| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Psychiatry (C.D., L.M., F.C.), Manchester University, Manchester; and Psychology Department (D.C.-C.), Staffordshire University, Staffordshire, United Kingdom.
Address reprint requests to: Dr. Chris Dickens, Senior Lecturer in Psychological Medicine, Department of Psychiatry, Rawnsley Building, Manchester Royal Infirmary, Oxford Road, Manchester M13 9WL, United Kingdom. Email: c.dickens{at}man.ac.uk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
METHODS: CD-ROM databases and bibliographies were searched to identify all studies comparing depression in patients with rheumatoid arthritis and control subjects using standardized assessments. Effect sizes (Pearson’s r) and probabilities were combined across studies. We examined the extent to which the association between rheumatoid arthritis and depression could be attributed to level of pain (using contrasts), sociodemographic differences between groups (combining methodologically restricted studies), and methods of assessing depression (examining heterogeneity across studies).
RESULTS: Twelve independent studies comparing depression in patients with rheumatoid arthritis with depression in healthy control subjects were found. Effect sizes for depression were small to moderate (r = .21, p < .0001; heterogeneous). This effect was not reduced in studies controlling for sociodemographic characteristics (r = .27, p < .0001). The effect sizes did vary in a linear manner in proportion to the effect size for pain (z = 2.67, p = .0064). The effect sizes produced by different measures of depression were heterogeneous (
2 for Fisher’s Z = 24.6, p = .0002), with the Hospital Anxiety and Depression Scale giving effect sizes most dissimilar to those of other measures.
CONCLUSIONS: Depression is more common in patients with rheumatoid arthritis than in healthy individuals. This difference is not due to sociodemographic differences between groups, but it may be attributable, in part, to the levels of pain experienced. Variation in the methods of assessing depression partly accounts for the differences among studies examining the levels of depression in patients with rheumatoid arthritis.
Key Words: pain, • depression, • rheumatoid arthritis, • meta-analysis, • systematic review.
Abbreviations: AIMS = Arthritis Impacts Measurement Scale;; BDI = Beck Depression Inventory;; CES-D = Center for Epidemiological Studies Depression Scale;; GDS = Geriatric Depression Scale;; GWB-D = General Well-Being Depression Scale;; HADS = Hospital Anxiety and Depression Scale;; IDD = Inventory to Diagnose Depression;; MMPI = Minnesota Multiphasic Personality Inventory;; PDI = Psychiatric Diagnostic Interview;; POMS = Profile of Mood States;; RA = rheumatoid arthritis.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
Cross-sectional studies have examined the degree of association between pain and depression in patients with RA. Such studies have shown that the degree of depression varies in proportion to the level of pain. Correlation coefficients between pain and depression of approximately 0.4 have been found (1, 5, 6). However, the magnitude of this association is weakened and can become nonsignificant once the effects of other variables, such as demographic factors and disability, are controlled (7). Few longitudinal studies of the relationship between changes in pain and depression have been performed. One longitudinal study found that changes in depression are associated with changes in pain (8), but others failed to confirm this association (9–11).
The variability in the findings of previous studies likely resulted from differences in the samples and measures used, because scores on some measures of depression may be more affected by the presence of pain than others (12, 13). Furthermore, depression is associated with other RA-associated factors, such as degree of physical disability (5, 7, 8, 14), disease activity (5), and duration of disease (7). Gender, levels of social stress experienced, support available, and perceived ability to cope with the illness are also associated with depression in this group (2, 5, 7). The extent to which these other covariates are controlled likely affected the apparent strength of the association between pain and depression.
To clarify the magnitude of the association between pain and depression in patients with RA, we performed a systematic review of the literature with meta-analysis. Specifically, the review aimed to answer the following questions: 1) Is there an association between RA and depression, and, if so, what is the magnitude of this relationship? 2) Does the magnitude of the relationship between RA and depression depend on how depression is defined and measured? 3) Does the relationship between RA and depression persist after proper adjustment for confounders? 4) Is the relationship between RA and depression related to the level of pain experienced? 5) Are other pain conditions associated with depression to the same extent as RA?
