Ideally you’d use matched historical controls from before 2019. Depending on the study, that’s a common source for “controls”. But it depends on using measureme

Ideally you’d use matched historical controls from before 2019. Depending on the study, that’s a common source for “controls”. But it depends on using measurements/diagnoses that are roughly standard. For example, prior to 2019, 30-40 years old made up X% or people diagnosed with a cardiomyopathy. Post 2020, that number is Y%.