| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Wolfson Unit for Prevention of Peripheral Vascular Diseases (M.C.W., F.G.R.F.), Public Health Sciences, and Department of Psychology (M.C.W., I.J.D.), University of Edinburgh, Edinburgh, Scotland.
Address reprint requests to: Ian J. Deary, FRCPE, Department of Psychology, University of Edinburgh, 7 George Square, Edinburgh EH8 9JZ, Scotland. Email: Ian.Deary{at}ed.ac.uk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
METHODS: In the Edinburgh Artery Study, 1592 men and women were randomly sampled from the general population, and their ABPI was measured at baseline and at the end of a 5-year follow-up period. A low ABPI suggests the presence of peripheral arterial disease. The revised Bedford-Foulds Personality Deviance Scale was administered at baseline to assess submissiveness and hostility. Data on other baseline risk factors, including physiological and social factors, were also collected.
RESULTS: Change in ABPI over 5 years was negatively correlated with age in both men and women (men, r = -0.10; women, r = -0.25). In multiple linear regression models, smoking, alcohol consumption, and submissiveness together accounted for 2% of the variance in ABPI change in men; in women, only age was related to change, accounting for 6% of the variance. Well-fitting structural equation models in both sexes showed that age influenced baseline ABPI and change in ABPI; that smoking and social deprivation directly affected baseline ABPI; and that the effect of hostility, and some of the effect of social deprivation, was mediated by smoking.
CONCLUSIONS: Social and personality factors were associated directly with baseline ABPI levels and indirectly with progression of atherosclerosis. Structural equation models revealed that associations among personality, social factors, and atherosclerotic progression were complex, involving mediation through other variables.
Key Words: atherosclerotic progression • personality • psychosocial factors • anklebrachial pressure index • structural equation modeling
Abbreviations: ABPI = ankle brachial pressure index; ABPI-1 = baselineABPI; ABPI-2 = follow-up ABPI; CABPI = change in ABPI; EAS = Edinburgh Artery Study; IMT = intima-media thickness; NEO-FFI = Neurotism, Extraversion, Openness–Five FactorInventory; PAD = peripheral arterial disease; PDS =Bedford-Foulds Personality Scale; PDS-R = revised Bedford-FouldsPersonality Deviance Scale; SEM = structural equation modeling; STAXI = State-Trait Anxiety Inventory.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
Noninvasive techniques can be applied to a wider population that includes asymptomatic and "healthy" individuals. The ABPI is the ratio of arm systolic pressure to ankle systolic pressure and is used to assess the extent of both symptomatic and asymptomatic lower limb PAD. The ABPI is an objective, noninvasive measure, a good indicator of generalized atherosclerosis, and a predictor of nonfatal and fatal coronary or cerebral vascular events (16–19). Ultrasound scanning of the carotid arteries, another noninvasive technique, allows simple, direct assessment of subclinical vascular disease by measuring either the presence of atherosclerotic plaques and the percentage of stenosis in the carotid arteries or by measuring the IMT (20–22). Greater IMT relates to an increased risk of myocardial infarction and stroke (23). The IMT and ABPI are significantly associated with each other (21, 22).
There are few studies of psychosocial factors and these measures of atherosclerotic disease processes. Cross-sectional associations have been reported between hostility and lower limb PAD, with an 80% increase in risk observed in the highest hostility quartile in comparison to the lowest (24) and a 40% increase in risk of PAD with a 1-SD rise in hostility score (25). An increase (of 20–50%) in the progression of carotid atherosclerosis has been related to higher hostility (26), hopelessness (27), increased workplace demands (28–30), and mental stress (31). These studies can help clarify the extent to which psychosocial factors are related to the chronic progression of atherosclerosis as opposed to the incidence of acute cardiovascular events.
Well-established risk factors for atherosclerotic disease include increased age, raised serum cholesterol levels, hypertension, and smoking (32). Many health outcomes, including coronary heart disease, have been found to be more prevalent in socially deprived groups (33–38). It is therefore essential to measure all of these factors so that the strength of their independent and combined contributions to risk can be assessed concomitantly with personality measures.
In cross-sectional analyses in the EAS (39), higher hostility in men was associated with increased prevalence of symptomatic PAD (40). The adjusted odds ratio for a 1-SD increase in hostile acts score was 1.41 (95% CI, 1.01–1.96); ie, there was a 41% increase in the risk of a man having intermittent claudication with each standard deviation rise in his hostile acts score (mean, 14.3; SD, 2.8).
