Research
Print page Print page
Switch language
The Capital Region of Denmark - a part of Copenhagen University Hospital
Published

Assessing baseline dependency of anchor-based minimal important change (MIC): don't stratify on the baseline score!

Research output: Contribution to journalJournal articleResearchpeer-review

  1. The longitudinal validity of proxy-reported CHU9D

    Research output: Contribution to journalJournal articleResearchpeer-review

  2. Development and content validation of a questionnaire measuring patient empowerment in cancer follow-up

    Research output: Contribution to journalJournal articleResearchpeer-review

  3. International validation of the EORTC CAT Core: a new adaptive instrument for measuring core quality of life domains in cancer

    Research output: Contribution to journalJournal articleResearchpeer-review

View graph of relations

PURPOSE: The minimal important change (MIC) of a patient-reported outcome measure (PROM) is often suspected to be baseline dependent, typically in the sense that patients who are in a poorer baseline health condition need greater improvement to qualify as minimally important. Testing MIC baseline dependency is commonly performed by creating two or more subgroups, stratified on the baseline PROM score. This study's purpose was to show that this practice produces biased subgroup MIC estimates resulting in spurious MIC baseline dependency, and to develop alternative methods to evaluate MIC baseline dependency.

METHODS: Datasets with PROM baseline and follow-up scores and transition ratings were simulated with and without MIC baseline dependency. Mean change MICs, ROC-based MICs, predictive MICs, and adjusted MICs were estimated before and after stratification on the baseline score. Three alternative methods were developed and evaluated. The methods were applied in a real data example for illustration.

RESULTS: Baseline stratification resulted in biased subgroup MIC estimates and the false impression of MIC baseline dependency, due to redistribution of measurement error. Two of the alternative methods require a second baseline measurement with the same PROM or another correlated PROM. The third method involves the construction of two parallel tests based on splitting the PROM's item set. Two methods could be applied to the real data.

CONCLUSION: MIC baseline dependency should not be tested in subgroups based on stratification on the baseline PROM score. Instead, one or more of the suggested alternative methods should be used.

Original languageEnglish
JournalQuality of life research : an international journal of quality of life aspects of treatment, care and rehabilitation
Volume30
Issue number10
Pages (from-to)2773-2782
Number of pages10
ISSN0962-9343
DOIs
Publication statusPublished - Oct 2021

Bibliographical note

Publisher Copyright:
© 2021, The Author(s).

Copyright:
Copyright 2021 Elsevier B.V., All rights reserved.

    Research areas

  • Baseline dependency, Mean change method, Minimal important change, Minimal important difference, Predictive modeling method, ROC method

ID: 65796241