Schizophrenia is a severe lifetime illness. Since the introduction of antipsychotic agents in the 1950s, early prediction of treatment outcome has been an increasingly important issue. Clinical prediction models assist physicians in making early decisions because they are comprised of clinical variables that show strong predictive value for outcome. Early prediction for poor outcome to antipsychotics could prevent unnecessary persistence with ineffectual agents, diminish risk of adverse events, and reduce the financial burden in treating the illness. Thus, physicians could adjust management programs and timely provide treatment targets.1
In clinical trials for antipsychotic agents, response is a commonly used outcome variable, usually defined as a clinically meaningful percentage of reduction after treatment (eg, 20%, 30%, 40%, or 50%)2,3 in validated rating scales for schizophrenia, such as the Brief Psychiatric Rating Scale (BPRS)4 or the Positive and Negative Syndrome Scale (PANSS).5 However, a schizophrenic patient who responds with a 20%, 30%, 40%, or 50% reduction of BPRS or PANSS scores may still have significant symptoms, particularly if the patient is severely ill before treatment, with, for example, a baseline BPRS score of 90 or a baseline PANSS score of 150.
For schizophrenia, remission, albeit a challenging concept, is an achievable goal for a proportion of patients.6,7 The Remission in Schizophrenia Working Group has proposed operationally defined criteria for remission in schizophrenia.8 These criteria consist of 2 elements: a symptomatic criterion and a time criterion. Symptomatic remission is defined by using an absolute threshold of severity for selected rating scale items representing the “core symptoms” of schizophrenia, rather than the percentage of symptom reductions from baseline. A patient is in symptomatic remission if 7 items of the BPRS (4 items on psychoticism, 2 items on disorganization, and one item on negative symptoms) or 8 items of the PANSS (3 items on psychoticism, 2 items on disorganization, and 3 items on negative symptoms) are rated “3 or less”. These 7 items of the BPRS or 8 items of the PANSS have been named “remission items”. A time criterion requires that the symptomatic criterion must be sustained for at least 6 months. This definition of symptomatic remission affirms that remission is a stricter standard than response. In short, remission is not synonymous with response.
Several studies have shown that after using an antipsychotic agent, there is a significant symptom reduction in the first 2 weeks compared with subsequent treatment weeks.9–11 For example, a meta-analytic study of 42 published studies10 has demonstrated that a greater reduction of symptoms occurred during the first 2 weeks than during the rest of the treatment’s duration using the same antipsychotic agent. According to this “early onset hypothesis of antipsychotic action,”11 several studies12–17 have used a percentage of symptom reduction from the PANSS or BPRS baselines during the first 2 weeks (also known as “early improvement”) to predict both response and remission for schizophrenic patients undergoing short-term treatment. They found that the percentage of symptom reduction at week 2 can significantly predict subsequent response and remission. Therefore, the phenomenon of an early improvement to medication is a stable predictor of subsequent remission to medication for schizophrenic patients.
A response criterion is used to reflect percentage symptom reduction irrespective of whether the patient is still symptomatic. However, a symptomatic remission criterion is used to reflect whether the residual core symptoms are below a predefined level after treatment.18 We hypothesize that the absolute score of BPRS remission items (ie, the level of residual core symptoms) after the first 2 weeks is an alternative predictor of remission.
The first purpose of this study was to identify valid predictors for remission from schizophrenia by comparing the percentage of symptom reduction (ie, early improvement) and remission-items scores after 2 weeks of treatment. The second purpose was to maximize both sensitivity and specificity to determine the optimal cutoff point of predictors for eventual remission/nonremission at week 4 among schizophrenic inpatients taking zotepine, an atypical antipsychotic agent.
MATERIALS AND METHODS
This study was conducted in the inpatient unit of Kai-Syuan Psychiatric Hospital, a major psychiatric center in Kaohsiung, Taiwan, from June 2004 to April 2005. The study was approved by the facility’s institutional review board and conducted in accordance with the Declaration of Helsinki.
