| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ALL ISSUES | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Regular Article |
Received May 30, 2002; revised August 28, 2002; accepted September 12, 2002. From Columbia University Department of Psychiatry and the New York State Psychiatric Institute. Address correspondence to Dr. Mohlman, Syracuse University, Department of Psychology, 430 Huntington Hall, Syracuse, NY 13244. e-mail: jmohlman{at}psych.syr.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Key Words: Anxiety Disorders (General) • Cognitive Therapy • Outcome Studies
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
An early study by Stanley et al.3 compared 14 90-minute sessions of group CBT with a supportive-therapy control condition (ST) for treating GAD in 48 adults age 55 and over (mean age: 68.3 years). Both interventions significantly reduced symptoms of GAD and depression. Using criteria of at least 20% reduction on 75% of outcome measures,4 the authors reported higher response rates in the ST group as compared with the CBT group at posttreatment (54% for ST, 28% for CBT) and 6-month follow-up (77% for ST, 50% for CBT).
A second study by Stanley et al.5 tested 15 90-minute sessions of group CBT against a wait-list condition in a large sample of older adults with either principal or co-principal GAD (n=85; mean age: 66.2). Results indicated significantly better outcome in the CBT group on measures of worry, anxiety, depression, and quality of life. Moreover, 45% of the CBT group and 8% of the control group were classified as responders at posttreatment. However, posttreatment scores for CBT participants failed to reach normal levels of functioning.
Wetherell et al.6 tested group CBT against a discussion group (DG) focused on topics related to aging and a wait-list control condition over a 12-week period. Participants were 75 older adults with principal diagnoses of GAD (mean age: 67.0). CBT had larger effect sizes than DG both at post-treatment and 6-month follow up, and both interventions led to significant improvements on outcome measures; one-third of participants in each active-treatment group met the criteria for responders described above.4
A recent study tested the efficacy of individual-format CBT administered either in a primary care setting or patients' own homes.7 CBT (n=5) was superior to a wait-list control condition (n=4) in reducing the frequency and severity of worry in this small sample of older medical patients with GAD. The CBT group also showed significant decreases on self-report measures of depression and health, versus no significant improvement in the wait-list group. All five participants in the CBT group attained responder status at the end of the eight sessions, as did one in the wait-list condition.
Two additional studies, one unpublished (Gorenstein EE et al.) and one published,8 found that individual CBT for mixed anxiety disorders (including GAD) was effective. However, in each of the above studies, CBT was administered in a non-traditional manner or was augmented with additional components that could account for some portion of the variance in outcome, leading to plausible alternative hypotheses. In the first three studies summarized above, a group format was used. Areán9 has argued that groups offer positive aspects that individual therapy does not; thus beneficial effects of participating in groups could have obscured the benefits of CBT relative to control conditions. This is contrary to data from younger adults, however, which indicate that individual CBT fares better than group for the treatment of GAD.10
In the study by Stanley et al.7 and in the unpublished study by Gorenstein et al., treatment was augmented with medical aspects (meeting weekly with an M.D., sessions held in primary-care clinic) not usually found in standard CBT protocols. Because health-related concerns are highly salient in GAD patients,11 it is possible that these medical components interacted with features of CBT, leading to enhanced outcome. A subset of Stanley et al.'s7 and all but one of Barrowclough et al.'s8 participants were treated in their own homes. Whereas this is standard practice in the United Kingdom, it is not currently a common format for interventions in the United States, thus limiting the generalizability of these data. Finally, a subset of Wetherell et al.'s,6 Gorenstein et al.'s, and Barrowclough et al.'s8 participants were taking antianxiety medications during CBT, making interpretation of benefits difficult.
