Journal of Consulting and Clinical Psychology 1991, Vol. 59, No. 4, 547-557

In the public domain

Marital Therapy as a Treatment for Depression Neil S. Jacobson

Keith Dobson University of Calgary

University of Washington

Alan E. Fruzzetti

Karen B. Schmaling

University of Washington

Department of Medicine, National Jewish Center for Immunology and Respiratory Medicine Department of Psychiatry,: University of Colorado Health Sciences Center

Sheppard Salusky

University of Washington The purpose of this study was to compare cognitive-behavioral therapy (CT; n = 20), behavioral marital therapy (BMT; n = 19), and a treatment combining BMT and CT (CO; n = 21) in the alleviation of wives' depression and the enhancement of marital satisfaction. BMT was less effective than CT for depression in maritally nondistressed couples, whereas for maritally distressed couples the two treatments were equally effective. BMT was the only treatment to have a significant positive impact on relationship satisfaction in distressed couples, whereas CO was the only treatment to enhance the marital satisfaction of nondistressed couples. On marital interaction measures CO was the only treatment to significantly reduce both husband and wife aversive behavior and to significantly increase wife facilitative behavior.

Although many pharmacological and psychosocial treatments have been shown to be effective in alleviating depression, the treatments examined to date appear to have significant limitations. Pharmacological treatments produce only temporary effects, and relapse rates are high even among those who are maintained on medication subsequent to recovery (Prien et al, 1984). Although various psychosocial treatments such as cognitive-behavioral therapy may have prophylactic effects (Blackburn, Eunson, & Bishop, 1986; Evans et al., 1990; Simons, Murphy, Levine, & Wetzel, 1986), there is no definitive evidence that any existing treatment prevents recurrence. Moreover, when appropriately conservative criteria for recovery were invoked to evaluate the clinical significance of treatment effects, the National Institute of Mental Health (NIMH) Treatment of Depression Collaborative Research Program (Elkin et al., 1989) reported recovery rates of 57% for pharmacology plus clinical

management, 55% for interpersonal psychotherapy, and 51% for cognitive-behavioral therapy. Although these rates exceeded the 29% recovery rate found in a placebo/clinical management condition, almost half of the treated outpatients remained at least somewhat depressed at the conclusion of therapy. There is room for improvement in the technology for treating depression. Many of the treatments studied in clinical trials are intrapersonal in focus, treating the depressed person in relative isolation from other family members. Focusing on the depressed individual is in keeping with the prevailing theories of depression. However, there is reason to suspect that the involvement of family members, and in particular the spouses of depressed individuals, could have a facilitative effect on treatment outcome (Jacobson, Holtzworth-Munroe, & Schmaling, 1989). First, studies have reported a significant inverse relationship between depression and marital satisfaction (e.g., Coleman & Miller, 1975). Second, disruptions in close relationships may be the single most common precipitant of depressive episodes (Paykel et al., 1969). Third, marital distress is a good predictor of depressive relapse following recovery from an acute episode (e.g., Hooley & Teasdale, 1989). Fourth, the extent to which marital satisfaction improves during the course of therapy bears an inverse relationship to relapse (e.g., Rounsaville, Weissman, Prusoff, Herceg-Baron, 1979). Finally, a close, confiding relationship with a spouse can buffer the otherwise depressogenic effects of stressful life events (e.g., Brown & Harris, 1978). Thus, it seems reasonable to hypothesize that marital therapy has a role to play in the treatment of depression. The primary purpose of the present study was to investigate the effects of marital therapy (the behavioral approach de-

This research was supported by Grant MH33838 from the National Institute of Mental Health, awarded to Neil S. Jacobson. We wish to thank many valuable contributors to this research program: Sandra Coffman, Joyce Victor, Steve Sholl, David Coppel, Redmond Reams, and Leanne Wilson, who served as therapists; Victoria Follette and Mark Whisman, who served as project coordinators; Jennifer Waltz, Kelly Koerner, and Michael Addis, who rated therapy session tapes; Joan Giacomini, who served as program assistant; and Hyman Hops, who supervised the coding of marital interaction. We are especially grateful to three anonymous reviewers, whose comments on our initial draft led to a greatly improved article. Correspondence concerning this article should be addressed to Neil S. Jacobson, Department of Psychology NI-25, University of Washington, Seattle, Washington 98195. 547

