| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of General Internal and Psychosomatic Medicine, University of Heidelberg, Medical Hospital, Heidelberg, Germany (B.L., L.W., W.E., W.H.); Regenstrief Institute and Indiana University School of Medicine, Indianapolis, Indiana (B.L.); Department of Psychosomatic Medicine and Psychotherapy, University of Tübingen, Medical Hospital, Tübingen, Germany (S.Z.); and Department of Haematology, Oncology, and Rheumatology, University of Heidelberg, Medical Hospital, Heidelberg, Germany (A.D.H., C.F.).
Address correspondence and reprint requests to Bernd Löwe, MD, Dipl.-Psych., Regenstrief Institute for Health Care, RG-6, Indiana University School of Medicine, 1050 Wishard Boulevard, Indianapolis, IN 46202. E-mail: bloewe{at}regenstrief.org
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
METHOD: The patient group with IRD (73.9% female; mean age, 42.7 years) and the control group (73.9% female; mean age, 42.2 years) consisted of 356 outpatients each. Psychiatric comorbidity was diagnosed using the Patient Health Questionnaire (PHQ), functional disability was measured with the 12-item Short-Form Health Survey (SF-12), and severity of illness was assessed by the treating physicians. Stepwise logistic regression analyses were performed to identify independent predictors of work disability controlling for psychiatric comorbidity, functional disability, severity of illness, additional clinical variables, and demographics.
RESULTS: Psychiatric comorbidity increased work disability in patients with severe IRD from 25% to 50%, and from 5% to 17% in mild disease. Physical functioning (OR = 0.6; p < .0001), depression severity (OR = 1.6; p = .005), and illness severity (OR = 1.8; p = .006) were identified as independent predictors of work disability in patients with IRD. Similar predictors of work disability were identified in the control group, but depression severity did not meet statistical significance at the 5% level.
CONCLUSIONS: This study provides evidence that depression is an independent risk factor of work disability in patients with IRD, but this result cannot be generalized to other medical conditions. Improved diagnosis and treatment of depressive disorders in patients with IRD may help avoid work disability.
Key Words: work disability; • depressive disorder; • anxiety disorders; • rheumatic diseases; • depression; • diagnosis; • questionnaires;
Abbreviations: CI = confidence intervals;; IRD = inflammatory rheumatic diseases;; OR = odds ratio;; PCS-12 = Physical Component Summary from the SF-12;; PHQ = Patient Health Questionnaire;; PHQ-9 = Depression severity scale from the Patient Health Questionnaire;; RA = rheumatoid arthritis;; SF-12 = Medical Outcomes Study 12-item Short-Form Health Survey;; SF-36 = Medical Outcomes Study 36-item Short-Form Health Survey;; SLE = systemic lupus erythematosus.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
For several groups of medical patients, psychiatric comorbidity is strongly associated with disability, increased health care costs, and elevated risk of mortality (7–12). To our knowledge, our study is the first to show the relationship between work disability and psychiatric comorbidity in patients with IRD while controlling for physical variables such as disease severity and physical functioning. Increased rates of work disability in patients with IRD remain a significant public health concern. In fact, 25 to 40% of those with IRD become disabled during the first 5 years of their disease (13–17).
Identifying the predictors of work disability may help target treatments aimed at improving work function. The majority of the studies investigating work disability in IRD have focused on patients with RA. Multivariate analyses in the different studies have revealed heterogeneous results. Generally, most studies have identified greater patient age, low educational level, physically demanding jobs, employment by someone else, and physical functioning as independent predictors of work disability in patients with RA (15,17–22). In contrast, gender, duration of illness, marital status, and laboratory or radiological parameters did not predict work disability significantly. Few studies have incorporated psychosocial factors such as anxiety or depression, but remarkably, no study has attempted to demonstrate an independent impact of these variables in multivariate analyses. In their 18-year longitudinal study of patients with RA, Wolfe and Hawley (17) found that self-reported anxiety and depression are associated with subsequent work disability in a subset of patients with early RA. Nevertheless, it was not reported if depression and anxiety had an independent impact on work disability. The authors conclude that attention to self-reports of anxiety and depression will identify patients at high risk for work disability. In another prospective follow-up study, Fex et al (22) found that handicaps in terms of reduced leisure time activities, and difficulties performing different social roles are associated with psychological distress. However, this study did not include psychological distress or other psychosocial variables in analyses to identify predictors of work disability. Overall, these studies support a multifactorial model related to work disability in patients with IRD, but the independent contribution of psychiatric comorbidity to work disability remains unknown. Furthermore, it is unclear if predictors of work disability are unique to patients with certain medical conditions, or if these predictors are similar in patients, independent from their type of illness. If the latter is true, the causes of work disability and treatment aimed at improving work function could be generalized to many different conditions.