These studies are quite valuable as initial contributions to what is known about psychosocial treatments for late-life anxiety and indicate that CBT is effective when delivered in group or other nonstandard formats. However, the efficacy of standard, individual CBT for late-life GAD administered in a mental health clinic remains unknown. Although some have argued that few older adults present to psychiatric clinics for treatment,12 it is our expectation that, with ongoing efforts to disseminate information on CBT and other empirically validated treatments, this trend may change in the future. The two pilot studies reported here are, to our knowledge, the first investigations of CBT efficacy delivered in a mental health clinic in individual format. Study 1 tested a standard version, and Study 2 tested a version of CBT that was enhanced with learning and memory aids that were designed to make the therapy more effective for elderly clients.13–15 In both studies, CBT was expected to result in significantly greater decreases on self-report measures, rates of GAD, and severity ratings at posttreatment than reported in the wait-list groups.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
The Structured Clinical Interview for DSM-IV Axis I Disorders (SCID16) was administered to confirm diagnoses and provide data on associated features, comorbidity, clinical history, and physical health. All participants met DSM-IV17 criteria for a principal diagnosis of GAD. The samples evidenced a moderate rate of comorbidity (58%; n=14, in Study 1, and 57%; n=8, in Study 2). Thirty percent of Study 1 participants had comorbid major depressive disorder (MDD) but were not in a current episode; 22% had social phobia; 21%, panic disorder; 19%, dysthymia; and 7%, specific phobia. Twenty-seven percent of Study 2 participants had comorbid dysthymia; 20%, MDD but were not in a current episode; 13%, social phobia; and 6%, panic disorder.
Exclusion criteria for Studies 1 and 2 were current use of antianxiety medications, active suicidality over the previous 6 months, acute current MDD episode, lifetime psychotic symptoms, organic brain disease, current use of medications that can cause anxiety-like symptoms (e.g., bronchodilators), or raw scores of 131 or below on the Dementia Rating Scale (DRS18), a clinician-administered measure of cognitive functioning.
A random sample of audiotaped SCID interviews (n=7) was rated by a second assessor to estimate interrater reliability. For principal diagnoses of GAD, kappa coefficients indicated excellent reliability (0.87). However, all participants completed an initial phone screen assessing the presence of GAD symptoms; therefore, this estimate may be somewhat inflated, partly because of raters' increased expectancy for GAD-positive cases.
Measures
A packet of self-report psychometric scales was administered, comprising the Beck Anxiety Inventory (BAI19), Beck Depression Inventory (BDI20), the Trait scale of the Spielberger State–Trait Anxiety Inventory–Form Y (STAI–Trait21), the Penn State Worry Questionnaire (PSWQ22), and the Revised Hopkins Symptom Checklist–90 (SCL23). All measures have been shown to have sound psychometric properties in older adults,24,25 with internal consistency coefficients ranging from 0.80 to 0.93 in Study 1 and 0.80 to 0.89 in Study 2.
Procedures
All participants read and signed a consent form and then completed the SCID interview. The DRS was administered after the SCID, and the self-report measures were completed last. The assessment lasted between 2 and 3 hours. Participants were free to take short breaks or complete the assessment over two separate visits.
After the assessment, Study 1 participants were randomly assigned to either 13 50-minute weekly sessions of CBT (n=14) followed by monthly booster sessions for 6 months or a 13-week wait-list condition (n=13). Study 2 participants were randomly assigned to either 13 50-minute sessions of enhanced CBT (ECBT; n=8), followed by monthly booster sessions for 6 months, or a 13-week wait-list condition (n=7).
Standard CBT Condition: Study 1
CBT consisted of psycho-education about GAD and mood, progressive muscle relaxation, cognitive-restructuring training, worry behavior prevention, exposure to worrisome situations, problem-solving training, daily structure exercises, and sleep hygiene. Therapists introduced one new concept or skill in each treatment session and assigned related homework each week.
After the acute-treatment phase, CBT participants attended monthly booster sessions for 6 consecutive months. Booster sessions were used to prepare patients for independent management of mood and behavior, reinforce continued use of techniques, and review material imparted in weekly sessions.
