Presentation on theme: "Method Introduction Results Relationships Between Attributions, Insight into Mental Disorder, and Treatment Response in Rehabilitation for Serious Mental."— Presentation transcript:
Method Introduction Results Relationships Between Attributions, Insight into Mental Disorder, and Treatment Response in Rehabilitation for Serious Mental Illness Jason E. Peer, Anita H. Sim, Eric Strachan, A. Jocelyn Ritchie, and Will Spaulding. University of Nebraska – Lincoln Social cognition in serious mental illness (SMI) has recently received increased attention in both theoretical and applied research domains. It has been argued that an investigation of social cognitive variables in SMI may yield a greater understanding of these disorders (Penn et al. 1997) and that social cognitive variables may mediate treatment outcomes (Green and Nuechterlein 1999). Some research in rehabilitation settings for SMI supports these assertions. For example, negative attributions about self-efficacy have been associated with poorer treatment response (Hoffman et al. 2000) and self-efficacy attributions have been found to improve as a result of psychosocial rehabilitation (Schaub et al. 1999). It has also been suggested that insight into mental disorder is a form of “meta social cognition” and may mediate treatment outcomes (Green and Nuechterlein 1999), although this relationship is considerably complex (i.e., McEvoy et al. 1989). Lack of insight has been found to be associated with a variety of variables in SMI populations including treatment adherence and neurocognitive functioning. While research has evaluated the relationships between treatment and insight and attribution style independently, to date little research has evaluated the combined contributions of these variables to treatment response. The current study sought to evaluate the relative and combined contributions of attributions and insight to treatment response in an intensive psychosocial rehabilitation setting for SMI. Based on previous research (Hoffman et al. 2000) it was hypothesized that a negative attribution style (as measured by I-SEE and IPSAQ) is associated with poorer functioning in the rehabilitation program. In addition, it was hypothesized that personalizing bias (i.e., the tendency to blame negative events on other people) would be associated with poor functioning in the rehabilitation program. It was also hypothesized that there would be an interaction between personalizing bias and insight scores which would predict poorer functioning in the rehabilitation program during the first six months of treatment. Participants The participants were 35 inpatients engaged in an intensive psychiatric rehabilitation program for individuals with SMI. The treatment regimen of this program consists of optimal psychopharmacological interventions, intensive training in areas of social, occupation and leisure skills and rehabilitation interventions designed to increase participants management of their mental disorder. The treatment is conducted in the context of a social learning program that uses contingency management (CM) on an individualized basis to maximize participants’ engagement in treatment. Measures Attributions: The Inventory for Self-Efficacy and Externality (I-SEE) (Krampen 1991) was used to assess a more global attributional style (locus of control). This measure consists of 4 primary scales: Internality, Self-Concept of Own Competence, Powerful Others’ Control Beliefs and Chance Control Beliefs. It yields two composite scales Self Efficacy and Externality. The Internal, Personal, Situational Attribution Questionnaire (IPSAQ) (Kinderman and Bentall 1997) assesses a more interpersonal attributional style based on participants explanations of positive and negative social scenarios. It yields two measures: an externalizing bias (Eb) score (the degree to which persons attribute negative events to external factors and attribute positive events to themselves) and personalizing bias (Pb) (the degree to which persons attribute negative events to other people as opposed to situational factors). Insight: Insight was assessed based on the total score on the Insight Scale (IS) (Birchwood et al. 1993), an eight item self- report measure of insight into mental disorder. Treatment variables: “Instances of noncompliance per week” (RNC) (i.e., refusing to attend treatment groups, not following program rules etc.) were counted as an index of participants’ response to treatment. The SMI rehabilitation program’s infrastructure includes a computerized system for collection of this behavioral data based on the CM interventions. The data collection is subject to monthly fidelity checks by trained graduate students to ensure behaviors are being recorded accurately and contingencies are being implemented correctly. The Nurse Observation Scale of Inpatient Evaluation-30 (NOSIE-30) (Honigfeld et al. 1966) was used to measure general psychosocial functioning. The mean weekly RNC instances and mean monthly total NOSIE scores were calculated for the first six months of rehabilitation and used in the subsequent analyses. Procedure Behavioral observational data were collected continuously over the first 6 months of rehabilitation. Attribution and insight measures were administered within the first month of admission to the rehabilitation program. Complete assessment data were not available for all participants which resulted in a smaller pool of participants for some of the subsequent analyses. All data were screened for outliers and checked for normality.. The outlier analysis yielded outliers on two variables, average weekly instances of RNC and the I-SEE Self-Efficacy composite scale. These extreme values were addressed using a windsorizing procedure. Table 1 shows the correlations between the social cognitive variables and the treatment variables. The I-SEE Externality composite score and associated subscales (Powerful Others’ Control Beliefs and Chance Control Beliefs) were significantly and positively