Triangulation in biomedical research: navigating an ocean of uncertainty

instrument-based methods (e.g., Mendelian randomisation and natural experiments), to bias demonstration and quantification techniques (e.g., negative exposure or outcome controls).The article nicely outlines how these different approaches all come with their own strengths and limitations.Even the 'holy grail' of evidence-based medicine, the randomised controlled trial, is not without risk of selection bias due to attrition, and often hampered by poor representativeness [4,5].As eloquently summarised by the authors: "The central tenet of evidence triangulation is to identify the most important weaknesses for any given study approach [] and, if necessary, to identify which new sources of evidence that do not share these weaknesses are required".In other words, we look to change a certain causal assumption, while keeping others constant, to estimate its impact on the observed association.The authors rightfully caution that concordance between methods might still reflect similar bias.This could be characteristic to the method, but also be a trait of the data.For example, misclassification of outcome in one data source might bias various methods in the same direction.Triangulation therefore benefits from diversity in applied methods as well as data. Comparable studies and comparable resultsGutierrez et al. demonstrate that the process of triangulation relies on many of the same principles as meta-analysis.Comparability of estimands, definitions of exposure and outcome, and population differences all are important when considering consistency of the presented evidence.With advances in methodology, we start to move beyond qualitative comparisons to translating estimands in-for example-Mendelian randomisation, to those from regression discontinuity or trials.The underlying assumptions are still challenging to meet, and different methods may not