Choice of control group treatments in hand osteoarthritis trials: A systematic review and meta-analysis

To assess how patient characteristics and study design influence the effectiveness of control interventions in hand OA trials.

The study protocol was registered in PROSPERO (CRD42020163473). Two authors independently searched four electronic databases from their inception to December 31, 2019. Randomized and non-randomized controlled hand OA trials were included if pain intensity was assessed using a validated scale. We allocated control groups into one of the following: placebo, add-on treatment, no treatment, or active treatment. The standardized mean differences (d) of pain, as well as subjective function and hand strength, were pooled with 95% confidence intervals (CI) and 90% prediction intervals using random-effects models. Meta-regression and post-hoc subgroup analyses were performed to investigate which factors potentially impacted placebo analgesia and between-study heterogeneity.

Thirty-one placebo, 11 add-on, 12 no-treatment, and 10 active-treatment controls were included in meta-analyses. Effective pain relief was observed in placebo (d = -0.50, 95% CI -0.63 to -0.37), add-on (d = -0.35, 95% CI -0.59 to -0.12), and active-treatment (d = -0.92, 95% CI -1.35 to -0.48) groups. In subjective function, these treatments had smaller but beneficial effects; hand strength, contrastingly, was not improved. Placebo effects were larger when flare designs were used (d = -0.96) and more homogeneous when minimum pain thresholds were set (d = -0.46, 90% prediction intervals -0.79 to -0.14).

Placebo, add-on, and active control treatments were more effective than the no treatment control in relieving hand pain and improving subjective function. By choosing minimum pain thresholds and flare requirements at patient enrollment, moderate pain relief may be replicated among control participants in future randomized placebo-controlled trials.