Available online at www.sciencedirect.

com

Journal of Psychiatric Research 42 (2008) 10421049

JOURNAL OF PSYCHIATRIC RESEARCH

www.elsevier.com/locate/jpsychires

Examining quality of life in patients with generalized anxiety disorder: Clinical relevance and response to duloxetine treatment
Mark H. Pollack a,*, Jean Endicott b, Michael Liebowitz b, James Russell c, Michael Detke c,d,e, Melissa Spann c, Susan Ball c,e, Ralph Swindle c
a

Center for Anxiety and Traumatic Stress Disorders, Massachusetts General Hospital/Harvard Medical School, Simches Research Building, 185 Cambridge Street, Suite 2200, 2nd Floor, Boston, MA 02114, United States b New York State Psychiatric Institute/College of Physicians and Surgeons, Columbia University, New York, NY, United States c Lilly Research Laboratories, Indianapolis, IN, United States d McLean Hospital/Harvard Medical School, Boston, MA, United States e Indiana University School of Medicine, Indianapolis, IN, United States Received 14 February 2007; received in revised form 19 November 2007; accepted 21 November 2007

Abstract Background: Duloxetine, a serotonergic noradrenergic reuptake inhibitor, improved functional outcomes in each of three clinical studies for the treatment of adults with generalized anxiety disorder (GAD). Using comparison norms, the current work describes the clinical relevance of these functional improvements in terms of return to normative functioning and symptom remission. Methods: Data were pooled at the individual patient level from three double-blind, placebo-controlled trials of duloxetine treatment (910 weeks acute therapy, dose ranges 60120 mg). Inclusion/exclusion criteria were consistent across studies, and outcome measures included the Sheehan Disability Scale (SDS), Quality of Life Enjoyment and Satisfaction Questionnaire Short Form (Q-LES-Q-SF), and European Quality of Life 5 Dimensions (EQ-5D). Results: Adult patients (mean age = 42.4 years; 65% women) were randomly assigned to duloxetine (N = 668) or placebo (N = 495). At baseline, the majority of patients were impaired on the SDS global functioning (89%), Q-LES-Q-SF maximum percent (95%), and EQ5D (76%) scores. On each measure, a greater percentage of duloxetine-treated patients converted from an impaired baseline to a normative endpoint score than did placebo-treated patients (p 6 0.001, all comparisons). Remission dened as a HAMA total score at endpoint of 610, compared with 67, captured a greater proportion of patients who were functionally in remission. Conclusions: GAD is associated with substantial impairment in functioning and subjective well-being, and patients treated with duloxetine 60120 mg/day, compared with placebo, experienced a greater return to normative functioning. Attention to role functioning and quality of life may rene our denition of remission when using standard symptom measures of anxiety. 2007 Elsevier Ltd. All rights reserved.
Keywords: Quality of life; Generalized anxiety disorder; Duloxetine; Clinical relevance

1. Introduction As part of evaluating the eectiveness of treatment interventions, clinicians and researchers concur that impact on an individuals quality of life is a crucial outcome. Despite
*

Corresponding author. Tel.: +1 617 724 0844; fax: +1 617 643 3080. E-mail address: mpollack@partners.org (M.H. Pollack).

this importance, quality of life remains a complicated construct for measurement (Holmes, 2005). Dierent conceptualizations have been proposed, but most denitions of quality of life incorporate three key areas: role functioning, subjective well-being, and overall health status (Patrick and Chiang, 2000). Role functioning signies the ability to perform ones duties, engage in social relationships, and manage self-care. Subjective well-being is more broadly

0022-3956/$ - see front matter 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.jpsychires.2007.11.006

