Can Machine Learning Algorithms Predict Which Patients Will Achieve Minimally Clinically Important Differences From Total Joint Arthroplasty? :

Objectives To present estimates of clinically meaningful or minimal important changes for the Oxford Hip Score (OHS) and the Oxford Knee Score (OKS) after joint replacement surgery. Study Design and Setting Secondary data analysis of the NHS patient-reported outcome measures data set that included 82,415 patients listed for hip replacement surgery and 94,015 patients listed for knee replacement surgery was performed. Results Anchor-based methods revealed that meaningful change indices at the group level [minimal important change (MIC)], for example in cohort studies, were ∼11 points for the OHS and ∼9 points for the OKS. For assessment of individual patients, receiver operating characteristic analysis produced MICs of 8 and 7 points for OHS and OKS, respectively. Additionally, the between group minimal important difference (MID), which allows the estimation of a clinically relevant difference in change scores from baseline when comparing two groups, that is, for clinical trials, was estimated to be ∼5 points for both the OKS and the OHS. The distribution-based minimal detectable change (MDC90) estimates for the OKS and OHS were 4 and 5 points, respectively. Conclusion This study has produced and discussed estimates of minimal important change/difference for the OKS/OHS. These estimates should be used in the power calculations and the interpretation of studies using the OKS and OHS. The MDC90 (∼4 points OKS and ∼5 points OHS) represents the smallest possible detectable change for each of these instruments, thus indicating that any lower value would fall within measurement error.

To study responsiveness and establish the minimal clinically important differences (MCIDs) and minimal detectable change (MDC) in patients undergoing total knee replacement (TKR) using the Short Form 36 (SF-36) and Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC). Prospective observational study in three public hospitals of all consecutive patients on waiting lists to undergo TKR intervention with diagnosis of knee osteoarthritis (OA). Patients were asked to complete before the intervention and at 6 months and 2 years afterward the SF-36 and the WOMAC health-related quality of life questionnaires (HRQoL), and additional transition questions which measured the changes in their joint at 6 months. In both questionnaires the possible range of values is from 0 to 100 points. In WOMAC improvement at 6 months after a TKR was between 27 (stiffness) and 31 points (pain). The SF-36 showed improvements between the 28.3 points of role physical and 2.79 of general health. From 6 months to 2 years, WOMAC improvements were between 2 and 6 points. The MCID ranged from 14.52 (stiffness) to 22.87 (pain) on the WOMAC and in the physical domains of SF-36 from 11.56 (physical function) to 16.86 (bodily pain). On the WOMAC, the MDC ranged from 13.11 (function) to 29.12 (stiffness), and on SF-36 from 19.50 (physical function) to 41.23 (social functioning). The MCID for TKR is around 15 on WOMAC, while with the SF-36 of at least 10 points. These values should not be considered as absolute thresholds.

The minimal clinically important difference (MCID) for an instrument is a much sought after, but elusive figure. In this review we will highlight new findings in this area, including taxonomy of MCID, methods used to ascertain MCID, the perspective taken for evaluating importance, and other sources of variation for MCID values. In the end we believe the MCID will be a context-specific value rather than a fixed number. The review highlights the need to do methodological research in this area, especially concurrent comparisons between approaches, or across different patient groups. There are many faces to the MCID, it is not a simple concept, nor simple to calculate.