Behav. Res. Thu. Vol. 31, No. 3, pp. 325-330, 1993 OOOS-7967/93 S6.00 + 0.00 Printed in Great Britain. All rights mservcd Copyright 0 1993 Pergamon Press Ltd

Outcome profiles in the treatment of unipolar depression

STEVEN TAYLOR and PETER MCLEAN

Department of Psychiatry, University of British Columbia, 2255 Westbrook Mall, Vancouver, EC, Canada V6T 24 I

(Received 25 March 1992)

Summary-Treatment efficacy is typically evaluated by examining group means and pre-post change scores. Although informative, such analyses may obscure individual or subgroup differences in response (outcome profiles). The present study used two different methods to define treatment outcome profiles- rationally-derived criteria (Frank et al., Archives of General Psychiatry 48, 851-455, 1991) and dynamic clustering-to evaluate four treatments of unipolar depression: behaviour therapy, amitriptyline, psycho- dynamic psychotherapy and relaxation training (attention placebo). The profiling methods yielded similar results. Regardless of treatment, the majority of patients displayed either a recovery or nonremission outcome profile, with relatively few instances of remission followed by a recurrence of depression. These findings challenge the view that any of the treatments are associated with a strong tendency to relapse, at least over the 3-month follow-up period. To further characterize the major outcome profiles, discriminant analysis was performed. Results indicated that recovery and nonremission profiles differed in that the latter was associated with a longer and more severe index episode and greater neuroticism. A number of variables, including family history of depression and therapists’ prediction of outcome, failed to distinguish recovered from unremitted patients.

INTRODUCI’ION

There are several ways of characterizing treatment efficacy. The most common approach has been to measure the degree of improvement from pre- to post-treatment, and to determine whether these gains are maintained over a given follow-up interval. An important, complementary approach to therapy evaluation is to determine the major outcome profiles for a given type of treatment. That is, what proportion of Ss show treatment outcome patterns such as nonremission, remission without relapse, remission with relapse and so forth? These patterns may be obscured in the analysis of group means. For example, consider a comparison of group means for two treatments. Although the means may suggest that the treatments are of equivalent efficacy, it may be that one treatment yields of two types of treatment outcome-response and nonresponse-whereas the other treatment is moderately effective for all patients. Thus, the identification of major outcome profiles for a given treatment, and the comparison of these profiles with other treatments, enables one to more fully understand the treatment outcome characteristics and patient predictors of response to the various therapies.

There is suggestive evidence that behavioural-cognitive therapies are associated with fewer relapses or recurrences than tricyclic phannacotherapy (Hollon, Shelton & Loosen, 1991). This may occur because the former provides patients with skills in adaptive problem-solving and mood-management (Beck, Rush, Shaw % Emery, 1979, McLean, 1976), and suggests that a treatment response-then-recurrence (response -t recurrence) profile is unlikely to be a common outcome pattern for behavioural-cognitive therapies. The question remains as to the proportion of patients showing response + recurrence profiles for other treatments such as the pharmacotherapies. A further question is how the outcome profiles vary with the pre-treatment severity of depression. Elkin et al. (1989) found that tricyclic pharmacotherapy was of comparable efficacy to cognitive-behavioural therapy for mild depression, but that the former was more effective for severe. depression. Recent studies by McLean and Taylor (1992) and Hollon et al. (1989, unpublished, cited in Hollon, Shelton & Loosen, 1991) failed to replicate this finding. McLean and Taylor (1992) found that behaviour therapy was generally more effective than amitriptyline, and both treatments were less efiective for patients who were initially more depressed. These findings suggest that for both treatments, the more depressed patients show a nonresponse profile, whereas the less depressed patients are more likelv to resoond to treatment.