M.H. Pollack et al. / Journal of Psychiatric Research 42 (2008) 10421049

1043

characterized, but usually involves happiness and satisfaction with life activities. Overall health status may refer to variables related to the specic disease process, but more often reects a holistic assessment as well (Holmes, 2005). Clinical trials have increasingly included measures of these aspects of quality of life in evaluation of pharmacological treatments, but the dilemma then advances from the detection of to the interpretation of change. Given the subjective nature of quality of life indices, the mean changes associated with treatment interventions can be difcult to translate into personal or clinical relevance for the individual patient. Several methods for assigning meaningfulness to change have been advocated, including the use of eect sizes (mean changes in terms of standard deviations), comparison of the self-report with an external criterion (e.g., a global measure of change), or statistical methods (e.g., standard error of measurement) (Patrick and Chiang, 2000; Wyrwich et al., 1999). For most patients, an intuitive meaningful change would be a return to a premorbid state of well-being; however, chronic illnesses may result in a diminished baseline that can obscure this reference point. Under these conditions, the use of an absolute reference, such as normative values from healthy controls, can provide a more meaningful context for the interpretation of within-subject change. Generalized anxiety disorder (GAD) is a chronic anxiety illness that has been shown to result in diminished quality of life as conceptualized by role functioning, well-being, and health state (Mogotsi et al., 2000). In the present study, we pooled data from three double-blind, placebo-controlled clinical trials that examined the ecacy of duloxetine, a serotonergic noradrenergic reuptake inhibitor, for the treatment of adults with GAD. In each of these trials, duloxetine treatment signicantly improved anxiety symptom severity for patients with GAD (Hartford et al., 2007; Koponen et al., 2007; Rynn et al., 2007). Measures of role functioning, subjective well-being, and health status were also administered within these 910 weeks acute treatment trials. Results from each study showed that patients who were treated with duloxetine experienced signicant improvements (dened as mean changes in total scores) compared with placebo in these aspects of quality of life (Endicott et al., 2007). For this manuscript, the primary objective was to place these quality of life improvements into an appropriate clinical context by using comparison norms to describe their relevance in patients with GAD. A secondary objective was to describe the association between quality of life and ecacy measures to provide clinical guidance on the magnitude of symptom change associated with a return to normal functioning. 2. Materials and methods 2.1. Study designs Data were pooled from three randomized, double-blind, placebo-controlled, multicenter trials conducted in adult

outpatient centers in seven countries, including the United States. Study 1 was a 9-week acute treatment study that compared xed-dose duloxetine 60 mg (N = 168), duloxetine 120 mg (N = 170), or placebo (N = 175). Studies 2 and 3 were exible-dose studies of duloxetine 60120 mg (N = 330) or placebo (N = 320) with 10-week acute therapy. Study 3 also included an active comparator arm, venlafaxine XR 75-225 mg daily, which was not included in the duloxetine pooled analyses. 2.2. Study patients Men and women P18 years of age were recruited from outpatient centers. Patients were administered the Mini International Neuropsychiatric Interview (MINI; Sheehan et al., 1998) and had to meet the Diagnostic and Statistical Manual, 4th edition, Text Revision (DSM-IV-TR, American Psychiatric Association, 2000) criteria for GAD, and their diagnosis was also conrmed by study psychiatrist. The GAD illness was required to be at least moderately severe as indicated by a Hospital Anxiety Depression Scale (HADS; Zigmond and Snaith, 1983) anxiety subscale score P10 and a score of P4 on the Clinical Global ImpressionSeverity scale (CGI-S; Guy, 1976). To ensure predominance of anxiety symptoms, patients ratings on the Covi anxiety scale (Covi; Lipman and Covi, 1976) had to be higher than their score on the Raskin Depression Scale (RDS; Raskin et al., 1969) with no item on the RDS rated >3. Reasons for diagnostic exclusion in each study included any primary DSM-IV Axis I diagnosis other than GAD (including major depressive disorder) within the past 6 months; panic disorder, post-traumatic stress disorder, or an eating disorder within the past year; any past or current obsessive-compulsive disorder, bipolar disorder, psychosis, factitious disorder, or somatoform disorders; or a DSMIV-dened history of alcohol or any psychoactive substance abuse/dependence within the prior 6 months. Other exclusion criteria were benzodiazepine use 2 weeks prior to randomization; serious suicide risk, initiation of psychotherapy, any medical illness that contraindicated treatment with duloxetine, or the use of any excluded concomitant or other psychotropic medications. 2.3. Ecacy measures The Hamilton Anxiety Rating Scale (HAMA; Hamilton, 1959) was administered at each visit using the Structured Interview Guide for HAMA (SIGH-A; Shear et al., 2001). The HAMA consists of 14 items that are rated by a clinician using a 0 (not at all present) to 4 (severely disabling) scale; higher total scores indicate greater distress and impairment. The HADS is a self-report measure that consists of two, 7-item subscales (anxiety and depression). Patients rated each item for its frequency. For example, worrying thoughts go through my head is rated from 0 (very little) to 3 (a great deal of the time). HADS anxiety

1044

M.H. Pollack et al. / Journal of Psychiatric Research 42 (2008) 10421049

subscale scores range from 0 to 21, with higher scores indicating greater symptom severity. 2.4. Functional outcome measures

patients rated their own health state. Mean VAS values have been reported to range from 82.8 to 84.4 within the general population (Johnson et al., 2005). 2.5. Procedures

