- Email: [email protected]
Response trajectories in “real-world” naturalistically treated schizophrenia patients
Schizophrenia Research 139 (2012) 218–224
Contents lists available at SciVerse ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
Response trajectories in “real-world” naturalistically treated schizophrenia patients Rebecca Schennach a,⁎, Sebastian Meyer a, Florian Seemüller a, Markus Jäger a, Max Schmauss b, Gerd Laux c, Herbert Pfeiffer d, Dieter Naber e, Lutz G. Schmidt f, Wolfgang Gaebel g, Joachim Klosterkötter h, Isabella Heuser i, Wolfgang Maier j, Matthias R. Lemke k, Eckart Rüther l, Stefan Klingberg m, Markus Gastpar n, Richard Musil a, Hans-Jürgen Möller a, Michael Riedel a, o a
Department of Psychiatry and Psychotherapy, Ludwig-Maximilians-University, Munich, Germany Psychiatric Clinic, District Hospital Augsburg, Germany c Psychiatric Clinic, Inn-Salzach Hospital Wasserburg/Inn, Germany d Psychiatric Clinic, Isar-Amper Hospital, Munich-Haar, Germany e Department of Psychiatry and Psychotherapy, University of Hamburg, Germany f Department of Psychiatry and Psychotherapy, University of Mainz, Germany g Department of Psychiatry and Psychotherapy, Medical Faculty, Heinrich-Heine-University Duesseldorf, Germany h Department of Psychiatry and Psychotherapy, University of Cologne, Germany i Department of Psychiatry and Psychotherapy, Charite Berlin, Campus Benjamin Franklin, Germany j Department of Psychiatry and Psychotherapy, University of Bonn, Germany k Department of Psychiatry, Alsterdorf Hospital, Hamburg, Germany l Department of Psychiatry and Psychotherapy, University of Göttingen, Germany m Department of Psychiatry and Psychotherapy, University of Tübingen, Germany n Department of Psychiatry and Psychotherapy, University of Essen, Germany o Psychiatric Clinic, Vinzenz-von-Paul-Hospital, Rottweil, Germany b
a r t i c l e
i n f o
Article history: Received 21 November 2011 Received in revised form 13 April 2012 Accepted 7 May 2012 Available online 2 June 2012 Keywords: Schizophrenia Response trajectory Heterogeneity Group membership
a b s t r a c t Background: To date, research has identified distinct antipsychotic response trajectories yet focussing on data from randomized-controlled trials (RCTs). Therefore, the heterogeneity of response in “real-world” schizophrenia patients is still unknown. Methods: Antipsychotic response was evaluated in 399 patients suffering from a schizophrenia spectrum disorder within a naturalistic multicenter study of the Competence Network on Schizophrenia using latent class regression. Baseline and illness-related variables were compared between the different trajectory classes as well as currently proposed outcome definitions (early improvement, response, remission) using univariate tests. In order to predict the trajectory group membership classification and regression tree analysis were furthermore performed. Results: Five distinct trajectories of antipsychotic response were identified: Class 1 (15%) showing an early and considerable improvement, Class 2 (14%) incorporating patients with the greatest response to treatment, Class 3 (34%) again showing an early improvement to treatment yet with a slightly lower degree of improvement, Class 4 (22%) featuring patients gradually responding to treatment, and Class 5 (15%) with the poorest antipsychotic response. Fewer depressive symptoms at admission, better functioning, a shorter duration of illness and less previous hospitalizations were found to be significant predictors of good response. No considerable differences were found comparing the present results to the previous trajectory analyses deriving from RCTs. Conclusion: Our results underline the heterogeneous course of response independent of the study or treatment design suggesting that the diversity in schizophrenia response and outcome is determined primarily by different pathophysiological underpinnings. © 2012 Elsevier B.V. All rights reserved.
1. Introduction
⁎ Corresponding author at: Department of Psychiatry and Psychotherapy, LudwigMaximilians-University Munich, Nussbaumstrasse 7, 80336 München, Germany. Tel.: + 49 89 5160 5511; fax: + 49 89 5160 5728. E-mail address: [email protected] (R. Schennach). 0920-9964/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.schres.2012.05.004
In modern psychiatry, schizophrenia is believed to be a severe and heterogeneous disorder resulting in a favorable outcome in some patients although others may suffer from a deteriorating course of the illness (Wiersma et al., 1998). Since the introduction of antipsychotic treatment clinicians and researchers have tried to understand
R. Schennach et al. / Schizophrenia Research 139 (2012) 218–224
patterns of treatment response in schizophrenia and to identify and support patients in danger of a poor response in order to adopt specific treatment adaptations improving course and outcome of the illness (Gaebel, 1996). However, despite significant research in this field factors determining response to antipsychotic treatment are still not fully understood with inconclusive and contradictory results (Correll et al., 2003). But, understanding the characteristics of patients not responding well to treatment would be the first step to find better treatment strategies for them. To identify patients responding well or poorly to antipsychotic treatment, traditionally predefined thresholds are applied to categorize patients into responders on one side and non-responders on the other (Stauffer et al., 2011). This dichotomizing approach however bears numerous problems. Often the applied cut-offs are arbitrarily chosen and dichotomizing a continuous score was found to lead to inefficient analyses (Uher et al., 2010). Therefore, in order to shed light on actions of antipsychotic response in schizophrenia a new statistical method has been introduced lately to quantify the extent of heterogeneity in trajectories of response using so called mixed mode latent regression modeling. This approach bears the advantage of categorizing patients based on temporal patterns of change searching for heterogeneity as it naturally occurs in clinical data (Muthen et al., 2002). Levine and Leucht were among the first to perform growth mixture modeling in schizophrenia patients and reported five distinct treatment response trajectories characterized by varied amelioration levels (Levine and Leucht, 2010). Three trajectory classes showed a treatment response trend of amelioration, one class only a small reduction on the Brief Psychotic Rating Scale and one class featured a considerable symptom reduction during the first two weeks (Levine and Leucht, 2010). Also, Case et al. calculated response trajectories finding four distinct response trajectories in contrast to Levine and Leucht, however, identifying very similar response patterns with most patients achieving a moderate improvement with rapid symptom improvement in 12.5% of the patients and 2.3% with unsustained improvement (Case et al., 2010). The authors suggested that the observed heterogeneity might represent specific endophenotypes of response with different etiologic underpinnings (Case et al., 2010). In the meanwhile, several authors calculated treatment response trajectories in schizophrenia patients, however, all previously published articles focus solely on data deriving from randomized controlled trials examining different atypical antipsychotics (Levine and Rabinowitz, 2010; Marques et al., 2010; Levine et al., 2011; Stauffer et al., 2011). This limits the clinically relevant implications concurrently leaving out results of treatment with typical compounds or combination treatments. Besides, in terms of predicting the trajectory group membership only few variables have been examined so far regarding their predictive validity (Levine and Rabinowitz, 2010). Also, it is unclear how far the response trajectories overlap with currently established response and outcome definitions which might help to evaluate their clinical adequacy and use. Therefore, to expand prior research, we wanted to perform growth mixture modeling in schizophrenia patients treated within a naturalistic trial analyzing a variety of sociodemographic and clinical variables in terms of their value to predict the patient's response trajectory class. 2. Methods 2.1. Subjects Data were collected at eleven psychiatric university hospitals and three psychiatric district hospitals within a multicenter follow-up program by the German Research Network on Schizophrenia (Wolwer et al., 2003). All patients admitted to one of the above mentioned hospitals between January 2001 and December 2004 with a diagnosis of schizophrenia, schizophreniform disorder, delusional
219
disorder and schizoaffective disorder according to DSM-IV criteria were selected for inclusion. Subjects were aged between 18 and 65 years. Exclusion criteria were the presence of a head injury, a history of major medical illness and alcohol or drug dependency. An informed written consent had to be provided to participate in the study. The study protocol was approved by the local ethics committees (Jager et al., 2007). 2.2. Assessments DSM-IV diagnoses were established by clinical researchers on the basis of the German version of the Structured Clinical Interview for DSM-IV (American Psychiatric Association, 1994). Sociodemographic and course-related variables were collected using a standardized documentation system (Cording, 1998). To assess symptom severity and the level of antipsychotic response the Positive and Negative Syndrome Scale for Schizophrenia (PANSS) (Kay et al., 1988) was applied. To compare the patients within the different response trajectories further rating scales were used. The 17-item version of the Hamilton Depression Rating Scale (Hamilton, 1960) and the Udvalg for Kliniske Undersogelser (UKU) Side Effect Rating Scale (Lingjaerde et al., 1987) were applied to evaluate the level of depressive symptoms and adverse events, respectively. The patient's functioning was examined via the Global Assessment of Functioning Scale (GAF) (American Psychiatric Association, 1994) and the Social Occupational Functioning Scale (SOFAS) (American Psychiatric Association, 1994). To evaluate the patients' premorbid adjustment the short-scale for premorbid social-personal adjustment of the Phillips Scale was applied (Phillips, 1953). Ratings were assessed by trained clinicians at baseline and subsequently every two weeks until discharge. All raters had been trained using the applied scales. A high inter-rater reliability was achieved (ANOVA-ICC > 0.8 based on the PANSS). In order to establish and maintain the high interrater reliability interactive video-based rater-training sessions were regularly performed throughout the study period in every participating hospital. To further assure an accurate and consistent documentation the collected data were sent to the study center on a regular basis to be checked for completeness and coherence. Rater seminars between the centers were held before the study began. 2.3. Statistical analysis In a first step, different courses of treatment response were identified using latent class regression, also known as latent class growth analysis (LCGA). To model the course of treatment response, 49 different parameterizations with orthogonal polynomials (1–7°), semiparametric B-splines (1–7°, 2–9 df), or natural cubic splines (3–9 df) were investigated. Trading off between goodness-of-fit and complexity, the model minimizing the Bayesian Information Criterion (BIC) was chosen. Models were estimated by the classification–expectation– maximization (CEM) algorithm. However, in order to obtain regular maximum likelihood estimates, the selected model was finally reestimated by the original CEM algorithm using the maximum a posteriori (MAP) classification of the best CEM fit as initial configuration. The identified clusters resulting from the MAP classification were numbered according to the percentage reduction in the PANSS total score in each class. In a second step, the identified subgroups of patients were compared on various clinical and baseline characteristics using Fisher's exact test for categorical variables, and the Kruskal–Wallis test or the ANOVA F-test for metric variables, depending on the assumption of normality. The selection of these clinical and baseline characteristics goes back to previous publications of response trajectories deriving from randomized controlled trials in order to compare whether there would be a difference in the potential influencing variables
