Abstract
Patient‐reported outcome (PRO) data present unique statistical challenges that may not be encountered with other types of endpoints used in randomized clinical trials. Statisticians, with their expertise in the design of clinical trials and the analysis of clinical trial data, contribute to the science of how studies can use PRO endpoints to provide data that are intended to support approval of medical products and PRO labeling claims. This article provides a brief regulatory history of PROs and highlights topics that statisticians who design and analyze clinical studies with PRO endpoints need to consider.
We discuss study design features that may help improve the likelihood of labeling claims for PRO data. Specification of the scientific question and estimand, definition of PRO endpoint, collection of data that aid PRO analyses and interpretation, and multiplicity adjustments are critical elements. In addition, PRO data have distinctive missing data considerations because PRO scores are typically constructed from many items and, often, from daily diary data. We discuss how missing items and missing diary days can affect efficacy analyses. Finally, we provide considerations for the interpretation of PRO results and suggest topics for further statistical research.
We discuss study design features that may help improve the likelihood of labeling claims for PRO data. Specification of the scientific question and estimand, definition of PRO endpoint, collection of data that aid PRO analyses and interpretation, and multiplicity adjustments are critical elements. In addition, PRO data have distinctive missing data considerations because PRO scores are typically constructed from many items and, often, from daily diary data. We discuss how missing items and missing diary days can affect efficacy analyses. Finally, we provide considerations for the interpretation of PRO results and suggest topics for further statistical research.
Original language | English |
---|---|
Type | Entry in Wiley StatsRef: Statistics Reference Online |
Media of output | Online |
Publisher | Wiley |
DOIs | |
Publication status | E-pub ahead of print - 15 May 2017 |
Keywords
- patient-reported outcomes
- longitudinal data
- missing data
- interpretation
- clinical trials