Measurement of quality of life in patients with lung cancer in multicenter trials of new therapies. Psychometric assessment of the lung cancer symptom scale

This study continued the development and psychometric testing of the Lung Cancer Symptom Scale (LCSS), a disease- and site-specific instrument primarily measuring the physical and functional dimensions of quality of life for individuals with lung cancer. The instrument contains two scales, one for patients and a counterpart for health professionals as observers.Feasibility, reliability, construct validity, and criterion-related validity were evaluated with 207 patients with non-small cell lung cancer (NSCLC) from six cancer centers. Within an interview with an observer, patients completed part of a battery of instruments by self-report and were interviewed for the remaining measures. Observers also completed measures after the interview.Feasibility, reliability, and validity were well supported for this lung cancer population. Feasibility was demonstrated by patient and staff compliance in completion at all six cancer centers. Internal consistency was good, with coefficient alphas of 0.82 for the patient scale and 0.75 for the observer scale. Construct validity was supported by 1. contrasted groups approach: regression lines (with 95% confidence bands) were obtained between the Karnofsky performance scale (KPS) and each of the two LCSS scales; 2. as a refinement, relationship testing: significant correlations between the LCSS and KPS for each item (except hemoptysis for the patient scale); and 3. multitrait-multimethod approach: good reliability (alphas ranging from 0.75 to 0.93), good convergent validity for the two LCSS scales (r = 0.77), and a good discriminant validity pattern from the Brief Symptom Inventory (BSI). Criterion-related validity with relevant gold standard measures (American Thoracic Society Questionnaire [ATS] and McGill Pain questionnaire, KPS, Profiles of Mood States [POMS], and Sickness Impact Profile [SIP]) was supported with significant correlations (0.40-0.67 for the LCSS patient scale; 0.54-0.65 for the LCSS observer scale).These psychometric properties demonstrate that the LCSS patient and observer scales are feasible, reliable, and valid quality of life measures that are ready for research and clinical use with lung cancer populations.