220
R. Schennach et al. / Schizophrenia Research 139 (2012) 218–224
depending on the study design (naturalistic versus randomized controlled study design). In a third step, classification and regression tree (CART) analysis was performed using a series of if-then rules (Breiman et al., 1984; Clark and Pregibon, 1992). Early improvement was defined as a 20% PANSS total score reduction from admission to week 2 (Kinon et al., 2010) and response as a 50% PANSS total score reduction from admission to discharge (Leucht et al., 2007). We followed the recommendations of Obermeier et al. concerning the calculation of percent changes in the PANSS (Obermeier et al., 2009). Remission was defined according to the symptom-severity component of the consensus remission criteria proposed by the Remission in Schizophrenia Working Group (Andreasen et al., 2005). All analyses were performed using the statistical software environment R 2.13.0 (R Development Core Team, 2008) and the contributed package flexmix (Leisch, 2004). 3. Results 3.1. Patients In the entire multicenter study 474 patients were enrolled. 46 patients had to drop out for different reasons, another 29 patients were excluded from analysis: 28 patients because they were discharged from the hospital within 7 days after admission and 1 patient due to incomplete psychopathology ratings. Therefore, the available sample comprised 399 subjects (225 males/174 females). The mean age was 35.4 years (±11.1), the mean duration of illness was 7.1 years (±12) and the mean duration of the current hospitalization was 64.7 days (±46). 31% of these patients suffered from their first illness episode. At admission, 20% of the patients suffered from suicidality. During the study patients were treated under naturalistic conditions as follows: 81% of the patients received first-generation antipsychotics, 80% of patients had second-generation antipsychotic treatment and 65% of the patients were treated with first as well as second-generation antipsychotics. Tranquilizers were administered in 79% of the patients and mood stabilizers in 16%. 27% of the patients were also treated with antidepressants. 3.2. Assessments 3.2.1. Response trajectory classes (Fig. 1)
• Trajectory Class 1 — (=early and considerable response) This class comprised 61 patients (15%) with an average PANSS baseline score of 45.8 points indicating that these patients suffered from only mild symptoms already at admission. Despite the low PANSS total score at admission the relative change of the PANSS score during inpatient treatment was 63.9%. 80% of the patients were early treatment improvers, 77% were responders and as many as 95% of the patients in Class 1 achieved the symptom severity component of the remission criteria at discharge (see Fig. 2). • Trajectory Class 2 — (=rapid and dramatic response) 54 patients (14%) were found to belong to response Class 2 with an initially high PANSS total score at admission and a dramatic improvement during treatment (62% PANSS total score change). This class is characterized by a very early and dramatic response with considerable psychopathological improvement in the first treatment weeks yet with some patients showing an unsustained response pattern (Fig. 2). • Trajectory Class 3 — (=early and satisfying response) Class 3 comprised 137 patients (34%) with an average PANSS baseline score of 64.1 points. The relative change of the PANSS score during inpatient treatment was 57.7%. Patients in this class showed an early and considerable response to treatment with an
Fig. 1. Response trajectories of naturalistically treated schizophrenia inpatients.
unsustained course of symptoms after the initial response to treatment in some patients of this class (Fig. 2). • Trajectory Class 4 — (=gradual response) Class 4 comprised 89 patients (22%) with an initially moderate PANSS total score at admission (72.6 points) and a relative change of the PANSS total score during treatment of 31%. Response to treatment occurred very gradually in this class. Fig. 2 shows the course of response of Class 4 indicating that possibly the degree of response might have been more satisfying if the study's observational time would have been longer for the response pattern in fact is gradual, but continuous. • Trajectory Class 5 — (=partial response) Class 5 comprises 58 patients (15%) with the worst response to treatment. Patients belonging to this class were rather symptomatic at admission (92.8 points) with only a relative change of the PANSS total score of 26.3%. Of the 58 patients 24% were early improvers with a similar number of patients being in response at discharge. The greatest variability of response and courses of response was found in this response class (see Fig. 2). 3.2.2. Baseline differences and outcome comparing the different trajectory groups Several significant differences in terms of sociodemographic and illness related variables were found between the five response classes (see Table 1). Patients with the greatest response to treatment were the ones with less severe positive and negative symptoms and better functioning at admission and with less illness chronicity (shorter duration of illness, fewer illness episodes and less numbers of previous hospitalizations). The patients with the earliest onset of their illness were found to be mainly in Class 5 with the poorest response to treatment. No significant differences were found comparing the different trajectory groups in terms of age, gender and the diagnostic subtype. The kind of antipsychotic treatment (atypical/typical/atypical plus typical) was compared between the different response classes yet without finding a significant difference between them (p = 0.87) (see Table 1). 3.2.3. Predicting the trajectory class membership Using a classification and regression tree analysis we wanted to identify the variables that would best separate the different response classes. Less depressive symptoms at admission, a higher level of functioning and less illness chronicity (fewer number of previous
R. Schennach et al. / Schizophrenia Research 139 (2012) 218–224
221
Fig. 2. Trajectories of each individual patient for the identified response trajectory classes.
hospitalizations/shorter duration of illness) were found to be significant predictors of a favorable trajectory class membership (Classes 1–3) (see Fig. 3).
4. Discussion 4.1. Response patterns within a “real-world” naturalistic treatment of schizophrenia patients In naturalistically treated schizophrenia patients five different response trajectories could be identified ranging from patients with an early and clinically considerable improvement to patients with only poor improvement underlining the heterogeneity of antipsychotic response. Response was understood to be an improvement of the patients' general psychopathological symptoms and did not focus on positive symptoms. Today response to treatment is believed to be heterogeneous with a variety of different courses of the illness with pathobiological and pathophysiological differences responsible for the differing patterns (Lieberman et al., 1993). This diversity in response and outcome in schizophrenia patients is also underlined by the results of cluster analyses stating that there is no “good” versus “bad” outcome dichotomy, but that outcome in schizophrenia is multidimensional (Lipkovich et al., 2009). Lipkovich et al. for example reported five different groups when analyzing the variety of outcome in 1449 schizophrenia patients treated up to 30 weeks ranging from patients with non‐existent to minimal symptoms and mild functional impairments (25.6%) to patients with moderate to severe psychiatric symptoms and severe functional deficits (14.8%) (Lipkovich et al., 2009). Also, long-term results (36 months of follow-up) emphasize the multifaceted course of schizophrenia and the wide range of outcomes (good/intermediate/poor) (Di and Bolino, 2004). Interestingly, our results found a rather early improvement in almost all patients (except those in Class 5) which indicate that despite
the diversity in treatment response there are also similarities in the response patterns of schizophrenia patients. This underlines the early-onset hypothesis proposed by Agid et al. stating that generally greater improvement occurs in the first 2 weeks of treatment than in the subsequent 2 treatment weeks (Agid et al., 2003). To our knowledge, this is one of the first studies explicitly evaluating the association between the response trajectory and different predefined outcome criteria. As one would expect we found patients of the very favorable response classes (Classes 1–3) to easily achieve the recommended criteria of early improvement at week 2 and response at discharge which suggests that these criteria adequately mirror the natural response to antipsychotic treatment. We found more patients achieving the symptom severity component of the remission criteria compared to the response criterion in Classes 1 and 3. It has been reported before that an at least 50% PANSS reduction is more stringent than the remission criteria in a short-term analysis (Beitinger et al., 2008). Given that response and remission provide different information on the patient's condition they should always both be implemented in schizophrenia trials as recommended by Leucht and Kane (2006).
