Time course of response to antidepressants: Predictive value of early improvement and effect of additional psychotherapy

Methods 124 patients with a major depression referred for hospitalized care were randomized to 5 weeks of sertraline (or amitriptyline as a second choice) plus either additional Interpersonal Psychotherapy modified for inpatients (IPT) or Clinical Management (CM). “Improvement” was defined as a decrease of ≥ 20% on the 17-item Hamilton Rating Scale for Depression (HAMD). “Onset of response” was defined as sustained improvement (without any subsequent increase in the HAMD) culminating in 50% decrease on the HAMD by week 5. Results Early improvement within two weeks was highly predictive of later stable response (≥ 50% decrease on the HAMD at weeks 4 and 5) or stable remission (HAMD score of ≤ 7 at weeks 4 and 5), irrespective of the type of medication or additional IPT or CM. Survival analysis of the ITT sample revealed that patients of the IPT group had a shorter time to “onset of response” than patients in the CM group (median: 12 vs. 30 days; p = 0.041, Log Rank). However, there was no significant difference in the time to onset of response, when more stringent conditions were used. Limitations Due to ethical restrictions a comparison with an untreated placebo group could not be performed. Conclusions Early improvement is highly predictive for later stable response or remission in more severely depressed inpatients. In combination therapy, the additional benefit of psychotherapy occurs at least as rapid as the response to antidepressants. Keywords Interpersonal Psychotherapy Survival analysis Amitriptyline Sertraline 1 Introduction Despite considerable progress in both the pharmacological and the psychotherapeutic treatment of major depression, recurrent depressive illness remains the predominant cause of disability worldwide ( McIntyre and O'Donovan, 2004 ). Recent attempts to improve outcome have used both modalities, both sequentially (for review see Fava et al., 2003 ) or in combination (for review see Hollon et al., 2005 ). These efforts have been particularly successful in the treatment of chronic depression ( Fava et al., 2002; Keller et al., 2000 ) or with patients between 59 and 70 years of age ( Lenze et al., 2002; Reynolds et al., 1999 ). It is generally believed that the superiority of combination therapy over monotherapy is due to the complementary benefits of both modalities. These include a robust response to pharmacotherapy, particularly in more biologically driven depressions, and sustained improvements in interpersonal, social and cognitive skills with psychotherapy ( Hollon et al., 2005 ). One major problem encountered both in pharmacological and psychological treatments of depression is the apparent delay in onset of the therapeutic effect, which is traditionally thought to take several weeks for antidepressants ( Quitkin et al., 1996 ) and even longer for psychotherapy ( Hollon et al., 2005; Schulberg et al., 1996 ). There is some evidence, however, that pharmacotherapy may have more immediate effects; recent meta-analyses of double-blind, placebo-controlled studies revealed pronounced drug placebo differences in the first 2 weeks of treatment ( Papakostas et al., 2006; Posternak and Zimmerman, 2005b ). Earlier work by several groups ( Katz et al., 2004; Nierenberg et al., 2000; Stassen et al., 1993, 1996; Szegedi et al., 2003 ) also questioned the validity of the delayed onset hypothesis of the effect of antidepressants. Furthermore, it provided evidence that early improvement with antidepressive treatment is a sensitive predictor of later stable response ( Katz et al., 2004; Szegedi et al., 2003 ). It is, however, still a matter of debate whether these early improvements represent “true” drug effects or are merely due to unspecific placebo-like responses ( Posternak and Zimmerman, 2005a; Quitkin et al., 2005 ). Even less clear is the potential influence of psychotherapy on the time course of improvement in depression. Time to response appears to be longer with psychotherapy than pharmacotherapy ( DiMascio et al., 1979; Hollon et al., 2005; Schulberg et al., 1996; Watkins et al., 1993 ). To our knowledge, it has not been systematically assessed how the combination of pharmacotherapy and psychotherapy affects the time course of improvement. To answer this question, we reanalyzed data of a recent study in which the response of hospitalized more severely depressed patients to combined treatment with antidepressants and Interpersonal Psychotherapy (IPT) was compared with the response to antidepressants plus intensive clinical management (CM) ( Schramm et al., 2007 ). 