ORIGINAL CONTRIBUTION
Relationship Between Symptomatic Improvement and Overall Illness Severity in Patients With Schizophrenia Gagan Fervaha, BSc,*† Ofer Agid, MD,*†‡ Hiroyoshi Takeuchi, MD, PhD,*‡ Jimmy Lee, MBBS, MMed,*‡ George Foussias, MD, PhD,*†‡ and Gary Remington, MD, PhD*†‡ Background: Whether improvement on ratings of global illness severity is differentially associated with improvement in specific symptom domains in patients with schizophrenia is not well understood. The present study examined the independent relationships between improvement in specific symptom clusters and change in global impressions of illness severity. Methods: This study included 589 patients with chronic schizophrenia who were assessed at baseline and after 6 months of antipsychotic treatment. Both clinicians and patients completed the Clinical Global Impressions–Severity of Illness Scale (CGI-S). Symptom severity was assessed using factor scores derived from the Positive and Negative Syndrome Scale. Results: Change in illness severity ratings made by the clinician and those made by the patient demonstrated moderate overlap. Nearly half of the patients were evaluated as clinically improved during the 6-month period, as rated by the clinician, with less than a third of patients experiencing a reduction in illness severity as determined by both the clinician and themselves. Improvements in clinician-rated CGI-S scores were most strongly associated with reduction in positive symptom severity. In contrast, change in patient-rated CGI-S scores was not linked to reduction in positive symptoms but rather to improvement in depressive and anxiety symptoms. This latter finding remained in a subsample of patients with relatively preserved insight into illness, suggesting that lack of insight cannot account for these findings. Finally, reduction in positive symptoms beyond 2 to 3 points was found to be clinically meaningful. Conclusions: In conclusion, change in overall illness severity, as determined by clinicians, is not necessarily interchangeable with patients' view of improvement of their own clinical status. Moreover, changes in the 2 evaluations of illness severity are associated with changes in different symptom domains. Key Words: psychotic disorder, psychosis, psychopathology, Clinical Global Impression, rating scales, minimum clinically important difference (J Clin Psychopharmacol 2015;35: 128–133)
I
llness severity in psychotic disorders is often evaluated by way of overall impression using the Clinical Global Impressions– Severity of Illness Scale (CGI-S),1,2 as well as more comprehensive rating scales that evaluate severity of specific symptoms such as the Brief Psychiatric Rating Scale (BPRS)3 or the Positive and Negative Syndrome Scale (PANSS).4 Although these latter scales are often used as primary endpoints in the evaluation of treatment efficacy,5,6 clinically important changes in illness presentation are typically based on change in global impressions.7–9 It is not known whether changes in specific symptom domains From the *Schizophrenia Division, Centre for Addiction and Mental Health; and †Institute of Medical Science, and ‡Department of Psychiatry, University of Toronto, Toronto, Ontario, Canada. Received July 28, 2014; accepted after revision December 29, 2014. Reprints: Gagan Fervaha, BSc, Schizophrenia Division, Centre for Addiction and Mental Health, 250 College St, Room 320, Toronto, Ontario, Canada M5T 1R8 (e‐mail: [email protected]). Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved. ISSN: 0271-0749 DOI: 10.1097/JCP.0000000000000286
128
www.psychopharmacology.com
are differentially associated with changes in overall impression of illness severity. Several previous studies have established that improvements in global impressions of illness severity are highly associated with a reduction in overall symptom severity as measured by a total score on a comprehensive rating scale such as the PANSS or the BPRS.7–14 Comparatively less is known about the impact of symptomatic improvement in specific domains and their relationship with changes in overall illness severity ratings. One previous study evaluated change in illness severity after an acute illness exacerbation and for a mean of 20 days until discharge.15 Change in positive symptoms was the strongest predictor of change in overall illness severity, with relatively smaller yet independent contributions made by change in negative and agitation symptoms. Another study examining inpatients followed for 4 weeks reported that change in positive psychotic symptoms explained the largest amount of variance in change in illness severity; however, the independent predictive value of different symptom domains was not evaluated concurrently.16 Both of these investigations were restricted to an acutely ill inpatient sample and had relatively short follow-up, limiting generalizability of the findings. Clinical Global Impressions–Severity of Illness Scale ratings are routinely completed by clinicians; relatively little is known about changes in illness severity as evaluated by patients themselves. One study has been published to date examining the link between total PANSS scores and change in patient-rated illness severity ratings.12 This study found that changes in patient-rated CGI-S could be linked to changes in PANSS total scores but that the link was slightly different than for clinician-rated CGI-S scores. Another study failed to find an association between changes in clinical status as rated by patients and ratings by the clinician; however, associations with changes in symptom severity were not reported.17 No study to date has examined the independent association between change in severity of specific symptom clusters and changes in CGI-S scores self-reported by patients with schizophrenia. The present study therefore sought to examine the independent predictive value of symptomatic improvement in predicting changes in overall illness severity as rated by both clinicians and patients.