4.2. Comparing response classes of naturalistically treated patients with previously performed trajectory studies in randomized-controlled trials Comparing the present results to previously performed response trajectory analyses exclusively based on data deriving from randomized-controlled trials (RCTs) we did not find any considerable differences in terms of the number and the course of the trajectory classes (Case et al., 2010; Levine and Leucht, 2010; Levine and Rabinowitz, 2010; Levine et al., 2010; Marques et al., 2010; Stauffer et al., 2011). This is interesting, for one could have hypothesized that due to the strict inclusion and exclusion criteria of randomizedcontrolled trials more severely ill, yet less chronically ill patients
222
R. Schennach et al. / Schizophrenia Research 139 (2012) 218–224
Table 1 Baseline and course-related variables, psychopathology and functioning comparing the different response trajectories. Class 1 (N = 15%)
Class 2 (N = 14%)
Class 3 (N = 34%)
Class 4 (N = 22%)
Class 5 (N = 15%)
46%
67%
51%
65%
57%
0.052
75% 11% 14% 36% 88% 66%
92% 6% 2% 31% 72% 47%
82% 11% 7% 40% 69% 55%
82% 16% 2% 17% 74% 69%
86% 12% 2% 22% 57% 64%
0.081
75% 7% 11% 7% 15%
77% 5% 10% 8% 19%
72% 9% 9% 10% 14%
69% 6% 21% 4% 27%
79% 5% 9% 7% 28%
Mean (SD) Age (years) Age at onset (years)
36.2 (± 10.24) 30.6 (± 10.34)
35.2 (± 10.96) 29 (± 8.29)
36.1 (± 11.25) 29.5 (± 10.07)
34.2 (± 11.62) 23.9 (± 7.20)
34.9 (± 11.10) 25.1 (± 8.47)
0.71 b 0.0001
Illness related variables [median (IQR)] Duration of illness (years) Number of previous hospitalizations Duration of current hospitalization (days) Premorbid adjustment
2 (± 7) 1 (± 2) 35 (± 32) 2 (± 3)
2 (± 11) 0 (± 2) 63 (± 42) 2 (± 2)
2 (± 10) 0 (± 3) 51 (± 37) 2 (± 2)
4 (± 16) 2.5 (± 5) 64 (± 52) 3 (± 2)
7 (± 16) 1 (± 4.75) 77 (± 78.5) 3 (± 1.50)
0.0045 0.0011 b 0.0001 0.0038
Psychopathology at admission PANSS total score PANSS positive subscore PANSS negative subscore PANSS general psychopathology Insight into illness (PANSS G12 item) HAMD
45.8 (± 6.2) 13.6 (± 3.9) 9.4 (± 2.4) 22.8 (± 3.3) 6.8 (± 5.61)
92.4 24.2 23.4 44.8 17.6
64.1 (± 10.3) 18.5 (± 5.9) 14.6 (± 5.2) 31 (± 6.1) 11.9 (± 8.18)
72.6 18.3 19.4 34.9 14.2
92.8 22.5 24.7 45.7 21.3
b 0.0001a b 0.0001a b 0.0001a b 0.0001a b 0.0001
Treatment tolerability (UKU)
3.7 (± 4.91)
4.8 (± 7.37)
4.5 (± 6.00)
6.3 (± 6.27)
6.9 (± 6.99)
Functioning at admission GAF SOFAS
51.6 (± 12.44) 43.2 (± 11.46)
40.7 (± 11.24) 42.1 (± 11.24)
42.5 (± 11.22) 52.3 (± 10.69)
42.1 (± 10.53) 43.1 (± 10.71)
35.3 (± 10.01) 41.1 (± 10.57)
Antipsychotic treatment during hospitalization Atypical antipsychotics Typical antipsychotics Atypical and typical antipsychotics
18% 24% 58%
13% 13% 74%
17% 18% 65%
15% 17.5% 67.5%
13% 15% 72%
Sociodemographic variables [%] Gender (male) Diagnostic subtype Schizophrenia Schizoaffective Brief psychotic disorder First illness episode Duration of current episode b 6 months Antipsychotic premedication Reason for admission Psychopathological worsening Suicidality Acute psychosocial crisis Other Suicidality at admission
a
(± 14.5) (± 5.7) (± 5.5) (± 9.1) (± 8.29)
(± 8.5) (± 5.6) (± 6.1) (± 5.6) (± 7.69)
p-value
0.002 0.0068 0.047 0.65
0.058
(± 15.5) (± 5.8) (± 6.3) (± 9) (± 8.69)
0.02
b 0.0001 b 0.0001
0.87
The PANSS was used to group the trajectory classes so that regular test conditions to compare these items between the classes are not fulfilled.
might have been included in those studies resulting in different response patterns. However, in line with the study by Stauffer et al., Levine and Leucht and Levine and Rabinowitz we found five different
response trajectories enclosing dramatic responders, partial and unsustained responders and delayed and poor responders (Levine and Leucht, 2010; Levine and Rabinowitz, 2010; Stauffer et al., 2011).
Fig. 3. Predicting the response trajectory class.
R. Schennach et al. / Schizophrenia Research 139 (2012) 218–224
Other authors reported of only four response classes, however, the course and description of the trajectory classes remain the same (Case et al., 2010). Interestingly, all published trajectory analyses match especially in finding one response trajectory class with a rapid and significant response to treatment, often representing around 10–20% of the patient sample (Case et al., 2010; Levine and Leucht, 2010; Levine and Rabinowitz, 2010; Levine et al., 2010; Marques et al., 2010; Stauffer et al., 2011). In their analysis of 628 patients with schizophrenia and schizoaffective disorder Case et al. found the rapid response class to be uniquely represented by 100% early improvers (≥20% PANSS total score improvement within the first two treatment weeks) (Case et al., 2010). Marques et al. examined different trajectories of antipsychotic response in comparison to placebo and reported that in their subset of patients with rapid response all patients were from the drug-treated group showing more than a 70% improvement (Marques et al., 2010). In the present study patients with dramatic response showed a relative change of 62% in the PANSS total score which lies in between the results of the comparative studies. 4.3. Predicting the response trajectory group membership We were able to identify several clinical significant predictors of the membership of a favorable response trajectory, namely less baseline depressive symptoms, better baseline functioning, a shorter duration of illness and less previous hospitalizations in the psychiatric history. Depressive symptoms are believed to be a major problem for illness course and outcome in schizophrenia patients (Addington et al., 1998) and have been associated with poor quality of life and an increased risk of relapse (Zisook et al., 1999). Also, impairments in functioning evaluated via two different rating scales (GAF and SOFAS) were found to significantly differ between the response trajectories. Generally, the patient's functioning is linked to the severity of psychopathological symptoms and the course of outcome (Schennach-Wolff et al., 2009). However, it should be kept in mind that not only the level of functioning influences the course of symptoms, but that in turn the degree of symptoms also influences the level of functioning so that both of these domains somehow predict each other. Variables like a longer duration of illness or a greater number of hospitalizations generally stand for a less favorable course of schizophrenia and have consistently been associated with a lower chance to respond to treatment as well as to achieve remission and recovery (Marshall et al., 2005; Haro et al., 2008). In terms of the above discussed comparative trajectory analyses only few authors have performed prediction models in order to examine the membership in the best or worst treatment response trajectories. Levine and Rabinowitz calculated binary logistic regression models to examine the class membership and found a better premorbid functioning, better cognitive functioning and not having a diagnosis of schizophrenia to be predictive of belonging to the best response trajectory (Levine and Rabinowitz, 2010). However, it is unclear what potential influencing variables were examined in their study which limits the significance of their results. In a different analysis on 236 recentepisode psychosis patients Levine et al. reported that the membership with the least improvement was significantly predicted by higher PANSS baseline scores and again membership in the trajectory with the most improvement was predicted by better premorbid functioning, the absence of a schizophrenia diagnosis and a higher cognitive score (Levine et al., 2010). 4.4. Strengths and limitations The major strength of this study is that it was performed within a naturalistic study design mirroring the “real-world” treatment conditions of schizophrenia patients allowing reliable clinical conclusions. Also, due to the liberal inclusion and exclusion criteria, findings of
223
this study on treatment seeking patients might be more generalizable and exhibit higher external validity than results from RCTs. However, such a design does not allow sufficient control of study results for the effect of different pharmacological treatments which limits specific association between the treatment applied and the response pattern. Besides, given that specific health system aspects influence the practice of hospitalizing and treating mentally ill patients, our study results might not be generalizable and comparable to others, especially those of non-European countries. 4.5. Conclusion Response trajectory analysis of naturalistically treated schizophrenia patients identified five different response trajectories. Only a smaller proportion of patients suffered from a poor response with a very similar number of patients achieving rapid and considerable improvement. Generally, a more favorable response to treatment was associated with and predicted by a shorter duration and later onset of illness, less baseline illness severity and better functioning. Response trajectory analysis allows the evaluation of different response patterns without using predefined cut-offs or time-points emphasizing the heterogeneity of response in schizophrenia patients. Role of funding source The study was performed within the framework of the German Research Network on Schizophrenia, which is funded by the German Federal Ministry for Education and Research BMBF (grant 01 GI 0233). Contributors The German Research Network on Schizophrenia designed the study and wrote the protocol. All authors have contributed to and have approved the final manuscript. Authors are as follows: Rebecca Schennach, Sebastian Meyer, Florian Seemüller, Markus Jäger, Richard Musil, Hans-Jürgen Möller, Michael Riedel from the Department of Psychiatry and Psychotherapy, Ludwig-Maximilians-University, Munich, Germany; Max Schmauss from the Psychiatric Clinic, District Hospital Augsburg, Germany; Gerd Laux from the Psychiatric Clinic, Inn-Salzach Hospital Wasserburg/Inn, Germany; Herbert Pfeiffer from the Psychiatric Clinic, Isar-Amper Hospital, Munich-Haar, Germany; Dieter Naber from the Department of Psychiatry and Psychotherapy, University of Hamburg, Germany; Lutz G. Schmidt from the Department of Psychiatry and Psychotherapy, University of Mainz, Germany; Wolfgang Gaebel from the Department of Psychiatry and Psychotherapy, Heinrich-Heine-University Duesseldorf, Germany; Joachim Klosterkötter from the Department of Psychiatry and Psychotherapy, University of Cologne, Germany; Isabella Heuser from the Department of Psychiatry and Psychotherapy, Charite Berlin, Campus Benjamin Franklin, Germany; Wolfgang Maier from the Department of Psychiatry and Psychotherapy, University of Bonn, Germany; Matthias R. Lemke from the Department of Psychiatry, Alsterdorf Hospital, Hamburg, Germany; Eckart Rüther from the Department of Psychiatry and Psychotherapy, University of Göttingen, Germany; Stefan Klingberg from the Department of Psychiatry and Psychotherapy, University of Tübingen, Germany and Markus Gastpar from the Department of Psychiatry and Psychotherapy, University of Essen, Germany. Conflict of interest All authors declare that they have no conflicts of interest within the context of this article. Acknowledgments The study was performed within the framework of the German Research network on Schizophrenia, which is funded by the German Federal Ministry for Education and Research BMBF (grant 01 GI 0233).