2 Methods 2.1 Patients Details of the patient sample have been presented elsewhere ( Schramm et al., 2007 ). In brief, the sample consisted of 124 depressed inpatients referred by the treating psychiatrist to the Department of Psychiatry and Psychotherapy at the University Medical Centre Freiburg, Germany, for hospitalized care. Eligible were patients between 18 and 65 years with a diagnosis of Major Depression using the Structured Clinical Interview for DSM IV (SCID) ( First et al., 1997 ). Exclusion criteria were (1) a Hamilton Depression Rating Scale (HAMD) score lower than 16 on the 17-item version, (2) bipolar I disorder and bipolar II if currently hypomanic or mixed, (3) psychotic symptoms, (4) current diagnosis of primary substance abuse or dependency, (5) mental disorder due to organic factors, (6) severe cognitive impairment, (7) borderline or antisocial personality disorder, (8) contraindications to sertraline or amitriptylin, (9) being actively suicidal. The study was approved by the local ethics committee of the Medical Faculty at the University of Freiburg. After complete description of the study to the subjects, written informed consent was obtained. 2.2 Design overview 2.2.1 Randomization Patients were randomized by a computer program using the stratification variables age, gender, unipolar vs. bipolar II, comorbidity on axis I, duration of index-episode and number of episodes. The allocation sequence was not predictable to any of the investigators. 2.2.2 Treatment conditions IPT : The IPT program was conducted according to a modified version of the original IPT manual by Klerman et al. (1984) for use in an inpatient setting (IPT). The program included three weekly individual sessions of 50 minutes duration (all videotaped) and eight additional IPT group sessions. CM : The patients in the CM condition also received three weekly sessions according to a brief guideline (adapted from Reynolds and Perel, unpublished manuscript, 1988; see Schramm et al., 2007 ). CM was defined as psychoeducative, supportive and empathic intervention of 15–20 min in length and was delivered by psychiatric residents. Pharmacotherapy : Both treatment groups received standardized pharmacotherapy in addition to IPT or CM (for details see Schramm et al., 2007 ). Briefly, the first line treatment was sertraline ( n = 81 in the ITT sample; n = 68 among completers) with an initial dosage of 25 mg on day one, and 50 mg/day on days 2–14. After day 14, the dosage could be increased by the treating physician according to clinical judgment and/or plasma level with no upper limit. In case of a known non-response to sertraline in the patient's treatment history, amitriptyline or amitriptyline- N -oxide was applied as the second line treatment ( n = 35 in the ITT sample; n = 30 among completers). In the ITT sample the mean final dosage for sertraline was 90.2 mg/day (SD, 43.9; range: 50–250 mg/day) and among completers it was 86.03 mg/day (SD, 41.9; range: 50–250 mg/day). For amitriptyline or amitriptyline- N -oxide the mean final dosage in the ITT sample was 175.43 mg/day (SD, 66.9; range: 75–360 mg/day), among completers it was 182.17 mg/day (SD, 69.7; range: 75–360 mg/day). The number of patients using sertraline or amitriptyline did not differ by treatment condition. Furthermore, there was no significant difference in the mean final dosage of either type of medication between IPT and CM patients. 2.2.3 Assessments Severity of depressive symptoms : The 17-item version of the Hamilton Depression Rating Scale (HAMD; Hamilton, 1960 ) served as the primary outcome measure. Measures of treatment response and predictive value : Treatment response was defined as a 50% decrease on the HAMD-17 from baseline to endpoint; remission was defined as a final HAMD-17 score of 7 or lower ( Angst et al., 1993; Frank et al., 1990 ). “Stable response” was defined as ≥ 50% reduction in HAMD-17 score at weeks 4 and 5, “stable remission” was defined as HAMD-17 score of 7 or lower at weeks 4 and 5. To assess the time course of response to treatment we made use of the two measures suggested by Nierenberg et al. (2000) and Szegedi et al. (2003) : improvement , which is defined as a decrease in HAMD score of at least 20% ( Szegedi et al., 2003 ) or 30% ( Nierenberg et al., 2000 ) from baseline and sustained improvement defined as improvement without any subsequent increase in the HAMD score. Defining sustained improvement in this way excludes patients who demonstrate a “placebo-like pattern” of nonsustained response ( Quitkin et al., 1991 ). The time until appearance of sustained