548

JACOBSON, DOBSON, FRUZZETTI, SCHMALING, SALUSKY

scribed by Jacobson & Margolin, 1979). Evidence based on 17 clinical trials supports the efficacy of behavioral marital therapy (e.g., Hahlweg & Markman, 1988; Jacobson, 1978). We compared behavioral marital therapy (BMT) with a cognitive-behavioral treatment (CT) focusing on the depressed spouse alone (Beck, Rush, Shaw, & Emery, 1979). This treatment has been shown to be at least as effective (Hollon et al., 1990; Murphy, Simons, Wetzel, & Lustman, 1984) if not more effective (Rush, Beck, Kovacs, & Hollon, 1977) than other existing treatments, including tricyclic antidepressants (Dobson, 1989). In addition to the comparison between BMT and CT, we included a third condition that combined the two component treatments (CO). Because CT is focused directly on depressive symptoms but only indirectly on marital satisfaction and because BMT is directed toward marital satisfaction rather than depressive symptoms, it was thought that a combination treatment might maximize the strength of both treatments. Recently, O'Leary and Beach (1990) reported findings from a study similar to this one, in which married couples complaining of both depression in the wife and marital discord were randomly assigned to their version of BMT, CT, or a waitinglist control group. They reported that although both active treatments were effective and equally so in alleviating depression, only BMT was successful in enhancing marital satisfaction. Our study complements the work of O'Leary and Beach in a number of ways. First, the present study includes a CO treatment. Second, our study treated a sample that had to attain more stringent severity criteria for depression (criteria that approximate the typical cutoff points used in other trials; Elkin et al., 1989). Third, the present sample of couples was heterogeneous with respect to marital satisfaction. Whereas the O'Leary and Beach study selected only couples who were both depressed and complaining of marital discord, our study selected on the basis of clinical depression and thus included couples with a range on measures of marital satisfaction. Although at first glance it might seem counterintuitive to include maritally nondistressed couples in a study of marital therapy for depression, both our own interaction research (e.g., Schmaling & Jacobson, 1990) and various systemic theories of depression (e.g., Jacobson et al., 1989) suggest that there may be dysfunctional interaction patterns even in those couples with a depressed spouse who report high levels of marital satisfaction. We expected BMT to be at least as effective as CT in alleviating depression, especially for couples complaining of marital distress. We also expected BMT to be more effective than CT at improving marital satisfaction among distressed couples. Finally, we expected the CO treatment to have more generalized effects than the component treatments. We expected it to outperform CT as a treatment for marital problems and to outperform both BMT and CT as treatments for depression.

