Symptoms and Neurocognition as Rate Limiters in Skills Training for Psychotic Patients

Subjects were recruited on admission to an inpatient unit. Inclusion criteria included 1) a diagnosis of schizophrenia or schizoaffective disorder according to the Structured Clinical Interview for DSM-III-R; 2) age 18–55 years; and 3) IQ greater than 70. Subjects were excluded if they were experiencing acute withdrawal from drugs or alcohol. The final group included 32 subjects (14 women) with an average age of 34.6 years. The mean age at illness onset was 18.2 years, and subjects had an average of 7.6 prior hospitalizations.

After complete description of the study to the subjects, written informed consent was obtained. Subjects then completed a pretreatment assessment of symptoms, neurocognition, and community reintegration skills before being randomly assigned to receive either skills training or supportive group psychotherapy. The skills training cohort participated in the Community Re-Entry Program (6), which uses behavioral techniques and engages patients in efforts to define discharge readiness, identify symptoms and medication effects, and assist with discharge planning. The program, which was shown to improve both skill level and postdischarge adjustment in a prior study (7), includes a manual describing 16 sessions that were scheduled daily (Monday–Friday) for 60 minutes. The comparison group received daily supportive group therapy involving guided discussions of issues relevant to subjects’ daily experiences. All treatment groups ran continuously throughout the study, with subjects joining after completion of the pretreatment assessment. Following completion of the group treatment and before discharge, each subject completed a follow-up assessment of symptoms and skill level.

All subjects received antipsychotic medications in an uncontrolled fashion. In the skills training cohort, six patients received conventional agents, six received atypical agents (risperidone or olanza­pine), and four received clozapine. For those randomly assigned to supportive group therapy, 10 received conventional agents, three atypical agents, and three clozapine. Results of a chi-square test for medication differences were not significant.

Symptoms were assessed by using the expanded Brief Psychiatric Rating Scale (BPRS) (8) and the Scale for the Assessment of Negative Symptoms (SANS) (9). BPRS item scores were summed to create a total score and three dimension scores: positive symptoms, anxiety/depression, and mania. The global subscale ratings from the SANS were averaged to create a negative symptom score. Community reintegration skills were assessed by using a test provided with the Community Re-Entry Program and previously validated (7). The test includes 21 questions, and items are scored 0 or 1 on the basis of the presence of an acceptable answer. Interrater reliability was documented, with intraclass correlation coefficients of 0.73 for total skill scores and 0.42 to 0.98 for the symptom variables.

The neurocognition battery, administered before treatment, included measures of attention and executive functioning (Trail Making Test A and B). Verbal memory was assessed with the California Verbal Learning Test, with the sum of trials 1–5 variable used in analyses.

All subjects attended at least five sessions (mean=7.9 sessions for the skills training cohort and 8.8 for the supportive group therapy cohort). The two groups were initially compared on demographic, pretreatment symptom, and neurocognitive data. Only the age at illness onset variable was significantly different; the skills training group had a mean age at onset of 20.7 years (SD=7.4), compared with 15.2 years (SD=3.7) for the supportive group therapy group (t=–2.36, df=24, p=0.03).

Pretreatment skill scores did not distinguish the two cohorts (mean=8.4, SD=4.2, for skills training and mean=8.6, SD=4.9, for supportive group therapy). After training, the skills training cohort had significantly higher skills training test scores than the supportive group therapy group (mean=12.8, SD=3.7, versus mean=8.1, SD=4.4) (t=–3.04, df=26, p<0.005). Correlation coefficients were calculated between change scores for the skill test and the five symptom measures. None was significant. When pretreatment symptom levels were correlated with posttreatment skill, only the anxiety/depression variable was significant (r=0.51, df=26, p<0.01). All three neurocognitive variables correlated significantly with posttreatment skill (Trails A: r=–0.45, df=18, p<0.05; Trails B: r=–0.53, df=22, p<0.01; and California Verbal Learning Test: r=0.54, df=23, p<0.01). The anxiety/depression variable did not significantly correlate with any of the neurocognitive variables.

Regression equations were created with posttreatment skill level as the dependent variable and pretreatment skill level, group membership, and the symptom and neurocognition variables entered simultaneously as predictors. The best model was significant (F=31.6, df=3, 21, p<0.001) and accounted for 79% of the variance in posttraining skill ratings. Significant predictors included pretreatment skill level (beta=0.60, t=6.03, df=21, p<0.001), group membership (beta=0.48, t=5.09, df=21, p<0.001), and verbal memory (beta=0.29, t=2.87, df=21, p<0.01). A stepwise analysis indicated that pretreatment skill level accounted for 47% of the variance in posttreatment skill. The skills training intervention and memory score accounted for an additional 26% and 6% of the variance, respectively. Because the age at illness onset variable differentiated the two groups, a stepwise regression equation was created that added this variable to the predictors used previously. The age at illness onset variable failed to enter, and the model did not change.

Subjects receiving skills training acquired skills; this finding is consistent with a prior study of the Community Re-Entry Program (7). Verbal memory capacity was more predictive of skill acquisition than were symptoms. This also replicates previous findings but with important methodological advancements. We used a validated, manualized skills training protocol and included a comparison group to allow for documentation of a learning effect; this was not done in any previous studies. We also controlled for improvement related to symptom fluctuations by incorporating longitudinal assessments; this was addressed in only one prior study (5).

Pretreatment skill level and the skills training intervention did account for more of the variance in posttraining skill level than did neurocognitive deficits. Premorbid functioning is a known predictor of outcome in schizophrenia (10), and future studies of skills training should incorporate measures of baseline knowledge and competence. In addition, the fact that the skills training intervention accounted for more outcome variance in our study suggests that although neurocognitive deficits are rate limiting, they do not eliminate the effect of the intervention.

The lack of control for medication effects on neurocognitive functioning and skill level is another important limiting factor in this study. Nonetheless, we have replicated the finding of Mueser et al. (5) that short-term memory is a significant predictor of response to skills training in schizophrenia. Further research should guide efforts to develop strategies for matching patients with neurocognitive deficits to focused remediation or skills training treatments or both.

Received Dec. 28, 1998; revision received May 10, 1999; accepted May 13, 1999. From New York Presbyterian Hospital, Westchester Division. Address reprint requests to Dr. Smith, Weill Medical College of Cornell University, New York Presbyterian Hospital, Westchester Division, 21 Bloomingdale Rd., White Plains, NY 10605; [email protected] (e-mail). Supported by NIMH grant MH-01359 to Dr. Smith and by funds established in The New York Community Trust by DeWitt-Wallace. The authors thank Drs. Charles Wallace and Sally MacKain for their consultation and Lorna Starkey, Sarah Pratt, and Rosa Lim for their assistance with the project.