriorating (assessed by the health subscale of the SSWO questionnaire)were more likely to have a high level of pessimism, as would be expected,because these measures are not independent.22,33 Optimisticsubjects may be biased toward reporting better health and vice versa. Whenwe included the health subscale in a separate multivariate analysis, the associationsbetween optimism and mortality were attenuated but not eliminated. When weincluded the more objective measures of chronic disease or physical disabilityin the analyses, there was less attenuation. Thus, baseline health may explainonly part of the relationship between optimism and mortality.
Third, optimists may cope differently and more effectively than pessimistsdo.9,22,31,46 Optimismcorrelated positively with problem-focused coping and seeking social support.47 Therefore, an optimistic person may be more likelyto have habits that enhance health or a recovery process; for example, theymay be more compliant with their medical treatment regimens.
Fourth, other biological mechanisms that have been suggested in thelink between major depression and excess mortality include effects of geneticfactors, the immune system, sympathoadrenal system, endorphins, steroidalhormones in the hypothalamic-pituitary-adrenal axis, heart rate variability,and platelet function.7,31,48 Thefindings that dispositional optimism in our study and hopelessness in anotherstudy25 were related to both cardiovascularand all-cause mortality suggest that a combination of factors are involved.To provide further evidence for causality, randomized studies are needed tofind out whether stimulating future optimism and hope increases life expectancyin elderly individuals, yet such psychosocial intervention studies are difficultto perform and interpret.33,49
The potential limitations of our cohort study merit careful consideration.We used only a single questionnaire; thus, our results do not necessarilyapply to other scales of optimism. The optimism subscale has not been validatedagainst the Life Orientation Test, which measures dispositional optimism.11 The lack of face-to-face interviews to complete questionnairesmay have limited the precision. Potential confounders were assessed only ina subgroup of participants, leading to reduced statistical power in multivariateanalyses. The association between optimism and sociodemographic characteristicsand other risk factors may be subtle, and the overall association with mortalitycould be (partly) ascribed to residual confounding and reverse causation.33 Moreover, we cannot exclude the possibility thatthe apparent interaction of optimism with sex is a function of different levelsof power between the sexes because mortality rates were higher in men thanwomen. The strengths of this study are the longitudinal design with a follow-upperiod of 9 years and the hard clinical end point of death. Our mortalityrate was generally similar to that reported in the Netherlands, suggestinga wide relevance. Further validation is suggested by the expected baselineassociations of optimism with sociodemographic variables. We also includedrisk factors known to predict mortality, suggesting that the relationshipbetween optimism and mortality was not due to confounding.33
In conclusion, we found that the trait of optimism was an importantlong-term determinant of all-cause and cardiovascular mortality in elderlysubjects independent of sociodemographic characteristics and cardiovascularrisk factors. A predisposition toward optimism seemed to provide a survivalbenefit in elderly subjects with relatively short life expectancies otherwise.Our results, combined with the finding that hopelessness was associated withan increased incidence or progression of disease,1,24-26 suggestthat dispositional optimism affects the progression of cardiovascular disease.Although optimism reduces the risk of cardiovascular death through mechanismslargely unaffected by baseline values of physical activity, obesity, smoking,hypertension, and lipid profile, pessimistic subjects may be more prone tochanges across time in risk factors that affect the progression of cardiovasculardisease (eg, the development of smoking habits, obesity, or hypertension)than optimistic subjects. Dispositional optimism may also be associated withbetter coping strategies that are adhered to throughout life.
Submitted for Publication: January 22, 2004;final revision received April 19, 2004; accepted April 21, 2004.
Correspondence: Erik J. Giltay, MD, PhD,Psychiatric Center GGZ Delfland, PO Box 5016, 2600 GA Delft, the Netherlands(giltay@dds.nl).
Funding/Support: This study was part of theresearch program Lifestyle and Health in the Elderly supported by the DutchMinistry of Health, Welfare, and Sport (The Hague, the Netherlands).
1.Anda
RWilliamson
DJones
DMacera
CEaker
EGlassman
AMarks
J Depressed affect, hopelessness, and the risk of ischemic heart diseasein a cohort of US adults.
Epidemiology 1993;4285- 294
PubMedGoogle ScholarCrossref 2.Glassman
AHShapiro
PA Depression and the course of coronary artery disease.
Am J Psychiatry 1998;1554- 11
PubMedGoogle Scholar 3.Penninx
BWGeerlings
SWDeeg
DJvan Eijk
JTvan Tilburg
WBeekman
AT Minor and major depression and the risk of death in older persons.
Arch Gen Psychiatry 1999;56889- 895
PubMedGoogle ScholarCrossref 4.Cohen
HWMadhavan
SAlderman
MH History of treatment for depression: risk factor for myocardial infarctionin hypertensive patients.