Therefore, in the EAS, we hypothesized that there would be longitudinal associations between baseline hostility (and physical risk factors) and the progression of PAD. Measurement of ABPI is objective and is not biased by symptom reporting; thus, the ABPI is a useful outcome measure in studies of psychological factors and objectively assessed disease. Both personality and ABPI were assessed in subjects at their baseline medical examinations in 1988. The participants were followed up and their ABPI was reassessed after 5 years. The aim of the present study was to examine the relationships among physical and psychosocial factors measured at baseline and the change in ABPI over 5 years. This allowed us to explore the contributions of these risk factors to the severity and/or progression of objectively measured atherosclerosis.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
Baseline Examination
Baseline medical examinations were conducted between August 1987 and September 1988. At the examination, subjects underwent a series of clinical procedures and completed a self-administered questionnaire.
Ankle brachial pressure index.
An ABPI of <0.9 is up to 95% sensitive in detecting disease that would be detectable on an angiogram, and a ratio of 
0.9 is nearly 100% specific in identifying healthy subjects (16, 17). To obtain the ABPI reading, systolic and diastolic (phase V) blood pressures were measured in the right arm using a Hawksley random zero sphygmomanometer. Systolic ankle pressures were then measured, first in the right leg and then in the left, using the random zero sphygmomanometer and a Sonicaid Doppler probe. The lower of the two measures, regardless of whether it was taken in the right or left leg, was used in all analyses. Variability of the ABPI is similar to that of arm systolic pressure (16), and the test-retest reliability of the ABPI in different observers showed that observer bias was minimal (17).
Questionnaire.
The self-administered questionnaire comprised items on medical history, symptoms, smoking habits, alcohol consumption, diet, and personal characteristics, such as occupation and marital status. Medical history included recall of physician-diagnosed angina or myocardial infarction. Chest pain and leg pain symptoms were assessed using the World Health Organization angina and intermittent claudication questionnaires (41). Smoking habits were self-reported, and the information for each subject was converted into pack-years by calculating the number of 20-cigarette packs smoked per day and then multiplying by the number of years as a smoker. Participants were also asked to record their alcohol intake over the past week. One unit of alcohol was defined as half a pint of beer, one glass of wine, or a single measure of spirits. Social class categories (I-V) were assigned using the Registrar General’s classification (42). A Carstairs deprivation index score, based on the postal code district classification from the 1981 census (43), was also assigned. Deprivation scores are calculated using the proportions of people in an area who 1) live in overcrowded conditions, 2) are unemployed (men only), 3) are in social class IV or V, or 4) have no car (43). Scores at either end of the distribution indicate areas that are more homogenous (ie, either very deprived or very affluent). Using an independent rating system, areas ranked "worst" or "best" are consistently identified as very deprived or very affluent by the Carstairs scores. The scores reflect the prevailing social and economic conditions of different areas in which people live, regardless of their social class.
Personality measurement.
The PDS (44) was administered as part of the whole questionnaire. The scales were developed originally to elicit three personality traits: extrapunitiveness, by measurement on two primary scales, hostile thoughts and denigratory attitude toward others; intropunitiveness, by measuring lack of self-confidence and dependence on others; and dominance, by measuring hostile acts and a domineering attitude. The scales comprise 36 items, with 6 questions on each of six primary scales. Each question has four possible responses: very often (scoring 4), often (3), seldom (2), and never (1) (these are reverse-scored when appropriate). The range of possible scores for each primary scale is therefore 6 to 24, and that for each main scale is 12 to 48. The instructions specifically asked participants to "choose the one which best describes you for most of your life " (emphasis in original). This helped to ensure that the scales represented trait rather than state measures. The PDS manual (44) and Deary et al. (45) reported information about concurrent validity of the scales with the 16 personality factors (16PF).
Two revised scales, hostility and submissiveness/low self-confidence, were later derived from item-level factor analyses of the PDS (45) (see Appendix). These revised scales (PDS-R) are orthogonal. PDS-R hostility is calculated from items that assess hostile acts and hostile thoughts and is based on eight questions, allowing a score range of 8 to 32. PDS-R submissiveness/low self-confidence (for brevity, referred to as submissiveness) is based on nine questions drawn from a combination of the domineering attitude and lack of self-confidence scales; the range of scores is therefore 9 to 36.
