When to start dialysis?: Clinical and methodological issues involved

Abstract

In this thesis we showed the clinical impact of several methodological issues that should be taken into account when studying chronic kidney disease progression and in order to find an answer to the question when to start dialysis. From a methodological point of view, the observations from this thesis provide several recommendations for future research with regard to studying associations of risk factors with CKD progression: the use of an inception cohort of incident patients, and using linear mixed effects models for analyses. Ultimately, patients with kidney failure could be subjected to dialysis. The optimal timing for dialysis initiation is unclear. Ideally, a randomized trial with all possible starting moments would be performed. However, this seems to be unfeasible due to the high patient number needed in each treatment arm. We have to rely on observational data, where methodological issues such as lead-time and immortal time bias arise. In this thesis, we showed the clinical impact of lead-time bias and how both types of bias were avoided by emulating a randomized trial using observational data in a pilot study. The methodology and recommendations provided in the thesis will be highly useful to find a more definitive answer in future research.