- Email: [email protected]
Generalization of training effects in schizophrenia
Schizophrenia Research 48 (2001) 255±262
www.elsevier.com/locate/schres
Generalization of training effects in schizophrenia Alan S. Bellack a,*, Lance S. Weinhardt b,1, James M. Gold b, Jean S. Gearon b a
VA Capitol Network MIRECC, University of Maryland School of Medicine, 737 West Lombard Street, Baltimore, MD 21201, USA b University of Maryland School of Medicine, Baltimore, MD 21201, USA Received 17 December 1999; accepted 25 February 2000
Abstract The primary goal of this study was to investigate transfer of training (generalization) in patients with schizophrenia. We randomly assigned 33 schizophrenia subjects to one of three conditions: training on the Wisconsin Card Sort Test (WCST-T), training on the Halstead Category Test (CAT-T), or no training (No-T). The WCST and CAT were administered to all subjects at baseline. Subjects in the WCST-T and CAT-T groups then received training on the respective test, while the No-T group received additional untrained trials. All participants were subsequently retested on the WCST and CAT, and completed a brief neuropsychological battery. As hypothesized, the WCST-T and CAT-T groups exhibited large improvements on the trained test and moderate improvement on the untrained test, while the No-T group failed to show improvement on either test. These results suggest that the training paradigm did produce generalization, and that the changes were not due to practice effects. The extent of generalization across both training groups was strongly associated with neuropsychological test performance (Spearman's rho 0.56, P , 0.05). The implications of these ®ndings for rehabilitation programs were discussed, and recommendations were made for future research. q 2001 Elsevier Science B.V. All rights reserved. Keywords: Generalization; Rehabilitation; Schizophrenia; Training effects
1. Introduction Neurocognitive de®cits are now widely recognized as a central feature of schizophrenia, and one of the primary factors contributing to the social impairment that is characteristic of the disorder (Bellack et al., 1999; Gold et al., 1999; Green, 1996, 1999). There is not complete consistency across studies in terms of which speci®c cognitive measures correlate with which speci®c measures of social-role functioning in the community, but the literature is consistent with the * Corresponding author. Tel.: 11-215-842-4550; fax: 11-215843-3341. E-mail address: [email protected] (A.S. Bellack). 1 Present address: Medical College of Wisconsin, Center for AIDS Intervention Research (CAIR), 2071N. Summit Ave., Milwaukee, WI 53202.
hypothesis that an impairment of basic cognitive functions may be an important determinant of both socialskill level and the ability to pro®t from skills-training interventions (Dickerson, et al., 1996; Green, 1996). There is strong evidence that many of these impairments are `trait-like', being largely stable over time despite ¯uctuations in symptomatic state, and there is reason to believe that certain cognitive impairments may act as `rate limiting' factors, compromising ability to bene®t from a variety of intervention approaches (Green, 1996). There is good evidence that people with schizophrenia can learn new skills in highly structured training settings, including social and problem-solving strategies that require higher level cognitive abilities (Bellack et al., 1999; Bellack and Mueser, 1993). However, there are few data to suggest that these newly acquired skills generalize to other
0920-9964/01/$ - see front matter q 2001 Elsevier Science B.V. All rights reserved. PII: S 0920-996 4(00)00066-9
256
A.S. Bellack et al. / Schizophrenia Research 48 (2001) 255±262
settings, and we know of no replicated studies to date documenting that the effects of cognitive skills training generalize. In fact, given the magnitude, breadth, and trait-like nature of cognitive impairments, it may be hypothesized that the ability to adaptively apply newly acquired cognitive abilities across situations is markedly limited (Bellack et al., 1999). A number of studies have examined this issue in the context of the Wisconsin Card Sorting Test (WCST), demonstrating that de®cits on the WCST can be reduced over short periods of time (Bellack et al., 1990; Green et al., 1992; Metz et al., 1994; Vollema et al., 1994; Young and Freyslinger, 1995). Two recent studies have examined the generalizability of training effects across problem-solving tasks that require similar cognitive operations, but that present different types of stimuli and operate according to different rules. Young and Freyslinger (1995) found that improvement in a WCST training protocol was associated with better performance on the Halstead Category Test, but their study did not include a baseline measure of Category Test performance so it is not certain that subjects improved as a function of being trained on the WCST. Bellack et al. (1996) conducted a similar study that examined transfer of training between the WCST and the Vygotsky Category Test (Wang, 1987). The WCST training was highly effective, but the data did not demonstrate cross-test generalization as the Vygotsky test proved to be comparatively easy and all groups exhibited some pre±post improvement. The current study was designed to circumvent the methodological problems with the Bellack et al. (1996) and Young and Freyslinger (1995) studies. Our primary hypotheses were: (1) that training would produce large improvements in performance on the trained test; and (2) that training effects would generalize (transfer) to the untrained test, but that performance on the generalization test would not improve as much as on the trained test.
2. Method 2.1. Subjects Subjects included 33 patients with schizophrenia and one with schizoaffective disorder, who did not
meet criteria for current abuse or dependence on alcohol or street drugs, and for whom there was no chart evidence of neurologic illness or injury. Diagnoses were determined by SCID interviews performed by experienced diagnosticians who had been trained to a criterion on a standardized set of video-tapes. Given the small sample size and the experience of the interviewers, reliability was not separately calculated for this trial. Seven subjects were stabilized inpatients on a long-term stay unit, nine were outpatients attending a day treatment program, and 17 were outpatients in a community psychiatry clinic (these subgroups did not differ on primary demographic measures or on WCST or CAT performance). There were 21 males and 12 females; 26 subjects were African American, six were Caucasian, and one was Asian American. The mean age was 40.91 years (S.D. 10.58) and subjects had completed an average of 12.1 years of education (S.D. 2.29). The mean duration of illness was 15.68 years (S.D. 9.21) and mean number of prior hospitalizations was 5.17 (S.D. 3.52). Half of the sample were receiving typical antipsychotics in a standard therapeutic range, and half were maintained on atypicals. 2.2. Measures 2.2.1. Wisconsin Card Sorting Test (WCST) (Heaton et al., 1993) This 128-item test is a widely used measure of problem solving and cognitive ¯exibility. It requires subjects to match stimulus cards containing different geometric shapes that vary in color and the number of items per card. The examiner does not inform the subject of the rule for correct matching (e.g., color, shape, or number of items on the card), but simply responds ªRightº or ªWrongº after each response. In addition, the rule is changed without warning after the subject has matched 10 consecutive cards. 2.2.2. Halstead Category Test (CAT) (Lezak, 1995) This test is widely considered as a measure of novel problem solving and hypothesis testing. It requires subjects to identify a principle for categorizing geometric shapes and symbols, and select the correct match from among a set of four choices. The test consists of a series of subtests, each of which employs different types of stimuli involving progressively
A.S. Bellack et al. / Schizophrenia Research 48 (2001) 255±262