The two new orthogonal scales were compared (46) to the NEO-FFI (47) and the STAXI (48). PDS-R hostility correlated negatively with NEO-FFI agreeableness (-0.38) and weakly but positively with neuroticism (0.15). PDS-R submissiveness was positively correlated with NEO-FFI neuroticism (0.43) and negatively correlated with extraversion (-0.35) and conscientiousness (-0.26). Correlations between the PDS-R and STAXI scales showed that PDS-R hostility was positively related to anger-out (0.47), anger expression (0.43), angry temperament (0.48), and total anger (0.47) and negatively related to anger control (-0.29). PDR-R submissiveness had a weak positive relation to anger-in (0.23) but was not correlated with any of the other anger scales.
Therefore, the PDS-R showed acceptable concurrent validity with other widely used personality scales. High hostility correlated with low agreeableness on the NEO-FFI and with high anger on the STAXI. Submissiveness showed little relation to STAXI anger but was positively correlated with NEO-FFI neuroticism and negatively correlated with NEO-FFI extraversion. The 1995 factor analysis was the first report of the item level factor structure of the PDS questionnaire, and the resulting solution on the two main scales was psychometrically valid (45). We used the two PDS-R scales for all analyses reported here.
Five-Year Follow-Up Examination
At the end of the 5-year follow-up period, participants were invited to a second medical examination. A full account of the follow-up protocol and invitation procedure was previously published (19). At the second examination, subjects were asked to complete a self-administered questionnaire to update their marital and employment status, medical history and symptoms, and smoking habits. Blood pressure in the arm and ankle were measured using the same procedures and equipment used at the baseline examination.
Data Analysis
Data on the questionnaires and recording forms were checked by the clinic staff, coded, and entered into the database. All data were entered twice, once each by two researchers, to control error rates. Discrepancies were corrected by referring to original records. The PDS responses were coded by the research staff, entered into the database, and then checked.
For all but the structural equation models, data were analyzed using SPSS for Windows (version 8.0). Descriptive statistics were calculated for the PDS (original and revised scales). All analyses were performed separately for men and women because of differences in the ABPI and most baseline risk factors. Smoking and alcohol distributions were skewed; therefore, both were adjusted using square-root transformations.
Pearson correlation coefficients were calculated among ABPI-1, ABPI-2, CABPI, and, as predictors, the PDS-R and baseline physical and social factors. CABPI was calculated by regressing the baseline measure on the follow-up measure and saving the residuals as the measure of change. This residual change estimate was included because the level of atherosclerosis at baseline, measured by the ABPI, may affect progression of disease and therefore the follow-up ABPI. Saving the residuals statistically adjusts for the baseline level of disease. Next, forward, stepwise multiple linear regression models were tested in both men and women. The dependent variables were CABPI with covariates of baseline age, social class, social deprivation, total serum cholesterol, smoking, and alcohol consumption.
Structural equation modeling.
Finally, SEM was performed to gain a better understanding of the multivariable associations among medical, psychological, and demographic variables and the change in the ABPI (progression of cardiovascular disease) over time. Multiple regression retains as predictors only those variables that make significant, independent contributions to the variance in the outcome variable (here, CABPI). Ultimately, the aim of research is to understand the complex network of associations among many correlated variables. SEM facilitates this because the technique combines factor analysis, path analysis, and multiple regression (49) in a single set of procedures that allow specification and testing of models of association among many variables. The approach is becoming more widely used in medical settings, where many factors influence disease (50, 51), and was the appropriate choice to investigate the multiple associations among risk factors and progression of disease assessed using the ABPI.
Model specification.
Bentler’s (52) EQS program was used to construct and test models. The variables included cardiovascular outcomes (ABPI-1 and APBI-2); cholesterol; demographic variables (age, social class, which was defined according to occupational status, and social deprivation); lifestyle and health behavior factors (smoking and alcohol intake); and personality factors. The data used for modeling are the variances and covariances among the variables of interest (Tables 2 and 3). The models achieve their power by being 1) economical, because they posit fewer parameters to account for the associations among variables than are found in the covariance matrix, and 2) able to test explicitly theoretical assumptions about the data. Because the present analyses are novel in applying SEM to cardiovascular and psychosocial data, there are relatively few guidelines for model building. Therefore, partly to achieve validation of models, data were first fitted in the men’s data and then the main assumptions of the model were tested in the women’s data. The following four constraints were imposed on the models: 1) Of necessity, ABPI-1 was assumed to be a cause of ABPI-2, because causation cannot proceed backward in time. 2) Variables were first assumed to influence only ABPI-1. If there was also a significant effect on ABPI-2, then this was evidence of an effect on ABPI change in addition to ABPI level at baseline. 3) Hostility was assumed to be mediated, at least in part, by lifestyle factors such as smoking (53). 4) Age, social class, and social deprivation were assumed to act only as antecedent variables.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
In the whole baseline sample of 1592, 121 (15%) men and 154 (19.6%) women had an ABPI <0.90. These prevalence and incidence rates reflect the earlier onset of vascular disease in men (ie, more men would have died before recruitment) and the higher rate of cardiovascular deaths that occurred in low-ABPI men during follow-up. A low ABPI was associated with the risk of both cardiovascular deaths and noncardiovascular deaths over the follow-up period (18).
