| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Epidemiology and Public Health, International Centre for Health and Society, University College, London, UK (A.N., M.M.); and the Department of Epidemiology, Biostatistics and Occupational Health, McGill University, Quebec, Canada (R.F.).
Address correspondence and reprint requests to Amanda Nicholson, Department of Epidemiology & Public Health, University College London, 1-19 Torrington Place, London WC1E 6BT. E-mail: amanda.nicholson{at}ucl.ac.uk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSAPPENDIX.NOTESREFERENCES |
|---|
Method: 5449 men in an occupational cohort (79% of the total), with at least two prior measurements of the General Health Questionnaire (GHQ-30), were followed for CHD events (including CHD death, nonfatal myocardial infarction (MI), and angina) for (mean) 6.8 years. Psychological distress was measured using the GHQ-30, and general/anxiety, depression and sleep subscales were created based on a principal components analysis.
Results: Psychological distress increased the risk of CHD events, with the risk highest in men with recent onset of distress. Age-adjusted hazard ratios were 1.48 (1.03–2.13) for persistent and 1.77 (1.13–2.78) for new distress. Angina events accounted for much of the observed associations. This increased risk was independent of conventional CHD risk factors, markers of underlying CHD, or measures of reporting bias, and it was related to anxiety items and sleep disturbance rather than depressive symptoms.
Conclusions: Psychological distress increases the risk of a future diagnosis of angina in men. This risk is not accounted for by the presence of underlying CHD. These results highlight the importance of identifying both the role of underlying atherosclerosis in the pathway linking distress to heart disease and the timing of action of the components of psychological distress.
Key Words: psychological distress • CHD • angina • persistence • acceleration • atherogenic
Abbreviations: CHD = coronary heart disease; GHQ-30 = 30-item General Health Questionnaire; MI = myocardial infarction; BMI = body mass index; ECG = electrocardiogram; MC = Minnesota code.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSAPPENDIX.NOTESREFERENCES |
|---|
First is an atherogenic pathway with distress acting high up the causal pathway and leading to the development of atherosclerotic plaques. Several studies, showing long-term effects of depression persisting for up to 30 years, support this model (3,4). The literature consistently reports that the associations are not affected by adjusting for conventional CHD risk factors, suggesting the distress is not working via these mediators to cause atheroma.
A second alternative model involves emotional distress/depression operating to accelerate the progression of underlying atherosclerosis or to precipitate an event in those with underlying disease, rather than to promote atherogenesis (5). Few studies have addressed this directly. Two reports have published results for depressive features stratified by baseline disease. Aromaa indicated that the effect of depression was greater in those with baseline disease (6), whereas Penninx reported similar effect sizes (7). Studies on phobic anxiety predicting sudden death have demonstrated a short-term effect, and this may be interpreted as indicating an effect only in those with underlying CHD (8–10), but the question has not been examined by stratifying the population at risk by evidence of ischemia. This acceleration model requires further exploration.
A third explanation for the association between distress/ depression and heart disease suggests that the observed relationship between distress and CHD events is attributable to the presence of undetected cardiac disease leading to distress (i.e., an example of reversal causality). Several reports have found that only recent or increased symptoms are predictive of future CHD (11–13) or all-cause mortality (14), raising the possibility that psychological symptoms have arisen because of undiagnosed CHD, and hence causality may be reversed. There is a need for further work examining the role of persistence of distress in the etiology of CHD (15).
The existing literature on depression and CHD has not examined these mechanisms systematically. The duration of effect, the role of underlying atherosclerosis in the associations, and the persistence of depressive symptoms required for an adverse effect all provide insight into the potential mechanisms but need to be considered together. For instance, an adverse effect restricted to new symptoms, which declines with time and which is larger in participants with evidence of underlying disease would strongly support reverse causality.
The contribution of personality factors and other negative emotions, such as anxiety, to the observed associations between CHD and depression is also unresolved. The literature on anxiety as a risk factor for CHD is smaller than that on depression but suggests that anxiety acts over a shorter time span than depressive symptoms, with an acceleration role (8–10). Given that anxiety and depression often occur together and are difficult to separate, it is possible that any observed associations represent the combined effect of different psychosocial influences acting at different points along the causal pathway. We are not aware of studies that have examined the impact of the course of both anxiety and depression over time on the etiology of CHD.
