Observational cohort studies and the challenges of in silico experiments
Menée aux Etats-Unis à partir de données portant sur 9 272 patients atteints d'un cancer de la prostate (âge moyen : 66,9 ans), cette étude de cohorte évalue les effets d'un traitement anti-androgénique sur le risque ultérieur de démence (durée médiane de suivi : 3,4 ans)
In this issue of JAMA Oncology, an investigation by Nead and colleagues1 describes an association between use of androgen deprivation therapy and risk of dementia in a cohort of 9272 men. Of the 1826 men who received androgen deprivation therapy from 1994 to 2013, the increased absolute risk of dementia in those who received androgen deprivation therapy was 4.4% at 5 years. The authors conducted an observational cohort study and derived their cohort data elements from words, or patterns of words, found in more than 11 million largely unstructured clinical notes and reports from pathologic studies, radiologic images, and transcriptions. Although the article provides details about the methods used to derive these data elements, the authors rely on eTables in the Supplement and references to the peer-reviewed literature to provide the reader with assurance about these methods that may be relatively unfamiliar to some audiences.
JAMA Oncology , commentaire, 2015