Test–retest reliability of multidimensional dyspnea profile recall ratings in the emergency department: a prospective, longitudinal study

Background Dyspnea is among the most common reasons for emergency department (ED) visits by patients with cardiopulmonary disease who are commonly asked to recall the symptoms that prompted them to come to the ED. The reliability of recalled dyspnea has not been systematically investigated in ED patients. Methods Patients with chronic or acute cardiopulmonary conditions who came to the ED with dyspnea (N = 154) completed the Multidimensional Dyspnea Profile (MDP) several times during the visit and in a follow-up visit 4 to 6 weeks later (n = 68). The MDP has 12 items with numerical ratings of intensity, unpleasantness, sensory qualities, and emotions associated with how breathing felt when participants decided to come to the ED (recall MDP) or at the time of administration (“now” MDP). The recall MDP was administered twice in the ED and once during the follow-up visit. Principal components analysis (PCA) with varimax rotation was used to assess domain structure of the recall MDP. Internal consistency reliability was assessed with Cronbach’s alpha. Test–retest reliability was assessed with intraclass correlation coefficients (ICCs) for absolute agreement for individual items and domains. Results PCA of the recall MDP was consistent with two domains (Immediate Perception, 7 items, Cronbach’s alpha = .89 to .94; Emotional Response, 5 items; Cronbach’s alpha = .81 to .85). Test–retest ICCs for the recall MDP during the ED visit ranged from .70 to .87 for individual items and were .93 and .94 for the Immediate Perception and Emotional Response domains. ICCs were much lower for the interval between the ED visit and follow-up, both for individual items (.28 to .66) and for the Immediate Perception and Emotional Response domains (.72 and .78, respectively). Conclusions During an ED visit, recall MDP ratings of dyspnea at the time participants decided to seek care in the ED are reliable and sufficiently stable, both for individual items and the two domains, that a time lag between arrival and questionnaire administration does not critically affect recall of perceptual and emotional characteristics immediately prior to the visit. However, test–retest reliability of recall over a 4- to 6-week interval is poor for individual items and significantly attenuated for the two domains.