associated with average weekly instances of RNC during the first six months of rehabilitation. That is, a higher degree of external locus of control was related to higher instances of noncompliance. There were no significant correlations between social cognitive variables and average monthly NOSIE total assets score. In order to evaluate the unique contributions of the social cognitive variables to the treatment variables five multiple regression analyses were run. Table 2 shows the relative contributions of social cognitive variables to average weekly instances of RNC during the first six months of rehabilitation. Social cognitive variables accounted for a significant portion of the variance in average RNC (R 2 = 0.541, F = 4.003, p = 0.014). Evaluation of the Beta weights indicated that the I-SEE Externality composite scale was the only variable which contributed significantly. A subsequent regression was run to evaluate the relative and unique contributions of the two I-SEE Externality subscales. Together these subscales accounted for 41% of the variance in average RNC (F = 8.85, p = 0.001). Evaluation of the Beta weights indicate that only the Chance Control Beliefs subscale made a unique contribution to average RNC (Beta = 0.424, t = 2.175, p = 0.039). Table 3 shows the relative contributions of social cognitive variables to average NOSIE total asset scores. Social cognitive variables did not account for a significant portion of the variance in average NOSIE total asset scores (R 2 = 0.461, F = 2.393, p = 0.091). Although the overall model was not significant, evaluation of the Beta weights indicated that total Insight Score made a significant contribution to the model (Beta = 0.498, t = 2.230, p = 0.043) such that increased insight was related to increased psychosocial functioning. Table 4 shows the relative contributions of personalizing bias, insight and their interaction to NOSIE total assets score. This regression analysis tested a partial model (individual variables only) against a full model (individual variables and their interaction term). The partial model was not significant (R 2 = 0.23, F = 2.546, p = 0.108). However, the addition of the interaction term resulted in a significant R 2 change (R 2 change = 0.17, F change = 4.566, p = 0.048). Thus, the interaction term accounted for 17% of the variance in psychosocial functioning during the first six months of rehabilitation. Evaluation of the Beta weights in the full model indicated that only the interaction term made a unique contribution to the full model (Beta = -0.47, t = -2.137, p = 0.048). The pattern of the interaction indicates that an increase in insight in combination with an increase in personalizing bias results in poorer psychosocial functioning, whereas poorer insight in combination with greater personalizing bias resulted in better psychosocial functioning. The same analysis was conducted using average RNC as the criterion variable. The results were not significant (R 2 = 0.217, F = 1.754, p = 0.19). Discussion The results indicate some support for the initial hypotheses of the study. First, as predicted, a negative attribution style(external Locus of Control (LOC)) as measured by the I-SEE was associated with poorer functioning in the rehabilitation program. This finding represents a replication of previous research using the I-SEE which found that an external LOC was associated with poorer outcomes in a vocational rehabilitation program for SMI (Hoffmann, Kupper et al. 2000). However, I-SEE externality was not associated with the NOSIE, a more global measure of psychosocial functioning. Nor was there a significant negative correlation between negative attribution style as measured by the IPSAQ externalizing bias score and the treatment variables as hypothesized. Second, as predicted, the interaction between insight and personalizing bias (Pb) was associated with poorer functioning in the rehabilitation program. The fact that an external locus of control was related to treatment noncompliance may, in part, be a result of the CM interventions utilized in the rehabilitation program. Specifically, CM seeks to modify behavior with reinforcers that are salient to the individual. Theoretically, the therapeutic effect of CM depends not only on an individual’s ability to identify response-outcome contingencies (i.e., make an association between appropriate behavior and reinforcement) but also their motivation to obtain or achieve a given reinforcer. It has been argued that an external locus of control is indicative of a hopelessness or depression (Hoffmann et al. 2000). One could predict that individuals maintaining this outlook may not benefit from CM due to poor motivation to obtain reinforcers (i.e., anhedonia). In turn, given an external locus of control, it may also be predicted that these individuals demonstrate a cognitive deficit in making associations between their behavior and a given reinforcer. That is, when obtaining reinforcement, they may associate it with some external factor as opposed to their own behavior. The interaction of insight and personalizing bias represents an intriguing finding. It was initially hypothesized that personalizing bias (i.e., the tendency to blame negative events on other people) would be associated with poor functioning in the rehabilitation program. Evaluation of the regression models did not support this hypothesis. In fact, the bivariate correlations, although not significant, were in the opposite direction than predicted (greater personalizing bias was associated with a higher level of functioning). However, this relationship was reversed in the presence of greater insight. Given that insight has been associated with neurocognitive functioning, it is appealing to consider that the observed relationships may reflect neurocognitive distinctions such that a personalizing bias combined with better neurocognitive functioning results in poorer functioning in the rehabilitation program. These findings warrant further investigation.