ISOQOL 2023

Meet up with our experts in Calgary

Amy M. Barrett, MSPH, MA

Senior Director, Patient-Centered Outcomes Assessment

Contact

Marci Clark, PharmD

Senior Director, Patient-Centered Outcomes Assessment

Contact

Dana DiBenedetti, PhD

Vice President, Patient-Centered Outcomes Assessment

Contact

Lynda Doward, MRes

VP, European Head Patient-Centered Outcomes Assessment

Contact

Lori McLeod, PhD

Vice President, Patient-Centered Outcomes Assessment

Contact

Nicholas Rockwood, PhD

Senior Psychometrician, Patient-Centered Outcomes Assessment

Contact

Shanshan Qin, PhD

Director, Psychometrics, Patient-Centered Outcomes Assessment

Contact

Video Transcript

When using clinician-reported outcomes within clinical trials, it is important to ensure the ratings provided by the clinicians (or raters) are reliable. Interrater reliability can be quantified by computing the intraclass correlation coefficient, or ICC. The rater effects can be treated as either random or fixed when computing the ICC.

Traditionally, the rater effects are treated as random when researchers wish to generalize the reliability findings to a much larger population of raters. When interest only lies in the specific raters within the sample, then it is appropriate to incorporate raters into the model as a fixed effect. In this study, we introduce a third option: A finite population ICC, which is useful when the researcher wishes to generalize to some relatively small population of potential raters. This ICC could be used, for example, if a clinical trial uses 5 raters, but the corresponding interrater reliability study only uses ratings provided by 3 of these raters.

An agreement-based ICC for estimating interrater reliability in a finite population of raters

Session: Thursday Poster Session II
Presentation date and time: 10/19/2023 at 3:20 pm
Presenting author: Nicholas Rockwood

Video Transcript

Hi, my name is Dana DiBenedetti, and I’m Vice President of Patient-Centered Outcomes Assessment at RTI Health Solutions. Our experts will be discussing design considerations for successful in-trial interviews at the ISOQOL Symposium on novel approaches to capturing patient experience in quality-of-life research at ISOQOL 2023.

The inclusion of study participant interviews within clinical trials is an evolving field of research. Interview data collected from patients or caregivers can supplement, support, and facilitate the interpretation of clinical trial data. However, one size does not fit all for studies that include in-trial interviews. This symposium presentation will focus on factors critical in the design of qualitative interviews in clinical trial settings, including the selection of appropriate methods to address the study objectives and operational issues critical to the success of such studies.

Please join us at ISOQOL to learn more about key considerations needed to plan for successfully commissioning or implementing in-trial interviews.

We hope to see you there.

Design considerations for successful in-trial interviews: the who, what, when, where, and how

Presentation date and time: 10/21/23 at 8:05 am
Presenter: Dana DiBenedetti

Video Transcript

The ability to measure the responsiveness, or the sensitivity to changes in scores, of a clinical outcomes assessment, or COA, is a key step to determining whether it is fit for purpose. Ordinal COAs, such as ratings of symptom severity, are commonplace in clinical trials. The observed response to an ordinal COA can typically be viewed as a coarse measurement of the underlying, continuous construct being measured, and Polychoric correlations can be used to estimate the relationships among these underlying constructs.

However, when interest is in the correlations among change scores, such as when assessing the responsiveness of the measure or the relationship between a measure and a global anchor, neither Polychoric nor Pearson correlations are appropriate because change scores for ordinal measures are no longer ordinal, or continuous. Within this project, we demonstrate that the standard practice of computing correlations for ordinal change scores results in an underestimation of the true relationship between changes on the underlying constructs being measured. We also propose a new method, which relies upon the bivariate relationships among each measure at each time point, to accurately estimate the correlations between true changes on the underlying constructs.

Estimating change score/responsiveness correlations among ordinal measures

Session: Oral Brief Session 202: Methodological Work
Session date and time: 10/20/23 at 1:20 pm
Presenting author: Nicholas Rockwood

ISOQOL 2023

30 Years of ISOQOL: Quality of Life – making it relevant.

Meet up with our experts in Calgary