References Addington, D., Addington, J., Patten, S., 1998. Depression in people with first-episode schizophrenia. Br. J. Psychiatry Suppl. 172, 90–92. Agid, O., Kapur, S., Arenovich, T., Zipursky, R.B., 2003. Delayed-onset hypothesis of antipsychotic action: a hypothesis tested and rejected. Arch. Gen. Psychiatry 60, 1228–1235. American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Mental Disorders, Fourth edition. American Psychiatric Association, Washington DC. Andreasen, N.C., Carpenter Jr., W.T., Kane, J.M., Lasser, R.A., Marder, S.R., Weinberger, D.R., 2005. Remission in schizophrenia: proposed criteria and rationale for consensus. Am. J. Psychiatry 162, 441–449. Beitinger, R., Lin, J., Kissling, W., Leucht, S., 2008. Comparative remission rates of schizophrenic patients using various remission criteria. Prog. Neuropsychopharmacol. Biol. Psychiatry 32, 1643–1651.
224
R. Schennach et al. / Schizophrenia Research 139 (2012) 218–224
Breiman, L., Friedman, H.J., Olshen, R.A., Stone, C.J., 1984. Classification and Regression Trees. Wadsworth International Group, Belmont, CA. Case, M., Stauffer, V.L., scher-Svanum, H., Conley, R., Kapur, S., Kane, J.M., KollackWalker, S., Jacob, J., Kinon, B.J., 2010. The heterogeneity of antipsychotic response in the treatment of schizophrenia. Psychol. Med. 1–10. Clark, L.A., Pregibon, D., 1992. Tree-based models. In: Chambers, J., Hastie, T. (Eds.), Statistical Models. Wadsworth & Books/Cole Advanced Books & Software, Pacific Grove, CA. Cording, C., 1998. Conceptual aspects in development and implementation of basic psychiatric documentation. Psychiatr. Prax. 25, 175–178. Correll, C.U., Malhotra, A.K., Kaushik, S., McMeniman, M., Kane, J.M., 2003. Early prediction of antipsychotic response in schizophrenia. Am. J. Psychiatry 160, 2063–2065. Di, M.V., Bolino, F., 2004. The natural course of schizophrenia and psychopathological predictors of outcome. A community-based cohort study. Psychopathology 37, 98–104. Gaebel, W., 1996. Prediction of response to acute neuroleptic treatment in schizophrenia. Int. Clin. Psychopharmacol. 11 (Suppl. 2), 47–54. Hamilton, M., 1960. A rating scale for depression. J. Neurol. Neurosurg. Psychiatry 23, 56–62. Haro, J.M., Novick, D., Suarez, D., Ochoa, S., Roca, M., 2008. Predictors of the course of illness in outpatients with schizophrenia: a prospective three year study. Prog. Neuropsychopharmacol. Biol. Psychiatry 32, 1287–1292. Jager, M., Riedel, M., Messer, T., Laux, G., Pfeiffer, H., Naber, D., Schmidt, L.G., Gaebel, W., Huff, W., Heuser, I., Kuhn, K.U., Lemke, M.R., Ruther, E., Buchkremer, G., Gastpar, M., Bottlender, R., Strauss, A., Moller, H.J., 2007. Psychopathological characteristics and treatment response of first episode compared with multiple episode schizophrenic disorders. Eur. Arch. Psychiatry Clin. Neurosci. 257, 47–53. Kay, S.R., Opler, L.A., Lindenmayer, J.P., 1988. Reliability and validity of the positive and negative syndrome scale for schizophrenics. Psychiatry Res. 23, 99–110. Kinon, B.J., Chen, L., Ascher-Svanum, H., Stauffer, V.L., Kollack-Walker, S., Zhou, W., Kapur, S., Kane, J.M., 2010. Early response to antipsychotic drug therapy as a clinical marker of subsequent response in the treatment of schizophrenia. Neuropsychopharmacology 35, 581–590. Leisch, F., 2004. FlexMix: a general framework for finite mixture models and latent class regression in R. J. Stat. Softw. 11, 1–18. Leucht, S., Kane, J.M., 2006. Measurement-based psychiatry: definitions of response, remission, stability, and relapse in schizophrenia. J. Clin. Psychiatry 67, 1813–1814. Leucht, S., Davis, J.M., Engel, R.R., Kane, J.M., Wagenpfeil, S., 2007. Defining ‘response’ in antipsychotic drug trials: recommendations for the use of scale-derived cutoffs. Neuropsychopharmacology 32, 1903–1910. Levine, S.Z., Leucht, S., 2010. Elaboration on the early-onset hypothesis of antipsychotic drug action: treatment response trajectories. Biol. Psychiatry 68, 86–92. Levine, S.Z., Rabinowitz, J., 2010. Trajectories and antecedents of treatment response over time in early-episode psychosis. Schizophr. Bull. 36, 624–632. Levine, S.Z., Rabinowitz, J., Case, M., scher-Svanum, H., 2010. Treatment response trajectories and their antecedents in recent-onset psychosis: a 2-year prospective study. J. Clin. Psychopharmacol. 30, 446–449. Levine, S.Z., Lurie, I., Kohn, R., Levav, I., 2011. Trajectories of the course of schizophrenia: from progressive deterioration to amelioration over three decades. Schizophr. Res. 126, 184–191.
Lieberman, J., Jody, D., Geisler, S., Alvir, J., Loebel, A., Szymanski, S., Woerner, M., Borenstein, M., 1993. Time course and biologic correlates of treatment response in first-episode schizophrenia. Arch. Gen. Psychiatry 50, 369–376. Lingjaerde, O., Ahlfors, U.G., Bech, P., Dencker, S.J., Elgen, K., 1987. The UKU side effect rating scale. A new comprehensive rating scale for psychotropic drugs and a crosssectional study of side effects in neuroleptic-treated patients. Acta Psychiatr. Scand. Suppl. 334 (1–100), 1–100. Lipkovich, I.A., Deberdt, W., Csernansky, J.G., Buckley, P., Peuskens, J., Kollack-Walker, S., Rotelli, M., Houston, J.P., 2009. Defining “good” and “poor” outcomes in patients with schizophrenia or schizoaffective disorder: a multidimensional data-driven approach. Psychiatry Res. 170, 161–167. Marques, T.R., Arenovich, T., Agid, O., Sajeev, G., Muthen, B., Chen, L., Kinon, B.J., Kapur, S., 2010. The different trajectories of antipsychotic response: antipsychotics versus placebo. Psychol. Med. 41 (7), 1481–1488. Marshall, M., Lewis, S., Lockwood, A., Drake, R., Jones, P., Croudace, T., 2005. Association between duration of untreated psychosis and outcome in cohorts of first-episode patients: a systematic review. Arch. Gen. Psychiatry 62, 975–983. Muthen, B., Brown, C.H., Masyn, K., Jo, B., Khoo, S.T., Yang, C.C., Wang, C.P., Kellam, S.G., Carlin, J.B., Liao, J., 2002. General growth mixture modeling for randomized preventive interventions. Biostatistics 3, 459–475. Obermeier, M., Mayr, A., Schennach-Wolff, R., Seemuller, F., Moller, H.J., Riedel, M., 2010. Should the PANSS be rescaled? Schizophr. Bull. 36 (3), 455–460. Phillips, L., 1953. Case history data and prognosis in schizophrenia. J. Nerv. Ment. Dis. 515–525. R Development Core Team, 2008. A Language and Environment for Statistical Computing. Foundation for Statistical Computing. Ref Type: Computer Program, Vienna, Austria. URL http/:www.R-project.org. Schennach-Wolff, R., Jager, M., Seemuller, F., Obermeier, M., Messer, T., Laux, G., Pfeiffer, H., Naber, D., Schmidt, L.G., Gaebel, W., Huff, W., Heuser, I., Maier, W., Lemke, M.R., Ruther, E., Buchkremer, G., Gastpar, M., Moller, H.J., Riedel, M., 2009. Defining and predicting functional outcome in schizophrenia and schizophrenia spectrum disorders. Schizophr Res. Stauffer, V., Case, M., Kollack-Walker, S., scher-Svanum, H., Ball, T., Kapur, S., Kinon, B.J., 2011. Trajectories of response to treatment with atypical antipsychotic medication in patients with schizophrenia pooled from 6 double-blind, randomized clinical trials. Schizophr. Res. 130 (1–3), 11–19. Uher, R., Muthen, B., Souery, D., Mors, O., Jaracz, J., Placentino, A., Petrovic, A., Zobel, A., Henigsberg, N., Rietschel, M., Aitchison, K.J., Farmer, A., McGuffin, P., 2010. Trajectories of change in depression severity during treatment with antidepressants. Psychol. Med. 40, 1367–1377. Wiersma, D., Nienhuis, F.J., Slooff, C.J., Giel, R., 1998. Natural course of schizophrenic disorders: a 15-year followup of a Dutch incidence cohort. Schizophr. Bull. 24, 75–85. Wolwer, W., Buchkremer, G., Hafner, H., Klosterkotter, J., Maier, W., Moller, H.J., Gaebel, W., 2003. German research network on schizophrenia-bridging the gap between research and care. Eur. Arch. Psychiatry Clin. Neurosci. 253, 321–329. Zisook, S., McAdams, L.A., Kuck, J., Harris, M.J., Bailey, A., Patterson, T.L., Judd, L.L., Jeste, D.V., 1999. Depressive symptoms in schizophrenia. Am. J. Psychiatry 156, 1736–1743.