Psychosom Med 2001;63203- 209
PubMedGoogle ScholarCrossref 5.Penninx
BWBeekman
ATHonig
ADeeg
DJSchoevers
RAvan Eijk
JTvan Tilburg
W Depression and cardiac mortality: results from a community-based longitudinalstudy.
Arch Gen Psychiatry 2001;58221- 227
PubMedGoogle ScholarCrossref 6.Ganguli
MDodge
HHMulsant
BH Rates and predictors of mortality in an aging, rural, community-basedcohort: the role of depression.
Arch Gen Psychiatry 2002;591046- 1052
PubMedGoogle ScholarCrossref 7.May
MMcCarron
PStansfeld
SBen-Shlomo
YGallacher
JYarnell
JSmith
G DaveyElwood
PEbrahim
S Does psychological distress predict the risk of ischemic stroke andtransient ischemic attack? the Caerphilly Study.
Stroke 2002;337- 12
PubMedGoogle ScholarCrossref 9.Peterson
CSeligman
MEVaillant
GE Pessimistic explanatory style is a risk factor for physical illness:a thirty-five-year longitudinal study.
J Pers Soc Psychol 1988;5523- 27
PubMedGoogle ScholarCrossref 10.Peterson
CSeligman
MEYurko
KHMartin
LRFriedman
HS Catastrophizing and untimely death.
Psychol Sci 1998;9127- 130
Google ScholarCrossref 11.Scheier
MFCarver
CS Optimism, coping, and health: assessment and implications of generalizedoutcome expectancies.
Health Psychol 1985;4219- 247
PubMedGoogle ScholarCrossref 12.Scheier
MFCarver
CS Dispositional optimism and physical well-being: the influence of generalizedoutcome expectancies on health.
J Pers 1987;55169- 210
PubMedGoogle ScholarCrossref 13.Scheier
MFMatthews
KAOwens
JFMagovern
GJ
SrLefebvre
RCAbbott
RACarver
CS Dispositional optimism and recovery from coronary artery bypass surgery:the beneficial effects on physical and psychological well-being.
J Pers Soc Psychol 1989;571024- 1040
PubMedGoogle ScholarCrossref 14.Scheier
MFMatthews
KAOwens
JFSchulz
RBridges
MWMagovern
GJCarver
CS Optimism and rehospitalization after coronary artery bypass graft surgery.
Arch Intern Med 1999;159829- 835
PubMedGoogle ScholarCrossref 15.Maruta
TColligan
RCMalinchoc
MOfford
KP Optimists vs pessimists: survival rate among medical patients overa 30-year period.
Mayo Clin Proc 2000;75140- 143
PubMedGoogle ScholarCrossref 16.Kubzansky
LDSparrow
DVokonas
PKawachi
I Is the glass half empty or half full? a prospective study of optimismand coronary heart disease in the Normative Aging Study.
Psychosom Med 2001;63910- 916
PubMedGoogle ScholarCrossref 17.Maruta
TColligan
RCMalinchoc
MOfford
KP Optimism-pessimism assessed in the 1960s and self-reported health status30 years later.
Mayo Clin Proc 2002;77748- 753
PubMedGoogle ScholarCrossref 18.Todaro
JFShen
BJNiaura
RSpiro
AWard
KD Effect of negative emotions on frequency of coronary heart disease(the Normative Aging Study).
Am J Cardiol 2003;92901- 906
PubMedGoogle ScholarCrossref 19.Leedham
BMeyerowitz
BEMuirhead
JFrist
WH Positive expectations predict health after heart transplantation.
Health Psychol 1995;1474- 79
PubMedGoogle ScholarCrossref 20.Agarwal
MDalal
AKAgarwal
DKAgarwal
RK Positive life orientation and recovery from myocardial infarction.
Soc Sci Med 1995;40125- 130
PubMedGoogle ScholarCrossref 22.Tomakowsky
JLumley
MAMarkowitz
NFrank
C Optimistic explanatory style and dispositional optimism in HIV-infectedmen.
J Psychosom Res 2001;51577- 587
PubMedGoogle ScholarCrossref 25.Everson
SAGoldberg
DEKaplan
GACohen
RDPukkala
ETuomilehto
JSalonen
JT Hopelessness and risk of mortality and incidence of myocardial infarctionand cancer.
Psychosom Med 1996;58113- 121
PubMedGoogle ScholarCrossref 26.Everson
SAKaplan
GAGoldberg
DESalonen
RSalonen
JT Hopelessness and 4-year progression of carotid atherosclerosis: theKuopio Ischemic Heart Disease Risk Factor Study.
