Journal of Consulting and Clinical Psychology Copyright 1993 by the American Psychological Association, Inc.

1993, Vol. 61, No. 6, 1003-1010 0022-006X/93/S3.00

SPECIAL POPULATIONS

Comparative Effectiveness of Social Problem-Solving Therapy and

Reminiscence Therapy as Treatments for Depression in Older Adults
Patricia A. Arean, Michael G. Perri, Arthur M. Nezu, Rebecca L. Schein,
Frima Christopher, and Thomas X. Joseph

Compared the effects of 2 psychotherapies based on divergent conceptualizations of depression in

This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

later life. Seventy-five older adults diagnosed with major depressive disorder were assigned randomly
This document is copyrighted by the American Psychological Association or one of its allied publishers.

to problem-solving therapy (PST), reminiscence therapy (RT), or a waiting-list control (WLC) con-
dition. Participants in PST and RT were provided with 12 weekly sessions of group treatment. De-
pendent measures, taken at baseline, posttreatment, and 3-month follow-up, included self-report
and observer-based assessments of depressive symptomatology. At posttreatment, both the PST and
the RT conditions produced significant reductions in depressive symptoms, compared with the WLC
group, and PST participants experienced significantly less depression than RT subjects. Moreover, a
significantly greater proportion of participants in PST versus RT demonstrated sufficient positive
change to warrant classification of their depression as improved or in remission at the posttreatment
and follow-up evaluations.

Depression constitutes the most common emotional disorder on the benefits of psychological treatments for depression in the
found in older people (Butler, Lewis, & Sunderland, 1991). Es- elderly (Adler, 1992).
timates of the prevalence of major depressive disorder in the Various psychological conceptualizations have been proposed
elderly range from 2% to 10% (Blazer, Hughes, & George, 1987), to explain and to treat depression in the elderly. These range
with milder forms of depression such as dysthymia and dyspho- from the developmental-existential perspective of reminiscence
ria affecting 20% to 30% of older adults (Butler et al., 1991). therapy (RT; Butler, 1974) to the social reinforcement formula-
Moreover, the clinical significance of depression in the elderly is tion of behavior therapy (Teri & Lewinsohn, 1982). Although
underscored by the consistent finding that suicide occurs more the research literature on psychotherapy for depression in the
frequently in the elderly than in any other age group (Rich, elderly is not extensive, several controlled studies have sup-
Young, & Fowler, 1986). ported the effectiveness of various psychosocial treatments, in-
Recently, a consensus development panel of the National In- cluding RT (e.g., Goldwasser, Auerbach, & Harkins, 1987), psy-
stitutes of Health (NIH; 1992) urged vigorous treatment of de- chodynamic psychotherapy (e.g., Steuer et al., 1984; Thomp-
pressed elderly with somatic therapies (i.e., drugs, electrocon- son, Gallagher, & Breckenridge, 1987), behavior therapy (e.g.,
vulsive therapy [ECT], or both); psychosocial therapies were Brand & Clingempeel, 1992; Gallagher & Thompson, 1982),
recommended as secondary or supplemental interventions. In and cognitive therapy (e.g., Beutler et al., 1987; Steuer et al.,
justifying these recommendations, members of the NIH panel 1984). Further research is warranted by the scope and serious-
noted that, in comparison with the substantial research base ness of depression in the elderly and by limitations of the exist-
supportive of somatic therapies (for a review, see Gerson, Plot- ing research base (i.e., few studies, small samples, failure to in-
kin, & Jarvik, 1988), there exists a relative dearth of research clude measures specific to the nature of depression in the el-
derly, and an overreliance on self-reports as outcome measures).
Research on psychological interventions is also needed because
Patricia A. Arean, Department of Psychiatry, University of Califor- somatic therapies are contraindicated in the treatment of many
nia, San Francisco; Michael G. Perri, Department of Clinical and older adults, particularly those who are medically ill and cannot
Health Psychology, University of Florida; Arthur M. Nezu, Department tolerate the side effects of antidepressant medications or ECT
of Mental Health Sciences, Hahnemann University; Rebecca L. Schein, (Butler etal., 1991; Winstead, Mielke, & O'Neill, 1990).
Center for Aging, Duke University; Frima Christopher, Goldwater Me- Because age-related psychosocial factors often contribute to
morial Hospital, New \brk; and Thomas X. Joseph, Department of Psy- the occurrence of depression in the elderly (Ruegg, Zisook, &
chology, Fairleigh Dickinson University. Swendlow, 1988), successful treatment (and prevention of re-
We are grateful to Christopher Capuano, Daniel Mahoney, and Mi-
lapse) may require modification of those psychological factors
chael Petronko for their helpful comments regarding the design and
completion of this study. that are etiologically related to depression in later life. Nezu and
Correspondence concerning this article should be addressed to Patri- his colleagues (Nezu, 1987; Nezu, Nezu, & Perri, 1989) have
cia A. Arean, Department of Psychiatry, University of California, San recently articulated a problem-solving model of unipolar de-
Francisco, 401 Parnassus Avenue—CPT, San Francisco, California pression that may hold particular relevance for understanding
94143-0984. and treating depression in older adults. Within this formula-
1003
 1004 AREAN, PERRI, NEZU, SCHEIN, CHRISTOPHER, JOSEPH

