- Email: [email protected]
Limited practice effects and evaluation of expectation for change: MATRICS Consensus Cognitive Battery
Schizophrenia Research 159 (2014) 188–192
Contents lists available at ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
Limited practice effects and evaluation of expectation for change: MATRICS Consensus Cognitive Battery Jarett E. Roseberry ⁎, S. Kristian Hill Department of Psychology, Rosalind Franklin University of Medicine and Science, United States
a r t i c l e
i n f o
Article history: Received 14 March 2014 Received in revised form 9 August 2014 Accepted 12 August 2014 Available online 31 August 2014 Keywords: Schizophrenia Practice effects Cognitive abilities Neuropsychology
a b s t r a c t Understanding of the impact of antipsychotic medications on cognition requires differentiating between treatment effects and practice effects. This prospective study examines expectations for change on neuropsychological assessments and possible differential practice effects in community dwelling schizophrenia patients (n = 27) who are clinically stable and on a stable medication regimen when compared to demographically similar psychiatrically healthy controls (n = 29). All participants were administered the MATRICS Consensus Cognitive Battery (MCCB) twice over a period of four weeks. The use of Regression Based Norms for Change (RBNC) and Reliable Change Index (RCI) was completed to anchor estimates of meaningful change to a demographically similar control group. A repeated measures ANOVA was used to examine the effects of time and diagnosis on MCCB composite scores. A repeated measures MANOVA was used to examine the effects of time and diagnosis, and their interaction for MCCB subtests. Estimates of meaningful change are provided. A significant main effect was observed for time; no significant interactions were observed. There was no support for differential practice effects. In the absence of any behavioral, cognitive, or pharmaceutical interventions, these findings suggest limited change in performance over time in either group. © 2014 Elsevier B.V. All rights reserved.
1. Introduction
1.1. Cognition in schizophrenia
Differentiating between treatment effects and practice effects is essential to understanding the impact of antipsychotic medications and cognitive rehabilitation techniques on cognition. One challenge facing schizophrenia researchers is estimating the extent of practice effects (Goldberg et al., 2010). Few clinical trials have included healthy control groups to quantify the impact that repeated exposure to the testing environment. Further, it is impractical/unethical to delay study treatments to establish practice effects on cognitive outcome measures in untreated patients. Despite the potential confounds associated with practice effects, improved cognitive performance has often been attributed to pharmacological treatments. This has led some authors to contend that improved test performance may not differ from practice effects observed in healthy controls. One approach to addressing this issue is to assess practice effects with schizophrenia patients who are clinically stable and on a stable medication regimen compared to a demographically similar psychiatrically healthy control group tested at the same interval (Goldberg et al., 2010). By assessing change in this matter, one can better isolate practice effects from treatment effects in the context of clinical trials.
Cognitive impairments have been well documented in patients with schizophrenia (Blanchard and Neale, 1994; Bilder et al., 2000). These impairments are present across a wide range of cognitive domains, including executive function, visual attention, auditory attention, verbal memory, visual memory, motor skills, and visual perception (Heinrichs and Zakzanis, 1998; Hill et al., 2004), yet impairments can be heterogeneous in nature (Heinrichs et al., 2008). Level of severity typically ranges from moderate to severe (Heinrichs and Zakzanis, 1998), and these generalized cognitive impairments seem to play a critical role in functional outcome.
⁎ Corresponding author at: Rosalind Franklin University of Medicine and Science, Department of Psychology, 3333 Green Bay Road, North Chicago, IL 60064, United States. Tel.: +1 847 578 8749; fax: +1 847 578 8765. E-mail address: [email protected] (J.E. Roseberry).
http://dx.doi.org/10.1016/j.schres.2014.08.004 0920-9964/© 2014 Elsevier B.V. All rights reserved.
1.2. Practice effects When accounting for improved performance over time, medication effects and/or practice effects are two of the primary explanations in schizophrenia spectrum populations. When investigating specific cognitive domains, practice effects have been observed across a broad range of cognitive domains in the context of clinical trials and research reports within schizophrenia spectrum populations (Hill et al., 2004; Goldberg et al., 2007; Goldberg et al., 2010). Interestingly, minimal practice effects have been observed on the composite score of the MCCB (Keefe et al., 2011). One methodological challenge facing clinical trials evaluating either first generation antipsychotics (FGA) or second generation antipsychotics
J.E. Roseberry, S. Kristian Hill / Schizophrenia Research 159 (2014) 188–192
(SGA) has been the absence of a control group. By including a control group, one can quantify the impact that exposure to the testing environment has on improving performance during subsequent assessments. When nonpsychotic groups were compared to schizophrenia patients using SGA at similar testing intervals, the apparent cognitive benefit of SGA was consistent with practice effects observed in healthy populations (Goldberg et al., 2007). Thus, performance on cognitive measures may be artificially inflated at retest due to familiarity with the tests, increased comfort and confidence in test taking strategies, and other practice effects. Differentiating between treatment effects and practice effects can be challenging. However, this is an area that is critical to understanding the cognitive impact of antipsychotic treatments and cognitive rehabilitation interventions. One way to address the issue of differentiating between treatment effects and practice effects is to establish expectations for meaningful change on neuropsychological measures. By establishing expectation for meaningful change through assessing stable schizophrenics on a stable medication regimen with no behavioral or therapeutic interventions, with a demographically similar control group, one would be able to better estimate practice effects. A unique way to complete this is to use conservative statistical techniques, such as RBNC and RCI, to anchor retest scores to demographically similar controls to establish expectations for change. Prior studies assessing practice effects in schizophrenia are complicated by changes in medication status (clinical trials) or have failed to include controls to provide normative expectations for practice effects. Thus, this study was designed to prospectively examine practice effects in both clinically stable schizophrenia patients on stable medication regimens and demographically similar healthy controls. Further, to compliment that approach the present study also aims to provide estimates for meaningful change on the MCCB, based on RCI estimates anchored to the control group with RBNC statistics. This will provide future studies with a reference point for a range of expected variability on the MCCB. As with prior studies (Goldberg et al., 2007; Nuechterlein et al., 2008; Goldberg et al., 2010) practice effects were expected to be similar for both groups and no differential practice effects (i.e., one group experiencing greater practice effects) were predicted. 2. Methods 2.1. Participants Participants were 27 community dwelling individuals who met criteria for schizophrenia spectrum disorders and 29 psychiatrically healthy individuals based on Structured Clinical Interview for DSM-IV (SCID). Patients were referred by primary care physicians, who provided medication status and dosage for each patient. Healthy participants were recruited from the community via local advertisements and a research registry. To limit effects of both acute illness and recent changes to medication treatments, all patients were clinically stable, meaning there was no acute symptomatology, significant change in positive symptom severity, or changes in pharmacotherapy regimen for one month prior to and during the study. All patients were assessed with the Positive and Negative Syndrome Scale (PANSS) (Table 1), and medication status and dosage were recorded at each visit. Furthermore, no participant was excluded at follow-up assessment due to hospitalization, substance use/abuse or dependence. All patients were treated with either SGA (n = 24) or FGA (n = 3). Concomitant medications included selective serotonin reuptake inhibitors (SSRI) (n = 5), mood stabilizers (n = 3), lithium (n = 1), anticholinergics (n = 2), benzodiazepine (n = 2), and serotonin receptor partial agonist (n = 1). Correlations between overall neuropsychological scores and clinical symptom ratings, medication status, and Chlorpromazine equivalents (Andreasen et al., 2010) were small and nonsignificant. Thus, these measures were not included as covariates for group comparisons. All participants were free of substance abuse within the last three months,
