Early improvement is a predictor of treatment ... - Commonweb

Journal of Affective Disorders 120 (2010) 86–93

Contents lists available at ScienceDirect

Journal of Affective Disorders j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j a d

Research report

Early improvement is a predictor of treatment outcome in patients with mild major, minor or subsyndromal depression André Tadić a,⁎, Isabella Helmreich a, Roland Mergl b, Martin Hautzinger c, Ralf Kohnen d, Verena Henkel e, Ulrich Hegerl b a b c d e

Department of Psychiatry and Psychotherapy, University of Mainz, Germany Department of Psychiatry, University of Leipzig, Germany Institute of Psychology, Department of Clinical and Developmental Psychology, University of Tübingen, Germany IMEREM (Institute for Medical Research Management and Biometrics GmbH), Nuremberg, Germany Department of Psychiatry and Psychotherapy, University of Munich, Germany

a r t i c l e

i n f o

Article history: Received 30 January 2009 Received in revised form 15 April 2009 Accepted 15 April 2009 Available online 9 May 2009 Keywords: Depression Early improvement Outcome prediction Pharmacotherapy Psychotherapy Primary care

a b s t r a c t Background: There is substantial evidence that early improvement (EI) under antidepressant treatment is a clinically useful predictor of later treatment outcome in patients with major depressive disorders. The aim of this study was to test whether EI can also be used as a predictor for treatment outcome in patients with mild major, minor or subsyndromal depression, i.e. patients, who are typically treated by general practitioners. Methods: Analyses were carried out using data from 223 patients of a 10-weeks randomized, placebo-controlled trial comparing the effectiveness of sertraline and cognitive-behavioural therapy (CBT) in patients with mild major, minor or subsyndromal depression. EI was defined as a reduction of ≥ 20% on the 17-item Hamilton Rating Scale for Depression (HAMD-17) compared with baseline within the first 2 weeks of treatment. The predictive value of EI for stable response at week 8 and 10 (≥ 50% HAMD-17 sum score reduction at weeks 8 and 10) and stable remission (HAMD-17 sum score ≤ 7 at weeks 8 and 10) was evaluated. Results: In both the sertraline- and CBT-treatment group, EI was a highly sensitive predictor for later stable response (76% and 82%, respectively) and stable remission (70% and 75%, respectively). In patients without EI, only a small proportion of sertraline or CBT-treated patients achieved stable response (20.9% and 5.9%, respectively) or stable remission (18.6% and 8.8%, respectively). Patients with EI were by far more likely to achieve stable response or stable remission than patients without as indicated by high odds ratios (95% confidence interval) of 8.1 (3.0–21.8) and 3.8 (1.4–10.1) for sertraline, and 11.1 (2.1–58.4) and 7.2 (1.7–30.8) for CBT-treated patients, respectively. Limitations: Sample sizes were relatively low in different treatment groups. Conclusion: The identification of early improvement might be useful in clinical decision making in the early course of treatment of patients with mild major, minor and subthreshold depression. © 2009 Elsevier B.V. All rights reserved.

1. Introduction Minor and mild to moderate depressive syndromes are common in primary care (Ackermann and Williams, 2002; Pincus et al., 1999; Williams et al., 1995) and associated with

⁎ Corresponding author. Department of Psychiatry and Psychotherapy, University of Mainz, Untere Zahlbacher Str. 8, 55131 Mainz, Germany. Tel.: +49 6131 17 3950; fax: +49 6131 17 3487. E-mail address: [email protected] (A. Tadić). 0165-0327/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jad.2009.04.014

considerable functional morbidity, suicidality and service utilization (Lyness et al., 2006; Rapaport et al., 2002; Wagner et al., 2000). The public health significance of these so-called ‘minor’ depression forms is high, but often underestimated (Backenstrass et al., 2006; Hermens et al., 2004; Lyness et al., 2006). The scientific evidence on the treatment of minor depression forms has been interpreted differently. Regarding mild major depression, British guidelines favor some restraint in the prescription of antidepressants (NICE, 2004), but American