METHODS Study Design and Participants Patients in the present study participated in the National Institute of Mental Health-funded multisite Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) schizophrenia study. Details of the study design and rationale,18 as well as primary findings,19 have been presented elsewhere. The primary purpose of the CATIE study was to compare the effectiveness of different atypical and a single typical antipsychotic medication through a randomized controlled trial conducted between January 2001 and December 2004 at 57 sites in the United States (16 university clinics, 10 state mental health agencies, 7 Veterans Affairs medical centers, 6 private nonprofit agencies, 4 private-practice sites, and
Journal of Clinical Psychopharmacology • Volume 35, Number 2, April 2015
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Journal of Clinical Psychopharmacology • Volume 35, Number 2, April 2015
14 mixed-system sites). A total of 1493 patients were initially randomized to receive olanzapine (7.5–30 mg/d), perphenazine (8–32 mg/d), quetiapine (200–800 mg/d), risperidone (1.5–6 mg/d), or ziprasidone (40–160 mg/d) under double-blind conditions and were followed up to 18 months or until treatment was discontinued for any reason (phase 1 endpoint).18 The primary sample for the present analysis included 589 patients with both baseline and 6-month follow-up data. This endpoint was selected rather than the phase 1 endpoint because of its variable length and the fact that much of the clinical response is evidenced within the initial 6 months of antipsychotic treatment.20 The study inclusion criteria have been reported previously.18 Briefly, participants were eligible if they were between the ages of 18 and 65 years and had a diagnosis of schizophrenia confirmed using the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Fourth EditionAxis I Disorders.21 Participants were excluded from the study if they had a diagnosis of schizoaffective disorder, mental retardation, or other cognitive disorders; had only 1 episode of schizophrenia; had a history of treatment resistance; were pregnant or breast-feeding; or had a serious and unstable medical condition. Notably, patients with comorbid conditions such as a substance use disorder were not excluded. The study was approved by the institutional ethics review board at each site, and written informed consent was obtained from the patients or their legal guardians. All participants demonstrated adequate decision-making capacity in regard to participating in the study as determined by the MacArthur Competence Assessment Tool.22
Measures The CGI-S is a 7-point anchored scale.1 Two variations of the scale were used in the CATIE study, one rated by the clinician and the other rated by the patient.18 For the clinician-rated CGI-S, the clinician was asked to rate how mentally ill the patient is based on his/her “total clinical experience with the particular population.” The following scores could be given: 1, normal, not at all ill; 2, borderline mentally ill; 3, mildly ill; 4, moderately ill; 5, markedly ill; 6, severely ill; and 7, very severely ill. For the patient-rated CGI-S, the patient was asked, “on a scale of ‘1’ to ‘7,’ where ‘1’ is not at all ill and ‘7’ is the worst that your illness has ever been, how would you rate the severity of your symptoms of schizophrenia (or patient equivalent)?” The patient-rated CGI-S had an identical 7-point anchored scoring system (ie, 1, normal, not at all ill; 2, borderline mentally ill; 3, mildly ill; 4, moderately ill; 5, markedly ill; 6, severely ill; and 7, very severely ill). The PANSS was used to assess discrete aspects of psychopathology using derived factor scores.4,23 Severity