improvement (20%) that led to response (50% reduction of HAMD score at week 5) was thus considered to represent the time until onset of true response ( Nierenberg et al., 2000 ). For the assessment of the predictive value of early improvement we calculated sensitivity, specificity, negative predictive value (NPV) and positive predictive value (PPV) following the procedure of Szegedi et al. (2003) . The screening interview and the weekly HAMD ratings were performed by trained clinical psychologists. Evaluations at 5 weeks were carried out by trained, independent raters (clinical psychologists) who were blind to treatment condition. Interrater reliability for the HAMD was assessed by rating a random sample of 13 audio-or videotaped HAMD interviews and was found to be excellent (intraclass correlation coefficient 0.95). 2.3 Statistical methods Analyses of the ITT sample ( n = 124) are reported. When attrition occurred between baseline and posttreatment “last observation carried forward” was used. Survival analysis ( Kaplan and Meier, 1958 ) was used to assess the time until improvement, sustained improvement, onset of true response, and the time until response or remission (for review see Montgomery et al., 2002 ). Log-rank tests were then employed to test whether the survival curves were equivalent. The use of the original Kaplan–Meier technique for the assessment of time to response has been criticised on the grounds that the actual time to response would be distorted by the censured data from the patients that fail to respond ( Laska and Siegel, 1995 ). In order to account for this, Laska and Siegel (1995) proposed a survival cure model to characterize time to response. A further development is the 2-dimensional cure model proposed by Tamura et al. (2000) which extends the original 1-dimensional cure model to distinguish between incidence and latency, the two central aspects of psychotropic drug response. This is of major relevance as unconditional time to improvement curves combine (1) the proportion of patients who improve, with (2) the time to improvement. Therefore, between-treatment comparison tests show statistical significance for differing proportions of improvers/responders even when the time characteristics of improvement are identical. To account for this issue we made additional analyses using the model of Tamura et al. (2000) . The Cramér–von Mises statistic for two-sample comparisons of conditional survival curves ( Li and Feng, 2005 ) was used, and statistical significance assessed using the Wald test. Calculations were made in “R” ( R Development Core Team, 2004 ). For all tests performed, the significance level was set at 0.05, two-tailed. 3 Results 3.1 Patients A detailed description of the patients and the patient flow is given elsewhere ( Schramm et al., 2007 ). In brief, 130 patients were randomized, 65 to IPT plus pharmacotherapy and 65 to pharmacotherapy plus CM. Of those, 124 patients were successfully assigned (63 to IPT, 61 to CM) and 105 completed the trial (53 IPT, and 52 CM patients). The two treatment groups were comparable on background demographic and clinical characteristics. Patients of both groups were at baseline more severely depressed (CM: HAMD 21.9 ± 4.1; BDI: 30.1 ± 10.2; IPT: HAMD 25.1 ± 5.1; BDI 29.5 ± 7.9; Schramm et al., 2007 ). There were also no statistical differences between the completer ( n = 105) and the non-completer ( n = 19) sample on any of the demographic or clinical variables. 3.2 Overall response to treatment 3.2.1 IPT vs. CM Participants in both treatment conditions demonstrated significant improvement on the HAMD after 5 weeks. Analysis of covariance showed that patients treated with IPT had significantly fewer symptoms on the HAMD at week 5 than those in the CM group ( p = 0.011). The rates of response (77% vs. 58%; p = 0.02) and of remission, (58% vs. 40%; p = 0.04) differed significantly between groups, favouring therapy with IPT. The differences in response rates correspond to a number needed to treat (NNT) for ITT of 6 (95% CI 3-21) (see Schramm et al., 2007 for a complete description of response and remission rates). 3.2.2 Sertraline vs. amitriptyline There were no differences between the two medications (sertraline or amitriptyline) regarding the rate of response or remission, either in the ITT or in the completer sample ( Schramm et al., 2007 ). This is in accordance with earlier studies ( Lydiard et al., 1997; Moller et al., 1998; Reimherr et al., 1990 ). 