Double-Blind Switch Study of Imipramine Or Sertraline Treatment of Antidepressant-Resistant Chronic Depression

ORIGINAL ARTICLE Double-blind Switch Study of Imipramine or Sertraline Treatment of Antidepressant-Resistant Chronic Depression

Michael E. Thase, MD; A. John Rush, MD; Robert H. Howland, MD; Susan G. Kornstein, MD; James H. Kocsis, MD; Alan J. Gelenberg, MD; Alan F. Schatzberg, MD; Lorrin M. Koran, MD; Martin B. Keller, MD; James M. Russell, MD; Robert M. A. Hirschfeld, MD; Lisa M. LaVange, PhD; Daniel N. Klein, PhD; Jan Fawcett, MD; Wilma Harrison, MD

Background: Although various strategies have been pro- verse effects. Although sertraline treatment also re- posed to treat antidepressant nonresponders, little con- sulted in significantly higher response rates in the intent- trolled research has been published that examines pro- to-treat samples (60% in the sertraline group and 44% spectively the use of switching to an alternate antidepressant. in the imipramine group), neither the intent-to-treat remission rates nor the response and remission rates among Methods: This was a multisite study in which outpa- study completers differed significantly. Moreover, after tients with chronic major depression (with or without considering the effect of attrition, there were no signifi- concurrent dysthymia), who failed to respond to 12 weeks cant treatment effects on the more comprehensive gen- of double-blind treatment with either sertraline hydro- eralized estimating equation analyses of the continuous chloride (n=117) or imipramine hydrochloride (n=51), dependent measures. were crossed over or switched to 12 additional weeks of double-blind treatment with the alternate medication. Conclusions: More than 50% of chronically depressed Outcome measures included the 24-item Hamilton Rat- antidepressant nonresponders benefited from a switch ing Scale for Depression and the Clinical Global Impres- from imipramine to sertraline, or vice versa, despite a high sions–Severity and Improvement scales. degree of chronicity. As in the initial trial, sertraline was generally better tolerated than imipramine. Switching to Results: The switch from sertraline to imipramine (mean a standard antidepressant of a different class is a useful dosage, 221 mg/d) and from imipramine to sertraline treatment strategy for antidepressant nonresponders and (mean dosage, 163 mg/d) resulted in clinically and sta- could be considered a standard of comparison for fu- tistically significant improvements. The switch to ser- ture studies of novel alternate strategies. traline treatment was associated with fewer adverse effect complaints and significantly less attrition owing to ad- Arch Gen Psychiatry. 2002;59:233-239

T LEAST 40% of patients is important to document the efficacy of each treatedfordepressiondonot medicationclass,insequence,whentheother respond to the initial trial of has failed. antidepressant medication.1 Despite the public health importance ThaseandRush2,3 character- of antidepressant nonresponse, surpris- izedA this as stage I antidepressant resistance, ingly few switch studies2,3 have been con- the first of a sequence of 5 stages defined by ducted. Only a handful have confirmed pro- treatment history. Numerous treatment op- spectively the adequacy of the ineffective tions are available for such individuals, in- medication trial or have used controlled, cluding switching between or within classes double-blind methods. We report herein the of medications, as well as various augmen- results of a large, double-blind study of pa- tation strategies. In fact, the major issue fac- tients with chronic forms of major depres- ing clinicians is trying to decide which “next sion who, after failing to respond to 12 step” strategy to choose.3,4 Some clinicians weeks of double-blind treatment with ser- favoraugmentationstrategiesbecauseofease traline hydrochloride or imipramine hy- ofimplementation(eg,notaperingisneeded), drochloride, were crossed over or switched whereas others view switching to a different to the alternate compound. antidepressant, especially one from a differ- 3 ent class, as a more prudent option. In an RESULTS era in which selective serotonin reuptake in- Author affiliations are listed hibitors and tricyclic antidepressants are the Of the 207 nonresponders who com- at the end of this article. most widely prescribed antidepressants,5-7 it pleted the acute-phase trial, 168 (81%) be-

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, MAR 2002 WWW.ARCHGENPSYCHIATRY.COM 233

