Factors Contributing to Variability in Longitudinal Pain Scores in Osteoarthritis Randomised Clinical Trials: A...

Osteoarthritis is a prevalent joint disease causing pain and functional decline, but high variability in patient-reported treatment outcomes in clinical trials challenges detection of treatment effects. This systematic review aimed to evaluate variability in pain reporting and identify trial characteristics, placebo comparator attributes, and participant clinical characteristics associated with this variability.

A systematic review of placebo-controlled RCTs for pharmacological interventions in knee osteoarthritis was conducted, adhering to PRISMA guidelines and registered at Open Science Framework. MEDLINE and EMBASE databases were searched for trials from 2010 to October 2024. The primary outcome was variability of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain subscale. A random-effects meta-analysis estimated pooled variability of WOMAC pain change.

From 1318 records, 87 reports involving 36,246 participants were included. The overall mean variability (SD) of WOMAC pain change was 20.7 on a normalised 0-100 scale, with an absolute mean change of 17.3. Variability was linked to trial size, number of sites, study arms, and baseline pain scores. Smaller studies and those with fewer sites showed lower variability. Topical and intra-articular placebo administration were associated with lower variability. Baseline BMI and pain scores influenced variability, while age and sex did not.

This review highlights significant variability in WOMAC pain reporting in osteoarthritis trials, affecting statistical power and trial design. Key factors influencing variability include trial design, administration route, and participant characteristics. Integrating these variability estimates into sample-size calculations can enhance the efficiency of future pain trials.

This first systematic quantification of variability in WOMAC pain across knee osteoarthritis trials shows substantial heterogeneity. Variability directly alters the detectable magnitude of treatment effect and trial sensitivity, and by shifting attention from mean treatment effects to outcome dispersion as a modifiable design element, this critical and insufficiently characterised driver of trial sensitivity provides a new methodological lens for pain trials. The resulting empiric variability benchmarks enables better study design planning, interpretability, and support more efficient development of new therapies.