Industry sponsorship bias in randomized controlled trials of digital cognitive behavioral therapy for insomnia: a meta-research study based on the 2023 Brazilian guidelines on the diagnosis and treatment of insomnia in adults.

Background: Industry sponsorship bias refers to the tendency of a study, most likely clinical trials, to produce results that favor the sponsor's interest. It is especially relevant in cases in which a study is funded by companies or organizations with a commercial interest in the product or technology being evaluated. Digital Cognitive Behavioral Therapy for Insomnia (dCBT-I) is a widely used nonpharmacological treatment, and research in this area is often funded by organizations that have a commercial interest in this treatment. This study aimed to assess whether industry sponsorship bias in dCBT-I trials is associated with more favorable outcomes. Methods: This study was based on the sample of randomized controlled trials (RCTs) included at the "2023 Brazilian Guidelines on the Diagnosis and Treatment of Insomnia in Adults." This guideline was based on a systematic review conducted in the PubMed and Web of Science databases, searching for randomized controlled trials (RCTs) on dCBT-I. Inclusion criteria included 1. Studies performed with adults with non-comorbid insomnia, diagnosed using the International Classification of Sleep Disorders 3rd edition (ICSD), the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5), or with moderate to severe insomnia symptoms evaluated using the Insomnia Severity Index (ISI) or the Athens Insomnia Scale (AIS). 2. dCBT-I as intervention. 3. Other forms of CBT-I or negative control groups (no treatment, placebo, waiting list, or minimal intervention) as comparators, and 4. ISI as main outcome. For each included study, sponsorship bias was analyzed in a 5-points scale, considering the role of the sponsor (from "no funding received" to "all author authors are affiliated to the company developing the dCBT-I"). Immediate post-treatment ISI data was extracted for both intervention and control groups in each included study, and the between-groups Cohen's d effect size was calculated for each included study. Methodological quality in each included RCT was evaluated using the van Tulder scale. Statistical analyses were performed to investigate possible associations between the levels of sponsorship bias and the results of the studies. Results: Twenty-eight analyses of RCTs were included. Interventions such as SHUTi (Sleep Healthy Using the Internet) (39.28%) and Online CBT-I (28.57%) were the most common, with comparators such as minimal intervention (50%) and waiting list (32.14%). There was a significant association between the risk of sponsorship bias and open access publication [X Conclusion: A greater risk of sponsorship bias was associated with lower methodological quality articles and open access publication.