Method

ing to attend therapy. Subjects were recruited from a number of sources; 42% came from referrals (e.g., from mental health agencies), and the others (58%) were solicited in newspapers offering treatment as part of a research project.1 Sixty subjects completed treatment (20 in CT, 19 in BMT, and 21 in CO), and analyses included only those subjects. An additional 12 subjects began but dropped out of treatment. (Six were from maritally nondistressed couples assigned to BMT, 2 were from maritally distressed couples assigned to BMT, one was from a distressed couple assigned to CT, and 3 were from nondistressed couples assigned to CT; none of the CO subjects dropped out). A chi-square test for differential dropout rate was not significant: x2(2, N=\\) = 7.5, ns. Two additional analyses were conducted to see whether the exclusion of dropouts in previous analyses biased the results. In one, only the original 60 subjects (the first 20 assigned to each condition) were analyzed. Because all dropouts provided BDI scores at the time of termination, we were able to include all of them in these analyses. This analysis is uncontaminated by attrition and thus preserves random assignment. In the other reanalysis, dropouts were included, but so were their replacements. Although the replacement of dropouts does not undo the damage done to randomization because of attrition, it does increase the prospects of group equivalence when compared with an attrition without replacement strategy (Hollon, 1990). Both of these analyses yielded results that were virtually identical to those on completers only. Thus, it was concluded that attrition did not distort or produce bias in the findings. Analyses of variance (ANOVAs) revealed no significant differences between treatment conditions in wife's age (M= 38.5, SD = 8.5); husband's age (M = 40.5, SD = 9.7); wife's education (M = 14.6 years, SD = 2.2); husband's education (M = 15.6, SD = 2.2); number of years married (M = 11.3, SD - 9.5); or number of children (M = 1.7, S£>=1.3). Although comorbidity was minimized by our exclusionary criteria (see below), 43.3% of our sample had a coexisting generalized anxiety disorder, and 11% met criteria for panic disorder. On the basis of data from the Millon Clinical Multiaxial Inventory (3rd ed., MCMI; MilIon, 1983), virtually all of the depressed women had at least one elevated score: (e.g., 58.3% on the Borderline-Cycloid scale; 35% on the Schizoid-Asocial scale; 36.7% on the Avoidant scale; 46.7% on the Dependent-Submissive scale; and 73.3% on the Passive Aggressive-Negativistic subscale). Finally, the median number of previous depressive episodes in our sample was 4, with a range of from 1 to 96. Spouses of the depressed wives were similarly excluded from the study if they met any of the exclusionary criteria described above. In addition, husbands were excluded if they scored above 20 on the BDI, regardless of whether or not they met DSM-III criteria for major depression. Finally, all husbands completed the Symptom Checklist (SCL-90; Derogatis, 1983) and the MCMI. On the SCL-90, no husband scored in the clinical range (T score < 70) on any scale. However, 36.7% of the husbands had elevated scores on the Histrionic-Gregarious scale of the MCMI, whereas 33.3% had elevations on the Narcissistic subscale.

Measures Diagnostic Interview Schedule (DIS; Robins, Helzer, Croughan, & Ratcliff, 1981). The DIS is a structured interview that provides diag-

Subjects Subjects for this study were married women who met Diagnostic and Statistical Manual of Mental Disorders (3rd ed., DSM-III; American Psychiatric Association, 1980) criteria for major depression; scored 20 or greater on the Beck Depression Inventory (BDI; Beck, Rush, Shaw, & Emery, 1979) and 14 or greater on the Hamilton Rating Scale for Depression (HRSD; Hamilton, 1960); and whose partners were will-

1 Subjects did not differ in any obvious way as a function of referral source. Although there was a trend for couples solicited through the media to be more severely depressed, this trend did not reach statistical significance. Recovery rates did not differ as a function of referral source, nor were there any interactions between treatment effect and source of referral.

MARITAL THERAPY noses consistent with the DSM-HI. The DIS can be administered by paraprofessional interviewers (see Hesselbrock, Stabenau, Hesselbrock, Mirkin, & Meyer, 1982). Diagnoses were determined by trained interviewers (clinical psychology graduate students), and a subset of these interviews were subsequently also rated by an experienced psychiatrist. Reliability was determined by the proportion of times in which interviewer and independent rater were in agreement as to whether or not the subject met inclusionary criteria for participation in the study. Agreement occurred 93% of the time on the basis of 13 jointly rated interviews. Exclusionary criteria included serious and imminent suicide potential, drug or substance use disorder, or lifetime diagnoses of organic mental disorder, schizophrenia, or bipolar disorder. Subjects were also excluded if the husband had been physically abusive toward his wife within the past year, or if the husband or wife was concurrently in psychotherapy with a professional. Subjects who were taking antidepressant medications were offered the opportunity to discontinue their medication. The washout period lasted for 2 weeks. Four subjects went through the discontinuation procedure, and 3 of the 4 remained eligible for the study following the procedure. Subjects were excluded if they insisted on continuing their medication. Hamilton Rating Scale for Depression (HRSD). The HRSD is an interview-based measure of the severity of depression. It is a widely used measure of depression severity that has been shown to have high concurrent validity (Shaw, Vallis, & McCabe, 1985). In the current study, a structured version of the 17-item HRSD was added to the DIS. As we reported elsewhere (Whisman et al, 1989), the HRSD was shown to have good psychometric properties and to correlate highly with the Hamilton ratings of an experienced psychiatrist. The HRSD was used not only as a criterion for inclusion but also as a primary outcome measure of improvement in depressive symptoms. Beck Depression Inventory (BDI). The BDI is the most commonly employed self-report measure of depression. It has been shown to be internally consistent and to have high concurrent validity with other severity measures of depression. Dyadic Adjustment Scale (DAS; Spanier, 1976). The DAS is a frequently employed self-report measure of marital satisfaction. It has been shown to have high internal consistency and discriminant efficiency and, compared with other measures of global marital satisfaction, to be sensitive to treatment effects (Whisman & Jacobson, 1991). Proportions of facilitative, aversive, dysphoric, and guidance behaviors. To study the effects of various treatments on the quality of marital interaction, we videotaped couples discussing moderate marital problems for 15 min before and after therapy. These data were coded on the LIFE (Living in Family Environments) coding system (Biglan et al, 1985; Hops et al, 1987). The LIFE system was designed to code the marital and family interaction of depressives and their relatives. It combines many of the content and affect codes from previous coding systems with some codes designed to be specific to depression (e.g., dysphoria). Previous research has shown that the LIFE system generates useful data and possesses discriminant, concurrent, and predictive validity (Biglan etal, 1985; Hopsetal, 1987). Our data were sent to the LIFE developers to be coded. The following summary codes were used: facilitative behavior, aversive behavior, dysphoric behavior, and guidance. The proportions of each delivered by each spouse at pretest and posttest served as the units of analysis. The coders reported that reliability data indicated Cohen's kappas averaging 0.62 for wives and 0.54 for husbands. In addition, intraclass correlation coefficients averaged 0.53 for wives and 0.86 for husbands.