of specific items is rated on a 7-point scale (with higher scores indicating more severe symptoms), and from the 30 items included in the PANSS, scores were computed for 5 psychopathological dimensions as follows: positive, negative, disorganized thought, excitement and hostility, and depressive and anxiety symptoms.23 The following items were summed to compute the respective factor score: (1) positive: delusions, hallucinatory behavior, grandiosity, suspiciousness, stereotyped thinking, somatic concern, unusual thought content, and lack of judgment and insight; (2) negative: blunted affect, emotional withdrawal, poor rapport, passive social withdrawal, lack of spontaneity, motor retardation, and active social avoidance; (3) disorganized thought: conceptual disorganization, difficulty in abstract thinking, mannerisms and posturing, poor attention, disturbance in volition, preoccupation, and disorientation; (4) excitement and hostility: excitement, hostility, uncooperativeness, and poor impulse © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Symptom Improvement and CGI-S
control; and (5) depressive and anxiety symptoms: anxiety, guilt, tension, and depression.23
Statistical Analyses Change in clinical status and symptom severity was evaluated using paired-samples t tests. Spearman rank-order correlations were computed to examine the bivariate relationship between changes in the 2 CGI-S scales. Multiple regression modeling with forced entry was used to examine independent predictors of change in both the clinician-rated and patient-rated CGI-S scales separately. To ensure that the independent predictors of patientrated CGI-S scores were not a consequence of impaired clinical insight, the multiple regression analysis with patient-rated CGI-S scores was repeated while excluding patients with moderate-to-severe lack of insight (as defined by a PANSS clinical insight [G12] score > 3). Next, the minimum clinically important difference in positive symptoms was estimated using both a distribution-based method24 and an anchor-based method.25 This distribution-based method defines a change in scores beyond the standard error of measurement (SEM) as clinically relevant, an approach adopted by a previous report estimating this difference for PANSS total scores.12 The SEM is a measure of variability of scores derived from a specific instrument and represents the error intrinsic to an observed score related to measurement with a particular test (ie, the PANSS) that obscured the “true” score. The SEMp isffiffiffiffiffiffiffiffiffiffi typi-ffi cally calculated using the following formula: SEM¼σ 1−rxx where σ represents the SD and rxx represents the test-test reliability measured by an intraclass correlation coefficient (ICC).24 The ICC was calculated using a 2-way mixed model examining absolute agreement between severity ratings at baseline with those made at 6 months. Consistent with previous work,12,26 a clinically stable subsample that did not experience a change in clinical severity was examined for this analysis. For the anchor-based method, change scores for individuals experiencing a 1-point improvement in overall clinical status were compared with scores for patients who did not experience a change, an approach that has been used previously.27 A 2-sided P value of less than 0.05 was considered statistically significant. Statistical analyses were carried out using SPSS version 20 (IBM Corporation, Armonk, NY).
RESULTS Patient Characteristics Baseline demographic and clinical characteristics of the sample are presented in Table 1. Notably, the patients experienced significant improvement in their clinical status as assessed by both the clinician and the patients themselves (Table 2).