©2002 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/02/2021 SUBJECTS AND METHODS INITIAL TREATMENT TRIAL After providing written informed consent, a complete medi- SUBJECTS cal history, physical examination, electrocardiography, and laboratory screening battery were completed to confirm This protocol was part of a multiphase, collaborative re- medical eligibility. Patients began a 1-week, single-blind search program studying chronic depressive disorders. A placebo lead-in, during which the only psychotropic medi- full description of the study design, rationale, and meth- cations permitted were chloral hydrate or temazepam, used ods has been previously published.8 Briefly, outpatients aged sparingly for severe insomnia. Patients whose Clinical Global 21 to 65 years were eligible to enroll at 1 of 12 centers if Impressions–Improvement (CGI-I)14 score was 3 or higher they met DSM-III-R criteria for chronic major depressive and whose score on the 24-item Hamilton Rating Scale for disorder (ie, current major depressive episode of Ն2 years Depression (HAM-D)15 was 18 or higher at the end of the in duration) or “double depression” (ie, a current major lead-in period were randomized to double-blind treat- depressive episode superimposed on an antecedent dys- ment with either sertraline (n=426) or imipramine (n=209). thymic disorder). The findings of the main study9-11 have This 2:1 proportion was used because responders to ser- been described elsewhere. traline, but not imipramine, could subsequently partici- Intake diagnoses were based on interviews using pate in a placebo-controlled maintenance-phase trial.11 the Structured Clinical Interview for DSM-III-R12 and All patients were seen initially for 6 weekly visits, Structured Clinicial Interview for DSM-III-R Personality followed by visits every other week for the balance of the Disorders.13 Patients were excluded if they had organic 12-week trial. Imipramine was started at 50 mg/d and ti- mental syndromes, bipolar disorder or cyclothymia, trated by 50 mg/d per week, as tolerated, up to a maxi- schizophrenia or other psychotic disorder, obsessive mum of 300 mg/d. Sertraline was also initiated at 50 mg/d, compulsive disorder, or schizotypal, antisocial, or severe but a dosage increase was not permitted until the end of borderline personality disorder. Those with principal week 3. Thereafter, weekly increases of 50 mg/d were pos- current diagnoses of panic, generalized anxiety, or post- sible, if indicated and tolerated, up to a maximum of 200 traumatic stress disorders were also excluded. Patients mg/d. Therefore, maximum dosages of either study drug were not eligible if they had abused alcohol or other could be reached by week 6. Patients who could not tol- drugs within 6 months or had experienced bulimia or erate at least 50 mg/d of either study medication were with- anorexia nervosa within 1 year of intake. Patients con- drawn from the trial. Mean±SD dosages of 141±59 mg/d sidered to be an immediate suicide risk, to have medical of sertraline and 200±82 mg/d of imipramine were achieved contraindications to antidepressant therapy, or to have at the end point of the initial acute-phase trial. significant, unstable general medical disorders were also excluded. Patients who had not responded previously to ASSESSMENT OF RESPONSE minimally adequate trials of sertraline or imipramine (ie, at least 4 weeks of Ն50 mg of sertraline or Ն150 mg of Vital signs and adverse events (volunteered or observed) were imipramine daily) were ineligible. Patients could not assessed at each visit. The principal dependent measures, the have been treated with anxiolytic or other antidepressant CGI-I (Improvement) and the 24-item HAM-D, were obtained medication within 2 weeks, monoamine oxidase inhibi- at each visit. Secondary measures included the CGI-S (Sever- tors within 3 weeks, fluoxetine hydrochloride within 1 ity), Montgomery-A˚ sberg Depression Rating Scale,16 Cornell month, electroconvulsive therapy within 3 months, or Dysthymia Scale,17 and self-report 21-item Beck Depression depot neuroleptics within 6 months. Psychotherapy was Inventory.18 All clinical ratings were completed by a blinded, permitted during the study only if it had been ongoing independent evaluator. A satisfactory therapeutic response for at least 3 months before intake. (hereafter referred to as response) was noted if the final CGI-I

gan the alternate antidepressant. There were no signifi- and sertraline received at the end point of the second trial cant differences in sociodemographic or clinical measures were 221±84 mg/d and 163±48 mg/d, respectively. between those who did or did not participate in the second treatment trial. Reasons for not enrolling in the switch ATTRITION AND TOLERABILITY phase were not recorded systematically. However, several patients improved significantly during the washout Thirty-four patients dropped out, including 10% (5/51) period, and 1 patient experienced a significant clinical of those switched to sertraline treatment and 25% worsening necessitating emergency hospitalization. (29/117) of those switched to imipramine treatment ␹2 ( 1=4.94, P=.03). This difference was largely the re- SAMPLE CHARACTERISTICS sult of a significantly greater number of patients discon- tinuing imipramine because of intolerable adverse ef- Fifty-one patients were switched from imipramine to ser- fects (9% vs 0%; Fisher exact test, P=.04). traline, and 117 sertraline nonresponders were switched Adverse effects reported during the first and second to imipramine. These groups did not differ significantly on antidepressant trials are summarized in Table 2. Anticho- any relevant characteristic (Table 1). There was, how- linergic adverse effects, including dry mouth, constipation, ever, a trend for women to be overrepresented in the group and urinary complaints, were significantly more common switched from imipramine to sertraline (78% vs 62%; in the group that switched to imipramine. There were also ␹2 1=3.50, P=.06). The mean±SD dosages of imipramine significantly more complaints of sweating, dizziness, and

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, MAR 2002 WWW.ARCHGENPSYCHIATRY.COM 234

©2002 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/02/2021 score was 1 or 2, there was at least a 50% reduction in total alternate medication. Visits, assessments, medication HAM-D score to a final score of 15 or lower, and the CGI-S dosaging, and outcome definitions for the switch study score was 3 or lower (ie, the patient was no more than mildly followed the same protocol as that used in the initial acute- ill). Patients not meeting these criteria were considered non- phase trial. responders. A full remission was defined as a final HAM-D score of 7 or lower and a CGI-I score of 2 or lower. STATISTICAL ANALYSES