This paper uses the General Health Questionnaire-30 (GHQ-30) as an exposure measure. The GHQ-30 has been shown to predict all-cause mortality (16) and to influence prognosis after a cardiac event (17), but there has been limited work examining its role as a predictor of CHD events (18–20). Factor analyses can be used to examine different dimensions of distress within the GHQ-30. This paper extends the current literature by examining both the role of persistence of distress in the association with future CHD events and the contribution that undetected CHD at baseline makes to the observed associations. In addition, the contribution of separate components of psychological distress to future CHD events and the effect of their time course is examined. Results will be discussed in the context of the three mechanistic models outlined above, and potential analytic strategies for future research suggested.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSAPPENDIX.NOTESREFERENCES |
|---|
For this investigation, analyses have been restricted to men. There were a small number of CHD events in women, and exploratory analyses indicated that there was nonproportionality in the risk associated with psychological distress in women. Hence it was considered inappropriate to amalgamate the sexes. Because persistence of psychological distress is the major focus of this paper, only men with repeated measures of the GHQ-30 were included in analyses, and Phase 3 was used as the baseline.
Measurement of Psychological Distress
The GHQ-30 was included in questionnaires at Phases 1, 2, and 3. The GHQ-30 is a well-respected and widely used measure of psychiatric ill-health. It was designed as a self-administered questionnaire to detect undifferentiated emotional distress in community populations rather than psychiatric patients. It acts as a screening instrument for minor psychiatric disorder, especially anxiety and depressive illnesses. Questions ask whether a range of symptoms have recently been worse or better than usual. The 30-item version of the questionnaire was prepared from the full 60 items using the best discriminators for psychiatric caseness and removing somatic items (22,23). The original criterion validity of the GHQ-30 as an identifier of psychiatric caseness has been repeated many times with a reported median sensitivity of 81% and a specificity of 80% (24). The factor structure of the GHQ-30 has been extensively investigated (24–28). The general factor explains the bulk of the variance and contains many anxiety items, but a separate depression factor has been a consistent finding. Other factors that have been reported in the GHQ-30 relate to competence, social functioning, and sleep disturbance.
In study questionnaires, the four level responses for each question were arranged so that for all questions, the later options represented adverse responses. The responses were coded as either GHQ scores (0-0-1-1), with a response of worse or much worse than usual scoring 1, or as modified Likert scores (1-2-3-4). When scores were totaled over the entire 30 items during recoding, one missing answer was allowed, and the observed total was then multiplied by 1.035 (30/29). In accordance with recommendations, a validation study was performed within Whitehall II (29). On the basis of these results, participants scoring 
5 on the summed GHQ score were classified as cases and defined as having psychological distress. This threshold was used at all Phases.
GHQ Over Time: Definition of Exposure Groups
Participants were divided into groups according to GHQ status at Phases 1 to 3. These analyses were restricted to participants who had completed the GHQ-30 at Phase 1 and 3, with or without Phase 2. The groups compared at Phase 3 are as shown in Table 1, and these groups are summarized in Table 2.
|
|
GHQ Subscales
To asses the effect of different component of psychological distress on CHD risk, subscales were created based on a previous principal components analysis with varimax rotation of Phase 1 questionnaires (see Appendix: full details available from author). Five factors with an eigenvalue >1 were identified as follows: general (explaining 17.9% of the variance), competence (12.6%), depression (11.8%), social interaction (7.3%), and sleep disturbance (5.3%). Initial analyses using factor scores found that the sleep and general factors were associated with CHD events, but a depression subscale was also used in this paper, given the potential importance of depression in CHD etiology. Subscales were created at each Phase by summing the Likert scores of items that had a factor loading of 0.50 on each factor. The resulting scale was divided at an appropriate quartile, and exposure groups over time were created in the same way as for the whole GHQ scale. These subscales are summarized in Table 3.
