Introduction Nested case-control studies are classically used in epidemiology to reduce time and cost for data collection while minimising bias induced by sample size reduction. However, if the continuous exposure of interest have a highly skewed probability distribution, rare exposure values are less likely to be selected. Therefore, the sampling process can miss important information in the long tail of the exposure distribution. Counter-matching design attempt to maximise the information on known exposure of interest during sampling process by the selection of controls in several categories covering all the range of the exposure. This study attempt to assess the influence of the choice of the categorisation method on the risk estimation.
Method The categorisation of continuous exposure is provided by several classification or clustering Methods: a priori thresholds, quantiles and k-means. To investigate the robustness of these methods, a simulation study is realised with several shapes of the probability distribution of the exposure, addition of covariate and missing collected data. Theoretical results will be illustrated with the example of a case-control study of cardiovascular diseases mortality (440 cases, 5 controls per cases (1:5)) nested within the French cohort of uranium miners (including 5086 men followed-up between 1945 and 2007) where cases and controls are matched on age and birth cohort and counter-matched on cumulative radon exposure.
Results 5000 samples 1:5 have been generated from the French uranium miners cohort for each methods applied on the cumulated radiation exposure. Relatively to the risk estimated from the cohort, the bias was systematically lower by using the k-means method.
Perspectives Results of the simulation study and application to the case-control study nested within the French uranium miners cohort could confirm this trend and contribute to the improvement of the efficiency of nested case-control study for the assessment of risk of exposure.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.