The Cognitive Biases Questionnaire for Psychosis (CBQ-P) and the Davos Assessment of Cognitive Biases (DACOBS): Validation in a Flemish sample of psychotic patients and healthy controls

Abstract

Objective

A large body of research has demonstrated the importance of cognitive biases in the development and maintenance of psychosis. Self-report scales for routine clinical practice have been developed only recently. Two new instruments on cognitive biases are evaluated: the Cognitive Biases Questionnaire for Psychosis and the Davos Assessment of Cognitive Biases Scale.

Methods

In a Flemish sample of 98 patients diagnosed with schizophrenia and 152 healthy controls, we investigated (1) the factor structure, (2) the reliability (internal consistency), (3) the discriminative power and (4) the convergent validity of the Dutch CBQ-P and the DACOBS.

Results

Using Confirmatory Factor Analysis, a 1-factor solution provided the best fit for the CBQ-P, and a 3-factor solution for the DACOBS. The CBQ-P Total Scale and the three scales of the DACOBS showed good internal consistencies. The CBQ-P Total Scale and all three DACOBS subscales were able to differentiate between healthy controls and patients diagnosed with schizophrenia, when controlling for age and years of education. The CBQ-P and DACOBS scales showed moderate correlations, confirming the convergent validity of both scales.

Conclusions

The CPQ-P and DACOBS appear to be psychometrical sound instruments to assess general thinking bias in psychosis within a Flemish population. Implications for future research are discussed.

Introduction

A large body of research has demonstrated the importance of cognitive biases in the development and maintenance of psychosis and delusions more specifically. Bentall and colleagues (Bentall et al., 1994, Kinderman and Bentall, 1997) found persecutory delusions to be associated with a tendency towards defensive, explanatory, personalizing attributional biases, conceptualized as the “attribution–self-representation cycle” (Bentall et al., 2001). As far back as 1988, Huq et al. described biases in data gathering, also known as Jumping To Conclusions (JTC), which appear to be present in ongoing delusional psychosis (Garety et al., 2005) as well as in at-risk state for psychosis (Broome et al., 2007). Also, JTC-bias is found to be still detectable in people recovered from delusions (Peters and Garety, 2006, Bentall et al., 2007). Woodward et al., 2006a, Woodward et al., 2006b described a bias against disconfirmatory evidence (BADE), or a tendency not to take disconfirmatory information into account when assessing one's own beliefs, as a risk factor for delusion formation. In addition, Riccaboni et al. (2012) also found a bias against confirmatory evidence (BACE) in deluded as well as non-deluded patients with schizophrenia, with BADE as well as BACE measures differentiating (deluded as well as non-deluded) patients with schizophrenia from healthy controls. While interesting work has been done on which bias comes into play under which particular circumstances (Speechley et al., 2008), others have focused on cognitive problems in delusions and psychosis on the whole concerning Theory of Mind (Bora et al., 2009). Still others have investigated the perpetuating role of safety behavior in psychosis through a process of avoidance learning in addition to an interpretation bias regarding negative events (Moutoussis et al., 2007). Recent cognitive models on psychosis describe reasoning biases as “key components to the development of delusions” (Garety et al., 2007). This “by way of their influence on the appraisal of disturbing anomalous experiences and adverse events” (Garety et al., 2007), the latter stemming from the ‘aberrant salience’-phenomenon as described by Van Os (2009) and Van Os and Kapur (2009). Although well-established by means of experimental tasks such as the classic “beads” task (Huq et al., 1988), variations on the “beads” task (Speechley et al., 2008) or more recently the Picture To Decision Task (Rubio et al., 2011) for establishing JTC-bias, research on cognitive biases in psychosis has only recently given rise to the development of self-report scales for cognitive biases that are sufficiently practical in routine clinical practice. Two new instruments are of special interest in this matter: the Cognitive Biases Questionnaire for Psychosis (CBQ-P; Peters et al., 2010, Peters et al., 2013) and the Davos Assessment of Cognitive Biases Scale (DACOBS; van der Gaag et al., 2013).