The primary purpose of our study was to identify independent predictors of work disability in patients with IRD with particular interest in psychiatric comorbidity. Parallel analyses were also performed in a matched control group without IRD to investigate if predictors of work disability are independent from the presence of IRD. For both patient groups, it was hypothesized that the rate of work disability is associated with psychiatric comorbidity and different levels of illness severity. We further hypothesized that psychiatric comorbidity is among the independent predictors of work disability in patients with IRD as well as in controls, corresponding to findings in several other patient groups. Finally, we assessed both groups for the proportion of patients receiving psychotherapeutic or psychopharmacological treatment.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
For the patient sample with IRD, a total of 603 patients were approached in the rheumatology outpatient clinic. One hundred forty-five patients declined to participate or did not return the questionnaire (participation rate = 76%). The most frequent reasons for nonparticipation were lack of time and perceived inconvenient experiences with other clinical studies or questionnaires. One hundred two patients were excluded because the medical examination failed to confirm IRD diagnosis. Thus, the final sample consisted of 356 patients with IRD.
Within the other medical outpatient clinics, 656 patients were eligible for the control group. The participation rate was 79%, and the reasons for nonparticipation were similar to the patient group with IRD. Of the participating 518 patients, 41 were excluded because they were diagnosed as having a current or past history of IRD. Complete questionnaires were received from 477 patients without IRD. To obtain a control group approximately equal in basic demographic characteristics to the patient group with IRD, the control group was matched by age and sex, using the patient group with IRD as a reference. Thus, a parallel control group of 356 patients without IRD and matched by age and sex was used for our analyses.
Measures
The patients of both groups were diagnosed by their physicians according to the International Classification of Diseases 10th Revision (ICD-10) (23). Additionally, the American College of Rheumatology (ACR) (24) and Tan et al criteria (25) were used for the diagnosis of RA and systemic lupus erythematosus (SLE), respectively. The treating physician diagnoses were extracted from the medical records of patients. In addition, the participating physicians completed a short questionnaire indicating the reason for the current patient visit and severity of illness evaluation. The severity of illness was assessed using a 5-step rating scale, ranging from [1] "completely healthy" to [5] "seriously sick."
The patient questionnaire for both patient groups consisted of sociodemographic data, information regarding work disability, employment status, characteristics of work, psychopharmacological medication, psychotherapy, and standardized self-report questionnaires. Psychiatric comorbidity was assessed using the Patient Health Questionnaire (PHQ) (26–28). PHQ is a self-report instrument specifically developed for use in primary care and other nonpsychiatric settings. For diagnoses, the PHQ offers categorical algorithms for 8 mental disorders based on modified criteria of each disorder, according to the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) (29). In addition, the PHQ includes a continuous scale for measuring depression severity, which is called PHQ-9 (28). Thus, the PHQ-9 offers both the opportunity to establish depression diagnosis using the categorical algorithm and to measure depression severity using the continuous sum score. Excellent operating characteristics have been demonstrated for the American (26, 28), German (30–32), and Spanish versions (33) of the PHQ. Superior criterion validity of the PHQ compared with 2 other established self-report questionnaires has recently been confirmed with respect to the diagnoses of "major depressive disorder" and "panic disorder" made by a standard interview in assessing psychiatric disorders (30–32). The PHQ has been applied to a large number of different patient samples, including patients with rheumatic diseases (1). In our study, psychiatric comorbidity was defined as "present," when patients had a depressive ("major depressive disorder" or "other depressive disorder") or an anxiety disorder ("panic disorder" or "other anxiety disorder") diagnosed with the PHQ.