ACUTE-PHASE OUTCOME Categorical analyses (eg, adverse effects, attrition, and response rates) were performed using Fisher exact probabil- Among the 635 randomized patients, 302 completed the ity test, simple ␹2 tests, McNemar test (for within-subject 12-week acute-phase trial as responders (205 for sertra- comparisons), or Cochran-Mantel-Haenszel ␹2 test, as ap- line and 97 for imipramine). Two hundred seven com- propriate. The latter test was stratified by study site and pleted the study as nonresponders. One hundred twenty- depression subtype. All statistical tests used 2-tailed prob- six patients discontinued the study prematurely (76 in the ability values, with unadjusted significance levels of PՅ.05. sertraline-treated group and 50 in the imipramine-treated Response and remission rates are reported at the end point group), of whom 22 patients were responders at the time for the intent-to-treat sample and for study completers. of study discontinuation (15 in the sertraline-treated group The original data analysis plan called for analyses of and 7 in the imipramine-treated group) and of whom 12 covariance of the continuous dependent measures for the patients discontinued before the first evaluation after ran- intent-to-treat sample, using the last observation carried domization (5 in the sertraline-treated group and 7 in the forward method, and the completers sample. The avail- imipramine-treated group). There was no difference in re- ability of more sophisticated statistical methods, coupled sponse rates between treatments in the intent-to-treat with a significant, nonrandom difference in study comple- samples (52% in the sertraline group and 51% in the imip- tion rates, led us to pursue a different analytic strategy. ramine group).10 Of the 311 nonresponders, attrition was Tabulated results of the more conventional analyses of significantly greater from imipramine (41% [43/105]) than covariance are available from one of us (M.E.T.). ␹2 sertraline (30% [61/206]) ( 1=4.02, P=.045). To combine information for continuous outcome measures from all patient visits in a single analysis, repeated- SWITCH PROTOCOL measures models were fit using generalized estimating equation (GEE) methods.19,20 An independent correlation model Two hundred seven nonresponders completed the initial was assumed for the analyses reported herein. Separate GEE trial. The 62 imipramine nonresponders received a models were fit for each dependent measure, using data from mean±SD final dosage of 235±66 mg/d (range, 100-300 all available visits and including main effects for medication mg/d). The 145 sertraline nonresponders received a group and week of treatment. Because missing data were not mean±SD final dosage of 158±57 mg/d (range, 50-200 randomly dispersed (eg, virtually all missing data were the mg/d). In both cases, nonresponders received signifi- result of premature attrition, which was associated with poorer cantly higher mean±SD dosages than responders (205 ser- outcomes), we also included completion status as a main effect traline responders: 144±55 mg/d, F1,348=6.23, P=.01; 97 in the GEE analyses. Although repeated-measures analysis imipramine responders: 218±72 mg/d, F1,157=4.30, P=.04). of such data could lead to biased results (ie, informative cen- Nonresponders were tapered off study medication over soring), adjustment for completion status may partially re- 1 to 2 weeks, with the double blind maintained. After at duce this risk.21 z Score statistics based on GEE empirical SEs least 1 week of not taking any study medications, patients were used to quantify main effects and interaction terms (in who continued to meet nonresponse criteria and who gave the context of repeated-measures models), controlling for base- renewed written informed consent were crossed over to the line severity, study site, and type of depression.

somnolence in the imipramine group. Diarrhea and insom- while the imipramine intent-to-treat response rate was 44% ␹2 nia were significantly more common among the group that (52/117) ( 1=4.90, P=.03). Remission rates were 32% for ␹2 switched to sertraline therapy. Other common adverse sertraline and 23% for imipramine ( 1=2.28, P=.13). effects, including headache, nausea, fatigue, and sexual Among the 134 completers, response rates were 63% in the dysfunction, were comparable in both groups. sertraline group and 55% in the imipramine group ␹2 Comparing adverse effect reports between the 2 treat- ( 1=1.96, P=.16). Remission rates also did not differ sig- ment phases revealed that patients who switched from nificantly among completers (35% for sertraline and 30% ␹2 sertraline to imipramine experienced significant reduc- for imipramine; 1=0.94, P=.33). tions in 3 adverse effects, but significant increases in 8 adverse effects (columns 2 through 4 of Table 2). In con- ANALYSES OF CONTINUOUS MEASURES trast, the switch from imipramine to sertraline resulted in statistically significant reductions in 7 adverse ef- Mean scores at baseline and end point and mean change fects, but no significant increase in any adverse effect col- scores for continuous clinical outcome measures are pre- umns 5 through 7 of Table 2). sented in Table 3. Results of the repeated-measures GEE analyses indicated that the patients in both treatment RESPONSE AND REMISSION RATES groups improved significantly across time, and com- Response status could not be determined for 1 patient. The pleters improved significantly more than noncom- sertraline intent-to-treat response rate was 60% (30/50), pleters on all measures. After averaging across the study

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, MAR 2002 WWW.ARCHGENPSYCHIATRY.COM 235

Total Sertraline → Imipramine Imipramine → Sertraline Characteristic (N = 168) (n = 117) (n = 51) Mean (SD) age, y 40.5 (10.2) 40.9 (9.8) 39.6 (11.1) Female sex 112 (66.7) 72 (61.5) 40 (78.4) White ethnicity 156 (92.9) 109 (93.2) 47 (92.2) Employed 129 (76.8) 91 (77.8) 38 (74.5) Marital status Married 53 (31.5) 39 (33.3) 14 (27.5) Divorced or separated 46 (27.4) 32 (27.4) 14 (27.5) Never married 59 (35.1) 42 (35.9) 17 (33.3) Other 9 (5.4) 4 (3.4) 5 (9.8) Education High school or less 38 (22.6) 27 (23.1) 11 (1.6) Some college 60 (35.7) 41 (35.0) 19 (37.3) College graduate or more 67 (39.9) 47 (40.2) 20 (39.2) Major depressive disorder Mean (SD) age of onset, y 24.4 (12.3) 24.8 (12.6) 23.6 (11.8) Mean (SD) duration of current episode, y 6.1 (7.1) 5.6 (7.6) 7.3 (5.7) Recurrent 100 (59.5) 69 (59.0) 31 (60.8) Chronic major subtype 84 (50.0) 54 (46.2) 30 (58.8) Dysthymia Double depression subtype 84 (50.0) 63 (53.8) 21 (41.2) Mean (SD) age of onset, y 16.2 (13.0) 16.4 (12.8) 15.6 (13.8) Mean (SD) initial acute phase HAM-D 25.3 (5.3) 25.4 (5.3) 25.1 (5.3) Mean (SD) week-12 acute phase HAM-D 21.7 (6.0) 21.7 (6.1) 21.6 (5.8)

*Data are given as number (percentage) unless otherwise indicated. HAM-D indicates Hamilton Rating Scale for Depression.

weeks and adjusting for completion status, baseline value, customary 4-6 weeks) and by permitting imipramine and site, and depression type, there were no significant differ- sertraline dosages to be increased gradually, if tolerated, ences between the medication groups for each outcome. to the highest levels permitted within the manufactur- In addition, when the effect of attrition was considered, there ers’ therapeutic indications. were no significant week-by-medication group interac- Compared with augmentation strategies for antide- tions, indicating that the groups did not differ in their pat- pressant nonresponders, switching antidepressant classes tern of symptomatic improvement over time. has the advantages of simplicity and parsimony (eg, lower cost and little risk of drug-drug interactions). Given the COMMENT tolerability problems observed in the group that switched from sertraline to imipramine, it may be more useful to This double-blind switch study provides further evi- switch within the selective serotonin reuptake inhibi- dence that chronic depressions are responsive to anti- tors class26,27 or to a dissimilar newer antidepressant (eg, depressant monotherapy after nonresponse to a vigor- bupropion hydrochloride, mirtazapine, nefazodone hy- ous 12-week initial trial. Overall, more than 50% of the drochloride, reboxetine, or venlafaxine hydrochloride).3 patients who began treatment with a second antidepres- Augmenting the initial antidepressant with lithium sant responded. These results are particularly notewor- salts,28,29 thyroid hormone,29 pindolol,30 or buspirone hy- thy as the patients had a mean duration of more than 6 drochloride31 is also an option. Although it is unlikely years of continuous major depressive disorder. that the augmentation strategies would have produced As in the initial trial,10 sertraline was significantly better outcomes than those observed in the present study, better tolerated than imipramine. The better tolerability some clinicians favor them because of quicker imple- of sertraline was reflected by a significantly lower attri- mentation. Combining antidepressants (eg, adding ser- tion rate owing to adverse effects and a greater reduc- traline to imipramine, or vice versa) is another widely tion in adverse effect burden following the switch to the used approach, although there is the potential for drug- alternate antidepressant. The difference in tolerability also drug interactions and there are no large, well- helped to explain a significantly higher intent-to-treat re- controlled, prospective studies of this strategy.32 sponse rate in the sertraline group, although a smaller Despite a favorable antidepressant response rate difference in intent-to-treat remission rates was not sta- observed following the switch of antidepressants, low per- tistically significant. centages of patients who began the second treatment trial Overall, we observed response rates that were simi- (32% in the sertraline group and 23% in the imipramine lar to those in earlier published studies2,22-25 of switch- group) achieved full remission. In other words, about one ing across antidepressant classes. Response to the sec- half of the responders had significant residual symp- ond medication was maximized in the present study by toms after 12 weeks of pharmacotherapy. Such patients providing 12 weeks of pharmacotherapy (rather than the are unlikely to have robust social recoveries,33 and they

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, MAR 2002 WWW.ARCHGENPSYCHIATRY.COM 236

©2002 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/02/2021 Table 2. Most Common Treatment-Related Adverse Events During Acute-Phase and Switch-Phase Treatment*

