Achieving certainty about the structure of intolerance of uncertainty in a treatment-seeking sample with anxiety and depression

https://doi.org/10.1016/j.janxdis.2010.08.010Get rights and content

Abstract

Evidence is accumulating that intolerance of uncertainty (IU) may be a transdiagnostic maintaining factor across the anxiety disorders and depression. However, psychometric studies of the most commonly used measure of IU have typically used undergraduate students, and the factor structure has been highly inconsistent. Previous studies have also tended to focus on one diagnostic subgroup or related symptom, thereby limiting transdiagnostic comparisons. The first aim of this study was to test the latent structure of a commonly used measure of IU in a treatment-seeking sample with anxiety and depression (n = 463). The second aim was to examine psychometric properties of the best fitting solution, including internal reliability, convergent validity, and discriminant validity. Confirmatory factor analysis was used to compare the goodness of fit of five models previously found with undergraduate and community samples. A two-factor solution, comprising of prospective anxiety and inhibitory anxiety, was the best fitting model. The total scale and subscales demonstrated excellent internal reliability. Convergent validity was demonstrated by the scales correlating with symptoms associated with five anxiety disorders and depression, as well as neuroticism, distress and disability. IU explained unique variance in all symptom measures, even after controlling for neuroticism and other symptom measures. Evidence of discriminant validity was also found for each IU subscale. Findings support reliability and validity of the two-factor solution, and are consistent with IU being a transdiagnostic maintaining factor.

Research highlights

▶ A 12-item two-factor model of intolerance of uncertainty (IU) provided superior fit. ▶ Prospective anxiety was uniquely associated with GAD and OCD. ▶ Inhibitory anxiety was uniquely associated with phobic disorders and depression. ▶ IU appears to be a transdiagnostic construct.

Introduction

Individuals who are intolerant of uncertainty experience possibility of negative future events as threatening and unacceptable, regardless of the probability of the events actually occurring (Dugas, Gosselin, & Ladouceur, 2001). As a consequence, they tend to respond negatively to uncertain or ambiguous situations on an emotional, cognitive and behavioral level (Dugas, Buhr, & Ladouceur, 2004). For example, there is considerable evidence that intolerance of uncertainty (IU) is a cognitive vulnerability factor for worry (Koemer and Dugas, 2008, Ladouceur et al., 2000, Sexton et al., 2003, van der Heiden et al., 2010) and an important maintaining factor for generalized anxiety disorder (GAD; Behar et al., 2009, Dugas et al., 1998). However, IU has also been found to be associated with symptoms of obsessive-compulsive disorder (Holaway et al., 2006, Lind and Boschen, 2009, Tolin et al., 2003), and more recently other anxiety disorders including symptoms of social anxiety (Boelen and Reijntjes, 2009, Carleton et al., 2010, Riskind et al., 2007), panic disorder and agoraphobia (Buhr & Dugas, 2009), and depression (de Jong-Meyer et al., 2009, Dugas et al., 2004, van der Heiden et al., 2010). Thus, evidence is accumulating the IU may represent an important transdiagnostic maintaining factor (Starcevic & Berle, 2006).

The Intolerance of Uncertainty Scale (IUS; Freeston, Rhéaume, Letarte, Dugas, & Ladouceur, 1994) is the most widely used measure of IU and has been translated into English (Buhr & Dugas, 2002) and Dutch (de Bruin, Rassin, van der Heiden, & Muris, 2006), and has recently been adapted for children (Comer et al., 2009). The original French version of the IUS was designed to distinguish between differing levels of GAD pathology and it correlates significantly with worry (as measured by the Penn State Worry Quqestionnaire, PSWQ, Meyer, Miller, Metzger, & Borkovec, 1990). In student samples, the measure has consistently demonstrated excellent internal consistency and convergent validity with respect to relationships with worry, neuroticism, anxiety and depression (Berenbaum et al., 2008, Buhr and Dugas, 2002, Norton, 2005, Sexton and Dugas, 2009).

An important limitation of the IUS is that its factor structure has been highly inconsistent across studies (see Table 1). Freeston et al.’s (1994) initial exploratory factor analysis (EFA) on the French version yielded five factors reflecting different beliefs about and responses to uncertainty. In an English translation, Buhr and Dugas (2002) found a four-factor structure, although six items cross-loaded on multiple factors. Both Freeston et al. and Buhr and Dugas recommended against the factors being used as subscales and rather advocated for the unitary model. Freeston et al. made this decision based on high internal reliability for the whole scale and eigenvalue differences between the subscales, although the actual eigenvalues and the nature of the differences were not reported and only the Scree test was used to identify the number of factors. Buhr and Dugas recommended the use of the whole scale due to overlapping factors and item cross-loadings. Norton (2005) found five- and six-factor solutions, although the models differed significantly across ethnic groups and there was evidence of multi-vocal loading and poor interpretability. Subsequent studies employing EFA have found different solutions with varying congruency with previous findings (Berenbaum et al., 2008).

One explanation for the failure of these studies to identify robustly separable and replicable factors is that IU is best conceptualized as a unitary construct. Alternatively, responses to the IUS may be more homogenous at sub-threshold levels than at clinical levels. Therefore, use of undergraduate samples in these studies may have militated against the detection of clinically relevant multidimensional aspects of IU that are only separable at pathological levels. It is also noteworthy that all of these studies used EFA, which, unlike confirmatory factor analysis (CFA), precludes the specification of a priori pathway parameters. While it is appropriate to use EFA as an initial step in scale construction, EFA is data-driven and thus vulnerable to variability across samples. On the other hand, CFA enables the researcher to specify a model and determine goodness of fit to a particular dataset, thereby reducing the likelihood of spurious solutions that are unique to a particular sample.

Only two studies have used CFA to test the structure of the IUS. In an initial CFA, Carleton, Norton, and Asmundson (2007) examined the unitary, four- and five-factor models previously reported and found that none of these models provided an adequate fit to the data. A reduced 12-item version of the IUS was proposed, which demonstrated a stable two-factor structure, high internal consistency and its total score was correlated highly (r = .96) with that of the full 27-item IUS. Moreover, the reduction in items did not result in significant losses to convergent validity. In a large sample with both undergraduate students and community members, Sexton and Dugas (2009) identified a two-factor solution with all 27 items using EFA, which was then confirmed with CFA. Factors demonstrated moderate but significant differences regarding relationships with depression, neuroticism and GAD pathology. It is noteworthy that items in the two scales found by Carleton, Norton, et al. (2007) also loaded separately on Sexton and Dugas’ two scales, despite the latter including all 27 items (see Table 1). Thus, the two studies using CFA have identified two-factor solutions that were recommended as subscales. However, given that neither study used a clinical sample, the question remains about which version is most valid and reliable for use in clinical samples.

In sum, existing studies have predominantly employed student samples and EFA, which have resulted in numerous discrepancies in the IUS factor structure. Given the potential relevance of IU for anxiety and depressive disorders, it is important to investigate the factor structure of the IUS in a clinical sample. Rather than produce yet another factor solution using a data-driven EFA, we sought to examine if any of the existing models provide an adequate fit clinical data. As such, the current study used CFA to examine existing latent models of the IUS in a sample of clinically anxious and/or depressed patients. Previous studies have also typically used generic measures of anxiety (e.g. Beck Anxiety Inventory), which may be more valid for a subset of diagnoses than others (Cox, Cohen, Direnfeld, & Swinson, 1996), and thus precludes the identification of transdiagnostic relationships across specific anxiety disorders. Therefore, we used diagnosis-specific symptom measures for generalized anxiety disorder, social anxiety, panic disorder and agoraphobia, obsessive-compulsive disorder, and depression in order to examine relationships between IU and symptoms of each disorder. In addition, the association between the IUS and measures of non-specific psychological distress and disability were explored.

The first aim of this study was therefore to use CFA in a clinical sample to compare latent models of the IUS found in previous studies with undergraduate students. The second aim was to evaluate psychometric properties of the best fitting model, including internal reliability, convergent and discriminant validity. Convergent validity will be demonstrated if the IUS factors are significantly associated with neuroticism and diagnosis-specific symptom measures, as well as with distress and disability. Moreover, if IU is a transdiagnostic construct then the IUS should account for unique variance in symptoms related to each diagnosis above and beyond neuroticism, even when other symptoms are controlled for. Given that the IUS is expected to influence distress and disability through symptoms of anxiety and depression, discriminant validity will be demonstrated if the IUS does not explain variance in distress and disability above and beyond that explained by neuroticism and the symptom measures. Moreover, consistent with previous findings, the IUS should correlate weakly and negatively with the personality dimension of extraversion (Berenbaum et al., 2008).

Section snippets

Participants

Participants (n = 463, 55% women) were referred to a specialist anxiety disorders treatment service by General Practitioners or Psychiatrists. At the initial assessment participants completed the standard questionnaire battery and were diagnosed via a semi-structured clinical interview with a Consultant Psychiatrist. CIDI-Auto diagnoses were available for 121 of the participants to provide an indication of the frequency of anxiety and affective disorders. Proportions of these patients meeting

Data screening

Prior to data analyses, item distributions, skewness and kurtosis were examined. Items were generally normally distributed with no item demonstrating problematic levels of skewness (maximum skewness = .56) or kurtosis (maximum kurtosis = 1.23). Previous studies using non-clinical samples have identified higher levels of item skewness. Items were then screened for univariate and multivariate outliers. No standardized item scores exceeded 3 (minimum absolute value = 1.50, maximum absolute value = 2.44),

Discussion

The first aim of this study was to use confirmatory factor analysis (CFA) in a treatment-seeking clinical sample to compare latent models of the IUS previously found with undergraduate samples. We compared seven models and found that the best fitting model, and the only one with most goodness of fit indices meeting conventional cutoffs, was the 12-item two-factor version of the IUS extracted by Carleton, Norton, et al. (2007). The two factors were labeled prospective anxiety and inhibitory

References (68)

  • R.N. Carleton et al.

    “It's not just the judgments—It's that I don’t know”: intolerance of uncertainty as a predictor of social anxiety

    Journal of Anxiety Disorders

    (2010)
  • R.N. Carleton et al.

    Fearing the unknown. A short version of the Intolerance of Uncertainty Scale

    Journal of Anxiety Disorders

    (2007)
  • B.J. Cox et al.

    Does the Beck Anxiety Inventory measure anything beyond panic attack symptoms?

    Behaviour Research and Therapy

    (1996)
  • R. de Jong-Meyer et al.

    Relationships between rumination, worry, intolerance of uncertainty and metacognitive beliefs

    Personality and Individual Differences

    (2009)
  • M.J. Dugas et al.

    Generalized anxiety disorder: a preliminary test of a conceptual model

    Behaviour Research and Therapy

    (1998)
  • M.J. Dugas et al.

    Can the components of a cognitive model predict the severity of generalized anxiety disorder?

    Behavior Therapy

    (2007)
  • M.H. Freeston et al.

    Why do people worry?

    Personality and Individual Differences

    (1994)
  • R.M. Holaway et al.

    A comparison of intolerance of uncertainty in analogue obsessive-compulsive disorder and generalised anxiety disorder

    Journal of Anxiety Disorders

    (2006)
  • R. Ladouceur et al.

    Specificity of generalized anxiety disorder symptoms and processes

    Behavior Therapy

    (1999)
  • R. Ladouceur et al.

    Experimental manipulation of intolerance of uncertainty: a study of a theoretical model of worry

    Behaviour Research and Therapy

    (2000)
  • C. Lind et al.

    Intolerance of uncertainty mediates the relationship between responsibility beliefs and compulsive checking

    Journal of Anxiety Disorders

    (2009)
  • R.P. Mattick et al.

    Development and validation of measures of social phobia scrutiny fear and social interaction anxiety

    Behaviour Research and Therapy

    (1998)
  • R.P. Mattick et al.

    Exposure and cognitive restructuring for severe social phobia: a controlled study

    Behavior Therapy

    (1989)
  • T.J. Meyer et al.

    Development and validation of the Penn State Worry Questionnaire

    Behaviour Research and Therapy

    (1990)
  • P. Norton

    A psychometric analysis of the Intolerance of Uncertainty Scale among four racial groups

    Journal of Anxiety Disorders

    (2005)
  • J.H. Riskind et al.

    Short-term predictive effects of the looming cognitive style on anxiety disorder symptoms under restrictive methodological conditions

    Behaviour Research and Therapy

    (2007)
  • D. Tolin et al.

    Intolerance of uncertainty in obsessive-compulsive disorder

    Journal of Anxiety Disorders

    (2003)
  • P. van Oppen et al.

    The structure of obsessive-compulsive symptoms

    Behaviour Research and Therapy

    (1995)
  • H.-U. Wittchen

    Reliability and validity studies of the WHO—Composite International Diagnostic Interview (CIDI): a critical review

    Journal of Psychiatric Research

    (1994)
  • American Psychiatric Association

    Diagnostic and Statistical Manual of mental disorders

    (1994)
  • G. Andrews et al.

    The psychometric properties of the Composite International Diagnostic Interview

    Social Psychiatry and Psychiatric Epidemiology

    (1998)
  • G. Andrews et al.

    Normative data for the 12 item WHO Disability Assessment Schedule 2.0

    PLoS One

    (2009)
  • A.T. Beck et al.

    Beck Depression Inventory-II Manual

    (1996)
  • E.J. Brown et al.

    Validation of the Social Interaction Anxiety Scale and the Social Phobia Scale across anxiety disorders

    Psychological Assessment

    (1997)
  • Cited by (275)

    View all citing articles on Scopus
    View full text