more dif®cult categorization principles. The examiner responds ªCorrectº or ªIncorrectº after each answer. We used the 108-item short form of the Category Test, which excludes Subtests V and VI (Calsyn et al., 1980). 2.2.3. Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) The RBANS (Randolph, 1998) is a multifactorial 30-min neuropsychological screening battery. Gold and colleagues (Gold et al., 1999; Hobart et al., 1999) have demonstrated it to be reliable and valid for assessing cognitive status in schizophrenia. The battery yields a total scaled score as well as ®ve subscales: Immediate memory, Language, Visuospatial, Attention, and Delayed recall. 2.2.4. Wechsler Adult Intelligence Scale Ð Revised Intelligence was estimated based on age-scaled scores on the Block Design and Vocabulary subscales of the WAIS-R. 2.3. Procedure Eligible subjects who provided informed consent were randomly assigned to one of three groups: WCST Training (WCST-T), Category Test Training (CAT-T), or No Training (No-T). 2.3.1. Baseline assessment On the ®rst day of the protocol, subjects were ®rst tested on the WCST and CAT (in counterbalanced order). All subjects received both instruments with standard instructions. 2.3.2. Training Following a 15-min break, subjects assigned to the WCST-T and CAT-T groups received training on either the WCST or CAT. Detailed training manuals, developed in pilot research, were used to teach each test. Training was based on the procedure employed in our prior studies (Bellack et al., 1990, 1996). Each task was broken down into simple steps, brief didactic instruction was provided, the correct response was illustrated by the trainer, and the subject was required to explain the rule and demonstrate accurate responding before moving on to the next step. Systematic review and corrective feedback was built into the
257
Table 1 Problem-solving mnemonic Steps for solving problems 1. Figure out what your task or goal is 2. Figure out what your choices are 3. Develop a plan or strategy 4. Stick with your strategy if it is working 5. Change your strategy if it is not working 6. Remember your last guess and whether it was right or wrong
process. The CAT training protocol was developed to be a parallel to the WCST procedure. In addition, we presented the training in the context of a six-step problem-solving mnemonic (see Table 1) that could be applied to diverse problem-solving tasks. Subjects were not explicitly instructed to apply the mnemonic to the untrained task. However, they were told that it was a general strategy that could be used in other situations. Subjects assigned to the No-T group were randomly assigned to receive half of the items on each of the tests during the training phase to control for time spent working on the task in the training groups. 2.3.3. Retesting In this phase, administered on the following day, subjects in all three groups were retested on the WCST and CAT using standard instructions (i.e. no training). The order of administration for each subject was the same as that employed at baseline. Upon completion of those instruments, the RBANS and WAIS subtests were administered. 3. Results Two of 13 subjects assigned to the WCST-T group and four of 11 No-T subjects were dropped after the Baseline phase because they performed above normative levels on the WCST (®ve or more categories) or on the CAT (15 or fewer errors) at baseline, as it was assumed that they could not show a training effect. All other subjects performed below normative levels on both tests at baseline. One subject assigned to the WCST-T group did not complete the second day of testing and was not included in the data analyses. A series of ANOVAs and chi-squares (dichotomous
258
A.S. Bellack et al. / Schizophrenia Research 48 (2001) 255±262
Table 2 Baseline and post-training performance a WCST group (n 10) Mean (S.D.) CAT, No. of errors Baseline Post % Improvement WCST, No. correct Baseline Post % Improvement WCST, persev. errors Baseline Post % Improvement WCST, categories Baseline Post a
CAT group (n 9) t
Mean (S.D.)
2.72**
54.33 (12.82) 19.00 (14.50) 66.91 (21.64)
3.53***
59.11 (25.02) 60.78 (24.28) 5.02 (17.66)
47.10 (25.56) 31.67 (20.40) 30.80 (37.41) 1.20 (1.61) 4.50 (3.37)
53.50 (16.37) 48.00 (13.04) 7.11 (18.09) 52.50 (18.76) 78.90 (22.94) 66.89 (79.05)
Control group (n 7) t
Mean (S.D.)
t
61.86 (4.01) 57.71 (16.51) 7.52 (24.74)
0.79
0.67
63.43 (14.29) 48.57 (19.31) 2 25.35 (21.44)
2 3.14
2.35*
50.44 (30.62) 48.78 (30.79) 1.06 (23.84)
0.51
39.14 (11.80) 41.86 (11.49) 2 17.65 (50.17)
2 0.33
3.19**
1.44 (1.42) 1.89 (1.69)
1.84 ²
8.30****
2.43 (1.51) 1.14 (1.68)
²2.12
²
Signi®cance: P 0.052; *P , 0.05; **P , 0.01; ***P , 0.005; ****P , 0.0001.
variables) was conducted to compare the three groups on demographic characteristics, and baseline performance indices on the WCST, CAT, RBANS, and IQ. There were no signi®cant group differences on any of these variables. There was also no difference in the proportion of subjects in each group on typical versus atypical antipsychotics, in the proportion receiving anticholinergic medications, or in the proportion recruited from inpatient versus outpatient settings. The relationship among the neuropsychological measures at Baseline was examined with Spearman rank-order correlations (Spearman's are more robust than Pearson's with small samples). IQ was related to baseline performance on the CAT [rho (r ) 0.46, P , 0.01], but not the WCST (r 0.17, P 0.35). RBANS score was unrelated to baseline performance on either the CAT (r 0.23, P 0.23) or WCST (r 0.18, P 0.37). 3.1. Training A series of paired t tests was ®rst conducted for the Control group to examine the effects of practice. As indicated in Table 2, they did not evidence improvement on any measure across trials (their performance actually declined on several measures), and they were
not included in subsequent analyses. The effectiveness of each training protocol in improving performance on the trained test was examined with a parallel series of analyses. We opted for withingroup tests rather than an omnibus approach (i.e. an ANOVA) given the small sample size and the fact that we were primarily interested in within-group changes (i.e. did subjects learn and generalize, rather than which group learned more or generalized more). Both protocols were highly effective. As detailed in Table 2, subjects trained on the WCST had signi®cant increases in number of correct responses [t (9) 3.53, P , 0.005] and completed categories [t (9) 3.19, P , 0.01], and signi®cant decreases in number of perseverative errors [t (9) 2.35, P , 0.05]. Eight of the 10 subjects trained on the WCST improved by at least 10% on each measure. Subjects trained on the CAT had a signi®cant decrease in the number of errors (the primary dependent variable on the CAT) [t (8) 8.30 P , 0.0001]. All nine subjects trained on the CAT improved by at least 10%. Learning (i.e. the proportion improvement on the WCST number of items correct in the WCST-T group and proportion improvement on the CAT in the CATT group) was not signi®cantly correlated with either IQ (r 0.13, P . 0.50), RBANS total score
A.S. Bellack et al. / Schizophrenia Research 48 (2001) 255±262
(r 0.29, P 0.28), or any of the RBANS subscales: Immediate Memory (r 0.39, P 0.13), Visuospatial/Constructual (r 0.24, P 0.36), Attention P . 0.50), Language (r 0.04, (r 0.11, P . 0.50), and Delayed Memory (r 20.01, P . 0.50). However, conclusions about the RBANS should be tempered given the low power. There was no difference in the magnitude of learning between subjects on typical versus atypical antipsychotics [t (16) 1.46, P 0.16]. 3.2. Evidence for generalization effects Generalization would be expected only among subjects who exhibited a noticeable amount of improvement on the trained task. We employed an operational criterion of 10% improvement on the trained task as suf®cient evidence of learning to warrant examination of the extent of generalization. Two WCST-T subjects and no CAT-T subjects were excluded from the analyses below on the basis of this criterion. Six of the eight subjects who improved on the WCST had 10% or greater reduction in errors on the CAT; one other subject evidenced marginal improvement, and one exhibited no change. Two CAT-T subjects had meaningful increases in correct responses on the WCST (9.87% and 48.48%), three had marginal improvement, and three had more errors at post-test than at baseline. Improvement of WCST-T subjects on CAT, and CAT-T subjects on the WCST, was compared with the performance of No-T subjects on the respective task using Wilcoxon's rank sum test (i.e. Mann±Whitney U). The WCST-T group (n 8, median 11.6) improved signi®cantly more on the CAT than the No-T group (n 7, median 24.69): z 2.49, P 0.01. The CAT-T group (n 9, median 3.22) similarly evidenced greater improvement on the WCST (No. of correct responses) than the No-T group (n 7, median 223.6): z 1.76, P 0.05. Overall, there was considerable variability in the extent of generalization, and it appears that subjects were less able to generalize from the CAT to the WCST than vice versa. Of course, the WCST and CAT are not equated for dif®culty level, and it is not certain that a unit of change on one of the tests is equivalent to a unit of change on the other, so this ®nding should be interpreted with caution. There was no difference in the magnitude of generalization
259