Sertraline → Imipramine Imipramine → Sertraline Switch-Phase Acute Switch-Phase Acute Switch-Phase Treatment Treatment Treatment Within Treatment Treatment Within Comparison Adverse Event (Sertraline) (Imipramine) Subject† ␹2 (Imipramine) (Sertraline) Subject† ␹2 (␹2 or P Value) Headache 46.2 39.3 1.52 29.4 41.2 2.25 0.05 Dry mouth 37.6 83.8 47.03‡ 80.4 31.4 21.55‡ 44.69‡ Nausea 32.5 23.9 3.13 29.4 27.5 0.05 0.24 Insomnia 29.1 18.8 4.80§ 25.5 39.2 2.88 7.89࿣ Diarrhea 28.2 8.5 17.07‡ 11.8 25.5 2.88 8.63࿣ Somnolence 26.5 30.8 0.61 39.2 15.7 8.00࿣ 4.18§ Increased sweating 17.9 38.5 13.09‡ 27.5 7.8 8.33࿣ 16.12‡ Fatigue 17.1 15.4 0.20 15.7 19.6 0.33 0.46 Constipation 16.2 38.5 17.79‡ 41.2 9.8 11.64‡ 13.95‡ Dizziness 16.2 33.3 10.53‡ 45.1 17.6 10.89‡ 4.28§ Abdominal pain 14.5 6.0 6.25§ 3.9 3.9 0.00 P = .72 Dyspepsia 14.5 18.8 0.86 21.6 19.6 0.11 0.02 Sexual dysfunction 7.7 6.8 0.14 5.9 3.9 0.33 P = .73 Anxiety 11.1 5.1 3.77 7.8 2.0 3.00 P = .68 Flatulence 11.1 10.3 0.06 3.9 7.8 0.67 P = .78 Nervousness 11.1 6.8 1.92 21.6 11.8 2.27 P = .36 Tremor 6.8 17.9 8.90࿣ 27.5 13.7 3.77 0.46 Palpitation 4.3 1.7 1.29 11.8 5.9 1.29 P = .16 Abnormal taste 2.6 8.5 7.00࿣ 11.8 3.9 4.00§ P = .35 Increased appetite 1.7 10.3 8.33࿣ 7.8 3.9 0.67 P = .23 Urinary complaints 1.7 17.9 19.00‡ 11.8 3.9 2.67 5.91§ Postural hypotension 0.9 5.1 3.57 13.7 2.0 6.00§ P = .68

*All data are given as percentages unless otherwise indicated. Incidence of adverse events Ն10% that started in either the acute or switch phase for the sertraline hydrochloride and imipramine hydrochloride treatment groups (“other known causes” and “concurrent illness” omitted). 2 †␹ 1 Values for within-subject changes in adverse events comparing the first and second medication trials are from McNemar test. P values in column 8 report results of Fisher exact probability tests. ‡PՅ.001. §PՅ.05. ࿣PՅ.01.

Table 3. Mean Scores on Continuous Dependent Measures at Baseline and End Point and Results of Repeated-Measures Analyses Using a Generalized Estimating Equation (GEE) Model Method*

Switch-Phase Treatment

Imipramine Hydrochloride (n = 117) Sertraline Hydrochloride (n = 50) z Scores and P Values Based on GEE GEE Model GEE Model Model Empirical SEs Baseline End Point End Point Mean Adjusted Mean Baseline End Point End Point Mean Adjusted Mean Outcome Mean Mean Improvement Improvement Mean Mean Improvement Improvement Completion Measure (SD) (SD) (SE) (SE)† (SD) (SD) (SE) (SE)† Treatment Status HAM-D 24.6 (6.4) 15.3 (9.4) 9.3 (11.0) 6.8 (0.5) 25.0 (6.0) 12.8 (8.4) 12.1 (9.3) 7.3 (0.8) z = 0.57 z = 3.64 P = .57 PϽ.001 MÅDRS 24.8 (8.4) 14.9 (11.0) 9.9 (12.7) 7.2 (0.9) 25.8 (7.4) 13.1 (9.3) 12.8 (11.5) 7.3 (1.2) z = 0.08 z = 3.44 P = .94 PϽ.001 CDRS 38.7 (11.2) 25.3 (15.5) 13.5 (17.4) 10.1 (0.9) 39.7 (9.5) 21.9 (14.1) 17.8 (15.7) 10.4 (1.5) z = 0.21 z = 3.02 P = .83 P = .003 BDI 22.8 (8.9) 15.3 (10.9) 7.5 (10.6) 4.6 (0.9) 22.3 (10.0) 12.2 (10.2) 9.8 (10.0) 5.6 (1.3) z = 0.78 z = 2.85 P = .43 P = .004 CGI-S 4.2 (0.7) 2.9 (1.3) 1.3 (1.3) 0.8 (0.1) 4.3 (0.6) 2.5 (1.3) 1.7 (1.2) 0.8 (0.1) z = 0.04 z = 2.57 P = .96 P = .01