The study design has been described previously.19 All newly hospitalized schizophrenic patients with acute exacerbation were screened and evaluated by research psychiatrists. The Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV)20 was used in making diagnoses. Han Chinese patients in Taiwan entered into this study if they (1) were physically healthy and had all laboratory parameters within normal limits, (2) were between 18 and 65 years old, (3) satisfied DSM-IV criteria for schizophrenia, (4) were rated as moderate or worse on at least 1 of the 4 BPRS psychotic symptom items (ie, hallucinations, unusual thoughts, conceptual disorganization, or suspiciousness), (5) had no DSM-IV diagnosis of substance abuse or dependence, including alcohol, (6) had not received depot antipsychotics for the preceding 3 months, and (8) gave written informed consent after the procedures had been fully explained. Patients were excluded from the study if they had a history of severe adverse reaction to antipsychotics or if they had had a diagnosis of treatment-resistant schizophrenia.21
After a washout period of at least 72 hours, the patients received open-labeled zotepine treatment at a fixed dosage of 150 mg daily for 4 weeks. Benzodiazepine was allowed as needed for insomnia or agitation as was trihexyphenidyl for extrapyramidal adverse effects. No other psychotropic agents were used.
Symptom severity was assessed at baseline and again at weeks 1, 2, 3, and 4 by trained and experienced psychiatrists using the 18-item BPRS, with scores ranging from 1 (symptoms not present) to 7 (extremely severe symptoms). The BPRS score ranges from 18 to 126. The BPRS contains 3 clusters—positive, negative, and general symptoms—according to the pivotal clozapine trial.21 The intraclass correlation coefficient of reliability was 0.96 between the raters.
The research psychiatrists who conducted the clinical ratings did not know details of the study’s design. Adverse effects were evaluated by the Udvalg for Kliniske Undersøgelser (UKU) Side Effect Rating Scale,22 with scores ranging from zero (none) to 3 (severe). A score of 1 (0 = no or doubtful symptoms) or higher on any UKU item indicate cases of adverse effects.23
Remission of symptoms was chosen as the outcome measure. Symptomatic remission was defined, according to the criteria proposed by the Remission in Schizophrenia Working Group,8 as a BPRS score of 3 or less on the following items: grandiosity (item 8), suspiciousness (item 11), unusual thought content (item 15), hallucinatory behavior (item 12), conceptual disorganization (item 4), mannerisms/posturing (item 7), and blunted affect (item 16).8 The time criterion of the consensus remission could not be applied owing to the 4-week treatment. Consequently, status of symptomatic remission was a dichotomous variable, operationally defined as remission or nonremission.
Initially, remitters and nonremitters at week 4 were compared with regard to demographic data (sex, age, and years of education), age at onset, baseline BPRS score, percentages of BPRS score reduction at week 2, percentage of cluster (ie, positive, negative, and general symptoms) score at week 2 and BPRS remission-items score at week 2. Age at onset of illness was regarded as the age at which the first psychotic symptoms appeared. To demonstrate the odds ratios (ORs) for the variables, we used univariate logistic regression to identify variables associated with remission.
Second, if multiple potential predictive variables were identified from the first step, backward stepwise logistic regression model was used to determine the best predictors for clinical response. Third, continuous variables proved to be significant predictors were entered in a receiver operating characteristic (ROC) curve to determine both the best model and best predictor. The ROC curve was used to determine the best cutoff point of predictor between the remitters and nonremitters, maximizing both the sensitivity and specificity of the predictor variable so that false-positive and false-negative rates could be minimized. The ROC curve is created by plotting 1 − specificity (ie, false-positive rate) against sensitivity (ie, true-positive rate) to distinguish remission from nonremission. The area under the ROC curve (AUC) is a measure of the overall discriminative power. In practice, an AUC generally falls somewhere between 0.50 and 1. An AUC value of 0.7 and 0.8 is considered reasonable, and a value greater than 0.8 is considered to have a good discriminative capacity.24,25 All tests were 2-tailed, and significance was defined as P < 0.05. All data were processed by SPSS version 17.0 for Windows (SPSS Inc, Chicago, Ill) and MedCalc (MedCalc Software, Belgium). The MedCalc software is a program that implements several statistical procedures, including ROC analysis.