189
Table 1 Participant demographic characteristics. Healthy controls: Patients:
Age (years): Sex: Male Female Education: Parental SES: WASI (2 subtest IQ): WRAT-III reading: Clinical data: Illness duration (years): PANSS total: PANSS positive: PANSS negative: Side effect ratings: AIMS total: ESRS total:
F/χ2
df
p
n = 29
n = 27
37.28 (11.32)
33.59 (9.34)
1.75
1.54
0.19
62.1% 37.9% 13.41 3.29 (1.15) 98.76 (10.96) 93.43 (11.09)
66.7% 33.3% 14.41 3.00 (0.98) 101.59 (13.78) 98.67 (13.66)
1.25
1
0.26
2.59 0.96 0.73 2.45
1.54 1.52 1.54 1.54
0.11 0.33 0.40 0.12
11.85 (10.78) 38.00 (6.57) 17.73 (4.06) 19.27 (5.16) 0.80 (1.32) 4.07 (4.37)
lifetime history of substance dependence, neurological disease, head injury with loss of consciousness greater than 10 min, and systemic disorders known to affect brain function. There were no group differences on education level, premorbid functioning, age, sex, and parental SES (Table 1). 2.2. Measures/procedures The Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS) initiative was an NIMH supported program for the development of a consensus cognitive battery for use in clinical trials. The MATRICS Neurocognition Committee evaluated more than 90 tests in seven cognitive domains to determine which tests were best suited for examining cognitive change in schizophrenia spectrum disorders. 20 tests were chosen for a beta battery, which were then administered to 176 schizophrenia spectrum patients, and readministered after a four-week interval (n = 167) to compare test–retest reliability among candidate tests (Nuechterlein et al., 2008). Ten tests across seven cognitive domains were selected to comprise the MCCB. In the present study all subjects were administered the MCCB twice over a period of four weeks. Alternative forms were used for the Brief Visuospatial Memory Test—Revised (BVMT-R), per MCCB administration suggestions. 2.3. Statistical analysis A repeated measures ANOVA was used to compute an omnibus test for differential practice effects among groups on the composite score of the MCCB. A repeated measures MANOVA was used for differential practice effects among groups or subtests of the MCCB, and to provide a detailed analysis of potential practice effects. Given the limited power of this small sample size, the probability of a Type II error is increased. To improve power and reduce the probability of a Type II error, a bootstrapping technique was used. For each subtest, estimates of meaningful change were computed using two conservative statistical approaches (RBNC and RCI). The RCI approach allows one to compute the amount of change necessary to reach significance (p b .05) by taking into account the test–retest correlation coefficient (Jacobson and Truax, 1991; Salinsky et al., 2001). RCI was compared to a 90% confidence interval, and scores falling outside that confidence interval would be considered meaningful (Jacobson and Truax, 1991). The RCI approach allows one to compute the amount of change necessary to reach significance (p b .05) by taking into account the test–retest correlation coefficient (Jacobson and Truax, 1991; Salinsky et al., 2001). The RBNC approach takes into account individual demographic characteristics which may influence retest scores,
190
J.E. Roseberry, S. Kristian Hill / Schizophrenia Research 159 (2014) 188–192
and regression based norms that utilize baseline test performance to predict retest scores (McSweeny et al., 1993). RBNC makes a conservative effort to correct several sources of measurement error, which may impact retest scores. Also, RBNC examines observed results versus expected test–retest changes. This approach enables one to estimate the relative magnitude of change across cognitive domains to assist in determining whether change scores are simply practice effects or meaningful change/cognitive improvement (Herman et al., 1996). For RBNC statistics, baseline scores for the healthy control group on age, education, and WRAT-III reading scores were entered into a regression analysis as predictors of retest scores. The mean differences between observed and predicted retest scores were computed as an indicator of performance change across tests. Bootstrapping techniques can be used to reduce the probability of a Type-II error by modeling a larger sample size. 3. Results 3.1. Repeated measures MANOVA Repeated measures ANOVA for the MCCB composite score revealed a significant main effect for time [F (1, 54) = 13.40, p = .001], (effect size: d = 0.21; ŋp2 = 0.20) indicating that all groups improved over time. There was no main effect of diagnosis [F (1, 54) = 0.47, p = 0.50], (effect size: d = 0.18; ŋp2 = 0.01). The time by diagnosis interaction was non-significant [F (1, 54) = 0.08, p = 0.78] (effect size: d = 0.15; ŋp2 = 0.002), thus there was no evidence of differential practice effects over time. A repeated measures MANOVA was computed to assess for differential practice effects among individual subtests of the
MCCB. This revealed a significant main effect of time [F (1, 54) = 9.47, p = 0.003] (effect size: d = 0.24; ŋp2 = 0.15). Additionally, there was a significant main effect of test [F (9, 46) = 9.21, p = .000] (effect size: d = 0.23; ŋp2 = 0.64) indicating variability in performance among MCCB subtests. None of the two- or three-way interactions involving time and diagnosis were significant, thus there was no evidence of differential practice effects. Overall, there was evidence of modest statistical improvement over time in both patient and control groups (see Fig. 1). 3.2. Regression Based Norms for Change and Reliable Change Index When using demographically similar controls, RBNC indicated that baseline performance on the MCCB was the primary predictor, and among demographic variables, only one subtest (BVMT-R) was significantly related to age (in controls). To be consistent with prior studies (Chelune et al., 1993), the RCI threshold for change was set at ± 1.64 standard deviations (90% confidence interval). Meaningful change estimated based on RCI computations for each subtest of the MCCB can be seen on Table 2 compared to a prior report (Goldberg et al., 2010). Within the sampled population, zero to six schizophrenia patients demonstrated meaningful change on any subtest, possibly indicating variable effort on baseline assessment or meaningful change in the absence of behavioral, therapeutic, or pharmacological intervention. 3.3. Bootstrapping In general, bootstrapping did not significantly improve the model in terms of estimating significant change over time for any individual
Patient and Healthy Control mean differences between Time 1 and Time 2 assessments and 90% RCI threshold needed for meaningful change 15
(14.27) (13.96)
12
(11.53)
(11.35)
(10.14)
(9.92)
(9.87)
(9.92)
(9.97)
(10.51)
9
T-Score
6 3 0 -3 Patients Controls
-6 -9
s ail
Tr
g R an Tdin Sp Co VL at. . H p m S Sy
S LN
ze Ma
s*
* y T -IP nc -R EI PT lue SC MT C F M . V t B Ca