|
CHD Events
These analyses use a combined endpoint of CHD death, nonfatal MI, and angina. CHD death: 10,300 (99.9%) of the original participants were flagged at the National Health Service Central Registry, who notified us of the date and cause of death. Participants were classified as having a coronary death if the underlying cause had an ICD-9 code 410 to 414. Nonfatal MIs and angina events were ascertained by questionnaire items on chest pain (WHO Rose questionnaire (30)); recall of a doctor’s diagnosis; investigation (exercise electrocardiography, stress testing, or angiography); and treatment (nitrates or revascularization); or from the study twelve lead resting electrocardiograms (Simmons Mingorec) classified according to the Minnesota code. When potential cases had been identified from these sources, details of physician diagnoses were sought and classification of events was carried out blind to other study data independently by two trained coders, with adjudication by a third in the (rare) event of disagreement. CHD death, MI, and angina cases which met diagnostic criteria for definite, probable, or possible events were included in these analyses.
Death (based on Simon (31)) Definite required death certificate plus post-mortem reports, probable required death certificate plus evidence in clinical notes, and possible required death certificate or clinical notes only. Study data did not contribute to the death classification.
MI (based on MONICA criteria (32)) Definite required a definite ECG or hospital records with two out of three of the following: symptoms, ischemic ECG, and enzyme changes. Probable required record of diagnosis in clinical notes. Possible required self-reported MI on study questionnaire, with or without abnormal study ECG.
Angina Definite angina required the combination of typical symptoms and an abnormal test result; probable was assigned in the absence of a test but with a record of diagnosis in clinical notes; possible angina was defined by self-report only with no test or record of diagnosis in clinical notes. In these analyses, Rose questionnaire angina alone was not included in the possible category.
CHD death/nonfatal MI and angina endpoints were divided into two separate groups for some analyses. Groups were hierarchical so that men with both MI and angina codes were classed as MI. In analyses with separate endpoints, men with the other endpoint were excluded. For CHD death/MI analyses, the event date was amended if a participant had an angina event before MI or death. For some analyses the possible cases of MI or angina were removed. This led to some reclassification of endpoint as 13 men with possible MI had definite/probable angina.
CHD events were collected until the end of 1999 and in this paper have been analyzed from the Phase 3 baseline (with mean 6.8 years follow-up), excluding all those with CHD events before Phase 3. Five thousand four hundred and forty-nine men (79% of 6895) were included in the distress exposure groups. There were 248 (197 definite/probable) CHD events, comprised of 122 CHD death/nonfatal MI (90 definite/probable) and 126 angina cases (107 definite/probable) during follow-up.
Markers of Underlying Disease
Men with validated CHD events at baseline were excluded, but to estimate undiagnosed CHD at baseline, the presence of either a positive Rose questionnaire (30) or an ischemic ECG at baseline was considered as a marker of underlying CHD. The presence of any of the following Minnesota codes on the computer ECG coding was taken as evidence of ischemia: diagnostic Q waves (MC all 1-1, 1-2-1 to 1-2-5 and 1-2-7); equivocal Q waves (MC 1-2-8 and all 1-3); ST segment depression (MC 4-1 to 4-3); T wave inversion (MC 5-1 to 5-3); or left bundle branch block (MC 7-1-1). All computer abnormals and a sample of computer normals were reviewed by an experienced manual Minnesota coder to confirm the computer code and to exclude technically poor readings. The ECGs were classified by a combination of computer and manual coding. Manual MCs were assigned if there was a computer code for ischemia.
Other Variables
Potentially confounding covariates included health behaviors such as smoking, classified as never ex- and current smoking, plus number of manufactured cigarettes per day for current smokers; hours of vigorous exercise per week; alcohol consumption; systolic and diastolic blood pressure (average of two readings using Hawksley random-zero sphygmomanometer); cholesterol; and body mass-index. Social class was assessed by grade of employment in the civil service. The Somatosensory Amplification Scale, designed to measure symptom amplification, has also been shown to correlate with negative affectivity and distress (33). The Phase 3 questionnaire included 4 of the original 10 items that assessed how sensitive the participant was to loud noises, temperature, hunger, and pain, each rated on a 5 point scale. These were summed to form a symptom amplification score (range 4–20) that was normally distributed and used to adjust for reporting bias.
