Cognitive-behavioral therapy for late-life generalized anxiety disorder: Who gets better?

Cognitive-Behavioral Therapy for Late-Life Generalized Anxiety Disorder: Who Gets Better? Julie Loebach Wetherell, University of California, San Diego, and VA San Diego Healthcare System Derek R. Hopko, University of Tennessee, Knoxville Gretchen J. Diefenbach, Institute of Living, Hartford Hospital Patricia M. Averill, University of Texas-Houston Health Science Center J. Gayle Beck, State University of NewYork, Buffalo Michelle G. Craske, University of California, Los Angeles Margaret Gatz, University of Southern California Diane M. Novy, University of Texas-Houston Health Science Center Melinda A. Stanley, Baylor College of Medicine

The authors pooled data from three independently conducted treatment outcome studies to examine predictors of outcome from group-administered cognitive-behavioral therapy (CBT) for older adults with generalized anxiety disorder (GAD). Data were collected from 65 patients with a mean age of 67.7 years (SD = 6.6). Average reliable change indices (RCI) based on 3 outcome measures were calculated at posttreatment and at 6-month follow-up. Approximately half of patients achieved a significant RCI at posttreatment and two-thirds achieved a significant RCI at follow-up. Factors associated with better outcomes included better homework adherence, higher baseline GAD severity, and presence of a comorbid psychiatric diagnosis. Results suggest that at-home practice is associated with better and longer-lasting outcomes from CBT in older adults with GAD.

(GAD) may affect three times as many older adults as does major depression (Beekman et al., 1995; Beekman et al., 1998). Among older people, anxiety symptoms and disorders, including GAD, are associated with physical limitations and disability, poorer self-reported health and well-being, and increased use of medical G E N E R A L I Z E D ANXIETY DISORDER

This research was supported in part by National Institute of Mental Health grants F31 MHl1972, R01 MH53932, K23 MH67643, and the Texas Higher Education Coordinating Board

(003652-075). Address correspondence to Julie Wetherell, Department of Psychiatry, University of California, San Diego, 9500 Gilman Drive, Dept. 0603V, La Jolla, CA 92093-0603; e-mail:[email protected]. BEHAVIORTHERAPY36, 147--156, 2005

005-7894/05/0147~)15651.00/0 Copyright2005 by Associationfor Advancementof BehaviorTherapy All rights for reproductionin any form reserved.

specialists and benzodiazepine medications (De Beurs et al., 1999; Wetherell, Thorp, et al., 2004). Psychotherapeutic treatment, particularly cognitivebehavioral therapy (CBT), has been demonstrated to be effective with younger adults, and several randomized controlled trials have investigated the efficacy of group and individual CBT for late-life GAD (Gorenstein et al., in press; Mohlman et al., 2003; Stanley, Beck, & Glassco, 1996; Stanley, Beck, et al., 2003; Wetherell, Gatz, & Craske, 2003). Although these studies have shown clinically as well as statistically significant reductions in mean levels of anxiety symptoms, most participants remain in the clinical range on measures of psychopathology even after treatment. Research on psychotherapy for late-life anxiety is important because many older adults report a preference for psychotherapeutic treatment rather than medications (Wetherell, Kaplan, et al., 2004). Furthermore, benzodiazepines continue to be frequently prescribed for late-life anxiety despite associated dangers, supporting the need for empirically tested nonpharmacological interventions (Klap, Unroe, & Uniitzer, 2003). Investigations of factors predicting outcome from psychotherapy are important in order to suggest mediators or mechanisms of treatment response, which may lead to modifications to enhance the effectiveness of psychotherapy for late-life anxiety. Demographic variables, clinical variables, baseline symptomatology, and treatment factors have been associated with treatment outcome in younger adults with GAD and similar disorders. In a study of individuals diagnosed with GAD, panic disorder, or dysthymia, for example, poorer prognosis was associated with older age, earlier onset, and the presence of personality disorder or "general neu-

148

WETHERELL

rotic syndrome" (Seivewright, Tyrer, & Johnson, 1998). In a reanalysis of data from Durham and Turvey (1987), therapeutic outcome was worse for behavior therapy patients with GAD who were taking benzodiazepines, although there was no relationship between drug status and improvement for cognitive therapy patients (Wardle, 1990). Butler (1993; Butler & Anastasiades, 1988) reported better outcome for less anxious and more depressed patients in behavior therapy and CBT, although these results were not replicated in a subsequent study incorporating a similar patient sample and treatment protocol (Butler, Fennell, Robson, & Gelder, 1991). Butler (1993) also reported an inverse relation between the chronicity of GAD and positive response to psychotherapy, regardless of treatment method. Compared with analytic psychotherapy and anxiety management training, younger adults with GAD may be more responsive to cognitive therapy, particularly if they are married and have no coexisting Axis I psychological disorders (Durham, Allan, & Hackett, 1997). Interestingly, in contrast to the findings of Butler and colleagues (Butler, 1993; Butler & Anastasiades, 1988), these researchers found no relation between treatment outcome and pretreatment severity of anxiety as assessed with the Hamilton Anxiety Rating Scale (Hamilton, 1959). Finally, expectancy for improvement has also been associated with better outcome from CBT for GAD (Borkovec & Costello, 1993). Considering the equivocal findings in the younger adult literature and the absence of analogous research among older cohorts of patients with GAD, further exploration of variables associated with treatment outcome is warranted to identify factors that may predict success for anxious older adults treated with CBT. The current study represents a pooled analysis of data from three recent trials of CBT for GAD in older people (Stanley et al., 1996; Stanley, Beck, et al., 2003; Wetherell et al., 2003). We examined predictors of outcome immediately after treatment and at 6-month follow-up in 65 adults over the age of 55 who completed 12 to 15 weeks of groupadministered CBT. In addition to exploring factors that have demonstrated a positive relation to treatment outcome in younger adults with GAD (e.g., pretreatment anxiety and depression severity, comorbid Axis I disorders, marital status, expectancy, homework adherence), we also investigated variables that might be more specific to understanding treatment outcome in an older adult cohort (e.g., cognitive impairment). Because this study represents the first investigation of predictors of outcome in a sample of older adults completing psychotherapeutic treatment for

