Randomization minimizes the chance of bias by balancing both known and unknown prognostic factors between trial arms. However, the results can be biased when the data from ‘non-adherent’ participants are excluded from the analysis to evaluate treatment effect.
The aim of the study is to demonstrate the risk of bias associated with missing participants in a randomized controlled trial of cell therapy for peripheral vascular disease (PVD) patients.
We performed systematic searches from inception to December 2017 in MEDLINE and Embase. We conducted meta-analysis using complete case analysis as the primary analysis. To test the robustness of the results, we conducted sensitivity analyses using both worst-case scenario and adjusting zero events to evaluate the magnitude of impact of risk of bias on the reported results.
In the complete case analysis, the cell therapy showed a reduced the risk of amputation by 43% (risk ratio (RR) 0.43; 95% confidence interval (CI) 0.27 to 0.68). In the worst-case scenario, the treatment effect on prevention of amputation was decreased (RR 0.61 0.36; 95% CI 0.39 to 0.97). In addition, adjusting zero events yielded a wide confidence interval (RR 0.44; 95% CI 0.20 to 0.97).
Even when we assumed a worst-case scenario, the results remained statistically significant; cell therapy for PVD may reduce the risk of amputation and may be guaranteed as a true treatment effect for cell therapy.
This research was supported by a grant (18172MFDS182) from the Ministry of Food & Drug Safety in 2018.
Patient or healthcare consumer involvement: