ABSTRACT: Chronic obstructive pulmonary disease (COPD) is characterized by non-reversible airflow limitation. A common symptom of COPD is dyspnea or shortness of breath. Dyspnea may vary daily, with a large impact on patients' lives. Previous clinical trials used patient-reported outcome (PRO) measures that quantified dyspnea at discrete intervals and hence did not reflect this variability. Recently the Shortness of Breath with Daily Activities (SOBDA) questionnaire was developed as a PRO measure of dyspnea utilizing a daily diary. This confirmatory post hoc meta-analysis of SOBDA data from a large clinical study program further supports the questionnaire and clarifies the minimum threshold of SOBDA response.Data from four clinical trials (DB2113361, NCT01313637; DB2113373, NCT01313650; DB2113360, NCT01316900; DB2113374, NCT01316913) were analyzed. These 24-week trials were randomized, blinded studies investigating the efficacy and safety of several COPD treatments. These post hoc analyses focused on the SOBDA questionnaire properties. This electronic-diary consists of 13 items completed daily, in which patients rate their breathlessness level during common daily activities. Resultant SOBDA scores were compared with related, commonly used assessments: modified Medical Research Council Research Dyspnea Scale (mMRC), Baseline Dyspnea Index (BDI), Transition Dyspnea Index (TDI), St George's Respiratory Questionnaire (SGRQ), COPD Assessment Test (CAT), and trough forced expiratory volume in 1 s (FEV1). The consistency, reliability, validity (convergent, known groups), and responsiveness of the SOBDA questionnaire was assessed.In total, 4967 patients with COPD provided data for these analyses. The SOBDA questionnaire had high internal consistency (Cronbach's alpha?=?0.936), high test-retest reliability (Pearson's correlation coefficient?=?0.86) and convergent validity with related measures (SGRQ total score, Pearson's correlation coefficient?=?0.59; CAT, Spearman rank-order correlation coefficient?=?0.50). SOBDA scores were statistically significantly lower in responders (as defined by TDI, SGRQ, CAT, and trough FEV1 levels) versus non-responders (p?