Longitudinal evaluation of clustering of chronic sinonasal and related symptoms using exploratory factor analysis

• Published in: Allergy (Copenhagen) Vol. 73; no. 8; pp. 1715 - 1723
• Main Authors: Cole, M., Bandeen‐Roche, K., Hirsch, A. G., Kuiper, J. R., Sundaresan, A. S., Tan, B. K., Schleimer, R. P., Kern, R. C., Schwartz, B. S.
• Format: Journal Article
• Language: English
• Published: Denmark Blackwell Publishing Ltd 01.08.2018
• Subjects: Allergies; Asthma; chronic rhinosinusitis; Epidemiology; exploratory factor analysis; Factor analysis; longitudinal; nasal and sinus symptoms; Pain; population‐based epidemiology; Rhinitis; Rhinosinusitis; Sinusitis; chronic rhinosinusitis; exploratory factor analysis; population-based epidemiology; longitudinal; nasal and sinus symptoms
• ISSN: 0105-4538; 1398-9995; 1398-9995
• DOI: 10.1111/all.13470

Background Sinonasal symptoms are common and can have several underlying causes. When symptoms occur in specified patterns lasting 3 months or more they meet criteria for chronic rhinosinusitis (CRS). Approaches to CRS symptom measurement do not specify how to measure symptoms and treat specified sinonasal symptoms as generally interchangeable, suggesting that such symptoms should cluster on 1 or 2 latent factors. Methods We used questionnaire responses to 37 questions on the presence, severity, bother, and frequency of cardinal sinonasal and related symptoms lasting 3 months, from 3535 subjects at 3 time points over 16 months. We completed 5 exploratory factor analyses (EFA) to identify symptom clustering, 1 for each time point and 2 for the differences between adjacent questionnaires. The baseline EFA was used to provide factor scores that were described longitudinally and examined by CRS status. Results Five EFAs identified the same 5 factors (blockage and discharge, pain and pressure, asthma and cold/flu symptoms, smell loss, and ear and eye [mainly allergy] symptoms), with clustering determined by symptom frequency, severity, and degree of bother. Responses to individual questions showed changes over time but when combined into factor scores showed less longitudinal change. All symptom factor scores were progressively higher from never to past to current CRS status. Conclusions Although the current approaches to symptom characterization in CRS imply a single underlying latent construct, our results suggest that there are at least 3 latent constructs relevant to CRS. Further studies are needed to evaluate whether these clusters have identifiable underlying pathobiologies.