All the authors except for Dr. Wetherell are affiliated with the Health Services Research and Development Center for Mental Healthcare and Outcomes Research, Central Arkansas Veterans Healthcare System, Little Rock, and with the Department of Psychiatry and Behavioral Sciences, University of Arkansas for Medical Sciences, South Central Mental Illness Research, Education, and Clinical Center, Little Rock. Dr. Wetherell is with the Advanced Center for Intervention and Services Research, University of California, San Diego. Send correspondence to Dr. Mittal at the Health Services Research and Development Center for Mental Healthcare and Outcomes Research, Central Arkansas Veterans Healthcare System, 2200 Fort Roots Drive, Building 58 (152/NLR), Little Rock, AR 72114 (e-mail: dmittal@uams.edu).

Major depressive disorder and anxiety disorders are the most prevalent psychiatric disorders treated in primary care clinics (1). Anxiety symptoms frequently coexist with major depression (1,2,3), but reported prevalence rates vary. Clayton and colleagues (4) reported that among 327 patients with depression, nearly two-thirds had anxiety symptoms. In another primary care study a majority of patients had mixed anxiety-depressive syndromes (42 percent) or depression with comorbid anxiety (19 percent), whereas 13 percent had anxiety only and 10 percent had depression only (5). Compared with each disorder alone, comorbid depression and anxiety is associated with poorer prognosis (6), greater chronicity and severity of illness (7,8), lower rates of treatment response (9,10), higher suicide risk (11,12), and greater functional disability (9,13,14).

Health outcomes are increasingly being evaluated from the patient's perspective, and patients frequently rank improved quality of life as more important than decreased symptoms (15). Many studies have investigated the health-related quality of life of patients with either depression or anxiety disorders, but less is known about the combined impact of comorbid major depressive disorder and anxiety disorders on health-related quality of life.

Five studies have examined the impact of comorbid anxiety disorders on health-related quality of life among primary care patients with major depressive disorder (9,13,16,17,18). Andrews and colleagues (16) examined data from the Australian epidemiologic survey (N=10,641) and found that health-related quality of life was better for respondents with major depressive disorder alone compared with those with one or more comorbid anxiety disorders. The study did not report how health-related quality of life was affected by specific anxiety disorders—such as generalized anxiety disorder, panic disorder, posttraumatic stress disorder (PTSD), or social anxiety disorder.

Sherbourne and colleagues (13) used data from the Medical Outcomes Study (19) (N=875) and reported nonsignificant impairment in health-related quality of life, as measured by the 36-item Medical Outcomes Health Survey Short Form (SF-36), among primary care patients with depression and comorbid panic disorder or generalized anxiety disorder compared with those with depression alone. That study did not evaluate the impact of comorbid PTSD.

Wittchen and colleagues (17) examined the impact of major depression and comorbid generalized anxiety disorder by using data from a German epidemiologic survey (N= 4,181) and found that compared with patients with major depression alone, patients with comorbid generalized anxiety disorder had scores reflecting more problems on the mental health component summary scale but not on the physical health component summary scale of the SF-36.

Using a single-item measure of health-related quality of life, Hegel and colleagues (18) investigated the impact of comorbid panic disorder and PTSD on depression treatment outcomes among older primary care patients (N=1,801). At baseline comorbid panic disorder and PTSD among depressed patients resulted in significant additional impairment in health-related quality of life. Similarly, Brown and colleagues (9) found that depressed primary care patients with lifetime comorbid anxiety disorder (generalized anxiety disorder or panic disorder) had worse health-related quality of life, as measured by the SF-36, than those with depression alone (N=157).

These studies provide evidence that anxiety disorders comorbid with major depressive disorder are associated with a measurable decline in health-related quality of life. However, the results are not consistent, and at least one study reported nonsignificant impairment (13). Our study contributes to the literature by simultaneously examining the relative impact of three common comorbid anxiety disorders (generalized anxiety disorder, panic disorder, and PTSD) in a large sample of primary care patients with major depression who received care from the Department of Veterans Affairs. Anxiety disorders were identified by using a structured diagnostic interview, and we estimated the impact of comorbid anxiety on health-related quality of life by using both the commonly employed SF-12V (a shorter version of the SF-36 modified for veterans) and the more comprehensive self-administered version of the Quality of Well-Being Scale (QWB-SA). We hypothesized that generalized anxiety disorder, PTSD, or panic disorder comorbid with major depressive disorder would impair health-related quality of life above and beyond that of major depressive disorder alone.