Clinician-Rated Versus Patient-Rated CGI Change in clinician-rated CGI-S scores was significantly and positively correlated with change in patient-rated CGI-S scores (rs = 0.30, P < 0.001). Approximately 46.2% of the patients experienced improvement in clinical status, as rated by the clinician (ie, at least a 1-point decrease in CGI-S), and 46.9% of the patients rated themselves as less severe; however, only 30.0% of the patients demonstrated clinical improvement as rated by both themselves and the clinician. Restricting these analyses to individuals with relatively preserved insight into illness (n = 391) revealed a similar pattern of findings; specifically, change in clinician-rated CGI-S scores www.psychopharmacology.com
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
129
Journal of Clinical Psychopharmacology • Volume 35, Number 2, April 2015
Fervaha et al
TABLE 1. Sociodemographic and Clinical Characteristics of the Study Sample Variable
Mean (SD) or %
Range
41.5 (11.2) 74.6 62.5 83.8 12.1 (2.2) 14.8 (11.3) 3.9 (1.0) 3.5 (1.5) 74.0 (17.3) 21.3 (6.6) 19.3 (6.8) 16.5 (5.3) 6.7 (2.8) 10.2 (3.8)
18–66 — — — 3–21 0–56 1–7 1–7 32–133 8–42 7–40 7–37 4–17 4–22
Age, y Sex (male) Race (white) Employment status (unemployed) Patient's education, y Illness duration (years since the first prescribed antipsychotic medication) Clinician-rated CGI-S Patient-rated CGI-S PANSS (total score) PANSS–positive factor score PANSS–negative factor score PANSS–disorganization factor score PANSS–excitement factor score PANSS–anxiety/depression factor score
was still moderately correlated with patient-rated CGI-S scores in this subsample of patients (rs = 0.29, P < 0.001).
Determinants of Change in Clinician-Rated CGI-S We constructed a multiple regression model to examine the independent predictors of change in clinician-rated CGI-S scores. This model revealed that change in positive psychotic symptoms was the strongest predictor of clinician-rated illness severity scores, with additional contributions made by change in excitement and hostility scores (Table 3). Changes in severity of other symptom domains were not independently associated with change in clinician-rated illness severity scores. Given the broad inclusion criteria of the CATIE study, without imposition of an illness severity criterion,18 we repeated the above analysis to include individuals with an above average baseline PANSS total score. This model was largely consistent with the model computed with the full sample, in that change in positive symptoms emerged as the strongest predictor of clinician-rated illness severity scores (Table 3); however, there was no additional contribution made by change in excitement and hostility scores; rather, changes in disorganization symptoms emerged as a significant independent predictor in these more severely ill patients.
Estimating Minimum Clinically Important Difference in Positive Symptoms A subsample of 243 patients were rated by clinicians to have the same level of illness severity at 6 months as they did at
baseline. These individuals had a mean baseline PANSS positive factor score of 20.5 (SD, 6.3), and importantly, the SD of this score at 6 months was not different (mean, 18.8; SD, 6.3). The ICC for these scores was 0.82, which translates to an SEM for this score of 2.7 points. The individuals with a 1-point improvement in clinician-evaluated illness severity differed from those whose global clinical status did not change in terms of PANSS positive factor scores by a mean of 2.2 points (independent-sample t test; t444 = 4.8, P < 0.001). This analysis was repeated in patients with more severe illness, and the minimum clinically important difference was similar. These analyses were conducted in a subsample of 108 patients who were rated by clinicians to have the same level of illness severity at 6 months as they did at baseline and had a PANSS total score greater than the baseline mean score. These patients had a mean baseline PANSS positive factor score of 26.5 (SD, 5.1). The ICC for the baseline and follow-up scores was 0.67, which translates to an SEM for this score of 2.9 points. Moreover, within this subsample, the individuals with a 1-point improvement in clinician-evaluated illness severity differed from those whose global clinical status did not change in terms of PANSS positive factor scores by a mean of 3.1 points (independent-sample t test; t198 = 4.3, P < 0.001).
