Currently, my main projects are:
- Amyotrophic lateral sclerosis (ALS) in the Danish registries. We use national registry data from Denmark, including hospital diagnosis history, residential history, occupational history and mortality data to assess behavioral, occupational and other potential risk factors for ALS.
- Bias amplification. We are currently working on assessing when and to what degree bias amplification is a problem - in both conventional analyses and analyses of mixtures. Bias amplification occurs when unmeasured confounding is amplified by improperly conditioning on an instrumental variable or instrument-like variable.
- Shifted diagnosis times. In certain situations, exposures can shift the time at which a disease would be diagnosed, but not affect any of the etiological processes or the "true" underlying disease progression (much like lead-time bias in screening epidemiology). We consider the impact that such shifting of diagnosis times might have on both incidence and survival estimates in exposure epidemiology, particularly in large health-care datasets.