ET

AL.

anxiety, we included a wide variety of potential predictors in the analyses. To reduce the number of analyses performed, we classified predictors into four groups: demographics, clinical variables, baseline levels of anxiety and depressive symptoms, and treatment variables. Results should be interpreted as exploratory and will require independent replication.

Method PARTICIPANTS

Participants included 65 older adults with GAD who completed group CBT in independently conducted treatment outcome studies in Houston, Texas (Stanley et al., 1996, n = 18; Stanley, Beck, et al., 2003, n = 29) and Los Angeles, California (Wetherell et al., 2003, n = 18). Participants were recruited through media advertisements, hospital-affiliated health education programs, and community and religious groups (Akkerman et al., 2001; Wetherell & Gatz, 2001). Attrition rates for the CBT condition from the parent treatment studies were 26% for Stanley, Beck, et al. (2003) and 31% for both Stanley et al. (1996) and Wetherell et al. (2003). Eligible participants had a principal or co-principal diagnosis of GAD according to DSM-III-R or DSM-IV criteria (American Psychiatric Association, 1987, 1994) that was based on administration of the Anxiety Disorders Interview Schedule for DSM-III-R or DSM-IV (ADIS-R; DiNardo & Barlow, 1988; ADIS-IV; DiNardo, Brown, & Barlow, 1994). Reliability of the diagnostic interviews was adequate across all three studies, as assessed by blind review of videotapes (Stanley et al., 1996: kappa = 1.00; Stanley, Beck, et al., 2003: kappa = .78), audiotapes (Wetherell et al., 2003: kappa = .75), and multiple administrations (Stanley, Beck, et al., 2003: kappa = .60). Exclusion criteria were age under 60 (Stanley, Beck, et al., 2003) or 55 (Stanley et al., 1996, and Wetherell et al., 2003), history of mania or psychosis, cognitive impairment as indicated by a score of less than 23 (Stanley et al., 1996), 24 (Wetherell et al., 2003), or 25 (Stanley, Beck, et al., 2003) on the Mini-Mental State Examination (MMSE; Folstein, Folstein, & McHugh, 1975), current participation in psychotherapy, and current alcohol or other substance abuse. With regard to psychotropic medication, in the Texas studies, participants were asked to discontinue antianxiety or antidepressant medication under supervision of the prescribing physician at least 2 weeks prior to the initial diagnostic interview. Exceptions were made for individuals taking occasional sedative-hypnotic medication for sleep problems (e.g., less than four times per week), low levels of psychotropic medication

CBT

FOR

LATE-LIFE

for pain management, or beta-blockers for heart conditions. For the California study, individuals were excluded if they initiated pharmacological treatment within the past 2 months. Individuals stabilized on psychotropic medication for at least 2 months (i.e., five individuals, or 8% of the present sample) were asked not to change their dose or type of medication for the duration of the study. Participants included 45 women (69%) and 20 men (31%) with a mean age of 67.7 years (SD = 6.6, range = 55 to 83). They were mostly Caucasian (83%) and well-educated (mean years of education = 14.7, SD = 2.4). With respect to marital status, 55% were married, 22% were widowed, and 17% were divorced. With respect to employment status, 55% were retired, 34% were working, and the rest identified themselves as homemakers, disabled, or unemployed. The mean MMSE score was 28.7 (SD = 1.1). Mean age of GAD onset was 36.1 (SD = 26.5) and mean duration was 31.7 years (SD = 26.8). More than half (57%) had no comorbid Axis I disorders, 31% had one, 9% had two, and 3 % had three. The most c o m m o n comorbidities were depression (23%), social phobia (15%), and specific phobia (9%). ASSESSMENT

MEASURES

Anxiety. The ADIS-IV includes a clinical severity rating on a 0 (no distress or disablement) to 8 (very severe distress and disablement) scale that is assigned to each cluster of diagnostic symptoms (DiNardo et al., 1994). Interrater reliability (Pearson's r) for the GAD severity rating ranged from .85 to .87 across the three studies. The Penn State Worry Questionnaire (PSWQ; Meyer, Miller, Metzger, & Borkovec, 1990) is a 16item scale designed to assess worry severity independent of worry content. In older adults, internal consistency and test-retest reliability have been adequate (Stanley et al., 2001). Cronbach's alphas were over .80 in all three of the studies combined in the present report. The Hamilton Rating Scale for Anxiety (HAMA; Hamilton, 1959) served as a measure of clinicianrated anxiety. Previous reports have documented adequate interrater reliability, internal consistency, and discriminative validity for the H A M A in older adults with and without GAD (Beck, Stanley, & Zebb, 1999; Diefenbach et al., 2001). Interrater reliability (r) of the H A M A in the present samples ranged from .81 to .82. Depression. The Beck Depression Inventory (BDI; Beck & Steer, 1987) consists of 21 items, each of which is rated on a 4-point Likert scale. Among older adults with GAD (Snyder et al.,