MCCB Subtests Fig. 1. − − − represents 90th %tile RCI threshold (raw scores were converted to T-scores) needed for meaningful change. * represents significant change from Time 1 to Time 2 for both patients and control. No differential practice effects were observed; however both groups demonstrated similar rates of change on most measures. Although RCI estimates are designed to be used with individual data, the cut-off for meaningful change was added to each column for illustrative purposes. For symbol coding, no individuals exceeded the threshold. For Trail Making Test—A, and Letter—Number Span, one schizophrenia patient and one control exceeded the threshold per subtest. For HVLT-R, MSCEIT—Managing Emotions, and CPT-IP one schizophrenic patient and two healthy controls exceeded the threshold per subtest. For Spatial Span and Category Fluency, two schizophrenia patients and two healthy controls exceeded the threshold per subtest. For Mazes, four schizophrenia patients and four healthy controls exceeded the threshold per subtest. For BVMT-R, six schizophrenia patients and six healthy controls exceeded the threshold.
J.E. Roseberry, S. Kristian Hill / Schizophrenia Research 159 (2014) 188–192 Table 2 Raw score RCI values on the MCCB needed for meaningful change. Test
Adjusted RCI (90%)
Goldberg et al. (2010) RCI
MCCB composite score Trail Making Test—A BACS symbol coding Hopkins Verbal Learning Test—Revised (HVLT-R) WMS-III, Spatial Span University of Maryland Letter—Number Span NAB—Mazes Brief Visuospatial Memory Test—Revised (BVMT-R) Category Fluency MSCEIT—Managing Emotions CPT—Identical Pairs (CPT-IP)
±6.54 ±15.42 ±9.54 ±7.41 ±3.44 ±3.29 ±6.03 ±5.09 ±5.57 ±12.69 ±0.65
N/A 17.16 10.56 7.10 3.14 3.29 6.60 10.07 6.60 13.20 0.76
Adjusted RCI column represents the amount of raw score needed change for individual scores to indicate a meaningful change in cognition. BACS: Brief Assessment of Cognition in Schizophrenia. WMS-III: Wechsler Memory Scale—Third Edition. NAB: Neuropsychological Assessment Battery. MSCEIT: The Mayer–Salovey–Caruso Emotional Intelligence Test. CPT: Connors Continuous Performance Test.
subtests. The bootstrapping model was a marginal improvement compared to the observed model and offered little support for differential practice effects or other meaningful changes. 4. Discussion Differentiating between treatment effects and practice effects is essential to understanding the impact of antipsychotic medication on cognition. However, understanding the impact that practice effects have on cognition and cognitive improvement over time is a challenge for researchers in this area. One objective of this prospective study was to explore expectations for change due to practice effects by evaluating clinically stable schizophrenia patients on a stable medication regimen in the absence of behavioral or cognitive interventions. Furthermore, demographically similar psychiatrically healthy controls were used to anchor estimates of practice effects and meaningful change. The current findings indicate, at best, modest, but equivalent, practice effects for both groups on the MCCB, consistent with prior findings (Keefe et al., 2011). Specifically, both groups displayed similar rates of significant improvement on tests of visual memory and reasoning and problem solving, which is consistent with prior findings and near the same magnitude of improvement when compared to healthy controls also (Goldberg et al., 2007). These findings held true when bootstrapping simulated an increase in power. The present RCI findings were consistent with a prior study that based RCI estimates on normative tables from Nuechterlein et al. (2008) without the use of a demographically similar control group and use of RBNC anchored by demographic data (Goldberg et al., 2010) (Table 2). Overall, when compared to psychiatrically healthy controls, and in the absence of any behavioral, cognitive, or pharmaceutical intervention, schizophrenia patients displayed limited practice effects and there was no evidence of differential practice effects relative to healthy controls. Future studies may use these findings to differentiate genuine cognitive improvement from practice effects seen in healthy controls. The test–retest reliability estimates in the present patient sample were moderate to high and consistent with our previous report on this patient sample (Hill et al., 2013) and somewhat higher than the larger MCCB test selection and standardization report (Nuechterlein and Green, 2006; Nuechterlein et al., 2008). Prior research has suggested that high test–retest reliability on cognitive measures does not guarantee that the measure is guarded against practice effects (Goldberg et al., 2007). There are often unforeseen pragmatic issues in researching practice effects. Such as, practice effects differing across retest intervals (e.g., 4-week intervals compared to 12-week intervals), the use of alternative forms resulting in form effects rather than practice
191
effects, and the use of only one test–retest period rather than multiple retesting periods (Belinger et al., 2005; Goldberg et al., 2010). Thus, when studying practice effects, future research would benefit from using a demographically similar control sample, multiple retesting periods, clinically stable patients on consistent medication regimens, and at least one patient group with no behavioral treatment component. One limitation is the small sample size in the current study. Although the sample size was adequate for the statistical analyses and similar to prior results (Goldberg et al., 2007; Nuechterlein and Green, 2006; Nuechterlein et al., 2008), having a larger sample size may increase the likelihood of demonstrating differential practice effects between schizophrenia spectrum patients and psychiatrically healthy controls, should they exist and further estimating a Reliable Change Index. Another limitation of the present study is representativeness of the sample. A clinically stable sample, on a stable medication regimen, and not receiving behavioral or medical interventions were core features of the study design yet, these factors may also have limited the representativeness of the current sample. Additionally, the schizophrenia group was relatively high functioning, thus the present findings may not generalize to all schizophrenia patients, particularly those with more pronounced cognitive impairments, as practice effects for more pronounced cognitive impairments are likely to have a greater ceiling for improvement, thus demonstrating more cognitive improvement than a relatively stable patient sample. Several industry sponsored controlled trials and prior research on practice effects reported more global practice effects than observed presently (Keefe et al., 2006; Goldberg et al., 2007). One potential explanation for this discrepancy is that the current patient sample was not demographically different from psychiatrically healthy controls on age, education, premorbid intellectual functioning, or current intellectual functioning. Significant cognitive deficits in patients at baseline may confound findings of practice effects or differential practice effects at follow-up testing. That is, having a lower baseline performance allows for more room for improvement in patients than for the higher functioning healthy controls (particularly if they are performing at or near ceiling). The current study included schizophrenia spectrum patients on a stable antipsychotic regimen for one month prior the baseline assessment and throughout the study. This recruitment strategy limited certain confounds such as dosage changes or switching pharmaceutical treatment. Additionally, by assuring stable clinical status during the study, confounds associated with improved symptom control were also limited. Role of funding source This study was supported by funds received from NIH/NIMH (MH072767 & MH083888). The funding source played no role in data analysis or interpretation. Contributors Peter Weiden, MD: Subject recruitment services. Ellen Herbener, PhD: Clinical ratings and independent diagnostic evaluations. Conflict of interest There are no conflicts of interest to report. Acknowledgments This study was supported in part by NIMH grants MH072767 and MH083888.