*All GEE models include adjustments for study site, depression subtype (double vs chronic major), and baseline severity. Time (ie, change across weeks of treatment) was significant (PϽ.001) in all models. None of the interaction terms were significant; hence, they were not included in the final models. †GEE model adjusted mean is averaged across visits. Total number of observations: Hamilton Rating Scale for Depression (HAM-D),15 1089; Montgomery-Åsberg Depression Rating Scale (MÅDRS),16 321; Cornell Dysthymia Rating Scale (CDRS),17 772; Beck Depression Inventory (BDI),18 291; Clinical Global Impressions-Severity (CGI-S),14 1473. are at high risk for subsequent relapse.34,35 It is likely that another recent study36 of chronic depression suggests that patients with chronic depression warrant longer acute- the combination of psychotherapy and pharmaco- phase trials. Indeed, 42% of the patients who responded therapy also increases the likelihood of remission. but did not remit during the initial trial converted to full Our study has several methodological limitations that remission during continuation therapy.35 Evidence from affect interpretation of the results. The first limitation is

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, MAR 2002 WWW.ARCHGENPSYCHIATRY.COM 237

©2002 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/02/2021 the lack of a placebo control group, which precludes more small compared with those of the parent trial. The pub- definitive assessment of the true magnitude of the treat- lic health importance of treatment-resistant depression ment effects. We decided against using a placebo- provides a strong justification for broader use of the switch controlled design because of concerns about withholding designs, which would facilitate the study of promising active treatment from chronically depressed antidepres- strategies for antidepressant nonresponders. sant nonresponders. Most investigators studying treatment- The findings of this study illustrate the value of resistant depression have made the same choice.3 How- switching to an antidepressant of a different class after ever, placebo has been used in several studies29-31 of the failure of first-line pharmacotherapy. Switching non- augmentation strategies, and nonspecific response rates of responders to an alternate antidepressant monotherapy 10% to 20% are typically observed. Kocsis and col- is a straightforward strategy that could be considered an leagues37 similarly found a placebo response rate of only ethical standard of comparison for future studies of treat- 13% in their study of imipramine as the initial treatment ment-resistant depression. of chronic depression. If it is assumed that 20% of our patients would have responded to placebo, the effects of Submitted for publication November 3, 1998; final revision sertraline and imipramine would have been statistically received October 19, 2001; accepted October 24, 2001. significant. From the Departments of Psychiatry, University of Pitts- A second methodological problem is that switch- burgh School of Medicine, and Western Psychiatric Insti- ing the antidepressants of only the nonresponders is not tute and Clinic, Pittsburgh, Pa (Drs Thase and Howland), representative of a true crossover design, which would The University of Texas Southwestern Medical Center at require that responders also switch medications. Be- Dallas (Dr Rush), Medical College of Virginia, Virginia Com- cause the nonresponders are no longer necessarily rep- monwealth University, Richmond (Dr Kornstein), Cornell resentative of the initial randomized samples, we can- University School of Medicine, New York, NY (Dr Kocsis), not rule out that the 2 switch groups differed in some University of Arizona, Tucson (Dr Gelenberg), Stanford Uni- meaningful way. Confidence in the validity of compar- versity School of Medicine, Stanford, Calif (Drs Schatz- ing these groups is strengthened by the lack of signifi- berg and Koran), Butler Hospital, Brown University, Provi- cant differences in the sociodemographic and clinical char- dence, RI (Dr Keller), and The University of Texas Medi- acteristics, as well as a virtual lack of predictors of cal Branch at Galveston (Drs Russell and Hirschfeld); differential response in the main trial.38 Nevertheless, with- Quintiles, Inc, Research Triangle Park, North Carolina out rerandomization of nonresponders to alternate treat- (Dr LaVange); Departments of Psychiatry, State University ments, our comparative results should be viewed as of New York at Stony Brook (Dr Klein), and Rush- tentative. Presbyterian-St Luke’s Medical Center, Chicago, Ill A third limitation is the exclusion of more compli- (Dr Fawcett); Pfizer Inc and Department of Psychiatry, Col- cated and comorbid patients from the main trial. Be- lege of Physicians and Surgeons, Columbia University, New cause such patients tend to be overrepresented among York (Dr Harrison). The following authors have served as antidepressant nonresponders,1,2 the generalizability of paid consultants to these companies: Pfizer Inc (Drs Thase, our findings is somewhat attenuated. However, there is Rush, Kornstein, Kocsis, Gelenberg, Schatzberg, Koran, no reason to assume that excluded patients would have Keller, Russell, Hirschfeld, LaVange, and Fawcett) and Ab- responded preferentially to one or the other of the study bott Laboratories, North Chicago, Ill (Drs Hirschfeld and medications. Fawcett). Drs Kocsis, Gelenberg, and Harrison own stock in Finally, it could be argued that the slower dose- Pfizer Inc. titration schedule of imipramine during the first weeks This multicenter research project was supported in part of the protocol, while necessary, might have biased the by a series of grants from Pfizer Inc. results in favor of sertraline. However, because rapidity Corresponding author and reprints: Michael E. Thase, of response did not differ between the treatments and be- MD, Department of Psychiatry, University of Pittsburgh cause all imipramine-treated patients received up to 6 School of Medicine, Western Psychiatric Institute and Clinic, weeks of therapy at maximal dosage, it is unlikely that 3811 O’Hara St, Pittsburgh, PA 15213 (e-mail: thaseme speed of titration affected the final results. Faster titra- @msx.upmc.edu). tion of imipramine also may have exaggerated the tolerability problems observed with this medication. REFERENCES To our knowledge, this is the largest double-blind study of switching antidepressants conducted to date. 1. Depression Guideline Panel. Clinical Practice Guideline Number 5: Depression Given the magnitude of the problem of antidepressant in Primary Care, 2: Treatment of Major Depression. Rockville, Md: Agency for nonresponse and the large number of controlled studies Health Care Policy and Research, Public Health Service, US Dept of Health and of antidepressants under way at any given time, it is puz- Human Services; 1993. AHCPR publication 93-0551. 2. Thase ME, Rush AJ. Treatment-resistant depression. In: Bloom FE, Kupfer DJ, zling that switch designs are not used more widely. There eds. Psychopharmacology: The Fourth Generation of Progress. New York, NY: is an impression that these studies are too difficult to con- Raven Press; 1995:1081-1097. duct, because of perceived problems such as subject re- 3. Thase ME, Rush AJ. When at first you don’t succeed: sequential strategies for fusal, attrition, and inability to maintain the double blind. antidepressant nonresponders. J Clin Psychiatry. 1997;58:23-29. We experienced no such difficulties (eg, more than 80% 4. Rush AJ, Kupfer DJ. Strategies and tactics in the treatment of depression. In: Gabbard GO, ed. Treatment of Psychiatric Disorders. 3rd ed. Washington, DC: of patients consented to the switch study), and our pa- American Psychiatric Association; 1996. tients’ outcomes were as good as those observed in the 5. Nelson JC. Are the SSRIs really better tolerated than the TCAs for treatment of initial trial. Furthermore, the costs of a switch study are major depression? Psychiatr Ann. 1994;24:628-631.