While the anal&is of outcome profiles can make an important contribution in understanding treatment outcome characteristics, it has not met with widespread use.. The main reason for this is the lack of consensus about how to define profiles. Currently. there is no consensus regarding the criteria for defining recovery, recurrence or remission of depression (Frank et al., 1991; Prien, Carpenter & Kupfer, 1991). Frank et al. (1991) recommended a set of criteria (described below), although these are arbitrary in nature. Another approach is to empirically determine outcome profiles by means of dynamic clustering (Prochaska, Velicer, Guadagnoli, Rossi & DiClemente, 1991). This is a cluster analytic procedure where the units of clustering are not a collection of variables obtained at a given time. as in conventional cluster analysis, but rather based on the same variable measured at different points in timeTIn this way dynamic outcome typologies can be obtained (Prochaska et al., 1991). Unfortunately, this procedure contains the limitations inherent to all clustering procedures. The main shortcoming is that the cluster solution needs to be cross-validated (Morris, Blashfield & Satx, 1981).

Dynamic clustering and the use of the Frank et al. (1991) criteria both have limitations, yet they may yield important information on the patterns of outcome associated with the various treatments for depression. The purpose of the present study was to apply both methods to the analysis of outcome profiles, with a particular emphasis on convergent results. Specifically, our aims were to identify and compare the major outcome profiles for the following treatments for unipolar depressed outpatients: behaviour therapy, psychodynamic psychotherapy, amitriptyline and attention placebo (relaxation training). Treatment effects were examined at post-treatment and 3-month follow-up. The group differences were originally

aaT 3Ip-G ,,a

326 CASE HSI’CRl!3 AND SHORTER COMMUNICATIONS

Table I. Outcome profiles and their defining criteria (adapted from Frank er II/., 1991)

Defining cri:eria: BDI scores

Outcome orotile’ Post-treatment

(IO wk uost-intake) 3-Month follow-up (22 wk post-intake)

I. 2. 3.

4. 5. 6.

7. 8. 9.

Recovery, sustained Remission + partial recurrence Remission -+ recurrence

Partial remission - full remission Partial remission, sustained Partial remission + recurrence

Unremitted + remission Unremitted + partial remission Unremitted, sustained

98 <8 $8

9-14 9-14 %I4

215 .I5 ,I5

<a 9-14 215

<8 9-14 215

68 %I4 215

‘x-ry = status at post-treatment followed by status at follow-up.

reported by McLean and Hakstian (1979). The results clearly favoured behaviour therapy over the other treatments. However, the question of outcome profiles has yet to be examined.

According to the criteria of Frank et a/. (1991) a score of > 15 on the Beck Depression Inventory (BDI: Beck, Ward, Mendelson, Mock & Erbaugh, 1961) indicates a fully symptomatic depressive episode, and a score <8 indicates an asymptomatic status. A depressive episode is defined as 24 wk of being fully symptomatic. Full remission is defined as 3-15 wk asymptomatic and recovery is defined as > 16 wk asymptomatic. These criteria were adapted slightly to the design of the present study, which consisted of assessments at week 0 (me-treatment), week 10 (post-treatment) and week 22 (3-month follow-up). All patients were fully symptomatic on admission. The definitions for the outcome profiles used in the present study are shown in Table 1. The distinction between relapse (exacerbation of the index episode) and recurrence (onset of a new episode) could not be determined from the data available for the present study. Therefore, the term recurrence was used to describe an initial improvement followed by exacerbation in depression.

METHOD

Subjects

Patients between 20 and 60 yr of age were recruited through newspaper announcements. The diagnosis of moderate to severe unipolar depression was made according to the research criteria proposed by Feighner, Robins, Guze, Woodruff, Winokur and Munoz (1972). This was determined by a three-stage assessment: (a) telephone screening to ascertain whether depressed mood had been present for at least I month; (b) a semi-structured clinical interview by a psychologist or psychiatrist to determine whether the S met Feighner criteria for depression; and (c) psychometric assessment. For the latter, the potential S was required to score in the moderately depressed range on two of three measures: a score > 25 for males or 229.5 for females on Scale D of the MMPI (Hathaway & McKinley, 1951); a score 223 on the BDI; or a score 3 14 on the Depression Adjective Check List (Lubin, 1965).

Patients were required to be fluent in written and spoken English and not in treatment elsewhere for depression. A total of 541 potential Ss were screened and 196 were assigned to treatment. The sample of this study is restricted to 155 Ss for whom complete BDI responses were available. These Ss had a mean age of 39 yr (SD = 10 yr), 73% were female, 57% were married and 54% were employed. For further details on S selection and sample characteristics see McLean and Hakstian (1979).