References Andreasen, N.C., Pressler, M., Nopoulos, P., Miller, D., Ho, B.C., 2010. Antipsychotic dos equivalents and dose–years: a standardized method for comparing exposure to different drugs. Biol. Psychiatry 67, 255–262. Belinger, L.J., Gaydos, B., Tangphao-Daniels, O., Duff, K., Kareken, D.A., Siemers, E.R., 2005. Practice effects and the use of alternate forms in serial neuropsychological testing. Arch. Clin. Neuropsychol. 20, 517–529. Bilder, R.M., Goldman, R.S., Robinson, D., Reiter, G., Bell, L., Bates, J.A., Lieberman, J.A., 2000. Neuropsychology of first-episode schizophrenia: initial characterization and clinical correlates. Am. J. Psychiatry 157, 549–559.
192
J.E. Roseberry, S. Kristian Hill / Schizophrenia Research 159 (2014) 188–192
Blanchard, J.J., Neale, J.M., 1994. The neuropsychological signature of schizophrenia: generalized or differential deficit? Am. J. Psychiatry 151 (1), 40–48. Chelune, G.J., Naugle, R.I., Luders, H., Sedlak, J., Award, I.A., 1993. Individual change after epilepsy surgery: practice effects and base-rate information. Neuropsychology 7, 41–52. Goldberg, T.E., Goldman, R.S., Burdick, K.E., Malhotra, A.K., Lencz, T., Patel, R.C., Robinson, D.G., 2007. Cognitive improvement after treatment with second generation antipsychotic medications in first-episode schizophrenia, is it a practice effect? Arch. Gen. Psychiatry 64 (10), 1115–1122. Goldberg, T.E., Keefe, R.S.E., Goldman, R.S., Robinson, D.G., Harvey, P.D., 2010. Circumstances under which practice does not make perfect: a review of the practice effects literature in schizophrenia and its relevance to clinical treatment studies. Neuropsychopharmacology 1–10. Heinrichs, R.W., Zakzanis, K.K., 1998. Neurocognitive deficit in schizophrenia: a quantitative review of the evidence. Neuropsychology 12, 426–445. Heinrichs, R.W., Walter, A.A., Smith, D., Zargarian, T., Vaz, S., Ammari, N., 2008. Cognitive, clinical, and functional characteristics of verbally superior schizophrenia patients. Neuropsychology 22 (3), 321–328. Herman, B.P., Seidenberg, M., Schoenfeld, J., Peterson, J., Leveroni, C., Wyler, A.R., 1996. Empirical techniques for determining the reliability, magnitude, and pattern of neuropsychological change after epilepsy surgery. Epilepsia 37 (10), 942–950. Hill, S.K., Schuepback, D., Herbener, E.S., Keshavan, M.S., Sweeney, J.A., 2004. Pretreatment and longitudinal studies of neuropsychological deficits in antipsychotic-naive patients with schizophrenia. Schizophr. Res. 68, 49–63.
Hill, S.K., Bjorkquist, O., Carrathers, T., Roseberry, J.E., Hochberger, W., Bishop, J.R., 2013. Sequential processing deficits in schizophrenia: relationship to neuropsychology and genetics. Schizophr. Res. 151, 91–96. Jacobson, N.S., Truax, P., 1991. Clinical significance: a statistical approach to defining meaningful change in psychotherapy research. J. Consult. Clin. Psychol. 59 (1), 12–19. Keefe, R.S.E., Perkins, D.O., Gu, H., Zupirsky, R.B., Christensen, B.K., Lieberman, J.A., 2006. A longitudinal study of neurocognitive function in individuals at-risk for psychosis. Schizophr. Res. 88, 26–35. Keefe, R.S.E., Fox, K.H., Harvey, P.D., Cucchiaro, J., Siu, C., Loebel, A., 2011. Characteristics of the MATRICS Consensus Cognitive Battery in a 29-site antipsychotic schizophrenia clinical trial. Schizophr. Res. 125 (2–3), 161–168. McSweeny, A.J., Naugle, R.I., Chelune, G.J., Luders, H., 1993. T scores for change: an illustration of a regression approach to depicting change in clinical neuropsychology. Clin. Neuropsychol. 7 (3), 300–312. Nuechterlein, K.H., Green, M.F., 2006. MCCB: MATRICS Consensus Cognitive Battery Manual. MATRICS Assessment Inc., Los Angeles, CA. Nuechterlein, K.H., Green, M.F., Kern, R.S., Baade, L.E., Barch, D.M., Kraemer, H., 2008. The MATRICS consensus cognitive battery, part 1: test selection, reliability, and validity. Am. J. Psychiatry 165, 203–213. Salinsky, M.C., Storzbach, D., Dodrill, C.B., Binder, L.M., 2001. Test–retest bias, reliability, and regression equations for neuropsychological measures repeated over 12–16-week period. J. Int. Neuropsychol. Soc. 7, 597–605.
