Limitations of Minimal Clinically Important Difference Estimates and Potential Alternatives

Letter to the Editor: Defining Patient-relevant Thresholds and Change Scores for the HOOS JR and KOOS JR Anchored on the Patient-acceptable Symptom State Question

“…We have written on the limitations of MCIDs and PASS [13][14][15] and have elaborated on the many important weaknesses of MCIDs, many of which also, in our opinion, apply to PASS measures. For example, we wrote that MCIDs rely on arbitrary cut points; ignore random measurement error; demonstrate level of measurement bias; are subject to recall/present state bias; do not consider intervention side effects, inconveniences, or costs; and do not capture change with versus without a treatment alternative [14]. Additionally, MCIDs and PASS thresholds are non-falsifiable [14,19,20].…”

“…For example, we wrote that MCIDs rely on arbitrary cut points; ignore random measurement error; demonstrate level of measurement bias; are subject to recall/present state bias; do not consider intervention side effects, inconveniences, or costs; and do not capture change with versus without a treatment alternative [14]. Additionally, MCIDs and PASS thresholds are non-falsifiable [14,19,20]. That is, MCIDs and PASS scores cannot be subjected to formal hypothesis testing to determine which of the many estimates for both hip and knee arthroplasty are truly incorrect [1,2,11].…”

“…The authors concluded their study by stating: "…we suggest that routine collection of a single postoperative PASS may be an easier way to measure success than the routine collection of preoperative and postoperative PROMs" [3]. Given the psychometric gaps in MCID and PASS measures discussed in this letter and elsewhere [8][9][10][13][14][15][16], we believe this recommendation is ill advised.…”