A total of 135 acutely ill inpatients with schizophrenia were enrolled, with 100 (74.1%) of them completing the 4-week trial. Fifty-four (54.0%) of the completers were men and 46 (46.0%) were women. The mean (SD) age was 37.3 (9.1) years. The mean (SD) age at onset of illness was 23.5 (6.3) years. The remaining 35 patients did not complete the study: 13 patients were discharged from the hospital before week 4 owing to lack of cooperation, 10 patients received haloperidol IM or increased dosages of zotepine to control agitation, and 12 patients could not tolerate the adverse effects of dizziness (n = 7), sleepiness (n = 3), and asthenia (n = 2). Dropout patients (n = 35) and the completers (n = 100) were comparable for sex (17 men and 18 women vs 54 men and 46 women; P = 0.580), mean (SD) age (36.7 [9.2] vs 37.3 [9.1]; P = 0.740), mean (SD) age at onset of illness (24.9 [6.9] vs 23.5 [6.4]; P = 0.270), and total baseline BPRS score (56.6 [11.3] vs 55.7 [10.6]; t = 0.44; df = 133; P = 0.658).
Of the 100 completers, 21.0% (21) were classified as remitters, and 79.0% (79) were classified as nonremitters after the 4-week treatment. Remitters and nonremitters did not significantly differ in sex, age, age at onset, years of education, and baseline BPRS scores (Table 1). However, compared to nonremitters, remitters had significantly greater percentages of BPRS score reduction at week 2, greater percentages of BPRS positive symptoms score reduction at week 2, and lower BPRS remission-items scores at week 2 (Table 1). Because percentages of BPRS score at week 2, percentages of BPRS positive symptoms score reductions at week 2, and BPRS remission-items score at week 2 were potential predictors (Table 1), we then applied backward stepwise logistic regression model and found both percentages of BPRS score reductions at week 2 (OR, 1.05; 95% confidence interval, 1.00–1.11; P = 0.04) and BPRS remission-items score at week 2 (OR, 0.85; 95% confidence interval, 0.74–0.97; P = 0.01) to be the most influential predictors for remission (data not shown in the table). Remitters and nonremitters also had comparable rates of adverse effects at each selected UKU item (data not shown).
Receiver operating characteristic analysis was used to determine the cutoff point of valid predictors by plotting the proportion of true-positive results (sensitivity) versus the proportion of false-positive results (1 − specificity). The cutoff point, sensitivity, specificity, predictive power (the number of true positives plus the number of true negatives divided by the total number of patients), and the AUC of each valid predictor are shown in Table 2.
At week 2, a BPRS score reduction of 35% provided a sensitivity of 62%, a specificity of 84%, and an AUC of 0.81. Comparatively, a BPRS remission-items score of 18 at week 2 provided a sensitivity of 86%, a specificity of 65%, and an AUC of 0.82. The BPRS remission-items scores at week 2 predicted a remission at week 4 better than BPRS score reductions at week 2 because the AUC value was higher. At week 2, a BPRS remission-items score of 18 seemed to be the optimal cutoff point for predicting eventual symptomatic remission. Receiver operating characteristic curves of BPRS remission-items scores at week 2, together with the percentages of BPRS reduction at week 2, are presented in Figure 1.
This study showed that applying absolute remission-items scores at week 2 as the predictor for remission generated a marginally larger AUC than using the percentage of BPRS score reduction at week 2 (Table 2). An AUC represents discriminating power; therefore, using absolute scores of 7 remission items at week 2 as the predictor is slightly more accurate than using the percentage of score reduction of all 18 BPRS items.
The best cutoff point of predictor, determined by ROC curves at week 2, had a good discriminatory ability (ie, an AUC >0.8) in identifying remitters/nonremitters at the end of the 4-week treatment. At week 2, 86% of the patients who displayed a BPRS remission-items score of 18 or less were correctly classified as remitters at the end of the 4-week treatment. Conversely, 65% of the patients with a BPRS remission-items score greater than 18 ultimately met nonremission criteria. Because specificity of 65% was not very high, it denoted that false-positive rate (1 − specificity) was relatively high. Therefore, it is difficult to avoid unnecessarily continuing treatment with patients who would ultimately not remit (eg, the patients who were not in remission at week 4 actually achieved remission at week 2 but subsequently lost it). Because percentage of BPRS score reduction at week 2 was another predictor, a BPRS score reduction of 35% provided a sensitivity of 64% and specificity of 82%. That is, the false-negative rate was relatively high. If we balanced the sensitivities and specificities from these 2 potential predictors, patients with less than a 35% BPRS score reduction and a BPRS remission-items score larger than 18 during the first 2 weeks of treatment were unlikely to reach a final remission (Table 2). Physicians should determine whether a patient would be maintained on the initial drug or shifted to a new treatment.