In the literature, the terms "health-related quality of life," "quality of life," "functional status," and "health status" have been used interchangeably (20). In this article, we use the term health-related quality of life to describe the patient's self-reported perception of his or her physical, emotional, mental, and functional well-being, rather than the term quality of life, which also captures the patient's subjective experience of his or her social situation (non-health-related factors). We used two measures of health-related quality of life: the QWB-SA and the SF-12V. The QWB-SA integrates physical, mental, and functional health dimensions into a single score that can be compared across disorders, whereas the SF-12V (21) provides separate physical health and mental health component summary scores.

Our study utilized baseline data from the Telemedicine Enhanced Antidepressant Management (TEAM) Study, which evaluated the effectiveness of a telemedicine-based collaborative care intervention for depression (22). Baseline data were collected from March 2003 to October 2004. The TEAM Study was approved by the institutional review board at each of the participating sites.

Participants

The TEAM Study recruited patients from seven VA community-based outpatient clinics in Arkansas, Louisiana, Mississippi, and Texas. Patients with depression who received care from primary care providers and scored 12 or higher on the nine-item Patient Health Questionnaire depression screener were eligible for the study. The TEAM Study excluded veterans with severe cognitive impairment—defined as a score of 11 or higher on the Blessed Dementia Scale (23)—psychotic disorders, bipolar disorder, alcohol or drug dependence, and those receiving care in a VA mental health clinic. We excluded patients receiving specialty mental health care because the focus of our study was on improving depression treatment in the primary care setting. For this study, a subsample of 324 participants was extracted from the full sample (N=395) recruited for the TEAM Study. We excluded participants with minor depression (N=71) because comorbid anxiety may interact differently with minor depression and major depressive disorder (24,25).

Measures

We collected data on health-related quality of life, anxiety comorbidity, physical comorbidity, depression history, at-risk drinking, and demographic characteristics. All data were collected by using computer-assisted telephone interviews. The primary measure of health-related quality of life was the QWB-SA (26). Each item on the QWB-SA is preference weighted by using published preference weights from a representative community sample. The preference weights have been shown to be reliable, internally consistent, correlated with a wide variety of medical and psychosocial variables, and stable across patient groups and over time (27,28). The overall score ranges from 0 for death to 1.0 for asymptomatic optimal functioning. The QWB-SA is divided into five parts. For all sections, the participants are asked to consider the past three days while responding to the questions. Part I asks about 25 acute physical symptoms and 11 acute and chronic mental health symptoms. Part II assesses self-care, such as whether they need help caring for themselves or whether they required hospital services. Part III assesses mobility—for example, use of public transportation or driving. Part IV assesses physical functioning—for example, walking or confinement to a bed or chair. Part V assesses performance of usual activities—for example, work, school, or housework.

The SF-12V (21), our secondary measure of health-related quality of life, yields a mental health component summary score (MCS-12V) and a physical health component summary score (PCS-12V). Both scales are standardized to the U.S. population so that the scores have a direct interpretation in relation to the distribution of scores in the U.S. population, which has a mean±SD score of 50±10. Higher scores indicate better health. The SF-12V has been shown to have good reliability, internal consistency, and validity across groups with differing sociodemographic characteristics (21,29).

The Mini International Neuropsychiatric Interview (MINI) (30) was administered at baseline to detect the presence or absence of major depressive disorder and comorbid psychiatric disorders (including dysthymia, at-risk drinking, panic disorder, generalized anxiety disorder, and PTSD). The MINI is a structured diagnostic interview to assess DSM-IV and ICD-10 psychiatric disorders. It was specifically designed to be administered by nonclinicians in multicenter clinical trials and epidemiologic studies and used as a first step in outcomes tracking in nonresearch clinical settings. The MINI has good reliability and validity and has been used in international epidemiologic and treatment outcomes research (31).

Depression severity was assessed by using the 20-item Symptom Checklist (SCL-20), a self-report instrument that has good reliability and validity (32). Possible scores on the SCL-20 range from 0 to 4, with higher scores indicating greater severity.

We used the Depression Outcomes Module (33) to obtain information on demographic characteristics, physical comorbidity, and depression-related clinical information for all participants. The Depression Outcomes Module has been shown to have good validity and reliability.

Social support was measured by using the social support subscale of the Duke Social Support and Stress Scale (34). Possible scores range from 0 to 1, with 0 denoting the worst social support and 1 indicating the best. The social support subscale has been proven to be valid and reliable (35).