Contents lists available at ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
Limited practice effects and evaluation of expectation for change: MATRICS Consensus Cognitive Battery Jarett E. Roseberry ⁎, S. Kristian Hill Department of Psychology, Rosalind Franklin University of Medicine and Science, United States
a r t i c l e
i n f o
Article history: Received 14 March 2014 Received in revised form 9 August 2014 Accepted 12 August 2014 Available online 31 August 2014 Keywords: Schizophrenia Practice effects Cognitive abilities Neuropsychology
a b s t r a c t Understanding of the impact of antipsychotic medications on cognition requires differentiating between treatment effects and practice effects. This prospective study examines expectations for change on neuropsychological assessments and possible differential practice effects in community dwelling schizophrenia patients (n = 27) who are clinically stable and on a stable medication regimen when compared to demographically similar psychiatrically healthy controls (n = 29). All participants were administered the MATRICS Consensus Cognitive Battery (MCCB) twice over a period of four weeks. The use of Regression Based Norms for Change (RBNC) and Reliable Change Index (RCI) was completed to anchor estimates of meaningful change to a demographically similar control group. A repeated measures ANOVA was used to examine the effects of time and diagnosis on MCCB composite scores. A repeated measures MANOVA was used to examine the effects of time and diagnosis, and their interaction for MCCB subtests. Estimates of meaningful change are provided. A significant main effect was observed for time; no significant interactions were observed. There was no support for differential practice effects. In the absence of any behavioral, cognitive, or pharmaceutical interventions, these findings suggest limited change in performance over time in either group. © 2014 Elsevier B.V. All rights reserved.
1. Introduction
1.1. Cognition in schizophrenia
Differentiating between treatment effects and practice effects is essential to understanding the impact of antipsychotic medications and cognitive rehabilitation techniques on cognition. One challenge facing schizophrenia researchers is estimating the extent of practice effects (Goldberg et al., 2010). Few clinical trials have included healthy control groups to quantify the impact that repeated exposure to the testing environment. Further, it is impractical/unethical to delay study treatments to establish practice effects on cognitive outcome measures in untreated patients. Despite the potential confounds associated with practice effects, improved cognitive performance has often been attributed to pharmacological treatments. This has led some authors to contend that improved test performance may not differ from practice effects observed in healthy controls. One approach to addressing this issue is to assess practice effects with schizophrenia patients who are clinically stable and on a stable medication regimen compared to a demographically similar psychiatrically healthy control group tested at the same interval (Goldberg et al., 2010). By assessing change in this matter, one can better isolate practice effects from treatment effects in the context of clinical trials.
Cognitive impairments have been well documented in patients with schizophrenia (Blanchard and Neale, 1994; Bilder et al., 2000). These impairments are present across a wide range of cognitive domains, including executive function, visual attention, auditory attention, verbal memory, visual memory, motor skills, and visual perception (Heinrichs and Zakzanis, 1998; Hill et al., 2004), yet impairments can be heterogeneous in nature (Heinrichs et al., 2008). Level of severity typically ranges from moderate to severe (Heinrichs and Zakzanis, 1998), and these generalized cognitive impairments seem to play a critical role in functional outcome.
⁎ Corresponding author at: Rosalind Franklin University of Medicine and Science, Department of Psychology, 3333 Green Bay Road, North Chicago, IL 60064, United States. Tel.: +1 847 578 8749; fax: +1 847 578 8765. E-mail address: [email protected] (J.E. Roseberry).
http://dx.doi.org/10.1016/j.schres.2014.08.004 0920-9964/© 2014 Elsevier B.V. All rights reserved.
1.2. Practice effects When accounting for improved performance over time, medication effects and/or practice effects are two of the primary explanations in schizophrenia spectrum populations. When investigating specific cognitive domains, practice effects have been observed across a broad range of cognitive domains in the context of clinical trials and research reports within schizophrenia spectrum populations (Hill et al., 2004; Goldberg et al., 2007; Goldberg et al., 2010). Interestingly, minimal practice effects have been observed on the composite score of the MCCB (Keefe et al., 2011). One methodological challenge facing clinical trials evaluating either first generation antipsychotics (FGA) or second generation antipsychotics
J.E. Roseberry, S. Kristian Hill / Schizophrenia Research 159 (2014) 188–192
(SGA) has been the absence of a control group. By including a control group, one can quantify the impact that exposure to the testing environment has on improving performance during subsequent assessments. When nonpsychotic groups were compared to schizophrenia patients using SGA at similar testing intervals, the apparent cognitive benefit of SGA was consistent with practice effects observed in healthy populations (Goldberg et al., 2007). Thus, performance on cognitive measures may be artificially inflated at retest due to familiarity with the tests, increased comfort and confidence in test taking strategies, and other practice effects. Differentiating between treatment effects and practice effects can be challenging. However, this is an area that is critical to understanding the cognitive impact of antipsychotic treatments and cognitive rehabilitation interventions. One way to address the issue of differentiating between treatment effects and practice effects is to establish expectations for meaningful change on neuropsychological measures. By establishing expectation for meaningful change through assessing stable schizophrenics on a stable medication regimen with no behavioral or therapeutic interventions, with a demographically similar control group, one would be able to better estimate practice effects. A unique way to complete this is to use conservative statistical techniques, such as RBNC and RCI, to anchor retest scores to demographically similar controls to establish expectations for change. Prior studies assessing practice effects in schizophrenia are complicated by changes in medication status (clinical trials) or have failed to include controls to provide normative expectations for practice effects. Thus, this study was designed to prospectively examine practice effects in both clinically stable schizophrenia patients on stable medication regimens and demographically similar healthy controls. Further, to compliment that approach the present study also aims to provide estimates for meaningful change on the MCCB, based on RCI estimates anchored to the control group with RBNC statistics. This will provide future studies with a reference point for a range of expected variability on the MCCB. As with prior studies (Goldberg et al., 2007; Nuechterlein et al., 2008; Goldberg et al., 2010) practice effects were expected to be similar for both groups and no differential practice effects (i.e., one group experiencing greater practice effects) were predicted. 2. Methods 2.1. Participants Participants were 27 community dwelling individuals who met criteria for schizophrenia spectrum disorders and 29 psychiatrically healthy individuals based on Structured Clinical Interview for DSM-IV (SCID). Patients were referred by primary care physicians, who provided medication status and dosage for each patient. Healthy participants were recruited from the community via local advertisements and a research registry. To limit effects of both acute illness and recent changes to medication treatments, all patients were clinically stable, meaning there was no acute symptomatology, significant change in positive symptom severity, or changes in pharmacotherapy regimen for one month prior to and during the study. All patients were assessed with the Positive and Negative Syndrome Scale (PANSS) (Table 1), and medication status and dosage were recorded at each visit. Furthermore, no participant was excluded at follow-up assessment due to hospitalization, substance use/abuse or dependence. All patients were treated with either SGA (n = 24) or FGA (n = 3). Concomitant medications included selective serotonin reuptake inhibitors (SSRI) (n = 5), mood stabilizers (n = 3), lithium (n = 1), anticholinergics (n = 2), benzodiazepine (n = 2), and serotonin receptor partial agonist (n = 1). Correlations between overall neuropsychological scores and clinical symptom ratings, medication status, and Chlorpromazine equivalents (Andreasen et al., 2010) were small and nonsignificant. Thus, these measures were not included as covariates for group comparisons. All participants were free of substance abuse within the last three months,