The patients in the current study received the same fixed dosage, 150 mg daily, of zotepine treatment. This dosage, based on previous studies in western patients26 and in Taiwanese,27 is the optimal dose for treatment of schizophrenia. Certainly, it remains possible that a portion of patients require other doses of zotepine to reach symptomatic remission. However, schizophrenic patients taking 150 mg daily of zotepine reveal a striatal D2 receptor occupancy of 65.8% (SD, 6.2%),28 and brain imaging studies indicate that D2 receptor occupancy of 65% to 70% is correlated with maximal antipsychotic efficacy.29
Several strengths of this study should be addressed. First, the subjects were inpatients requiring short-term treatment. As inpatients, the subjects were carefully monitored for symptom assessment, development of adverse effects, and medical adherence. They also had similar environments. Second, the ROC curve was applied to determine the cutoff point for an early prediction of eventual remission, to obtain the highest sensitivity and specificity.
The limitations of this study also must be mentioned. First, it was a short-term open-labeled treatment design. It was also a single-center study using a single antipsychotic agent (ie, zotepine), and a single-center study. The ideal duration for an antipsychotic trial still remains controversial.15 However, studies11,30 have suggested that most patients with schizophrenia achieve the plateau of clinical improvement within 4 weeks of short-term treatment. For example, a meta-analysis of 1-year follow-up trials11 revealed total BPRS scores of 14.2 (12.4) at 4 weeks and 16.2 (5.8) after 1 year. The researchers concluded that a substantial proportion (88%) of the 1-year BPRS score reduction has already been reached within the first 4 weeks and that all antipsychotic agents would have a similar time course pattern. Another meta-analysis by McMahon et al31 found that trials as short as 4 weeks could have results similar to longer-term 6- to 8-week studies. Therefore, the authors31 suggested that trials of antipsychotics could be shortened to 3 to 4 weeks. However, a trial duration of 4 weeks leaves the possibility that some patients might have improved much more if the trial had lasted longer. Furthermore, our criteria of remission used only symptomatic criteria, not time criteria. However, Jager et al32 suggested that when the proposed remission criteria are applied to studies for short-term treatment, the time criterion can be waived. Actually, some studies for short-term treatment have used symptomatic criterion, without time criterion of the remission, for a “cross-sectional remission”.33–36 Finally, the BPRS does not contain adequate representation of negative symptoms.8 We did not use the Scale for Assessment of Negative Symptoms criteria for evaluating remission.
In addition to the BPRS, there are other widely used scales proposed by the Schizophrenia Working Group, such as PANSS5 or the Scale for Assessment of Negative Symptoms and Positive Symptoms.37,38 Studies using these scales are needed to confirm when using remission-items scores at week 2 to predict symptomatic remission is more accurate than using percentage of symptoms reduction.
The cutoff points and sensitivity and specificity varied in different cohorts.7–9,13,14,16,17,39–43 This nonreplication may have resulted from different phases of schizophrenia (first episode or not), measure scales (eg, BPRS or PANSS), definition of outcomes (response or remission), trial duration (eg, 4, 6, or 8 weeks’ stay in hospital or longer), enrolled subjects (inpatients or outpatients), or dose strategies (fixed or flexible).7–9,13,14,16,17,39–43 However, the hypothesis that absence of an early improvement to medication is a stable predictor of subsequent lack of improvement in patients with schizophrenia is still replicated. Further studies, preferably involving other antipsychotics, larger patient groups, frequent early symptom ratings, and 6-month duration, are needed to better determine the predictive value of remission-items scores at earlier assessments for ultimate treatment remission. Moreover, whether patients with higher remission-items score at week 2 would benefit from shifting to other antipsychotic agents earlier also deserves further research.
AUTHOR DISCLOSURE INFORMATION
The authors declare no conflicts of interest.
1. Emsley R, Chiliza B, Schoeman R. Predictors of long-term outcome in schizophrenia
. Curr Opin Psychiatry
. 2008; 21: 173–177.