Measures

Subjects completed the BDI at pre-treatment, post-treatment and follow-up, and a number of pretreatment measures. The latter included the Eysenck Personality Questionnaire (EPQ: Eysenck & Eysenck, 1976), life events questionnaire (Holmes & Rahe, 1967) and a structured psychiatric interview assessing age at first onset of depression, duration of current episode, family history of depression, number of previous inpatient admissions, and number of previous treatment episodes. On the basis of the information elicited during the intake interview the therapist predicted the patient’s post-treatment status on a IO-point scale (1 = worse, 10 = greatly improved).

Design

Patients were randomly assigned to one of four treatments: psychodynamic psychotherapy, relaxation training (attention placebo), behaviour therapy or phannacotherapy. The respective number of completers was 39,37,40 and 39. Patients were assessed prior to treatment, immediately after treatment and 3 months post-treatment.

Treatments

All treatments consisted of 10 weekly sessions conducted on an individual outpatient basis by a licensed and experienced psychologist, psychiatrist or physician. Psychodynamic psychotherapy used fundamental principles of short-term insight- oriented nondirective psychotherapy (Marmor, 1973, 1975; Wolberg, 1967). Relaxation training consisted of progressive muscle relaxation. Behaviour therapy used graduated practice and modelling techniques, with targets for intervention consisting of communication skills, behavioural productivity, social interaction, assertiveness, decision-making, problem solving and self-control (McLean, 1976). Daily homework tasks were assigned, and an emphasis was placed on the avoidance of depressive rumination by engaging in pleasant and productive activities. Pharmacotherapy consisted of amitriptyline, starting at 75 mg/day and graduating over a 10 day period to a fixed treatment dose of 150 mg/day. After 1 I wk at 150 mg/day, the patients were weaned at a rate of 25 mg/day. Further details on treatments are provided in the original report (McLean & Hakstian, 1979).

CASE HiSTORiES AND SHORTER ~~~ICATIO~ 327

RESULTS

Frank et al. (1991) Criteria

Table 2 shows the percentage of patients in each treatment group who met criteria for a given outcome profile. The number of patients in each condition are given in parentheses. The table shows that bchaviour therapy was associated with the greatest proportion of patients with sustained recovery. The next most efficacious treatment was amitriptyline. These results are in keeping with the group means for these data, as reported in the original study (McLean & Hakstian, 1979). Of greater interest for the present purposes is the dist~bution of profiles across treatments. The table shows that the distribution of profiles was bimodal for each treatment. That is. over half of the patients (53%) had one of two profiles: sustained recovery (22% of sample) or sustained unremitted depression (31%). There were very few patients who had a full remission followed by a recurrence (4%), and few patients with a partial remission followed by recurrence (9%). Aside from the greater efficacy of behaviour therapy and amitriptyline, the proportion of outcome patterns is similar across all four treatments.

The true proportion of patients with a recurrence in depression may have been somewhat underestimated since 26 patients (17%) sought additional treatment between the ~st-tr~tment and follow-up period. The proportions of these patients were equally distributed across treatment groups, x2 (df= 3, n = 26) c 1.00, P > 0.i. Thirteen of these patients were in the unremitted group (Profile 9) while the remainder were distributed approximately evenly across the other profiles.

Cluster analyses

Dynamic cluster analysis using Ward’s (1963) method was performed separately for each of the treatment groups. The clustering variabies were the BDI scores at pre-treatment, ~st-t~tment and follow-up. A two-cluster solution was obtained for the psychothempy group and three-cluster solutions for each of the remaining groups. The mean BDI scores for each cluster are presented in Fig. 1, along with the number of patients in each cluster. The cluster analyses replicated the findings obtained with the criteria of Frank et al. (1991). That is, the outcome profiles were similar across treatments and recovered and unremitted clusters where identified for each treatment group. There was no evidence that remission followed by relapse constituted a major outcome profile for any treatment group. However, some worsening after partial remission was found in clusters consisting of more severely depressed patients (Fig. 1). As the figure shows, once subjects became asymptomatic they tended not to have a recurrence of depression. Behaviour therapy and amitriptyline groups were associated with more patients in the ‘rccovemd’ clusters.