A. Tadić et al. / Journal of Affective Disorders 120 (2010) 86–93

guidelines favor its use (APA, 2000). Even more unclear as the use of antidepressants itself is the appropriate treatment duration, particularly in case of insufficient outcome. This uncertainty is reflected a) by widely varying guideline recommendations for the use of antidepressants in the primary care sector with time spans ranging from two (Bauer et al., 2007) to eight weeks (APA, 2000) or even no statement on this important aspect of treatment (NICE, 2004), and b) by the absence of a further differentiation considering the degree of illness severity. In other words, current national (AkdÄ, 2006) and international guidelines (APA, 2000; NICE, 2004) for the treatment of depression in the primary care sector usually do not contain recommendations for optimization strategies (re-assessment of diagnosis, dose adjustments, change of treatment, etc.) in the early course of treatment. In the treatment of major depressive disorder (MDD) the delayed-onset hypothesis (Quitkin et al., 1984) suggested that persistent or “true” drug response occurs mainly in the later course of treatment, i.e., week 3–4, while response occurring in the first 2 weeks was assumed to be unstable and due to placebo effects. Challenging the idea of a delayed onset of antidepressants' action, there is a substantial body of evidence from many studies with virtually all groups of antidepressants strongly suggesting that a true drug response can be observed within the first 14 days of treatment (Katz et al., 2004; Nierenberg et al., 2000; Papakostas et al., 2006; Posternak and Zimmerman, 2005; Stassen et al., 2007; Szegedi et al., 2003; Taylor et al., 2006). Moreover, the occurrence of improvement of depressive symptoms in the early course of treatment has been identified as being highly predictive for final treatment outcome. In 2004, Katz and colleagues reported on a randomized controlled trial (RCT) in which patients were treated with the selective serotonin reuptake inhibitor (SSRI) paroxetine, the tricyclic antidepressant (TCA) desipramine or placebo (Katz et al., 2004). In this study, early treatment-specific behavioural changes occurred that were not observed in the placebo-group, and these changes were highly predictive of final clinical response to antidepressant therapy. Nierenberg and colleagues have reported that N50% of patients who eventually respond to fluoxetine treatment started to improve during the first 2 weeks of treatment and that early non-response to fluoxetine treatment predicted poor 8-weeks outcomes (Nierenberg et al., 2000). In multiple studies between 1993 and 2007, Stassen and colleagues analyzed individual time courses of response in depressed patients treated with various antidepressants and placebo (Stassen et al., 2007, and references inside). Each of their analyses revealed that patients with improvement (defined as a reduction ≥20% of the 17-item version of Hamilton Depression Rating Scale baseline score) during the first 2 weeks (=early improvement) of antidepressant treatment showed substantial response at study endpoint. A recent meta-analysis of 2848 patients with MDD confirmed previous analyses showing that early improvers were far more likely to become responders than patients without early improvement (pooled odds ratio (OR) = 9.25, 95% confidence interval (95%-CI) = 7.79–10.98) (Stassen et al., 2007). In separate analyses, Szegedi et al. (2003) examined early improvement in a RCT comparing mirtazapine and paroxetine in MDD patients. Improvement1

1

Improvement was defined as HAMD-17 score reduction ≥20%.

87

occurred in a majority of patients within 2 weeks of treatment, and this improvement was a highly sensitive predictor of later stable response2 and stable remission3 for both drugs. Less than 10% of patients who had not improved after 2 weeks of treatment became stable responders or remitters over the course of the study. Apart from these studies analyzing data from RCTs, Henkel et al. (2008) recently reported on a large sample (N =795) of depressed in-patients treated with various antidepressants in a naturalistic setting. In their study, early improvement provided a sensitivity of 75% for response and 80% for remission. Time to response is believed to be longer with psychotherapy than with pharmacotherapy (for review, see Hollon et al., 2005). However, the principle of a relation between improvement in the early course of treatment and later treatment outcome has also been applied to the time course of symptom amelioration on psychotherapy. Keller and Hautzinger (2007) found early improvement in the course of cognitive behaviour therapy was predicting later outcome (odd ratios for response when no change in symptom score were .72 after week 2, 0.49 after week 3, and .09 after week 6). In this analysis of 304 depressed outpatients, after 3 weeks of psychotherapy, already 27% of patients had a 50% reduction of their starting Hamilton Depression score. Recently, the effect of additional interpersonal psychotherapy (IPT) on the time course of response to antidepressants has been investigated (van Calker et al., 2008). They observed that the benefit of combination treatment was already fully developed and indeed particularly pronounced in the second week of treatment with no further increase in the following weeks. The finding of early improvement being highly predictive for later outcome resulted in the idea that an effective antidepressant treatment triggers and maintains conditions necessary for recovery from the disorder (Stassen et al., 2007). It has been suggested that affectively ill patients possess a biological, “resilience”-like component that controls recovery from depression to a major extent. Once triggered, recovery seems to follow – independent of pharmacologic differences of the triggers – a uniform pattern of course. Consequently, the vast majority of patients showing a favorable later outcome experience the respective onset within the first 2 weeks of treatment. Inversely, non-improvement after 2 weeks of treatment seems to indicate that the selected antidepressant treatment did not trigger the resilience-like component and has strongly limited chances to do so, even if continued in the course of treatment. The aim of this study was to test I) whether improvement also regularly occurs in the first two weeks of antidepressant treatment in patients with mild major, minor or subsyndromal depression, and II) whether this early improvement can be used as a predictor for later treatment outcome in depressed patients, who are typically treated by general practitioners.

2 Stable response was defined as HAMD-17 score reduction ≥50% at week 4 and onward. 3 Stable remission was defined as HAMD-17 score ≤7 at week 4 and onward.

