Article type
Year
Abstract
Background: The placebo response varies across neuropathic pain trials making it challenging to demonstrate a treatment effect and difficult to compare trials of different medications.
Objectives: The primary objective was to determine the study-level characteristics which predict the placebo response in trials of painful diabetic neuropathy. A second objective was to identify characteristics that are associated with changes in relative effect, independent of the placebo response.
Methods: A systematic review of the English-language literature was performed and fair-good quality, randomized, placebo-controlled trials measuring ${\geq}$50% pain relief from baseline as an outcome, were identified. Medications included the tricyclic antidepressants, the serotonin-norepinephrine reuptake inhibitors, gabapentin, pregabalin, and other anticonvulsants. Meta-regression techniques were employed to identify predictors of placebo response and predictors of relative effect while controlling for the placebo response. Covariates included: drug studied, publication year, trial duration, size and design, gender distribution, age, duration of diabetes and neuropathy, number of treatment groups, presence of a washout period, use of additional pain medications, and rate of patient recruitment.
Results: Twenty-five trials of diabetic neuropathy were included. The best predictor of placebo rate was whether the study was conducted in the United States or Europe versus another country or region (p = 0.053). After adjusting for placebo response, predictors of relative risk were year of publication and length of trial. For example, assuming a 20% placebo rate, a 6-week trial published in 1997 is 1.7 times more likely to demonstrate pain relief versus placebo than a 12-week trial published in 2005. In contrast to previous research, neither baseline pain levels nor recruitment rate were found to predict placebo response.
Conclusions: Although the placebo response in neuropathic pain trials is an important consideration for systematic reviewers, trial duration and publication year may be equally important.
Objectives: The primary objective was to determine the study-level characteristics which predict the placebo response in trials of painful diabetic neuropathy. A second objective was to identify characteristics that are associated with changes in relative effect, independent of the placebo response.
Methods: A systematic review of the English-language literature was performed and fair-good quality, randomized, placebo-controlled trials measuring ${\geq}$50% pain relief from baseline as an outcome, were identified. Medications included the tricyclic antidepressants, the serotonin-norepinephrine reuptake inhibitors, gabapentin, pregabalin, and other anticonvulsants. Meta-regression techniques were employed to identify predictors of placebo response and predictors of relative effect while controlling for the placebo response. Covariates included: drug studied, publication year, trial duration, size and design, gender distribution, age, duration of diabetes and neuropathy, number of treatment groups, presence of a washout period, use of additional pain medications, and rate of patient recruitment.
Results: Twenty-five trials of diabetic neuropathy were included. The best predictor of placebo rate was whether the study was conducted in the United States or Europe versus another country or region (p = 0.053). After adjusting for placebo response, predictors of relative risk were year of publication and length of trial. For example, assuming a 20% placebo rate, a 6-week trial published in 1997 is 1.7 times more likely to demonstrate pain relief versus placebo than a 12-week trial published in 2005. In contrast to previous research, neither baseline pain levels nor recruitment rate were found to predict placebo response.
Conclusions: Although the placebo response in neuropathic pain trials is an important consideration for systematic reviewers, trial duration and publication year may be equally important.