3.3 Time to improvement and response There were no differences between the two medications in the time to improvement as assessed by Kaplan–Meier analysis ( χ 2 = 0.03, df = 1, p = 0.863 Log Rank). The median time until improvement was 9 days. After 2 weeks, 80% of the patients fulfilled this criterion, after 5 weeks 98%. Kaplan–Meier analysis also revealed no differences between the two medication groups in the time to sustained improvement ( χ 2 = 0.05, df = 1, p = 0.824; median: 12 days). After 2 weeks, 59% of the patients met this criterion, which improved to 76% after 5 weeks. Sustained improvement was highly associated with eventual response at week 5: Of the 82 patients that showed sustained improvement, 67 (82%) eventually responded. There was no difference between the two medication groups in the time to onset of response ( χ 2 = 0.6, df = 1, p = 0.416; median: 13 days). Onset of response was observed within the first 2 weeks in 51.2% of the patients of the ITT sample (83% of the responders). 3.3.1 Predictive value of early improvement We analyzed whether early improvement under treatment with sertraline or amiptriptyline was a reliable predictor of later stable response. As the response criterion, we chose 50% reduction in the HAMD-17 score that was maintained at least at weeks 4 and 5 (stable response). Following the procedure of Szegedi et al. (2003) we calculated sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of the test (early improvement) to predict later stable response and remission ( Table 1 A–D ). Early improvement turned out to be a very sensitive (but not very specific) predictor of later stable response ( Table 1 A): Thus, almost none of the patients who had not improved after 2 weeks became a stable responder after 4 weeks. However, a substantial number of patients who had shown early improvement did not become stable responders, indicating the limited specificity of the predictor. Using a 30% decrease in the HAMD values as the predictive criterion (as suggested by Nierenberg et al., 2000 ) tended to decrease sensitivity and to increase specificity ( Table 1 B). Nearly identical results were obtained when analysing response at week 5 instead of stable response at weeks 4 and 5 ( Table 1 C). When “stable remission” (i.e. a HAMD-17 total score of ≤ 7 at weeks 4 and 5) was chosen as the response criterion, the sensitivity of the predictor was even more pronounced ( Table 1 D). To illustrate the predictive value (sensitivity) of early improvement for later non-response but also its limited specificity for later stable response the data for all patients were combined and are summarized in Fig. 1 . The receiver operating characteristic (ROC) curves for percentual change in HAMD (of weeks 1, 2 and 3) as predictors of stable response are shown in the Supplementary Fig. 4 . The area under the ROC curve was for week 1 0.772 (95% CI:0.67–0.874) for week 2 0.82 (95% CI: 0.735–0.906) and for week 3 0.901 (95% CI: 0.838–0.964). The points at which sensitivity and specificity were optimal were at percentual change of 41%, 47% and 62% respectively. Since we did not find any apparent differences between the two medication groups in the occurrence of early improvement and its predictive value we combined them in order to assess the potential influence of psychotherapy on the predictive value of early improvement. As shown in Table 2 A–C , the predictive value of early improvement was similarly present both in the IPT and the CM groups, with higher sensitivity but lower specificity in the IPT group. We also analyzed, whether or not severity of depression at baseline had any influence on the predictive value of early improvement. To this end the group was splitted at the median (HAMD 24) in two subgroups of less severely (HAMD < 24) and very severely depressed patients (HAMD > 24) ( Supplementary Table 4 ). There was no difference in the sensitivity of prediction between both groups. However, in the very severely depressed patients the specificity of the predictor was even lower, i.e. even in the presence of early improvement a considerable number of patients, and particularly the very depressed, show now response. 