2. Leucht S, Kane JM, Kissling W, et al.. Clinical implications of Brief Psychiatric Rating Scale scores. Br J Psychiatry
. 2005; 187: 366–371.
3. Leucht S, Kane JM, Kissling W, et al.. What does the PANSS mean? Schizophr Res
. 2005; 79: 231–238.
4. Overall JE, Gorham DR. The Brief Psychiatric Rating Scale. Psychol Rep
. 1962; 10: 799–812.
5. Kay SR, Fiszbein A, Opler LA. The positive and negative syndrome scale (PANSS) for schizophrenia
. Schizophr Bull
. 1987; 13: 261–276.
6. van Os J, Burns T, Cavallaro R, et al.. Standardized remission
criteria in schizophrenia
. Acta Psychiatr Scand
. 2006; 113: 91–95.
7. Leucht S, Beitinger R, Kissling W. On the concept of remission
. Psychopharmacology (Berl)
. 2007; 194: 453–461.
8. Andreasen NC, Carpenter WT Jr, Kane JM, et al.. Remission
: proposed criteria and rationale for consensus. Am J Psychiatry
. 2005; 162: 441–449.
9. Kinon BJ, Chen L, Ascher-Svanum H, et al.. Early response to antipsychotic drug therapy as a clinical marker of subsequent response in the treatment of schizophrenia
. 2010; 35: 581–590.
10. Agid O, Kapur S, Arenovich T, et al.. Delayed-onset hypothesis of antipsychotic action: a hypothesis tested and rejected. Arch Gen Psychiatry
. 2003; 60: 1228–1235.
11. Leucht S, Busch R, Hamann J, et al.. Early-onset hypothesis of antipsychotic drug action: a hypothesis tested, confirmed and extended. Biol Psychiatry
. 2005; 57: 1543–1549.
12. Jager M, Schmauss M, Laux G, et al.. Early improvement as a predictor of remission
and response in schizophrenia
: results from a naturalistic study. Eur Psychiatry
. 2009; 24: 501–506.
13. Schennach-Wolff R, Seemuller FH, Mayr A, et al.. An early improvement threshold to predict response and remission
in first-episode schizophrenia
. Br J Psychiatry
. 2010; 196: 460–466.
14. Schennach-Wolff R, Jager M, Mayr A, et al.. Predictors of response and remission
in the acute treatment of first-episode schizophrenia
patients—is it all about early response? Eur Neuropsychopharmacol
. 2011; 21: 370–378.
15. Leucht S, Busch R, Kissling W, et al.. Early prediction
of antipsychotic nonresponse among patients with schizophrenia
. J Clin Psychiatry
. 2007; 68: 352–360.
16. Kinon BJ, Chen L, Ascher-Svanum H, et al.. Predicting response to atypical antipsychotics based on early response in the treatment of schizophrenia
. Schizophr Res
. 2008; 102: 230–240.
17. Chen L, Ascher-Svanum H, Stauffer V, et al.. Optimal thresholds of early response to atypical antipsychotics: application of signal detection methods. Schizophr Res
. 2009; 113: 34–40.
18. Leucht S, Kane JM. Measurement-based psychiatry: definitions of response, remission
, stability, and relapse in schizophrenia
. J Clin Psychiatry
. 2006; 67: 1813–1814.
19. Lin CH, Chou LS, Lin CH, et al.. Early prediction
of clinical response in schizophrenia
patients receiving the atypical antipsychotic zotepine. J Clin Psychiatry
. 2007; 68: 1522–1527.
20. APA. American Psychiatric Association: Structured Clinical Interview for DSM-IV
. Washington, DC: American Psychiatric Press; 1994.
21. Kane J, Honigfeld G, Singer J, et al.. Clozapine for the treatment-resistant schizophrenic. A double-blind comparison with chlorpromazine. Arch Gen Psychiatry
. 1988; 45: 789–796.
22. Lingjaerde O, Ahlfors UG, Bech P, et al.. The UKU side effect rating scale. A new comprehensive rating scale for psychotropic drugs and a cross-sectional study of side effects in neuroleptic-treated patients. Acta Psychiatr Scand Suppl
. 1987; 334: 1–100.