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
Identification of Relevant Articles
Electronic literature searches were performed on the Medline (1966 to present), Embase (1980 to present), Psychinfo (1984 to present), and Cinahl databases (1982 to present). Medical subject headings (MeSH) were used to identify all papers indexed as containing material relevant to "depression" and "pain." Text words were used to further identify studies that had included standardized measures of depression (by listing the full names of these measures and their accepted abbreviations). The last search of the CD-ROM databases was performed on April 20, 1999. Reference lists from identified papers and from reviews in the area were searched by hand. In addition, first authors of papers identified as containing data relevant to this review (30 in total) were asked for details of any other studies of which they were aware (published or unpublished) that might contain data useful for this review. Another 11 first authors were contacted for points of methodological clarification or additional data on studies that were reported in insufficient detail to enable their inclusion in the review.
In total, 108 studies were identified by these strategies. On detailed inspection, 30 of these studies were found to meet the criteria for inclusion and contained sufficient data to be included in the meta-analysis. For the purposes of this review, "sufficient data" meant that the study report provided the number of subjects in the patient and control groups and means and standard deviations for depression scores in both groups, thus enabling the calculation of an effect size. Studies were also included if they provided details of the size of each group along with the inferential statistic used for the comparison of depression between groups (eg, t value, F value, or 2 value) and degree of significance (p value). Least ideally, studies could be included if the sizes of patient and control groups and a measure of the degree of significance of the difference between groups with regard to the depression scores were known, even if the scores of the inferential statistic were not known. Effect sizes were calculated from inferential statistics and p values only if the authors failed to provide means and standard deviations for depression scores of each group when contacted.
Of the 78 studies excluded from this review, the most common reason for rejection (34 studies) was the failure to include a control group, which only became apparent when the full text of the study report was inspected. Other reasons for exclusion included failure to use a standardized measure of depression and failure to measure depression in control subjects as well as patients.
Among the 30 study reports, three pairs of studies used identical or highly related samples (Refs. 15 and 16, Refs. 17 and 18, and Refs. 19 and 20). To avoid studies with highly related samples introducing too great an influence on the results of the meta-analysis, the effect sizes and probabilities of highly related samples are averaged (unweighted). In this way highly related studies contributed only a single effect size to each meta-analysis. Thus, 27 independent studies were identified by the systematic review and included in the meta-analyses.
Data Extraction
Means and standard deviations of the depression scores for the RA patient and control groups were recorded using standardized data extraction sheets. If these were not reported, the inferential statistic (ie, t, F, or 2 value) comparing depression in the RA and control groups and the p value were recorded. Demographic details and measures of the severity of the underlying physical disorder were recorded.
To establish the reliability of the data extraction, 5 study reports were chosen at random from the 30 reports fulfilling inclusion criteria for this meta-analysis. Using the standardized data extraction sheets, a second researcher (L.M.), who was blind to the findings of the first researcher, recorded values for key variables. The key variables recorded were the number of patients with RA and control subjects, number of men and women, level of depression (mean and standard deviation) or number of depressed subjects, and levels of pain and disability if available. In total, 110 data points were checked by the second researcher. When the completed data sheets from both researchers were compared, only three data points had been recorded inaccurately. This indicated that the data extraction process had a high degree of reliability.
Meta-Analytic Techniques
Two main meta-analytic techniques were used: combination of effect sizes and probabilities. In addition, the effect sizes from individual studies were examined for heterogeneity.
Combination of effect size.
The effect size calculated for each study was Pearson’s r (21, 22). For the purposes of this review, positive effect sizes indicate that the value of a given variable is greater in the RA group than in the control group. Negative effect sizes indicate the opposite. In addition, effect sizes were classified according to the criteria of Cohen (23) as small (r = .1), medium (r = .3), or large (r = .5).
Combined effect size is calculated by transforming r to Fisher’s Z, taking a mean of Fisher’s Z across studies (weighting for study size), and then transforming back to obtain the combined effect size. Ninety-five percent confidence intervals were calculated for each combined effect size estimate. For each combined effect size, heterogeneity of the effect sizes combined was examined. This was achieved by testing whether any of the effect sizes differed significantly from the others (22). If the effect sizes were heterogeneous, the effect that was most different from the others was removed and the variability was recalculated. This process is repeated until a homogeneous group of effect sizes is identified. By identifying both the full set of studies and a homogeneous subset, where the full set is heterogeneous, the danger of one or a small minority of studies skewing the results and creating a false impression of the overall effect size is avoided.