Statistical analysis

We used multiple linear regression analyses to test the relationship between comorbid anxiety disorders (generalized anxiety disorder, panic disorder, or PTSD) and health-related quality of life (scores on the QWB-SA, PCS-12V, and MCS-12V). We controlled for demographic variables (age, gender, race, marital status, education, social support, and income), at-risk drinking, self-reported number of chronic medical conditions, and depression-related variables (presence versus absence of family history of depression, number of previous episodes, age of onset of depression, past depression treatment, current depression treatment, and depression severity). Overall, 2 percent of the data were missing. Missing data were imputed by using SAS PROC MI. Regression parameter estimates and standard errors were generated by using SAS PROC MIANALYZE. Sensitivity analyses were conducted by dropping observations with missing data (instead of imputing missing data), and the results were very similar.

The socioeconomic and clinical characteristics of the 324 study participants are provided in Table 1. A majority had a comorbid diagnosis of one or more anxiety disorders (69 percent). Specifically, 58 percent had generalized anxiety disorder, 29 percent had PTSD, and 11 percent had panic disorder. The mean score on the QWB-SA overall was .39±.11. The mean scores on the QWB-SA were similar to those of inpatients with depression (36). The mean score on the PCS-12V was 30.29, and the mean MCS-12V score was 34.13. These scores are far below those of the general population and are well below those of other veterans using VA primary care services (37). Study participants had multiple prior depression episodes, two-thirds (69 percent) had received prior depression treatment, and nearly half (43 percent) were receiving depression treatment at baseline. Of particular note is the extremely high number of physical comorbidities. Participants had a mean of 5.8 self-reported illnesses—for example, diabetes (108 participants, or 33 percent), heart disease (105 participants, or 32 percent), lung disease (65 participants, or 20 percent), stroke (36 participants, or 11 percent), and cancer (39 participants, or 12 percent).

Regression results are reported in Table 2. Generalized anxiety disorder and PTSD independently predicted the QWB-SA scores, whereas panic disorder did not. In addition, social support, depression severity, and the number of chronic medical conditions significantly predicted QWB-SA scores. Among the anxiety disorders, only PTSD predicted lower PCS-12V scores. As expected, the number of self-reported chronic physical health conditions significantly predicted PCS-12V scores. The number of depression episodes also significantly predicted higher PCS-12V scores. Surprisingly, none of the comorbid anxiety disorders predicted MCS-12V scores. Similar to the analyses of QWB-SA, social support and depression severity significantly predicted MCS-12V scores. Moreover, current (that is, prebaseline) depression treatment significantly predicted lower MCS-12V scores.

To our knowledge this study is the first to demonstrate the differential effects of three common comorbid anxiety disorders in the context of major depressive disorder by using comprehensive measures of health-related quality of life (QWB-SA and SF-12V). In our sample of primary care patients with depression, approximately 70 percent had a comorbid anxiety disorder. When the analyses controlled for demographic and clinical variables (including depression severity), we found that the presence of either generalized anxiety disorder or PTSD among persons with major depressive disorder resulted in additional impairment in health-related quality of life, as measured by the QWB-SA but that panic disorder does not. The effect sizes for generalized anxiety disorder (Cohen's d=.27) and PTSD (Cohen's d=.36) were in the moderate range.

These findings are consistent with prior studies that found worse health-related quality of life among depressed persons with comorbid generalized anxiety disorder or comorbid PTSD compared with those with major depressive disorder alone (9,13,16,17,18). Our finding that comorbid panic disorder did not result in additional impairment in health-related quality of life was consistent with the results of Sherbourne (13) but inconsistent with the findings of Hegel and colleagues (18). It is possible that our sample, and that of Sherbourne and colleagues, included too few participants with panic to detect anything but large effects.

The predictors of PCS-12V and MCS-12V scores were somewhat different from the predictors of QWB-SA scores. None of the comorbid anxiety disorders predicted scores on the MCS-12V, and only PTSD predicted scores on the PCS-12V. Thus our findings are mixed with respect to our health-related quality-of-life outcome measures. These differences may result from the slightly different aspects of health-related quality of life assessed by the three dependent measures. For example, the PCS-12V focuses on physical health and MCS-12V focuses on mental health; however, the QWB-SA combines both physical and mental health-related quality of life. Furthermore, the QWB-SA asks participants to respond to questions related to symptoms and functioning during the past three days, whereas the SF-12V asks respondents about the past four weeks.