Determinants of Change in Patient-Rated CGI-S Improvement in depressive and anxiety symptoms were found to be the only significant predictor of improvement in illness severity, as rated by the patient (β = 0.09, P = 0.05; overall
TABLE 2. Change in Clinical Variables During 6 Months Clinical Variable (Change Scores)
Estimated Mean Difference
SD
Effect Size*
Paired-Samples t Statistic
P
−0.47 −0.45 −3.12 −2.17 −1.73 0.35 −1.48
1.05 1.68 5.73 5.67 4.21 3.04 3.65
0.45 0.27 0.54 0.38 0.41 0.12 0.41
10.94 6.49 13.20 9.28 9.96 −2.79 9.85
Relationship Between Symptomatic Improvement and Overall Illness Severity in Patients With Schizophrenia Gagan Fervaha, BSc,*† Ofer Agid, MD,*†‡ Hiroyoshi Takeuchi, MD, PhD,*‡ Jimmy Lee, MBBS, MMed,*‡ George Foussias, MD, PhD,*†‡ and Gary Remington, MD, PhD*†‡ Background: Whether improvement on ratings of global illness severity is differentially associated with improvement in specific symptom domains in patients with schizophrenia is not well understood. The present study examined the independent relationships between improvement in specific symptom clusters and change in global impressions of illness severity. Methods: This study included 589 patients with chronic schizophrenia who were assessed at baseline and after 6 months of antipsychotic treatment. Both clinicians and patients completed the Clinical Global Impressions–Severity of Illness Scale (CGI-S). Symptom severity was assessed using factor scores derived from the Positive and Negative Syndrome Scale. Results: Change in illness severity ratings made by the clinician and those made by the patient demonstrated moderate overlap. Nearly half of the patients were evaluated as clinically improved during the 6-month period, as rated by the clinician, with less than a third of patients experiencing a reduction in illness severity as determined by both the clinician and themselves. Improvements in clinician-rated CGI-S scores were most strongly associated with reduction in positive symptom severity. In contrast, change in patient-rated CGI-S scores was not linked to reduction in positive symptoms but rather to improvement in depressive and anxiety symptoms. This latter finding remained in a subsample of patients with relatively preserved insight into illness, suggesting that lack of insight cannot account for these findings. Finally, reduction in positive symptoms beyond 2 to 3 points was found to be clinically meaningful. Conclusions: In conclusion, change in overall illness severity, as determined by clinicians, is not necessarily interchangeable with patients' view of improvement of their own clinical status. Moreover, changes in the 2 evaluations of illness severity are associated with changes in different symptom domains. Key Words: psychotic disorder, psychosis, psychopathology, Clinical Global Impression, rating scales, minimum clinically important difference (J Clin Psychopharmacol 2015;35: 128–133)
I
llness severity in psychotic disorders is often evaluated by way of overall impression using the Clinical Global Impressions– Severity of Illness Scale (CGI-S),1,2 as well as more comprehensive rating scales that evaluate severity of specific symptoms such as the Brief Psychiatric Rating Scale (BPRS)3 or the Positive and Negative Syndrome Scale (PANSS).4 Although these latter scales are often used as primary endpoints in the evaluation of treatment efficacy,5,6 clinically important changes in illness presentation are typically based on change in global impressions.7–9 It is not known whether changes in specific symptom domains From the *Schizophrenia Division, Centre for Addiction and Mental Health; and †Institute of Medical Science, and ‡Department of Psychiatry, University of Toronto, Toronto, Ontario, Canada. Received July 28, 2014; accepted after revision December 29, 2014. Reprints: Gagan Fervaha, BSc, Schizophrenia Division, Centre for Addiction and Mental Health, 250 College St, Room 320, Toronto, Ontario, Canada M5T 1R8 (e‐mail: [email protected]). Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved. ISSN: 0271-0749 DOI: 10.1097/JCP.0000000000000286
128
www.psychopharmacology.com
are differentially associated with changes in overall impression of illness severity. Several previous studies have established that improvements in global impressions of illness severity are highly associated with a reduction in overall symptom severity as measured by a total score on a comprehensive rating scale such as the PANSS or the BPRS.7–14 Comparatively less is known about the impact of symptomatic improvement in specific domains and their relationship with changes in overall illness severity ratings. One previous study evaluated change in illness severity after an acute illness exacerbation and for a mean of 20 days until discharge.15 Change in positive symptoms was the strongest predictor of change in overall illness severity, with relatively smaller yet independent contributions made by change in negative and agitation symptoms. Another study examining inpatients followed for 4 weeks reported that change in positive psychotic symptoms explained the largest amount of variance in change in illness severity; however, the independent predictive value of different symptom domains was not evaluated concurrently.16 Both of these investigations were restricted to an acutely ill inpatient sample and had relatively short follow-up, limiting generalizability of the findings. Clinical Global Impressions–Severity of Illness Scale ratings are routinely completed by clinicians; relatively little is known about changes in illness severity as evaluated by patients themselves. One study has been published to date examining the link between total PANSS scores and change in patient-rated illness severity ratings.12 This study found that changes in patient-rated CGI-S could be linked to changes in PANSS total scores but that the link was slightly different than for clinician-rated CGI-S scores. Another study failed to find an association between changes in clinical status as rated by patients and ratings by the clinician; however, associations with changes in symptom severity were not reported.17 No study to date has examined the independent association between change in severity of specific symptom clusters and changes in CGI-S scores self-reported by patients with schizophrenia. The present study therefore sought to examine the independent predictive value of symptomatic improvement in predicting changes in overall illness severity as rated by both clinicians and patients.