Discriminating recovered from unremitted patients

Table 2 and Fig. 1 clearly suggest a bimodality in the distribution of treatment responses. That is, regardless of treatment most patients tended to have either a sustained recovery or remained unremitted. It may be that nonresponders to one treatment (e.g. amit~ptyline) would have been responders to another treatment (e.g. bchaviour therapy). Alternatively, the findings may have revealed a subgroup of unipolar depressives that would have been unresponsive to any of the treatments. To further characterize the nature of these profiles we compared the patients in Profile 1 (sustained recovery) with the patients in Profile 9 (sustained nonremission) on the variables shown in Table 3. Patients with missing data were eliminated from the discriminant analysis, leaving a sample of 63. The number of patients in the psychotherapy, relaxation, behaviour therapy and amitriptyline groups were as follows. Profile 1: 5, 5.9 and 8. Profile 9: 12, 11, 5 and 8. The interactions between the profile type (1 vs 9) and treatment were nonsignificant at P 7 0.1 for all variables except extraversion, which was nonsignificant at P 7 0.057. Since the interactions were not significant, patients were pooled across treatment groups and di~minant analysis was performed to identify the variables that were most important in distinguishing patients who responded to any treatment from patients who failed to respond.

Table 3 shows the means for each variable and the discriminant function loadings. The hit rate was 78% and three predictors were significant in discriminating profiles I and 9. As suggested by the pattern of loadings in Table 3, these were pre-treatment BDI, F(l.55) = 7.29, P -z 0.009). duration (months) of the index episode, F(l,55) = 7.04, P < 0.01 and neuroticism, F(1,55) = 3.93, P c 0.053. Taken with the means shown in Table 3, it can be seen that compared with the patients who responded to any of the treatments, the nonresponders were more neurotic, depressed, had been depressed for longer. The therapists’ outcome predictions did not discriminate profiles 1 and 9, F(1,63) = 0.01, P > 0.1, and the accuracy of predictions did not differ across treatments, as indicated by the nonsi~ifi~nt profil~by-t~trn~t interaction, F(3,63) = 1.38, P 7 0.1.

Table 2. Percentage of patients in each treatment group with a given outcome profile (number of Ss in parentheses)

Treatment group

Qutcomc profik Psychotherapy Relaxation Rehaviour

th-PY Amitriptyline

1. Recovery, sustained 2. Remission -+ partial recurrence 3. Remission -+ recovery

4. Partial remission + full remission 5. Partial remission, sustained 6. Partial remission -+ recurrence

7. Unremitted + remission 8. Unremitted + partial remission

Unremitted, sustained 9.

15&l $9 (7) 23 (9) 130) S(2) 5 (2) 0 (0) 50) 8 (3) 30)

8 (3) 5 (2) 5 (2) 8 (3) 5 (2) 8 (3) to(4) 10 (4) 8 (3) 14(S) 30) t3c3l

8 (31 0 (0) 3 (1) 5 (21 5 (2) s (2) 5 (2) 8 (3)

38(H) 38 (14) 23 (9) 2s (IO)

Group ns (39) (37) W) (39)

328 CASE HISTORIES AND SHORTER COMMUNICATIONS

Psychotherapy

&, “=,7

\, *- n=22

Relaxation

training

0 L . Crll, n=6

n = 13

n = 18

Behaviour

therapy Amitriptyline

25

Pre Post F.U. Pre Post F.U

Fig. I. Mean BDI scores at pre-treatment, post-treatment and 3-month follow-up for the clusters of patients in each treatment condition.