In all three studies, role functioning was assessed using the Sheehan Disability Scale (SDS; Sheehan, 1983). Patients rated how much their anxiety illness impairs their role functioning in three major life areas: work, social life, and family/home management. Each domain is evaluated using a 010 scale where 0 = not at all and 10 = extremely. The three item ratings were summed into a global functioning score to indicate overall impairment; for patients who were not working or in school, the mean value from the other two domains was inputed for the work score. The SDS was originally developed for use with anxiety patients and has demonstrated moderate to good internal reliability (Placchi, 1997). Baseline SDS global functioning impairment mean values for outpatients with GAD have been reported to range from 14.0 to 17.4, with similar mean values across each of the three individual domains (Ansseau et al., 2005; Rickels et al., 2003). In a large study of primary care patients, patients with and without psychiatric illness completed the SDS using the reference of my emotional symptoms have disrupted my life. In this study, a cuto score of ve demonstrated the optimal sensitivity and specicity for dierentiating between primary care patients who were experiencing impairment due to emotional/psychiatric symptoms and primary care patients without a psychiatric illness (Leon et al., 1997). The Quality of Life Enjoyment and Satisfaction Questionnaire Short Form (Q-LES-Q-SF; Endicott et al., 1993) was used to measure patients general satisfaction with their life. The Q-LES-Q-SF is composed of the general activities questions from the parent scale, and the SF has been shown to be a valid proxy of the longer form (Schechter et al., 2007; Rapaport et al., 2005). Patients used a 5-point scale where 1 = very poor to 5 = very good, to indicate their satisfaction with physical health, mood, work, household activities, social relationship, family relationship, leisure time activities, ability to function, sexual interest and performance, economic status, living/ housing situation, mobility, vision, and overall sense of well-being. The Q-LES-Q-SF has a total score that is derived from 14 items so the maximum score is 70. Scores are typically expressed as the percentage of the maximum score, with higher percentages indicating greater enjoyment and satisfaction. Community adult norms for the Q-LES-Q-SF have established a mean maximum percent of 83%, and the denition of normative range has been specied as scoring within 10% of the normative mean (Rapaport et al., 2005). In studies 1 and 2, the European Quality of Life 5 Dimensions (EQ-5D; EuroQol Group, 1990) was also administered to assess perception of health. Using a Visual Analogue Scale (VAS) anchored by 0 (worst imaginable health state) and 100 (best imaginable health state),

Written informed consent was obtained from each patient prior to any study procedures. Each treatment centers institutional review board approved the conduct of a study; all three studies were developed in accordance with the ethical standards of Good Clinical Practice and the Declaration of Helsinki as revised in 2000. During baseline and screening phases, patients underwent diagnostic, medical, and psychiatric evaluations. Functional outcome and quality of life measures were administered at baseline and treatment endpoint or at early discontinuation visit if necessary. 2.6. Statistical methods Data from each study were pooled at the individual patient level using the intent-to-treat sample, which consisted of patients randomly assigned to treatment with at least one baseline and one postrandomization observation. Baseline was dened as the last non-missing measurement prior to treatment randomization whereas endpoint was the last non-missing postbaseline measurement (last observation carried forward). For each quality of life measure, patients scores were classied as impaired or normative at baseline and at endpoint. Denitions for impairment or normative were based on the reported values from the literature described above, and we used the convention of being within 10% of the normative mean. Impairment on the SDS was a global functioning score of >5, normative was 65; impairment on the Q-LES-Q-SF was a score of <75% (10% less than community norm of 83%), normative was P75%; and impairment on the EQ-5D VAS was a score of <76 (10% less than population normative mean of 84), normative was P76. Patients were categorized as normative converters if their score went from the impaired range at baseline to within a normative range at endpoint, and impaired converters if they went from a normative baseline to an impaired endpoint. The proportion of patients who were categorized as impaired at baseline was compared on each measure between the duloxetine and placebo groups using a CochranMantelHaenszel test controlling for study. This analysis was repeated for comparison of proportions between treatment groups on each measure for endpoint scores, normative conversion rates, and impaired conversion rates. The relationship between ecacy and functional outcome was examined using an analysis of variance (ANOVA) with mean change in HAMA total scores from baseline to endpoint as the dependent variable, and normative conversion status (yes/no) and treatment (duloxetine/ placebo) as xed eects. This analysis was also repeated

