Objectives: The European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Group is developing a computer-adaptive test (CAT) version of the EORTC Quality of Life Questionnaire (QLQ-C30). We evaluated the measurement properties of the CAT versions of physical functioning (PF) and fatigue (FA) and compared these with the corresponding QLQ-C30 scales.
Study design and setting: Based on international samples of more than 1,000 cancer patients, we simulated CAT administration of varying numbers of items and compared the resulting scores with those based on all items in the respective item pools. Furthermore, the relative validity (RV) of CATs was compared with that of the QLQ-C30 scales using known groups validity.
Results: For both dimensions, CATs of all lengths resulted in unbiased score estimates. CATs consisting of five or more items had reliability>0.90, correlated ≥ 0.97 with the full scale, and had root mean square error <0.25. The average RVs for these CATs ranged 1.02-1.33, indicating possible savings in sample size requirements of 3-42% using CAT.
Conclusion: The CAT versions of PF and FA exhibited high levels of measurement precision and efficiency. The potential savings in sample size requirements using CATs compared with those using the original QLQ-C30 scales were typically 20% or more.
Copyright © 2013 Elsevier Inc. All rights reserved.