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, MAR 2002 WWW.ARCHGENPSYCHIATRY.COM 238

©2002 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/02/2021 6. Hotopf M, Lewis G, Norman C. Are SSRIs a cost-effective alternative to tricy- the effects of oxiprotiline and fluvoxamine. Acta Psychiatr Scand. 1988;78:668- clics? Br J Psychiatry. 1996;168:404-409. 675. 7. Thase ME, Kupfer DJ. Recent developments in the pharmacotherapy of mood 23. Thase ME, Frank E, Mallinger AG, Hamer T, Kupfer DJ. Treatment of imipramine- disorders. J Consult Clin Psychol. 1996;64:646-659. resistant recurrent depression, III: efficacy of monoamine oxidase inhibitors. 8. Rush AJ, Koran LM, Keller MB, Markowitz JC, Harrison WM, Miceli RJ, Fawcett J Clin Psychiatry. 1992;53:5-11. JA, Gelenberg AJ, Hirschfeld RMA, Klein DN, Kocsis JH, McCullough JP, 24. Nierenberg AA, Feighner JP, Rudolph R, Cole JO, Sullivan J. Venlafaxine for treat- Schatzberg AF, Thase ME. The treatment of chronic depression: study design ment-resistant unipolar depression. J Clin Psychopharmacol. 1994;14:419- and rationale for evaluating the comparative efficacy of sertraline and imipra- 423. mine as acute, crossover, continuation, and maintenance phase therapies. J Clin 25. Cole JO, Schatzberg AF, Sniffin C, Zolner J, Cole JP. Trazodone in treatment- Psychiatry. 1998;59(pt 1):589-597. resistant depression: an open study. J Clin Psychopharmacol. 1981;1:49-54. 9. Keller MB, Harrison W, Fawcett JA, Gelenberg A, Hirschfeld RMA, Klein D, Koc- 26. Thase ME, Blomgren SL, Birkett MA, Apter JT, Tepner RG. Fluoxetine treatment sis JH, McCullough JP, Rush AJ, Schatzberg A, Thase ME. Treatment of chronic in patients with major depressive disorder who failed initial treatment with ser- depression with sertraline or imipramine: preliminary blinded response rates and traline. J Clin Psychiatry. 1997;58:16-21. high rates of undertreatment in the community. Psychopharmacol Bull. 1995; 27. Brown WA, Harrison W. Are patients who are intolerant to one serotonin selec- 31:205-212. tive reuptake inhibitor intolerant to another? J Clin Psychiatry. 1995;56:30-34. 10. Keller MB, Gelenberg AJ, Hirschfeld RMA, Rush AJ, Thase ME, Kocsis JH, Marko- 28. Stein G, Bernadt M. Lithium augmentation therapy in tricyclic resistant depres- witz JC, Fawcett JA, Koran LM, Klein DN, Russell JM, Kornstein SG, McCul- sion: a controlled trial using lithium in low and normal doses. Br J Psychiatry. lough JP, Davis SM, Harrison WM. The treatment of chronic depression: a double- 1992;162:634-640. blind, randomized trial of sertraline and imipramine. J Clin Psychiatry. 1998; 29. Joffe RT, Singer W, Levitt AJ, MacDonald C. A placebo-controlled comparison 59(pt 2):598-607. of lithium and triiodothyronine augmentation of tricyclic antidepressants in uni- 11. Keller MB, Kocsis JH, Thase ME, Gelenberg AJ, Rush AJ, Koran L, Schatzberg polar refractory depression. Arch Gen Psychiatry. 1993;50:387-393. A, Russell J, Hirschfeld R, Klein D, McCullough JP, Fawcett JA, Kornstein S, La- 30. McAskill R, Mir S, Taylor D. Pindolol augmentation of antidepressant therapy. Vange L, Harrison W, for the Sertraline Chronic Depression Study Group. Main- Br J Psychiatry. 1998;273:203-208. tenance phase efficacy of sertraline for chronic depression: a randomized 31. Landen M, Bjorling G, Agren H, Fahlen T. A randomized, double-blind, placebo- controlled trial. JAMA. 1998;280:1665-1672. controlled trial of buspirone in combination with an SSRI in patients with treatment- 12. Spitzer RL, Williams JBW, Gibbons M, First MB. The Structured Clinical Inter- refractory depression. J Clin Psychiatry. 