Validity of the Nijmegen Questionnaire for Hyperventilation Syndrome
The Nijmegen Questionnaire is an outcome measure used by health professionals for the assessment of symptoms associated with hyperventilation syndrome in the clinical setting and other areas of health research. However, a review of existing literature suggests evidence on the psychometric properties of the questionnaire is limited. Further, in some instances, the methods used in the investigation of these properties are questionable. The aim of this study was to investigate two specific psychometric properties of the Nijmegen Questionnaire: content validity and internal construct validity. Based on the principles of outcome measure development and testing, Qualitative Descriptive Methodology and Rasch analysis were employed. To assess content validity, data from six patients with hyperventilation syndrome and four health professionals with relevant clinical experience were collected using semi-structured interviews. Interview data were analysed using conventional content analysis to identify symptoms of hyperventilation syndrome, and organised into categories and sub-categories. Data were then mapped against current items of the Nijmegen Questionnaire. In addition, data from 239 questionnaires were collected and analysed using Rasch analysis to establish the internal construct validity of the Nijmegen Questionnaire. Results indicated that perceived symptoms of hyperventilation syndrome divided into three categories: breathing symptoms, psychological symptoms, and physical symptoms. Each category had various sub-categories, which included symptoms that were conceptually congruent with each other. There was only one questionnaire item that did not map onto symptoms identified by participants. However, there were noted differences in symptom reporting and language used between data and the existing questionnaire items. Likewise there were differences between patients and health professional reports. Rasch analysis showed the current Nijmegen Questionnaire did not fit the Rasch model. Issues were identified concerning a misfitting item which was under discriminating and demonstrated differential item functioning for gender. This item was deleted. All items had disordered thresholds. Response categories were collapsed and items with local dependencies between them which were conceptually similar were combined into testlets. The revised version of the Nijmegen Questionnaire contained 15 items, was unidimensional and fit the criteria for the Rasch model. Subsequently, a conversion table was created for transforming raw total scores of the Nijmegen Questionnaire. The study provides additional evidence on the psychometric properties of the Nijmegen Questionnaire that allows clinicians and researchers to ascertain the value of this outcome measure for the screening of hyperventilation syndrome. Given that all but one existing item mapped onto interview data, the study findings demonstrated that the Nijmegen Questionnaire technically meets criteria for content validity. However, there are some differences in the way patients and health professionals talk about symptoms which should be considered when interpreting findings. The Nijmegen Questionnaire did not meet strict criteria for internal construct validity until after deletion of one misfitting item and dealing with threshold disordering and local dependency. Therefore, those using this measure should rescore the items as done in this study, ignore responses provided to item NQ14 (Cold hands or feet) and then use the conversion table to convert ordinal raw scores to interval scores. Future studies are recommended to explore whether the questionnaire wording could be improved to better reflect how patients perceive their symptoms. Further research should include a more diverse range of patient participants by including males, individuals of different ethnic, and socioeconomic background to ensure these findings are transferable to these populations. Furthermore, future work should explore test-retest reliability and responsiveness of the Nijmegen Questionnaire.