tion, social problem-solving encompasses the processes by Method

which people develop effective means of coping with stressful
life events; deficits in problem-solving skill serve as one impor- Subjects
tant vulnerability factor for depression. When deficits in prob- Announcements of a university-sponsored depression program for
lem-solving lead to ineffective coping attempts under high levels adults over the age of 55 years were placed in community newspapers,
of stress (emanating either from major negative life events or churches, synagogues, and senior citizen centers. Interested individuals
from continuous daily problems), depression is likely to ensue were required to complete the Beck Depression Inventory (BDI; Beck,
(Nezuetal., 1989). Ward, Mendelson, Mock, & Erbaugh, 1961) and the Geriatric Depres-
Many of the changes that occur in later life often constitute sion Scale (GDS; Yesavitch et al., 1983). Individuals with scores of 20
significant stressors. For example, declining socioeconomic sta- or higher on the BDI and 10 or higher on the GDS were invited to par-
ticipate in a 1.5-hr semistructured clinical intake interview (scheduled
tus, deteriorating physical health, and the loss of loved ones can
2 weeks later) and to complete a battery of self-report inventories, in-
each have a powerful impact on overall mood and self-esteem cluding a second BDI and a second GDS. Intake interviewers were con-
(Butler et al., 1991; Ruegg et al., 1988). Whether such stressors ducted by advanced clinical psychology graduate students who had been
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

precipitate a major depression may be determined, in part, by trained to criterion in the use the Schedule of Affective Disorders and
This document is copyrighted by the American Psychological Association or one of its allied publishers.

the individual's ability to cope effectively with the major and Schizophrenia (SADS; Endicott & Spitzer, 1978). In addition to diag-
minor problems posed by these life changes (Lazarus, 1991; nostic decisions based on the Research Diagnostic Criteria (RDC;
Nezu et al., 1989). Indeed, some recent research has found that Spitzer, Endicott, & Robins, 1978), the interviewers also completed the
depressed older adults show deficits in social problem-solving 17-item version of the Hamilton Rating Scale for Depression (HRSD;
ability—deficiencies that may impair one's ability to cope with Hamilton, 1960) for each subject. All interviews were recorded on au-
diotapes and were rated by an independent assessor who was unaware
stressors related to depression (e.g., Fry, 1989).
of the interviewee's treatment status.
One corollary of the conceptual model proposed by Nezu et Criteria for inclusion in the study involved (a) meeting the RDC re-
al. (1989) suggests that problem-solving training will lead to de- quirements for a diagnosis of unipolar, major depressive disorder; (b)
creases in depressive symptomatology. Two outcome studies scores of 20 or greater on the BDI and 10 or greater on the GDS on both
(Nezu, 1986; Nezu & Perri, 1989) have provided a direct test of the screening and pretreatment evaluations; (c) HRSD scores of 18 and
this hypothesis and have demonstrated the clinical effectiveness above; and (d) meeting the age of 55 years or older, not exceeding 80
of problem-solving therapy (PST) in middle-aged adults with years. Exclusionary criteria included a diagnosis of bipolar disorder,
unipolar depression. Only one investigation (Hussian & Law- psychosis, dementia, depression secondary to a physical disorder (e.g.,
hypothyroidism; not simply a reaction to physical illness), borderline or
rence, 1981) has directly tested the effectiveness of PST in de-
antisocial personality disorder, active substance, or current involvement
pressed older adults. Hussian and Lawrence found that PST was in psychological or pharmacological treatment for depression.
superior to a social reinforcement approach for reducing de- A total of 156 people responded to the program announcements.
pression in institutionalized older adults, but the findings were Each was contacted by telephone and was provided with information
limited by an exclusive reliance on self-reports as dependent about the nature of the treatment and the requirements of the study;
measures. Although the results of these studies provide initial 16 felt that the treatment would not meet their needs; the other 140
support of the problem-solving model of depression, additional individuals were sent a packet of materials that included detailed infor-
research is needed regarding the effectiveness of PST as a treat- mation about the study, an informed consent statement, a demographic
ment for depression in older adults and as compared with other questionnaire, a BDI, and a GDS. One hundred fifteen people com-
pleted and returned the consent statement and questionnaires, and 105
treatments (e.g., RT) derived from alternative conceptualiza- of them met the requirements for face-to-face interviews. The 25 indi-
tions of depression in the elderly. viduals who did not return materials were contacted a second time to
RT, a commonly recommended psychotherapy for older ensure that they had received the study packet: 6 indicated that they
adults, is based on the premise that life review constitutes a nor- were no longer interested in participating in the program; 12 reported
mal developmental process brought about by increasing aware- receipt of the materials but failed to return them, even after repeated
ness of one's mortality (Butler, 1974). A failure to successfully reminders; and 7 did not return our phone calls.
integrate one's life experiences is viewed as contributing to de- Of the 105 people offered an interview, 93 appeared for their sched-
uled interview, and 75 of them met the criteria for major depressive
spair and depression (Erikson, Erikson, & Kinvick, 1986). RT
disorder. Telephone contacts with the 12 individuals who failed to show
entails a progressive return to an awareness of past experiences, for scheduled interviews indicated that 5 had sought treatment else-
both successful and unsuccessful, so that salient life experiences where, 4 were "feeling better," and 3 did not return our calls. Among
may be reexamined and reintegrated. The life review process the 18 people who were interviewed but excluded from participation, 1
gives older people opportunities to place their accomplishments met the criteria for bipolar disorder, 2 showed obvious signs of demen-
in perspective, to resolve lingering conflicts, and to find new sig- tia, and the remaining 15 showed some signs or symptoms of depression
nificance and meaning in their lives, thereby relieving the de- but did not meet all the criteria for a diagnosis of major depressive dis-
spair and depression that often accompany aging (Butler et al., order. The individuals not accepted into the program were provided
1991). Support for the effectiveness of RT as a treatment for with referral information to local mental health facilities. The demo-
graphic characteristics of the 75 individuals accepted for participation
depression in older people has been found in several studies in the study are summarized in Table 1.
(Goldwasser et al., 1987; Rattenberg& Stones, 1989).
Thus, in the present study, the comparative efficacy of two
psychotherapies for the treatment of depression in older adults Measures
was examined. We conducted a randomized, prospective inves- Depression. Severity of depression was assessed through both observ-
tigation to evaluate the effectiveness of PST and RT, compared er-based and self-report measures of depression. Observer-based mea-
with each other and to a waiting-list control (WLC) condition. sures of depression were obtained through a structured interview using
 DEPRESSION IN OLDER ADULTS 1005