Therapists Therapists were three licensed psychologists, one psychiatric social worker with an ACSW degree, and two advanced clinical psychology graduate students. Because the therapist factor was crossed with treatment condition, all therapists saw at least two cases in each of the three

549

treatment conditions. Before participation, all went through a year of training and treating pilot cases. The training included bibliotherapy, clinical workshops, peer meetings, and supervision by two of the co-authors for the BMT (NSJ) and CT (KSD) treatments. Because the site of the study is associated with BMT, steps were taken to counteract potential bias. First, all therapists were thoroughly trained and supervised by an experienced trainer (KSD), who had been one of the CT supervisors in the NIMH collaborative depression study. All therapists were trained until they scored consistently above 40 on the Cognitive Therapy Scale (CTS; Dobson, Shaw, & Vallis, 1985), a widely used measure of therapist competence in CT. Second, once therapists were certified as competent, their sessions continued to be rated and monitored, and remedial training was undertaken whenever a CTS score for a particular therapist dropped below 40 for two consecutive sessions (Shaw, 1984). Specifically, KSD rated five sessions from each case on the CTS, with the therapists kept blind as to which cases were being rated. The rated sessions were chosen randomly, although randomization was constrained to ensure that all segments of therapy were represented in the ratings for each case. Fourth, additional ratings were provided by a group of experts who had no connection with the study. Specifically, S. D. Hollon and his associates rated therapist tapes for competence, on the basis of the CTS, and adherence, on the basis of the Collaborative Study Psychotherapy Rating Scale (CSPRS), the scale developed by Hollon and his associates to assess adherence in the NIMH collaborative depression study (Hollon, Evans, Elkin, & Lowery, 1984). Specifically, Hollon and his associates rated three randomly chosen sessions from each pilot case for each of four therapists. The ratings by Hollon and his associates confirmed that the therapists in our study not only adhered to CT when implementing that condition but were competent CT therapists. The mean CTS scores were consistently above 40 and correlated significantly with Dobson's CTS ratings (rs between 0.50 and 0.70). In addition, on the CT scale of the CSPRS, our therapists' scores were at or above the mean for trained cognitive therapists in the collaborative study at early, middle, and late phases of the pilot year. Whereas the mean for cognitive therapists in the collaborative study was approximately 62, the mean for our therapists at early, middle, and late stages of training respectively was 70.67, 64.55, and 67.16 (Hollon et al., 1984).