M.H. Pollack et al. / Journal of Psychiatric Research 42 (2008) 10421049

1045

using the mean change in HADS anxiety scale from baseline to endpoint as the dependent variable. Pearson product correlations were computed between baseline ecacy (HAMA total, HADS anxiety) and quality of life measures (SDS global functioning, Q-LES-Q-SF maximum percent, and EQ-5D VAS scores). Patients were also classied per protocol as meeting remission criteria if they had a HAMA total endpoint score of 67; however, since discussion in the literature has also involved the use of a denition of 610, remission rates were also calculated using this criterion (Doyle and Pollack, 2003). Treatment-group comparisons of baseline clinical and demographic variables were examined using chi-square statistics for categorical variables and analysis of variance (with treatment and study as terms in the model) for continuous variables. All analyses were performed at twosided 0.05 signicance level. 3. Results 3.1. Patient characteristics A total of 1163 patients were randomly assigned to either duloxetine treatment (N = 668) or placebo treatment (N = 495). The mean age of the patients was 42.4 years (SD = 13.6 years) and 65% were female. At baseline, patients in the duloxetine group did not dier significantly from patients in the placebo group on their mean values for the illness severity and functional outcome measures (Table 1). Consistent with the inclusion criteria, the mean values on the HAMA and HADS anxiety were in the moderately severe range. Patients also reported similar severity of impairment across the 3 SDS domains of work, social, and family/home management. Baseline mean SDS global functioning, quality of life, and health status scores were each below community normative values.

3.2. Conversion from impaired to normative range At baseline, 89% of the patients had an SDS global functioning score of >5, with no dierence in the proportions between treatment groups (p = .419). At endpoint, a greater proportion of duloxetine-treated patients had SDS global functioning scores in the normative range than did placebo-treated patients (46.7% vs 27.7%, p 6 0.001). For the Q-LES-Q-SF maximum percent value, 95% of the sample scored <75% with no dierence in the baseline percentages between treatment groups (p = .543). At endpoint, a greater proportion of patients in the duloxetine group had Q-LES-Q-SF maximum percent scores within the community normative range compared with the placebo group (32.7% compared with 19.0%, p 6 0.001). In the subset of patients from studies 1 and 2 who were administered the EQ-5D VAS ratings and who had baseline and postbaseline observations (duloxetine, N = 459; placebo, N = 303), 76% of the sample was in the impaired range at baseline with no dierence between treatment groups. At endpoint, 58.0% of the duloxetine-treated patients scored within the normative range compared with 38.9% of the placebo-treated patients (p 6 0.001). On each measure, duloxetine-treated patients were more likely to improve from an impaired baseline score to a score within the normative range at endpoint compared with placebo-treated patients (all comparisons, p 6 0.001, Fig. 1). The percentage of patients who were impaired converters (e.g., worsened from baseline normative to endpoint impaired) was low (<7% on any measure), and the likelihood of impaired conversion did not dier between treatment groups except for EQ-5D VAS scores in which placebo-treated patients were more likely to worsen than duloxetine-treated patients (p 6 0.05). 3.3. Association between ecacy and quality of life measures At baseline, HAMA total scores were moderately correlated with SDS global functioning scores (r = 0.44, R2 = 19%, p 6 0.001), Q-LES-Q-SF scores (r = 0.45, R2 = 20%, p 6 0.001), and EQ-5D VAS scores

Table 1 Mean baseline severity of illness and quality of life scores by treatment group Baseline severity: mean, (SD) HAMA total score HADS anxiety SDS global functioning Work domain Social life Family/home management Q-LES-Q-SF Maximum % EQ-5D VAS Healtha Duloxetine (N = 668) 24.6 13.2 15.6 5.1 5.4 5.2 48.7 61.2 (7.0) (3.6) (7.3) (2.8) (2.8) (2.7) (15.2) (20.1) Placebo (N = 495) 24.8 13.2 15.8 5.0 5.4 5.3 48.3 59.9 (7.2) (3.6) (7.3) (2.7) (2.7) (2.7) (14.6) (21.2)

Key: HAMA: Hamilton Anxiety Scale; HADS: Hospital Anxiety Depression Scale; SDS: Sheehan Disability Scale; Q-LES-Q-SF Max %: Quality of Life Enjoyment and Satisfaction Scale Short Form maximum percent score; EQ-5D VAS: European Quality of Life 5 Dimensions Visual Analogue Scale score. a Administered only in Studies 1 and 2; duloxetine (N = 499) and placebo (N = 331).

Fig. 1. Percentage of patients who converted from an impaired baseline score to a normative score at endpoint by treatment group. ***p 6 0.001 Key: SDS: Sheehan Disability Scale; Q-LES-Q-SF Max %: Quality of Life Enjoyment and Satisfaction Scale Short Form maximum percent score; EQ-5D VAS: European Quality of Life 5 Dimensions Visual Analogue Scale score.

