Objective: Sample size (N) affects the robustness of psychometric results, but for evaluations of patient-reported outcome (PRO) measures, N is often a compromise between timelines and resources. Currently, there are no psychometric N guidelines or requirements for the development of PROs for use in clinical trials. The objective of this study is to review current N practices by conducting a systematic literature review of the psychometric methods and N choices made in the evaluation of PRO measures (for use in clinical trials) over the past 10 years.
Methods: This systematic literature review included abstracts that described the psychometric evaluation of PRO measures that were likely developed for use in clinical trials. The review included English-language journal abstracts published in the past 10 years and identified in PubMed. Characteristics of each study were tabulated including the number of items and dimensions in each PRO of interest, the psychometric methods employed (e.g., internal consistency, test-retest reliability, factor analysis, responsiveness), and N.
Results: The literature search yielded 252 abstracts describing studies conducted mostly in Europe, Canada, and the United States. Preliminary results indicate that Ns ranged from approximately 40 to 4,000. The most frequently reported psychometric method was Cronbach’s alpha to quantify internal consistency. Approximately thirty percent of studies employed methods that demand the largest Ns such as item response theory (IRT) and factor analysis (FA). Ns for studies using IRT or FA analysis ranged from approximately 100 to 4,000 participants.
Conclusions: A wide range of Ns were employed for the psychometric evaluation of PROs developed for use in clinical trials. Ideally, researchers should consider the complexity of the PRO measure, its intended use, and the purpose of the evaluation when deciding on a study N. Additional studies should work toward developing best practices for PRO N guidelines in clinical trials.