88


2. Methods 2.1. Patients Details of the patient sample have been described elsewhere (Hegerl et al., 2009). In brief, analyses were carried out using data (ITT population, N = 368) obtained from a prospective, single-center, randomized, placebo-controlled, parallel-group, ten-week clinical trial with five treatment arms to compare the effectiveness of sertraline and cognitivebehavioural therapy in patients with mild major, minor and subsyndromal depression. The study has been approved by an independent Ethics Review Committee. All subjects gave written informed consent after the procedures and possible side effects were fully explained to them. Patients were referred by the treating general practitioner to the study center (established within the ‘German Research Network on Depression and Suicidality’, subproject 2.1; Hegerl and Pfeiffer-Gerschel, 2005) for out-patient treatment. Main inclusion criteria were: age ≥18 years; diagnoses of minor depression (based on research criteria; also with comorbidity such as anxiety or somatoform disorders), dysthymia or major depressive disorder with mild or moderate severity according to DSM-IV criteria and Hamilton Rating Scale for Depression (HAMD-17; Hamilton, 1960) baseline total scores ≥8 and ≤22. Main exclusion criteria were: current psychotherapeutic or antidepressant treatment, acute suicidality, brief recurrent depression, bipolar affective disorders, addiction (i.e., dependency of alcohol, benzodiazepines (with the exception of “low dose dependency”), or illicit drugs), schizophrenia, schizotypal disorder, delusional disorders and schizo-affective disorders, obsessive–compulsive disorder, severe somatic diseases. Diagnoses according to DSM-IV criteria were based upon a fully structured clinical interview (Composite International Diagnostic Interview, CIDI; WHO, 1993), applied as a German computer-administered version (DIA-X) with high inter-rater reliability (κ ≥ 0.81; Wittchen and Pfister, 1997). Further exclusion criteria were antisocial personality disorder and borderline personality disorder, which were clinically assessed.

2.1.1. Randomization The randomization list for sertraline versus placebo treatment was developed by the biometrical department of a contract research organization named IMEREM™ (Institute for Medical Research Management and Biometrics) in advance using SAS™ ProcPlan. Randomization to the three psychotherapy arms and one blinded medication arm including sertraline and placebo was block stratified using a minimization method (Schouten, 1995) with four strata: age (≤65 years; N65 years), sex, HAMD-17 baseline total score (intensity ranges: 8–12; 13–17; 18–22), comorbidity regarding anxiety and/or somatoform disorders. Sertraline and placebo were comprised in one treatment arm “medication” to maintain double-blind conditions. Stratified randomization was performed with a computer program (TurboPascal) which had been developed by IMEREM™ according to Schouten (1995) and extensively validated by simulation studies. Medication was distributed to newly randomized patients in a sequential order.

2.1.2. Interventions: arm 1 (sertraline) and arm 2 (placebo) In these arms, the visits (at weeks 0, 2, 4, 6, 8 and 10) involved adjustment of medication and recording of medical history, symptoms, (severe) adverse events, weight, and blood pressure. If symptom reduction was achieved, patients continued with initial dose of the drug. Otherwise, an escalation in dose in steps of 50 mg up to 200 mg per day was possible at week 2, 4 and 6. Arm 3 (CBT) and arm 4 (GSG): Six experienced clinical psychotherapists with a specialization in CBT provided CBT and GSG and received continuous supervision. All had several years of experience in CBT (mean number of years after licencing as psychotherapist was 5.5 years, with a range from 4 up to 9 years; they were all full time in private practice since 3 up to 8 years) and had been additionally trained with the study manual. Short-term CBT (Hautzinger, 2003) and GSG were provided for ten weeks, in nine weekly group sessions (at 90 min each) with five to eight patients after an initial individual 50-minute session. In the GSG, patients were allowed to talk about their situation and daily life, but no specific psychotherapeutic interventions were allowed for the groupleader. All group sessions were videotaped and good to excellent adherence to treatment protocols was found for all therapists (for detailed results, see Hegerl et al., 2009), when randomly selected tapes were rated by trained raters. Arm 5 (patients' preference/choice arm): Patients randomized to the patients' choice condition could freely select a ten-week treatment with either sertraline or CBT. Severity of depressive symptoms was assessed in biweekly intervals with the HAMD-17. All HAMD-17 ratings were conducted by blinded and trained psychologists (Hegerl et al., 2009). 2.2. Statistical analyses For the present analysis, patients were excluded if there was no HAMD-17 rating available for week 2, week 4 and at Table 1 Patients' characteristics. Sertraline

M:F, % Age, mean ± SD, ys.1 Main diagnosis, % – Major depression – Dysthymic disorder – Double depression – Depressive disorder NOS – Subsyndromal depressive disorder Psychiatric comorbid diagnosis, % Previous depressive episodes, mean±SD, n2 HAM-D-17 score at baseline, mean ± SD 1

Placebo

CBT

GSG

(N = 84)

(N = 57)

(N = 56)

(N = 26)

40.5:59.5 44.0±15.2

31.6:68.4 48.1±14.9

23.2:76.8 51.3±14.0

23.1:76.9 51.3±12.9

34.5 1.2 45.2 17.9

24.6 5.3 40.4 26.3

26.8 1.8 50.0 19.6

42.3 3.8 42.3 7.7

1.2

3.5

1.8

3.8

34.5

26.3

57.1

61.5

1.9 ± 0.8

2.0 ± 0.9

2.2 ± 0.7

2.1 ± 0.7

16.9 ± 4.5

16.7 ± 4.1

16.4 ± 4.3

15.3 ± 4.3

Mean age was significantly different between groups (p=0.016); the CBT group had a higher mean age compared to the sertraline group (Tukey-HSD post-hoc p=0.02); 2Data were unavailable for 46.4% of sertraline-, 57.9% of placebo-, 50.0% of CBT-, and 46.2% of GSG-treated patients. CBT: Cognitive-behavioral therapy; F: female; GSG = guided self-help group treatment; HAM-D-17: 17-item version of the Hamilton Depression Rating Scale (Hamilton, 1960); M: male; n/N: sample size; SD: standard deviation.