The CBQ-P is a recently developed questionnaire that “measures cognitive distortions considered to be important in psychosis, targeting delusions specifically” (Peters et al., 2010, Peters et al., 2013). It is based on the Cognitive Style Test (CST; Blackburn et al., 1986), which was designed to measure common thinking biases in depression. The original CST is composed of 30 vignettes representing everyday scenarios. Respondents are asked to select their own cognitive response to each of these scenarios out of 4 possible reactions: a very negative one (scored 4), a somewhat negative (scored 3), a somewhat positive (scored 2) or a very positive (scored 1). The CBQ-P keeps the format of the CST and uses 30 adapted, psychosis-relevant vignettes, measuring five specific cognitive biases: Jumping-To-Conclusions (making firm decisions based on little evidence), Intentionalizing (interpreting events or behaviors as deliberate), Catastrophizing (worst-case-scenario thinking), Emotion-Based Reasoning (describing definite (threatening) meaning to one's feelings on a particular moment) and Dichotomous (i.e. ‘black or white’) Thinking. Fifteen of the 30 scenarios comprising the CBQ-P relate to ‘anomalous experiences’ and 15 pertain ‘threatening events’, both considered to be key dimensions in psychosis. Respondents have to select one out of three given statements as their most likely reaction to the situation, with one possible choice representing the presence of bias (scored 3), another the absence of bias (scored 1), and a third option the presence of bias with some doubt about it (scored 2). The CBQ-P items and scoring key can be found in Appendix 1.

To validate the CBQ-P, Peters et al. (2010) assessed 173 patients with psychosis, together with a group of 32 depressed patients and a group of 30 healthy controls by means of the CBQ-P. They compared a 1-factor, 2-factor and 5-factor model of the CBQ-P, the 5-factor model representing the five specific cognitive biases (Jumping-To-Conclusions, Intentionalizing, Catastrophizing, Emotion-Based Reasoning and Dichotomous Thinking), the 2-factor model relating to the two dimensions aforementioned (Anomalous experiences, Threatening events), and the 1-factor model representing a General Thinking Bias. Peters et al. found that the 1-factor model fitted the data best. Although the 2-factor and the 5-factor model showed good fit as well, the different factors were highly correlated, and could hardly be differentiated empirically. Based on these findings, the more parsimonious 1-factor model was preferred. The 1-factor solution showed a high internal consistency (Cronbach's alpha coefficient equals 0.89 for the total group). However, the CBQ-P did not relate to existing self-report measures and experimental tasks supposed to assess similar reasoning biases [Beads Task, Catastrophizing Interview (Startup et al., 2007), Ambiguous Intentions and Hostility Questionnaire (AIHQ; Combs et al., 2007); Dysfunctional Attitudes Scale (DAS; Weissman and Beck, 178))], suggesting that the CBQ-P does not conceptualize reasoning, judgment or decision-making processes, but “rather taps into a different construct, perhaps a bias of interpretation” (Peters et al., 2013).

The Davos Assessment of Cognitive Biases Scale (DACOBS; Appendix 3) was developed by Mark van der Gaag et al. (2013). It consists of 42 statements relating to seven (six-item) subscales constructed by means of exploratory factor analysis: (1) Jumping to conclusions bias, (2) Belief Inflexibility bias (i.e. confirmation bias), (3) Attention to threat bias, (4) External attribution bias, (5) Social cognition problems, (6) Subjective cognitive problems, and (7) Safety behaviors. These seven subscales subsequently group into three higher order scales: Cognitive biases (1–4), Cognitive Limitations (5–6), and Safety behaviors (identical to 7). Respondents score each statement using a 7 point rating scale, ranging from 1 ‘totally disagree’ to 7 ‘totally agree’ taking into account the past two weeks. Van der Gaag et al. (2013) found good reliability for the DACOBS (Cronbach's alpha = .90), with the scale differentiating adequately between schizophrenia spectrum patients and healthy control subjects. Five out of seven subscales showed significant correlations (ranging from .36 to .63) with related measures: Jumping to conclusions bias related to the Beads Task, Belief Inflexibility bias related to the Dogmatism Scale (DOG; Altemeyer, 2002), Attention to threat bias related to the Green Paranoid Thought Scale part A, External attribution bias related to the GTPS part B (GPTS; Green et al., 2008), and Safety behaviors related to the Safety Behaviors Questionnaire — Paranoid Delusions (SBQ-PD; Freeman et al., 2001) (van der Gaag et al., 2013).

The aims of the current study are (1) to investigate the factor structure of the Dutch CBQ-P and the DACOBS in a Flemish sample, (2) to investigate the reliability of both scales, (3) to explore the ability of both scales to discriminate between patients with schizophrenia versus normal controls, and (4) to calculate the convergent validity of the CBQ-P and DACOBS, based on their correlations.