Treatment Conditions Behavioral marital therapy (BMT). Subjects assigned to BMT were given the conceptualization that depression occurs in an interpersonal context and that this fact necessitated that both members of the marital dyad be included in therapy. The treatment program (20 sessions) was essentially that used in previous investigations (Jacobson & Margolin, 1979) but also involved more recent innovations (Jacobson & Holtzworth-Munroe, 1986). The treatment initially focused on behavioral exchange and then moved on to communication and problem-solving training, where couples were taught how to resolve conflicts around issues such as finances, sex and affection, parenting, and intimacy. The therapy approach was explicitly both cognitive and behavioral in nature. Although standard behavior therapy techniques such as behavior rehearsal and contingency management were emphasized, cognitive interventions such as refraining and other restructuring techniques were an explicit part of the protocol. However, the type of Socratic questioning and hypothesis testing that characterizes Beck's CT was not used. Cognitive-behavioral therapy (CT). This treatment consisted of the administration of Beck's 20-session cognitive therapy (Beck et al, 1979; DeRubeis & Beck, 1988). Consistent with other investigations (Dobson, 1989), therapists were trained in the assessment and modification of dysfunctional cognitions related to depression, as well as the assessment and modification of behavioral aspects of depression. Typically, the therapy involved a preliminary stage in which depressive be-

550

JACOBSON, DOBSON, FRUZZETTI, SCHMALING, SALUSKY

havior and its functional relationship to the maintenance of depression were assessed and modified, followed in turn by the assessment and modification of situation-specific automatic thoughts, and finally the assessment and modification of underlying beliefs and attitudes that may have precipitated or maintained the depressive episode. Combined treatment (CO). This treatment condition included 20 sessions of a combination of BMT and CT. For BMT sessions both spouses were required to attend, whereas for CT sessions only the depressed wife attended. In this treatment condition, patients were given the conceptualization that although depression is primarily experienced by one individual and therefore requires some individual treatment, it also has interpersonal consequences and potential causes, therefore also requiring the active participation of the spouse for at least some of the treatment program. Consideration was given to various strategies for combining the two treatment programs. It was determined that an optimal combination treatment was one that was somewhat flexible and permitted the optimal use of either individual or marital treatment sessions, as best fitted the patient's needs. On the other hand, in order to be a combined treatment, some minimal number of sessions of each type of treatment had to be provided. Thus, it was determined that for the combined treatment there had to be a minimum of 8 BMT sessions, and at least 6 CT sessions (total number of sessions was 20). The patterning of these sessions was left to the therapist's clinical judgment; however, a typical pattern was to begin with individual treatment and then to involve both spouses in the treatment plan.

Results Manipulation Check We developed a rating scale to measure treatment adherence. The final scale consisted of 51 items, 17 representing behaviors that were supposed to occur in CT, 17 representing BMT behaviors, and 17 representing clinical skills that were not specific to either component treatment. We referred to these latter items as representing "facilitative conditions." After listening to a tape, the rater rated each item on a scale ranging from did not occur at all (1) to extensively used (4). Three graduate students, blind to the treatment condition, served as raters. After the raters were trained to a reliability criterion of .70 (kappa), the tapes were divided between two of the three raters, with a third serving as a calibrator. Three tapes from each case were rated: one from the first third of the therapy regimen (randomly chosen from Sessions 2-7), one from the second third of the therapy regimen (randomly chosen from Sessions 8-14), and one chosen from the final third of the therapy regimen (randomly chosen from Sessions 15-19). Weekly meetings were held with the calibrator throughout the rating period to preserve reliability and avoid observer drift. First we compared the total BMT and CT scores for tapes rated from the component (BMT and CT) treatment conditions. The total BMT score was derived by summing the Likert values for the BMT items on the scale; the CT score was similarly derived. High scores meant high adherence. As expected, BMT scores were significantly higher in the BMT condition (M= 11.4, SD = 5.7) than in the CT condition (M= 1.3, SD = 1.4), F(\, 111) = 155.8, p < .001. Similarly, CT scores were significantly higher in the CT condition (M = 9.5, SD = 5.4) than in the BMT condition (M = 1.2, SD = 1.5), F(l, 111)= 109.4, p

Marital therapy as a treatment for depression.

The purpose of this study was to compare cognitive-behavioral therapy (CT; n = 20), behavioral marital therapy (BMT; n = 19), and a treatment combinin...
1MB Sizes 0 Downloads 0 Views