There were statistically significant (p 
.05) differences between men and women across the whole baseline sample (N = 1592; Table 1). These differences were in 1) ABPI (mean ABPI was higher in men than in women); 2) age (men were slightly older, by 9 months, than women); 3) cholesterol (men had a lower mean value); 4) alcohol consumption (men reported higher consumption); 5) smoking (men had accumulated more pack-years than women had); 6) submissiveness (men had lower scores than women); and 7) social class (the main difference in distribution was that the percentage of men in the IIIN category was lower than the percentage of women; therefore, the percentage of men in all other categories was higher).
|
The current sample of 1080 (ie, those with follow-up ABPI readings) showed a very similar pattern of male-female differences (Table 1). The current sample, however, differed (in expected directions) from the whole sample on several baseline characteristics. For instance, those who survived to the 5-year follow-up had higher ABPI scores at baseline, were slightly younger, were less socially deprived, smoked less, and were concentrated in the higher social class categories. Neither hostility nor submissiveness mean scores were significantly different between the whole sample and the current sample.
Correlation Coefficients Among All Risk Factors
The second step of the analysis was to examine the correlation coefficients among the change in ABPI over 5 years (using the saved residuals) and the physical and psychosocial variables. Correlations with both ABPI-1 and ABPI-2 were also obtained. Table 2 shows that, in men, ABPI-1 was correlated significantly, and most strongly, with age (-0.18), smoking (-0.26), social class (-0.17), deprivation (-0.13), and hostility (-0.10). The directions of the associations indicate that a lower (worse) ABPI was associated with higher age, increased smoking, lower social class, and higher hostility. Lower ABPI-2 values in men correlated most strongly with higher smoking (-0.17) and increased age (-0.14). The worsening of ABPI over time (CABPI) in men was associated with increased age (-0.10) and lower alcohol consumption (0.10).
In women (Table 3), a lower ABPI-1 was most strongly correlated with higher cholesterol (-0.11), increased smoking (-0.22), and increased hostility (-0.11). Lower values of ABPI-2 correlated with older age (-0.26) and higher cholesterol levels (-0.14); the same was true for a worsening in ABPI (CABPI), which was related to older age (-0.25) and a higher cholesterol level (-0.11).
|
|
Men.
The model in Figure 1 is the best-fitting model for the 499 men with complete data. Some conventions should be described. Single-headed arrows represent putative causal influences. Double-headed arrows represent correlations without the implication of causation. The values placed on or adjacent to the arrows are the parameter weights estimated by the model fitting procedures; they may be squared to determine the percentage of variance shared by two adjacent variables. The adequacy with which the models fit the data are assessed by a number of indices. Note that the number of paths among measured variables in Figure 1 is less than the number of correlations or covariances in the original matrix. This difference is the degrees of freedom in the model, and the fewer paths there are, the more economical the model is. Obviously, if all possible paths are allowed, all of the covariance in the matrix will have been accounted for, but this will be uninformative. The model in the figure has 22 df. The average of the off-diagonal absolute standardized residuals is 0.029. Typically, values of 0.04 are considered good. The 
2 value for the model is 29.09 (df = 22, p = .14). This indicates that the residual covariance is not significantly greater than zero, another indication of good model fit. All parameter weights in the model are significant. The model’s overall fit to the data may be gauged by three indices that take values between 0 and 1, with good fit indicated by values 0.9 in all cases. In the present case, the values were as follows: Bentler-Bonnett Normed Fit Index, 0.931; Bentler-Bonnett Nonnormed Fit Index, 0.970; and Comparative Fit Index, 0.982. All of these values are very good. Finally, the Wald test is used to indicate any paths in the model that might be removed without significantly reducing the fit of the model to the data, and the Lagrange Multiplier test indicates paths that might be included to improve the model. In neither case was the model altered from that shown in Figure 1. In summary, the model has comprehensively good indices of fit.
|
Women.