DISCUSSION

The Frank et al. (1991) criteria and the cluster analyses yielded similar results. Two outcome profiles-sustained recovery and nonremission-characterized the majority of patients. These results were consistent across treatment groups. Therapist pre-treatment predictions did not discriminate recovered from unremitted patients. Regardless of the type of therapy, treatment nonresponders were characterized by a greater severity and duration of depression, and more personality disturbance (neuroticism). Similar findings were reported by Pilkonis and Frank (I 988) and Sotsky, Glass, Shea, Pilkonis, Collins, Elkin, Watkins, Imber, Leber, Moyer and Oliver (1991). It is notable that variables that are predictive of the

Table 3. Results of discriminant analysis for distinguishing recovered patients (profile I) from chronically unremitted patients (profile 9)

Variable

Mean or %

Recovered Unremitted (Profile I) (Profile 9)

Loading on discriminant

function

Sex (% female) Age (yr) Marital status (% married) Employment status (% employed) Pre-treatment BDI Duration of current episode (months) Mean number of life events” Age at first depressive episode (yr) Family history of depression (% yes) Number of previous inpatient admissions Number of previous treated episodes EPO-Extraversion EPO-Neuroticism EPQ-Psychoticism

63% 37.4 63% 63% 24.3 12.5 5.3

25.6 28%

I.1 I.5 a.7

17.6 3.7

78% 0.22 39.3 0.12 47% -0.21 60% -0.04 29.2 0.48 20.2 0.39 4.1 -0.28

21.2 0.07 34% -0.04

1.2 -0.19 I.1 0.11 7.1 -0.22

19.3 0.29 3.5 -0.06

’ Weighted arithmetic mean of life events during treatment period (2.5 mos) and follow-up period (3 mos).

CASE HISTORIES AND SHORTER COMMUNICATIONS 329

vulnerability to depression, such as sex, marital status, and family history of depression, did not distinguish treatment responders from nonresponders.

Depression recurrence (i.e. initial improvement followed by exacerbation) did not emerge as a major outcome profile even though the proportion of recurrence in the present study was similar to that reported in previous studies over comparable follow-up periods (1522%: Keller, Shapiro, Lavori & Wolfe, 1982; Kupfer & Frank, 1987). Regardless of treatment, patients who were initially severely depressed tended to lose their slight treatment gains over the 3 month follow-up period. Patients who were moderately depressed at intake tended to be in remission at post-treatment, and these gains were maintained at follow-up. It is not known whether the nonresponders in the present study would have responded if they had been changed to another treatment. However, there are two reasons to suspect that our nonresponders were probably refractory to many forms of treatment. First, these patients were most likely to seek additional treatments during the follow-up period, yet they did not benefit from this. Second, the predictors of nonresponse did not vary over treatments, suggesting that the nonresponders for one treatment were similar to the nonresponders for other treatments. Further research is needed to characterize the patients who are refractory to both psychological and pharmacological interventions.

Our results failed to identify outcome profiles that are unique to any form of treatment. That is, the treatments differed in efficacy, but not in pattern. It remains to be Seen whether this conclusion holds for therapies not included in the present study, such as interpersonal psychotherapy (Klerman, Weissman, Rounsaville & Chevron, 1984), and whether treatment- unique outcome patterns can be identified with a longer-term follow-up. With a longer follow-up period it may be that a larger proportion of patients have relapse or recurrence profiles. Unfortunately, such studied are difficult to conduct since longer follow-up periods are associated with a greater incidence of subsequent treatment seeking. For the patients in the present study, for example, it was found that 57% of patients sought treatment over a 2.25 yr follow-up period (McLean & Hakstian, 1990). The development of methods to statistically correct for this effect may prove helpful in subsequent studies of outcome profiles.

REFERENCES

Beck, A. T., Ward, D., Mendelson, M., Mock, J. & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561-571.

Beck, A. T., Rush, A. J., Shaw, B. F. & Emery, G. (1979). Cognitive therapy of depression. New York: Guildford. Elkin, I., Shea, T., Watkins, J. T., Imber, S. D.[ Sotsky, S.-M., Collins, J. F,G&b. R., Pilkonis, P. A., Leber, W. R.,

Dochertv. J. P.. Fiester, S. J. & Parloff, M. B. (1989). National Institute of Mental Health Treatment of Denression Collaborative Research Program: general effectivene$s of treatments. Archiues of General Psychiatry, 46, 971:982.

Eysenck, H. J. & Eysenck, S. B. G. (1976). Eysenck Personality Questionnaire manual. San Diego, CA: Educational and Industrial Testing Service.

