Methods Projects

We carry out simulation studies to assess the suitability of the 'sequential Cox approach' for analyzing time-to-event data in the presence of a time-dependent confounder, as an alternative to the marginal structural Cox modelling approach.

We illustrate the practical implementation of a newly proposed double robust estimator “Targeted Maximum Likelihood Estimation” (TMLE) and demonstrated its application in common pharmacoepidemiological settings. A real-word example using the clinical practice research datalink is used with a high-dimensional covariate space.

We review some methodological challenges facing drug safety networks such as CNODES and Sentinel.

We illustrate the confounding that can occur due to drug formulary restrictions using the example of fluticasone/salmeterol combination therapy, whose use is restricted in the province of Quebec, Canada. We found that formulary restrictions can result in substantial and unexpected confounding and should be considered during the design and analysis of pharmacoepidemiologic studies.

Likelihood ratio meta-analysis (LRMA) provides an alternative technique for comparing alternative hypotheses, and the results of a LRMA can be presented in a format that is familiar to meta-analytic audience. We explain the LRMA method and illustrate its application to an existing meta-analysis of observation studies on the association between high potency statins and acute kidney injury.

We examine the standard propensity score and high-dimensional propensity score methods to identify which provides the best adjustment for confounding by indication. To demonstrate and compare the performance of each, we use both methods to assess the risk of diabetes among statin users in Quebec, Canada.