Combining probabilities.
Combining probabilities was achieved by calculating the standard normal deviate (standard z) for each study and then averaging standard z values across studies (weighting for study size) before converting back to an overall p value (21, 22).
Assessing publication bias.
The robustness of the meta-analyses to publication bias was estimated using "fail-safe N" (21). This method calculates the number of negative studies required to be added to the meta-analysis to render the finding nonsignificant. In addition, Rosenthal (21) provides a formula to estimate the likely number of unpublished, nonsignificant studies, the so-called "file drawer n." Comparing fail-safe N with file drawer n provides a measure of the robustness of the meta-analytic findings against the effects of publication bias.
The above calculations were performed using Microsoft Excel 97.
Planned Meta-Analyses
Studies meeting minimal criteria.
From the studies meeting minimal inclusion criteria, effect sizes for depression were calculated between patients with RA and 1) healthy control subjects, 2) pain patient control subjects with single diagnoses (eg, fibromyalgia or osteoarthritis), and 3) other patient control subjects.
Methodologically restricted studies.
Additional meta-analyses were planned on studies controlling for the effects of possible confounding variables. Variables considered likely to affect the prevalence or severity of depression in the groups under study included demographic variables (age, sex, educational status, work status, and social support) and illness-related variables (severity of disease, duration of disease, and disability). In these meta-analyses, we decided a priori to include studies controlling for age and sex when the control group was healthy and studies controlling for age, sex, and disability if the control subjects were patients. Too few studies had measured illness-related variables, such as disability, in patients and control groups, to enable a meta-analysis examining the association between pain and depression while controlling for differences in disability.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
|
Methodologically restricted studies.
Of the 12 studies comparing patients with RA with healthy control subjects, 8 had matched subjects for age and sex. This was achieved either by recruiting matched individuals or groups or by demonstrating no significant differences in these variables on statistical comparison after recruitment (14, 17–20, 24–27, 29, 30). One of these studies did not specify that groups were matched and had not compared groups statistically but had provided sufficient data for the researcher to be reasonably sure that the groups were similar in terms of age and sex distribution (19, 20). In these methodologically restricted studies, there was a small to moderate effect size for depression seen in the heterogeneous (r = +.19, p < .0001) and homogeneous (r = +.13, p < .0001) study sets (Table 1).
In addition to controlling for age and sex, two studies had matched subjects for socioeconomic factors indirectly by recruiting friends, neighbors, and relatives of the patients with RA as control subjects (17, 18, 30). There was a moderate effect size from these two studies (r = .27, p < .0001), but this finding may have resulted from publication bias (fail-safe N = 20, file drawer n = 20).2
Other single studies had controlled for education (24), work status (24), marital status (24), and verbal IQ (25), which did not permit meta-analysis.
Comparing different measures of depression.
Ten independent studies comparing patients with RA with healthy control subjects used a single measure of depression (Refs. 14–18 and 25–30; C. M. Dickens et al., unpublished). Three of these studies had used the Hospital Anxiety and Depression Scale (HADS) (Refs. 17, 18, and 25; C. M. Dickens et al., unpublished). The effect sizes (or rather the Fisher’s Z transformation of the effect size) and probabilities (or rather the appropriate value for standard z) of these three studies were averaged (unweighted). This resulted in eight independent effect sizes for depression when patients with RA were compared with healthy control subjects.
The effects for these eight measures (GDS, HADS, PDI, IDD, POMS, CES-D, AIMS, and the depression subscale of Psychological Symptom Checklist; the latter was used by Kasl and Cobb, Ref. 29) formed a heterogeneous group (2 for Fisher’s Z = 21.7, df = 7, p = .003). Exclusion of the effect size from studies using the HADS left a homogeneous group of effects. This suggests that effect sizes obtained from studies using the HADS were quantitatively different from those obtained using other methods of assessment. Inspection of the effect sizes obtained using different depression assessments revealed that the HADS produced an effect size that was positive in direction and larger in magnitude than the effect sizes for other studies. This indicates that the studies using the HADS found patients with RA to be more depressed than did studies using other methods of assessment (Table 2). Elimination of the effect size for HADS (furthest from the weighted mean for the group) left a homogeneous population of studies (2 for Fisher’s Z = 11.2, df = 6, p = .08) (Table 2).
|
Determining moderator variables on depression effect sizes.