189
Table 1 Participant demographic characteristics. Healthy controls: Patients:
Age (years): Sex: Male Female Education: Parental SES: WASI (2 subtest IQ): WRAT-III reading: Clinical data: Illness duration (years): PANSS total: PANSS positive: PANSS negative: Side effect ratings: AIMS total: ESRS total:
F/χ2
df
p
n = 29
n = 27
37.28 (11.32)
33.59 (9.34)
1.75
1.54
0.19
62.1% 37.9% 13.41 3.29 (1.15) 98.76 (10.96) 93.43 (11.09)
66.7% 33.3% 14.41 3.00 (0.98) 101.59 (13.78) 98.67 (13.66)
1.25
1
0.26
2.59 0.96 0.73 2.45
1.54 1.52 1.54 1.54
0.11 0.33 0.40 0.12
11.85 (10.78) 38.00 (6.57) 17.73 (4.06) 19.27 (5.16) 0.80 (1.32) 4.07 (4.37)
lifetime history of substance dependence, neurological disease, head injury with loss of consciousness greater than 10 min, and systemic disorders known to affect brain function. There were no group differences on education level, premorbid functioning, age, sex, and parental SES (Table 1). 2.2. Measures/procedures The Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS) initiative was an NIMH supported program for the development of a consensus cognitive battery for use in clinical trials. The MATRICS Neurocognition Committee evaluated more than 90 tests in seven cognitive domains to determine which tests were best suited for examining cognitive change in schizophrenia spectrum disorders. 20 tests were chosen for a beta battery, which were then administered to 176 schizophrenia spectrum patients, and readministered after a four-week interval (n = 167) to compare test–retest reliability among candidate tests (Nuechterlein et al., 2008). Ten tests across seven cognitive domains were selected to comprise the MCCB. In the present study all subjects were administered the MCCB twice over a period of four weeks. Alternative forms were used for the Brief Visuospatial Memory Test—Revised (BVMT-R), per MCCB administration suggestions. 2.3. Statistical analysis A repeated measures ANOVA was used to compute an omnibus test for differential practice effects among groups on the composite score of the MCCB. A repeated measures MANOVA was used for differential practice effects among groups or subtests of the MCCB, and to provide a detailed analysis of potential practice effects. Given the limited power of this small sample size, the probability of a Type II error is increased. To improve power and reduce the probability of a Type II error, a bootstrapping technique was used. For each subtest, estimates of meaningful change were computed using two conservative statistical approaches (RBNC and RCI). The RCI approach allows one to compute the amount of change necessary to reach significance (p b .05) by taking into account the test–retest correlation coefficient (Jacobson and Truax, 1991; Salinsky et al., 2001). RCI was compared to a 90% confidence interval, and scores falling outside that confidence interval would be considered meaningful (Jacobson and Truax, 1991). The RCI approach allows one to compute the amount of change necessary to reach significance (p b .05) by taking into account the test–retest correlation coefficient (Jacobson and Truax, 1991; Salinsky et al., 2001). The RBNC approach takes into account individual demographic characteristics which may influence retest scores,
190
J.E. Roseberry, S. Kristian Hill / Schizophrenia Research 159 (2014) 188–192
and regression based norms that utilize baseline test performance to predict retest scores (McSweeny et al., 1993). RBNC makes a conservative effort to correct several sources of measurement error, which may impact retest scores. Also, RBNC examines observed results versus expected test–retest changes. This approach enables one to estimate the relative magnitude of change across cognitive domains to assist in determining whether change scores are simply practice effects or meaningful change/cognitive improvement (Herman et al., 1996). For RBNC statistics, baseline scores for the healthy control group on age, education, and WRAT-III reading scores were entered into a regression analysis as predictors of retest scores. The mean differences between observed and predicted retest scores were computed as an indicator of performance change across tests. Bootstrapping techniques can be used to reduce the probability of a Type-II error by modeling a larger sample size. 3. Results 3.1. Repeated measures MANOVA Repeated measures ANOVA for the MCCB composite score revealed a significant main effect for time [F (1, 54) = 13.40, p = .001], (effect size: d = 0.21; ŋp2 = 0.20) indicating that all groups improved over time. There was no main effect of diagnosis [F (1, 54) = 0.47, p = 0.50], (effect size: d = 0.18; ŋp2 = 0.01). The time by diagnosis interaction was non-significant [F (1, 54) = 0.08, p = 0.78] (effect size: d = 0.15; ŋp2 = 0.002), thus there was no evidence of differential practice effects over time. A repeated measures MANOVA was computed to assess for differential practice effects among individual subtests of the
MCCB. This revealed a significant main effect of time [F (1, 54) = 9.47, p = 0.003] (effect size: d = 0.24; ŋp2 = 0.15). Additionally, there was a significant main effect of test [F (9, 46) = 9.21, p = .000] (effect size: d = 0.23; ŋp2 = 0.64) indicating variability in performance among MCCB subtests. None of the two- or three-way interactions involving time and diagnosis were significant, thus there was no evidence of differential practice effects. Overall, there was evidence of modest statistical improvement over time in both patient and control groups (see Fig. 1). 3.2. Regression Based Norms for Change and Reliable Change Index When using demographically similar controls, RBNC indicated that baseline performance on the MCCB was the primary predictor, and among demographic variables, only one subtest (BVMT-R) was significantly related to age (in controls). To be consistent with prior studies (Chelune et al., 1993), the RCI threshold for change was set at ± 1.64 standard deviations (90% confidence interval). Meaningful change estimated based on RCI computations for each subtest of the MCCB can be seen on Table 2 compared to a prior report (Goldberg et al., 2010). Within the sampled population, zero to six schizophrenia patients demonstrated meaningful change on any subtest, possibly indicating variable effort on baseline assessment or meaningful change in the absence of behavioral, therapeutic, or pharmacological intervention. 3.3. Bootstrapping In general, bootstrapping did not significantly improve the model in terms of estimating significant change over time for any individual
Patient and Healthy Control mean differences between Time 1 and Time 2 assessments and 90% RCI threshold needed for meaningful change 15
(14.27) (13.96)
12
(11.53)
(11.35)
(10.14)
(9.92)
(9.87)
(9.92)
(9.97)
(10.51)
9
T-Score
6 3 0 -3 Patients Controls
-6 -9
s ail
Tr
g R an Tdin Sp Co VL at. . H p m S Sy
S LN
ze Ma
s*
* y T -IP nc -R EI PT lue SC MT C F M . V t B Ca
