Methods for the estimation of the effects of chronic disease risk factors on mortality continue to be an area that generates confusion and controversy.
In response to the frequently observed U-or J-shaped relations between risk factors and mortality, some authors suggest that subjects dying during the first k years of follow-up (where k is some positive number less than the total length of follow-up) be excluded from statistical analyses.
By excluded, the authors mean completely removed from the data set.
The rationale is that persons dying during the first k years are likely to have a preexisting occult disease that confounds the relation between the risk factor under study and mortality.
Excluding persons dying during the first k years of follow-up purportedly reduces this confounding.
However, the authors are aware of no demonstration that this procedure effectively accomplishes its goal.
They show that excluding subjects who die during the first k years of follow-up does not necessarily lead to a reduction in bias in the estimated effect of a risk factor on mortality when this relation is confounded by the presence of occult disease.
Moreover, it is possible for such exclusion to exacerbate the confounding due to preexisting disease.
Thus, excluding subjects dying during the first k years of follow-up is not necessarily an effective strategy for dealing with confounding due to occult disease.
Investigators are encouraged to pursue alternative methods.
Mots-clés Pascal : Epidémiologie, Méthodologie, Facteur risque, Mortalité, Stade précoce, Modèle statistique, Homme, Antécédent, Facteur confusion
Mots-clés Pascal anglais : Epidemiology, Methodology, Risk factor, Mortality, Early stage, Statistical model, Human, Antecedent
Notice produite par :
Inist-CNRS - Institut de l'Information Scientifique et Technique
Cote : 97-0533016
Code Inist : 002B30A01A1. Création : 13/02/1998.