Psychometric Properties of the Swedish Cardiac Anxiety Questionnaire: a Rasch analysis

Background: Cardiac anxiety (CA) is a commonly occurring condition in populations with both cardiac and mental health conditions. It is assessed using the CA questionnaire (CAQ), which has been translated to several languages and its validity and reliability has been investigated using classical test theory. The factor structure of the CAQ remains uncertain, and no study has investigated the psychometric properties of the CAQ using modern test theory approaches. The aim of this study was to use Rasch analysis to test the validity and reliability of the original 18-item CAQ in a Swedish sample of patients with myocardial infarction, as well as other scales, based on previous findings. Methods: Participants were 754 patients after myocardial infarction with preserved cardiac function. Data were fitted to the Rasch model, investigating dimensionality, ordering of response categories, invariance, targeting and measurement uncertainty. Additionally, a confirmatory factor analysis (CFA) was conducted using a weighted least mean squares estimator and dynamic fit indices. Results: All three subscales indicated problems with dimensionality and several items had disordered response categories, and with some of them minimal distance between steps. Many items displayed local dependency. There were also small problems with differential item functioning. Person locations were lower than the item threshold locations, but there were no significant floor or ceiling effects. The CFA also demonstrated a poor model fit of the three subscales in the 18-item version of the CAQ. An iterative removal of items resulted in a modified 10-item version still with questionable poor reliability. Conclusions: The original 18-item version of the CAQ did not fit the Rasch model. The major problems with the 18-item version were dimensionality, local dependency, and disordered response categories. No other version of the CAQ achieved acceptable fit to the Rasch model. These results indicate that the CAQ is lacking in terms of reliability and a revision of the scale is called for.