between subjects on typical versus atypical antipsychotics [t (14) 0.69, P 0.5]. We hypothesized that there would be less generalization (i.e. cross-test improvement) than learning (within-test improvement). In order to examine this hypothesis we ®rst developed metrics for learning and generalization that were comparable across tests: a generalization score (GS) and a learning score (LS). GS equaled the proportion improvement on No. correct on the WCST for CAT-T subjects, and the proportion reduction in No. of errors on the CAT for the WCST-T subjects. LS was simply the proportion improvement on those scores within each test. GS and LS were compared for each training group with correlated t-tests for unequal sample variance. As predicted, LS was signi®cantly greater than GS for each training group: for WCST-T, t (9.94) 2.32, P , 0.02; for CAT-T, t (16) 6.64, P , 0.000. 3.3. Neuropsychological functioning The GS was also employed to examine the relationship of neuropsychological functioning to generalization. One subject in each training group did not complete the RBANS and they were not included in these analyses. The remaining subjects were merged into a single group (n 17 for correlations with IQ, and n 15 for correlations with RBANS). GS was signi®cantly correlated with total RBANS score (r 0.56, P 0.03) and with delayed memory (r 0.53, P 0.04), but failed to reach signi®cance with any other subscales or with IQ. Given the small sample size these ®ndings are susceptible to Type II error. In fact, while the correlations between generalization and the Language and Immediate Memory subscales were not statistically signi®cant at P , 0.05, the correlations represent moderate to large effect sizes (Cohen, 1988; p. 80). To further examine the potential signi®cance of these data, we determined the number of participants that would have been necessary to achieve statistical signi®cance at P , 0.05 with 80% power to avoid Type II error. This post-hoc power analysis revealed that n values of 31 and 43 would have been suf®cient to yield signi®cant results for the Language and Immediate Memory subscales, respectively. These are modest samples, and suggest that the ®ndings warrant further examination. In contrast, n values of 69 and 311 would have
260
A.S. Bellack et al. / Schizophrenia Research 48 (2001) 255±262
been necessary to achieve signi®cant results with the Attention and Visuospatial/Constructual scales, respectively. An n of 52 would have yielded signi®cance at the P , 0.05 level for IQ. 4. Discussion As expected, both training protocols were highly effective in improving performance on the respective tasks. Subjects trained on the CAT showed a large decrease in errors from pre-test to post-test, and subjects trained on the WCST exhibited signi®cant increases in the number of categories and correct responses, and decreases in perseverative errors on the WCST. These ®ndings are consistent with our own prior research on the WCST (Bellack et al., 1990), as well as a large number of other studies. Training effects on the CAT have not been reported previously, but this result extends a growing literature documenting that people with schizophrenia can improve their performance on a number of neuropsychological domains with modest instructional effort (Kern et al., 1995; Medalia et al., 1998). Several studies have demonstrated improvements on neuropsychological tests solely on the basis of extended practice (e.g., Stratta et al., 1994; Wexler et al., 1997). Our control group did not exhibit pre±post improvement on either test, suggesting that the changes in the training groups were not due to practice. Whether extended practice can produce equivalent gains to our brief training protocol is less relevant than the fact that these performance de®cits are malleable and can be reduced with a brief, simple training strategy. Performance on any neuropsychological test is an imperfect marker of the underlying neurocognitive capacity that the test purports to measure. Test performance is also affected by error, associated with factors such as motivation, effort, and test-taking experience. It remains to be determined if the changes seen in this and related studies are speci®c to the error component of test performance, or if they represent real change in the underlying neurocognitive domain. These ®ndings would not be of much theoretical or practical signi®cance if they simply demonstrated that patients can be motivated to be better at taking tests. This issue may best be addressed by studies that
examine neurometabolic or neurophysiological correlates of improvement. The durability and clinical signi®cance of these changes is also unclear and warrants further study. In regard to transfer of training, subjects trained on the WCST had a signi®cant pre±post reduction in errors on the CAT. As expected, the generalization effect was not as great as the training effect achieved by CAT-trained subjects. Similar, albeit more limited, results were found for the CAT-trained subjects. Subjects trained on the CAT had modest pre±post increases in WCST categories and correct responses, and decreases in perseverative errors. Changes in the WCST and CAT groups on the non-trained test were consistently better than in the No-T group. Overall, the data do demonstrate that a modest amount of transfer did occur, and they provide con®dence in rejecting the null hypothesis that people with schizophrenia are unable to learn generalizable skills. Several factors may actually have served to limit generalization in this analog study: training was quite brief, the generalization mnemonic was not validated and may not have been suf®ciently powerful, and the two tasks are not exact neurocognitive equivalents. It should also be underscored that our sample was relatively small and was composed of a large proportion of African American patients from an urban mental-health treatment facility. Thus, caution is advised in making extrapolations from our data; the ®ndings should be viewed as promising and suggest that future trials are warranted. The data underscore the wide inter-subject variability in both learning and generalization. While most people with schizophrenia perform poorly on the WCST, there are substantial individual differences and some 20% of patients have been shown to perform at normal levels (i.e. they achieve ®ve or more categories). Several studies have reported that variability in WCST performance is associated with broader differences in neurocognitive performance pro®les (Braff et al., 1991; Goldstein et al., 1996). The WCST training studies indicate that there is considerable variability in the extent to which subjects learn; our data illustrate that there is variability in the capacity for generalization as well: six of eight WCST-T subjects exhibited moderate or greater generalization to the CAT, while only two of eight CAT-T subjects exhibited such improvement on the WCST. These
A.S. Bellack et al. / Schizophrenia Research 48 (2001) 255±262
®ndings suggest that there may be important variations in the extent to which individual patients can learn and generalize, and therefore in the type of rehabilitation programming they require (Green, 1996). In a recent paper on the relationship between neuropsychological de®cits and functional outcomes, Green et al. (in press) proposed the use of dynamic assessment strategies that examine learning potential rather than static performance on neuropsychological tests. The learning and generalization paradigm employed here is an example of such a dynamic approach. The data provide preliminary information about the association of neuropsychological factors and generalization. There was a strong correlation between the overall RBANS score and generalization. The results also suggest that immediate and delayed memory and language ability may be important factors. These ®ndings must be viewed cautiously given the small sample size and the modest validity of individual RBANS scores, but they do suggest directions for future studies. It appears as if there may have been differential transfer from WCST training to the CAT versus from CAT training to the WCST. Such a difference could be due to a variety of factors, including differential effectiveness of the training paradigms and idiosyncrasies of the two tests. The instruments also use different metrics for scoring and a unit of change on one does not equal a unit of change on the other. The tests are also not equated for dif®culty level. Any directional effects of transfer of training may have both heuristic and clinical implications that extend beyond questions about speci®c instruments, and should be examined in future studies. Future research should also examine the predictive validity of our training paradigm vis-aÁ-vis performance in actual rehabilitation programs: i.e. does generalization in this experimental paradigm predict generalization from clinical programs? Acknowledgements This project was supported by the VA Capitol Network MIRECC (A.S.B., Director), and NIH grant DA09406 from NIDA to Dr Bellack.
