“…Sample sizes differed between 92,842 and ≈ 11,000,000. All reviewed studies used secondary data as their main data source: eight studies used health insurance claims data [8, 13, 14, 40, 41, 44, 45, 47], one used prescription data from a pharmacy computing centre [15], one from a disease analyser database [43], one from computerised patient records [42], and one from mixed sources (clinical and market research) [46]. Among the eight publications using health insurance claims records as data source, three obtained their data from BARMER GEK [8, 14, 40], four from regional branches of the AOK [13, 44, 45, 47], and one did not state the name of the statutory health insurance [41].…”