149

GAD

2000), there was good support for the internal consistency of the BDI (alpha = .82) and the measure correlated highly with the Geriatric Depression Scale (r = .78; Yesavage et al., 1983). Clinician-rated depression was assessed using the Hamilton Depression Scale (HAMD; Hamilton, 1960). Other studies have documented adequate psychometric properties of the H A M D among samples of older adults with GAD (Beck et al., 1999; Diefenbach et al., 2001 ). Interrater reliability (r) in the present samples ranged from .88 to .92. PROCEDURE

Participants in all three studies received groupadministered CBT based on a treatment manual that has been tested and used extensively in younger adult samples (Craske, Barlow, & O'Leary, 1992). Duration of treatment was 12 weekly sessions in Wetherell et al. (2003), 14 weekly sessions in Stanley et al. (1996), and 15 weekly sessions in Stanley, Beck, et al. (2003). Groups in all three studies were comprised of four to six participants and one leader. Leaders were postdoctoral fellows, psychology interns, and advanced doctoral students in clinical psychology. Groups were audiotaped or videotaped for supervision and for evaluation of treatment fidelity by licensed clinical psychologists. Components of treatment were very similar across studies and included psychoeducation; monitoring of anxiety symptoms and triggers; training in progressive muscle relaxation, passive muscle relaxation, and cue-controlled relaxation; cognitive restructuring; and exposure to catastrophic thoughts and imagery with behavior modification (e.g., prevention of excessively cautious behaviors). All participants were assigned at-home practice exercises to complete between sessions. STATISTICAL

ANALYSES

Data were analyzed using SPSS release 11.5. Data were tested for outliers, normality, linearity, heteroscedasticity, and multicollinearity; no adjustments were required based on these analyses. Missing data were as follows: one participant each was missing data from the baseline PSWQ, baseline BDI, follow-up GAD severity, follow-up PSWQ, and follow-up HAMA. Two participants were missing data on expectancy. Values were not imputed to replace missing data, so sample sizes for each analysis include only those participants with complete data, and these values are reflected in the reported degrees of freedom. To examine the comparability of the three component studies, data from the studies were compared using analysis of variance and chi-square tests.

150

WETHERELL

ET

AL.

T A B L E I Demographic and Clinical Information, Baseline Levels of Psychopathology, and Treatment Variables Across Three Studies of Group CBT for Late-Life G A D

Variable

Stanley et at,, 1996 (n = 18) M (SD) %

Stanley et al., 2003 (n = 29) M (SD) %

Wetherell et al., 2003 (n = 18) M (SD) %

Age

67.9 (7.3)

66.7 (5.3)

69.1 (7.8)

Gender Women Men

33.3%a 66.7%

86.2% b 13.8%

77.8% b 22.2%

Ethnicity Caucasian African American Latino Asian

88.9% I I. I% 0.0% 0.0%

89.7% 0,0% 6.9% 3,4%

66.7% I I. I% I I, I% I I, I%

Education (years)

15.4 (2.0)

I4.7 (2.5)

13,9 (2.7)

Madtal status Married Widowed Single Divorced

66,7% I I. I% 5.6% 16.7%

48.3% 24. I% 6,9% 20.7%

55.6% 27,8% 5,6% I I. I%

Work status Employed Retired Homemaker Disabled/unemployed

27.8% 66.7% 0.0% 5.6%

44,8% 41.4% 13,8% 0,0%

22,2% 66.7% 5,6% 5,6%

Age of onset of GAD (years)

28.9 (22,0)

39,8 (26,4)

37,3 (30,4)

Duration (years)

39.0 (23.1)

26.9 (25.2)

32.0 (32.1)

Number of comorbid psychiatric diagnoses 0 I 2 3

6 I. 1% 38.9% 0.0% 0,0%

55.2% 20.7% 17.2% 6.9%

55.6% 38.9% 5.6% 0.0%

For X2

df

0.75

2, 62

15.43

2

<.001

8.44

6

,2 I

1.60

2, 62

.21

2.65

6

.85

8.31

6

.22

0.98

2, 62

.38

.26

1.15

2, 62

.32

8.23

6

.22

Any depressive diagnosis

I I. 1%a

37.9%b

I I. t %a

6.5 I

2

.04

Social phobia

I I. 1%

24. 1%

5.6%

3.30

2

.19

Specific phobia

5.6%

10.3%

I I. 1%

0.41

2

.82

Panic disorder with or without agoraphobia

5.6%

0.0%

5.6%

1.66

2

.44

Posttraumatic stress disorder

5.6%

0.0%

5.6%

1.66

2

.44

Obsessive-compulsive disorder

0.0%

3.4%

I I. I%

2.69

2

.26

Cognitive impairment (MMSE) ADIS GAD severity

28.8 (I.0)

28,5 (I,2)

29,0 (I.0)

1.53

2, 62

.23

5. I (0.8)

5.2 (I.2)

4.9 (0.8)

0.68

2, 62

.5 I

Penn State Worry Questionnaire

59.5 (I 1,5) 60.1 (9.5)

63.6 (10.3)

0.88

2, 61

.42

Hamilton Anxiety Scale

19.3 (6.3) a

20.8 (7.3) a

14.2(4.9) b

5.98

2, 62

.004

Beck Depression Inventory

14.4 (7.0)

17.9 (6.7)

15.9 (6.8)

1.49

2, 61

.24

Hamilton Depression Scale

16.7 (5.4) a

18.9 (7.2) a

10,4 (5.1)b

10.56

2, 62

<.001

Expectancy

28.8 (4.8)

28.8 (5.7)

28.3 (7.0)

0,06

2, 60

.95

Sessions attended

12.2 (I.4) a

12.6 (I.9) a

10.4 (I.7) b

9.49

2, 62

<.001

Homework adherence (%)

71,9 (26.1)

76.0 (24.8)

61.6 (32.0)

1.55

2, 62

.22

Note. MMSE = Mini Mental State Examination;ADIS = Anxiety Disorders Interview Schedule. Superscripts indicate significant pairwise differences, p < .05.