261
References Bellack, A.S., Blanchard, J.J., Murphy, P., Podell, K., 1996. Generalization effects of training on the Wisconsin Card Sorting Test for schizophrenia patients. Schizophr. Res. 19, 189±194. Bellack, A.S., Gold, J.M., Buchanan, R.W., 1999. Cognitive rehabilitation for schizophrenia: problems, prospects, and strategies. Schizophr. Bull. 25, 257±274. Bellack, A.S., Mueser, K.T., 1993. Psychosocial treatment for schizophrenia. Schizophr. Bull. 19, 317±336. Bellack, A.S., Mueser, K.T., Morrison, R.L., Tierney, A., Podell, K., 1990. Remediation of cognitive de®cits in schizophrenia. Am. J. Psychiatry 147, 1650±1655. Braff, D.L., Heaton, R., Kuck, J., Cullum, M., Moranville, J., Grant, I., Zisook, S., 1991. The generalized pattern of neuropsychological de®cits in outpatients with chronic schizophrenia with heterogeneous Wisconsin Card Sorting Test results. Arch. Gen. Psychiatry 48, 891±898. Calsyn, D.A., O'Leary, M.R., Chaney, E.F., 1980. Shortening the category test. J. Consult. Clin. Psychol. 48, 788±789. Cohen, J., 1988. Statistical Power Analysis for the Behavioral Sciences. . second ed.Lawrence Erlbaum, Hillsdale, NJ. Dickerson, F., Boronow, J.J., Ringel, M., Parente, F., 1996. Neurcogonitive de®cits and social functioning in outpatients with schizophrenia. Schizophr. Res. 21, 75±83. Gold, J.M., Queern, C., Iannone, V.N., Buchanan, R.W., 1999. Repeatable battery for the Assessment of Neuropsychological Status as a screening test in schizophrenia, I: sensitivity, reliability, and validity. Am. J. Psychiatry 156, 1944±1950. Goldstein, G., Beers, S.R., Shemansky, W.J., 1996. Neuropsychological differences between schizophrenic patients with heterogeneous Wisconsin Card Sorting Test performance. Schizophr. Res. 21, 13±18. Green, M.F., 1996. What are the functional consequences of neurocognitive de®cits in schizophrenia? Am. J. Psychiatry 154, 321±330. Green, M.F., 1999. Interventions for neurocognitive de®cits Editor's introduction. Schizophr. Bull. 25, 197±200. Green, M.F., Kern, R.S., Braff, D.L., Mintz, J., 2000. Neurocognitive de®cits and functional outcome in schizophrenia: are we measuring the `right stuff'? Schizophr. Bull., 26 (1), 119±136. Green, M.F., Satz, P., Ganzell, S., Vaclav, J.F., 1992. Wisconsin Card Sorting Test performance in schizophrenia: remediation of a stubborn de®cit. Am. J. Psychiatry 149, 62±67. Heaton, R.K., Chelune, G.J., Talley, J.L., Kay, G.G., Curtiss, G., 1993. Wisconsin Card Sorting Test Manual. Psychological Assessment Resources, Odessa, FL. Hobart, M.P., Goldberg, R., Bartko, J.J., Gold, J.M., 1999. Repeatable battery for the Assessment of Neuropsychological Status as a screening test in schizophrenia, II: convergent/discriminant validity and diagnostic group comparisons. Am. J. Psychiatry 156, 1951±1957. Kern, R.S., Green, M.F., Goldstein, M.J., 1995. Modi®cation of performance on the span of apprehension, a putative marker of vulnerability to schizophrenia. J. Abnorm. Psychol. 104, 385±389. Lezak, M.D., 1995. Neuropsychological Assessment. third ed., Oxford University Press, New York.
262
A.S. Bellack et al. / Schizophrenia Research 48 (2001) 255±262
Medalia, A., Aluma, M., Tryon, W., Merriam, A.E., 1998. Effectiveness of attention training in schizophrenia. Schizophr. Bull. 24, 147±152. Metz, J.T., Johnson, M.D., Pliskin, N.H., Luchins, D.J., 1994. Maintenance of training effects on the Wisconsin Card Sorting Test by patients with schizophrenia or affective disorders. Am. J. Psychiatry 151, 120±122. Randolph, C., 1998. Repeatable Battery for the Assessment of Neurpsychological Status (RBANS). Psychological Corporation, San Antonio. Stratta, P., Mancini, F., Mattei, P., Casacchia, M., Rossi, A., 1994. Information processing strategy to remediate Wisconsin Card Sorting Test performance in schizophrenia: a pilot study. Am. J. Psychiatry 151, 915±918.