Table 1 conducted within a group format with one of three therapists. Each of
Baseline Demographic Characteristics of Participants the three therapists led one PST group and one RT group. Each treat-
ment was implemented over 12 weekly sessions, with each session last-
PST RT WLC ing approximately 1.5 hr. The therapists, who were advanced graduate
Characteristic (n = 28) (n = 27) (n = 20) students in clinical psychology, included one Hispanic woman, one
White woman, and one man of Indian descent; their ages were 28, 26,
Mean age (and SD) 67.0 (7.5) 66.7 (8.5) 65.5 (6.3) and 26 years, respectively. Their self-identified theoretical orientations
Mean months depressed were cognitive-behavioral, eclectic, and psychodynamic. All three ther-
(andSD) 11.6 (5.7) 12.0 (4.8) 11.0 (6.2)
apists had previous experience in group psychotherapy with depressed
PST RT WLC patients in general, and all were trained to criterion independently in
both treatment approaches. Each of the two treatment conditions was
Gender % n % n % n carried out in accord with a detailed manual describing the theoretical
% Male (and n) 21 6 29 8 25 5 underpinnings of the approach, the general strategies involved, the ma-
% Female (and n) 79 22 70 19 75 15 jor techniques that could be used, and suggestions for dealing with spe-
Ethnic group cific problems. Weekly supervision of the therapists regarding the im-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

% African American
plementation of PST and RT was provided independently by two li-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

(andn) 18 5 15 4 15 3
% Hispanic (and n ) 7 2 7 2 5 1 censed clinical psychologists, one experienced in PST (M. G. Perri), the
% Middle Eastern other proficient in the use of RT (F. Christopher).
(andn) 0 0 4 1 0 0 PST. This program was adapted from the problem-solving training
% White (and n) 75 21 77 21 80 16 procedures detailed by Nezu et al. (1989). PST began with intensive
Employment training in developing an appropriate orientation to coping with depres-
% Full time (and«) 25 7 22 6 15 3 sion and the problems associated with it (Nezu & Perri, 1989). Training
% Part time (and n) 18 5 19 5 20 4 in problem orientation was geared to provide participants with a ra-
% Retired (and n) 50 14 55 15 60 12 tional, positive, and constructive set toward problems in living and to
% Unemployed have participants view problem solving as a means of coping with the
(and«) 7 2 4 1 0 0
Marital status current stressors in their lives. Participants were taught to label emo-
% Single (and n ) 7 2 7 2 10 2 tions as cues for identifying the existence of a problem, to inhibit the
% Married (and n) 54 15 63 17 50 10 tendency to respond automatically to problems, and to engage instead
% Divorced/separated in the problem-solving process. Training in the remaining component
(andn) 18 5 19 5 25 5 skills of problem solving involved teaching participants (a) to better
% Widowed (and n) 21 6 15 4 15 3 define and formulate the nature of problems, (b) to generate a wide
range of alternative solutions, (c) to systematically evaluate the potential
Note. PST = problem-solving therapy; RT = reminiscence therapy; consequences of a solution and select the optimal ones to implement,
WLC = waiting-list control. and (d) to monitor and evaluate the actual solution outcome after its
implementation.
the SADS to generate HRSD ratings and RDC diagnostic classifica- RT. This program was adapted from the RT procedures described
tions. At posttreatment and follow-up, interviews were scheduled by a by Matteson (1984) and from Butler's (1974) recommendations for the
third party, and subjects were asked not to disclose whether they had adaptive use of reminiscence in life review therapy. Treatment was fo-
been in treatment or the type of treatment they had received. Interrater cused on specific themes derived from Erikson's conceptualization of
reliability for the SADS was .91 at pretreatment, .95 at posttreatment, later life as a time when individuals must grapple with a psychosocial
and .95 at follow-up. Interrater reliability for the HRSD was .87 at pre- crisis involving "integrity versus despair" (Erikson et al., 1986, p. 54).
treatment, .84 at posttreatment, and .84 at follow-up. Self-report mea- The goal of treatment was to have participants review their life histories
sures included the BDI (Beck et al., 1961), a 21-item self-report mea- so as to gain a greater sense of perspective and satisfaction with what
sure, and the CDS (Yesavitch et al., 1983), a 30-item self-report mea- they had and had not achieved during their lives. Weekly topics were
sure of depression in older adults. Psychometric evaluations of the BDI used to guide each participant through a life history review and to stim-
(Gallagher, Breckenridge, Steinmetz, & Thompson, 1983) and the GDS ulate discussion of the major positive and negative events in their lives.
(Dunn & Sacco, 1989) have shown these instruments to be reliable and Reminiscences were directed toward discussion and interpretation of
valid measures of depression in the elderly. how past events were similar or dissimilar to the problems that the par-
Social problem-solving ability. Ability to solve problems was mea- ticipants were currently facing. Discussion also included an examina-
sured using the Social Problem-Solving Inventory (SPSI; D'Zurilla & tion of past goals, lingering regrets, and future plans. The major objec-
Nezu, 1990), a 70-item self-report measure of problem-solving ability tives of the review process were (a) to facilitate acceptance of one's life
that yields scale scores for five component processes of problem solving: with both its successes and shortcomings, (b) to enhance resolution of
problem orientation; problem definition and formulation; generation unresolved conflicts, and (c) to encourage participants to pursue future
of alternative solutions; decision making; and solution implementation goals that would enhance the meaning of their lives.
and verification. D'Zurilla and Nezu have provided psychometric data WLC. Subjects in this condition were told that the program, because
supporting the reliability and validity of the SPSI. of limited capacity, was unable to accommodate any more members but
Integration of life events. The Life Integration Scale (LIS; Ryff& that at the end of 12 weeks they would be able to receive treatment.
Heinke, 1983) is a 64-item self-report measure that yields scale scores Each WLC member was contacted twice during the waiting period to
indicating degree of integrity (one's level of life integration), interiority assess the need for referral for immediate treatment as well as to provide
(one's ability to introspect), generativity (concerns of instructing a assurance that treatment would be available at the end of the waiting
younger generation), and complexity (rate of engagement in work and period. Subjects in the WLC group were provided with treatment after
activities). Ryff and Heinke have reported psychometric data supporting 12 weeks and were not evaluated at the 3-month follow-up assessment.
the reliability and validity of the LIS. Results
Procedure Preliminary Analyses
The 75 subjects were assigned randomly to one of three conditions: Before evaluating treatment outcome, we conducted a series
PST (n = 28); RT (n = 27); and a WLC group (n = 20). Treatment was of preliminary analyses geared to address various validity is-
 1006 AREAN, PERRI, NEZU, SCHEIN, CHRISTOPHER, JOSEPH