Physical functioning was measured by the PCS-12 from the Medical Outcomes Study SF-12 (34,35). The brief SF-12 is derived from the Medical Outcomes Study SF-36 (36). The SF-12 and SF-36 are among the most frequently used generic health status measures. Multiple studies have confirmed the validity of these measures, and good correlations have been found between the Physical Component Summary scales of the SF-12 and the SF-36. One of these studies included patients with osteoarthritis or RA (37). Altogether, the SF-12 appears to be a psychometrically sound tool for physical function assessment in patients with IRD.
Our study aimed to identify predictors of work disability, irrespective of whether the patient was currently employed or not. In Germany, employed patients who are unable to work due to a chronic medical illness are either certified unfit by their doctors, or they receive temporary or permanent disability pension. Those with permanent disability are not expected to return to work in the future. In contrast, unemployed persons such as homemakers, are not usually certified unfit to work by their doctors. Accordingly, in our study, patients were assigned to the group of "work-disabled" if one of the following conditions applied: 1) formerly employed, currently receiving temporary or permanent disability pension, 2) employed (full-time or part-time) with a sick leave of at least 4 weeks, or 3) unemployed and unable to do their usual activities for at least 4 weeks because of medical illness. Work disability of at least 4 weeks was required to only include patients significantly disabled by their chronic disease. Due to data protection rights in Germany, we had to rely on patient self-reports regarding work disability. However, for patients receiving disability pension, a verification of their self-report was possible, because these patients have a different insurance status than those not receiving a disability pension.
Statistical Methods
Differences of categorical patient characteristics between IRD patients and controls and between patients with and without work disability were tested using chi-squared tests. If cell counts were less than 5, Fisher’s exact tests were performed. With respect to continuous variables, subgroup differences were tested using 1-way analysis of variance according to the General Linear Model.
Potential predictors of work disability were drawn from the studies mentioned above and from theoretical considerations. In this study, demographic variables, characteristics of work, physicians’ assessment of disease severity, illness duration, physical functioning (PCS-12), and psychiatric comorbidity (PHQ) were included as potential predictors of work disability for both patient samples. First, univariate analyses were performed within both patient groups to investigate differences between patients with and without work disability. Second, the relationship between rate of work disability and psychiatric comorbidity was investigated for patients with severe disease (physicians’ rating 3–5) and for patients with mild disease (physicians’ rating 1–2), respectively. Third, the most important independent predictors of work disability in patients with IRD were identified using multivariate logistic regression analyses. A stepwise forward variable selection approach was used and included all variables with p < .20 in the univariate analyses (38). An identical logistic regression analysis (ie, same variables and approach) was conducted in the control group. This forced replication of IRD group results was performed to investigate if the same model of predictive variables also fits patient groups with conditions other than IRD. Furthermore, to investigate whether IRD status confers special status to some predictors, the same stepwise logistic regression analysis was performed for the total patient sample, with group status as an additional predictor. Interactions between group status and the various predictors were examined using separate non-stepwise multivariate regression analyses. The dependent variable for all logistic regression analyses was the presence or absence of work disability for more than 4 weeks. Applicable predictor variables were maintained as continuous instead of dichotomizing them for logistic regression (39). The final subset of variables for patients with or without IRD represents the most important independent predictors of work disability for each patient sample.
Sample size estimation for multivariate analysis was performed at an 
-level of 0.05, power of 0.80, and assumption that multiple correlations between predictors were 0.20 or higher. As a result, a sample size of at least 350 patients was calculated for a model incorporating 7 different predictor variables (40).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
Differences Between Work-Disabled and Non–Work-Disabled Patients
Table 2 shows the results of univariate analyses investigating differences between patients with and without work disability in those with IRD and controls. Within the IRD group, the work-disabled patients lived less frequently with a partner, had lower educational levels, a higher percentage of psychiatric comorbidity (PHQ depressive or anxiety disorder; 45.1% vs. 17.8%), higher severity of illness, lower physical functioning, and higher depression severity compared with the non–work-disabled patients. The corresponding analyses in the control group revealed similar differences regarding educational level, psychiatric comorbidity (PHQ depressive or anxiety disorder; 45.5% vs. 22.0%), severity of illness, physical functioning, and depression severity. However, controls with and without work disability did not differ significantly with respect to living with a partner, and the work-disabled controls had a physically demanding job significantly more often than the controls without work disability.