CBT

FOR

LATE-LIFE

Outcome was operationalized as the mean reliable change index (RCI) across three outcome variables used in all three studies: the GAD severity rating, PSWQ, and HAMA. The RCI represents a difference score corrected for the reliability of the measure, and values greater than 1.96 are considered to represent statistically significant change (McGlinchey, Atkins, & Jacobson, 2002). RCI is widely used as a measure of treatment gains and is preferred over classifying participants as remitted vs. improved vs. unchanged/deteriorated due to its greater statistical power as a continuous variable (Steketee & Chambless, 1992). The following values were used in the RCI formulae in the present study: GAD severity: SD = 1.0, reliability = .85; PSWQ: SD = 10.3, reliability = .81; HAMA: SD = 6.9, reliability = .81. The standard deviations represent the standard deviation of the pooled sample at baseline, and the reliability estimates represent the most conservative (i.e., lowest) of the reliability coefficients reported for each sample. Mean RCIs were calculated for change immediately after treatment and at 6-month follow-up, relative to baseline. Potential predictor variables included demographic information such as the specific study in which the participant was enrolled, gender, age, ethnicity, education, marital status, and work status (with ethnicity dummy-coded as Caucasian vs. non-Caucasian, marital status coded as married vs. not married, and work status coded as employed vs. not employed); clinical variables that included the presence of any comorbid Axis I psychiatric diagnoses, duration of GAD, and cognitive impairment as measured by the MMSE score; initial levels of anxiety and depressive symptoms as measured by baseline GAD severity, PSWQ, HAMA, BDI, and HAMD; and treatment variables such as expectancy (Borkovec & Nau, 1972), number of sessions attended, and homework adherence rates (the proportion of days on which each participant completed at least some at-home practice, as documented on forms turned in each week to group leaders). Each set of predictor variables (demographics, clinical variables, initial levels of symptom severity, and treatment variables) were entered in a block in a separate multiple regression model. The models were each run twice, once to predict mean RCI at posttreament and once to predict mean RCI at follow-up, for a total of eight multiple regression analyses.

significantly higher proportion of men than the other two studies, and the 2003 Stanley study had a significantly higher proportion of participants with comorbid depression than the other studies. Initial scores on both the H A M A and H A M D were significantly lower in the Wetherell study than in the other studies, although there were no significant differences in baseline GAD severity, PSWQ, or BDI among the samples. Participants in the Wetherell study also attended fewer sessions than participants in the Stanley studies, as would be expected based on the shorter protocol. There were no differences in age, ethnicity, education, marital status, employment status, psychiatric comorbidity (other than depression), cognitive impairment, age of GAD onset, duration, expectancy, or homework adherence rates among the samples. Because of the differences among the samples, two dummy-coded variables representing the specific study in which each participant was enrolled were included in the first block of each regression equation, prior to the entry of the other predictor variables. Power to detect a medium-sized effect for differences due to the specific study was calculated at .78. Figures 1 through 3 depict the mean scores across participants on the three outcome measures at baseline, posttreatment, and at 6-month followup. At posttreatment, the mean effect size (Cohen's d) for the three outcome measures was 1.12, and 34 patients (52%) had a mean RCI > 1.96, indicating statistically significant change. At 6-month follow-up, the mean effect size was 1.27, and 43 patients (67%) had a mean RCI > 1.96. Results of the regression models predicting mean RCI are presented in Tables 2 and 3 for posttreatment and 6-month follow-up, respectively. Demographic variables, including gender, age, ethnicity, 6

5

4

-

-

3

2

1

0

Pre

Results A statistical comparison of the three samples in terms of demographics and clinical variables is presented in Table 1. The 1996 Stanley study had a

151

GAD

Post

6-month

F I G U R E I Mean generalized anxiety disorder severity, as rated by the Anxiety Disorders Interview Schedule, in 65 older adults with GAD treated with 12 to 15 weeks of group-administered cognitive-behavioral therapy.

152

W E T H E R E L L

62 60 58 56 54 52

50 48 46 Pre

Post

6-month

F I G U R E 2 Mean Penn State Worry Questionnaire scores in 65 older adults with GAD treated with 12-15 weeks of groupadministered cognitive-behavioral therapy.

education, marital status, and employment status, did not predict mean RCI at posttreatment, F(8, 56) = 1.28, p = .27, or 6-month follow-up, F(8, 55) = 1.94, p = .07. Clinical variables, including comorbidity, duration of GAD, and cognitive impairment, did not predict mean RCI at posttreatment, F(5, 59) = 1.69, p = .15. However, the overall regression model for clinical variables was significant at 6-month follow-up, F(5, 58) = 2.55, p = .04. Variables within that model associated with significantly better treatment response at 6-month followup were enrollment in Stanley's 1996 study, beta = .342, p = .02, and the presence of a comorbid psychiatric diagnosis, beta -- .310, p = .01. Baseline severity of psychopathology did not predict mean RCI at posttreatment, F(7, 56) = 1.92, p = .08. However, the regression model including initial scores on anxiety and depressive symptoms did predict mean RCI at 6-month follow-up,

10 8 6

.....