Statistical Analyses
Statistical analyses were performed using SAS version 6.12 and version 8.04 computer software (SAS, Cary, NC). Baseline characteristics of the distress groups were age adjusted by direct standardization using the whole population as standard, with age divided into four 5-year age groups. The CHD risk in each psychological distress group was assessed by survival analysis. Differences in the survival function, compared with the never group, were evaluated using the log-rank test. These differences were quantified and adjusted for age (using four 5-year age groups entered as categorical variables) and other covariates using Cox proportional hazards models. The "never distress" was taken as the reference group. The assumption of proportionality of hazard was checked by examining log minus log plots and by tests of the interaction of time with exposure (separately for each distress group). There was some minor nonproportionality for new distress in the first 18 months of follow-up that was not considered to preclude the use of Cox models. Otherwise assumptions for proportionality were met. Markers of underlying disease were included both as covariates and used as stratification variables. Interactions were tested by including a product term in the Cox models.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSAPPENDIX.NOTESREFERENCES |
|---|
|
Figure 1 shows survival curves (without a CHD event) for men in the different distress groups, and Table 5 shows age-adjusted hazard ratios for the distress groups from Cox models. New distress at Phase 3 had a hazard ratio of 1.77 (1.13–2.78) for future CHD events. Persistent distress in men at Phase 3 was also significantly associated with future CHD events, and former distress approached significance. When CHD events were divided into two groups, CHD death/nonfatal MI and angina, and analyzed separately the effect of distress was stronger for angina events, with hazard ratios for CHD death/MI not reaching significance. Removal of possible events led to a weakening of the associations with CHD death/MI but little change in the associations with angina.
|
|
Table 6 summarizes the effect of adjusting for the marker of baseline disease, CHD risk factors and somatic amplification on the associations. These analyses use the combined CHD endpoint. The marker of underlying CHD was more prevalent in men with persistent distress, and adjustment reduced the effect slightly. New distress remained strongly associated with future CHD events. When analyses were repeated within strata of men with (n = 5148) and without markers (n = 300), there was a suggestion that the effect of new distress was greater in men with the marker of underlying disease. The hazard ratios for persistent, new, and former distress were 1.53 (1.04–2.24), 1.57 (0.95–2.59) and 1.21 (0.87–1.67), respectively, in men without markers and 0.99 (0.32–3.05), 2.60 (0.91–7.46) and 1.75 (0.81–3.79) in men with markers of underlying disease. Interaction tests were not significant, perhaps because of small numbers in the group with markers. Adjustment for CHD risk factors and somatic amplification had minimal impact on the associations with new distress, although somatic amplification reduced the effect size of persistent distress.
|
The effect of different subscales of the GHQ on future CHD events is summarized in Table 7. The general and sleep scales were most strongly associated with CHD events, although no association was observed for the depression subscale. As with the whole score, effects were greater with angina than with CHD death/MI. There were few differences in effect size between the distress groups, with the persistent group at only slightly higher risk. Adjustment for the presence of a marker of baseline disease reduced the effect of persistent sleep disturbance slightly so that it was no longer significant, but other associations were largely unchanged.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSAPPENDIX.NOTESREFERENCES |
|---|
One possible explanation for the observed association is that men with distress may not be more likely to develop CHD than men without distress, but rather are more likely to either report symptoms or to get a diagnosis after presentation (35). Harder endpoints such as MI are less prone to this bias and so the fact that the effect of distress was weaker for CHD death/MI than for incident angina raises the question of whether detection bias may account for findings. Five studies have reported results for depression separately for angina and MI and two reported similar effect sizes in MI and angina (11,36), and three were only significant for angina (37–39). We know of no other studies comparing the effect of the GHQ on MI and angina, but Yasuda et al. (18) found no association between GHQ scores and CHD death, whereas Rasul and Robinson (19,20) did find an association with CHD death. We have examined the potential role of detection bias by removing self-report only cases and by adjusting for somatic amplification, but we cannot rule out detection bias as an explanation for the findings. An alternative explanation is that the natural history of angina is different from MI and that psychological distress contributes to the narrowing of the coronary artery and hence anginal symptoms but not to obstruction. There are surprisingly few data examining differences in risk factors for angina and MI (40–42), but there is some evidence that some risk factors may be stronger for MI than angina.
These analyses included only men who completed the second screening examination (79% of the total) and hence differential loss-to-follow-up could result in selection bias. Detailed analyses of the likely impact of these losses (43) found that loss-to-follow-up was not related to baseline GHQ status and that the relationship between distress at baseline and future CHD events was similar in those who did and did not attend later Phases of the study, indicating that selection bias is unlikely to have influenced these results.
