Extensions to Epidemiological Designs

This workshop is aimed at biostatisticians and epidemiologists who work with population register data. It is assumed that participants are familiar with cohort and nested case-control studies, and the regression models used to analyse them (logistic regression and Cox regression). The focus of the workshop will be to introduce extensions of these designs that make use of the additional information in the population register to improve the efficiency of the relative risk estimates: in particular, case-cohort designs (including stratification) and counter-matching in nested case-control studies. Methods for re-use of data will also be introduced, where researchers wish to re-use data sampled from the population register(s) for other purposes than the original research question. With appropriate statistical methods, these designs offer the potential to address more research questions than the classical standard designs and/or to improve efficiency.  

The workshop will begin with a brief overview of the traditional epidemiological designs and the typical approaches for the estimation of relative risk. The presenters will then introduce the methodological developments that have led to more advanced study designs in the context where data are sampled from a well-defined  “study base”, such as a national register. Case-control design (simple and stratified), and extensions to case-control designs (eg. counter-matching) will be presented from a sampling perspective, and different analysis options considered. Methods for estimation of relative risks from sampled cohorts will be presented for the traditional and extended designs. The re-use of case-control data and the combination of matched and unmatched case-control data will be formulated as case-cohort designs with appropriate sampling weights. All designs will be illustrated with data examples from published or ongoing work.

After the workshop, participants should be aware of more advanced designs for data sampled from a population register, understand what effect measures they provide, which design is appropriate for a given research question, and how statistical efficiency might be improved by the choice of design.

The format of the workshop will be lectures, interactive discussions, group sessions and some contributions from workshop participants. All applicants are requested to submit a short (half-page) motivation for their interest in the workshop, with examples of their work, thereby enabling the organisers to tailor the workshop to the participants’ interests.  Applicants with an ongoing or planned research study that would be of interest to discuss at the workshop are encouraged to submit an abstract (maximum 1 A4-page) that outlines (i) the research question, (ii) the design, and (iii)any plans for analysis or preliminary results (if available).