sues. First, a series of three one-way multivariate analyses of moved to another location (1 in PST, 1 in RT); death in family
variance (MANOVAs) were conducted to determine possible (0 in PST, 1 in RT); conflict with another group participant (1
differences among initial levels of the dependent variables in PST, 0 in RT); discomfort in being the "only male subject in
across the three experimental conditions. One MANOVA in- the group" (1 in PST, 0 in RT); and reason unknown (1 in PST,
cluded the three depression measures, a second incorporated 0 in RT). Analyses indicated that subjects who dropped out did
the five scales of the SPSI, and the third comprised the four not differ from those who completed the program on any of the
scales of the LIS. No between-group differences were found to pretreatment measures (all ps > .10). Moreover, the rates for
be significant (all ps > .60). Additional ANOVAs and chi-square attrition and reasons for dropping out did not vary according to
tests were conducted regarding the various demographic vari- treatment condition. All 20 subjects in the WLC group returned
ables listed in Table 1. Once again, no significant between- for the posttreatment assessment.
group differences emerged. On the basis of these results, we con-
cluded that our randomization procedure resulted in compara- Evaluation of Differential Treatment Effects
ble groups of subjects.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

A second validity check involved possible differences in out- To evaluate overall and relative effects of the differing treat-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

come that may be attributable to subjects' perceptions of treat- ment approaches, we used a general statistical analytic strategy
ment efficacy, therapist competency, or both. To that end, sub- that incorporated a series of 3 X 2 (Condition X Pre- versus
jects in the PST and RT conditions were requested to anony- Posttreatment Analysis) repeated measures MANOVAs. One
mously complete a questionnaire at the conclusion of both the repeated measures MANOVA included the three measures of
1 st and 12th treatment sessions. Specifically, they were asked to depression (BDI, GDS, and HRSD), a second incorporated the
rate, using a 7-point Likert-type scale ranging from disagree five scales of the SPSI (Problem Orientation, Problem Defini-
very strongly (1) to agree very strongly (7), their level of tion and Formulation, Generation of Alternatives, Decision
agreement regarding the following four items: (a) "I believe this Making, and Solution Implementation and Verification), and
treatment program will help (has helped) me to become less the third comprised the four scales of the LIS (Integrity, Interi-
depressed"; (b) "I believe that my therapist is (was) competent ority, Complexity, and Generativity). Individual contrasts be-
and can (has been) effective in helping me to cope better with tween mean scores of interest were conducted using the New-
my problems"; (c) "I agree with the rationale that this program man-Keuls multiple range test on the basis of error terms gen-
is based upon"; and (d) "Based upon the first session (entire erated by the omnibus (Wilks's lambda estimates) and
program), I believe that I will be (have been) helped to become respective subsequent measure-specific univariate F tests.
less depressed." Table 2 contains the means and standard deviations for all
dependent variables over the three assessment points.
A 2 (conditions) X 2 (trials) X 3 (therapists) MANOVA incor-
Depression. This overall MANOVA initially yielded a non-
porating all four items indicated no significant differences as a
significant omnibus effect for condition, F(2, 56) = 2.25, p =
function of group assignment, therapist, or assessment point.
.12, but significant omnibus effects for both the Trials effect,
Although such ratings are subject to social desirability factors,
F(5, 52) = 45.82,p < .001, and the Interaction effect, F(\0,280)
these findings provide some support for the notion that any con-
= 8.06, p < .001. Individual contrasts indicated that subjects in
sequent differences between conditions were not likely attribut-
both the PST and RT conditions were found to exhibit signifi-
able to subjects' expectations, satisfaction, or perception of the
cantly less depression at posttreatment as compared with their
competency of their therapists.
pretreatment scores according to all three depression measures
In addition, because therapist effects have often been found (all ps < .05), whereas WLC subjects displayed no significant
to be stronger than differential treatment effects (Beutler et al, improvement. Furthermore, as again exemplified by all three
1991), additional analyses were conducted to investigate the measures, PST and RT participants reported significantly lower
equivalence of therapist effects for each treatment condition. depression scores at posttreatment than individuals in the WLC
Accordingly, a series of 2 (PST versus RT) X 3 (the three thera- condition (all ps < .05). Finally, PST participants were found to
pists) X 2 (pretreatment versus posttreatment) repeated mea- be significantly less depressed than RT subjects according their
sures MANOVAs were conducted on all relevant dependent posttreatment HRSD and GDS scores (ps < .05) but not ac-
variables, with the therapist factor treated as a random effect. In cording to their posttreatment BDI scores.
ruling out therapist effects in preliminary analyses, we followed Another strategy to evaluate differential treatment effects in-
the guidelines of Crits-Cristoph and Mintz (1991) in using an volves considering variations in the proportions of subjects at
alpha level of .3, rather than the standard .05, to determine sig- posttreatment who no longer meet the Diagnostic and Statisti-
nificance. In all of these analyses, no significant Therapist or cal Manual of Mental Disorders (3rd ed., rev.; DSM-III-R;
Therapist X Condition effect was found to be significant (all American Psychiatric Association, 1987) diagnostic criteria for
ps > .32). Given the absence of such significant differences, all major depressive disorder (see Table 3). The data are reported
subsequent analyses combined data across therapists. first for those subjects who completed treatment (i.e., excluding
The last validity check focused on subject attrition. During dropouts) and then for the entire sample of subjects who started
the course of the study, 16 subjects dropped out of treatment (9 treatment (i.e., including dropouts). In the latter set, missing
from the PST condition and 7 from the RT condition), repre- data were substituted using the conservative assumption that
senting an overall attrition rate of 29.1%. Participants were subjects who dropped out of treatment continued to remain de-
found to leave treatment for a variety of reasons: physical illness pressed. For both sets of data, chi-squared tests showed that at
(2 in PST, 1 in RT); "dissatisfaction with the treatment" (1 in posttreatment (and follow-up) a significantly greater proportion
PST, 3 in RT); unexpected time conflicts (2 in PST, 1 in RT); of participants in the PST condition, compared with those in
 DEPRESSION IN OLDER ADULTS 1007