4 2 0 Pre

Post

6-month

F I G U R E 3 Mean Hamilton Anxiety Rating Scale scores in 65 older adults with GAD treated with 12-15 weeks of groupadministered cognitive-behavioral therapy.

ET

AL.

F(7, 55) = 2.58, p = .02. The only significant variable within that model was baseline GAD severity, beta = .285, p = .04; higher levels of GAD severity were associated with better treatment response. Finally, the models containing treatment variables (expectancy, number of sessions attended, and homework completion) were significant both at posttreatment, F(5, 57) = 2.57, p = .04, and at follow-up, F(5, 56) = 2.78, p = .03. For each model, the only variable that achieved significance was the percent of days on which the participant reported completing at-home practice, beta = .291, p = .04 (posttreatment), and beta = .344, p = .02 (follow-up), with greater homework completion associated with better response.

Discussion This study used a pooled analysis of data from three independently conducted treatment outcome studies of group-administered CBT for GAD in older adults to examine predictors of change as measured by reliable change indices. Predictors examined included demographic and clinical variables, baseline severity of anxiety and depressive symptoms, and treatment variables such as expectancy, number of sessions attended, and adherence as measured by homework completion rates. Overall, findings suggest that initial severity of GAD symptoms, presence of psychiatric comorbidity, and at-home practice may predict change due to treatment with CBT for late-life GAD, especially at 6-month follow-up. The most consistent predictor of change in this sample was homework adherence. Both immediately after treatment and at 6-month follow-up, change in anxiety symptoms correlated with the proportion of days during treatment in which participants reported completing at least some athome practice. This finding is consistent with the results of a recent meta-analysis, which found a weighted mean effect size of .22 for homework compliance on therapy outcome (Kazantzis, Deane, & Ronan, 2000). These results suggest that either the mechanism of improvement in CBT for late-life GAD involves learning and practicing new skills, or that some other variable such as motivation is responsible for both homework adherence and improvement. Although we did not measure motivation per se, we did not find a relationship between the related construct of expectancy and outcome in these analyses. Given this finding, it seems appropriate that CBT protocols for older adults emphasize the importance of engaging in practice exercises. Motivational interviewing strategies may be helpful in encouraging participants to complete athome assignments.

CBT

TABLE 2

FOR

LATE-LIFE

153

GAD

Predictors of Mean Reliable Change Index at Posttreatment R

R2

AR2

F

df

Demographic variables Block I: Stanley 1996 vs. others Stanley 2003 vs. others Block 2: Gender Ethnlcity Marital status Employment status Age Education

.393

.154

.081

1.279

8, 56

Clinical variables Block I: Stanley 1996 vs. others Stanley 2003 vs. others Block 2: MMSE score Duration of GAD Psychiatric comorbidity

.354

Baseline psychopathology Block I: Stanley 1996 vs, others Stanley 2003 vs. others Block 2: ADIS GAD severity Penn State Worry Questionnaire Hamilton Anxiety Scale Beck Depression Inventory Hamilton Depression Scale

.440

Treatment variables Block I: Stanley 1996 vs. others Stanley 2003 vs. others Block 2: Expectancy Number of sessions attended Percent homework completed

.429

.125

.194

.184

.052

.141

,078

1.692

1.924

2.566

Beta

p

.328 .074 .052 -.241 .009 .021 - . 103 -.088

.273 .057 ,649 .722 .093 .944 .888 .503 .526

.306 -.009 .021 - . 197 .143

.15 I .043 .953 ,870 .120 .249

,316 ,093 ,381 ,093 ,141 ,050 -.353

.083 .063 .597 .010 .496 .490 .733 .147

.302 -.039 .109 -,064 .291

,037 .052 .808 ,379 .689 .041

5, 59

7, 56

5, 57

Note. AR2 = change in R2 due to the addition of Block 2 variablesto the model; GAD = generalizedanxiety disorder; MMSE = Mini Mental State Examination;ADIS = Anxiety Disorders Interview Schedule,

Patients with higher baseline GAD severity scores and patients with psychiatric comorbidity achieved greater levels of change at 6-month followup in this analysis. These findings are inconsistent with samples of younger adults (Butler, 1993; Butler & Anastasiades, 1988), in which lower levels of pretreatment anxiety have been associated with better outcome. However, these findings are consistent with previous findings (Butler & Anastasiades, 1988) in which higher pretreatment depressive symptoms were associated with positive outcome. It is important to note that the outcome measure used in the Butler studies was the posttreatment score on a single anxiety measure (the Leeds Anxiety score) rather than a measure of residualized or reliable change, which may limit comparability with the present study. It is possible that results in the present study represent the effect of regression to the mean, in which patients with more severe symptomatology at baseline, including psychiatric comorbidity, experience correspondingly greater

improvement than those with less severe initial psychopathology. It is also conceivable that older adults with more severe symptomatology or comorbid diagnoses may be more motivated for treatment. In either case, these results provide cause for optimism for clinicians working with more severely anxious or depressed older GAD patients. It is important to mention that most studies of CBT for GAD in younger adult samples have used individual therapy, whereas the trials included in the present investigation were of group-administered CBT. Thus, the comparison of predictors across age groups may be confounded by differences between individual and group modalities. Some investigators have argued that group mental health treatment is particularly beneficial for older adults due to loss of social networks (e.g., Are~in, 1993). A recent meta-analysis of psychotherapy trials for late-life anxiety (not limited to GAD) found no differences in effect sizes between group and individual treatments, providing further support

154

WETHERELL

TABLE 3

ET

AL.

