Methods in Evidence Synthesis seminar (MESS)
Click here to join the meeting on Teams
Summary: In 1995, Jha et al examined the effect of β-carotene on cardiovascular mortality in a systematic review. At the time of their study, there were in total 6 observational studies and 3 RCTs, and meta-analyses of these studies suggested a protective effect of β-carotene based on the observational studies, but an adverse effect based on the RCTs. Since 1995, many more studies have been published. We sought to triangulate the available studies, to tackle specific causal questions about the effect of a well-defined exposure on a well-defined outcome. In addition, we addressed plausible biases from each study by implementing risk of bias tools, RoB 2 and ROBINS-E for RCTs and observational studies respectively. RCTs and MR studies generally show no association between β-carotene and CVD, but observational studies indicate that high intake of β-carotene reduces the risk of CVD. Our work highlights the importance of defining research questions explicitly and raises several points of discussion when triangulating RCT and observational studies.
Bio: I am a Biostatistician, specialises in Genetic Epidemiology, and working at MRC Integrative Epidemiology Unit. My research interest is in Mendelian randomisations, Bayesian approaches, triangulation, and prior elicitation.