The specific approach in each study is detailed under the study-specific subheadings below. However, there were some consistent approaches across studies: we tested for internal reliability and validity using Cronbach's alpha and omega, and examining inter-item and item-total correlations; we performed exploratory factor analyses in Studies 1 and 2, then confirmatory factor analyses in subsequent Studies 3, 4 and 5. These methods are further detailed in the electronic supplementary material, along with our policy for item reduction during scale development.

An initial scale with 31 items was used. Further details on how the items were developed can be found in the electronic supplementary material.

We tested this first version of the questionnaire on a sample of 117 participants.

In this second study, we reduced the number of items in the scale, and simplified items which were considered too long or too complex. Details of item removal can be found in the electronic supplementary material.

Based on item reduction in Study 1, we tested a version of the scale with 25 items.

This iteration of the questionnaire was tested on 117 participants.

All participants completed the final version of the Catastrophizing Questionnaire along with convergent and discriminant validity questionnaires (all of which showed acceptable internal consistency in our sample, table 3).

Internal consistency indices for all measures used.

aGeneralized Anxiety Disorder Assessment.

bPersonal Health Questionnaire.

cPenn-State Worry Questionnaire.

dRuminative Responses Scale.

eSpielberger State-Trait Anxiety Inventory—Trait Subscale.

fTemporal Experience of Pleasure Scale, Anticipatory Subscale.

gTemporal Experience of Pleasure Scale, Consummatory Subscale.

hCognitive Distortion Scale: Catastrophizing Subscale.

iThere were too few items (n = 2) to calculate omega in the CDS-Catastrophizing scale.

jShort version of the full novel Catastrophizing Questionnaire reported in this paper.

kAlcohol Use Disorders Identification Test.

lShort Scales for Measuring Schizotypy—Unusual Experiences subscale.

mShort Scales for Measuring Schizotypy—Cognitive Disorganization subscale.

nShort Scales for Measuring Schizotypy—Introvertive Anhedonia subscale.

oShort Scales for Measuring Schizotypy—Impulsive Non-conformity subscale.

One item from the second version of the questionnaire was removed, resulting in a final version of the Catastrophizing Questionnaire with 24 items.

We developed a short version of the Catastrophizing Questionnaire, as there are a number of circumstances in which researchers or clinicians might need a briefer scale which corresponds to the full version. This short version consisted of four items. Three of them were from the Catastrophizing Questionnaire and were chosen based on their high factor loading and relevance to catastrophizing. The fourth item, ‘I turn minor issues into really big problems in my head’ which was derived from the first version of the questionnaire, reflected a more general definition of catastrophizing. We analysed the relationship between this questionnaire and its full version using a Pearson's correlation, and also analysed the relationship between this questionnaire without the fourth item and the full Catastrophizing Questionnaire.

Five hundred participants provided good quality data in Study 3 (for additional data quality assurance checks, see electronic supplementary material).

We collected data on a number of additional self-report questionnaires, in order to assess convergent and discriminant validity. We define construct validity here as positive associations between two measures of the same construct (i.e. two different questionnaires measuring the same thing), and define convergent validity as positive (but not large enough to indicate redundancy) associations between two measures of different constructs which are theoretically hypothesized to be related. We define discriminant validity as a lack of significant positive association between measures of constructs that are not hypothesized to be related to catastrophizing, but are related to mental health—such as anhedonia, schizotypy and alcohol use disorder. These measures are itemized below.

Cognitive Distortion Scale: Catastrophizing Subscale. The Cognitive Distortion Scale (CDS) conceived by Covin et al. [20] contains a two-item catastrophizing subscale. This subscale was used for comparison with our Catastrophizing Questionnaire to assess construct validity.

Anxiety. Catastrophizing is believed to relate to psychiatric diagnoses such as anxiety and depression, and as such should be related to scores on questionnaires measuring symptoms of these disorders. Therefore, we included the Generalized Anxiety Disorder Assessment (GAD-7), which measures symptoms of generalized anxiety disorder over the previous two weeks [26]. We also included the Trait Anxiety Subscale of the State-Trait Anxiety Inventory (STAI-T) which measures anxiety level as an enduring personality trait, and focuses on anxiety in general rather than on a specific anxiety disorder [27].

Depression. For the same reasons as above, we included the Patient Health Questionnaire (PHQ-9) which is a depression scale that asks about symptoms over the preceding two weeks [28,29]. As we were testing participants online and could not assess the risk of harm in person, we removed the question about suicidal ideation [30].

Worry. Worry is an important feature of anxiety disorders that may also be related to catastrophizing. Thus, we included the Penn State Worry Questionnaire (PSWQ), which measures traits of worry independently from anxiety [31]. It examines the excessiveness, generality and uncontrollability of worry.

Rumination. Rumination is related to depression, and we thus hypothesized it would also correlate with catastrophizing. However, it is thought to be separable from depression, so in addition to the PHQ-9, we included the Rumination Response Scale (RRS) which measures two aspects of rumination: brooding and reflective pondering [32].

Experience of pleasure. We expected the experience of pleasure to be orthogonal to catastrophizing, as it implies a positive conception of the anticipation of pleasure and enjoying the present moment. Therefore, we included the Temporal Experience of Pleasure Scale—Anticipatory (TEPS-ANT) and Consummatory (TEPS-CON) subscales, which assess pleasure experienced during anticipation of reward and on attaining rewards, respectively [33].

We performed Pearson's correlations between scores on different questionnaires to assess the magnitudes of relationships between the putative latent constructs. To assess the extent to which the convergent measures examine separable constructs, we performed an exploratory factor analysis (EFA) including all these items, and hypothesized that the items from the Catastrophizing Questionnaire would load onto a separate factor to other items. We chose to use an EFA as a model-free way of attempting to characterize the relationships between individual items from all questionnaires—rather than assuming that each questionnaire is separate, we wanted to test whether the Catastrophizing Questionnaire items coupled together with other questionnaires, or whether they loaded separately onto a factor of their own. To assess discriminant validity, we also used the heterotrait–monotrait ratio of correlations (HTMT), which is a particularly robust measure of discriminant validity [34], for which statistic a value of less than 0.85 can be considered discriminant.

We also present an exploratory CFA analysis, performed using lavaan, in the electronic supplementary material, in which each ‘convergent’ questionnaire was mapped to a latent factor, such that each individual item within that questionnaire became an indicator for that latent factor, and subsequently estimated the covariances between each latent variable. We anticipated that this would result in heightened correlations between questionnaires, due to the disattenuation for measurement error for each indicator. The aim of this analysis was to assess whether the anticipated heightened correlations between items were specifically increased for correlations with the Catastrophizing Questionnaire, which might have indicated redundancy with the other questionnaire measures we used.

We also examined participants' responses to questions on their mental health and their medication history. We performed binomial logistic regression analyses to establish the incremental validity gained in predicting self-reported diagnostic status by using the Catastrophizing Questionnaire over two questionnaires which are commonly obtained in primary care settings in the UK—the GAD-7 and PHQ-9. A null model predicting the self-report diagnosis status of each participant using GAD-7 and PHQ-9 was compared with a model of interest which included the GAD-7, PHQ-9 and Catastrophizing Questionnaire scores. Similarly, we also performed multinomial logistic regression analyses on the self-reported medication status of each participant. We hypothesized that the Catastrophizing Questionnaire would show incremental validity over these questionnaires in predicting both self-report medication usage and diagnostic status.

A number of participants (100) who had completed Study 3 repeated the final 24-item version of the Catastrophizing Questionnaire two months later. We did not systematically select these participants—the study was placed online and open to all participants who had completed Study 3, and recruitment proceeded in a ‘first-come, first-served’ manner.

This study was completed by an unselected, first-come, first-served group of 100 participants who had completed Study 3.

We firstly calculated the Pearson's correlation coefficient between participants' scores in Studies 3 and 4. We also calculated the intraclass correlation coefficient (ICC). More details on the ICC can be found in the electronic supplementary material.

A number of participants (264) who had completed Study 3 repeated the final 24-item version of the Catastrophizing Questionnaire 10 months later. They also completed two additional questionnaires to measure discriminant validity, and repeated the PHQ-9, GAD-7, PSWQ and RRS.

This study was completed by a subset of participants who had completed Study 3. As in Study 4, we did not systematically select these participants.

We collected data on two additional questionnaires: the Alcohol Use Disorder Identification Test (AUDIT), and the Short Scales for Measuring Schizotypy, both of which we did not expect to be highly related to catastrophizing, particularly over-and-above general psychiatric distress. More details on these can be found in the electronic supplementary material.

Note: The content above has been extracted from a research article, so it may not display correctly.



Q&A
Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.



We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.