3.3.2 Influence of psychotherapy on the time course of improvement and response Since there was no difference between the two medication groups in time to improvement, time to sustained improvement, and time to “onset of response” both were combined to assess the speed of response under treatment with IPT or CM. Classical Kaplan–Meier analysis ( Fig. 2 A–C ) revealed no significant difference between the IPT and CM groups in the time to improvement ( Fig. 2 A) ( χ 2 = 1.23, df = 1, p = 0.268, Log Rank), although there were more improvers in the IPT group, particularly in the second week ( Fig. 3 A ). However there was a highly significant difference between the two groups in the time to sustained improvement and time to onset of response. During the second week a significant greater number of patients in the IPT group exhibited sustained improvement ( χ 2 = 6.584, p = 0.01) ( Fig. 3 B) and apparent onset of response ( χ 2 = 9.677, p = 0.002) ( Fig. 3 C) than in the CM group. Kaplan–Meier analysis revealed an apparent more rapid occurrence of sustained improvement in the IPT group (median: 12 days) as compared to the CM group (median: 17 days; χ 2 = 3.95, df = 1, p = 0.047, Log Rank) ( Fig. 2 B). Similarly, time to onset of response ( Fig. 2 C) was also far more rapid in the IPT group than in the CM group (median: 12 vs. 30 days; χ 2 = 4.18, df = 1, p = 0.041, Log Rank). Similar results were obtained for a 30% reduction in the HAMD-17 as the criterion for sustained improvement and onset of response respectively (data not shown). The right panel of Fig. 2 (D–F) displays the non-parametric conditional survival distributions for the various timing measures separated by treatment. These were not statistically different by the Cramér–von Mises statistic (see Table 3 , latency). The median for the different measures ranged from 8 to 11 days. On the other hand, the estimated proportion of “sustained improvers” and of responders were statistically significantly higher for IPT than for CM (see Table 3 , incidence). 4 Discussion The results of the present study are significant with respect to two different issues: First, they add to the growing body of evidence that – in contrast to the long held delayed onset hypothesis – effects of antidepressants can be observed within the first 2 weeks of treatment. In striking similarity to data obtained with other antidepressants ( Katz et al., 2004; Nierenberg et al., 2000; Szegedi et al., 2003; Stassen et al., 2007 ), we found substantial improvements in patients treated with amitriptyline or sertraline already during the first 2 weeks. As shown previously for other antidepressants ( Katz et al., 2004; Nierenberg et al., 2000; Szegedi et al., 2003 ) these early improvements were a very sensitive predictor of later stable response or remission. In this study, very few patients without early improvement attained stable response or remission at a later time point ( Fig. 1 ). Our results reveal that the predictive value of early improvement also holds true for more severely depressed hospitalized patients with or without additional psychotherapy (see also Supplementary table 4 ). On the other hand, early improvement is not specific for later stable response or remission since a substantial number of patients who showed early improvement did not attain response or remission in the later course of therapy. These patients are probably those with a “placebo-like response pattern”, as described by Quitkin et al. (1996, 2005) . Besides confirming earlier findings for an inpatient sample, the present work provides new and unprecedented information regarding the impact of additional psychotherapy on the time course of improvement and response. Previous work has shown that psychotherapy (CBT or IPT) takes longer than pharmacotherapy to work (for review see Hollon et al., 2005 ). If the advantage of combined therapy over antidepressants alone (with CM to control for unspecific effects) reported in this study ( Schramm et al., 2007 ) was exclusively due to an additive effect of IPT and independent of the antidepressant, one would expect a delayed onset of this additional action. However, we observed that the benefit of combination treatment was already fully developed and indeed particularly pronounced in the second week of treatment ( Fig. 3 ), with no further increase in the following weeks. Additional IPT shortened the time to sustained improvement and onset of response ( Fig. 2 . left panel). This was apparently due to the large increase in both measures occurring in the second week in the combination group only, which entirely explains the higher response rate of this group at endpoint. However, the use of the ITT sample in the survival analytical assessment of time to response has been criticised, on the grounds that the ITT sample also comprises patients who might never respond to the treatment. This would lead (in the case of different overall responses in the two comparison groups) to false conclusions about the time to response for responders ( Laska and Siegel, 1995 ). The 5 weeks of treatment implemented in our study may indeed be too short to evaluate overall response since a substantial part of patients may respond only at or after 8 weeks ( Quitkin et al., 2003; Trivedi et al., 2006 ). Nonetheless, to account for the problem of potential different overall response we have in addition to the classical survival analysis also applied a modified version, which makes allowance for this issue ( Tamura et al., 2000 ). This statistical method, the “2-dimensional cure model” of Tamura et al. (2000) , allows to separate the two central aspects of psychotropic drug response, namely “incidence” (the proportion of patients in whom an effect is induced) and “latency” (the time characteristics of recovery, e.g. time to onset of improvement) and therefore allows an exact determination of the time course of improvement and response (“latency”) by use of appropriate statistics such as Cramer von Mises or Kolmogorov–Smirnov ( Stassen et al., 2007 ). The application of this advanced methodology revealed no significant differences in the time to onset of response between the two groups ( Fig. 2 right panel and Table 3 ). Thus, we conclude that the additional advantage of combination treatment is at least as rapid in onset as pharmacotherapy alone and is particularly pronounced in the second week. This finding is difficult to reconcile with a mere addition of the effect of IPT to that of the antidepressants but rather indicates a positive interaction between the two modalities. The additional benefit of combination therapy was fully developed after 6 IPT sessions representing the initial phase of IPT. In this phase hope and relief is instilled, the patient gets the sick role, and there is a mutual agreement on the therapeutic focus. This phase is crucial for the formation of the therapeutic alliance and expectation, and may be particularly important in the synergetic interaction of both treatments. Psychotherapy and even placebo are known to elicit specific alterations in neurobiological functioning which correlate with treatment response and appear to show features both common to and distinct from those evoked by pharmacotherapy (for review see Mayberg, 2003 ). Thus in combination therapy, the neurobiological actions of pharmacotherapy and psychotherapy may complement each other to induce a particular efficacious and rapid response. Several limitations of the present study should be borne in mind. The sample comprised severely depressed patients that were hospitalized because of suicidal ideation, inability to care for themselves and/or non-response to outpatient treatment. Thus, inclusion of a placebo group was not possible for ethical reasons. It is therefore open to question whether or not the improvements observed during the first 2 weeks of treatment were because of “true” effects of medication and/or psychotherapy, or were “placebo-like” responses. Nevertheless these early improvements proved to be sensitive (although not specific) predictors of eventual response or remission ( Fig. 1 ). They may thus represent indicators of a principle disposition to improve irrespective of the type of treatment ( Stassen et al., 2007 ) rather than specific predictors of response to medication and/or psychotherapy. Nevertheless, the presence or absence of early improvement can help the clinician to decide much earlier than traditionally thought possible whether to continue or change treatment ( Stassen et al., 1993 ). Another potential shortcoming of the present study is that though the final assessment at week 5 was performed by blinded raters, the weekly HAMD-17 scores were assessed by non-blinded raters. However, this is unlikely to have caused a bias in favour of IPT, since an effect of IPT as early as week 2 was in contrast to our expectations. In fact, since an additional benefit of IPT was expected to become evident much later, the time point at week 5 was designated for blind assessment. Thus, if anything, raters may have been biased against the actual finding. In conclusion, our data suggest that combined treatment with medication and psychotherapy not only increases the overall efficacy as compared to pharmacotherapy alone, but provides this additional benefit very early in the course of improvement. Further studies specifically designed to examine this issue are necessary to corroborate these findings. Role of funding source Funding for this study was provided by the Deutsche Forschungsgemeinschaft (DFG); the DFG had no further role in the study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication. Conflict of interest Author Dietrich van Calker has received honoraria for lecturing from various pharmaceutical companies. All other authors declare that they have no conflicts of interest. Acknowledgements The authors are indebted to Prof. Charles Reynolds for his valuable advice. We thank Mr. James Balmford for reading the manuscript and his valuable comments. Appendix A Supplementary data Appendix A Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jad.2008.07.023 . References Angst et al., 1993 Angst J. Delini-Stula A. Stabl M. Stassen H.H. Is a cut-off score a suitable measure of treatment outcome in short-term trials in depression? A methodological meta-analysis Hum. Psychopharmacol. 