Article type
Year
Abstract
Background: in published studies, industry funding has been associated with outcomes that favour the commercial funder, and these differences could not be explained by differences on the Cochrane 'Risk of bias' tool. This has led to calls to include an additional ‘funding bias’ domain in this tool. Examining trial characteristics at the stage of trial registration may give insight to the origins of funding bias.
Objective: the aim of the current study was to determine whether industry involvement (industry sponsorship, funding, or collaboration) is associated with trial characteristics at the time of trial registration.
Methods: this was a cross-sectional analysis of all trials registered on the Australian New Zealand Clinical Trials Registry (ANZCTR) in 2017. We classified trials by whether there was any industry involvement, defined as any industry funding, sponsorship or collaboration. We analysed whether industry involvement was associated with differences in type of control, sample size, study phase, randomisation, registration timing, and purpose of study.
Results: 300 (21%) of the 1433 included trials reported industry involvement. As shown in the Figure, there were differences in characteristics of trials with and without industry involvement. Trials with industry involvement were less likely to use an active control compared to trials without industry involvement (40% versus 58%, odds ratio (OR) 0.49, 95% confidence interval (CI) 0.38 to 0.63), and sample size was smaller in trials with industry involvement (median (interquartile range (IQR)) industry = 45 (24 to 100), median (IQR) non-industry = 70 (35 to 160)). There was a higher likelihood for industry trials to be earlier phase trials (Chi2 (df) = 71.46 (4), P < 0.001). There was no difference in use of randomised allocation, but industry trials were more likely to be prospectively registered than non-industry trials (OR 2.02, 95%CI 1.47 to 2.82). There was a large difference in primary study purpose: 83% of industry-funded trials (compared to 65% of non-industry-funded trials) assessed treatment instead of prevention, education or diagnosis (OR 3.02, 95% CI 2.17 to 4.32).
Conclusion: this study gives novel insights into differences in study design by industry involvement. Higher use of non-active controls in industry trials offers a potential explanation for funding bias, since comparing a treatment to placebo as opposed to an active control is likely to yield larger effect sizes. However, the results presented here are descriptive associations and thus need to be interpreted with caution. Systematic review authors should be mindful of use of control when assessing studies. Our findings also highlight the importance of non-industry-funded research to ensure not only treatment, but also prevention, education and diagnosis questions are addressed.
Patient or healthcare consumer involvement: we will invite healthcare consumers to comment on this research.
Objective: the aim of the current study was to determine whether industry involvement (industry sponsorship, funding, or collaboration) is associated with trial characteristics at the time of trial registration.
Methods: this was a cross-sectional analysis of all trials registered on the Australian New Zealand Clinical Trials Registry (ANZCTR) in 2017. We classified trials by whether there was any industry involvement, defined as any industry funding, sponsorship or collaboration. We analysed whether industry involvement was associated with differences in type of control, sample size, study phase, randomisation, registration timing, and purpose of study.
Results: 300 (21%) of the 1433 included trials reported industry involvement. As shown in the Figure, there were differences in characteristics of trials with and without industry involvement. Trials with industry involvement were less likely to use an active control compared to trials without industry involvement (40% versus 58%, odds ratio (OR) 0.49, 95% confidence interval (CI) 0.38 to 0.63), and sample size was smaller in trials with industry involvement (median (interquartile range (IQR)) industry = 45 (24 to 100), median (IQR) non-industry = 70 (35 to 160)). There was a higher likelihood for industry trials to be earlier phase trials (Chi2 (df) = 71.46 (4), P < 0.001). There was no difference in use of randomised allocation, but industry trials were more likely to be prospectively registered than non-industry trials (OR 2.02, 95%CI 1.47 to 2.82). There was a large difference in primary study purpose: 83% of industry-funded trials (compared to 65% of non-industry-funded trials) assessed treatment instead of prevention, education or diagnosis (OR 3.02, 95% CI 2.17 to 4.32).
Conclusion: this study gives novel insights into differences in study design by industry involvement. Higher use of non-active controls in industry trials offers a potential explanation for funding bias, since comparing a treatment to placebo as opposed to an active control is likely to yield larger effect sizes. However, the results presented here are descriptive associations and thus need to be interpreted with caution. Systematic review authors should be mindful of use of control when assessing studies. Our findings also highlight the importance of non-industry-funded research to ensure not only treatment, but also prevention, education and diagnosis questions are addressed.
Patient or healthcare consumer involvement: we will invite healthcare consumers to comment on this research.