1046

M.H. Pollack et al. / Journal of Psychiatric Research 42 (2008) 10421049

(r = 0.38, R2 = 14%, p 6 0.001). Similarly, HADS anxiety scores were also correlated at baseline with the SDS global functioning score (r = 0.47, R2 = 22%, p 6 0.001), the Q-LES-Q-SF score (r = 0.42, R2 = 18%, p 6 0.001), and with the EQ-5D VAS score (r = 0.39, R2 = 15%, p 6 0.001). Within the functional outcome measures, baseline SDS global functioning was correlated with Q-LES-QSF maximum percent scores (r = 0.59, R2 = 35%, p 6 0.001) and with the EQ-5D VAS scores (r = 0.47, R2 = 22%, p 6 0.001). The Q-LES-Q-SF maximum percent score was moderately correlated with the EQ-5D VAS (r = 0.54, R2 = 29%, p 6 0.001). Patients who converted from an impaired score at baseline to a normative score by endpoint did not dier in their percent improvements on the HAMA total score based on treatment group, except for the Q-LES-Q-SF (Fig. 2). For patients who converted on the SDS, the mean HAMA total score at endpoint was 7.7; for patients whose scores normalized on the Q-LES-Q-SF maximum percent, the mean HAMA total score at endpoint was 6.9; and for patients who improved into the normal range on the EQ-5D VAS scores, the mean HAMA total score at endpoint was 9.0. These mean endpoint scores represented a HAMA improvement from baseline to endpoint of at least 60% across the three measures. Similar results were seen using the HADS anxiety subscale (Fig. 3). For patients who improved into the normative range on the SDS, the mean HADS anxiety score at endpoint was 5.3; for patients whose score converted from impaired to within normative values on the Q-LES-Q-SF, the mean HADS anxiety score at endpoint was 4.8; and for patients who improved to within normative health range on the EQ-5D VAS, the mean HADS anxiety score at endpoint was 6.1. These mean HADS anxiety endpoint scores represented an improvement of 60.6% for SDS global functioning score converters; 62.8% for Q-LES-Q-SF score converters, and 52.3% for EQ-5D VAS score converters. Overall, patients whose global functioning improved from an impaired range to within a normative range were more likely to meet HAMA remission criteria, dened as

Fig. 3. Percent improvement in HADS anxiety subscale score from baseline for patients who converted from impaired to normative score. ** p 6 0.01, duloxetine vs placebo Key: SDS: Sheehan Disability Scale; QLES-Q-SF Max %: Quality of Life Enjoyment and Satisfaction Scale Short Form maximum percent score; EQ-5D VAS: European Quality of Life 5 Dimensions Visual Analogue Scale score.

Fig. 4. Percent of patients who met HAMA remission criteria for patients who converted from impaired to normative score on the SDS global functioning score following treatment. *** p 6 0.001, converters vs. nonconverters.

either an endpoint score of 67 or 610 (Fig. 4). Approximately 74% of patients whose SDS global functioning score went from a baseline >5 to 65 at endpoint had a HAMA endpoint score of 610; the value was 55% when using a HAMA endpoint score of 67. For patients who improved to within normative values on the Q-LES-Q-SF, 79% had HAMA endpoint score of 610 and 61% had an endpoint score of 67. On the EQ-5D VAS ratings, patients whose scores improved into a normative range were also more likely to meet remission criteria, with 66% having a HAMA total score at endpoint of 610 and 53% with an endpoint score of 67. 4. Discussion Consistent with previous studies, the present work found that patients with GAD present with impaired role functioning and diminished well-being. At baseline, patients reported similar disruptions in their ability to engage in work, social relationships, and management of home and family responsibilities. Using norms derived from primary care controls (Leon et al., 1997), more than 89% of the patients had a SDS global functioning score that indicated impairment due to psychiatric illness. Similarly, patients experienced low satisfaction with their life

Fig. 2. Percent improvement in HAMA total score from baseline to endpoint for patients who converted from impaired to normative score. * p 6 0.05, duloxetine vs placebo Key: SDS: Sheehan Disability Scale; QLES-Q-SF Max %: Quality of Life Enjoyment and Satisfaction Scale Short Form Maximum Percent Score; EQ-5D VAS: European Quality of Life 5 Dimensions Visual Analogue Scale score.