The therapist manual was a published protocol26 written by one of the study co-investigators (E.E.G.; electronic manual available upon request from the second author), and it included treatment techniques for anxiety and depression (e.g., cognitive techniques, in-vivo and imaginal exposure, self-monitoring) found in several standard CBT anxiety treatment manuals.27,28 In an earlier, unpublished study, use of this CBT manual produced significant improvement in elderly anxiety patients who were also medication users. Although all techniques were introduced to each patient, therapists had the flexibility to emphasize certain techniques over others or delay the application of a given technique. Patient workbooks were not used; however, handouts were given to illustrate key concepts. Weekly homework was assigned, but there were no specific requirements for compliance; even those who did not complete any assignments were retained in the study. However, all participants were encouraged to do as much homework as possible, which was discussed and collected each week by the therapist.
Enhanced CBT Condition: Study 2
ECBT included the same therapists' manual and modules as in Study 1, with the addition of learning and memory aids designed to 1) increase homework compliance; 2) strengthen memory for techniques; and 3) facilitate the use of these techniques. These consisted of weekly reading assignments in the "Mastery of Your Anxiety and Worry Client Workbook,"27 (meant to reinforce session material); graphing exercises in which patients charted numerical mood ratings averaged over each week (meant to highlight gradual progress, reveal patterns in moods, and facilitate discussion); mid-week homework reminder/troubleshooting phone calls from the therapist for the first four assignments (meant to alleviate patients' ambivalence about asking for help and review the procedures and goals of the assignment); and the inclusion of a homework-compliance requirement of no more than three missed assignments (meant to emphasize the importance of consistent, independent work). All ECBT participants were also asked to use a perspective-taking strategy to facilitate evidence generation in cognitive-restructuring exercises. First, each participant generated a list of three to five individuals who they believed were good problem-solvers. Then, evidence that refuted automatic thoughts was generated from the perspective of each person on the list. This technique was meant to reduce automatic negative thoughts through the enhancement of generative thinking abilities and broadened perspectives. Also, all homework assignments were photocopied and returned with ample feedback for correcting mistakes and improving CBT skills.
During the last 5 minutes of each session, the therapist led an "expanding review" of all concepts (e.g., the three-systems model of emotion) and techniques (e.g., progressive muscle relaxation) learned to date, with a brief summary of when techniques should be used (e.g., "Whenever you have a catastrophic or worrisome thought, write it down, then weigh the evidence and turn it into a less negative thought"). At the midpoint of treatment, responsibility for leading expanding review was transferred to the client. This technique was intended to strengthen memory for, and facilitate use of, CBT concepts and skills. The enhancements were derived from articles and workshop material focused on tailoring CBT to older adults or were devised by the first author.13,14
Wait-List Conditions: Studies 1 and 2
Those assigned to wait-list conditions were contacted every month by phone. Any participant who experienced a worsening of symptoms was referred to a clinic in the community for immediate treatment (n=1). All wait-list participants were invited to begin CBT immediately after the waiting period.
Posttreatment and Follow-Up Assessments
Immediately after the 13-week treatment or wait-period, participants completed a posttreatment assessment comprising the SCID interview, DRS, and psychometric scales. The posttreatment assessment lasted between 1 and 2 hours. Participants in the CBT and ECBT groups repeated this assessment immediately after the last monthly booster session.
Therapists and Assessors
Therapists in the study were two doctoral-level clinicians who had specialized training in CBT. Posttreatment assessors were doctoral-level clinicians and were unaware of participants' group assignments. All assessors had at least 2 years of formal training in administering the SCID interview.
Because the treatment protocols were flexible in Studies 1 and 2, clinicians tabulated the CBT methods used during each session on a checklist to ensure that none were omitted. An independent clinician who was unaware of these tabulations rated 10 randomly selected session tapes to assess therapist adherence. The kappa coefficient was 0.86, indicating a high degree of concordance between treating clinicians' and independent raters' tabulations. According to adherence ratings, therapists covered 89% of essential elements in the 10 randomly selected sessions.
