Determining minimal important differences for patient-reported outcome measures in shoulder, lateral elbow, patellar and Achilles tendinopathies using distribution-based methods

Abstract

Background

Minimal important difference (MID) is a concept used inconsistently and arbitrarily in tendinopathy research. Our aim was to determine the MIDs for the most commonly used tendinopathy outcome measures using data-driven approaches.

Methods

Recently published systematic reviews of randomised controlled trials (RCTs) on tendinopathy management were identified and used for extraction of eligible studies through a literature search. Each eligible RCT was used to obtain information on MID where this was used and it also contributed data for the calculation of the baseline pooled standard deviation (SD) for each tendinopathy (shoulder, lateral elbow, patellar and Achilles). The rule of “half SD” was used for the computation of MIDs for patient-reported pain (visual analogue scale, VAS 0–10, single-item questionnaire) and function (multi-item questionnaires) and the rule of “one standard error of measurement (SEM)” was additionally used for the multi-item functional outcome measures.

Results

A total of 119 RCTs were included for the 4 tendinopathies. MID was defined and used by 58 studies (49%) and there were significant inconsistencies amongst studies where the same outcome measure was used as MID. From our data-driven methods the following suggested MIDs were obtained: a) Shoulder tendinopathy, pain VAS (combined) 1.3 points, Constant-Murley score 6.9 (half SD) and 7.0 (one SEM) points; b) lateral elbow tendinopathy, pain VAS (combined) 1.0 point, Disabilities of Arm, Shoulder and Hand questionnaire 8.9 (half SD) and 4.1 (one SEM) points; c) Patellar tendinopathy, pain VAS (combined) 1.2 points, Victorian Institute of Sport Assessment – Patella (VISA-P) 7.3 (half SD) and 6.6 points (one SEM); d) Achilles tendinopathy, pain VAS (combined) 1.1 points, VISA-Achilles (VISA-A) 8.2 (half SD) and 7.8 points (one SEM). The rules of half SD and one SEM produced very similar MIDs except for DASH due to its very high internal consistency. MIDs were also calculated for different pain settings for each tendinopathy.

Conclusions

Our computed MIDs can be used in tendinopathy research to increase consistency. Clearly defined MIDs should be used with consistency in tendinopathy management studies in the future.