The model that fitted well to the data gathered from men was fitted to the data gathered from 480 women. The Wald and Lagrange multiplier tests were used to indicate paths that might be changed to achieve a better-fitting model, and this is shown in Figure 2. The average of the off-diagonal absolute standardized residuals of the model is 0.017. The 2 value is 15.54 (df = 19, p = .69). The Bentler-Bonnett Normed Fit Index is 0.962; the Bentler-Bonnett Nonnormed Fit Index, 1.02; and the Comparative Fit Index, 1.00. The Wald and Lagrange multiplier tests indicated that no parameters could be removed or added to improve the model. Thus, the model fits the data very well.
|
Multisample SEM
The validity of the structural equation models tested here rests on three factors: the considerations underlying their construction, the comprehensive goodness-of-fit indices, and the fact that very similar models fit men and women separately. A further, even more rigorous, step was taken to evaluate model validity. The 10 parameter values were tested for equality in the models for men and women. Because the sample sizes were large, this test was sensitive to small differences in parameter weights across the two samples, providing a stringent test of model equivalence between men and women. Eight of the 10 parameter weights were equal (Table 5), with p .05 used as the cutoff value. There was a marginally significant difference in the age–ABPI-1 association between the sexes, possibly because of the mediating effect of cholesterol in one sex and not the other. The other difference was in the age–alcohol association, which was significantly larger in men than in women. The other eight parameters did not differ significantly across the sexes, adding considerably to the validity of the models.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
Therefore, in this sample, only age or health behaviors (smoking and alcohol consumption) had independent effects on objectively assessed levels of incident arterial disease. However, these statistical techniques may be overly simplistic given our true aims, which were to understand how all of the factors associate and to determine the point at which personality influenced health behaviors. There are many ways in which a factor may affect the disease process, and only one way is direct. With multiple linear regression, the "chains of causation and the different levels at which factors operate are often ignored" (54). SEM allowed further exploration of these relationships by examining the effects of mediating variables. In previous analyses of the EAS group, smoking and alcohol consumption were associated with hostility, social class, and social deprivation in both men and women (53). Those findings, although cross-sectional, suggested that relationships between hostility and disease might have been mediated by smoking and alcohol consumption. The structural equation models of the current analysis showed that this does seem to be the case.
In the structural equation model for men, only age affected both ABPI-1 and ABPI-2, indicating that it was acting directly on the actual change over time. However, smoking and alcohol consumption, which seemed to be independent predictors in the multiple regression models, were shown to have their impact through ABPI-1, which in turn was related to ABPI-2. Hostility, not appearing as an independent predictor in multiple linear regression, was shown in the structural equation model to be mediated by smoking and social class, precisely what had been suggested by the previous analysis of data from the EAS group (53). In women, too, age was directly related to the change in ABPI, as was found in the multiple linear regression models, but SEM showed the combined effects of hostility, social class, social deprivation, and smoking on ABPI-1 and, indirectly, on ABPI-2. In addition, the SEM technique provides theoretical models that can be tested explicitly by other researchers using data from other study populations. The models, therefore, can be validated or refuted by other researchers. Even a well-fitting model is not the definitive answer; alternatives can always be tested.
The substantial effect of social class on health is well documented (33, 34). Deprivation in the current study emerged as a predictor of the severity of atherosclerosis. Brunner (55) noted that a person’s experiences throughout life, especially deprivation, influence many biological variables that are particularly important for coronary disease, including hormonal stress mechanisms. Deprivation influenced both disease and personality, and both in turn were related to ABPI-1 and, indirectly, to ABPI-2. This may be one of the reasons why personality in multiple linear regression models did not emerge as an independent predictor of change in ABPI.
The problem of disease-based spectrum bias (11) may apply here. In the EAS, Leng et al. (19) found that asymptomatic PAD, defined using the ABPI and a reactive hyperemia test, was a significant predictor of both cardiovascular deaths (RR, 2.19; 95% CI, 1.33–3.59) and total death (RR, 2.44; 95% CI, 1.59–3.74). In the current analysis, those that died did not have a follow-up measure of ABPI, and the results reflect the change in ABPI in those who were less severely diseased. This, however, would tend to attenuate associations between risk factors and disease.
Use of objective measures of disease, such as the ABPI, has advantages over use of more subjective techniques (56). In some studies of cardiovascular disease, angina has been defined as the presence or absence of chest pain as reported by the subject (57). This may result in confounding of signs of disease with symptoms of disease. Bias due to symptom reporting should, however, be minimal if the outcome is subclinical (such as ABPI), provided the sample was randomly selected (56). Associations reported here were with objectively assessed, subclinical disease. The ABPI alone was used as the outcome; there were no categories of disease based on symptoms.