Contents lists available at SciVerse ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
Response trajectories in “real-world” naturalistically treated schizophrenia patients Rebecca Schennach a,⁎, Sebastian Meyer a, Florian Seemüller a, Markus Jäger a, Max Schmauss b, Gerd Laux c, Herbert Pfeiffer d, Dieter Naber e, Lutz G. Schmidt f, Wolfgang Gaebel g, Joachim Klosterkötter h, Isabella Heuser i, Wolfgang Maier j, Matthias R. Lemke k, Eckart Rüther l, Stefan Klingberg m, Markus Gastpar n, Richard Musil a, Hans-Jürgen Möller a, Michael Riedel a, o a
Department of Psychiatry and Psychotherapy, Ludwig-Maximilians-University, Munich, Germany Psychiatric Clinic, District Hospital Augsburg, Germany c Psychiatric Clinic, Inn-Salzach Hospital Wasserburg/Inn, Germany d Psychiatric Clinic, Isar-Amper Hospital, Munich-Haar, Germany e Department of Psychiatry and Psychotherapy, University of Hamburg, Germany f Department of Psychiatry and Psychotherapy, University of Mainz, Germany g Department of Psychiatry and Psychotherapy, Medical Faculty, Heinrich-Heine-University Duesseldorf, Germany h Department of Psychiatry and Psychotherapy, University of Cologne, Germany i Department of Psychiatry and Psychotherapy, Charite Berlin, Campus Benjamin Franklin, Germany j Department of Psychiatry and Psychotherapy, University of Bonn, Germany k Department of Psychiatry, Alsterdorf Hospital, Hamburg, Germany l Department of Psychiatry and Psychotherapy, University of Göttingen, Germany m Department of Psychiatry and Psychotherapy, University of Tübingen, Germany n Department of Psychiatry and Psychotherapy, University of Essen, Germany o Psychiatric Clinic, Vinzenz-von-Paul-Hospital, Rottweil, Germany b
a r t i c l e
i n f o
Article history: Received 21 November 2011 Received in revised form 13 April 2012 Accepted 7 May 2012 Available online 2 June 2012 Keywords: Schizophrenia Response trajectory Heterogeneity Group membership
a b s t r a c t Background: To date, research has identified distinct antipsychotic response trajectories yet focussing on data from randomized-controlled trials (RCTs). Therefore, the heterogeneity of response in “real-world” schizophrenia patients is still unknown. Methods: Antipsychotic response was evaluated in 399 patients suffering from a schizophrenia spectrum disorder within a naturalistic multicenter study of the Competence Network on Schizophrenia using latent class regression. Baseline and illness-related variables were compared between the different trajectory classes as well as currently proposed outcome definitions (early improvement, response, remission) using univariate tests. In order to predict the trajectory group membership classification and regression tree analysis were furthermore performed. Results: Five distinct trajectories of antipsychotic response were identified: Class 1 (15%) showing an early and considerable improvement, Class 2 (14%) incorporating patients with the greatest response to treatment, Class 3 (34%) again showing an early improvement to treatment yet with a slightly lower degree of improvement, Class 4 (22%) featuring patients gradually responding to treatment, and Class 5 (15%) with the poorest antipsychotic response. Fewer depressive symptoms at admission, better functioning, a shorter duration of illness and less previous hospitalizations were found to be significant predictors of good response. No considerable differences were found comparing the present results to the previous trajectory analyses deriving from RCTs. Conclusion: Our results underline the heterogeneous course of response independent of the study or treatment design suggesting that the diversity in schizophrenia response and outcome is determined primarily by different pathophysiological underpinnings. © 2012 Elsevier B.V. All rights reserved.
1. Introduction
⁎ Corresponding author at: Department of Psychiatry and Psychotherapy, LudwigMaximilians-University Munich, Nussbaumstrasse 7, 80336 München, Germany. Tel.: + 49 89 5160 5511; fax: + 49 89 5160 5728. E-mail address: [email protected] (R. Schennach). 0920-9964/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.schres.2012.05.004
In modern psychiatry, schizophrenia is believed to be a severe and heterogeneous disorder resulting in a favorable outcome in some patients although others may suffer from a deteriorating course of the illness (Wiersma et al., 1998). Since the introduction of antipsychotic treatment clinicians and researchers have tried to understand
R. Schennach et al. / Schizophrenia Research 139 (2012) 218–224
patterns of treatment response in schizophrenia and to identify and support patients in danger of a poor response in order to adopt specific treatment adaptations improving course and outcome of the illness (Gaebel, 1996). However, despite significant research in this field factors determining response to antipsychotic treatment are still not fully understood with inconclusive and contradictory results (Correll et al., 2003). But, understanding the characteristics of patients not responding well to treatment would be the first step to find better treatment strategies for them. To identify patients responding well or poorly to antipsychotic treatment, traditionally predefined thresholds are applied to categorize patients into responders on one side and non-responders on the other (Stauffer et al., 2011). This dichotomizing approach however bears numerous problems. Often the applied cut-offs are arbitrarily chosen and dichotomizing a continuous score was found to lead to inefficient analyses (Uher et al., 2010). Therefore, in order to shed light on actions of antipsychotic response in schizophrenia a new statistical method has been introduced lately to quantify the extent of heterogeneity in trajectories of response using so called mixed mode latent regression modeling. This approach bears the advantage of categorizing patients based on temporal patterns of change searching for heterogeneity as it naturally occurs in clinical data (Muthen et al., 2002). Levine and Leucht were among the first to perform growth mixture modeling in schizophrenia patients and reported five distinct treatment response trajectories characterized by varied amelioration levels (Levine and Leucht, 2010). Three trajectory classes showed a treatment response trend of amelioration, one class only a small reduction on the Brief Psychotic Rating Scale and one class featured a considerable symptom reduction during the first two weeks (Levine and Leucht, 2010). Also, Case et al. calculated response trajectories finding four distinct response trajectories in contrast to Levine and Leucht, however, identifying very similar response patterns with most patients achieving a moderate improvement with rapid symptom improvement in 12.5% of the patients and 2.3% with unsustained improvement (Case et al., 2010). The authors suggested that the observed heterogeneity might represent specific endophenotypes of response with different etiologic underpinnings (Case et al., 2010). In the meanwhile, several authors calculated treatment response trajectories in schizophrenia patients, however, all previously published articles focus solely on data deriving from randomized controlled trials examining different atypical antipsychotics (Levine and Rabinowitz, 2010; Marques et al., 2010; Levine et al., 2011; Stauffer et al., 2011). This limits the clinically relevant implications concurrently leaving out results of treatment with typical compounds or combination treatments. Besides, in terms of predicting the trajectory group membership only few variables have been examined so far regarding their predictive validity (Levine and Rabinowitz, 2010). Also, it is unclear how far the response trajectories overlap with currently established response and outcome definitions which might help to evaluate their clinical adequacy and use. Therefore, to expand prior research, we wanted to perform growth mixture modeling in schizophrenia patients treated within a naturalistic trial analyzing a variety of sociodemographic and clinical variables in terms of their value to predict the patient's response trajectory class. 2. Methods 2.1. Subjects Data were collected at eleven psychiatric university hospitals and three psychiatric district hospitals within a multicenter follow-up program by the German Research Network on Schizophrenia (Wolwer et al., 2003). All patients admitted to one of the above mentioned hospitals between January 2001 and December 2004 with a diagnosis of schizophrenia, schizophreniform disorder, delusional
219
disorder and schizoaffective disorder according to DSM-IV criteria were selected for inclusion. Subjects were aged between 18 and 65 years. Exclusion criteria were the presence of a head injury, a history of major medical illness and alcohol or drug dependency. An informed written consent had to be provided to participate in the study. The study protocol was approved by the local ethics committees (Jager et al., 2007). 2.2. Assessments DSM-IV diagnoses were established by clinical researchers on the basis of the German version of the Structured Clinical Interview for DSM-IV (American Psychiatric Association, 1994). Sociodemographic and course-related variables were collected using a standardized documentation system (Cording, 1998). To assess symptom severity and the level of antipsychotic response the Positive and Negative Syndrome Scale for Schizophrenia (PANSS) (Kay et al., 1988) was applied. To compare the patients within the different response trajectories further rating scales were used. The 17-item version of the Hamilton Depression Rating Scale (Hamilton, 1960) and the Udvalg for Kliniske Undersogelser (UKU) Side Effect Rating Scale (Lingjaerde et al., 1987) were applied to evaluate the level of depressive symptoms and adverse events, respectively. The patient's functioning was examined via the Global Assessment of Functioning Scale (GAF) (American Psychiatric Association, 1994) and the Social Occupational Functioning Scale (SOFAS) (American Psychiatric Association, 1994). To evaluate the patients' premorbid adjustment the short-scale for premorbid social-personal adjustment of the Phillips Scale was applied (Phillips, 1953). Ratings were assessed by trained clinicians at baseline and subsequently every two weeks until discharge. All raters had been trained using the applied scales. A high inter-rater reliability was achieved (ANOVA-ICC > 0.8 based on the PANSS). In order to establish and maintain the high interrater reliability interactive video-based rater-training sessions were regularly performed throughout the study period in every participating hospital. To further assure an accurate and consistent documentation the collected data were sent to the study center on a regular basis to be checked for completeness and coherence. Rater seminars between the centers were held before the study began. 2.3. Statistical analysis In a first step, different courses of treatment response were identified using latent class regression, also known as latent class growth analysis (LCGA). To model the course of treatment response, 49 different parameterizations with orthogonal polynomials (1–7°), semiparametric B-splines (1–7°, 2–9 df), or natural cubic splines (3–9 df) were investigated. Trading off between goodness-of-fit and complexity, the model minimizing the Bayesian Information Criterion (BIC) was chosen. Models were estimated by the classification–expectation– maximization (CEM) algorithm. However, in order to obtain regular maximum likelihood estimates, the selected model was finally reestimated by the original CEM algorithm using the maximum a posteriori (MAP) classification of the best CEM fit as initial configuration. The identified clusters resulting from the MAP classification were numbered according to the percentage reduction in the PANSS total score in each class. In a second step, the identified subgroups of patients were compared on various clinical and baseline characteristics using Fisher's exact test for categorical variables, and the Kruskal–Wallis test or the ANOVA F-test for metric variables, depending on the assumption of normality. The selection of these clinical and baseline characteristics goes back to previous publications of response trajectories deriving from randomized controlled trials in order to compare whether there would be a difference in the potential influencing variables
