Purpose <p>Many clinical trials of metastatic non-small cell lung cancer (mNSCLC) include a composite endpoint of cough, chest pain and dyspnea to assess patient-reported symptomatic deterioration using time-to-confirmed deterioration (TTCD) analyses. We sought to determine the robustness of this composite, and whether composites involving other patient-relevant symptoms of mNSCLC are more sensitive to progression.</p> Methods <p>Data from nine mNSCLC trials with the EORTC QLQ-C30 and QLQ-LC13 were pooled. Spearman's correlation coefficients were used to assess associations between symptoms. Composites were evaluated based on the number of confirmed deterioration (CD) events, median TTCD, and positive predictive value (PPV) against progression-free survival (PFS).</p> Results <p>5,157 patients with mNSCLC with at least one PRO assessment were included. Cough and dyspnea were frequently reported at baseline (79.8% and 67.8%, respectively) and moderately correlated (0.37), while chest pain was less common (41.5%) and showed weaker associations. The composite of cough, chest pain, and dyspnea had a median TTCD of 8 three-week treatment cycles and PPV of 78.3%. Composites including symptoms such as fatigue and general pain reached median TTCD earlier while maintaining high PPVs (78.3–81.3%). The median TTCD for single-item dyspnea was 50 cycles but was not reached for cough and chest pain.</p> Conclusions <p>The composite of cough, chest pain, and dyspnea displayed strong properties; however, composites obscure the time course of included symptoms. Although median TTCD is reached earlier with a composite endpoint, greater clarity is obtained when symptoms are studied individually.</p>