Table 2
Means and Standard Deviations for All Dependent Measures by Condition
at Pretreatment, Posttreatment, and Follow-Up Assessments
PST(«= 19) RT(n = 28) WLC (n = 20)

Measure M SD M SD M SD

Hamilton Rating Scale for Depression

Pretreatment 25.2^ 5.7 25.3Aa 6.1 22.4Aa 5.5
Posttreatment 8.8ft, 6.3 17.6Bb 9.9 22.0AC 4.5
Follow-up 8.5ea 6.0 I8.4Bb 8.8 — —
Geriatric Depression Scale
Pretreatment 19.8Aa 4.8 19.2Aa 6.6 17.0Aa 5.1
Posttreatment 13.6,,. 6.6 16.5Bb 6.8 18.4AC 4.3
Follow-up 13.8ft, 6.2 6.2 —
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

14.5Ba
—
Beck Depression Inventory
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Pretreatment 23.7^ 5.2 23.6Aa 6.5 23.0Aa 4.3

Posttreatment IS.?*, 6.9 16.9s, 9.5 21.2Aa 6.0
Follow-up 16.7ft, 9.9 15.6B, 9.0 — —
Social Problem Solving Inventory
Problem orientation
Pretreatment 59.1 A 24.6 51.8A 20.6 59.8A 27.1
Posttreatment 65.6AB 26.4 55.2A 23.2 65.0A 23.3
Follow-up 67.7B 26.8 59.4A 24.9 — —
Problem definition and formulation
Pretreatment 16.0A 7.3 18.8A 10.2 16.3A 6.9
Posttreatment 19.5, 7.0 17.1A 9.7 16.0A 6.9
Follow-up 20.5B 7.4 19.1A 11.0 — —
Generation of alternatives
Pretreatment 17.7A 8.6 19.0A 9.0 18.4A 7.8
Posttreatment 21.6B 7.9 20.0A 9.9 17.6A 7.7
Follow-up 23.0B 7.4 22.1A 6.2 — —
Decision making
Pretreatment 20.8A 7.4 20.7A 9.1 20.8A 6.2
Posttreatment 23.0B 8.1 19.9A 8.8 20. 1A 6.6
Follow-up 23.5B 8.2 19.4A 6.9 —
Solution implementation and verification —
Pretreatment 18.8A 8.3 18.7A 9.2 19.0A 7.1
Posttreatment 21.9A 8.6 19.0A 8.0 19.3A 7.9
Follow-up 22.1A 8.4 20.7A 9.6 —
Life Integration Scale —
Integrity
Pretreatment 7.4A 3.4 5.6A 3.7 6-9A 3.7
Posttreatment 9.0B 3.5 6.0A 3.9 7.6A 3.6
Follow-up 9.0AB 3.9 6.4A 4.2 —
Inferiority
Pretreatment 7.7A 3.0 7.2A 3.4 7.4A 2.7
Posttreatment 7.7A 2.8 8.1A 3.3 7.3A 2.7
Follow-up 8.0A 2.8 7.3A 3.7 —
Complexity
Pretreatment 4.5A 2.6 4.3A 3.2 5.0A 3.0
Posttreatment S.IAB 3.0 6.3B 4.4 5.0A 3.1
Follow-up 6.1B 3.3 6.4B 3.6
Generativity
Pretreatment 4.5A 2.6 7.8A 3.2 8.0A 3.0
Posttreatment S.IAB 2.9 9.1B 4.3 8.2A 3.1
Follow-up 9.8B 3.4 9.1B 4.0 —
Note. Follow-up data were not collected on WLC subjects. For each scale, dissimilar uppercase subscripts
(down columns) indicate significant differences within a condition across differing assessment points; dis-
similar lowercase subscripts (across rows) indicate significant differences between conditions within the
same testing period. PST = problem-solving therapy; RT = reminiscence therapy; WLC = waiting-list
control.

the RT and WLC groups, no longer experienced the constella- for both the Condition effect, F(2, 56) = 0.32, p = .73, and the
tion of symptoms indicative of major depressive disorder (ps < Interaction effect, F(18, 96) = 1.02, p = .45. The Trials effect,
.01; see Table 3). however, was found to be significant, omnibus F(9,48) = 29.52,
Social problem solving. Initial results from the 3 X 2 repeated p < .001. Subsequent individual contrasts revealed that only
measures MANOVA indicated nonsignificant omnibus F ratios PST participants showed any significant improvement in prob-
 1008 AREAN, PERRI, NEZU, SCHEIN, CHRISTOPHER, JOSEPH