220
R. Schennach et al. / Schizophrenia Research 139 (2012) 218–224
depending on the study design (naturalistic versus randomized controlled study design). In a third step, classification and regression tree (CART) analysis was performed using a series of if-then rules (Breiman et al., 1984; Clark and Pregibon, 1992). Early improvement was defined as a 20% PANSS total score reduction from admission to week 2 (Kinon et al., 2010) and response as a 50% PANSS total score reduction from admission to discharge (Leucht et al., 2007). We followed the recommendations of Obermeier et al. concerning the calculation of percent changes in the PANSS (Obermeier et al., 2009). Remission was defined according to the symptom-severity component of the consensus remission criteria proposed by the Remission in Schizophrenia Working Group (Andreasen et al., 2005). All analyses were performed using the statistical software environment R 2.13.0 (R Development Core Team, 2008) and the contributed package flexmix (Leisch, 2004). 3. Results 3.1. Patients In the entire multicenter study 474 patients were enrolled. 46 patients had to drop out for different reasons, another 29 patients were excluded from analysis: 28 patients because they were discharged from the hospital within 7 days after admission and 1 patient due to incomplete psychopathology ratings. Therefore, the available sample comprised 399 subjects (225 males/174 females). The mean age was 35.4 years (±11.1), the mean duration of illness was 7.1 years (±12) and the mean duration of the current hospitalization was 64.7 days (±46). 31% of these patients suffered from their first illness episode. At admission, 20% of the patients suffered from suicidality. During the study patients were treated under naturalistic conditions as follows: 81% of the patients received first-generation antipsychotics, 80% of patients had second-generation antipsychotic treatment and 65% of the patients were treated with first as well as second-generation antipsychotics. Tranquilizers were administered in 79% of the patients and mood stabilizers in 16%. 27% of the patients were also treated with antidepressants. 3.2. Assessments 3.2.1. Response trajectory classes (Fig. 1)
• Trajectory Class 1 — (=early and considerable response) This class comprised 61 patients (15%) with an average PANSS baseline score of 45.8 points indicating that these patients suffered from only mild symptoms already at admission. Despite the low PANSS total score at admission the relative change of the PANSS score during inpatient treatment was 63.9%. 80% of the patients were early treatment improvers, 77% were responders and as many as 95% of the patients in Class 1 achieved the symptom severity component of the remission criteria at discharge (see Fig. 2). • Trajectory Class 2 — (=rapid and dramatic response) 54 patients (14%) were found to belong to response Class 2 with an initially high PANSS total score at admission and a dramatic improvement during treatment (62% PANSS total score change). This class is characterized by a very early and dramatic response with considerable psychopathological improvement in the first treatment weeks yet with some patients showing an unsustained response pattern (Fig. 2). • Trajectory Class 3 — (=early and satisfying response) Class 3 comprised 137 patients (34%) with an average PANSS baseline score of 64.1 points. The relative change of the PANSS score during inpatient treatment was 57.7%. Patients in this class showed an early and considerable response to treatment with an
Fig. 1. Response trajectories of naturalistically treated schizophrenia inpatients.
unsustained course of symptoms after the initial response to treatment in some patients of this class (Fig. 2). • Trajectory Class 4 — (=gradual response) Class 4 comprised 89 patients (22%) with an initially moderate PANSS total score at admission (72.6 points) and a relative change of the PANSS total score during treatment of 31%. Response to treatment occurred very gradually in this class. Fig. 2 shows the course of response of Class 4 indicating that possibly the degree of response might have been more satisfying if the study's observational time would have been longer for the response pattern in fact is gradual, but continuous. • Trajectory Class 5 — (=partial response) Class 5 comprises 58 patients (15%) with the worst response to treatment. Patients belonging to this class were rather symptomatic at admission (92.8 points) with only a relative change of the PANSS total score of 26.3%. Of the 58 patients 24% were early improvers with a similar number of patients being in response at discharge. The greatest variability of response and courses of response was found in this response class (see Fig. 2). 3.2.2. Baseline differences and outcome comparing the different trajectory groups Several significant differences in terms of sociodemographic and illness related variables were found between the five response classes (see Table 1). Patients with the greatest response to treatment were the ones with less severe positive and negative symptoms and better functioning at admission and with less illness chronicity (shorter duration of illness, fewer illness episodes and less numbers of previous hospitalizations). The patients with the earliest onset of their illness were found to be mainly in Class 5 with the poorest response to treatment. No significant differences were found comparing the different trajectory groups in terms of age, gender and the diagnostic subtype. The kind of antipsychotic treatment (atypical/typical/atypical plus typical) was compared between the different response classes yet without finding a significant difference between them (p = 0.87) (see Table 1). 3.2.3. Predicting the trajectory class membership Using a classification and regression tree analysis we wanted to identify the variables that would best separate the different response classes. Less depressive symptoms at admission, a higher level of functioning and less illness chronicity (fewer number of previous
R. Schennach et al. / Schizophrenia Research 139 (2012) 218–224
221
Fig. 2. Trajectories of each individual patient for the identified response trajectory classes.
hospitalizations/shorter duration of illness) were found to be significant predictors of a favorable trajectory class membership (Classes 1–3) (see Fig. 3).
4. Discussion 4.1. Response patterns within a “real-world” naturalistic treatment of schizophrenia patients In naturalistically treated schizophrenia patients five different response trajectories could be identified ranging from patients with an early and clinically considerable improvement to patients with only poor improvement underlining the heterogeneity of antipsychotic response. Response was understood to be an improvement of the patients' general psychopathological symptoms and did not focus on positive symptoms. Today response to treatment is believed to be heterogeneous with a variety of different courses of the illness with pathobiological and pathophysiological differences responsible for the differing patterns (Lieberman et al., 1993). This diversity in response and outcome in schizophrenia patients is also underlined by the results of cluster analyses stating that there is no “good” versus “bad” outcome dichotomy, but that outcome in schizophrenia is multidimensional (Lipkovich et al., 2009). Lipkovich et al. for example reported five different groups when analyzing the variety of outcome in 1449 schizophrenia patients treated up to 30 weeks ranging from patients with non‐existent to minimal symptoms and mild functional impairments (25.6%) to patients with moderate to severe psychiatric symptoms and severe functional deficits (14.8%) (Lipkovich et al., 2009). Also, long-term results (36 months of follow-up) emphasize the multifaceted course of schizophrenia and the wide range of outcomes (good/intermediate/poor) (Di and Bolino, 2004). Interestingly, our results found a rather early improvement in almost all patients (except those in Class 5) which indicate that despite
the diversity in treatment response there are also similarities in the response patterns of schizophrenia patients. This underlines the early-onset hypothesis proposed by Agid et al. stating that generally greater improvement occurs in the first 2 weeks of treatment than in the subsequent 2 treatment weeks (Agid et al., 2003). To our knowledge, this is one of the first studies explicitly evaluating the association between the response trajectory and different predefined outcome criteria. As one would expect we found patients of the very favorable response classes (Classes 1–3) to easily achieve the recommended criteria of early improvement at week 2 and response at discharge which suggests that these criteria adequately mirror the natural response to antipsychotic treatment. We found more patients achieving the symptom severity component of the remission criteria compared to the response criterion in Classes 1 and 3. It has been reported before that an at least 50% PANSS reduction is more stringent than the remission criteria in a short-term analysis (Beitinger et al., 2008). Given that response and remission provide different information on the patient's condition they should always both be implemented in schizophrenia trials as recommended by Leucht and Kane (2006).