METHODS Study Design and Participants Patients in the present study participated in the National Institute of Mental Health-funded multisite Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) schizophrenia study. Details of the study design and rationale,18 as well as primary findings,19 have been presented elsewhere. The primary purpose of the CATIE study was to compare the effectiveness of different atypical and a single typical antipsychotic medication through a randomized controlled trial conducted between January 2001 and December 2004 at 57 sites in the United States (16 university clinics, 10 state mental health agencies, 7 Veterans Affairs medical centers, 6 private nonprofit agencies, 4 private-practice sites, and
Journal of Clinical Psychopharmacology • Volume 35, Number 2, April 2015
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Journal of Clinical Psychopharmacology • Volume 35, Number 2, April 2015
14 mixed-system sites). A total of 1493 patients were initially randomized to receive olanzapine (7.5–30 mg/d), perphenazine (8–32 mg/d), quetiapine (200–800 mg/d), risperidone (1.5–6 mg/d), or ziprasidone (40–160 mg/d) under double-blind conditions and were followed up to 18 months or until treatment was discontinued for any reason (phase 1 endpoint).18 The primary sample for the present analysis included 589 patients with both baseline and 6-month follow-up data. This endpoint was selected rather than the phase 1 endpoint because of its variable length and the fact that much of the clinical response is evidenced within the initial 6 months of antipsychotic treatment.20 The study inclusion criteria have been reported previously.18 Briefly, participants were eligible if they were between the ages of 18 and 65 years and had a diagnosis of schizophrenia confirmed using the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Fourth EditionAxis I Disorders.21 Participants were excluded from the study if they had a diagnosis of schizoaffective disorder, mental retardation, or other cognitive disorders; had only 1 episode of schizophrenia; had a history of treatment resistance; were pregnant or breast-feeding; or had a serious and unstable medical condition. Notably, patients with comorbid conditions such as a substance use disorder were not excluded. The study was approved by the institutional ethics review board at each site, and written informed consent was obtained from the patients or their legal guardians. All participants demonstrated adequate decision-making capacity in regard to participating in the study as determined by the MacArthur Competence Assessment Tool.22
Measures The CGI-S is a 7-point anchored scale.1 Two variations of the scale were used in the CATIE study, one rated by the clinician and the other rated by the patient.18 For the clinician-rated CGI-S, the clinician was asked to rate how mentally ill the patient is based on his/her “total clinical experience with the particular population.” The following scores could be given: 1, normal, not at all ill; 2, borderline mentally ill; 3, mildly ill; 4, moderately ill; 5, markedly ill; 6, severely ill; and 7, very severely ill. For the patient-rated CGI-S, the patient was asked, “on a scale of ‘1’ to ‘7,’ where ‘1’ is not at all ill and ‘7’ is the worst that your illness has ever been, how would you rate the severity of your symptoms of schizophrenia (or patient equivalent)?” The patient-rated CGI-S had an identical 7-point anchored scoring system (ie, 1, normal, not at all ill; 2, borderline mentally ill; 3, mildly ill; 4, moderately ill; 5, markedly ill; 6, severely ill; and 7, very severely ill). The PANSS was used to assess discrete aspects of psychopathology using derived factor scores.4,23 Severity