89

Fig. 1. a: Percentage of Improvers of the total sample during treatment (≥20% decrease of HAMD-17 sum score). b: Percentage of Stable Responders (≥50% decrease of HAMD-17 sum score) Starting at Week X and subsequently maintained to Week 10. c: Percentage of Stable Remitters (HAMD-17 sum score ≤ 7) Starting at Week X and subsequently maintained to Week 10.

90


least once beyond week 4. The method of last observation carried forward (LOCF) was used for missing HAM-D-17 values. Patients randomized to the patients' choice arm had treatment outcomes similar to patients who received the identical treatment after randomization to sertraline or CBT, resp (Hegerl et al., 2009). Therefore, group differences in the relation of early improvement and treatment outcome are not to be expected. For this reason, we assigned patients randomized to the patients' choice condition (arm 5) to the study arm with identical treatment (sertraline or CBT, respectively), resulting in 4 groups with a total number of 223 patients to analyze. Baseline clinical and demographical data were analyzed according to the scale level with oneway analysis-of-variance or with chi-square test. All tests were 2-sided and statistical significance was defined as p ≤ 0.05. To evaluate treatment outcome we defined Improvement as a reduction in HAMD-17 score of ≥ 20% compared to baseline (Stassen et al., 2007; Szegedi et al., 2003). An overall comparison of the rates of early improvement in the four treatment groups was performed with chisquare test (df = 3). Differences if the rates of early improvement between specific treatment groups (sertraline vs. placebo; CBT vs. GSG; sertraline vs. CBT; placebo vs. GSG) were evaluated with chi-square tests and df = 1. Stable symptom changes are more important than temporary symptom changes; therefore, we defined and evaluated stable response (= HAMD-17 score reduction ≥ 50% compared to baseline at weeks 8 and 10) and stable remission (= HAMD-17 score ≤ 7 points at weeks 8 and 10) as outcome criteria. In order to calculate the predictive value of improvement in the early course of treatment, the number of early improvers, stable responders and stable remitters were entered into contingency tables. We calculated the indices sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) following the principles of analysis for individual patients proposed by Szegedi et al. (2003). Additionally, we evaluated the outcomes of patients with and without early improvement and calculated Odds ratios (OR) and 95%-confidence intervals (95% CI) for achieving stable response and stable remission. Statistical analysis was performed by using SPSS software, version 12.0 (SPSS Inc., Chicago, Illinois).

3. Results 3.1. Patients characteristics and treatment outcome The four treatment groups were comparable on background clinical characteristics (p N 0.05 for each analysis), with the exception of a significant difference in age (p = 0.016; see Table 1). The criterion of improvement, defined as a ≥20% reduction in HAMD-17 total score, was fulfilled by 39% and 49% of patients across treatment groups (sertraline and CBT, respectively) and by 35% and 40% across control groups (placebo and GSG) at week 2. Across treatment groups, improvement occurred in a sufficiently high number of patients after 2 weeks of treatment and only a relatively little proportion of patients additionally developed improvement at weeks 4, 6, 8 and 10 (Fig. 1a). The overall comparison of rates of early improvement revealed no significant differences between treatment groups (Χ23 = 2.4; p = 0.499). There were also no significant differences in the rates of early improvement between specific treatment groups (sertraline vs. placebo; CBT vs. GSG; sertraline vs. CBT; placebo vs. GSG; p ≥ 0.267 for each analysis). Regarding the overall treatment outcome, stable response and stable remission developed slowly during the treatment period (see Fig. 1b and c, respectively) and were achieved at week 8 and 10 in 44% and 32% of sertraline-, 20% and 22% of CBT, 19% and 23% of placebo-, and 19% and 15% of GSG-treated patients, respectively. 3.2. Predictive value of early improvement After demonstrating that early improvement occurs in a high percentage of patients with mild major, minor and subthreshold depression, it was analyzed whether early improvement is a reliable predictor of later outcome. Sensitivity, Specificity, Positive Predictive Value (PPV) and Negative Predictive Value (NPV) were calculated. The predictive values of early improvement for the treatment outcomes stable response and stable remission are shown in Table 2. In both treatment groups, early improvement was a highly sensitive predictor for later stable response (range: 76– 82%) and stable remission (range: 70–75%). Specificity for later stable response (range 71–72%) and stable remission

Table 2 Indices of predictive value of early improvement for later treatment outcome. Sensitivity (95% CI)

Specificity (95% CI)

PPV (95% CI)

NPV (95% CI)

A. Predicted outcome: stable response (response at weeks 8 and 10) Sertraline 0.76 (0.59–0.89) 0.72 Placebo 0.82 (0.48–0.98) 0.70 CBT 0.82 (0.48–0.98) 0.71 GSG 0.40 (0.05–0.85) 0.67

(0.57–0.84) (0.55–0.83) (0.56–0.84) (0.43–0.86)

0.68 0.39 0.41 0.22

(0.52–0.82) (0.20–0.61) (0.21–0.64) (0.03–0.60)

0.79 0.94 0.94 0.82

(0.64–0.90) (0.80–0.99) (0.80–0.99) (0.56–0.96)