4.2. Comparing response classes of naturalistically treated patients with previously performed trajectory studies in randomized-controlled trials Comparing the present results to previously performed response trajectory analyses exclusively based on data deriving from randomized-controlled trials (RCTs) we did not find any considerable differences in terms of the number and the course of the trajectory classes (Case et al., 2010; Levine and Leucht, 2010; Levine and Rabinowitz, 2010; Levine et al., 2010; Marques et al., 2010; Stauffer et al., 2011). This is interesting, for one could have hypothesized that due to the strict inclusion and exclusion criteria of randomizedcontrolled trials more severely ill, yet less chronically ill patients
222
R. Schennach et al. / Schizophrenia Research 139 (2012) 218–224
Table 1 Baseline and course-related variables, psychopathology and functioning comparing the different response trajectories. Class 1 (N = 15%)
Class 2 (N = 14%)
Class 3 (N = 34%)
Class 4 (N = 22%)
Class 5 (N = 15%)
46%
67%
51%
65%
57%
0.052
75% 11% 14% 36% 88% 66%
92% 6% 2% 31% 72% 47%
82% 11% 7% 40% 69% 55%
82% 16% 2% 17% 74% 69%
86% 12% 2% 22% 57% 64%
0.081
75% 7% 11% 7% 15%
77% 5% 10% 8% 19%
72% 9% 9% 10% 14%
69% 6% 21% 4% 27%
79% 5% 9% 7% 28%
Mean (SD) Age (years) Age at onset (years)
36.2 (± 10.24) 30.6 (± 10.34)
35.2 (± 10.96) 29 (± 8.29)
36.1 (± 11.25) 29.5 (± 10.07)
34.2 (± 11.62) 23.9 (± 7.20)
34.9 (± 11.10) 25.1 (± 8.47)
0.71 b 0.0001
Illness related variables [median (IQR)] Duration of illness (years) Number of previous hospitalizations Duration of current hospitalization (days) Premorbid adjustment
2 (± 7) 1 (± 2) 35 (± 32) 2 (± 3)
2 (± 11) 0 (± 2) 63 (± 42) 2 (± 2)
2 (± 10) 0 (± 3) 51 (± 37) 2 (± 2)
4 (± 16) 2.5 (± 5) 64 (± 52) 3 (± 2)
7 (± 16) 1 (± 4.75) 77 (± 78.5) 3 (± 1.50)
0.0045 0.0011 b 0.0001 0.0038
Psychopathology at admission PANSS total score PANSS positive subscore PANSS negative subscore PANSS general psychopathology Insight into illness (PANSS G12 item) HAMD
45.8 (± 6.2) 13.6 (± 3.9) 9.4 (± 2.4) 22.8 (± 3.3) 6.8 (± 5.61)
92.4 24.2 23.4 44.8 17.6
64.1 (± 10.3) 18.5 (± 5.9) 14.6 (± 5.2) 31 (± 6.1) 11.9 (± 8.18)
72.6 18.3 19.4 34.9 14.2
92.8 22.5 24.7 45.7 21.3
b 0.0001a b 0.0001a b 0.0001a b 0.0001a b 0.0001
Treatment tolerability (UKU)
3.7 (± 4.91)
4.8 (± 7.37)
4.5 (± 6.00)
6.3 (± 6.27)
6.9 (± 6.99)
Functioning at admission GAF SOFAS
51.6 (± 12.44) 43.2 (± 11.46)
40.7 (± 11.24) 42.1 (± 11.24)
42.5 (± 11.22) 52.3 (± 10.69)
42.1 (± 10.53) 43.1 (± 10.71)
35.3 (± 10.01) 41.1 (± 10.57)
Antipsychotic treatment during hospitalization Atypical antipsychotics Typical antipsychotics Atypical and typical antipsychotics
18% 24% 58%
13% 13% 74%
17% 18% 65%
15% 17.5% 67.5%
13% 15% 72%
Sociodemographic variables [%] Gender (male) Diagnostic subtype Schizophrenia Schizoaffective Brief psychotic disorder First illness episode Duration of current episode b 6 months Antipsychotic premedication Reason for admission Psychopathological worsening Suicidality Acute psychosocial crisis Other Suicidality at admission
a
(± 14.5) (± 5.7) (± 5.5) (± 9.1) (± 8.29)
(± 8.5) (± 5.6) (± 6.1) (± 5.6) (± 7.69)
p-value
0.002 0.0068 0.047 0.65
0.058
(± 15.5) (± 5.8) (± 6.3) (± 9) (± 8.69)
0.02
b 0.0001 b 0.0001
0.87
The PANSS was used to group the trajectory classes so that regular test conditions to compare these items between the classes are not fulfilled.
might have been included in those studies resulting in different response patterns. However, in line with the study by Stauffer et al., Levine and Leucht and Levine and Rabinowitz we found five different
response trajectories enclosing dramatic responders, partial and unsustained responders and delayed and poor responders (Levine and Leucht, 2010; Levine and Rabinowitz, 2010; Stauffer et al., 2011).
Fig. 3. Predicting the response trajectory class.
R. Schennach et al. / Schizophrenia Research 139 (2012) 218–224
Other authors reported of only four response classes, however, the course and description of the trajectory classes remain the same (Case et al., 2010). Interestingly, all published trajectory analyses match especially in finding one response trajectory class with a rapid and significant response to treatment, often representing around 10–20% of the patient sample (Case et al., 2010; Levine and Leucht, 2010; Levine and Rabinowitz, 2010; Levine et al., 2010; Marques et al., 2010; Stauffer et al., 2011). In their analysis of 628 patients with schizophrenia and schizoaffective disorder Case et al. found the rapid response class to be uniquely represented by 100% early improvers (≥20% PANSS total score improvement within the first two treatment weeks) (Case et al., 2010). Marques et al. examined different trajectories of antipsychotic response in comparison to placebo and reported that in their subset of patients with rapid response all patients were from the drug-treated group showing more than a 70% improvement (Marques et al., 2010). In the present study patients with dramatic response showed a relative change of 62% in the PANSS total score which lies in between the results of the comparative studies. 4.3. Predicting the response trajectory group membership We were able to identify several clinical significant predictors of the membership of a favorable response trajectory, namely less baseline depressive symptoms, better baseline functioning, a shorter duration of illness and less previous hospitalizations in the psychiatric history. Depressive symptoms are believed to be a major problem for illness course and outcome in schizophrenia patients (Addington et al., 1998) and have been associated with poor quality of life and an increased risk of relapse (Zisook et al., 1999). Also, impairments in functioning evaluated via two different rating scales (GAF and SOFAS) were found to significantly differ between the response trajectories. Generally, the patient's functioning is linked to the severity of psychopathological symptoms and the course of outcome (Schennach-Wolff et al., 2009). However, it should be kept in mind that not only the level of functioning influences the course of symptoms, but that in turn the degree of symptoms also influences the level of functioning so that both of these domains somehow predict each other. Variables like a longer duration of illness or a greater number of hospitalizations generally stand for a less favorable course of schizophrenia and have consistently been associated with a lower chance to respond to treatment as well as to achieve remission and recovery (Marshall et al., 2005; Haro et al., 2008). In terms of the above discussed comparative trajectory analyses only few authors have performed prediction models in order to examine the membership in the best or worst treatment response trajectories. Levine and Rabinowitz calculated binary logistic regression models to examine the class membership and found a better premorbid functioning, better cognitive functioning and not having a diagnosis of schizophrenia to be predictive of belonging to the best response trajectory (Levine and Rabinowitz, 2010). However, it is unclear what potential influencing variables were examined in their study which limits the significance of their results. In a different analysis on 236 recentepisode psychosis patients Levine et al. reported that the membership with the least improvement was significantly predicted by higher PANSS baseline scores and again membership in the trajectory with the most improvement was predicted by better premorbid functioning, the absence of a schizophrenia diagnosis and a higher cognitive score (Levine et al., 2010). 4.4. Strengths and limitations The major strength of this study is that it was performed within a naturalistic study design mirroring the “real-world” treatment conditions of schizophrenia patients allowing reliable clinical conclusions. Also, due to the liberal inclusion and exclusion criteria, findings of
223
this study on treatment seeking patients might be more generalizable and exhibit higher external validity than results from RCTs. However, such a design does not allow sufficient control of study results for the effect of different pharmacological treatments which limits specific association between the treatment applied and the response pattern. Besides, given that specific health system aspects influence the practice of hospitalizing and treating mentally ill patients, our study results might not be generalizable and comparable to others, especially those of non-European countries. 4.5. Conclusion Response trajectory analysis of naturalistically treated schizophrenia patients identified five different response trajectories. Only a smaller proportion of patients suffered from a poor response with a very similar number of patients achieving rapid and considerable improvement. Generally, a more favorable response to treatment was associated with and predicted by a shorter duration and later onset of illness, less baseline illness severity and better functioning. Response trajectory analysis allows the evaluation of different response patterns without using predefined cut-offs or time-points emphasizing the heterogeneity of response in schizophrenia patients. Role of funding source The study was performed within the framework of the German Research Network on Schizophrenia, which is funded by the German Federal Ministry for Education and Research BMBF (grant 01 GI 0233). Contributors The German Research Network on Schizophrenia designed the study and wrote the protocol. All authors have contributed to and have approved the final manuscript. Authors are as follows: Rebecca Schennach, Sebastian Meyer, Florian Seemüller, Markus Jäger, Richard Musil, Hans-Jürgen Möller, Michael Riedel from the Department of Psychiatry and Psychotherapy, Ludwig-Maximilians-University, Munich, Germany; Max Schmauss from the Psychiatric Clinic, District Hospital Augsburg, Germany; Gerd Laux from the Psychiatric Clinic, Inn-Salzach Hospital Wasserburg/Inn, Germany; Herbert Pfeiffer from the Psychiatric Clinic, Isar-Amper Hospital, Munich-Haar, Germany; Dieter Naber from the Department of Psychiatry and Psychotherapy, University of Hamburg, Germany; Lutz G. Schmidt from the Department of Psychiatry and Psychotherapy, University of Mainz, Germany; Wolfgang Gaebel from the Department of Psychiatry and Psychotherapy, Heinrich-Heine-University Duesseldorf, Germany; Joachim Klosterkötter from the Department of Psychiatry and Psychotherapy, University of Cologne, Germany; Isabella Heuser from the Department of Psychiatry and Psychotherapy, Charite Berlin, Campus Benjamin Franklin, Germany; Wolfgang Maier from the Department of Psychiatry and Psychotherapy, University of Bonn, Germany; Matthias R. Lemke from the Department of Psychiatry, Alsterdorf Hospital, Hamburg, Germany; Eckart Rüther from the Department of Psychiatry and Psychotherapy, University of Göttingen, Germany; Stefan Klingberg from the Department of Psychiatry and Psychotherapy, University of Tübingen, Germany and Markus Gastpar from the Department of Psychiatry and Psychotherapy, University of Essen, Germany. Conflict of interest All authors declare that they have no conflicts of interest within the context of this article. Acknowledgments The study was performed within the framework of the German Research network on Schizophrenia, which is funded by the German Federal Ministry for Education and Research BMBF (grant 01 GI 0233).