8 1993 311 317 DiMascio et al., 1979 DiMascio A. Weissman M.M. Prusoff B.A. Neu C. Zwilling M. Klerman G.L. Differential symptom reduction by drugs and psychotherapy in acute depression Arch. Gen. Psychiatry 36 1979 1450 1456 Fava et al., 2002 Fava G.A. Fabbri S. Sonino N. Residual symptoms in depression: an emerging therapeutic target Prog. Neuro-Psychopharmacol. Biol. Psychiatry 26 2002 1019 1027 Fava et al., 2003 Fava G.A. Ruini C. Sonino N. Treatment of recurrent depression: a sequential psychotherapeutic and psychopharmacological approach CNS Drugs 17 2003 1109 1117 First et al., 1997 First M. Spitzer R. Gibbon M. Williams J. Structured Clinical Interview for DSM-IV Axis I Disorders: patient edition 1997 Biometrics Research Dept, New York State Psychiatric Institute New York Frank et al., 1990 Frank E. Kupfer D.J. Perel J.M. Cornes C. Jarrett D.B. Mallinger A.G. Thase M.E. McEachran A.B. Grochocinski V.J. Three-year outcomes for maintenance therapies in recurrent depression Arch. Gen. Psychiatry 47 1990 1093 1099 Hamilton, 1960 Hamilton M.A. A rating scale for depression J. Neurol. Neurosurg. Psychiatry 23 1960 56 63 Hollon et al., 2005 Hollon S.D. Jarrett R.B. Nierenberg A.A. Thase M.E. Trivedi M. Rush A.J. Psychotherapy and medication in the treatment of adult and geriatric depression: which monotherapy or combined treatment? J. Clin. Psychiatry 66 2005 455 468 Kaplan and Meier, 1958 Kaplan E.L. Meier P. Nonparametric estimation from incomplete observations J. Am. Stat. Assoc. 53 1958 457 481 Katz et al., 2004 Katz M.M. Tekell J.L. Bowden C.L. Brannan S. Houston J.P. Berman N. Frazer A. Onset and early behavioral effects of pharmacologically different antidepressants and placebo in depression Neuropsychopharmacology 29 2004 566 579 Keller et al., 2000 Keller M.B. McCullough J.P. Klein D.N. Arnow B. Dunner D.L. Gelenberg A.J. Markowitz J.C. Nemeroff C.B. Russell J.M. Thase M.E. Trivedi M.H. Zajecka J. A comparison of nefazodone, the cognitive behavioral-analysis system of psychotherapy, and their combination for the treatment of chronic depression N. Engl. J. Med. 342 2000 1462 1470 Klerman et al., 1984 Klerman G.L. Weissman M. Rounsaville B.J. Chevron E.S. Interpersonal Psychotherapy of Depression 1984 Basic Books New York Laska and Siegel, 1995 Laska E.M. Siegel C. Characterizing onset in psychopharmacological clinical trials Psychopharmacol. Bull. 31 1995 29 35 Lenze et al., 2002 Lenze E.J. Dew M.A. Mazumdar S. Begley A.E. Cornes C. Miller M.D. Imber S.D. Frank E. Kupfer D.J. Reynolds C.F. III Combined pharmacotherapy and psychotherapy as maintenance treatment for late-life depression: effects on social adjustment Am. J. Psychiatry 159 2002 466 468 Li and Feng, 2005 Li Y. Feng J. A nonparametric comparison of conditional distributions with nonnegligible cure fractions Lifetime Data Anal. 11 2005 367 387 Lydiard et al., 1997 Lydiard R.B. Stahl S.M. Hertzman M. Harrison W.M. A double-blind, placebo-controlled study comparing the effects of sertraline versus amitriptyline in the treatment of major depression J. Clin. Psychiatry 58 1997 484 491 Mayberg, 2003 Mayberg H.S. Modulating dysfunctional limbic-cortical circuits in depression: towards development of brain-based algorithms for diagnosis and optimised treatment Br. Med. Bull. 65 2003 193 207 McIntyre and O'Donovan, 2004 McIntyre R.S. O'Donovan C. The human cost of not achieving full remission in depression Can. J. Psychiatry 49 2004 10S 16S Moller et al., 1998 Moller H.J. Gallinat J. Hegerl U. Arato M. Janka Z. Pflug B. Bauer H. Double-blind, multicenter comparative study of sertraline and amitriptyline in hospitalized patients with major depression Pharmacopsychiatry 31 1998 170 177 Montgomery et al., 2002 Montgomery S.A. Bech P. Blier P. Moller H.J. Nierenberg A.A. Pinder R.M. Quitkin F.M. Reimitz P.E. Rosenbaum J.F. Rush A.J. Stassen H.H. Thase M.E. Selecting methodologies for the evaluation of differences in time to response between antidepressants J. Clin. Psychiatry 63 2002 694 699 Nierenberg et al., 2000 Nierenberg A.A. Farabaugh A.H. Alpert J.E. Gordon J. Worthington J.J. Rosenbaum J.F. Fava M. Timing of onset of antidepressant response with fluoxetine treatment Am. J. Psychiatry 157 2000 1423 1428 Papakostas et al., 2006 Papakostas G.I. Perlis R.H. Scalia M.J. Petersen T.J. Fava M. A meta-analysis of early sustained response rates between antidepressants and placebo for the treatment of major depressive disorder J. Clin. Psychopharmacol. 