1998;59:664-668. view for DSM-III-R (SCID), I: history, rationale, and description. Arch Gen 32. Fava M, Rosenbaum JF, McGrath PJ, Stewart JW, Amsterdam JD, Quitkin FM. Psychiatry. 1992;49:624-629. Lithium and tricyclic augmentation of fluoxetine treatment for resistant major 13. First MB, Spitzer RL, Gibbons M, et al. The Structured Clinical Interview for DSM- depression: a double-blind, controlled study. Am J Pyschiatry. 1994;151:1372- III-R Personality Disorders (SCID-II), I: description. J Personal Disord. 1995;9: 1374. 83-91. 33. Miller IW, Keitner GI, Schatzberg AF, Klein DN, Thase ME, Rush AJ, Markowitz 14. Guy W. ECDEU Assessment Manual for Psychopharmacology. Rockville, Md: Al- JC, Schlager DS, Kornstein SG, Davis SM, Harrison WM, Keller MB. The treat- cohol, Drug Abuse, and Mental Health Administration, Public Health Service, US ment of chronic depression, III: psychosocial functioning before and after treat- Dept of Health, Education, and Welfare; 1976. ADM publication 76-338. ment with sertraline or imipramine. J Clin Psychiatry. 1998;59:608-619. 15. Hamilton M. A rating scale for depression. J Neurol Neurosurg Psychiatry. 1960; 34. Paykel ES, Scott J, Teasdale JD, Johnson AL, Garland A, Moore R, Jenaway A, 23:56-62. Cornwall PL, Hayhurst H, Abbott R, Pope M. Prevention of relapse in residual 16. Montgomery SA, A˚ sberg M. A new depression scale designed to be sensitive to depression by cognitive therapy. Arch Gen Psychiatry. 1999;56:829-835. change. Br J Psychiatry. 1979;134:382-389. 35. Koran LM, Gelenberg A, Kornstein SG, Howland RH, Friedman RA, DeBattista C, 17. Mason BJ, Kocsis JH, Leon AC, Thompson S, Frances AJ, Morgan RO, Parides Keller MB, Klein D, Kocsis J, Schatzberg A, Thase ME, Rush AJ, LaVange L. Ser- MK. Measurement of severity and treatment response in dysthymia. Psychiatr traline vs. imipramine in continuation treatment of chronic depression. J Affect Ann. 1993;23:625-631. Disord. 2001;65:27-36. 18. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measur- 36. Keller MB, McCullough JP, Klein DN, Arnow B, Dunner DL, Gelenberg AJ, Marko- ing depression. Arch Gen Psychiatry. 1961;4:561-571. witz J, Nemeroff CB, Russell JM, Thase ME, Trivedi MH, Zajecka J. A compari- 19. Liang KY, Zeger SL. Longitudinal data analysis for discrete and continuous out- son of nefazodone, the cognitive behavioral-analysis system of psychotherapy, comes. Biometrics. 1986;43:121-130. and their combination for the treatment of chronic depression. N Engl J Med. 20. Diggle PJ, Liang KY, Zeger SL. Analysis of Longitudinal Data. Oxford, England: 2000;342:1462-1470. Oxford University Press; 1994. 37. Kocsis JH, Frances AJ, Voss CB, Mann JJ, Mason BJ, Sweeney J. Imipramine 21. Schluchter MD. Methods for the analysis of informatively censored longitudinal treatment for chronic depression. Arch Gen Psychiatry. 1988;45:253-257. data. Stat Med. 1992;11:1861-1870. 38. Hirschfeld RMA, Russell JM, Delgado PL, Fawcett J, Friedman RA, Harrison WM, 22. Nolen WA, van de Putte JJ, Dijken WA, Kamp JS, Blansjaar BA, Kramer HJ, Haff- Koran LM, Miller IW, Thase ME, Howland RH, Connolly MA, Miceli RJ. Predic- mans J. Treatment strategy in depression, I: non-tricyclic and selective reup- tors of response to acute treatment of chronic and double depression with ser- take inhibitors in resistant depression: a double-blind partial crossover study on traline or imipramine. J Clin Psychiatry. 1998;59:669-675.

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, MAR 2002 WWW.ARCHGENPSYCHIATRY.COM 239