Development and psychometric validation of a patient-reported core symptom set for lung cancer patients undergoing chemotherapy

Background Patients with lung cancer receiving chemotherapy commonly experience multiple concurrent symptoms, yet many patient-reported outcome (PRO) instruments are either too lengthy or insufficiently targeted for high-frequency monitoring in routine care. We aimed to develop a brief, clinically meaningful core symptom set and evaluate its psychometric performance. Methodology In this single-centre prospective observational study, instrument development was conducted in 24 patients who completed the M.D. Anderson Symptom Inventory–Lung Cancer module (MDASI-LC) daily from day 1 to day 21 of one chemotherapy cycle. Symptoms were categorized as none (0), mild (1–3), moderate (4–6), or severe (7–10). Candidate high-burden symptoms were identified using descriptive distributions and hierarchical cluster analysis, and finalized by a 12-member expert panel using anonymous voting (consensus threshold: ≥75%). Psychometric evaluation was then performed in an independent cohort of 150 patients. Highest daily symptom scores from days 1–7 and 8–14 were used as two adjacent analytic windows, and the EQ-5D-5L visual analogue scale (VAS; day 7) was used as an external criterion. Reliability, validity, and measurement precision were assessed using classical test theory (CTT) and graded response item response theory (IRT) models. Results Six symptoms were retained: insomnia, fatigue, cough, nausea, constipation, and shortness of breath. The core set exhibited robust internal consistency (Cronbach’s α = 0.842), with item-deletion analyses confirming that the removal of any single item would result in a lower coefficient. Agreement across adjacent windows was acceptable (all ICCs ≥0.542, p<0.001). Exploratory factor analysis supported unidimensionality, with one factor explaining 56.5% of variance. Higher core symptom scores were moderately associated with lower EQ-5D-5L VAS scores, supporting convergent validity. Scores decreased from week 1 to week 2 with small-to-moderate effect sizes (Cohen’s d=0.217–0.628), supporting responsiveness. Item response theory (IRT) analyses demonstrated satisfactory item discrimination, ordered thresholds, and optimal measurement precision at mild-to-moderate symptom burden. Conclusions This six-item core symptom set demonstrates reliable psychometric performance and clinical feasibility, suggesting its potential to support high-frequency PRO-based symptom monitoring for patients with lung cancer receiving chemotherapy in routine care. Multicentre external validation is warranted.