|
2 = 45.4; df = 3; p < .0001) and for controls (2 = 24.7; df = 3; p < .0001). Similar results emerged if the patients’ self-report of physical functioning (SF-12) was used instead of the physicians’ assessment of illness severity.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
The findings from our study support this hypothesis. Along with physical functioning and illness severity, depression severity independently predicted work disability in patients with IRD. Psychiatric comorbidity, defined as the presence of depressive or anxiety disorders substantially increased work disability by 25% in patients with severe IRD and by 12% in those with mild disease. Alarmingly, only 25% of the IRD patients with psychiatric comorbidity and 36% of the affected controls had their psychiatric comorbidity treated by psychotherapy, medication, or both. In accordance with preliminary studies (15,17–22), other variables, such as low educational level and not living with a partner, were also associated with work disability. However, these variables had no independent contribution to work disability in the multivariate analysis.
Identical analyses were performed to investigate if the results from IRD patients are transferable to other patient groups with no diagnosis of IRD. The results from the logistic regression analysis in the IRD patients were entirely replicated by the analysis in the total sample. We found neither group status nor interactions between predictor variables as contributing independently to work disability. Thus, the results from the total sample support the hypothesis that predictors of work disability are similar in patients with and without IRD. However, the analysis of the control group revealed only physical functioning and physicians’ rating of illness severity as independent predictors of work disability. Depression severity was also positively associated with work disability, but it did not reach statistical significance. Considering that neither group status nor interactions contributed to work disability in the total sample, one suitable explanation for the weaker association of depression severity with work disability in the controls might be the smaller rate of work disability in controls compared with IRD patients. In addition, the heterogeneous composition of the control group might have influenced this result. Finally, physical functioning and illness severity are strongly associated with work disability and may contribute more significantly to work disability than depression severity. We would caution to not discount the potential impact of depression severity on work disability, even though it did not reach statistical significance in the control group. As shown by the univariate analyses, levels of depression severity were significantly higher in work-disabled controls compared with non–work-disabled controls, and depression severity might be confounded by self-perceived physical functioning and physicians’ assessment of illness severity. As such, an improvement in depression severity might also lead to an improvement in work functioning in the control group.
The case mix of our patients with IRD was comparable to those at other university-based rheumatology departments in Germany, with the exception that the proportion of patients with RA was considerably lower in our sample (21% vs. 37%) (16). In rheumatology outpatient clinics at general hospitals, in individual rheumatology practices, and in rehabilitation hospitals, the proportion of patients with RA might exceed 50% (16). Twenty-four percent of our IRD patients were work-disabled after a mean illness duration of 6.5 years, which is similar to the rate reported for RA patients 6.4 years after disease onset (17). Naturally, this rate is higher in studies with longer duration of illness (16,22). As expected, the rate of work disability in controls was significantly lower than in IRD patients. The 25% prevalence rate of depressive and anxiety disorders in both patient groups corresponds well to other studies in primary care (26,41). However, even higher prevalence rates can be expected inpatient groups with more severe cases and/or longer courses of disease (6).
Some limitations of the study have to be mentioned. First, psychiatric comorbidity and work disability were assessed simultaneously in our study. Due to the cross-sectional design, we cannot distinguish whether the identified predictor variables cause work disability or vice versa. Another possibility, however, is that work disability and the predictor variables are caused by a common underlying factor. Undoubtedly, work disability and depression are closely linked, and either variable could precede the other. Thus, prospective studies in patients with IRD with a subsequent evaluation of predictor variables, such as depression severity and work disability, would be required to distinguish cause from effect. A second limitation of our study is the reliance on patient self-reports. Only medical diagnoses and severity of illness were reported from the treating physicians. However, an external cross-validation of self-reported work disability in patients receiving disability pension confirmed the validity of the patients’ statements. For the measurement of psychiatric comorbidity and physical functioning, diagnostic instruments with proven validity have been applied. The composition of our control group is a third potential limitation of our study. The control group was matched by age and sex to the IRD group and was similar in illness duration and rate of psychiatric comorbidity. However, it differed with respect to other potential confounders like sociodemographic characteristics, rate of work disability, physical demands on job, and illness severity from the IRD group. In addition, although the presence of IRD was excluded in any patient, the control group comprises patients with heterogeneous symptoms and diagnoses. Therefore, the control group does not represent a patient group with a particular diagnosis, and it remains unclear if these results can be transferred to other patient groups. Finally, although our sample size was sufficient for the inclusion of a maximum of 7 predictors into the multivariate regression analyses (40), statistical significance is dependent on sample size, and larger sample sizes might potentially have led to the identification of additional predictors of work disability in both patient groups.
