Comparison of anchor-based methods for estimating thresholds of meaningful within-patient change using simulated PROMIS PF 20a data under various joint distribution characteristic conditions

作     者：Qin, Shanshan Nelson, Lauren Williams, Nicole Williams, Valerie Bender, Randall McLeod, Lori 

作者机构：RTI Hlth Solut 3040 Cornwallis Rd Res Triangle Pk NC 27709 USA ICF Int 1902 Reston Metro Plaza Reston VA 20190 USA 

出 版 物：《QUALITY OF LIFE RESEARCH》 (生命特质研究)

年 卷 期：2023年第32卷第5期

页      面：1277-1293页

核心收录：

学科分类：12[管理学] 1204[管理学-公共管理] 120402[管理学-社会医学与卫生事业管理(可授管理学、医学学位)] 1004[医学-公共卫生与预防医学(可授医学、理学学位)] 10[医学] 

基　　金：RTI Health Solutions 

主　　题：Patient-reported outcome Distribution-based method Anchor-based method Responder definition Meaningful change Minimal important difference MID Minimal clinically important difference MCID 

摘      要：Purpose To compare the performance of anchor-based methods for estimating thresholds of meaningful within-patient change (i.e., individual change) of clinical outcome assessments in conditions reflecting data characteristics of small- to medium-sized clinical trials. Methods Datasets were generated from the joint distributions of the PROMIS PF 20a T-score changes and a seven-point global change anchor measure. The 108 simulation conditions (1000 replications per condition) included combinations of three marginal distributions of T-score changes, three improvement percentages in the anchor measure, four levels of responsiveness correlations, and three sample sizes. Threshold estimation methods included mean change, median change, ROC curve, predictive modeling, half SD, and SEM. Relative bias, precision, accuracy, and measurement significance of the estimates were evaluated based on comparison with true thresholds and IRT-based individual reliable changes of PROMIS scores. Quantile regression models were applied to select and interpret effects of simulation conditions on estimation bias. Results When PROMIS T-score changes were distributed normally, the predictive modeling method performed best with 50% or more responders identified by the anchor;the mean and median methods were preferred with 30% responders. For skewed distributions, the median method and ROC method gained more advantages. Among the evaluated study conditions, the improvement percentage condition had the most obvious effects on estimation bias. Conclusion To establish accurate and precise thresholds, clinical researchers are recommended to prioritize study designs with at least 50% anchor-defined responders and strongly responsive target endpoints with highly reliable scoring calibration and to select optimal anchor-based methods given the data characteristics.