Background: Seeking local evidence on treatment efficacy is necessary if cultural factors are involved, as in psychotherapy for chronic pain (CP). Yet, local evidence is known to be prone to bias, making it difficult to reach reliable conclusions.
Objectives: This study aimed to critically evaluate our local evidence on the efficacy of psychotherapy on quality of life and disability in CP. We elaborate that, with some requirements, common meta-analytic tools can be utilized to detect and correct local evidence bias. We then provide a corrected estimate of the treatment efficacy.
Methods: The protocol was registered on PROSPERO, Record [deleted for blind review]. Elmnet, Pubmed, and ProQuest were searched for randomized trials. A multilevel meta-analysis was used to capture the hierarchical structure of the data, and robust variance estimation was used for inference. Several moderation analyses were conducted, and publication and other related sources of bias were examined.
Results: Forty-two trials were initially included. Six were excluded before the analysis due to serious reporting problems undermining their validity. The SMD from 185 effect sizes was 1.08 [.87, 1.3]. The funnel plot showed a strong bias. The bias-corrected estimate from a regression-based method was 0.45 [0.04, 0.87], and from the trim-and-fill was 0.75 [0.48, 1.0].
Conclusion: While our original estimate was large, the corrected estimate showed a medium effect, fairly comparable to the international estimates. Current evidence on different sources of bias in our literature suggests low quality and questionable research practice as the first suspects for our local evidence bias.