[Clinical relevance of retrospective observations]

Many retrospective observational studies have been published in the recent literature. In the current issue of this journal two papers deal with retrospective analysis of observations after different surgical interventions: one concerns surgery for hilar cholangiocarcinoma, the other surgery for oesophageal carcinoma. Studies of this design are hard to interpret due to a number of methodological issues: (a) cohort effects: both the patient population and diagnostic procedures may have changed over time; (b) the learning curve of the healthcare provider, in this case the surgeon; and (c) confounding by indication: the data studied is biased by medical decision-making. However, the results of these studies inform healthcare providers and patients about survival and complication rates, among other things. Although they are not properly designed according to basic evidence-based medicine guidelines, these studies still provide us with clinically relevant information about current experience of medical interventions.