Vollema, M.G., Geurtsen, G.J., van Voorst, A.J. P., 1994. Durable improvements in Wisconsin Card Sorting Test performance in schizophrenic patients. Schizophr. Res. 16, 209±215. Wang, P.L., 1987. The search for a clinical frontal lobe test. In: Perecman, E. (Ed.), Concept Formation and Frontal Lobe Function. IRBN Press, New York, pp. 189±205. Wexler, B.E., Hawkins, K.A., Rounsaville, B., Anderson, M., Sernyak, M.J., Green, M.F., 1997. Normal neurocognitive performance after extended practice in patients with schizophrenia. Schizophr. Res. 26, 173±180. Young, D.A., Freyslinger, M.G., 1995. Scaffolded instruction and the remediation of Wisconsin Card Sorting Test de®cits in chronic schizophrenia. Schizophr. Res. 16, 199±207.
www.elsevier.com/locate/schres
Generalization of training effects in schizophrenia Alan S. Bellack a,*, Lance S. Weinhardt b,1, James M. Gold b, Jean S. Gearon b a
VA Capitol Network MIRECC, University of Maryland School of Medicine, 737 West Lombard Street, Baltimore, MD 21201, USA b University of Maryland School of Medicine, Baltimore, MD 21201, USA Received 17 December 1999; accepted 25 February 2000
Abstract The primary goal of this study was to investigate transfer of training (generalization) in patients with schizophrenia. We randomly assigned 33 schizophrenia subjects to one of three conditions: training on the Wisconsin Card Sort Test (WCST-T), training on the Halstead Category Test (CAT-T), or no training (No-T). The WCST and CAT were administered to all subjects at baseline. Subjects in the WCST-T and CAT-T groups then received training on the respective test, while the No-T group received additional untrained trials. All participants were subsequently retested on the WCST and CAT, and completed a brief neuropsychological battery. As hypothesized, the WCST-T and CAT-T groups exhibited large improvements on the trained test and moderate improvement on the untrained test, while the No-T group failed to show improvement on either test. These results suggest that the training paradigm did produce generalization, and that the changes were not due to practice effects. The extent of generalization across both training groups was strongly associated with neuropsychological test performance (Spearman's rho 0.56, P , 0.05). The implications of these ®ndings for rehabilitation programs were discussed, and recommendations were made for future research. q 2001 Elsevier Science B.V. All rights reserved. Keywords: Generalization; Rehabilitation; Schizophrenia; Training effects
1. Introduction Neurocognitive de®cits are now widely recognized as a central feature of schizophrenia, and one of the primary factors contributing to the social impairment that is characteristic of the disorder (Bellack et al., 1999; Gold et al., 1999; Green, 1996, 1999). There is not complete consistency across studies in terms of which speci®c cognitive measures correlate with which speci®c measures of social-role functioning in the community, but the literature is consistent with the * Corresponding author. Tel.: 11-215-842-4550; fax: 11-215843-3341. E-mail address: [email protected] (A.S. Bellack). 1 Present address: Medical College of Wisconsin, Center for AIDS Intervention Research (CAIR), 2071N. Summit Ave., Milwaukee, WI 53202.
hypothesis that an impairment of basic cognitive functions may be an important determinant of both socialskill level and the ability to pro®t from skills-training interventions (Dickerson, et al., 1996; Green, 1996). There is strong evidence that many of these impairments are `trait-like', being largely stable over time despite ¯uctuations in symptomatic state, and there is reason to believe that certain cognitive impairments may act as `rate limiting' factors, compromising ability to bene®t from a variety of intervention approaches (Green, 1996). There is good evidence that people with schizophrenia can learn new skills in highly structured training settings, including social and problem-solving strategies that require higher level cognitive abilities (Bellack et al., 1999; Bellack and Mueser, 1993). However, there are few data to suggest that these newly acquired skills generalize to other
0920-9964/01/$ - see front matter q 2001 Elsevier Science B.V. All rights reserved. PII: S 0920-996 4(00)00066-9
256
A.S. Bellack et al. / Schizophrenia Research 48 (2001) 255±262
settings, and we know of no replicated studies to date documenting that the effects of cognitive skills training generalize. In fact, given the magnitude, breadth, and trait-like nature of cognitive impairments, it may be hypothesized that the ability to adaptively apply newly acquired cognitive abilities across situations is markedly limited (Bellack et al., 1999). A number of studies have examined this issue in the context of the Wisconsin Card Sorting Test (WCST), demonstrating that de®cits on the WCST can be reduced over short periods of time (Bellack et al., 1990; Green et al., 1992; Metz et al., 1994; Vollema et al., 1994; Young and Freyslinger, 1995). Two recent studies have examined the generalizability of training effects across problem-solving tasks that require similar cognitive operations, but that present different types of stimuli and operate according to different rules. Young and Freyslinger (1995) found that improvement in a WCST training protocol was associated with better performance on the Halstead Category Test, but their study did not include a baseline measure of Category Test performance so it is not certain that subjects improved as a function of being trained on the WCST. Bellack et al. (1996) conducted a similar study that examined transfer of training between the WCST and the Vygotsky Category Test (Wang, 1987). The WCST training was highly effective, but the data did not demonstrate cross-test generalization as the Vygotsky test proved to be comparatively easy and all groups exhibited some pre±post improvement. The current study was designed to circumvent the methodological problems with the Bellack et al. (1996) and Young and Freyslinger (1995) studies. Our primary hypotheses were: (1) that training would produce large improvements in performance on the trained test; and (2) that training effects would generalize (transfer) to the untrained test, but that performance on the generalization test would not improve as much as on the trained test.
2. Method 2.1. Subjects Subjects included 33 patients with schizophrenia and one with schizoaffective disorder, who did not
meet criteria for current abuse or dependence on alcohol or street drugs, and for whom there was no chart evidence of neurologic illness or injury. Diagnoses were determined by SCID interviews performed by experienced diagnosticians who had been trained to a criterion on a standardized set of video-tapes. Given the small sample size and the experience of the interviewers, reliability was not separately calculated for this trial. Seven subjects were stabilized inpatients on a long-term stay unit, nine were outpatients attending a day treatment program, and 17 were outpatients in a community psychiatry clinic (these subgroups did not differ on primary demographic measures or on WCST or CAT performance). There were 21 males and 12 females; 26 subjects were African American, six were Caucasian, and one was Asian American. The mean age was 40.91 years (S.D. 10.58) and subjects had completed an average of 12.1 years of education (S.D. 2.29). The mean duration of illness was 15.68 years (S.D. 9.21) and mean number of prior hospitalizations was 5.17 (S.D. 3.52). Half of the sample were receiving typical antipsychotics in a standard therapeutic range, and half were maintained on atypicals. 2.2. Measures 2.2.1. Wisconsin Card Sorting Test (WCST) (Heaton et al., 1993) This 128-item test is a widely used measure of problem solving and cognitive ¯exibility. It requires subjects to match stimulus cards containing different geometric shapes that vary in color and the number of items per card. The examiner does not inform the subject of the rule for correct matching (e.g., color, shape, or number of items on the card), but simply responds ªRightº or ªWrongº after each response. In addition, the rule is changed without warning after the subject has matched 10 consecutive cards. 2.2.2. Halstead Category Test (CAT) (Lezak, 1995) This test is widely considered as a measure of novel problem solving and hypothesis testing. It requires subjects to identify a principle for categorizing geometric shapes and symbols, and select the correct match from among a set of four choices. The test consists of a series of subtests, each of which employs different types of stimuli involving progressively
A.S. Bellack et al. / Schizophrenia Research 48 (2001) 255±262