Further focused tests were performed to establish whether the effect sizes for depression (between patients with RA and healthy control subjects) varied in proportion to the degree of pain (21). Of the 12 studies comparing patients with RA with healthy control subjects, only 4 included measures of pain for both the RA and control groups (25, 26, 27, 30). Using contrasts (21, 22), the effect sizes for depression were found to increase in a linear manner in proportion to the effect size for pain (z = 2.67, p = .0064).
Comparisons of Patients With RA and Patients With Other Painful Conditions
Sixteen studies that compared patients with RA with subjects suffering from other painful conditions were found. Fibromyalgia, osteoarthritis, and back pain were the only painful disorders that were compared with RA in more than one study, enabling separate meta-analyses.
Patients with RA versus patients with fibromyalgia.
Nine studies comparing depression between patients with RA and those with fibromyalgia were found (19, 20, 26, 31–37). A small to moderate negative effect was seen when these groups were compared (r = -.14, p < .0001). This set of studies was homogeneous for effect sizes (Table 3). This indicates that the patients with fibromyalgia were more depressed than the patients with RA, but the difference between groups was small. This result can be considered to be robust to the effects of any publication bias (Table 3).
|
Seven methodologically restricted studies controlled for age and sex, resulting in significant effect sizes (r = -.21 (homogeneous), p < .0001), and two controlled for age, sex, and education (r = .28, p = .002). Both of these combined effects may have resulted from publication bias. Three studies controlled for age, sex, and duration of symptoms (r = .13 (homogeneous), p = .06). No studies controlled for age, sex, and disability (Table 3).
Patients with RA versus patients with osteoarthritis.
Four studies compared patients with RA and patients with osteoarthritis (37–40). These studies gave effect sizes that were small in magnitude but highly significant when the groups were compared with regard to depression (r = +.09, p < .0001). This effect remained small but significant when homogeneous groups of studies were examined (r = .08, p < .0001). Although these effects indicate that subjects with RA were more depressed than subjects with osteoarthritis, the difference between subjects was very small, and both results were vulnerable to publication bias (fail-safe N = 24, file drawer n = 30 and fail-safe N = 8, file drawer n = 25, respectively). Nonsignificant effects were found for pain (r = -.0001, p = 0.5). None of the studies compared the two groups by disability.
Patients with RA versus patients with back pain.
Only two studies comparing patients with RA and patients with back pain were found (25, 37). There was a statistically significant negative effect when the groups were compared for depression (r = -.05, p = .02). The effect size was very small, however, and it was not robust to publication bias (Fail-safe N = 2, file drawer n = 20). Only one study compared RA patients with back pain patients by pain and disability, so meta-analyses were not performed for these variables.
Patients with RA versus control subjects with no pain.
Only two studies comparing patients with RA and groups of subjects with nonpainful, chronic disabling conditions (spinal injury and multiple sclerosis) (15, 16, 41) were found. A small but significant negative effect was seen when these groups were compared for depression (r = -.12, p = .012). This result was not robust and may be the result of publication bias (fail-safe N = 6, estimated file drawer n = 20).
A large and highly significant positive effect was found between groups for pain (r = .48, p < .0001), but the effect for disability was nonsignificant (r = .09, p = .11).
Sources of Heterogeneity
Meta-analyses comparing RA patients with healthy control subjects, with fibromyalgia patients, and with osteoarthritis patients were examined to identify any methodological features that might account for heterogeneity of effect sizes. The methodological features examined were age, sex, duration of symptoms, and source of recruitment of patients with RA, total number of subjects included, whether standardized diagnostic criteria for RA were used, methods of assessment of depression, and year of publication of the study. On inspection of these variables, no consistent differences that could account for the heterogeneity in the different meta-analyses could be found between the homogeneous studies and others.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
Our findings indicate that subjects with RA are more likely to be depressed than healthy control subjects, and the combined effect size was small to moderate in magnitude. The effect size for depression was not substantially reduced in magnitude on controlling for age, sex, and socioeconomic status (the latter approximately controlled by recruiting friends and neighbors of the RA patients). Thus, the excess of depression in the patients with RA when compared with healthy control subjects was likely due to the presence of RA and not the result of sociodemographic differences between patients and control subjects.