In this paper the definition of the various exposure groups was designed to enable the examination of different effects of transient distress and more long-term distress on future CHD events to be examined. This approach is essential for research questions regarding chronicity and the time-span of effect to be addressed. The effect size for new distress was larger than that for former or persistent distress. Although confidence intervals overlap, this is consistent with other reports in the literature (11–13) that recent-onset symptoms are the strongest predictors of CHD. The pattern over time for subscales differed from that for whole GHQ score. For the subscales, the persistent group had largest effect size, with no further increased risk seen for men with new symptoms. This may reflect the fact the GHQ scoring was used for the whole score, but Likert scores were used for the subscales to make full use of the data. Early validation studies found that GHQ scoring was better at case identification, and hence the whole score is likely to be detecting more new cases (22).
Reverse causality is a potential explanation for the increased risk seen with recent –onset depression (11–13). Recent evidence suggests reverse causality may contribute to the observed effect of the GHQ (20). This paper does not support that model. New distress was not associated with underlying CHD markers, and the effect was not reduced by adjustment. It is possible that distress is acting as a more sensitive indicator of health than our marker, and hence the association might retain some residual confounding. Further adjustment for more subjective and more sensitive measures of health, such as self-rated health or longstanding illness, leads to issues of overadjustment because these instruments are acting to some extent as measures of distress.
Men were divided into those with and without markers of underlying disease, and any effect restricted only to men with markers might suggest an acceleration role whereas an effect observed in men without markers would support an atherogenic role. An increased risk associated with new distress was seen in men with markers, indicating that recent onset distress has a greater effect in the presence of underlying disease. The small size of the stratum means this is not conclusive evidence of an acceleration role. Previous work has suggested that anxiety may act as an accelerator. In this study the effect of the general/anxiety factor was demonstrated in the larger stratum of men without markers (n = 5123, effect sizes for persistent, new and former were 1.68 (1.15–2.45), 1.49 (0.90–2.46) and 1.17 (0.84–1.64), respectively), indicating that anxiety is not restricted to an acceleration role.
However, this paper was unable to investigate atherogenic or atherosclerosis models adequately. Apart from the limits of strata size, such stratification analyses can only have a limited role in examining these mechanistic issues, because the indicators used will not separate men completely into those with and those without underlying atherosclerosis (44). Hence, an effect of distress in men without markers may not mean that distress is acting as an atherogenetic factor. Detection bias may also be involved, since distress may have a greater effect in men with markers because of response to symptoms rather than natural history. Further work should aim to examine the effect of distress in men (and women) with documented levels of atherosclerosis to differentiate the atherogenic and acceleration models.
Factor analysis of the GHQ-30 was largely consistent with other reports (24). The depression factor was similar to that reported by other investigators and included the four items contained in the GHQ-30 from the seven-item depression scale of the GHQ-28 (45), encompassing the cognitive symptoms of depression and hopelessness. The first general factor is likely to be measuring anxiety, but may also include some of the affective symptoms of depression. A distinct sleep factor has been reported previously (46), but in other studies, the sleep questions are incorporated into the other factors.
Little work has been published on factors of GHQ-30 predicting disease endpoints, but in the Japanese study none of the factors from the GHQ-30 predicted mortality from heart disease in men, although depression and apathy within the GHQ-30 were related to all-cause and cerebrovascular mortality (18). Whittington and Huppert (47), who did not use a conventional factor analysis, found that failure to endorse positive items, such as finding it easy to get along with people or feeling that you were doing things well, was associated with all-cause mortality. Our study is consistent with that of Yasudain that no association was found with CHD mortality.
The lack of an association between a high score on the depression factor and increased CHD risk is an unexpected finding and contrary to many published reports (48–50) that have found that similar items to these were associated with future CHD. In this study, the general factor and sleep questions within the GHQ-30 were strong predictors of cardiac events. Previous studies have examined the role of somatic symptoms, such as appetite and sleep disturbance (5,50–52) within depression instruments and have found that these are not as important as the cognitive and affective symptoms so that our findings are contrary to expectations. Comparison between different questionnaires could account for these discrepancies. The depression factor of the GHQ-30 may be measuring only certain cognitive components of depressive illness, and the affective and biological aspects may be picked up by the general and sleep factors, hence explaining their associations with increased CHD.
