Knowledge framework to study and appraise the immunogenicity related to switch. The figure classifies the designs for the studies that can provide evidence about the immunogenicity of switching. The x-axis represents time (flowing left to right). The y-axis represents risk of bias, from low (top, in green) to high (bottom, in red). The space results partitioned into 4 quadrants by combination of study perspective (retrospective/prospective) and rigor of observation (controlled/uncontrolled). The panel on the right hand side describes how the baseline risk for inhibitors is accounted for. Randomization is the main mechanism to reduce risk of bias and control for the baseline risk of events. Multivariable analysis might be a good adjunct if size permits, and it is optimal for nested case control studies, which sum up more events by design). The risk of bias in registries and cohort studies lies in the nonrandomized assignment to “active” or control group and in incomplete follow up data. Complete data collections are indeed more powerful then nested case control studies, but are rarely available. The control group for prospective studies might be historical (studies across the switch, prone to secular variation in rate of events) or parallel. An ongoing example of prospective parallel controlled cohort is the EUHASS study. For each and any of the study designs, a much better insight might be obtained using as a “magnifier glass” a proper combination of baseline assessment for preexisting inhibitors; standardized assay methodology; observation time frame; and testing frequency.