Arterioscler Thromb Vasc Biol 1997;171490- 1495
PubMedGoogle ScholarCrossref 27.Tempelman
CJJ
Welbevinden bij ouderen: konstruktie van een meetinstrument [Well-being in the elderly: development of theScale Subjective Well-being Older Persons] [dissertation]. Groningen,the Netherlands University of Groningen1987;
Google Scholar 28.Farmer
MELocke
BZMoscicki
EKDannenberg
ALLarson
DBRadloff
LS Physical activity and depressive symptoms: the NHANES I EpidemiologicFollow-up Study.
Am J Epidemiol 1988;1281340- 1351
PubMedGoogle Scholar 29.Glassman
AHHelzer
JECovey
LSCottler
LBStetner
FTipp
JEJohnson
J Smoking, smoking cessation, and major depression.
JAMA 1990;2641546- 1549
PubMedGoogle ScholarCrossref 30.Prescott
CANeale
MCCorey
LAKendler
KS Predictors of problem drinking and alcohol dependence in a population-basedsample of female twins.
J Stud Alcohol 1997;58167- 181
PubMedGoogle Scholar 32.Harper
SLynch
JHsu
WLEverson
SAHillemeier
MMRaghunathan
TESalonen
JTKaplan
GA Life course socioeconomic conditions and adult psychosocial functioning.
Int J Epidemiol 2002;31395- 403
PubMedGoogle ScholarCrossref 33.Macleod
JSmith
G Davey Psychosocial factors and public health: a suitable case for treatment?
J Epidemiol Community Health 2003;57565- 570
PubMedGoogle ScholarCrossref 34.van den Hombergh
CESchouten
EGvan Staveren
WAvan Amelsvoort
LGKok
FJ Physical activities of noninstitutionalized Dutch elderly and characteristicsof inactive elderly.
Med Sci Sports Exerc 1995;27334- 339
PubMedGoogle ScholarCrossref 35.Caris-Verhallen
WMde Gruijter
IMKerkstra
ABensing
JM Factors related to nurse communication with elderly people.
J Adv Nurs 1999;301106- 1117
PubMedGoogle ScholarCrossref 36.Chin
APaw
MJde Jong
NSchouten
EGvan Staveren
WAKok
FJ Physical exercise or micronutrient supplementation for the wellbeingof the frail elderly? a randomised controlled trial.
Br J Sports Med 2002;36126- 131
PubMedGoogle ScholarCrossref 37.Mattsson
NBWilliams
HVRickels
KLipman
RSUhlenhuth
EH Dimensions of symptom distress in anxious neurotic outpatients.
Psychopharmacol Bull 1969;519- 32
PubMedGoogle Scholar 38.Voorrips
LERavelli
ACDongelmans
PCDeurenberg
PVan Staveren
WA A physical activity questionnaire for the elderly.
Med Sci Sports Exerc 1991;23974- 979
PubMedGoogle ScholarCrossref 39.Van den Hombergh
CEJ Physical Activity and Health Characteristics: A SurveyAmong Dutch Elderly Women and Men [dissertation] Wageningen, the Netherlands Wageningen University1995;
40.van Staveren
WABurema
JLivingstone
MBvan den Broek
TKaaks
R Evaluation of the dietary history method used in the SENECA Study.
Eur J Clin Nutr 1996;50
((suppl 2))
S47- S55
PubMedGoogle Scholar 41.Fresh Foods AGB,
Beschrijvend rapport inzake opzet en uitvoering vande voedselconsumptiepeiling [Food Consumption Survey]. Dongen, the Netherlands Fresh Foods AGB1993;
Google Scholar 42. NEVO [database online]. The Hague Netherlands Nutrition Centre1990; 2001
43.Hoekstra
TGeleijnse
JMSchouten
EGKluft
C Diurnal variation in PAI-1 activity predominantly confined to the 4G-alleleof the PAI-1 gene.
Thromb Haemost 2002;88794- 798
PubMedGoogle Scholar 44.Field
AP Discovering Statistics Using SPSS for Windows. London, England SAGE Publications2000;
47.Scheier
MFWeintraub
JKCarver
CS Coping with stress: divergent strategies of optimists and pessimists.
J Pers Soc Psychol 1986;511257- 1264
PubMedGoogle ScholarCrossref 48.Musselman
DLEvans
DLNemeroff
CB The relationship of depression to cardiovascular disease: epidemiology,biology, and treatment.
Arch Gen Psychiatry 1998;55580- 592
PubMedGoogle ScholarCrossref 49.Patel
CMarmot
MGTerry
DJCarruthers
MHunt
BPatel
M Trial of relaxation in reducing coronary risk: four year follow up.
BMJ 1985;2901103- 1106
PubMedGoogle ScholarCrossref