To maintain quality of care, supervision of specific cases took place on an as-needed basis, headed by one of the senior investigators (E.E.G.). Clinicians also met weekly as a group to discuss study progress and questions regarding the administration of CBT or the assessment tools.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
There were no dropouts or participants withdrawn from either condition during the weekly treatment or follow-up phase in Study 2.
Characteristics of the Samples: Study 1 and Study 2 (Standard and Enhanced CBT Versus Wait-List)
Continuous demographic and clinical variables were compared between the groups by use of multivariate analyses of variance (MANOVA) based on strength of correlation between the variables. This resulted in two clusters of dependent variables examined in separate MANOVAs, with age, age at onset, number of minor health problems, and number of alcoholic drinks per month in the first cluster and the number of previous anxiety treatments, number of non-anxiety medications taken daily, number of comorbid diagnoses, and number of major medical problems in the second cluster. The omnibus tests failed to reach significance.
Fisher's exact tests indicated a significant difference between the groups on gender composition (p<0.05), with a higher proportion of women in the CBT group, but nonsignificant values for level of education, marital status, ethnicity, and occupational status. Results are shown in Table 1.
|
Analysis of Self-Report Measures: Study 1 Completers
Five out of the six participants who dropped out of the study did not complete posttreatment measures; therefore, outcome data are reported for completers only. Because the correlation between the PSWQ and STAI-Trait measures was 0.74, p<0.001, scores were standardized and added to produce a composite called "Trait Worry." This strategy was used to enhance statistical power in the analysis and avoid problems (e.g., statistical instability) caused by redundant outcome measures.28 Correlations on remaining outcome measures ranged from 0.28 to 0.59 and are displayed in Table 2. There were no differences between groups on any of the outcome measures at pretest. Kolmogorov-Smirnov tests of normality indicated that the distribution of data for each measure was not significantly different from normal in Studies 1 and 2; thus, Group (CBT, Wait-List) x Time (Pre-, Post-) repeated-measures analyses of variance (ANOVAs) were used to evaluate treatment effects.
|
2=0.28. (2 represents effect size, or the proportion of variance explained in the analysis.28) Means and standard deviations (SD) are shown in Table 3.
|
2=0.27) but not in the wait-list group. Using criteria from previous studies4 (free of GAD and 20% reduction on 75% of outcome measures), 40% of the CBT group and 9% of the wait-list group were classified as responders at posttreatment. Therapists' ratings of homework compliance (quantity and quality) were also assessed. The CBT group completed an average of 7 out of 12 assignments, with a mean rating of homework quality of 1.02 (SD: 0.63), on a 0 to 3 scale (0: poor, 1: fair, 2: good, 3: excellent). Means and SDs are shown in Table 3.
Analysis of Self-Report Measures: Study 2 Completers
There were no differences between the ECBT and wait-list groups on any of the outcome measures at pretest. Because the BAI, SCL Anxiety subscale, and PSWQ were correlated (0.74; p<0.005), scores on these measures were standardized and added to produce a composite outcome variable called "Anxiety and Worry." Correlations between the remaining variables ranged from 0.22 to 0.65 and are shown in Table 2.
A Group (ECBT; Wait-List) x Time (Pre-; Post-) repeated-measures ANOVA revealed a significant interaction on the Anxiety and Worry composite (F[1,10]=13.08; p<0.005; 2=0.47). The ECBT group showed a significant decrease (t[7]=3.23; p<0.02; 2=0.49), with no significant change in the wait-list group from pre- to posttreatment.
There was also a significant Group x Time interaction on the SCL Global Severity Index (GSI) subscale (F[1,10]=12.28; p<0.005; 2=0.42). Tests of simple effects indicated that the ECBT group showed a significant decrease on the subscale from pre- to posttreatment (t[7]=2.98; p<0.02; 2=0.40), whereas the wait-list group showed no significant change.