M.H. Pollack et al. / Journal of Psychiatric Research 42 (2008) 10421049

1047

activities and well-being as over 95% scored below the community norms for the Q-LES-Q-SF. The EQ-5D VAS has been demonstrated to be more indicative of satisfaction with physical rather than psychological health (van de Willige et al., 2005). Given the exclusion criteria of signicant medical illness, the baseline scores reect the generally physically healthy state of the study sample. Nonetheless, 75% of the patients still fell below population normative values at baseline, indicating that patients with GAD perceive themselves as having poorer health than the general population. Patients treated with duloxetine were more likely to return to normative functioning compared with patients treated with placebo. Within these acute trials of 910 weeks, approximately 1/3 to 1/2 of the patients improved to within the community norms at endpoint. The recovery rate for duloxetine-treated patients was twice the rate seen for placebo-treated patients. Despite the low baseline scores and the chronic nature of the illness, the use of the absolute reference points from community controls indicates that the patients with GAD experienced an objective return to functioning. Further, given that GAD typically requires long-term management to achieve remission (Pollack, 2001), the improvement of scores to normative values within the acute time frame is particularly notable. Conceptually, quality of life encompasses the interplay between the emotional and behavioral consequences of an illness. Among patients treated with duloxetine, the improvement from an impaired baseline to a normative score at endpoint occurred more frequently for global functioning (43%) than for subjective satisfaction and wellbeing (31%) scores. The impact of treatment may vary by functional status. That is, role functioning may be a direct measure of illness impact whereas subjective well-being may be a measure of the interaction of illness impact with other factors; thus, well-being may require positive experience under the condition of improved role functioning and take longer to achieve (Patrick and Chiang, 2000). Support for this possibility is given by the modest baseline correlations among the dierent quality of life and functional measures, indicating that these instruments were assessing dierent dimensions of the global construct. Further, the moderate correlations between the ecacy and the quality of life measures also suggests that quality of life is not synonymous with symptomatic severity. At baseline, the shared variance between HAMA total score and each measure was approximately 20% for each measure, which is comparable with other studies (Rapaport et al., 2005). Greater emphasis has been placed in recent years on the goal of remission of illness as the primary objective in treatment rather than settling for response to treatment. Different denitions of response and remission though have been proposed (Doyle and Pollack, 2003). A recent analysis indicated that when using CGI-I ratings as the external criterion, a HAMA reduction of 42% could be dened as a treatment response, in contrast to the typical denition of a P50% reduction (Bandelow et al., 2006). The standard

denition of remission for GAD has been a HAMA total score at endpoint of 67 or 610, but these remission guidelines were derived from the depression literature and are somewhat arbitrary rather than being developed from an empirical basis (Doyle and Pollack, 2003). Remission should include not only a reduction of anxiety symptoms to a minimal or non-existent level, but should also require a return to functional status, preferably within the range of the community norm. Using these criteria, the nding of the present study indicates that HAMA total score reductions should be at least 60% from baseline to endpoint. For each measure, patients who improved from impaired to normative scores demonstrated this degree of change on both the clinician and self-report ecacy measures. When combining functional remission on the three scales with symptom remission, the distribution suggests that a HAMA total endpoint score of 610 would be more preferable, as this cut-o score captured a greater proportion of the patients who were remitted functionally (6679%) than did a cut-o score of 67 (5361%). The results of the present study need to be interpreted within the context of the selected normative values. The SDS has been used to characterize impairment in a number of disorders (Kennedy et al., 2002), but has not been administered to a large epidemiological sample. We relied upon the ndings from the large dataset of primary care patients with and without psychiatric illness, but the global functioning of non-psychiatric primary care patients may dier from the general population (Leon et al., 1997). Indeed, in establishing the clinical relevance of quality of life, there appears to be a continuum of normative values. For example, Q-LES-Q-SF scores have been compared between subjects without any history of psychiatric illness, subjects with a history of minor psychiatric illness (1 episode of brief mood disorder <16 weeks), subjects who did not meet criteria for a current illness but had a signicant history of psychiatric illness, and subjects with a current psychiatric illness (Schechter et al., 2007). Subjects who did not have a current psychiatric illness, but had a past psychiatric illness, had a higher mean Q-LES-Q score than subjects with a current psychiatric illness; however, they still scored signicantly lower than the subjects without a history of psychiatric illness or those who had only a minor previous episode. Thus, current quality of life satisfaction reects ones history of psychiatric illness and supports that the pre-morbid quality of life for GAD patients may be diminished. In the present study, we used the normative values associated with subjects without a history of psychiatric illness to provide the most stringent reference point for absolute gain. However, we did not take other types of objective measures into account, such as missed work or medical utilization, and therefore the ndings of this study should be considered to reect primarily the patients subjective perception of functioning and well-being. Another factor for consideration is that our sample was pooled from global studies and therefore included patients from both US and non-US countries. Although certain