B. Predicted outcome: stable remission (remission at weeks Sertraline 0.70 (0.49–0.86) Placebo 0.77 (0.46–0.95) CBT 0.75 (0.43–0.95) GSG 0.25 (0.01–0.81)

(0.47–0.74) (0.54–0.83) (0.54–0.83) (0.41–0.83)

0.46 0.43 0.41 0.11

(0.30–0.62) (0.23–0.65) (0.21–0.64) (0.00–0.48)

0.81 0.91 0.91 0.82

(0.66–0.91) (0.76–0.98) (0.76–0.98) (0.56–0.96)

8 and 10) 0.61 0.70 0.70 0.64

Notes: We defined and evaluated stable response (= HAMD-17 score reduction ≥ 50% compared to baseline at weeks 8 and 10) and stable remission (= HAMD-17 score ≤ 7 points at weeks 8 and 10) as outcome criteria. CBT: cognitive-behavioral therapy; CI: confidence interval; GSG: guided self-help group treatment; HAMD-17: 17-item version of the Hamilton Depression Rating Scale (Hamilton, 1960); NPV: negative predictive value; PPV: positive predictive value.

A. Tadić et al. / Journal of Affective Disorders 120 (2010) 86–93 Table 3 Treatment outcome and ORs (95% CI) for treatment outcomes of patients with early improvement compared to patients without early improvement. Stable response (%)

Sertraline Placebo CBT GSG

Stable remission (%)

EI+

EI−

OR (95% CI)

EI+

EI−

OR

68.3 39.1 40.9 22.2

20.9 5.9 5.9 17.6

8.1 (3.0–21.8) 10.3 (2.0–53.9) 11.1 (2.1–58.4) 1.33 (0.2–9.9)

46.3 43.5 40.9 11.1

18.6 8.8 8.8 17.6

3.78 8.0 7.2 0.6

(1.4–10.1) (1.9–33.7) (1.7–30.8) (0.1–6.6)

Notes: We defined and evaluated stable response (= HAMD-17 score reduction ≥ 50% compared to baseline at weeks 8 and 10) and stable remission (= HAMD-17 score ≤7 points at weeks 8 and 10) as main outcome criteria. CBT: cognitive-behavioral therapy; 95% CI: 95% confidence interval; EI+: with early improvement; EI−: no early improvement; GSG: guided self-help group treatment; HAMD-17: 17-item version of the Hamilton Depression Rating Scale (Hamilton, 1960); OR: odds ratio.

(range 61–70%) was lower, but also quite high across treatment conditions. NPV for stable response (range: 79– 94%) and stable remission (81–91%) were far higher than PPV for these treatment groups (ranges: 41–68%; 41–46%, respectively). In placebo-treated patients, we found similar indices of the predictive value of early improvement for later treatment outcome, but not in patients participating in the guided self-help group (see Table 2). 3.3. Outcomes of early improvement We further examined the outcomes of patients with and without early improvement. The Odds ratios (ORs) for achieving a favorable outcome for patients with early improvement compared to patients without early improvement were remarkably high across both treatment groups (see Table 3). These results clearly demonstrate that patients who have a HAMD-17 score decrease ≥20% within the first two weeks of antidepressant therapy are far more likely to achieve stable response and stable remission compared to patients who do not have a HAMD-17 score decrease ≥20% within the first two weeks of antidepressant therapy. This relation between early improvement and later treatment outcome was also observed in placebo-treated patients, but not in GSG-treated patients (see Table 3). 4. Discussion We found that improvement, defined as a decrease of ≥20% of the HAMD-17 total score, occurred during the first 2 weeks of antidepressant treatment in patients with mild major, minor and subthreshold depression. Thus, we deliver further evidence against the delayed-onset hypothesis for antidepressant treatment response (Quitkin et al., 1984) and further support the view that improvement in the early course of treatment can be expected with various antidepressant drugs (Stassen et al., 2007) and in various psychiatric conditions (Leucht et al., 2008; Pollack et al., 2008). Additionally, we provide evidence that early improvement can be expected with CBT. Until recently, it was believed that time to onset of improvement with psychotherapy is longer than with pharmacotherapy (for review, see Hollon et al., 2005). Recently, this has been questioned by the analysis of the effects of additional IPT on the time course of response to

