Sampling bias in an internet treatment trial for depression

Abstract

Internet psychological interventions are efficacious and may reduce traditional access barriers. No studies have evaluated whether any sampling bias exists in these trials that may limit the translation of the results of these trials into real-world application. We identified 7999 potentially eligible trial participants from a community-based health cohort study and invited them to participate in a randomized controlled trial of an online cognitive behavioural therapy programme for people with depression. We compared those who consented to being assessed for trial inclusion with nonconsenters on demographic, clinical and behavioural indicators captured in the health study. Any potentially biasing factors were then assessed for their association with depression outcome among trial participants to evaluate the existence of sampling bias. Of the 35 health survey variables explored, only 4 were independently associated with higher likelihood of consenting—female sex (odds ratio (OR) 1.11, 95% confidence interval (CI) 1.05–1.19), speaking English at home (OR 1.48, 95% CI 1.15–1.90) higher education (OR 1.67, 95% CI 1.46–1.92) and a prior diagnosis of depression (OR 1.37, 95% CI 1.22–1.55). The multivariate model accounted for limited variance (C-statistic 0.6) in explaining participation. These four factors were not significantly associated with either the primary trial outcome measure or any differential impact by intervention arm. This demonstrates that, among eligible trial participants, few factors were associated with the consent to participate. There was no indication that such self-selection biased the trial results or would limit the generalizability and translation into a public or clinical setting.