The results suggest that anxiety may be an important contributor to the etiology of CHD but do not support an acceleration role for anxiety acting as a trigger for events in those with underlying disease. Effects were stronger in the persistent group, adjustment for the marker of underlying disease did not reduce effects substantially, and effects were seen in men with no marker present.
The finding that the sleep factor predicts CHD events is not entirely unexpected. Vital exhaustion has been identified as an independent risk factor for MI by Appels et al. (53,54). Vital exhaustion is characterized by fatigue rather than insomnia, but others have reported that insomnia is related to future CHD (55,56). The sleep questions within the GHQ focus on insomnia rather than tiredness and may be a measure of depression. The effect of sleep was strongest in men with persistent sleep problems, which may reflect general ill-health. Underlying CHD did not account for the findings but other health problems may contribute to the effect.
The biological pathway of action of distress remains uncertain. These results support other reports in finding that distress does not work through recognized CHD risk factors such as health behaviors, blood pressure, or cholesterol. It has been suggested that the metabolic syndrome (57), influenced by the hypothalamic pituitary axis, may be involved in the relationship between distress and CHD, and these potential mediators will be examined in a subsequent paper.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSAPPENDIX.NOTESREFERENCES |
|---|
We thank all participating civil service departments and welfare, personnel, and establishment officers; the Occupational Health and Safety Agency; the Council of Civil Service Unions; all participating civil servants in the Whitehall II study; and all members of the Whitehall II study team.
|
APPENDIX. |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSAPPENDIX.NOTESREFERENCES |
|---|
|
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSAPPENDIX.NOTESREFERENCES |
|---|
Received for publication March 25, 2004; revision received February 3, 2005.
DOI:10.1097/01.psy.0000171159.86446.9e
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSAPPENDIX.NOTESREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  T. Meyer, U. Buss, and C. Herrmann-Lingen Role of Cardiac Disease Severity in the Predictive Value of Anxiety for All-Cause Mortality Psychosom Med, January 1, 2010; 72(1): 9 - 15. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Bankier, J. Barajas, A. Martinez-Rumayor, and J. L. Januzzi Association Between Anxiety and C-Reactive Protein Levels in Stable Coronary Heart Disease Patients Psychosomatics, July 1, 2009; 50(4): 347 - 353. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Walters, G. Rait, I. Petersen, R. Williams, and I. Nazareth Panic disorder and risk of new onset coronary heart disease, acute myocardial infarction, and cardiac mortality: cohort study using the general practice research database Eur. Heart J., December 2, 2008; 29(24): 2981 - 2988. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. V. Igna, J. Julkunen, H. Vanhanen, P. Keskivaara, and M. Verkasalo Depressive Symptoms and Serum Lipid Fractions in Middle-Aged Men: Physiologic and Health Behavior Links Psychosom Med, November 1, 2008; 70(9): 960 - 966. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. P. Roy-Byrne, K. W. Davidson, R. C. Kessler, G. J.G. Asmundson, R. D. Goodwin, L. Kubzansky, R. B. Lydiard, M. J. Massie, W. Katon, S. K. Laden, et al. Anxiety Disorders and Comorbid Medical Illness Focus, October 1, 2008; 6(4): 467 - 485. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. M. Nezu, C. M. Nezu, D. Jain, M. S. Xanthopoulos, T. A. Cos, J. Friedman, and M. Lee Social Problem Solving and Noncardiac Chest Pain Psychosom Med, November 1, 2007; 69(9): 944 - 951. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Ahnquist, P. Fredlund, and S. P Wamala Is cumulative exposure to economic hardships more hazardous to women's health than men's? A 16-year follow-up study of the Swedish Survey of Living Conditions J Epidemiol Community Health, April 1, 2007; 61(4): 331 - 336. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Nicholson, H. Kuper, and H. Hemingway Depression as an aetiologic and prognostic factor in coronary heart disease: a meta-analysis of 6362 events among 146 538 participants in 54 observational studies Eur. Heart J., December 1, 2006; 27(23): 2763 - 2774. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
