Assessment of Dyspnea in Asthma: Validation of the Dyspnea-12

Abstract

Background. Dyspnea is a prominent symptom in asthma. The Dyspnea-12 (D-12), an instrument that quantifies breathlessness using 12 descriptors that tap the physical and affective aspects, has shown promise for the measurement of dyspnea in cardiorespiratory disease. Objective. We report the results of a study designed to test the validity and reliability of the D-12 in a population of patients with asthma. Methods. This cross-sectional study included 102 patients with asthma. Subjects completed the D-12, Hospital Anxiety and Depression scale, St. George’s Respiratory Questionnaire (SGRQ), and Medical Research Council scale. Confirmatory factor analysis confirmed the two-component structure of the D-12 (i.e., seven items that tap the physical aspects of breathlessness and five items that tap the affective aspects). Results. The D-12 subscales had excellent internal reliability (Cronbach’s alpha for the “physical” score was 0.94 and the affective score was 0.95). The D-12 physical component was more strongly correlated with SGRQ Symptoms (r = 0.648), SGRQ Activities (r = 0.635) and Medical Research Council grade (r = 0.636), while the affective component was more strongly correlated with SGRQ Impacts (r = 0.765) and Hospital Anxiety and Depression scale scores (anxiety r = 0.641 and depression r = 0.602). Conclusion. This study supports validity of the D-12 for use in the assessment of dyspnea of patients with asthma. It assesses one of the most pertinent symptoms of asthma from two viewpoints—physical and affective.

• adults
• asthma
• dyspnea
• patient-reported outcome measures
• quality of life
• symptoms

Acknowledgments

The authors are grateful to all patients at Pennine Acute Trust who took time to take part in this study. We are grateful to Dr. David Weir and Dr. Jon Miles for assisting with patient recruitment and thoughtful discussions during the conduct of this study.

This study was funded by Action Medical Research UK (SP4244). Dr. J. Swigris is supported in part by a Career Development Award from the NIH (K23 HL092227).

Author contributions: Study conceptualization and design: J. Yorke, P.W. Jones, C. Haigh, C. Shuldham; statistical analyses: J. Yorke, J. Swigris; manuscript preparation: J. Yorke, A.-M. Russell, N. Rochnia, J. Swigris, C. Haigh, C. Shuldham, J. Hoyle, P.W. Jones.

Declaration of Interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

Online Supplement—Dyspnea-12 in asthma

Methods for Confirmatory Factor Analysis

Confirmatory Factor Analysis (CFA). CFA is used to examine the relationship hypothesized to exist between observed variables and their underlying latent constructs (i.e., factor structure). In this study, the latent constructs would be the physical and affective perceptions of dyspnea (1). This is in contrast to exploratory factor analysis (EFA) which involves an “orderly simplification of interrelated measures” in which there is no preconceived or hypothesized structure of how the variables relate (1). CFA requires an a priori specification of the model, and the analysis tests how well the data fit this model. It is recommended that 5–20 cases per variable (i.e., questionnaire item) be used to perform a CFA.

CFA involves a battery of tests to determine the adequacy of model fit to the data. Results may be discordant, but fit is considered acceptable if the majority of tests support adequate fit to the model. The statistics that are often used to determine model fit with CFA include the following: (1) chi-square—assesses the difference between observed and expected covariance matrices, and values close to 0 indicate acceptable fit; (2) the comparative fit index (CFI) is the same as the discrepancy function adjusted for sample size, and larger values (closer to 1 and ≥0.90) indicate acceptable fit; (3) root mean square error of approximation (RMSEA) relates to the residual of the model, and smaller values (close to 0 and <0.06) indicate acceptable fit; (4) Bentler and Bonnett’s normed fit index (NFI) (2) tests the null hypothesis that the model is one in which all of the correlations or covariances are 0, and values ≥0.90 indicate acceptable fit; and (5) Bentler and Bonnett’s non-normed index (NNI) touted to better reflect fit at all sample sizes, with values ≥0.90 indicating acceptable fit (2).

We used PROC CALIS in SAS version 9.1.3 (SAS, Inc., Cary, NC, USA) to test the hypothesized two-factor structure (items 1–7 would load on a factor describing the latent variable “physical perceptions of dyspnea”; items 8–12 would load on a factor describing the latent variable “affective perceptions of dyspnea”).

Results

(i) Manifest variable equations with estimates.

Table

(ii) Variances of exogenous variables.

Table

(iii) Covariances among exogenous variables.

Table

(iv) Correlations among exogenous variables.

Table

(v) Fit statistics.

Table