23. Hummer M, Holzmeister R, Kemmler G, et al.. Attitudes of patients with schizophrenia
toward placebo-controlled clinical trials. J Clin Psychiatry
. 2003; 64: 277–281.
24. Hosmer DW, Lemeshow S. Applied Logistic Regression
. 2nd ed. New York, NY: John Wiley & Sons; 1999.
25. Weinstein MC, Fineberg HV. Clinical Decisions Analysis
. Philadelphia, PA: WB Saunders; 1980.
26. Cooper SJ, Tweed J, Raniwalla J, et al.. A placebo-controlled comparison of zotepine versus chlorpromazine in patients with acute exacerbation of schizophrenia
. Acta Psychiatr Scand
. 2000; 101: 218–225.
27. Hwang TJ, Lin SK, Lin HN. Efficacy and safety of zotepine for the treatment of Taiwanese schizophrenic patients: a double-blind comparison with haloperidol. J Formos Med Assoc
. 2001; 100: 811–816.
28. Barnas C, Quiner S, Tauscher J, et al.. In vivo (123)I IBZM SPECT imaging of striatal dopamine 2 receptor occupancy in schizophrenic patients. Psychopharmacology (Berl)
. 2001; 157: 236–242.
29. Kapur S, Seeman P. Does fast dissociation from the dopamine d(2) receptor explain the action of atypical antipsychotics? A new hypothesis. Am J Psychiatry
. 2001; 158: 360–369.
30. Lane HY, Chang YC, Cheng YC, et al.. Effects of patient demographics, risperidone dosage, and clinical outcome on body weight in acutely exacerbated schizophrenia
. J Clin Psychiatry
. 2003; 64: 316–320.
31. McMahon RP, Kelly DL, Boggs DL, et al.. Feasibility of reducing the duration of placebo-controlled trials in schizophrenia
research. Schizophr Bull
. 2008; 34: 292–301.
32. Jager M, Messer T, Laux G, et al.. Standardized remission
criteria in schizophrenia
: descriptive validity and comparability with previously used outcome measures. Pharmacopsychiatry
. 2008; 41: 190–195.
33. Lasser RA, Bossie CA, Gharabawi GM, et al.. Remission
: results from a 1-year study of long-acting risperidone injection. Schizophr Res
. 2005; 77: 215–227.
34. Sethuraman G, Taylor CC, Enerson M, et al.. A retrospective comparison of cumulative time spent in remission
during treatment with olanzapine or risperidone among patients with schizophrenia
. Schizophr Res
. 2005; 79: 337–340.
35. van Os J, Drukker M, a Campo J, et al.. Validation of remission
criteria for schizophrenia
. Am J Psychiatry
. 2006; 163: 2000–2002.
36. Jager M, Riedel M, Schmauss M, et al.. Prediction
of symptom remission
during inpatient treatment. World J Biol Psychiatry
. 2009; 10: 426–434.
37. Andreasen NC, Olsen S. Negative v positive schizophrenia
. Definition and validation. Arch Gen Psychiatry
. 1982; 39: 789–794.
38. Andreasen NC. Scales for the Assessment of Negative Symptoms (SANS)
. Iowa City, IA: University of Iowa; 1983.
39. Stauffer VL, Case M, Kinon BJ, et al.. Early response to antipsychotic therapy as a clinical marker of subsequent response in the treatment of patients with first-episode psychosis. Psychiatry Res
. 2011; 187: 42–48.
40. O’Gorman C, Kapur S, Kolluri S, et al.. Early improvement on antipsychotic treatment as a predictor of subsequent response in schizophrenia
: analyses from ziprasidone clinical studies. Hum Psychopharmacol
. 2011; 26: 282–290.
41. Leucht S, Davis JM, Engel RR, et al.. Definitions of response and remission
: recommendations for their use and their presentation. Acta Psychiatr Scand Suppl
. 2009; 438: 7–14.
42. Schennach-Wolff R, Meyer S, Seemuller F, et al.. Influencing factors and predictors of early improvement in the acute treatment of schizophrenia
spectrum disorder. J Psychiatr Res
. 2011; 45: 1639–1647.
43. Beitinger R, Lin J, Kissling W, et al.. Comparative remission
rates of schizophrenic patients using various remission
criteria. Prog Neuropsychopharmacol Biol Psychiatry
. 2008; 32: 1643–1651.