Predictors o f Mean Reliable Change Index at 6 - M o n t h Follow-up

Demographic variables

R

R2

AR2

F

df

.469

.220

.155

1.938

8, 55

Block I: Stanley t996 vs. others Stanley 2003 vs. others Block 2: Gender Ethnicity

Beta

p .072

.446 .144 -.078 -.276

.008 .36 I .580 .048

-.051 .225

.688 . I 15

Marital status Employment status Age

-.088

.553

Education

-.203

.134

Clinical variables Block I: Stanley 1996 vs. others

.424

.180

. 115

2.546

5, 58

Stanley 2003 vs. others Block 2: HMSE score Duration o f G A D Psychiatric comorbidity Baseline psychopathology

.497

.247

.183

2.579

.342

.038 .02 I

. t 08 .078 - . 167

.463 .528 .178

.310

.0I 2

7, 55

.023

Block I: Stanley 1996 vs. others

.237

.150

Stanley 2003 vs. others Block 2: ADIS G A D severity Penn State W o r r y Questionnaire

.003 .285 .121

.986 .042 .368

.052 .047 .093

.795 .745 .692

Hamilton Anxiety Scale Beck Depression Inventory Hamilton Depression Scale Treatment variables

.446

.199

. 117

2.775

5, 56

.026

Block I: Stanley 1996 vs. others Stanley 2003 vs. others

.288 .043

.064 .790

Block 2: Expectancy N u m b e r o f sessions attended

.132 -.044

291 .784

.344

.017

Percent h o m e w o r k completed

Note. AR2 = change in R2 due to the addition of Block 2 variables to the model; GAD = generalized anxiety disorder; MMSE = Mini Mental State Examination; ADIS = Anxiety Disorders Interview Schedule.

for the use of efficient group interventions (Nordhus & Pallesen, 2003). To date, unfortunately, psychotherapy for late-life anxiety in either group or individual formats remains underutilized and often unavailable in community settings. It is hoped that studies like those included in the present analysis will serve to promote these types of interventions and ultimately make them more widely available to older adults with anxiety disorders. This study uses a pooled sample of older adults receiving group CBT for GAD across three different studies. Although pooled analyses are c o m m o n in the pharmacological, assessment, and epidemiological literatures (e.g., Andersen et al., 1999; Hopko et al., 2003; Katz, Reynolds, Alexopoulos, & Hackett, 2002), and investigators have called for the pooling of results to increase power to detect predictors of treatment outcome (Steketee & Chambless, 1992), the pooling of participants across psychotherapy studies to explore predictors of outcome raises certain methodological issues. The

comparability of the studies combined in the present analysis is limited by the fact that they were conducted by different research teams using separate research protocols. However, they used treatment protocols based on the same manual, similar inclusion and exclusion criteria, and many of the same demographic, clinical, treatment, and outcome variables. Meta-analyses of psychotherapy studies, as well as pooled analyses of pharmacotherapy studies, frequently combine more dissimilar data than were merged in the present study. Moreover, the specific treatment study in which participants were enrolled was included as a predictor in the first step of all regression analyses. Two other very important limitations are the post-hoc nature of the analyses and the relatively high number of predictors examined, especially relative to the number of participants. Because of these limitations, these results must be considered preliminary and need to be verified in independent replications. Another limitation is the fact that par-

CBT FOR

LATE-LIFE

ticipants were "young-old" and active, potentially limiting generalizability to disabled or more medically ill older adults. Additional treatment outcome research on late-life anxiety is sorely needed, particularly among the oldest old. Attrition rates in the parent studies were high, ranging from 26% to 31%; these rates are much higher than in comparable studies with younger adults and may have affected the results of the present analyses, particularly if data on certain predictors of outcome were missing due to attrition of participants who were not responding (Borkovec & Ruscio, 2001; Borkovec & Whisman, 1996; Gould et al., 2004). Finally, the lack of measures of health status or disability, which may be important factors in the development and maintenance of pathological anxiety in later life, is a major limitation of the present study. Overall, however, because this study included data on a variety of demographic, clinical, and treatment variables from a relatively large number of older adults with GAD, an important yet understudied group, it may prove useful for generating hypotheses to be tested in future research. In conclusion, the present investigation was designed to explore factors associated with a positive treatment response among older adults with GAD. Although a number of pertinent variables were identified, additional research in larger samples is necessary to clarify and replicate these findings in order to maximize benefits of CBT treatment for older adults with GAD. Additionally, subsequent studies should systematically explore potentially relevant variables that may affect treatment outcome but were not assessed in the present context, such as health and disability, social support, religiosity, and comorbidity with Axis II disorders. Additional research is also necessary to refine treatment strategies for late-life anxiety to improve response. New treatments currently under development for late-life GAD include enhanced CBT with reminders and mnemonic strategies to compensate for age-related cognitive changes, adaptations of CBT for use in the primary care setting, and protocols that include behavioral activation strategies to alleviate depressive symptoms in latelife GAD (Mohlman et al., 2003; Stanley, Hopko, et al., 2003; Wetherell et al., in press). References

Akkerman, R. L., Stanley, M. A., Averill, P. M., Novy, D. M., Snyder, A. G., & Diefenbach, G. J. (2001). Recruiting older adults with generalized anxiety disorder. Journal of Mental Health & Aging, 7, 385-394. American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington, DC: Author.