Feighner, J. P., Robins, E., Guze, S. B., Woodruff, R. A., Winokur, R. A. & Munoz, R. (1972). Diagnostic criteria for use in psychiatric research. Archives of General Psychiarry, 26, 57-63.

Frank, E., Prien, R. F., Jarrett, R. B., Keller, M. B., Kupfer, D. J., Lavori, P. W., Rush, A. J. & Weissman, M. M. (1991). Conceptualization and rationale for consensus definitions of terms in major depressive disorder. Archives of General Psychiatry, 48, 851-855.

Hathaway, S. R. & McKinley, J. C. (1951). Minnesota Multiphasic Personality Inventory manual. New York: Psychological Corporation.

Holmes, T. H. & Rahe, R. H. (1967). The social readjustment rating scale. Journal of Psychosomatic Research, I I, 213-218. Hollon, S. D., Shelton, R. C. & Loosen, P. T. (1991). Cognitive therapy and pharmacotherapy for depression. Journal of

Consulting and Clinical Psychology, 59, 88-99. Keller, M. B., Shapiro, R. W., Lavori, P. W. & Wolfe, N. (1982). Relapse in major depressive disorder: analysis with life

tables. Archiues of General Psychiatry, 39, 91 l-915. Klerman, G. L., Weissman, M. M., Rounsaville, B. J. & Chevron, E. S. (1984). Interpersonal psychotherapy of depression.

New York: Basic Books. Kupfer, D. J. & Frank, E. (1987). Relapse in recurrent unipolar depression. American Journal of Psychiatry, /44, 86-88. L&in, B. (1965). Adjective checklists for measurement of depression. Archives of General Psychiatry, 12, j7-62. Marmor, J. (1973). Psychiatry in transition. New York: Brunner/Mazel. Marmor, J. (1975). The nature of the psychotherapeutic process’ revisited. Canadian Psychiatric Association Journal, 20,

557-565. McLean, P. D. (1976). Decision-making in the behavioral management of depression. In P. 0. Davidson (Ed.), Behavioral

management of anxiety, depression, and pain (pp. 54-89). New York: Brunner/Mazel. McLean, P. D. & Hakstian, A. R. (1979). Clinical depression: comparative efficacy of outpatient treatments. Journal of

Consulring and Clinical Psychology, 47, 818-836. McLean, P. D. & Hakstian, A. R. (1990). Relative endurance of unipolar depression treatment effects: longitudinal

follow-up. Journal of Consulting and Clinical Psychology, 58, 482488. McLean, P. D. & Taylor, S. (1992). Severity of unipolar depression and choice of treatment. Eehauiour Research and

Therapy, 30, 443-45 1. Morris, R., Blashfield, R. & Sal& P. (1981). Neuropsychology and cluster analysis: potentials and problems. Journal of

Clinical Neuropsychology, 3, 79-99. Pilkonis, P. A. & Frank, E. (1988). Personality pathology in recurrent depression: nature, prevalence, and relationship to

treatment response. American Journal of Psychiatry, 145, 435-l. Prien, R. F., Carpenter, L. L. & Kupfer, D. J. (1991). The definition and operational criteria for treatment outcome of

major depressive disorder. Archives of Genera/ Psychiatry, 48, 796800. Prochaska, J. O., Velicer, W. F., Guadagnoli, E., Rossi, J. S. & DiClemente, C. C. (1991). Patterns of change: dynamic

typology applied to smoking cessation. Mulfiuariate Behavioral Research, 26, 83-107. Sotsky, S. M., Glass, D. R., Shea, M. T., Pilkonis, P. A., Collins, J. F., Elkin, I., Watkins, J. T., Imber, S. D., Leber, W. R.,

Moyer, J. & Oliver, M. E. (1991). Patient predictors of response to psychotherapy and pharmacotherapy: findings in the NIMH treatment of depression collaborative research program. American Journal of Psychiatry, 148, 997-1008.

330 CASEHIS~~~UE~ ANDSHORTERC~~~~WNICA~ONS

Ward, J. H. (1963). Hierarchical grouping to optimize an objective function. Journal ofthe American S!arisfica/ Association, S& 236-244

Wolberg, L. R. (1967). Short-term ~s~chQtherupy. New York: Grune & Stratton.