Nevertheless, the association of work disability and psychiatric comorbidity in both patient groups, the independent contribution of depression severity to work disability in patients with IRD, and the low treatment rate, all underline the necessity for a more comprehensive assessment of psychiatric comorbidity in patients with IRD. The use of self-report screening instruments like the Patient Health Questionnaire (PHQ) can contribute to a better diagnosis and treatment of psychiatric comorbidity in busy medical settings (31), and it is well accepted by the large majority of the patients (32). The depression scale of the PHQ has been shown to be sensitive to change and can also be administered to monitor response to treatment (42). Alternatively, simple screening questions can be used to identify patients with psychiatric comorbidity in busy medical settings: Questioning for 1) nervousness, anxiety, or worries; 2) depressed mood; 3) multiple somatic symptoms; and 4) interpersonal distress could identify as many as 86% of cases with psychiatric comorbidity in medical outpatients (43). However, screening and diagnosing depression or other psychiatric disorders in medical patients is a futile exercise unless it is closely linked to treatment initiation. The US Preventive Services Task Force accordingly recommends screening for depression in "clinical practices that have systems in place to assure accurate diagnosis, effective treatment, and follow-up" (44). Effective and efficient psychosocial and pharmacological interventions are available for the treatment of depression in most medical outpatients (45,46), but there have been few attempts to measure the efficacy of these interventions in patients with IRD. At the present time, it can only be speculated that these interventions have positive consequences on both depression and work disability in patients with IRD. Nevertheless, additional efforts to improve diagnosis and treatment of psychiatric comorbidity in patients with IRD are necessary. These efforts may help achieve better patient-centered outcomes and reduce societal costs related to psychiatric comorbidity.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
We thank the patients and our colleagues who collaborated in this study. We would also like to express our thanks to Kerstin Gräfe, University of Heidelberg, Medical Center, for her support during all phases of this study. In addition, we greatly appreciate the valuable contributions of Donna Fadden, Matt Bair, Tony Perkins, and Kurt Kroenke, Regenstrief Institute, Indiana University School of Medicine, Indianapolis, to the final version of the manuscript.
Received for publication January 22, 2003.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S. L. Hider, W. Tanveer, A. Brownfield, D. L. Mattey, and J. C. Packham Depression in RA patients treated with anti-TNF is common and under-recognized in the rheumatology clinic Rheumatology, September 1, 2009; 48(9): 1152 - 1154. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Levinson and Y. Lerner Employment Among Persons With Past and Current Mood and Anxiety Disorders in the Israel National Health Survey Psychiatr Serv, May 1, 2009; 60(5): 655 - 662. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Schmitz, J. Wang, A. Malla, and A. Lesage Joint Effect of Depression and Chronic Conditions on Disability: Results From a Population-Based Study Psychosom Med, May 1, 2007; 69(4): 332 - 338. [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]
|
|
|
|
|
|
P. E. Stang, N. A. Brandenburg, M. C. Lane, K. R. Merikangas, M. R. Von Korff, and R. C. Kessler Mental and Physical Comorbid Conditions and Days in Role Among Persons With Arthritis Psychosom Med, January 1, 2006; 68(1): 152 - 158. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Lowe, K. Grafe, C. Ufer, K. Kroenke, E. Grunig, W. Herzog, and M. M. Borst Anxiety and Depression in Patients With Pulmonary Hypertension Psychosom Med, November 1, 2004; 66(6): 831 - 836. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