1048

M.H. Pollack et al. / Journal of Psychiatric Research 42 (2008) 10421049

aspects of quality of life are likely to be context independent, there are also cultural inuences that aect how patients perceive their health and their preferences. Our normative values for the dierent measures were taken predominantly from US studies and therefore may not reect the community norms for individuals from other countries. A nal consideration is that our sample was composed of patients with a primary diagnosis of GAD. The high comorbidity with other psychiatric illnesses is often cited as a major factor in the impairment associated with GAD, which then leads to diculty in separating the eects of one disorder from another (Stein et al., 2005). However, in the National Comorbidity Survey, impairment with pure GAD was shown to be equivalent to impairment demonstrated by patients with pure depression (Kessler et al., 1999). In the present work, depressive and other anxiety comorbidities were excluded diagnoses, which supports that the baseline ndings reected the specic burden of GAD illness. In summary, the ndings from this large dataset indicate that patients with GAD experience signicant and substantial impairment in three domains of quality of life: global role functioning, subjective well-being, and perceived health. Using the criterion of improvement to community values, patients treated with duloxetine were more likely to achieve this outcome on each quality of life measure. Further, this criterion for improvement may help rene our denition of the appropriate cuto for remission when using standard symptom measures of anxiety. Contributions Drs. Russell and Detke directed and executed the primary clinical trials. Drs. Pollack, Endicott, Liebowitz, Russell, Detke, Spann, Ball, and Swindle developed the study concept for examining clinical relevance, outlined the design of analyses, and participated in interpretation of data. Statistical analyses were directed by Dr. Spann. All authors participated in writing team meetings and contributed to the rst draft. All authors have also read and approved the nal manuscript version. Financial disclosures Dr. Pollack has either served on the advisory board, received research grant support, or has been a member of the Speaker Bureau from the following: Bristol Myers Squibb, Cephalon, Forrest Laboratories, Glaxo Smith Kline, Janssen, Eli Lilly and Company, Novartis, Otsuka Pharmaceuticals, Pzer, Predix, Roche Laboratories, Sepracor, UCB Pharma, and Wyeth-Ayerst Pharmaceuticals. Dr. Endicott has received research support from or has served as a consultant/advisory board member for Abbott, AstraZeneca, Berlex, Bristol-Meyers Squib, Cyberonics, Eli Lilly and Company, GlaxoSmithKline, Interneuron, Merck, Parke-Davis, Novartis, Otsuaka, Janssen, Pzer, Sano-Synthelabo Research, UpJohn,

and Wyeth-Ayerst pharmaceuticals. Dr. Liebowitz has received support for being a speaker, consulting, licensing, or for clinical trials from: Avera, Astra Zeneca, Bristol Myers Squibb, Cephalon, Eli Lilly and Company, Forrest, Glaxo Smith Kline, Indevus, Novartis, Pherin, Pzer, Sano, Solvay, UCB, and Wyeth-Ayerst pharmaceuticals. Drs. Russell, Detke, Spann, Ball, and Swindle are employees and/or shareholders of Eli Lilly and Company. Source of funding This research was funded by Lilly Research Laboratories, Eli Lilly and Company. Funding support included personnel and materials involved in study design; collection, analysis and interpretation of data; and the writing of the report. Acknowledgements Data were pooled from three separate clinical trials. One study was completed prior to the requirement to post trials at initiation (ongoing as of 1-July-2005) and does not have a registration number; the other two studies have registration identication numbers: NCT00122824 and NCT00122850, at http:www.clinicaltrials.gov. Portions of this manuscript have been presented as a poster at the scientic meeting, Anxiety Disorders Association of America, St. Louis, MO (March 29April 1, 2007). We thank the principal investigators, their sta, and all of the patients who participated in these trials. We appreciate the statistical assistance and review provided by Janelle Erickson, Hong Liu-Seifert, Wenqi You, Christine Young, Nicholas Okpokho, and Steve Gelwicks. References
American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 4th ed. Text-Revision Washington, DC: American Psychiatric Press; 2000. Ansseau M, Fischler B, Dierick M, Mignon A, Leyman S. Prevalence and impact of generalized anxiety disorder and major depression in primary care in Belgium and Luxemburg: the GADIS study. European Psychiatry 2005;20:22935. Bandelow B, Baldwin DS, Dolberg OT, et al. What is the threshold for symptomatic response and remission for major depressive disorder, panic disorder, social anxiety disorder, and generalized anxiety disorder? Journal of Clinical Psychiatry 2006;67:142834. Doyle AC, Pollack MH. Establishment of remission criteria for anxiety disorders. Journal of Clinical Psychiatry 2003;64(Suppl. 15):405. Endicott J, Russell JM, Raskin JR, Detke M, Erickson J, Ball SG, et al. Duloxetine treatment for role functioning improvement in generalized anxiety disorder: three independent studies. Journal of Clinical Psychiatry 2007;68:51824. Endicott J, Nee J, Harrison W, Blumenthal R. Quality of life enjoyment and satisfaction questionnaire: a new measure. Psychopharmacology Bulletin 1993;29:3216. EuroQol Group. EuroQol: a new facility for the measurement of healthrelated quality of life. Health Policy 1990;16:199208. Guy W. The clinician global severity and impression scales. ECDEU assessment manual for psychopharmacology. Rockville, MD: US Department Health, Education, and Welfare; 1976(76-338). p. 21822.

