ABSTRACT
OBJECTIVE: To examine the pain relief effects of comparators (placebos and untreated control groups) in hand osteoarthritis trials and the impact of contextual factors. METHODS: We systematically searched PubMed, EMBASE and CENTRAL from inception to December 26, 2021. We included randomised controlled trials of people with hand osteoarthritis with a placebo or an untreated control group. We assessed the Risk of Bias with Cochrane Risk-of-Bias tool version 2. Each comparator was contrasted with a null-arm, imputed as having a zero change from baseline with the same standard deviation as the comparator. We combined the standardised mean differences with a random effects meta-analysis. The contextual factors' effect was explored in meta-regression and stratified models with pain as the dependent variable. RESULTS: 84 trials (7262 participants) were eligible for quantitative synthesis, of which 76 (6462 participants) were eligible for the stratified analyses. Placebos were superior to their matched null-arms in relieving pain with an effect size of -0.51 (95% confidence interval -0.61 to -0.42), while untreated control groups were not. When analysing all comparators, blinded trial designs and low risk of bias were associated with higher pain relief compared to an open-label trial design and some concern or high risk of bias. CONCLUSION: The placebo response on pain for people with hand osteoarthritis was increased by appropriate blinding and a lower risk of bias assessment. Placebos were superior to a null-arm, while untreated control groups were not. Results emphasise the importance of using appropriate comparators in clinical trials. PROSPERO REGISTRATION ID: CRD42022298984.