of specific items is rated on a 7-point scale (with higher scores indicating more severe symptoms), and from the 30 items included in the PANSS, scores were computed for 5 psychopathological dimensions as follows: positive, negative, disorganized thought, excitement and hostility, and depressive and anxiety symptoms.23 The following items were summed to compute the respective factor score: (1) positive: delusions, hallucinatory behavior, grandiosity, suspiciousness, stereotyped thinking, somatic concern, unusual thought content, and lack of judgment and insight; (2) negative: blunted affect, emotional withdrawal, poor rapport, passive social withdrawal, lack of spontaneity, motor retardation, and active social avoidance; (3) disorganized thought: conceptual disorganization, difficulty in abstract thinking, mannerisms and posturing, poor attention, disturbance in volition, preoccupation, and disorientation; (4) excitement and hostility: excitement, hostility, uncooperativeness, and poor impulse © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Symptom Improvement and CGI-S
control; and (5) depressive and anxiety symptoms: anxiety, guilt, tension, and depression.23
Statistical Analyses Change in clinical status and symptom severity was evaluated using paired-samples t tests. Spearman rank-order correlations were computed to examine the bivariate relationship between changes in the 2 CGI-S scales. Multiple regression modeling with forced entry was used to examine independent predictors of change in both the clinician-rated and patient-rated CGI-S scales separately. To ensure that the independent predictors of patientrated CGI-S scores were not a consequence of impaired clinical insight, the multiple regression analysis with patient-rated CGI-S scores was repeated while excluding patients with moderate-to-severe lack of insight (as defined by a PANSS clinical insight [G12] score > 3). Next, the minimum clinically important difference in positive symptoms was estimated using both a distribution-based method24 and an anchor-based method.25 This distribution-based method defines a change in scores beyond the standard error of measurement (SEM) as clinically relevant, an approach adopted by a previous report estimating this difference for PANSS total scores.12 The SEM is a measure of variability of scores derived from a specific instrument and represents the error intrinsic to an observed score related to measurement with a particular test (ie, the PANSS) that obscured the “true” score. The SEMp isffiffiffiffiffiffiffiffiffiffi typi-ffi cally calculated using the following formula: SEM¼σ 1−rxx where σ represents the SD and rxx represents the test-test reliability measured by an intraclass correlation coefficient (ICC).24 The ICC was calculated using a 2-way mixed model examining absolute agreement between severity ratings at baseline with those made at 6 months. Consistent with previous work,12,26 a clinically stable subsample that did not experience a change in clinical severity was examined for this analysis. For the anchor-based method, change scores for individuals experiencing a 1-point improvement in overall clinical status were compared with scores for patients who did not experience a change, an approach that has been used previously.27 A 2-sided P value of less than 0.05 was considered statistically significant. Statistical analyses were carried out using SPSS version 20 (IBM Corporation, Armonk, NY).
RESULTS Patient Characteristics Baseline demographic and clinical characteristics of the sample are presented in Table 1. Notably, the patients experienced significant improvement in their clinical status as assessed by both the clinician and the patients themselves (Table 2).