M.H. Pollack et al. / Journal of Psychiatric Research 42 (2008) 10421049 Hamilton M. The assessment of anxiety states by rating. British Journal of Medical Psychology 1959;32:505. Hartford J, Kornstein S, Liebowitz M, et al. Duloxetine as an SNRI treatment for generalized anxiety disorder: results from a placebo- and active-controlled trial. International Clinical Psychopharmacology 2007;22:16774. Holmes S. Assessing the quality of life-reality or impossible dream? International Journal of Nursing Studies 2005;42:493501. Johnson JA, Luo N, Shaw JW, et al. Valuations of EQ-5D health states: are the United States and United Kingdom dierent? Medical Care 2005;43:2218. Kennedy BL, Lin Y, Schwab JJ. Work, social, and family disabilities of subjects with anxiety and depression. Southern Medical Journal 2002;95:14247. Kessler RC, DuPont RL, Berglund P, Wittchen HU. Impairment in pure and comorbid generalized anxiety disorder and major depression at 12 months in two national surveys. American Journal of Psychiatry 1999;156:191523. Koponen H, Allgulander C, Erickson J, et al. Ecacy of duloxetine for the treatment of generalized anxiety disorder: implications for primary care physicians. Primary Care Companion 2007;9:1007. Leon AC, Olfson M, Portera L, et al. Assessing psychiatric impairment in primary care with the Sheehan disability scale. International Journal of Psychiatry in Medicine 1997;27:93105. Lipman R, Covi L. Outpatient treatment of neurotic depression: medication and group psychotherapy. In: Spitzer RL, Klein DL, editors. Evaluation of the psychological therapies. Baltimore, MD: John Hopkins University Press; 1976. Mogotsi M, Kaminer D, Stein DJ. Quality of life in the anxiety disorders. Harvard Review of Psychiatry 2000;8:27382. Patrick DL, Chiang Y. Measurement of health outcomes in treatment eectiveness evaluations: conceptual and methodological challenges. Medical Care 2000;38(9 Suppl.):II-1425. Placchi M. Measuring disability in subjects with anxiety disorders. European Psychiatry 1997;12(Suppl. 3):249s53s. Pollack MH. Optimizing pharmacotherapy of generalized anxiety disorder to achieve remission. Journal of Clinical Psychiatry 2001;62(Suppl. 19): 205.

1049

Rapaport MH, Clary C, Fayyad R, Endicott J. Quality of life impairment in depressive and anxiety disorders. American Journal of Psychiatry 2005;162:11718. Raskin A, Schulterbrandt J, Reatig N, McKeon JJ. Replication of factors of psychopathology in interview, ward behavior and self-report ratings of hospitalized depressives. Journal of Nervous and Mental Disease 1969;148:8798. Rickels K, Zaninelli R, McCaerty J, Bellew K, Iyengar M, Sheehan D. Paroxetine treatment of generalized anxiety disorder: a double-blind, placebo-controlled study. American Journal of Psychiatry 2003;160(4):74956. Rynn MA, Russell JM, Erickson J, Detke M, Ball SG, Dinkel J, et al. Ecacy and safety of duloxetine in the treatment of generalized anxiety disorder: A exible-dose, progressive-titration, placebo-controlled trial. Depression and Anxiety 2007. Schechter D, Endicott J, Nee J. Quality of life of normal controls: association with lifetime history of mental illness. Psychiatry Research. 2007;152:4554. Shear MK, Vander Bilt J, Rucci P, et al. Reliability and validity of a structured interview guide for the Hamilton anxiety rating scale (SIGH-A). Depression and Anxiety 2001;13:16678. Sheehan DV. The anxiety disease. New York: Charles Scribner and Sons; 1983. Sheehan DV, Lecrubier Y, Sheehan KH, et al. The mini-international neuropsychiatric interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD10. Journal of Clinical Psychiatry 1998;59(Suppl. 20):3457. Stein MB, Roy-Byrne P, Craske MG, et al. Functional impact and health utility of anxiety disorders in primary care outpatients. Medical Care 2005;43:116470. van de Willige G, Wiersma D, Nienhuis F, Jenner JA. Changes in quality of life in chronic psychiatric patients: a comparison between EuroQol (EQ-5D) and WHOQoL. Quality of Life Research 2005;14:44151. Wyrwich KW, Nienaber NA, Tierney WM, Wolinksy FD. Linking clinical relevance and statistical signicance in evaluating intra-individual changes in health-related quality of life. Medical Care 1999;37:46978. Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatrica Scandinavica 1983;67:36170.