Methods Projects

Using plasmode and Monte-Carlo simulation studies, we evaluates the performance of TMLE compared to that of IPW estimators based on a point-exposure cohort study of the marginal causal effect of post-myocardial infarction statin use on the 1-year risk of all-cause mortality from the Clinical Practice Research Datalink. Near violations of the positivity assumption are investigated in a high-dimensional covariate setting for both methods.

In order to examine the claim that the high-dimensional propensity score algorithm can adjust for unmeasured confounding, we hide information from the algorithm to examine its ability to compensate by selecting proxies of what was hidden. Performance of the algorithm within the hidden information context is then compared to its performance within the full information context.

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.

The use of incretin-based drugs was not associated with an overall increased risk of pancreatitis, compared with other oral antidiabetic drugs.

We describe how three different propensity score methods, propensity score matching, stratification on the propensity score and inverse probability of treatment weighting using the propensity score, can be used to estimate absolute measures of treatment effect on survival outcomes.

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.