91

pharmacotherapy. In our study, the increase in improvement with CBT was greatest between start of treatment and week two. These patients had one individual and two group sessions by week 2 (in contrast to sertraline and placebo, which were administered daily). This raises the question about the underlying effects of psychotherapy in this early period. One might suggest that the patient accepts the sick role, experiences relief of some symptoms, develops hope and an alliance with the therapist. In general, the rates of early improvement observed in our sample were lower compared to previous reports (e.g., Stassen et al., 2007; Szegedi et al., 2003). However, these previous studies investigated patients with major depression, characterized by higher depression scores and less psychiatric comorbid diseases. These population differences might account for differences in rates of early improvement. The rates of stable response and stable remission were lower in the CBT group than in the sertraline group. This may be due to the fact that a marked antidepressive effect of sertraline was already reached in the first sixth week (with only marginal changes of the mean HAMD sum scores in the following weeks), whereas in the CBT group a pronounced change of the mean HAMD sum score is later apparent (compare Hegerl et al., 2009; Fig. 2). Nevertheless, improvement at week 2 was a highly sensitive predictor for later stable response and stable remission. This is consistent with findings from several studies in major depression, where early improvement was highly predictive of later outcome (Katz et al., 2004; Nierenberg et al., 2000; Stassen et al., 2007; Szegedi et al., 2003). It is also consistent with recent studies in other psychiatric diseases like generalized anxiety disorder (Pollack et al., 2008) and schizophrenia (Leucht et al., 2008), where improvement in the early course of treatment was highly predictive for later outcome. Furthermore, our results are consistent with studies evaluating psychotherapeutic strategies in the treatment of depressive disorders (Keller and Hautzinger, 2007; Keller, 2003; van Calker et al., 2008; Van et al., 2008), where improvement in the early course of treatment also predicted later outcome. On the other hand, early improvement was not necessarily associated with a favorable later outcome. This is indicated by the limited specificity of early improvement. The consistently higher negative predictive values show that the absence of early improvement is more predictive of a later lack of a favorable outcome with continued treatment. As indicated by PPV and NPV (see also Table 3), respectively, early improvement predicted stable response and stable remission in 41–68%; whereas non-improvement after two weeks of treatment predicted an unsuccessful treatment outcome in 79–94% across treatment groups. From a clinical point of view, the most immediate practical consequence evolving from our analysis is that early improvement in a patient with mild major, minor or subthreshold depression may predict a favorable outcome after 8 and 10 weeks of treatment. Probably even more important, that lack of improvement after 2 weeks should lead to strategies to treatment optimization, because remission can hardly be expected after 8 and 10 weeks of treatment. Our data thereby provide an empirical basis for guidelines for the treatment of minor depression, particularly for the treatment duration in case of insufficient outcome. This is important, because the

92


use of antidepressants in this population is still debated and its appropriate treatment duration is largely unclear, which is reflected by widely varying guideline recommendations (AkdÄ, 2006; APA, 2000; Bauer et al., 2007; NICE, 2004). One advantage of our study is the treatment duration of ten weeks. Other studies investigating the predictive value of early improvement under antidepressant treatment had a duration of usually 5 to 6 weeks (Katz et al., 2004; Stassen et al., 2007; Szegedi et al., 2003; van Calker et al., 2008). Regarding the long treatment duration of our study, one could have expected more limited predictive value indices of early improvement for later stable response and remission compared to indices reported in short-term trials. However, the herein reported indices are well comparable to those reported in short-term trials and thus underscore the usefulness of the identification of improvement in the early course of treatment, even for the prediction of treatment outcome 6–8 weeks later. Furthermore, we were able to compare active antidepressant treatment strategies like sertraline and CBT with corresponding control groups, i.e., a placebo- and a guided self-help group, respectively. We found lower rates of improvers in placebo-compared to sertraline-treated patients, but similar indices of the predictive value of early improvement for later favorable outcome. This result is fully in line with recent work of research groups focusing on placebo-drug differences (Papakostas et al., 2006; Posternak and Zimmerman, 2005; Stassen et al., 2007; Taylor et al., 2006). E.g., a recent large meta-analysis (Papakostas et al., 2006) with data from 4076 patients randomized to active antidepressants (AD) and 3045 patients randomized to placebo showed that verumtreated patients were more likely to experience clinical response by 2 weeks or even 1 week of treatment than placebo-treated patients. A further meta-analysis comprising 3418 placebo- and 5158 AD treated patients found nearly identical time courses of improvement on placebo and active medication: 60.2% and 61.6% of the improvement that occurred on active medication and placebo, respectively, took place within the first 2 weeks. The differences between drug and placebo were most pronounced during the first 2 weeks of treatment and diminished thereafter (Posternak and Zimmerman, 2005). We also found lower rates of improvers in GSG-compared to CBT-treated patients; however, early improvement under GSG-treatment was not predictive of later outcome. The most likely explanation is the low sample size of 4 and 5 patients out of 26 showing stable remission and stable response, respectively. In contrast to pill placebo, GSG did not represent a true “psychotherapy placebo” condition, because blinding of treatment was not possible and patients in this group were aware of being only in the control group. This knowledge might have resulted in negative effects (“nocebo effect”). Moreover, the therapists had to abstain from specific interventions and to play a quite passive role in the sessions, with negative consequences for the patient–therapist interaction and outcome (Hegerl et al., 2009). Thus, it is plausible that early improvement under these conditions could not be maintained over ten weeks and was therefore not predictive of later response. The results of the herein reported study are thus in line with several other reports on the development of improvement in the early course of antidepressant treatment and