more dif®cult categorization principles. The examiner responds ªCorrectº or ªIncorrectº after each answer. We used the 108-item short form of the Category Test, which excludes Subtests V and VI (Calsyn et al., 1980). 2.2.3. Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) The RBANS (Randolph, 1998) is a multifactorial 30-min neuropsychological screening battery. Gold and colleagues (Gold et al., 1999; Hobart et al., 1999) have demonstrated it to be reliable and valid for assessing cognitive status in schizophrenia. The battery yields a total scaled score as well as ®ve subscales: Immediate memory, Language, Visuospatial, Attention, and Delayed recall. 2.2.4. Wechsler Adult Intelligence Scale Ð Revised Intelligence was estimated based on age-scaled scores on the Block Design and Vocabulary subscales of the WAIS-R. 2.3. Procedure Eligible subjects who provided informed consent were randomly assigned to one of three groups: WCST Training (WCST-T), Category Test Training (CAT-T), or No Training (No-T). 2.3.1. Baseline assessment On the ®rst day of the protocol, subjects were ®rst tested on the WCST and CAT (in counterbalanced order). All subjects received both instruments with standard instructions. 2.3.2. Training Following a 15-min break, subjects assigned to the WCST-T and CAT-T groups received training on either the WCST or CAT. Detailed training manuals, developed in pilot research, were used to teach each test. Training was based on the procedure employed in our prior studies (Bellack et al., 1990, 1996). Each task was broken down into simple steps, brief didactic instruction was provided, the correct response was illustrated by the trainer, and the subject was required to explain the rule and demonstrate accurate responding before moving on to the next step. Systematic review and corrective feedback was built into the
257
Table 1 Problem-solving mnemonic Steps for solving problems 1. Figure out what your task or goal is 2. Figure out what your choices are 3. Develop a plan or strategy 4. Stick with your strategy if it is working 5. Change your strategy if it is not working 6. Remember your last guess and whether it was right or wrong
process. The CAT training protocol was developed to be a parallel to the WCST procedure. In addition, we presented the training in the context of a six-step problem-solving mnemonic (see Table 1) that could be applied to diverse problem-solving tasks. Subjects were not explicitly instructed to apply the mnemonic to the untrained task. However, they were told that it was a general strategy that could be used in other situations. Subjects assigned to the No-T group were randomly assigned to receive half of the items on each of the tests during the training phase to control for time spent working on the task in the training groups. 2.3.3. Retesting In this phase, administered on the following day, subjects in all three groups were retested on the WCST and CAT using standard instructions (i.e. no training). The order of administration for each subject was the same as that employed at baseline. Upon completion of those instruments, the RBANS and WAIS subtests were administered. 3. Results Two of 13 subjects assigned to the WCST-T group and four of 11 No-T subjects were dropped after the Baseline phase because they performed above normative levels on the WCST (®ve or more categories) or on the CAT (15 or fewer errors) at baseline, as it was assumed that they could not show a training effect. All other subjects performed below normative levels on both tests at baseline. One subject assigned to the WCST-T group did not complete the second day of testing and was not included in the data analyses. A series of ANOVAs and chi-squares (dichotomous
258
A.S. Bellack et al. / Schizophrenia Research 48 (2001) 255±262
Table 2 Baseline and post-training performance a WCST group (n 10) Mean (S.D.) CAT, No. of errors Baseline Post % Improvement WCST, No. correct Baseline Post % Improvement WCST, persev. errors Baseline Post % Improvement WCST, categories Baseline Post a
CAT group (n 9) t
Mean (S.D.)
2.72**
54.33 (12.82) 19.00 (14.50) 66.91 (21.64)
3.53***
59.11 (25.02) 60.78 (24.28) 5.02 (17.66)
47.10 (25.56) 31.67 (20.40) 30.80 (37.41) 1.20 (1.61) 4.50 (3.37)
53.50 (16.37) 48.00 (13.04) 7.11 (18.09) 52.50 (18.76) 78.90 (22.94) 66.89 (79.05)
Control group (n 7) t
Mean (S.D.)
t
61.86 (4.01) 57.71 (16.51) 7.52 (24.74)
0.79
0.67
63.43 (14.29) 48.57 (19.31) 2 25.35 (21.44)
2 3.14
2.35*
50.44 (30.62) 48.78 (30.79) 1.06 (23.84)
0.51
39.14 (11.80) 41.86 (11.49) 2 17.65 (50.17)
2 0.33
3.19**
1.44 (1.42) 1.89 (1.69)
1.84 ²
8.30****
2.43 (1.51) 1.14 (1.68)
²2.12
²
Signi®cance: P 0.052; *P , 0.05; **P , 0.01; ***P , 0.005; ****P , 0.0001.
variables) was conducted to compare the three groups on demographic characteristics, and baseline performance indices on the WCST, CAT, RBANS, and IQ. There were no signi®cant group differences on any of these variables. There was also no difference in the proportion of subjects in each group on typical versus atypical antipsychotics, in the proportion receiving anticholinergic medications, or in the proportion recruited from inpatient versus outpatient settings. The relationship among the neuropsychological measures at Baseline was examined with Spearman rank-order correlations (Spearman's are more robust than Pearson's with small samples). IQ was related to baseline performance on the CAT [rho (r ) 0.46, P , 0.01], but not the WCST (r 0.17, P 0.35). RBANS score was unrelated to baseline performance on either the CAT (r 0.23, P 0.23) or WCST (r 0.18, P 0.37). 3.1. Training A series of paired t tests was ®rst conducted for the Control group to examine the effects of practice. As indicated in Table 2, they did not evidence improvement on any measure across trials (their performance actually declined on several measures), and they were
not included in subsequent analyses. The effectiveness of each training protocol in improving performance on the trained test was examined with a parallel series of analyses. We opted for withingroup tests rather than an omnibus approach (i.e. an ANOVA) given the small sample size and the fact that we were primarily interested in within-group changes (i.e. did subjects learn and generalize, rather than which group learned more or generalized more). Both protocols were highly effective. As detailed in Table 2, subjects trained on the WCST had signi®cant increases in number of correct responses [t (9) 3.53, P , 0.005] and completed categories [t (9) 3.19, P , 0.01], and signi®cant decreases in number of perseverative errors [t (9) 2.35, P , 0.05]. Eight of the 10 subjects trained on the WCST improved by at least 10% on each measure. Subjects trained on the CAT had a signi®cant decrease in the number of errors (the primary dependent variable on the CAT) [t (8) 8.30 P , 0.0001]. All nine subjects trained on the CAT improved by at least 10%. Learning (i.e. the proportion improvement on the WCST number of items correct in the WCST-T group and proportion improvement on the CAT in the CATT group) was not signi®cantly correlated with either IQ (r 0.13, P . 0.50), RBANS total score
A.S. Bellack et al. / Schizophrenia Research 48 (2001) 255±262
(r 0.29, P 0.28), or any of the RBANS subscales: Immediate Memory (r 0.39, P 0.13), Visuospatial/Constructual (r 0.24, P 0.36), Attention P . 0.50), Language (r 0.04, (r 0.11, P . 0.50), and Delayed Memory (r 20.01, P . 0.50). However, conclusions about the RBANS should be tempered given the low power. There was no difference in the magnitude of learning between subjects on typical versus atypical antipsychotics [t (16) 1.46, P 0.16]. 3.2. Evidence for generalization effects Generalization would be expected only among subjects who exhibited a noticeable amount of improvement on the trained task. We employed an operational criterion of 10% improvement on the trained task as suf®cient evidence of learning to warrant examination of the extent of generalization. Two WCST-T subjects and no CAT-T subjects were excluded from the analyses below on the basis of this criterion. Six of the eight subjects who improved on the WCST had 10% or greater reduction in errors on the CAT; one other subject evidenced marginal improvement, and one exhibited no change. Two CAT-T subjects had meaningful increases in correct responses on the WCST (9.87% and 48.48%), three had marginal improvement, and three had more errors at post-test than at baseline. Improvement of WCST-T subjects on CAT, and CAT-T subjects on the WCST, was compared with the performance of No-T subjects on the respective task using Wilcoxon's rank sum test (i.e. Mann±Whitney U). The WCST-T group (n 8, median 11.6) improved signi®cantly more on the CAT than the No-T group (n 7, median 24.69): z 2.49, P 0.01. The CAT-T group (n 9, median 3.22) similarly evidenced greater improvement on the WCST (No. of correct responses) than the No-T group (n 7, median 223.6): z 1.76, P 0.05. Overall, there was considerable variability in the extent of generalization, and it appears that subjects were less able to generalize from the CAT to the WCST than vice versa. Of course, the WCST and CAT are not equated for dif®culty level, and it is not certain that a unit of change on one of the tests is equivalent to a unit of change on the other, so this ®nding should be interpreted with caution. There was no difference in the magnitude of generalization
259
