Scoping review of response shift methods: current reporting practices and recommendations.

Area of Science:

• Health Outcomes Research
• Psychometrics
• Longitudinal Data Analysis

Background:

• Response shift (RS) refers to changes in self-evaluation meaning for health and quality of life.
• Statistical methods exist to detect RS in longitudinal data.
• Reviewing the application of these methods in patient-reported outcomes (PRO) literature is crucial.

Purpose of the Study:

• To review the uptake of statistical methods for detecting response shift in PRO literature.
• To identify trends in RS detection method usage and assumption testing.

Main Methods:

• Comprehensive literature search across multiple databases (CINAHL, EMBASE, Medline, ProQuest, PsycINFO, Web of Science) up to 2016.
• Extracted data included publication details, PRO measures, RS detection methods, RS types, and assumption testing.
• Systematic review of 101 included articles from 1032 identified.

Main Results:

• Only 9.8% of identified articles were included, indicating a niche research area.
• The Then-test was most frequently reported (54.5%), followed by structural equation modeling (SEM) (30.7%).
• Newer methods like relative importance analysis and random forest regression were infrequently used, and <25% reported assumption testing.

Conclusions:

• Despite methodological progress, there's a need for more research on RS detection for complex longitudinal data.
• Standardized reporting guidelines for RS methods in PRO research are recommended.
• Further investigation into the application and validation of advanced RS detection techniques is warranted.