Table 3 in reducing depressive symptoms for "poor problem solvers"

Percentage of Subjects in Each Condition Who Met (BDI and HRSD), "good problem solvers" (HRSD), "poor life
the Diagnostic Criteria for Major Depressive integrators" (HRSD), and "good life integrators" (BDI),
Disorder at Each Assessment whereas RT was found to be an effective intervention for indi-
viduals described as "good problem solvers" (BDI and HRSD),
PST RT WLC "poor problem solvers" (HRSD), and "good life integrators"
Assessment (HRSD; all ps < .05). However, RT was not found to be effective
for subjects characterized as "poor life integrators." Given the
Excluding dropouts post hoc nature of the analyses, these results should be viewed
Pretreatment 100 19 100 20 100 20 as quite tentative.
Posttreatment 11 2 60 12 90 18
Follow-up 11 2 70 6
Including dropouts3 Follow-up Analyses
Pretreatment 100 28 100 27 100 20
Posttreatment 39 11 70 19 90 18
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Because of ethical and clinical concerns, the 20 members of

Follow-up 39 11 78 2!
This document is copyrighted by the American Psychological Association or one of its allied publishers.

the WLC condition were provided with treatment at the end of

Note. PST = problem-solving therapy; RT = reminiscence therapy; the initial 12 weeks. Therefore, follow-up analyses included
WLC = waiting-list control. only those individuals in the PST and RT conditions. Because
" These calculations were based on the assumption that subjects who we were interested in determining whether positive treatment
dropped out of treatment met the criteria for major depressive disorder. effects evidenced at posttreatment were maintained at the 3-
month follow-up assessment, we were particularly interested in
possible significant trials effects within the 2 x 2 (Conditions X
lem solving as indicated by three SPSI scales (Problem Defini- Assessment Points) MANOVAs. Results across three separate
tion and Formulation, Generation of Alternatives, and Decision MANOVAs focusing on the three different measurement sets
Making). In other words, RT and WLC subjects did not im- (i.e., depression, problem solving, and life integration) indi-
prove on these scales. In addition, Newman-Keuls tests indi- cated nonsignificant omnibus F ratios (Wilks's lambda) repre-
cated that none of the conditions was characterized by signifi- senting the effects caused by time: For depression, F(5, 33) =
cant improvements on the remaining two problem-solving 2.01; for problem solving, F(9, 29) = 0.50; for life integration,
scales (problem orientation and solution verification). F(7, 30) = 1.92 (all ps > . 10). In essence, this suggests that the
Life integration. This initial repeated measures MANOVA overall treatment effects observed at posttreatment for both the
indicated nonsignificant omnibus effects regarding the Condi- PST and RT conditions were maintained 3 months after the
tion, F(2, 55) = 0.44, p = .65, and Interaction terms, F(14, 98) completion of treatment.
= 1.18, p = .30, respectively. However, the Trials effect was
found to be significant, omnibus F(l, 49) = 13.14, p < .001. Discussion
Individual contrasts revealed that (a) PST subjects were the only
participants to achieve a significant pre- to posttreatment im- Before discussing the implications of our findings, we note
provement regarding the Integrity scale (p < .05), (b) RT sub- here several caveats regarding interpretation of the results. First,
jects were the only individuals to show significant improve- although the subjects' ratings of perceived therapist competency
ments over time on the Complexity and Generativity scales (ps and perceived treatment efficacy at pretreatment and posttreat-
< .05), and (c) no differences were evidenced on the Interiority ment were found to be equivalent across both treatment condi-
tions, absent in our study was an independent means of assuring
scale.
treatment equivalence. It is possible that the therapists acquired
the skills necessary to implement the two protocols compe-
Additional Analyses tently at different rates or with differing preferences. Although
Although the primary intent of this study was to evaluate the preliminary analyses that included the therapist factor as a ran-
relative efficacy of PST versus RT, we were also interested in dom effect yielded no cause for concern (i.e., lack of significant
whether these two approaches worked differentially as a func- differences), the validity of these analyses is threatened by the
tion of pretreatment skill deficits. More specifically, one could small number of therapists involved in this investigation and
hypothesize that preexisting deficits in either problem solving by the consequent low power available when conducting such
or life integration may be predisposing factors for differential analyses (cf. Crits-Cristoph & Mintz, 1991).
response to the two therapies. To address this issue, we con- Second, similar to this concern, a recent review by Robinson,
ducted two additional MANOVAs, one focusing on "good" ver- Berman, and Neimeyer (1990) raises a question regarding the
sus "poor" problem solvers, and the second on "good" versus validity of psychotherapy research in which the principal inves-
"poor" life integrators, using median splits on the SPSI and LIS tigators have a strong "allegiance" to the treatment approach
measures, respectively. The dependent measures for both anal- under scrutiny. For example, in the present investigation, if our
yses included all three measures of depression (BDI, GDS, and predominant orientation was allied with a problem-solving ap-
HRSD) and involved a 2 X 2 X 2 (Time of Analysis X Condition proach, one might suspect that all that was accomplished by
X Predisposing Skill Deficit) repeated measures approach. such research was a confirmation of an allegiance bias. Al-
The results of these MANOVAs and subsequent follow-up though one of the present authors (A. M. Nezu) is a major pro-
analyses provided a somewhat mixed picture depending on the ponent of the problem-solving model, he was not involved in
measure of depression. In essence, PST was found to be effective the actual supervision of the therapists. Moreover, the two major
 DEPRESSION IN OLDER ADULTS 1009