its high predictive value for later outcome. Nevertheless, some limitations have to be taken into account. First, the sample sizes were relatively low in different treatment groups; therefore, the results of our study should be replicated in independent and optimally in larger populations. Second, the herein reported results are limited to the treatment with sertraline and CBT; in order to be able to generalize these results, prediction analyses have to be done with other antidepressant substances in comparable populations. An independent confirmation of our results would represent a step forward to profound changes of primary care treatment of depressed patients. Early and continued monitoring of treatment outcome could become a standard application, because it is time efficient and easy-to-use in the primary care sector. It can be easily applied and does not require expensive technical investment. Implementation only requires an assessment of depression severity at baseline and at (bi-)weekly intervals with an adequate scale. Given the potential for saving time and costs by using early improvement as a predictor of later outcome, a (bi-) weekly investment of 10–15 min for the rating of depressive symptoms would be reasonable. The treating physician would be prompted to optimize treatment in case of nonimprovement after 2–4 weeks. However, it should be pointed out that prospective randomized controlled trials comparing the treatment outcome in non-improvers who continue treatment compared to those who changed treatment in the early course are needed to validate this recommendation and to fully establish a critical point of decision after two weeks of antidepressant treatment to guide further strategy. In conclusion, early improvement occurs in a substantial proportion of patients with mild major, minor or subsyndromal depression under antidepressant treatment. Early improvement heralds a greater likelihood of response and remission in these patients. Furthermore, the lack of early improvement is associated with low rates of response and remission. Therefore, the identification of early improvement might be useful in clinical decision making in the early course of treatment in patients with mild major, minor or subsyndromal depression, who are typically treated by general practitioners. Role of funding source Funding for the study was provided by the German Ministry for Education and Research (BMBF; grant: 01 GI 9922/0222/0452) within the promotional emphasis “German Research Network on Depression and Suicidality”. The BMBF had no further role in 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 The authors have no conflicts of interest directly relevant to the content of this study.

Acknowledgements We want to thank the patients who participated in the study, the general practitioners who screened the patients, especially Veit Wambach, MD, and Vanadis Kamm-Kohl, MD, and the practice staff. We are indebted to Michael Schütze,


MD, Winfried Scheunemann, Ute Hägele, MD, and Patrick Bussfeld, MD, the physicians who investigated the patients at the specialized study center, Michaela König, MSc, Stephanie Lösch, MSc, Antje-Kathrin Allgaier, PhD, and Michael Stürmer, MSc who monitored the progress of the study and Professor Hans-Jürgen Möller, MD for his contributions to design and methods of the RCT reported here (trial registration: clinicaltrials. gov; URL: http://www.clinicaltrials.gov; registration number: NCT00226642). References Ackermann, R.T., Williams, J.W., 2002. Rational treatment choices for non-major depressions in primary care: an evidence-based review. J. Gen. Intern. Med.17, 293–301. American Psychiatric Association (APA), 2000. Practice guideline for the treatment of patients with major depressive disorder (revision). Am. J. Psychiatry 157 (suppl. 4), 1–45. Arzneimittelkommission der deutschen Ärzteschaft (AkdÄ), 2006. Empfehlungen zur Therapie der Depression. Arzneiverord. Prax. 33 (Sonderheft 1), 1–41. Backenstrass, M., Frank, A., Joest, K., Hingman, S., Mundt, C., Kronmuller, K.T., 2006. A comparative study of nonspecific depressive symptoms and minor depression regarding functional impairment and associated characteristics in primary care. Comput. Psychiatry 47, 35–41. Bauer, M., Bschor, T., Pfennig, A., Whybrow, P.C., Angst, J., Versiani, M., Möller, H.-J., WFSBP Task Force on Unipolar Depressive Disorders, 2007. World Federation of Societies of Biological Psychiatry (WFSBP) guidelines for biological treatment of unipolar depressive disorders in primary care. World J. Biol. Psychiatr. 8, 67–104. Hamilton, M., 1960. A rating scale for depression. J. Neurol. Neurosurg. Psychiatry 23, 56–62. Hautzinger, M., 2003. Kognitive Verhaltenstherapie bei Depressionen, 6th ed. Beltz/PVU, Weinheim. Hegerl, U., Pfeiffer-Gerschel, T., 2005. Das Kompetenznetz Depression, Suizidalität. Nervenheilkunde 24, 359–360. Hegerl, U., Hautzinger, M., Mergl, R., Kohnen, R., Schütze, M., Scheunemann, W., Allgaier, A.-K., Coyne, J., Henkel, V., 2009. Effects of pharmaco- and psychotherapy in depressed primary care patients. A randomized, controlled trial including a patient choice arm. Int. J. Neuropsychopharmacol. doi:10.1017/S1461145709000224. Henkel, V., Seemüller, F., Obermeier, M., Adli, M., Bauer, M., Mundt, C., Brieger, P., Laux, G., Bender, W., Heuser, I., Zeiler, J., Gaebel, W., Mayr, A., Möller, H.-J., Riedel, M., 2008. Does early improvement triggered by antidepressants predict response/remission? – analysis of data from a naturalistic study on a large sample of inpatients with major depression. J. Affect. Disord. doi:10.1016/j.jad.2008.10.011. Hermens, M.L.M., van Hout, H.P.J., Terluin, B., van der Windt, D.A.W.M., Beekman, A.T.F., van Dyck, R., de Haan, M., 2004. The prognosis of minor depression in the general population: a systematic review. Gen. Hosp. Psych. 26, 453–462. Hollon, S.D., Jarrett, R.B., Nierenberg, A.A., Thase, M.E., Trivedi, M., Rush, A.J., 2005. Psychotherapy and medication in the treatment of adult and geriatric depression: which monotherapy or combined treatment? J. Clin. Psychiatry 66, 455–468. Katz, M.M., Tekell, J.L., Bowden, C.L., Brannan, S., Houston, J.P., Berman, N., Frazer, A., 2004. Onset and early behavioral effects of pharmacologically different antidepressants and placebo in depression. Neuropsychopharmacology 29, 566–579. Keller, F., 2003. Analyse von Längsschnittdaten: Auswertungsmöglichkeiten mit hierarchischen linearen Modellen. Z. Klin. Psychol. Psychother. 32, 51–61.