References Addington, D., Addington, J., Patten, S., 1998. Depression in people with first-episode schizophrenia. Br. J. Psychiatry Suppl. 172, 90–92. Agid, O., Kapur, S., Arenovich, T., Zipursky, R.B., 2003. Delayed-onset hypothesis of antipsychotic action: a hypothesis tested and rejected. Arch. Gen. Psychiatry 60, 1228–1235. American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Mental Disorders, Fourth edition. American Psychiatric Association, Washington DC. Andreasen, N.C., Carpenter Jr., W.T., Kane, J.M., Lasser, R.A., Marder, S.R., Weinberger, D.R., 2005. Remission in schizophrenia: proposed criteria and rationale for consensus. Am. J. Psychiatry 162, 441–449. Beitinger, R., Lin, J., Kissling, W., Leucht, S., 2008. Comparative remission rates of schizophrenic patients using various remission criteria. Prog. Neuropsychopharmacol. Biol. Psychiatry 32, 1643–1651.
224
R. Schennach et al. / Schizophrenia Research 139 (2012) 218–224
Breiman, L., Friedman, H.J., Olshen, R.A., Stone, C.J., 1984. Classification and Regression Trees. Wadsworth International Group, Belmont, CA. Case, M., Stauffer, V.L., scher-Svanum, H., Conley, R., Kapur, S., Kane, J.M., KollackWalker, S., Jacob, J., Kinon, B.J., 2010. The heterogeneity of antipsychotic response in the treatment of schizophrenia. Psychol. Med. 1–10. Clark, L.A., Pregibon, D., 1992. Tree-based models. In: Chambers, J., Hastie, T. (Eds.), Statistical Models. Wadsworth & Books/Cole Advanced Books & Software, Pacific Grove, CA. Cording, C., 1998. Conceptual aspects in development and implementation of basic psychiatric documentation. Psychiatr. Prax. 25, 175–178. Correll, C.U., Malhotra, A.K., Kaushik, S., McMeniman, M., Kane, J.M., 2003. Early prediction of antipsychotic response in schizophrenia. Am. J. Psychiatry 160, 2063–2065. Di, M.V., Bolino, F., 2004. The natural course of schizophrenia and psychopathological predictors of outcome. A community-based cohort study. Psychopathology 37, 98–104. Gaebel, W., 1996. Prediction of response to acute neuroleptic treatment in schizophrenia. Int. Clin. Psychopharmacol. 11 (Suppl. 2), 47–54. Hamilton, M., 1960. A rating scale for depression. J. Neurol. Neurosurg. Psychiatry 23, 56–62. Haro, J.M., Novick, D., Suarez, D., Ochoa, S., Roca, M., 2008. Predictors of the course of illness in outpatients with schizophrenia: a prospective three year study. Prog. Neuropsychopharmacol. Biol. Psychiatry 32, 1287–1292. Jager, M., Riedel, M., Messer, T., Laux, G., Pfeiffer, H., Naber, D., Schmidt, L.G., Gaebel, W., Huff, W., Heuser, I., Kuhn, K.U., Lemke, M.R., Ruther, E., Buchkremer, G., Gastpar, M., Bottlender, R., Strauss, A., Moller, H.J., 2007. Psychopathological characteristics and treatment response of first episode compared with multiple episode schizophrenic disorders. Eur. Arch. Psychiatry Clin. Neurosci. 257, 47–53. Kay, S.R., Opler, L.A., Lindenmayer, J.P., 1988. Reliability and validity of the positive and negative syndrome scale for schizophrenics. Psychiatry Res. 23, 99–110. Kinon, B.J., Chen, L., Ascher-Svanum, H., Stauffer, V.L., Kollack-Walker, S., Zhou, W., Kapur, S., Kane, J.M., 2010. Early response to antipsychotic drug therapy as a clinical marker of subsequent response in the treatment of schizophrenia. Neuropsychopharmacology 35, 581–590. Leisch, F., 2004. FlexMix: a general framework for finite mixture models and latent class regression in R. J. Stat. Softw. 11, 1–18. Leucht, S., Kane, J.M., 2006. Measurement-based psychiatry: definitions of response, remission, stability, and relapse in schizophrenia. J. Clin. Psychiatry 67, 1813–1814. Leucht, S., Davis, J.M., Engel, R.R., Kane, J.M., Wagenpfeil, S., 2007. Defining ‘response’ in antipsychotic drug trials: recommendations for the use of scale-derived cutoffs. Neuropsychopharmacology 32, 1903–1910. Levine, S.Z., Leucht, S., 2010. Elaboration on the early-onset hypothesis of antipsychotic drug action: treatment response trajectories. Biol. Psychiatry 68, 86–92. Levine, S.Z., Rabinowitz, J., 2010. Trajectories and antecedents of treatment response over time in early-episode psychosis. Schizophr. Bull. 36, 624–632. Levine, S.Z., Rabinowitz, J., Case, M., scher-Svanum, H., 2010. Treatment response trajectories and their antecedents in recent-onset psychosis: a 2-year prospective study. J. Clin. Psychopharmacol. 30, 446–449. Levine, S.Z., Lurie, I., Kohn, R., Levav, I., 2011. Trajectories of the course of schizophrenia: from progressive deterioration to amelioration over three decades. Schizophr. Res. 126, 184–191.
Lieberman, J., Jody, D., Geisler, S., Alvir, J., Loebel, A., Szymanski, S., Woerner, M., Borenstein, M., 1993. Time course and biologic correlates of treatment response in first-episode schizophrenia. Arch. Gen. Psychiatry 50, 369–376. Lingjaerde, O., Ahlfors, U.G., Bech, P., Dencker, S.J., Elgen, K., 1987. The UKU side effect rating scale. A new comprehensive rating scale for psychotropic drugs and a crosssectional study of side effects in neuroleptic-treated patients. Acta Psychiatr. Scand. Suppl. 334 (1–100), 1–100. Lipkovich, I.A., Deberdt, W., Csernansky, J.G., Buckley, P., Peuskens, J., Kollack-Walker, S., Rotelli, M., Houston, J.P., 2009. Defining “good” and “poor” outcomes in patients with schizophrenia or schizoaffective disorder: a multidimensional data-driven approach. Psychiatry Res. 170, 161–167. Marques, T.R., Arenovich, T., Agid, O., Sajeev, G., Muthen, B., Chen, L., Kinon, B.J., Kapur, S., 2010. The different trajectories of antipsychotic response: antipsychotics versus placebo. Psychol. Med. 41 (7), 1481–1488. Marshall, M., Lewis, S., Lockwood, A., Drake, R., Jones, P., Croudace, T., 2005. Association between duration of untreated psychosis and outcome in cohorts of first-episode patients: a systematic review. Arch. Gen. Psychiatry 62, 975–983. Muthen, B., Brown, C.H., Masyn, K., Jo, B., Khoo, S.T., Yang, C.C., Wang, C.P., Kellam, S.G., Carlin, J.B., Liao, J., 2002. General growth mixture modeling for randomized preventive interventions. Biostatistics 3, 459–475. Obermeier, M., Mayr, A., Schennach-Wolff, R., Seemuller, F., Moller, H.J., Riedel, M., 2010. Should the PANSS be rescaled? Schizophr. Bull. 36 (3), 455–460. Phillips, L., 1953. Case history data and prognosis in schizophrenia. J. Nerv. Ment. Dis. 515–525. R Development Core Team, 2008. A Language and Environment for Statistical Computing. Foundation for Statistical Computing. Ref Type: Computer Program, Vienna, Austria. URL http/:www.R-project.org. Schennach-Wolff, R., Jager, M., Seemuller, F., Obermeier, M., Messer, T., Laux, G., Pfeiffer, H., Naber, D., Schmidt, L.G., Gaebel, W., Huff, W., Heuser, I., Maier, W., Lemke, M.R., Ruther, E., Buchkremer, G., Gastpar, M., Moller, H.J., Riedel, M., 2009. Defining and predicting functional outcome in schizophrenia and schizophrenia spectrum disorders. Schizophr Res. Stauffer, V., Case, M., Kollack-Walker, S., scher-Svanum, H., Ball, T., Kapur, S., Kinon, B.J., 2011. Trajectories of response to treatment with atypical antipsychotic medication in patients with schizophrenia pooled from 6 double-blind, randomized clinical trials. Schizophr. Res. 130 (1–3), 11–19. Uher, R., Muthen, B., Souery, D., Mors, O., Jaracz, J., Placentino, A., Petrovic, A., Zobel, A., Henigsberg, N., Rietschel, M., Aitchison, K.J., Farmer, A., McGuffin, P., 2010. Trajectories of change in depression severity during treatment with antidepressants. Psychol. Med. 40, 1367–1377. Wiersma, D., Nienhuis, F.J., Slooff, C.J., Giel, R., 1998. Natural course of schizophrenic disorders: a 15-year followup of a Dutch incidence cohort. Schizophr. Bull. 24, 75–85. Wolwer, W., Buchkremer, G., Hafner, H., Klosterkotter, J., Maier, W., Moller, H.J., Gaebel, W., 2003. German research network on schizophrenia-bridging the gap between research and care. Eur. Arch. Psychiatry Clin. Neurosci. 253, 321–329. Zisook, S., McAdams, L.A., Kuck, J., Harris, M.J., Bailey, A., Patterson, T.L., Judd, L.L., Jeste, D.V., 1999. Depressive symptoms in schizophrenia. Am. J. Psychiatry 156, 1736–1743.