Statistical Analysis of Patient-Reported Outcomes in Clinical Trials

Clinicians, researchers, funding agencies, regulatory agencies, and patients have long acknowledged the importance of patient-reported outcomes (PROs) in clinical trials. PROs refer to data provided directly by patients regarding their perceived health; presence, frequency, or severity of symptoms; health-related quality of life (HRQoL); or treatment satisfaction. Although definitional differences exist in the literature, HRQoL generally refers to the impact of disease, treatment, or perceived health on daily functioning. HRQoL and other PROs are measured by a single item or multiple items on questionnaires administered during clinic visits or completed by patients between clinic visits via various modes of administration including by paper, Internet-based survey, handheld device, mobile device application, automated telephone system, or interviewer (in-person or over the phone). PROs enhance clinicians’ and researchers’ understanding of patients’ experiences before, during, and after treatment, particularly when these experiences are difficult or impossible to observe. When designing clinical trials, researchers should select PROs that are appropriate for the patient population of interest and that have established and acceptable psychometric (i.e., measurement) properties. When carrying out clinical trials, missing data should be prospectively minimized. Although this chapter describes some considerations specific to PROs, most analysis principles that apply to other endpoints also apply to PROs. As with other endpoints, researchers should select analyses that match their hypotheses and data characteristics, consider multiplicity issues, and properly handle missing data. Clinicians and researchers should also assess the clinical significance of the results.