GAD

155

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Andersen, K., Launer, L. J., Dewey, M. E., Letenneur, L., Ott, A., Copeland, J. R. M., Dartigues, J.-F., Kragh-Sorensen, P., Balderschi, M., Brayne, C., Lobo, A., Martinez-Lage, J. M., Stijnen, T., Hofman, A., & EURODEM Incidence Research Group. (1999). Gender differences in the incidence of AD and vascular dementia. Neurology, 53, 1992-1997. Arefin, P. A. (1993). Cognitive behavioral therapy with older adults, the Behavior Therapist, 16, 236-239. Beck, A. T., & Steer, R. A. (1987). Beck Depression Inventory: Manual. San Antonio, TX: The Psychiatric Corporation. Beck, J. G., Stanley, M. A., & Zebb, B. J. (1999). Effectiveness of the Hamilton Anxiety Scale with older generalized anxiety disorder patients. Journal of Clinical Geropsychology, 5, 281-290. Beekman, A. T. E, Bremmer, M. A., Deeg, D. J. H., van Balkom, A. J. L. M., Smit, J. H., de Beurs, E., van Dyck, R., & van Tilburg, W. (1998). Anxiety disorders in later life: A report from the Longitudinal Aging Study Amsterdam. International Journal of Geriatric Psychiatry, 13,717-726. Beekman, A. T. E, Deeg, D. J. H., van Tilburg, T., Smit, J. H., Hooijer, C., & van Tilburg, W. (1995). Major and minor depression in later life: A study of prevalence and risk factors. Journal of Affective Disorders, 36, 65-75. Borkovec, T. D., & Costello, E. (1993). Efficacy of applied relaxation and cognitive-behavioral therapy in the treatment of generalized anxiety disorder. Journal of Consulting

and Clinical Psychology, 61, 611-619. Borkovec, T. D., & Nau, S. D. (1972). Credibility of analogue therapy rationales. Journal of Behaviour Therapy and

Experimental Psychiatry, 3,257-260. Borkovec, T. D., & Ruscio, A. M. (2001). Psychotherapy for generalized anxiety disorder. Journal of Clinical Psychiatry, 62(Suppl. 11), 37-42. Borkovec, T. D., & Whisman, M. A. (1996). Psychosocial treatment for generalized anxiety disorder. In M. R. Mavissakalian & R. F. Prien (Eds.), Long-term treatment of anxiety disorders (pp. 171-199). Washington, DC: American Psychiatric Press. Butler, G. (1993). Predicting outcome after treatment for generalised anxiety disorder. Behaviour Research and Therapy,

31,211-213. Butler, G., & Anastasiades, P. (1988). Predicting response to anxiety management in patients with generalised anxiety disorders. Behaviour Research & Therapy, 26, 531-534. Butler, G., Fennell, M., Robson, P., & Gelder, M. (1991). Comparison of behavior therapy and cognitive behavior therapy in the treatment of generalized anxiety disorder. Journal of Consulting and Clinical Psychology, 59, 167-175. Craske, M. G., Barlow, D. H., & O'Leary, T. A. (1992). Mastery of your anxiety and worry. Albany, NY: Graywind Publications. De Beurs, E., Beekman, A. T. E, van Balkom, A. J. L. M., Deeg, D. J. H., van Dyck, R., & van Tilburg, W. (1999). Consequences of anxiety in older persons: Its effect on disability, well-being and use of health services. Psychological Medicine, 29, 583-593. Diefenbach, G. J., Stanley, M. A., Beck, J. G., Novy, D. M., & Averill, P. M. (2001). Examination of the Hamilton Scales in assessment of anxious older adults: A replication and extension. Journal of Psychopathology and Behavioral Assessment, 23, 117-124. DiNardo, P. A., & Barlow, D. H. (1988). Anxiety Disorders Interview Schedule--Revised. Albany, NY: Center for Stress and Anxiety Related Disorders, State University of New York.