supervisors/trainers for two protocols each espouse the two ineffectiveness of the treatment, to a lack of sensitivity in the
varying models (i.e., M. G. Perri espouses problem solving, and measurement of life integration, or to a combination of both.
F. Christopher, reminiscence). As such, supervisor or mentor The effectiveness of PST in alleviating depression, on the
influence on the attitudes, dispositions, and therapy behaviors other hand, may be attributed in part to improvements in the
of the three therapists were equivalent across the two conditions. participants' ability to cope with the major and minor stressors
With these caveats in mind, we suggest that the data in this in their lives. Indeed, results from the SPSI showed that, over
study support the efficacy of psychotherapy in general for the the course of the study, PST subjects made significant improve-
treatment of major depressive disorder in older adults. The re- ments in three of the five component skills of problem solving.
sults showed significant improvements in depressive symptoms, These findings are consistent with the results of previous studies
as measured by self-reports and observer ratings, for partici- (Nezu, 1986; Nezu& Perri, 1989) that demonstrated an associ-
pants who completed either form of psychotherapy, compared ation between improvements in problem-solving ability and the
with those who received no treatment. Moreover, 64% of the amelioration of depression. Such findings provide support for
individuals who completed treatment showed substantial de- the proposition that ineffective coping with problems in living
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

grees of positive change such that, at posttreatment, their de- may contribute to the experience of depression in older adults
This document is copyrighted by the American Psychological Association or one of its allied publishers.

pression was classified as improved or in remission (based on (cf. Lazarus, 1991; Nezu, 1987; Nezu etal., 1989).
the SADS). These data are in agreement with findings of other The primary focus.of this study was not intended to involve a
studies (Gallagher & Thompson, 1982; Steuer et al., 1984; detailed analysis of psychotherapy process variables. Therefore,
Thompson et al., 1987) and suggest that the majority of de- our evaluation of the mechanisms of action responsible for
pressed older adults show substantial improvements when pro- treatment efficacy should be viewed as tentative. Beyond further
vided with psychotherapy. examination of a "predisposing skill deficit" hypothesis, other
The results from our WLC subjects revealed little evidence of process-oriented variables should be investigated in the future.
spontaneous remission. During the 12-week waiting period, the For example, Beutler et al. (1991) found that patients' predis-
control subjects showed minimal change in depressive symp- posing coping styles (externalizing versus internalizing) and de-
toms, whereas the treated patients improved significantly over fensiveness (resistant versus low defensiveness) were significant
time on all three measures of depression (HRSD, GDS, and predictors of differential treatment response regarding cogni-
BDI). Furthermore, 90% of the WLC subjects manifested tive, experiential, and self-directed therapy protocols. Future re-
sufficient symptomatology to warrant a diagnosis of major de- search should be directed toward such fine-grained analyses.
pressive disorder at the conclusion of the 12-week waiting pe-
riod. These findings highlight the observation that elderly de- References
pressed patients are not likely to improve unless they receive
Adler, T. (1992, February). For depressed elderly, drugs advised. APA
some type of treatment (Thompson et al., 1987). Monitor, pp. 16-17.
Furthermore, our examination of differential treatment Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, J.
effects between both experimental conditions provides initial (1961). An inventory for measuring depression. Archives of General
support of the superiority of a problem-solving approach as Psychiatry, 4, 561-571.
compared to reminiscence-focused intervention. Specifically, Beutler, L. E., Engle, D., Mohr, D., Daldrup, R. J., Bergan, J., Meredith,
PST subjects were found to report significantly less depression K.., & Merry, W. (1991). Predictors of differential response to cogni-
as measured by the GDS and the HRSD, compared with indi- tive, experiential, and self-directed psychotherapeutic procedures.
viduals receiving the RT protocol. Moreover, significantly fewer Journal of Consulting and Clinical Psychology, 59, 333-340.
Beutler, L. E., Scogin, E, Kirkish, P., Schretlen, D., Corbishley, A.,
participants in PST (11%) remained depressed at posttreatment Hamblin, D., Meredith, K., Potter, R., Bamford, C. R., & Levenson,
versus those in RT (60%). As such, both the degree of improve- A. I. (1987). Group cognitive therapy and alprazolam in the treat-
ment in depressive symptomatology, as well as the percentage of ment of depression in older adults. Journal of Consulting and Clinical
subjects who actually improved, compare favorably with suc- Psychology, 55, 550-557.
cess rates reported for other psychologically based interventions Blazer, D., Hughes, D. C., & George, L. K. (1987). The epidemiology of
for depression in the elderly (e.g., Beutler et al., 1987; Thomp- depression in an elderly community population. The Gerontologist,
son etal., 1987). 27,281-287.
However, contrary to the findings of Thompson et al. (1987) Brand, E., & Clingempeel, W. G. (1992). Group behavior therapy with
depressed geriatric inpatients: An assessment of incremental efficacy.
and Gallagher and Thompson (1982), who found no differences Behavior Therapy, 23, 475-482.
in treatment response to cognitive, behavioral, or psychody- Butler, R. (1974). Successful aging and the role of life review. Journal of
namic approaches, our study did detect significant differences the American Geriatric Society, 22, 529-535.
between a skills-oriented, cognitive-behavioral approach (i.e., Butler, R. N., Lewis, M. L, & Sunderland, T. (1991). Aging and mental
PST) and a more introspective, insight-oriented model of treat- health: Positive psychosocial and biomedical approaches (4th ed.).
ment (i.e., RT). Additional analyses indicated that, in part, the Columbus, OH: Charles E. Merrill.
lack of significant improvement in depressive symptomatology Crits-Cristoph, P., & Mintz, J. (1991). Implications of therapist effects
for RT subjects may be a function of the absence of a large im- for the design and analysis of comparative studies of psychotherapies.
pact of RT for individuals characterized as "poor life integ- Journal of Consulting and Clinical Psychology, 59, 20-26.
Dunn, V. K., & Sacco, W. P. (1989). Psychometric evaluation of the
rators." In other words, the changes in the variables hypothe- Geriatric Depression Scale and the Zung Self-Rating Depression
sized to mediate "integrity versus despair" did not occur (cf. Scale using an elderly community sample. Psychology and Aging, 4,
Erikson et al., 1986). This failure to engender major improve- 125-126.
ment on the life integration variables may be attributed to the D'Zurilla, T. J., & Nezu, A. M. (1990). Development and preliminary
 1010 AREAN, PERRI, NEZU, SCHEIN, CHRISTOPHER, JOSEPH