26 2006 56 60 Posternak and Zimmerman, 2005a Posternak M.A. Zimmerman M. Drs. Posternak and Zimmerman reply J. Clin. Psychiatry 66 2005 1193 Posternak and Zimmerman, 2005b Posternak M.A. Zimmerman M. Is there a delay in the antidepressant effect? A meta-analysis J. Clin. Psychiatry 66 2005 148 158 Quitkin et al., 1991 Quitkin F.M. Rabkin J.G. Stewart J.W. McGrath P.J. Harrison W. Ross D.C. Tricamo E. Fleiss J. Markowitz J. Klein D.F. Heterogeneity of clinical response during placebo treatment Am. J. Psychiatry 148 1991 193 196 Quitkin et al., 1996 Quitkin F.M. McGrath P.J. Stewart J.W. Ocepek-Welikson K. Taylor B.P. Nunes E. Deliyannides D. Agosti V. Donovan S.J. Petkova E. Klein D.F. Chronological milestones to guide drug change. When should clinicians switch antidepressants? Arch. Gen. Psychiatry 53 1996 785 792 Quitkin et al., 2003 Quitkin F.M. Petkova E. McGrath P.J. Taylor B. Beasley C. Stewart J. Amsterdam J. Fava M. Rosenbaum J. Reimherr F. Fawcett J. Chen Y. Klein D. When should a trial of fluoxetine for major depression be declared failed? Am. J. Psychiatry 160 2003 734 740 Quitkin et al., 2005 Quitkin F.M. McGrath P.J. Stewart J.W. Klein D.F. Ross D.C. Accurate meta-analytical assessment of “true antidepressant effects” needed J. Clin. Psychiatry 66 2005 1192 1193 R Development Core Team, 2004 R Development Core Team A Language and Environment for Statistical Computing 2004 R Foundation for Statistical Computing Vienna, Austria ISBN 3-900051-07-0, URL http://www.R-project.org Reimherr et al., 1990 Reimherr F.W. Chouinard G. Cohn C.K. Cole J.O. Itil T.M. LaPierre Y.D. Masco H.L. Mendels J. Antidepressant efficacy of sertraline: a double-blind, placebo- and amitriptyline-controlled, multicenter comparison study in outpatients with major depression J. Clin. Psychiatry 51 Suppl B 1990 18 27 Reynolds et al., 1999 Reynolds C.F. III Frank E. Perel J.M. Imber S.D. Cornes C. Miller M.D. Mazumdar S. Houck P.R. Dew M.A. Stack J.A. Pollock B.G. Kupfer D.J. Nortriptyline and interpersonal psychotherapy as maintenance therapies for recurrent major depression: a randomized controlled trial in patients older than 59 years JAMA 281 1999 39 45 Schramm et al., 2007 Schramm E. van Calker D. Dykierek P. Lieb K. Kech S. Zobel I. Leonhart R. Berger M. An intensive treatment program of interpersonal psychotherapy plus pharmacotherapy for depressed inpatients: acute and long-term results Am. J. Psychiatry 164 2007 768 777 Schulberg et al., 1996 Schulberg H.C. Block M.R. Madonia M.J. Scott C.P. Rodriguez E. Imber S.D. Perel J. Lave J. Houck P.R. Coulehan J.L. Treating major depression in primary care practice. Eight-month clinical outcomes Arch. Gen. Psychiatry 53 1996 913 919 Stassen et al., 1993 Stassen H.H. Delini-Stula A. Angst J. Time course of improvement under antidepressant treatment: a survival-analytical approach Eur. Neuropsychopharmacol. 3 1993 127 135 Stassen et al., 1996 Stassen H.H. Angst J. Delini-Stula A. Delayed onset of action of antidepressant drugs? Survey of results of Zurich meta-analyses Pharmacopsychiatry 29 1996 87 96 Stassen et al., 2007 Stassen H.H. Angst J. Hell D. Scharfetter C. Szegedi A. Is there a common resilience mechanism underlying antidepressant drug response? Evidence from 2848 patients J. Clin. Psychiatry 68 8 2007 1195 1205 Szegedi et al., 2003 Szegedi A. Muller M.J. Anghelescu I. Klawe C. Kohnen R. Benkert O. Early improvement under mirtazapine and paroxetine predicts later stable response and remission with high sensitivity in patients with major depression J. Clin. Psychiatry 64 2003 413 420 Tamura et al., 2000 Tamura R.N. Faries D.E. Feng J. Comparing time to onset of response in antidepressant clinical trials using the cure model and the Cramer–von Mises test Stat. Med. 19 2000 2169 2184 Trivedi et al., 2006 Trivedi M.H. Rush A.J. Wisniewski S.R. Nierenberg A.A. Warden D. Ritz L. Norquist G. Howland R.H. Lebowitz B. McGrath P.J. Shores-Wilson K. Biggs M.M. Balasubramani G.K. Fava M. Evaluation of outcomes with citalopram for depression using measurement-based care in STAR⁎D: implications for clinical practice Am. J. Psychiatry 163 2006 28 40 Watkins et al., 1993 Watkins J.T. Leber W.R. Imber S.D. Collins J.F. Elkin I. Pilkonis P.A. Sotsky S.M. Shea M.T. Glass D.R. Temporal course of change of depression J. Consult. Clin. Psychol. 61 1993 858 864