MCCB Subtests Fig. 1. − − − represents 90th %tile RCI threshold (raw scores were converted to T-scores) needed for meaningful change. * represents significant change from Time 1 to Time 2 for both patients and control. No differential practice effects were observed; however both groups demonstrated similar rates of change on most measures. Although RCI estimates are designed to be used with individual data, the cut-off for meaningful change was added to each column for illustrative purposes. For symbol coding, no individuals exceeded the threshold. For Trail Making Test—A, and Letter—Number Span, one schizophrenia patient and one control exceeded the threshold per subtest. For HVLT-R, MSCEIT—Managing Emotions, and CPT-IP one schizophrenic patient and two healthy controls exceeded the threshold per subtest. For Spatial Span and Category Fluency, two schizophrenia patients and two healthy controls exceeded the threshold per subtest. For Mazes, four schizophrenia patients and four healthy controls exceeded the threshold per subtest. For BVMT-R, six schizophrenia patients and six healthy controls exceeded the threshold.
J.E. Roseberry, S. Kristian Hill / Schizophrenia Research 159 (2014) 188–192 Table 2 Raw score RCI values on the MCCB needed for meaningful change. Test
Adjusted RCI (90%)
Goldberg et al. (2010) RCI
MCCB composite score Trail Making Test—A BACS symbol coding Hopkins Verbal Learning Test—Revised (HVLT-R) WMS-III, Spatial Span University of Maryland Letter—Number Span NAB—Mazes Brief Visuospatial Memory Test—Revised (BVMT-R) Category Fluency MSCEIT—Managing Emotions CPT—Identical Pairs (CPT-IP)
±6.54 ±15.42 ±9.54 ±7.41 ±3.44 ±3.29 ±6.03 ±5.09 ±5.57 ±12.69 ±0.65
N/A 17.16 10.56 7.10 3.14 3.29 6.60 10.07 6.60 13.20 0.76
Adjusted RCI column represents the amount of raw score needed change for individual scores to indicate a meaningful change in cognition. BACS: Brief Assessment of Cognition in Schizophrenia. WMS-III: Wechsler Memory Scale—Third Edition. NAB: Neuropsychological Assessment Battery. MSCEIT: The Mayer–Salovey–Caruso Emotional Intelligence Test. CPT: Connors Continuous Performance Test.
subtests. The bootstrapping model was a marginal improvement compared to the observed model and offered little support for differential practice effects or other meaningful changes. 4. Discussion Differentiating between treatment effects and practice effects is essential to understanding the impact of antipsychotic medication on cognition. However, understanding the impact that practice effects have on cognition and cognitive improvement over time is a challenge for researchers in this area. One objective of this prospective study was to explore expectations for change due to practice effects by evaluating clinically stable schizophrenia patients on a stable medication regimen in the absence of behavioral or cognitive interventions. Furthermore, demographically similar psychiatrically healthy controls were used to anchor estimates of practice effects and meaningful change. The current findings indicate, at best, modest, but equivalent, practice effects for both groups on the MCCB, consistent with prior findings (Keefe et al., 2011). Specifically, both groups displayed similar rates of significant improvement on tests of visual memory and reasoning and problem solving, which is consistent with prior findings and near the same magnitude of improvement when compared to healthy controls also (Goldberg et al., 2007). These findings held true when bootstrapping simulated an increase in power. The present RCI findings were consistent with a prior study that based RCI estimates on normative tables from Nuechterlein et al. (2008) without the use of a demographically similar control group and use of RBNC anchored by demographic data (Goldberg et al., 2010) (Table 2). Overall, when compared to psychiatrically healthy controls, and in the absence of any behavioral, cognitive, or pharmaceutical intervention, schizophrenia patients displayed limited practice effects and there was no evidence of differential practice effects relative to healthy controls. Future studies may use these findings to differentiate genuine cognitive improvement from practice effects seen in healthy controls. The test–retest reliability estimates in the present patient sample were moderate to high and consistent with our previous report on this patient sample (Hill et al., 2013) and somewhat higher than the larger MCCB test selection and standardization report (Nuechterlein and Green, 2006; Nuechterlein et al., 2008). Prior research has suggested that high test–retest reliability on cognitive measures does not guarantee that the measure is guarded against practice effects (Goldberg et al., 2007). There are often unforeseen pragmatic issues in researching practice effects. Such as, practice effects differing across retest intervals (e.g., 4-week intervals compared to 12-week intervals), the use of alternative forms resulting in form effects rather than practice
191
effects, and the use of only one test–retest period rather than multiple retesting periods (Belinger et al., 2005; Goldberg et al., 2010). Thus, when studying practice effects, future research would benefit from using a demographically similar control sample, multiple retesting periods, clinically stable patients on consistent medication regimens, and at least one patient group with no behavioral treatment component. One limitation is the small sample size in the current study. Although the sample size was adequate for the statistical analyses and similar to prior results (Goldberg et al., 2007; Nuechterlein and Green, 2006; Nuechterlein et al., 2008), having a larger sample size may increase the likelihood of demonstrating differential practice effects between schizophrenia spectrum patients and psychiatrically healthy controls, should they exist and further estimating a Reliable Change Index. Another limitation of the present study is representativeness of the sample. A clinically stable sample, on a stable medication regimen, and not receiving behavioral or medical interventions were core features of the study design yet, these factors may also have limited the representativeness of the current sample. Additionally, the schizophrenia group was relatively high functioning, thus the present findings may not generalize to all schizophrenia patients, particularly those with more pronounced cognitive impairments, as practice effects for more pronounced cognitive impairments are likely to have a greater ceiling for improvement, thus demonstrating more cognitive improvement than a relatively stable patient sample. Several industry sponsored controlled trials and prior research on practice effects reported more global practice effects than observed presently (Keefe et al., 2006; Goldberg et al., 2007). One potential explanation for this discrepancy is that the current patient sample was not demographically different from psychiatrically healthy controls on age, education, premorbid intellectual functioning, or current intellectual functioning. Significant cognitive deficits in patients at baseline may confound findings of practice effects or differential practice effects at follow-up testing. That is, having a lower baseline performance allows for more room for improvement in patients than for the higher functioning healthy controls (particularly if they are performing at or near ceiling). The current study included schizophrenia spectrum patients on a stable antipsychotic regimen for one month prior the baseline assessment and throughout the study. This recruitment strategy limited certain confounds such as dosage changes or switching pharmaceutical treatment. Additionally, by assuring stable clinical status during the study, confounds associated with improved symptom control were also limited. Role of funding source This study was supported by funds received from NIH/NIMH (MH072767 & MH083888). The funding source played no role in data analysis or interpretation. Contributors Peter Weiden, MD: Subject recruitment services. Ellen Herbener, PhD: Clinical ratings and independent diagnostic evaluations. Conflict of interest There are no conflicts of interest to report. Acknowledgments This study was supported in part by NIMH grants MH072767 and MH083888.
