An extension minimal important difference credibility item addressing construct proximity is a reliable alternative to the correlation item

Yuting Wang, Tahira Devji, Alonso Carrasco-Labra, Anila Qasim, Qiukui Hao, Elena Kum, Niveditha Devasenapathy, Madeleine T. King, Berend Terluin, Caroline B. Terwee, Michael Walsh, Toshi A. Furukawa, Yasushi Tsujimoto, Gordon H. Guyatt

Research output: Contribution to journalArticleAcademicpeer-review

1 Citation (Scopus)

Abstract

Objectives: Minimal important difference (MID), the smallest change or difference that patients perceive as important, aids interpretation of change in patient-reported outcome measure (PROM) scores. A credibility instrument that assesses the methodological rigor of an anchor-based MID includes one core item addressing the correlation between the PROM and the anchor. However, the majority of MID studies in the literature fail to report the correlation. To address this issue, we extended the anchor-based MID credibility instrument by adding an item addressing construct proximity as an alternative to the correlation item. Study Design and Setting: Informed by an MID methodological survey, we added an alternative item—a subjective assessment of similarity of the constructs (i.e., construct proximity) between PROM and anchor—to the correlation item and generated principles for the assessment. We sampled 101 MIDs and analyzed the assessments performed by each pair of raters. By calculating weighted Cohen's kappa, we assessed the reliability of the assessments. Results: Construct proximity assessment is based on the anticipated association between the anchor and PROM constructs: the closer the anticipated association, the higher the rating. Our detailed principles address the most frequently used anchors: transition ratings, measures of satisfaction, other PROMs, and clinical measures. The assessments showed acceptable agreement (weighted kappa 0.74, 95% CI 0.55–0.94) between raters. Conclusion: In the absence of a reported correlation coefficient, construct proximity assessment provides a useful alternative in the credibility assessment of anchor-based MID estimates.
Original languageEnglish
Pages (from-to)46-52
Number of pages7
JournalJournal of Clinical Epidemiology
Volume157
DOIs
Publication statusPublished - 1 May 2023

Keywords

  • Anchor-based minimal important difference
  • Construct proximity
  • Correlation
  • Credibility assessment
  • Patient-reported outcome measure
  • Reliability

Cite this