93

Keller, F., Hautzinger, M., 2007. Klassifikation von Verlaufskurven in der Depressionsbehandlung. Z. Klin. Psychol. Psychother. 36, 83–92. Leucht, S., Shamsi, S.A., Busch, R., Kissling, W., Kane, J.M., 2008. Predicting antipsychotic drug response – replication and extension to six weeks in an international olanzapine study. Schizophr. Res. 101, 312–319. Lyness, J.M., Moonseong, H., Datto, C.J., Ten Have, T.R., Katz, I.R., Drayer, R., Reynolds, C.F., Alexopoulos, G.S., Bruce, M.L., 2006. Outcomes of minor and subsyndromal depression among elderly patients in primary care settings. Ann. Intern. Med. 144, 496–504. National Institute for Health and Clinical Excellence, 2004. CG23 Depression – NICE Guideline 2004: Management of Depression in Primary and Secondary Care. [http://www.nice.org.uk/guidance/index.jsp?action=byID&o=10958]. Nierenberg, A.A., Farabaugh, A.H., Alpert, J.E., Gordon, J., Worthington, J.J., Rosenbaum, J.F., Fava, M., 2000. Timing of onset of antidepressant response with fluoxetine treatment. Am. J. Psychiatry 157, 1423–1428. Papakostas, G.I., Perlis, R.H., Scalia, M.J., Petersen, T.J., Fava, M., 2006. A metaanalysis of early sustained response rates between antidepressants and placebo for the treatment of major depressive disorder. J. Clin. Psychopharmacol. 26, 56–60. Pincus, H.A., Davis, W.W., McQueen, L.E., 1999. ‘Subthreshold’ mental disorders: a review and synthesis of studies on minor depression and other ‘brand names’. Br. J. Psychiatry 174, 288–296. Pollack, M.H., Kornstein, S.G., Spann, M.E., Crits-Christoph, P., Raskin, J., Russell, J.M., 2008. Early improvement during duloxetine treatment of generalized anxiety disorder predicts response and remission at endpoint. J. Psychiatr. Res. 42, 1176–1184. Posternak, M.A., Zimmerman, M., 2005. Is there a delay in the antidepressant effect? A meta-analysis. J. Clin. Psychiatry 66, 148–158. Quitkin, F.M., Rabkin, J.G., Ross, D., Stewart, J.W., 1984. Identification of true drug response to antidepressants. Use of pattern analysis. Arch. Gen. Psychiatry 41, 782–786. Rapaport, M.H., Judd, L.L., Schettler, P.J., Yonders, K.A., Thase, M.E., Kupfer, D.J., Frank, E., Plewes, J.M., Tollefson, G.D., Rush, A.J., 2002. A descriptive analysis of minor depression. Am. J. Psychiatry 159, 637–643. Schouten, H.J., 1995. Adaptive biased urn randomisation in small strata when blinding is impossible. Biometrics 51, 1529–1535. Stassen, H.H., Angst, J., Hell, D., Scharfetter, C., Szegedi, A., 2007. Is there a common resilience mechanism underlying antidepressant drug response? Evidence from 2848 patients. J. Clin. Psychiatry 68, 1195–1205. Szegedi, A., Müller, M.J., Anghelescu, I., Klawe, C., Kohnen, R., Benkert, O., 2003. Early improvement under mirtazapine and paroxetine predicts later stable response and remission with high sensitivity in patients with major depression. J. Clin. Psychiatry 64, 413–420. Taylor, M.J., Freemantle, N., Geddes, J.R., Bhagwagar, Z., 2006. Early onset of selective serotonin reuptake inhibitor antidepressant action: systematic review and meta-analysis. Arch. Gen. Psychiatry 63, 1217–1223. Van, H.L., Schoevers, R.A., Kool, S., Hendriksen, M., Peen, J., Dekker, J., 2008. Does early response predict outcome in psychotherapy and combined therapy for major depression? J.Affect. Disord. 105, 261–265. van Calker, D., Zobel, I., Dykierek, P., Deimel, C.M., Kech, S., Lieb, K., Berger, M., Schramm, E., 2008. Time course of response to antidepressants: predictive value of early improvement and effect of additional psychotherapy. J. Affect. Disord 114, 243–253. Wagner, H.R., Burns, B.J., Broadhead, W.E., Yarnall, K.S.H., Sigmon, A., Gaynes, B.N., 2000. Minor depression in family practice: functional morbidity, comorbidity, service utilization and outcomes. Psychol. Med. 30, 1377–1390. Williams Jr., J.W., Kerber, C.A., Mulrow, C.D., Medina, A., Aguilar, C., 1995. Depressive disorders in primary care: prevalence, functional disability, and identification. J. Gen. Intern. Med. 10, 7–12. Wittchen, H.-U., Pfister, H. (Eds.), 1997. DIA-X-Interview. Instruktionsmanual zur Durchführung von DIA-X Interviews. Swets & Zeitlinger, Frankfurt am Main. World Health Organization, 1993. Composite International Diagnostic Interview (CIDI), Version 1.1. American Psychiatric Press, Washington DC.