Studies analyzing the links between objective measures of atherosclerosis and psychological variables allow distinction between associations of personality and chronic vs. acute manifestations of disease. The Kuopio Ischemic Heart Disease Study of carotid atherosclerotic progression showed accelerated atherosclerosis to be associated with a range of personality variables, including hopelessness, hostility, workplace demands, cardiovascular reactivity, and mental stress (26–29). Measurement of the ABPI, like ultrasound scanning of the carotid arteries, is an accurate way to measure objectively the level of cardiovascular disease (22), so these end points will become increasingly useful if we are to advance our understanding of how personality contributes to the mechanisms of atherosclerotic progression. At the same time, it is vital not to underestimate the full impact of social class or social deprivation on both biological and personality-related risk factors for cardiovascular disease. In this capacity, SEM is a very useful technique.
The contribution of the present study is the description of social and personality predictors of disease progression in a longitudinal study with a near-representative sample of the age-relevant population. The sizes of the effects are concordant with the contributions of personality factors to disease status found in other studies (3). It is now accepted that cardiovascular disease processes have multifactorial influences and that progress is made by discovering each predictor that accounts for variance in the disease outcome and its associations with other predictors. In addition to biological factors, social and psychological variables have an impact, and future research can more clearly define their causal interactions.
|
APPENDIX |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
Hostility
I would have liked to get my own back on someone.
When I have wanted to have a row with someone, I have done so.
I have felt like telling people to go to blazes.
When I’ve disliked someone, I have shown it.
I have felt the urge to smash things.
I have wanted to give someone a piece of my mind.
I would have liked to pick a quarrel with someone.
When I’ve thought I was justified in losing my temper, I have done so in no uncertain terms.
Submissiveness/Low Self-Confidence
I have felt as capable as other people.
When in a group, I have been quite content to be led.
I have had confidence in myself.
I have been content to be dominated by someone else.
I have been very unsure of myself.
I have given up doing something because I thought too little of my own ability.
I have been happy to play second fiddle.
I have preferred to stay in the background.
I have felt pretty useless.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
Received for publication February 18, 1999.
Revision received February 24, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  A. E. Aquarius, K. G. Smolderen, J. F. Hamming, J. De Vries, P. W. Vriens, and J. Denollet Type D Personality and Mortality in Peripheral Arterial Disease: A Pilot Study Arch Surg, August 1, 2009; 144(8): 728 - 733. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. A. Shipley, A. Weiss, G. Der, M. D. Taylor, and I. J. Deary Neuroticism, Extraversion, and Mortality in the UK Health and Lifestyle Survey: A 21-Year Prospective Cohort Study Psychosom Med, November 1, 2007; 69(9): 923 - 931. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. E. Aquarius, J. Denollet, J. F. Hamming, D. P. Van Berge Henegouwen, and J. De Vries Type-D Personality and Ankle Brachial Index as Predictors of Impaired Quality of Life and Depressive Symptoms in Peripheral Arterial Disease Arch Surg, July 1, 2007; 142(7): 662 - 667. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. J. Lavie and R. V. Milani Adverse psychological and coronary risk profiles in young patients with coronary artery disease and benefits of formal cardiac rehabilitation. Arch Intern Med, September 25, 2006; 166(17): 1878 - 1883. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. E. Aquarius, J. De Vries, D. P. V. B. Henegouwen, and J. F. Hamming Clinical indicators and psychosocial aspects in peripheral arterial disease. Arch Surg, February 1, 2006; 141(2): 161 - 166. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. J. Lavie and R. V. Milani Prevalence of Hostility in Young Coronary Artery Disease Patients and Effects of Cardiac Rehabilitation and Exercise Training Mayo Clin. Proc., March 1, 2005; 80(3): 335 - 342. [Abstract] [PDF]
|
|
|
|
 |
|
 C. L. Leibson, J. E. Ransom, W. Olson, B. R. Zimmerman, W. M. O'Fallon, and P. J. Palumbo Peripheral Arterial Disease, Diabetes, and Mortality Diabetes Care, December 1, 2004; 27(12): 2843 - 2849. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. W. Lee and G. Y. H. Lip Effects of Lifestyle on Hemostasis, Fibrinolysis, and Platelet Reactivity: A Systematic Review Arch Intern Med, October 27, 2003; 163(19): 2368 - 2392. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
