This study examined neuropsychological predictors of functional outcome in a randomized clinical trial of CBSST for middle-aged and older people with schizophrenia or schizoaffective disorder. We previously reported that participants in CBSST showed greater skills acquisition and superior self-reported functioning relative to participants in TAU at end of treatment (Granholm et al. 2005
) and these gains were maintained at 1-year follow-up (Granholm et al. 2007
). The present study found that neuropsychological impairment was a nonspecific predictor of skills acquisition and functional outcome in CBSST. Greater severity of neuropsychological impairment predicted poorer skills acquisition and functional outcome at 12- month follow-up for both treatment groups. Neuropsychological impairment, however, was not found to be a significant moderator of outcome in CBSST. The necessary group X neuropsychological impairment interactions in the statistical models used to test for moderation were not significant (Kraemer et al. 2002
It is important to stress that the null effects found in this study cannot “prove” that neuropsychological impairment did not moderate outcome in CBSST relative to TAU. Effect sizes for the treatment group effect, however, were similar for participants with relatively mild (d
=.44–.64) and severe (d
=.29–.60) neurocognitive impairment, and effect sizes for the group X neuropsychological impairment interactions were close to zero or even in the opposite direction (std β = −.001 to +.033). Participants in CBSST, therefore, showed comparable benefit relative to TAU, regardless of severity of neuropsychological impairment. Given that neuropsychological impairment is more common in older people with psychosis (Jeste et al. 1999
), it is particularly important that adding CBSST to standard pharmacologic care improved functioning relative to standard care alone, even for older participants with severe neuropsychological impairment.
Referring clinicians may not recommend CBT or other psychosocial interventions for people with psychotic disorders, because neuropsychological impairment is common in these individuals and it is often assumed that this might be a barrier to success in psychotherapy. This may not be an accurate assumption. Studies of SST for schizophrenia, for example, found that greater severity of neuropsychological impairment was associated with poorer outcome (McKee et al. 1997
;Mueser et al. 1991
;Kern et al. 1992
;Smith et al. 1999
;Ucok et al. 2006
;Silverstein et al. 1998
), but these studies did not examine whether neuropsychological impairment differentially impacted outcome relative to a control group, which is required for moderation (Kraemer et al. 2002
). Neuropsychological impairment predicted poorer outcome in general in these studies (nonspecific predictor), but it cannot be concluded that participants with greater neuropsychological impairment would not benefit from adding SST to standard care, relative to standard care, alone. The present study, as well as one other study of CBT for psychosis (Leclerc et al. 2000
), found that neuropsychological impairment did not significantly moderate outcome in CBT relative to standard care. Further, more severe
neuropsychological impairment was associated with better
symptom outcome in another study of CBT for psychosis (Garety et al. 1997
). Published research, therefore, has provided no support for the assumption that people with neuropsychological impairment should be excluded from CBT for psychosis.
Consistent with previous research (McKee et al. 1997
;Kern et al. 1992
), participants with more severe neuropsychological impairment were less active participants in group and completed fewer homework assignments. Unlike prior research, neuropsychological impairment was not significantly correlated with attendance in sessions in this CBSST trial, but this may have been due to restricted range in this sample with excellent attendance (M
= 22 [92%] of 24 sessions; 95% C.I.= 21–23). Transportation to therapy groups was also provided in this CBSST trial, which may have compensated for the negative impact of neuropsychological impairment on attendance. Nonetheless, even when present in group, participants with more severe neuropsychological impairment participated less and attempted fewer homework assignments. We (Granholm et al. 2006
) previously found that greater CBSST group participation was associated greater improvement in total symptoms, and greater homework adherence was associated with greater skill acquisition. Improving group participation and homework adherence in participants with more severe neuropsychological impairment, therefore, may improve treatment outcomes.
Results of this study did not support the hypothesis that improvement in neuropsychological functioning mediates improvement in skills acquisition and independent living skills in CBSST. Neuropsychological test performance improved to a similar extent in CBSST and TAU and improvement in neuropsychological functioning was not significantly associated with improvement in skills acquisition or independent living skills. Other cognitive mechanisms, like changing dysfunctional performance beliefs (e.g., “Why bother, I will fail” or “It’s too hard; not worth the effort”) through cognitive therapy, may have contributed to improvement in functional outcome. Changing these cognitions may not require changing underlying basic neurocognitive abilities.
Some researchers (Brenner et al. 1992
;Perris and Skagerlind 1994
) have described two broad categories of cognitive interventions for schizophrenia: Cognitive remediation, which focuses on rehabilitation of specific cognitive functions (e.g., memory, attention), and CBT, which trains participants to use metacognitive strategies to identify and challenge unhelpful or inaccurate thoughts. Perris and Skagerlind (Perris and Skagerlind 1994
) point out that these areas of intervention have several factors in common and may lie on a continuum from more ‘molecular’ to more ‘molar’ cognitive rehabilitation approaches. At one end, cognitive remediation targets specific problems of neurocognitive processing, and at the other end, CBT targets more global cognitive constructs, such as thinking about thinking (metacognition) experiences and self-concept. Both approaches often involve repeated practice of skills, behavioral rewards for attention to task, self-instruction and inductive reasoning strategies applied to thinking and social perception, and hypothesis testing and problem-solving training to promote cognitive flexibility. CBSST incorporates all of these cognitive remediation factors, but the present study did not find significantly greater improvement in any neuropsychological domain in CBSST relative to improvements found for standard care. It is possible that the specific tests selected were not sufficiently sensitive to the types of neurocognitive improvements expected in CBT (e.g., reasoning, flexibility, problem-solving). It is also possible that incorporating repeated practice of specific neurocognitive skills might lead to greater improvement in neurocognition in CBSST, which may, in turn, improve clients’ ability to apply the reasoning, thought challenging and problem solving skills trained in CBSST.
Strengths of the study included a well-controlled clinical trial design, appropriate mediation/moderation analyses (Kraemer et al. 2002
), an understudied population, good retention of participants, and a relatively thorough evaluation of neuropsychological abilities associated with functional outcome in schizophrenia. Limitations included null findings in the context of a moderately small sample size, but comparable treatment effect sizes were found for subsamples with mild and severe neuropsychological impairment and effect sizes for interaction effects were very small or even in the opposite direction of moderation. In addition, exclusion of people with current comorbid substance dependence may reduce the generalization of the findings, but comorbid substance dependence declines with age in people with schizophrenia (Palmer et al. 1999
) and might interfere with neuropsychological testing. Participants also were not excluded for comorbid neurologic illness, which may interfere with neurocognitive rehabilitation, but accounting for neurologic illness in analyses did not alter the pattern of results. Finally, medications and symptoms also might impact neuropsychological test scores and functional outcomes, but adjusting for these variables in analyses also did not alter the pattern of results.