between subjects on typical versus atypical antipsychotics [t (14) 0.69, P 0.5]. We hypothesized that there would be less generalization (i.e. cross-test improvement) than learning (within-test improvement). In order to examine this hypothesis we ®rst developed metrics for learning and generalization that were comparable across tests: a generalization score (GS) and a learning score (LS). GS equaled the proportion improvement on No. correct on the WCST for CAT-T subjects, and the proportion reduction in No. of errors on the CAT for the WCST-T subjects. LS was simply the proportion improvement on those scores within each test. GS and LS were compared for each training group with correlated t-tests for unequal sample variance. As predicted, LS was signi®cantly greater than GS for each training group: for WCST-T, t (9.94) 2.32, P , 0.02; for CAT-T, t (16) 6.64, P , 0.000. 3.3. Neuropsychological functioning The GS was also employed to examine the relationship of neuropsychological functioning to generalization. One subject in each training group did not complete the RBANS and they were not included in these analyses. The remaining subjects were merged into a single group (n 17 for correlations with IQ, and n 15 for correlations with RBANS). GS was signi®cantly correlated with total RBANS score (r 0.56, P 0.03) and with delayed memory (r 0.53, P 0.04), but failed to reach signi®cance with any other subscales or with IQ. Given the small sample size these ®ndings are susceptible to Type II error. In fact, while the correlations between generalization and the Language and Immediate Memory subscales were not statistically signi®cant at P , 0.05, the correlations represent moderate to large effect sizes (Cohen, 1988; p. 80). To further examine the potential signi®cance of these data, we determined the number of participants that would have been necessary to achieve statistical signi®cance at P , 0.05 with 80% power to avoid Type II error. This post-hoc power analysis revealed that n values of 31 and 43 would have been suf®cient to yield signi®cant results for the Language and Immediate Memory subscales, respectively. These are modest samples, and suggest that the ®ndings warrant further examination. In contrast, n values of 69 and 311 would have
260
A.S. Bellack et al. / Schizophrenia Research 48 (2001) 255±262
been necessary to achieve signi®cant results with the Attention and Visuospatial/Constructual scales, respectively. An n of 52 would have yielded signi®cance at the P , 0.05 level for IQ. 4. Discussion As expected, both training protocols were highly effective in improving performance on the respective tasks. Subjects trained on the CAT showed a large decrease in errors from pre-test to post-test, and subjects trained on the WCST exhibited signi®cant increases in the number of categories and correct responses, and decreases in perseverative errors on the WCST. These ®ndings are consistent with our own prior research on the WCST (Bellack et al., 1990), as well as a large number of other studies. Training effects on the CAT have not been reported previously, but this result extends a growing literature documenting that people with schizophrenia can improve their performance on a number of neuropsychological domains with modest instructional effort (Kern et al., 1995; Medalia et al., 1998). Several studies have demonstrated improvements on neuropsychological tests solely on the basis of extended practice (e.g., Stratta et al., 1994; Wexler et al., 1997). Our control group did not exhibit pre±post improvement on either test, suggesting that the changes in the training groups were not due to practice. Whether extended practice can produce equivalent gains to our brief training protocol is less relevant than the fact that these performance de®cits are malleable and can be reduced with a brief, simple training strategy. Performance on any neuropsychological test is an imperfect marker of the underlying neurocognitive capacity that the test purports to measure. Test performance is also affected by error, associated with factors such as motivation, effort, and test-taking experience. It remains to be determined if the changes seen in this and related studies are speci®c to the error component of test performance, or if they represent real change in the underlying neurocognitive domain. These ®ndings would not be of much theoretical or practical signi®cance if they simply demonstrated that patients can be motivated to be better at taking tests. This issue may best be addressed by studies that
examine neurometabolic or neurophysiological correlates of improvement. The durability and clinical signi®cance of these changes is also unclear and warrants further study. In regard to transfer of training, subjects trained on the WCST had a signi®cant pre±post reduction in errors on the CAT. As expected, the generalization effect was not as great as the training effect achieved by CAT-trained subjects. Similar, albeit more limited, results were found for the CAT-trained subjects. Subjects trained on the CAT had modest pre±post increases in WCST categories and correct responses, and decreases in perseverative errors. Changes in the WCST and CAT groups on the non-trained test were consistently better than in the No-T group. Overall, the data do demonstrate that a modest amount of transfer did occur, and they provide con®dence in rejecting the null hypothesis that people with schizophrenia are unable to learn generalizable skills. Several factors may actually have served to limit generalization in this analog study: training was quite brief, the generalization mnemonic was not validated and may not have been suf®ciently powerful, and the two tasks are not exact neurocognitive equivalents. It should also be underscored that our sample was relatively small and was composed of a large proportion of African American patients from an urban mental-health treatment facility. Thus, caution is advised in making extrapolations from our data; the ®ndings should be viewed as promising and suggest that future trials are warranted. The data underscore the wide inter-subject variability in both learning and generalization. While most people with schizophrenia perform poorly on the WCST, there are substantial individual differences and some 20% of patients have been shown to perform at normal levels (i.e. they achieve ®ve or more categories). Several studies have reported that variability in WCST performance is associated with broader differences in neurocognitive performance pro®les (Braff et al., 1991; Goldstein et al., 1996). The WCST training studies indicate that there is considerable variability in the extent to which subjects learn; our data illustrate that there is variability in the capacity for generalization as well: six of eight WCST-T subjects exhibited moderate or greater generalization to the CAT, while only two of eight CAT-T subjects exhibited such improvement on the WCST. These
A.S. Bellack et al. / Schizophrenia Research 48 (2001) 255±262
®ndings suggest that there may be important variations in the extent to which individual patients can learn and generalize, and therefore in the type of rehabilitation programming they require (Green, 1996). In a recent paper on the relationship between neuropsychological de®cits and functional outcomes, Green et al. (in press) proposed the use of dynamic assessment strategies that examine learning potential rather than static performance on neuropsychological tests. The learning and generalization paradigm employed here is an example of such a dynamic approach. The data provide preliminary information about the association of neuropsychological factors and generalization. There was a strong correlation between the overall RBANS score and generalization. The results also suggest that immediate and delayed memory and language ability may be important factors. These ®ndings must be viewed cautiously given the small sample size and the modest validity of individual RBANS scores, but they do suggest directions for future studies. It appears as if there may have been differential transfer from WCST training to the CAT versus from CAT training to the WCST. Such a difference could be due to a variety of factors, including differential effectiveness of the training paradigms and idiosyncrasies of the two tests. The instruments also use different metrics for scoring and a unit of change on one does not equal a unit of change on the other. The tests are also not equated for dif®culty level. Any directional effects of transfer of training may have both heuristic and clinical implications that extend beyond questions about speci®c instruments, and should be examined in future studies. Future research should also examine the predictive validity of our training paradigm vis-aÁ-vis performance in actual rehabilitation programs: i.e. does generalization in this experimental paradigm predict generalization from clinical programs? Acknowledgements This project was supported by the VA Capitol Network MIRECC (A.S.B., Director), and NIH grant DA09406 from NIDA to Dr Bellack.