Clinician-Rated Versus Patient-Rated CGI Change in clinician-rated CGI-S scores was significantly and positively correlated with change in patient-rated CGI-S scores (rs = 0.30, P < 0.001). Approximately 46.2% of the patients experienced improvement in clinical status, as rated by the clinician (ie, at least a 1-point decrease in CGI-S), and 46.9% of the patients rated themselves as less severe; however, only 30.0% of the patients demonstrated clinical improvement as rated by both themselves and the clinician. Restricting these analyses to individuals with relatively preserved insight into illness (n = 391) revealed a similar pattern of findings; specifically, change in clinician-rated CGI-S scores www.psychopharmacology.com
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
129
Journal of Clinical Psychopharmacology • Volume 35, Number 2, April 2015
Fervaha et al
TABLE 1. Sociodemographic and Clinical Characteristics of the Study Sample Variable
Mean (SD) or %
Range
41.5 (11.2) 74.6 62.5 83.8 12.1 (2.2) 14.8 (11.3) 3.9 (1.0) 3.5 (1.5) 74.0 (17.3) 21.3 (6.6) 19.3 (6.8) 16.5 (5.3) 6.7 (2.8) 10.2 (3.8)
18–66 — — — 3–21 0–56 1–7 1–7 32–133 8–42 7–40 7–37 4–17 4–22
Age, y Sex (male) Race (white) Employment status (unemployed) Patient's education, y Illness duration (years since the first prescribed antipsychotic medication) Clinician-rated CGI-S Patient-rated CGI-S PANSS (total score) PANSS–positive factor score PANSS–negative factor score PANSS–disorganization factor score PANSS–excitement factor score PANSS–anxiety/depression factor score
was still moderately correlated with patient-rated CGI-S scores in this subsample of patients (rs = 0.29, P < 0.001).
Determinants of Change in Clinician-Rated CGI-S We constructed a multiple regression model to examine the independent predictors of change in clinician-rated CGI-S scores. This model revealed that change in positive psychotic symptoms was the strongest predictor of clinician-rated illness severity scores, with additional contributions made by change in excitement and hostility scores (Table 3). Changes in severity of other symptom domains were not independently associated with change in clinician-rated illness severity scores. Given the broad inclusion criteria of the CATIE study, without imposition of an illness severity criterion,18 we repeated the above analysis to include individuals with an above average baseline PANSS total score. This model was largely consistent with the model computed with the full sample, in that change in positive symptoms emerged as the strongest predictor of clinician-rated illness severity scores (Table 3); however, there was no additional contribution made by change in excitement and hostility scores; rather, changes in disorganization symptoms emerged as a significant independent predictor in these more severely ill patients.
Estimating Minimum Clinically Important Difference in Positive Symptoms A subsample of 243 patients were rated by clinicians to have the same level of illness severity at 6 months as they did at
baseline. These individuals had a mean baseline PANSS positive factor score of 20.5 (SD, 6.3), and importantly, the SD of this score at 6 months was not different (mean, 18.8; SD, 6.3). The ICC for these scores was 0.82, which translates to an SEM for this score of 2.7 points. The individuals with a 1-point improvement in clinician-evaluated illness severity differed from those whose global clinical status did not change in terms of PANSS positive factor scores by a mean of 2.2 points (independent-sample t test; t444 = 4.8, P < 0.001). This analysis was repeated in patients with more severe illness, and the minimum clinically important difference was similar. These analyses were conducted in a subsample of 108 patients who were rated by clinicians to have the same level of illness severity at 6 months as they did at baseline and had a PANSS total score greater than the baseline mean score. These patients had a mean baseline PANSS positive factor score of 26.5 (SD, 5.1). The ICC for the baseline and follow-up scores was 0.67, which translates to an SEM for this score of 2.9 points. Moreover, within this subsample, the individuals with a 1-point improvement in clinician-evaluated illness severity differed from those whose global clinical status did not change in terms of PANSS positive factor scores by a mean of 3.1 points (independent-sample t test; t198 = 4.3, P < 0.001).
Determinants of Change in Patient-Rated CGI-S Improvement in depressive and anxiety symptoms were found to be the only significant predictor of improvement in illness severity, as rated by the patient (β = 0.09, P = 0.05; overall
TABLE 2. Change in Clinical Variables During 6 Months Clinical Variable (Change Scores)
Estimated Mean Difference
SD
Effect Size*
Paired-Samples t Statistic
P
−0.47 −0.45 −3.12 −2.17 −1.73 0.35 −1.48
1.05 1.68 5.73 5.67 4.21 3.04 3.65
0.45 0.27 0.54 0.38 0.41 0.12 0.41
10.94 6.49 13.20 9.28 9.96 −2.79 9.85
