Questions méthodologiques en recherche clinique et épidémiologique sur le vieillissement

A lot of methological difficulties interfere with the design and the interpretation of data analysis in clinical and epidemiological research on diseases related to ageing. The first one is that normal and pathological ageing are often strongly related, with blurred limits. The selection bias is particularly frequent, related to numerous health care resources, to premature death, to deny or refusals. Data collection is also particularly difficult, due to memory problems related to ageing and prolonged lag time between exposure of a given factor and interview, or related to the disease of interest itself (in particular, dementia). Multiple confounding factors are frequent, due to polypathology and polymedication, also interference between biological, psychological and sociological factors. Another difficulty is related to the insidious evolution of many chronic pathologies related to ageing which interfere with the direction of the causality between factors and disease.