To evaluate the association between the degree of response to placebo in migraine studies and the observed difference between drug and placebo across studies of preventative treatments for migraine.
A systematic review was performed using MEDLINE and the Cochrane Central Register of Controlled Clinical Trials from January 1988 to June 2019. Randomized, double-blind, parallel-group, placebo-controlled trials on oral or injection preventative treatments for migraine were included. Single- and multi-variable linear regression analyses were performed on the placebo-subtracted response rate (i.e. placebo responders subtracted from active responders), and the proportion of placebo responders. Fisher\'s exact tests were performed on the level of placebo response and the success in meeting the study\'s primary endpoint.
After adjusting for route of administration and number of randomized subjects, there was a statistically significant association between the proportion of patients who were placebo responders and the placebo-subtracted response rate (b = -0.27, p = 0.02). There was a statistically significant difference in trial success rate (60%) between studies with ≤20% placebo responders and studies with > 30% placebo responders (p = 0.03).
Considering the detrimental impact that high placebo response can have on clinical trials, it is imperative to find effective solutions to decrease the placebo response and increase assay sensitivity.

Identifying methods to improve assay sensitivity in randomized clinical trials (RCTs) may facilitate the discovery of efficacious pain treatments. RCTs evaluating pain treatments typically use average pain intensity (API) or worst pain intensity (WPI) as the primary efficacy outcome. However, little evidence is available comparing the assay sensitivity of these 2 measures. In this systematic review and meta-analysis, we comprehensively reviewed all low back pain, osteoarthritis pain, fibromyalgia, diabetic peripheral neuropathy pain, and postherpetic neuralgia RCTs that used a parallel group design. Eligibility required: 1) primary RCT report published between 1980 and 2016, 2) comparing 1 or more active, efficacious pharmacologic pain treatment(s) with placebo, and 3) providing data on the standardized effect size (SES) for API as well as WPI for all treatment arms. Twenty-seven active versus placebo comparisons were identified in 23 eligible articles. Using a random-effects meta-analysis, API SES and WPI SES did not differ significantly (difference = -.021, 95% confidence interval = -.047 to .004, P = .12). The findings indicate that, depending on the objectives of the study, either API or WPI could be used as a primary outcome measure in clinical trials for the chronic pain conditions included in this analysis.
Understanding the comparative assay sensitivity of API and WPI may advance pain treatment research. A meta-analysis of trials of efficacious pharmacologic treatments in 5 pain conditions did not show a statistically significant difference between the assay sensitivity of API and WPI.