The magnitude of the effect size for depression did vary significantly depending on which measure of depression assessment was used. Inspection of the effect sizes for the different depression assessments (Table 2) reveals the much larger effect size for depression obtained when the HADS was used. This finding cannot be dismissed as an idiosyncratic finding of a single study because the HADS was used in three independent studies. The effect size for the HADS shown in Table 2 was obtained by averaging (unweighted) the Fisher’s Z from three independent studies (effect sizes r = .43, .34, and .51) and then converting back to Pearson’s r to give an overall effect size. Similarly, the effect size from the study using the MMPI failed to form part of the homogeneous group, though this assessment was used in only one study.
It may be argued that the larger effect sizes may have occurred because the HADS contains items such as "I feel as if I am slowed down," "I can sit at ease and feel relaxed," and "I still enjoy the things I used to [enjoy]." Patients with RA may have scored these items as the result of their physical illness independent of any psychological symptoms, thus inflating their depression scores. However, other self-rated measures of depression are recognized as being vulnerable to having their scores inflated by the inclusion of somatic items (ie, they have poor content validity); this phenomenon is by no means limited to the HADS (13, 47). Interestingly, only one study used a standardized psychiatric interview, the PDI, to establish the diagnosis of depression according to DSM-III criteria (26). Using a standardized interview should minimize the impact of mistakenly rating physical symptoms as indicative of psychological disorder. The PDI resulted in effect sizes similar to those obtained with the majority of self-rating questionnaires (Table 2), indicating that this standardized interview did not produce quantitatively different results from self-rated questionnaires in the studies included.
Despite the tendency of the HADS to give greater effect sizes for depression when patients with RA are compared with healthy control subjects, exclusion of the HADS leaves a homogeneous study set, giving a significant effect that was hardly reduced in magnitude (r = .14, p < .0001). Thus, our conclusion that patients with RA are more depressed than healthy control subjects remains valid. However, our findings indicate that the degree of depression detected in patients with RA may vary significantly depending on the measure of depression used, so care should be exercised when interpreting the results of single studies measuring depression in patients with RA. It is not clear why the HADS recorded levels of depression that are much larger than those obtained with other measures. Further studies are required to determine which measures are superior.
The trend for effects for depression to increase in proportion to increases in the effect size for pain provides evidence that the association between RA and depression may be attributed, in part, to the level of pain. However, because many of the physical factors in RA, such as disability, are associated with the level of pain, it is unclear to what extent depression is directly attributable to pain (as is often assumed in the clinical situation). We were unable to examine the level of association of depression with individual physical dimensions of RA due to the small number of studies recording the different physical dimensions in patients and control subjects. In any event, though this question has academic meaning, it has little clinical significance because physical factors such as pain and disability rarely occur independently in patients with RA.
Patients with RA were significantly less depressed than patients with fibromyalgia (r = -.14), even though these groups reported similar levels of pain, and RA patients were more depressed than osteoarthritis patients (r = .08). This supports the findings of Hawley and Wolfe (37) that the degree of depression varies significantly according to the diagnosis. The number of studies comparing RA with other painful rheumatological conditions, however, was extremely limited.
Only two studies that compared patients with RA with subjects with chronic disabilities whose primary complaint was not pain were found (15, 16, 41). Although subjects with RA experienced more pain than and similar levels of disability as these other chronically disabled control subjects, this latter control group experienced more depression. However, the findings of this comparison were not robust because of the effects of publication bias.