evaluation of the social problem-solving inventory. Psychological As- literature review and proposal of a pluralistic model. Clinical Psy-
sessment, 2, 156-163. chology Review, 7, 121-144.
Endicott, J., & Spitzer, R. (1978). A diagnostic interview for affective Nezu, A. M., Nezu, C. M., & Perri, M. G. (1989). Problem-solving ther-
disorders and schizophrenia. Archives of General Psychiatry, 35, 837- apy for depression: Theory, research, and clinical guidelines. New
844. York: Wiley.
Erikson, E. H., Erikson, J. M., & Kinvick, H. Q. (1986). Vital involve- Nezu, A. M., & Perri, M. G. (1989). Social problem-solving therapy for
unipolar depression: An initial dismantling investigation. Journal of
ment in old age: The experience of old age in our time. New York:
Consulting and Clinical Psychology, 57, 408-413.
Norton.
Rattenberg, C., & Stones, M. J. (1989). A controlled evaluation of rem-
Fry, P. S. (1989). Mediators of perceptions of stress among community- iniscence and current topics discussion groups in a nursing home
based elders. Psychological Reports, 65, 307-314. context. The Gerontologist, 29, 768-777.
Gallagher, D., Breckenridge, J., Steinmetz, J., & Thompson, L. (1983). Rich, C. L., Young, D., & Fowler, R. C. (1986). San Diego Suicide Study
The Beck Depression Inventory and Research Diagnostic Criteria: I: \oung versus old subjects. Archives of General Psychiatry, 43, 577-
Congruence in an older population. Journal oj Consulting and Clini- 582.
cal Psychology, 51, 945-946. Robinson, L. A., Berman, J. S., & Neimeyer, R. A. (1990). Psychother-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Gallagher, D. E., & Thompson, L. W. (1982). Treatment of major de- apy for the treatment of depression: A comprehensive review of con-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

pressive disorder in older adult outpatients with brief psychothera- trolled outcome research. Psychological Bulletin, 108, 30-49.
pies. Psychotherapy: Theory, Research and Practice, 19, 482-490. Ruegg, R. G., Zisook, S., & Swendlow, N. R. (1988). Depression in the
aged. Psychiatry Clinics of North America, 11, 83-89.
Gerson, S. C., Plotkin, D. A., & Jarvik, L. F. (1988). Antidepressant
Ryff, C. D., & Heinke, S. G. (1983). Subjective organization of person-
drug studies, 1964-1986: Empirical evidence for aging patients. Jour-
ality in adulthood and aging. Journal of Personality and Social Psy-
nal of Clinical Psychopharmacology, 8, 311-322.
chology, 44.801-816.
Goldwasser, A. N., Auerbach, S. M., & Harkins, S. W. (1987). Cognitive, Spitzer, R. L., Endicott, J., & Robins, E. (1978). Research diagnostic
affective, and behavioral effects of reminiscence group therapy on de- criteria: Rationale and reliability. Archives of General Psychiatry, 35,
mented elderly. International Journal of Aging and Human Develop- 773-782.
ment, 10, 555-557. Steuer, J., Mintz, J., Hammen, C., Hill, M. A., Jarvik, L. E, McCarely,
Hamilton, M. (1960). A rating scale for depression. Journal of Neurol- T, Motoike, P., & Rosen, R. (1984). Cognitive-behavioral and psy-
ogy, Neurosurgery and Psychiatry, 23, 56-62. chodynamic group psychotherapy in the treatment of geriatric de-
Hussian, R. A., & Lawrence, P. S. (1981). Social reinforcement of activ- pression. Journal of Consult ing and Clinical Psychology, 52, 180-189.
ity and social problem-solving training in the treatment of depressed Teri, L., & Lewinsohn, P. (1982). Modification of the pleasant and un-
institutionalized elderly. Cognitive Therapy and Research, 5, 57-69. pleasant events schedule for use with the elderly. Journal of Consult-
ing and Clinical Psychology, 50, 444-445.
Lazarus, R. S. (1991). Emotion and adaptation. New York: Oxford Uni-
Thompson, L. W., Gallagher, D., & Breckenridge, J. S. (1987). Compar-
versity Press. ative effectiveness of psychotherapies for depressed elders. Journal of
Matteson, M. A. (1984). Group reminiscing for depressed institutional- Consulting and Clinical Psychology, 55, 385-390.
ized elderly. In I. Burnside (Ed.), Working with the elderly: Group Winstead, D. K., Mielke, D. H., & O'Neill, P. T. (1990). Diagnosis and
process and techniques (2nd ed., pp. 287-297). Belmont, CA: Wads- treatment of depression in elderly: A review. Psychiatric Medicine, 8,
worth. 85-98.
National Institutes of Health. (1992). Diagnosis and treatment of de- Yesavitch, J., Brink, T, Rose, T, Lum, O., Huang, Q, Adey, V., & Leirer,
pression in late life. Journal of the American Medical Association, V. (1983). Development and validation of a geriatric screening scale:
268, 1018-1024. A preliminary report. Journal of Psychiatric Research, 17, 37-49.
Nezu, A. M. (1986). Efficacy of a social problem-solving therapy ap-
proach for unipolar depression. Journal of Consulting and Clinical Received November 3, 1992
Psychology, 54, 196-202. Revision received March 9, 1993
Nezu, A. M. (1987). A problem-solving formulation of depression: A Accepted March 9, 1993 •