References Andreasen, N.C., Pressler, M., Nopoulos, P., Miller, D., Ho, B.C., 2010. Antipsychotic dos equivalents and dose–years: a standardized method for comparing exposure to different drugs. Biol. Psychiatry 67, 255–262. Belinger, L.J., Gaydos, B., Tangphao-Daniels, O., Duff, K., Kareken, D.A., Siemers, E.R., 2005. Practice effects and the use of alternate forms in serial neuropsychological testing. Arch. Clin. Neuropsychol. 20, 517–529. Bilder, R.M., Goldman, R.S., Robinson, D., Reiter, G., Bell, L., Bates, J.A., Lieberman, J.A., 2000. Neuropsychology of first-episode schizophrenia: initial characterization and clinical correlates. Am. J. Psychiatry 157, 549–559.
192
J.E. Roseberry, S. Kristian Hill / Schizophrenia Research 159 (2014) 188–192
Blanchard, J.J., Neale, J.M., 1994. The neuropsychological signature of schizophrenia: generalized or differential deficit? Am. J. Psychiatry 151 (1), 40–48. Chelune, G.J., Naugle, R.I., Luders, H., Sedlak, J., Award, I.A., 1993. Individual change after epilepsy surgery: practice effects and base-rate information. Neuropsychology 7, 41–52. Goldberg, T.E., Goldman, R.S., Burdick, K.E., Malhotra, A.K., Lencz, T., Patel, R.C., Robinson, D.G., 2007. Cognitive improvement after treatment with second generation antipsychotic medications in first-episode schizophrenia, is it a practice effect? Arch. Gen. Psychiatry 64 (10), 1115–1122. Goldberg, T.E., Keefe, R.S.E., Goldman, R.S., Robinson, D.G., Harvey, P.D., 2010. Circumstances under which practice does not make perfect: a review of the practice effects literature in schizophrenia and its relevance to clinical treatment studies. Neuropsychopharmacology 1–10. Heinrichs, R.W., Zakzanis, K.K., 1998. Neurocognitive deficit in schizophrenia: a quantitative review of the evidence. Neuropsychology 12, 426–445. Heinrichs, R.W., Walter, A.A., Smith, D., Zargarian, T., Vaz, S., Ammari, N., 2008. Cognitive, clinical, and functional characteristics of verbally superior schizophrenia patients. Neuropsychology 22 (3), 321–328. Herman, B.P., Seidenberg, M., Schoenfeld, J., Peterson, J., Leveroni, C., Wyler, A.R., 1996. Empirical techniques for determining the reliability, magnitude, and pattern of neuropsychological change after epilepsy surgery. Epilepsia 37 (10), 942–950. Hill, S.K., Schuepback, D., Herbener, E.S., Keshavan, M.S., Sweeney, J.A., 2004. Pretreatment and longitudinal studies of neuropsychological deficits in antipsychotic-naive patients with schizophrenia. Schizophr. Res. 68, 49–63.
Hill, S.K., Bjorkquist, O., Carrathers, T., Roseberry, J.E., Hochberger, W., Bishop, J.R., 2013. Sequential processing deficits in schizophrenia: relationship to neuropsychology and genetics. Schizophr. Res. 151, 91–96. Jacobson, N.S., Truax, P., 1991. Clinical significance: a statistical approach to defining meaningful change in psychotherapy research. J. Consult. Clin. Psychol. 59 (1), 12–19. Keefe, R.S.E., Perkins, D.O., Gu, H., Zupirsky, R.B., Christensen, B.K., Lieberman, J.A., 2006. A longitudinal study of neurocognitive function in individuals at-risk for psychosis. Schizophr. Res. 88, 26–35. Keefe, R.S.E., Fox, K.H., Harvey, P.D., Cucchiaro, J., Siu, C., Loebel, A., 2011. Characteristics of the MATRICS Consensus Cognitive Battery in a 29-site antipsychotic schizophrenia clinical trial. Schizophr. Res. 125 (2–3), 161–168. McSweeny, A.J., Naugle, R.I., Chelune, G.J., Luders, H., 1993. T scores for change: an illustration of a regression approach to depicting change in clinical neuropsychology. Clin. Neuropsychol. 7 (3), 300–312. Nuechterlein, K.H., Green, M.F., 2006. MCCB: MATRICS Consensus Cognitive Battery Manual. MATRICS Assessment Inc., Los Angeles, CA. Nuechterlein, K.H., Green, M.F., Kern, R.S., Baade, L.E., Barch, D.M., Kraemer, H., 2008. The MATRICS consensus cognitive battery, part 1: test selection, reliability, and validity. Am. J. Psychiatry 165, 203–213. Salinsky, M.C., Storzbach, D., Dodrill, C.B., Binder, L.M., 2001. Test–retest bias, reliability, and regression equations for neuropsychological measures repeated over 12–16-week period. J. Int. Neuropsychol. Soc. 7, 597–605.