Further cross-sectional studies are required to clarify the relationship of depression with other physical and social factors in RA. Multivariate analyses will help differentiate the variables that are independently associated with depression from those associated with depression indirectly (eg, through an association with a third factor). Varying the patient group under study will determine whether the social and disease factors related to depression vary with the nature of the illness. Such cross-sectional studies will help identify which depression assessments are superior and aid in the development of new measures. Longitudinal studies are required to determine the direction of the causal association between depression and other factors. In RA, physical, psychological, and social variables fluctuate over time. Such studies could monitor changes in variables such as depression, disability, pain, and social contact to examine the temporal, and hence likely causal, association between these variables. Such studies will increase our understanding of how physical and psychosocial factors might interact to result in psychological disorders in these patients and help in the direction of treatments to have greatest effects.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
2 Homogeneity cannot be calculated for fewer than three studies.
Received for publication October 26, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. R. Piazza, S. T. Charles, and D. M. Almeida Living With Chronic Health Conditions: Age Differences in Affective Well-Being J. Gerontol. B. Psychol. Sci. Soc. Sci., November 1, 2007; 62(6): P313 - P321. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Odegard, A. Finset, P. Mowinckel, T. K. Kvien, and T. Uhlig Pain and psychological health status over a 10-year period in patients with recent onset rheumatoid arthritis Ann Rheum Dis, September 1, 2007; 66(9): 1195 - 1201. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A S Russell, G V Wallenstein, T Li, M C Martin, R Maclean, B Blaisdell, K Gajria, J C Cole, J-C Becker, and P Emery Abatacept improves both the physical and mental health of patients with rheumatoid arthritis who have inadequate response to methotrexate treatment Ann Rheum Dis, February 1, 2007; 66(2): 189 - 194. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Sheehy, E. Murphy, and M. Barry Depression in rheumatoid arthritis--underscoring the problem Rheumatology, November 1, 2006; 45(11): 1325 - 1327. [Full Text] [PDF]
|
|
|
|
 |
|
 E Kozora, M C Ellison, J A Waxmonsky, F S Wamboldt, and T L Patterson Major life stress, coping styles, and social support in relation to psychological distress in patients with systemic lupus erythematosus Lupus, May 1, 2005; 14(5): 363 - 372. [Abstract] [PDF]
|
|
|
|
 |
|
 N. T. James, C. W. Miller, K. C. Brown, and M. Weaver Pain Disability Among Older Adults With Arthritis J Aging Health, February 1, 2005; 17(1): 56 - 69. [Abstract] [PDF]
|
|
|
|
|
|
G. J. Treharne, E. D. Hale, A. C. Lyons, D. A. Booth, M. J. Banks, N. Erb, K. M. Douglas, D. L. Mitton, and G. D. Kitas Cardiovascular disease and psychological morbidity among rheumatoid arthritis patients Rheumatology, February 1, 2005; 44(2): 241 - 246. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. E. Egede Effect of Comorbid Chronic Diseases on Prevalence and Odds of Depression in Adults With Diabetes Psychosom Med, January 1, 2005; 67(1): 46 - 51. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Groarke, R. Curtis, R. Coughlan, and A. Gsel The role of perceived and actual disease status in adjustment to rheumatoid arthritis Rheumatology, September 1, 2004; 43(9): 1142 - 1149. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Lowe, L. Willand, W. Eich, S. Zipfel, A. D. Ho, W. Herzog, and C. Fiehn Psychiatric Comorbidity and Work Disability in Patients With Inflammatory Rheumatic Diseases Psychosom Med, May 1, 2004; 66(3): 395 - 402. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Covic, B. Adamson, D. Spencer, and G. Howe A biopsychosocial model of pain and depression in rheumatoid arthritis: a 12-month longitudinal study Rheumatology, November 1, 2003; 42(11): 1287 - 1294. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. A. Rosenkranz, D. C. Jackson, K. M. Dalton, I. Dolski, C. D. Ryff, B. H. Singer, D. Muller, N. H. Kalin, and R. J. Davidson Affective style and in vivo immune response: Neurobehavioral mechanisms PNAS, September 16, 2003; 100(19): 11148 - 11152. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Henningsen, T. Zimmermann, and H. Sattel Medically Unexplained Physical Symptoms, Anxiety, and Depression: A Meta-Analytic Review Psychosom Med, July 1, 2003; 65(4): 528 - 533. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Dickens, J. Jackson, B. Tomenson, E. Hay, and F. Creed Association of Depression and Rheumatoid Arthritis Psychosomatics, June 1, 2003; 44(3): 209 - 215. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
