Correspondence (letter to the editor): “Retrospective” as Well as “Retrolective”

Weiler et al. reported a secondary analysis of data on acute liver failure (1). They analyzed administrative data from the statutory health insurance company Allgemeine Ortskrankenkassen (AOK). The authors described their design as a population-based study using prospectively collected data. In actual fact, however, the study is a “scientific study with data whose primary use is not of a scientific nature” (2). The question was not defined before the data were collected, nor were data collected about the question (for example, by means of examination) or recorded from information regarding healthcare services for the question (for example, by means of a questionnaire). The mixture of data was therefore random with regard to the research question. Earlier, Lorenz already pointed out the problems in categorizing observational studies as prospective or retrospective (3). According to Lorenz, the analysis reported by Weiler et al. is retrospective, because it considers past events, as well as “retrolective,” because it does not collect new data on the events of interest. The difference between a prospective population based study of acute liver failure and the presented secondary analysis is also economically relevant. While carrying out the secondary analysis might have incurred about €100 000 in costs, a prospective population based study and 4652 cases would have cost several million Euros. An unambiguous and uniform terminology regarding study types is essential for interpreting empirically gained data, in view of the growing range of methods—for example, using Big Data or methods of artificial intelligence. Such a terminology would support recipients of publications in terms of awareness of strengths and limitations of the different study types. This terminology is currently lacking.