The analysis also yielded a nonsignificant interaction but significant main effect of Time on the BDI (F[1,10]=6.52; p<0.01; 2=0.66. Neither group showed significant change on the STAI-Trait. Means and SDs are shown in Table 4.
|
2=0.28). Using criteria described earlier,4 75% of the ECBT group and 14% of the wait-list group were classified as responders at posttreatment (free of GAD plus 20% decrease on 75% of outcome measure). The ECBT group showed a mean homework completion rate of 9 out of 12 assignments, with a mean quality rating of 2.1 (SD: 0.81; 0: poor, 1: fair, 2: good, 3: excellent). Results are shown in Table 4. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
ECBT participants improved significantly on two self-report measures (Anxiety and Worry composite and SCL–GSI subscale), rates of posttreatment GAD, and severity ratings, as compared with the wait-list group, who showed less improvement than ECBT on all measures. Effect sizes were slightly larger in ECBT than CBT. However, improvement on psychometric scales in both studies was somewhat modest relative to what is typically seen in younger GAD samples,1,2 suggesting either that late-life GAD is more refractory than GAD occurring in young adulthood or that CBT is not as effective in older GAD patients as it is in younger adults, or both. Because of the limitations of the current studies, additional evidence will be needed before either of these possibilities can be argued strongly.
We found increased homework compliance, based on clinicians' ratings of number and quality of completed assignments, in Study 2, as compared with Study 1. Homework compliance has shown an association with outcome in studies of CBT in younger adults.29 These results suggest that it may be worthwhile to emphasize independent work and offer assistance with out-of-session assignments when treating older adults. These were two of the main differences between our CBT and ECBT conditions. However, we also placed more emphasis on weekly changes in mood, provided detailed feedback on homework, attempted to strengthen patients' memory for CBT concepts and techniques, and attempted to enhance generative thinking via perspective-taking exercises. Thus, these modifications could also have contributed to outcome in Study 2.
It is unclear why the control group in Study 1 improved on the BDI in the absence of an intervention. Stanley et al.'s5 wait-list group also showed a significant decrease on the BDI, although not as large as the decrease in their CBT group. This effect in the current study did not appear to be due to changes in state anxiety in the wait-list group from pre- (mean: 47.71; SD: 9.50) to posttreatment (mean: 43.86; SD: 11.29). We speculate that the improvement might be attributable to fluctuations in symptoms of health problems or comorbid psychiatric conditions tapped by the measure,30 both of which were more frequent in the wait-list group than the CBT group. Statistical regression, pronounced symptoms at time of entry to the study, expectancy about upcoming treatment, or some combination thereof might also explain this effect.
Limitations of the present studies include small sample sizes, lack of minority participants, the biased gender breakdown in Study 1, use of wait-list control conditions rather than alternative therapies, and high-functioning status of the participants. Our sample also comprised a relatively high number of late-onset cases (cutoff age: 30), which may limit generalizability of findings.
|
CONCLUSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
This work was supported in part by National Institute of Mental Health grants MH01397 and R01MH53582 awarded to Dr. Papp.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  G. J. Diefenbach, D. F. Tolin, C. M. Gilliam, and S. A. Meunier Extending Cognitive-Behavioral Therapy for Late-Life Anxiety to Home Care: Program Development and Case Examples Behav Modif, September 1, 2008; 32(5): 595 - 610. [Abstract] [PDF]
|
|
|
|
 |
|
 M. Pinquart and P. R. Duberstein Treatment of Anxiety Disorders in Older Adults: A Meta-analytic Comparison of Behavioral and Pharmacological Interventions Am J Geriatr Psychiatry, August 1, 2007; 15(8): 639 - 651. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. A. Stanley, G. J. Diefenbach, and D. R. Hopko Cognitive Behavioral Treatment for Older Adults with Generalized Anxiety Disorder: A Therapist Manual for Primary Care Settings Behav Modif, January 1, 2004; 28(1): 73 - 117. [Abstract] [PDF]
|
|
|
|
|
|
P. A. Arean Advances in Psychotherapy for Mental Illness in Late Life Am J Geriatr Psychiatry, February 1, 2003; 11(1): 4 - 6. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ALL ISSUES | SEARCH | TABLE OF CONTENTS |