Moreover, the QWB-SA and SF-12V are scored by using different weighting methods—the QWB-SA uses preference weights and the SF-12V uses scoring coefficients from an orthogonal factor rotation method. Specifically, items from the mental health and role-emotion subscales of the SF-12V that contribute to the MCS-12V score also make negative contributions to PCS-12V, whereas items from the physical function, role-physical, and bodily pain subscales that contribute to the PCS-12V score also make negative contributions to MCS-12V (38). Therefore, improvements on physical function, role-physical, and bodily pain subscales may lower the scores on MCS-12V. The failure of comorbid anxiety disorders to predict MCS-12V scores may also be due to the possibility that impairment caused by comorbid anxiety disorders is subsumed within the impairment caused by major depressive disorder alone.

Another possible explanation is that the SF-12V was developed to detect large differences across heterogeneous populations, thus making the MCS-12V less sensitive to differences among patients in treatment for mental health disorders. Additionally, the SF-12V does not include any questions that specifically probe for anxiety symptoms, whereas the QWB-SA has at least two highly weighted items that specifically ask about anxiety symptoms. These differences in the two instruments may make the QWB-SA more sensitive to detecting impairment in health-related quality of life that is related to anxiety symptoms.

The high rates of comorbid depression and anxiety disorders and decreased health-related quality of life among patients found in this study and the poor response to depression treatment among these patients, as shown in the literature, suggest that this is an area in need of future study. At the very least, it may be useful to assess for anxiety disorders when an episode of major depressive disorder does not respond to first-line treatment. Current American Psychiatric Association guidelines for depression recommend the same pharmacological treatment for patients with comorbid anxiety and depression as for those with depression alone (39). They do not offer any specific recommendations for treating comorbid anxiety and major depressive disorder by using either pharmacological or psychotherapeutic modalities. We did not find any studies that examined the critical question of whether patients with comorbid anxiety should be treated differently than those without comorbid anxiety. Our findings suggest a need for randomized clinical trials that will provide empirical evidence about the efficacy of specific interventions that concurrently target anxiety disorders and will aim to improve health-related quality of life.

Symptomatic improvement is often viewed as a more proximal outcome, whereas health-related quality of life is a more distal outcome (40,41). One question that arises is whether there is a lag time between symptom improvement and improved health-related quality of life. For depression there is some evidence that the lag time may be very short or nonexistent (36,42). Health-related quality of life takes into account not only physical and mental health symptom severity but also psychosocial functioning (43). A single health-related quality of life construct facilitates the comparison of impairment associated with illnesses across the physical-mental health spectrum and the effectiveness and cost-effectiveness of various health care interventions (44).

A limitation of this study is that it was cross-sectional, and thus our results indicate only associations, not causality. Longitudinal studies are needed to better define the factors that lead to lower health-related quality of life among patients with major depressive disorder. The results of this study may be generalizable only to veterans and are not necessarily applicable to those who receive care in other systems, especially for women who were underrepresented in our sample. The cross-sectional design of our study prevented us from investigating the primary disorder that led to the comorbidity. At least two longitudinal studies have investigated this issue, and findings were consistent in both studies: anxiety typically precedes depression, suggesting that generalized anxiety disorder is, in the temporal sense, the primary disorder (45,46). Comorbid generalized anxiety disorder with depression is associated with lower remission rates and with longstanding, possibly genetic vulnerability (45,46). A strength of our study was that we were able to examine the relative impact of three common anxiety disorders, identified using a structured clinical assessment. Our hypotheses were tested by using a large sample, and because the prevalence rate of PTSD was relatively high, we had the statistical power to detect its independent impact on health-related quality of life. In addition, health-related quality of life was measured by using both the SF-12V and the more comprehensive QWB-SA.

Our study confirms that anxiety disorders are highly prevalent among patients with major depressive disorder and that generalized anxiety disorder or PTSD comorbid with depression impairs health-related quality of life as measured by the QWB-SA. Our study implies that there is a need for future studies to target both anxiety and depressive symptoms among patients with comorbid anxiety and depression.

This research was supported by grant VA-IIR-00-078-3 to Dr. Fortney and grant VA-NPI-01-006-1 to Dr. Pyne and by the Department of Veterans Affairs (VA) Health Services Research and Development Center for Mental Health and Outcomes Research and the VA South Central Mental Illness Research Education and Clinical Center (MIRECC). Dr. Edlund was supported by a VA Health Services Research and Development Research Career Development Award and by South Central MIRECC. Dr. Mittal's time was supported in part by the South Central Network Research and Career Development Grant Program.