Minimal important change of the knee injury and osteoarthritis outcome score in patients with mild to moderate knee osteoarthritis - using three different anchor-based methods

Henriette Killingrød Lundquist; Britt Elin Øiestad; Joseph Sexton; May Arna Risberg; Nina Østerås

doi:10.1016/j.ocarto.2025.100699

Minimal important change of the knee injury and osteoarthritis outcome score in patients with mild to moderate knee osteoarthritis - using three different anchor-based methods

Osteoarthr Cartil Open. 2025 Oct 29;7(4):100699. doi: 10.1016/j.ocarto.2025.100699. eCollection 2025 Dec.

Authors

Henriette Killingrød Lundquist¹, Britt Elin Øiestad², Joseph Sexton³, May Arna Risberg⁴, Nina Østerås³

Affiliations

¹ Department of Physiotherapy, Oslo Metropolitan University, Norway.
² Department of Rehabilitation Science and Health Technology, Oslo Metropolitan University, Norway.
³ Center for Treatment of Rheumatic and Musculoskeletal Diseases (REMEDY), Diakonhjemmet Hospital, Oslo, Norway.
⁴ Department of Sports Medicine, Norwegian School of Sport Sciences and Division of Orthopedic Surgery, Oslo University Hospital, Norway.

Abstract

Objective: To estimate Minimal Important Change (MIC) for improvement in the Knee injury and Osteoarthritis Outcome Score (KOOS) in patients with mild to moderate knee osteoarthritis (OA), using three recommended anchor-based methods, and examine how methodological choices influence these estimates.

Design: Secondary analysis of a three-arm randomized controlled trial. KOOS and a Global Rating of Change (GROC) scale were collected at baseline and 4-month follow-up. MIC values were estimated using predictive modeling, Mean Change, and Receiver Operating Characteristic (ROC) methods. Sensitivity analyses assessed the impact of different anchor cut-offs.

Results: Data were available for 131 patients undergoing non-surgical treatment (mean age 57.4 years, 50 % female). At follow-up, 19 % reported important improvement. Using the predictive modeling method, which allows adjustment for the low proportion of patients reporting important improvement, MICs were 11.3 (Pain), 12.1(Symptoms), 10.2 (ADL), 15.5 (Sport/Rec) and 13.2 (QoL). The Mean Change method yielded comparable MICs (range 10.6-16.1), but due to its reliance on a small subgroup, it is generally considered less robust and showed wider CIs in our sample. ROC-based MICs ranged from -0.4 to 12.5 and were associated with wide CIs, and high misclassification rates. Sensitivity analyses showed lower MICs with broader improvement definitions were used.

Conclusion: MIC estimates for KOOS varied considerably by method. Predictive modeling yielded the most precise MIC estimates and should be considered for future research, particularly when the proportion of improved patients deviates from 50 %. These results also highlight the importance of methodological transparency for interpreting PROMs in non-surgical knee OA treatment.

Keywords: Osteoarthritis; Patient reported outcomes; Statistical methods.