261
References Bellack, A.S., Blanchard, J.J., Murphy, P., Podell, K., 1996. Generalization effects of training on the Wisconsin Card Sorting Test for schizophrenia patients. Schizophr. Res. 19, 189±194. Bellack, A.S., Gold, J.M., Buchanan, R.W., 1999. Cognitive rehabilitation for schizophrenia: problems, prospects, and strategies. Schizophr. Bull. 25, 257±274. Bellack, A.S., Mueser, K.T., 1993. Psychosocial treatment for schizophrenia. Schizophr. Bull. 19, 317±336. Bellack, A.S., Mueser, K.T., Morrison, R.L., Tierney, A., Podell, K., 1990. Remediation of cognitive de®cits in schizophrenia. Am. J. Psychiatry 147, 1650±1655. Braff, D.L., Heaton, R., Kuck, J., Cullum, M., Moranville, J., Grant, I., Zisook, S., 1991. The generalized pattern of neuropsychological de®cits in outpatients with chronic schizophrenia with heterogeneous Wisconsin Card Sorting Test results. Arch. Gen. Psychiatry 48, 891±898. Calsyn, D.A., O'Leary, M.R., Chaney, E.F., 1980. Shortening the category test. J. Consult. Clin. Psychol. 48, 788±789. Cohen, J., 1988. Statistical Power Analysis for the Behavioral Sciences. . second ed.Lawrence Erlbaum, Hillsdale, NJ. Dickerson, F., Boronow, J.J., Ringel, M., Parente, F., 1996. Neurcogonitive de®cits and social functioning in outpatients with schizophrenia. Schizophr. Res. 21, 75±83. Gold, J.M., Queern, C., Iannone, V.N., Buchanan, R.W., 1999. Repeatable battery for the Assessment of Neuropsychological Status as a screening test in schizophrenia, I: sensitivity, reliability, and validity. Am. J. Psychiatry 156, 1944±1950. Goldstein, G., Beers, S.R., Shemansky, W.J., 1996. Neuropsychological differences between schizophrenic patients with heterogeneous Wisconsin Card Sorting Test performance. Schizophr. Res. 21, 13±18. Green, M.F., 1996. What are the functional consequences of neurocognitive de®cits in schizophrenia? Am. J. Psychiatry 154, 321±330. Green, M.F., 1999. Interventions for neurocognitive de®cits Editor's introduction. Schizophr. Bull. 25, 197±200. Green, M.F., Kern, R.S., Braff, D.L., Mintz, J., 2000. Neurocognitive de®cits and functional outcome in schizophrenia: are we measuring the `right stuff'? Schizophr. Bull., 26 (1), 119±136. Green, M.F., Satz, P., Ganzell, S., Vaclav, J.F., 1992. Wisconsin Card Sorting Test performance in schizophrenia: remediation of a stubborn de®cit. Am. J. Psychiatry 149, 62±67. Heaton, R.K., Chelune, G.J., Talley, J.L., Kay, G.G., Curtiss, G., 1993. Wisconsin Card Sorting Test Manual. Psychological Assessment Resources, Odessa, FL. Hobart, M.P., Goldberg, R., Bartko, J.J., Gold, J.M., 1999. Repeatable battery for the Assessment of Neuropsychological Status as a screening test in schizophrenia, II: convergent/discriminant validity and diagnostic group comparisons. Am. J. Psychiatry 156, 1951±1957. Kern, R.S., Green, M.F., Goldstein, M.J., 1995. Modi®cation of performance on the span of apprehension, a putative marker of vulnerability to schizophrenia. J. Abnorm. Psychol. 104, 385±389. Lezak, M.D., 1995. Neuropsychological Assessment. third ed., Oxford University Press, New York.
262
A.S. Bellack et al. / Schizophrenia Research 48 (2001) 255±262
Medalia, A., Aluma, M., Tryon, W., Merriam, A.E., 1998. Effectiveness of attention training in schizophrenia. Schizophr. Bull. 24, 147±152. Metz, J.T., Johnson, M.D., Pliskin, N.H., Luchins, D.J., 1994. Maintenance of training effects on the Wisconsin Card Sorting Test by patients with schizophrenia or affective disorders. Am. J. Psychiatry 151, 120±122. Randolph, C., 1998. Repeatable Battery for the Assessment of Neurpsychological Status (RBANS). Psychological Corporation, San Antonio. Stratta, P., Mancini, F., Mattei, P., Casacchia, M., Rossi, A., 1994. Information processing strategy to remediate Wisconsin Card Sorting Test performance in schizophrenia: a pilot study. Am. J. Psychiatry 151, 915±918.
Vollema, M.G., Geurtsen, G.J., van Voorst, A.J. P., 1994. Durable improvements in Wisconsin Card Sorting Test performance in schizophrenic patients. Schizophr. Res. 16, 209±215. Wang, P.L., 1987. The search for a clinical frontal lobe test. In: Perecman, E. (Ed.), Concept Formation and Frontal Lobe Function. IRBN Press, New York, pp. 189±205. Wexler, B.E., Hawkins, K.A., Rounsaville, B., Anderson, M., Sernyak, M.J., Green, M.F., 1997. Normal neurocognitive performance after extended practice in patients with schizophrenia. Schizophr. Res. 26, 173±180. Young, D.A., Freyslinger, M.G., 1995. Scaffolded instruction and the remediation of Wisconsin Card Sorting Test de®cits in chronic schizophrenia. Schizophr. Res. 16, 199±207.