156

WETHERELL

DiNardo, P. A., Brown, T. A., & Barlow, D. H. (1994). Anxiety Disorders Interview Schedule for DSM-IV. Boston: Center for Stress and Anxiety Related Disorders, Boston University. Durham, R. C., Allan, T., & Hackett, C. A. (1997). On predicting improvement and relapse in generalized anxiety disorder following psychotherapy. British Journal of Clinical Psychology, 36, 101-119. Durham, R. C., & Turvey, A. A. (1987). Cognitive therapy vs behaviour therapy in the treatment of chronic general anxiety. Behaviour Research and Therapy, 25, 229-234. Folstein, M. E, Folstein, S. E., & McHugh, P. R. (1975). "Mini-mental state": A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatry Research, 12, 189-198. Gorenstein, E. E., Papp, L. A., Kleber, M. S., Mohlman, J., DeJesus, M., & Gorman, J. M. (in press). Cognitive-behavioral therapy for management of anxiety and medication taper in older adults. American Journal of Geriatric Psychiatry. Gould, R. A., Safren, S. A., Washington, D. O., & Otto, M. W. (2004). A meta-analytic review of cognitive-behavioral treatments. In R. G. Heimberg, C. L. Turk, & D. S. Mennin (Eds.), Generalized anxiety disorder: Advances in research and practice (pp. 248-264). New York: The Guilford Press. Hamilton, M. (1959). The assessment of anxiety state by rating. British Journal of Medical Psychiatry, 32, 50-55. Hamilton, M. (1960). A rating scale for depression. Journal of Neurology, Neurosurgery, & Psychiatry, 23, 56-62. Hopko, D. R., Stanley, M. A., Reas, D. L., Wetherell, J. L., Beck, J. G., Novy, D. M., & Averill, E M. (2003). Assessing worry in older adults: Confirmatory factor analysis of the Penn State Worry Questionnaire and psychometric properties of an abbreviated model. PsychologicalAssessment, 15, 173-183. Katz, I. R., Reynolds, C. F. III, Alexopoulos, G. S., & Hackett, D. (2002). Venlafaxine ER as a treatment for generalized anxiety disorder in older adults: Pooled analysis of five randomized placebo-controlled clinical trials. Journal of the American Geriatrics Society, 50, 18-25. Katzantzis, N., Deane, F. P., & Ronan, K. R. (2000). Homework assignments in cognitive and behavioral therapy: A meta-analysis. Clinical Psychology: Science and Practice, 7, 189-202. Klap, R., Unroe, K. T., & Uniitzer, J. (2003). Caring for mental illness in the United States: A focus on older adults. American Journal of Geriatric Psychiatry, 11, 517-524. McGlinchey, J. B., Atkins, D. C., & Jacobson, N. S. (2002). Clinical significance methods: Which one to use and how useful are they? Behavior Therapy, 33, 529-550. Meyer, T. J., Miller, M. L., Metzger, R. L., & Borkovec, T. D. (1990). Development and validation of the Penn State Worry Questionnaire. Behaviour Research and Therapy, 28, 487-495. Mohlman, J., Gorenstein, E. E., Kleber, M., de Jesus, M., Gorman, J. M., & Papp, L. A. (2003). Standard and enhanced cognitive-behavior therapy for late-life generalized anxiety disorder: Two pilot investigations. American Journal of Geriatric Psychiatry, 11, 24-32. Nordhus, I. H., & Pallesen, S. (2003). Psychological treatment

ET A L . of late-life anxiety: An empirical review. Journal of Consult-

ing and Clinical Psychology, 71,643-651. Seivewright, H., Tyrer, P., & Johnson, T. (1998). Prediction of outcome in neurotic disorder: A 5-year prospective study. Psychological Medicine, 28, 1149-1157. Snyder, A. G., Stanley, M. A., Novy, D. M., Averill, P. M., & Beck, J. G. (2000). Measures of depression in older adults with generalized anxiety disorder: A psychometric evaluation. Depression and Anxiety, 11, 114-120. Stanley, M. A., Beck, J. G., & Glassco, J. D. (1996). Treatment of generalized anxiety in older adults: A preliminary comparison of cognitive-behavioral and supportive approaches. Behavior Therapy, 27, 565-581. Stanley, M. A., Beck, J. G., Novy, D. M., Averill, P. M., Swarm, A. C., Diefenbach, G. J., & Hopko, D. R. (2003). Cognitivebehavioral treatment of late-life generalized anxiety disorder. Journal of Consulting and ClinicalPsychology, 71,309-319. Stanley, M. A., Hopko, D. R., Diefenbach, G. J., Bourland, S. L., Rodriguez, H., & Wagener, E (2003). Cognitive-behavior therapy for late-life generalized anxiety disorder in primary care: Preliminary findings. American Journal of Geriatric Psychiatry, 11, 92-96. Stanley, M. A., Novy, D. M., Bourland, S. L., Beck, J. G., & Averill, P. M. (2001). Assessing older adults with generalized anxiety: A replication and extension. Behaviour Research and Therapy, 39, 221-235. Steketee, G., & Chambless, D. L. (1992). Methodological issues in prediction of treatment outcome. Clinical Psychol-

ogy Review, 12,387-400. Wardle, J. (1990). Behaviour therapy and benzodiazepines: Allies or antagonists? British Journal of Psychiatry, 156, 163-168. Wetherell, J. L., & Gatz, M. (2001). Recruiting anxious older adults for a psychotherapy outcome study. Journal of Clinical Geropsychology, 7, 29-38. Wetherell, J. L., Gatz, M., & Craske, M. G. (2003). Treatment of generalized anxiety disorder in older adults. Journal of Consulting and Clinical Psychology, 71, 31-40. Wetherell, J. L., Kaplan, R. M., Kallenberg, G., Dresselhaus, T. R., Sieber, W. J., & Lang, A. J. (2004). Mental health treatment preferences of older and younger primary care patients. International Journal of Psychiatry in Medicine, 34, 219-233. Wetherell, J. L., Sorrell, L. T., Thorp, S. R., & Patterson, T. L. (in press). Psychological interventions for late-life anxiety: A review and early lessons from the CALM study. Journal

of Geriatric Psychiatry and Neurology. Wetherell, J. L., Thorp, S. R., Patterson, T. L., Golshan, S., Jeste, D. V., & Gatz, M. (2004). Quality of life in geriatric generalized anxiety disorder: A preliminary investigation. Journal of Psychiatric Research, 38, 305-312. Yesavage, J. A., Brink, T. L., Rose, T. L., Lure, O., Huang, V., Adey, M., & Leirer, V. O. (1983). Development and validation of a Geriatric Depression Screening Scale: A preliminary report. Journal of Psychiatric Research, 17, 37-49. RECEIVED; December 15, 2003 ACCEPTED: June 21, 2004