Development of statistical methods for observational studies 

 There are ethical and practical limitations to the feasibility of vaccine RCTs. Hence, it is necessary to explore contradictions between expectations and findings in observational studies to improve our understanding of both specific and non-specific vaccine effects. Such observational studies pose many methodological problems as recently reviewed56. Data on interventions are likely to be incomplete in low-income settings with less information for children who die; this often gives rise to survival bias in survival analyses of interventions, as we have documented several times29. Development of time-dependent vaccination propensity-scores has been proposed as a means to control for inherent differences between vaccinated and unvaccinated children and to estimate average causal vaccination effects56. Another approach to causal inference is to use marginal structural models57.


 We will pursue both methods and develop statistical models which can address the important problems concerning observational vaccination studies.


 This work package provides important methodological input for WP5-6.

PhD student

 The PhD student will be placed at the Center, but will spend 1-2 days per week at Department for Biostatistics, University of Copenhagen. The PhD student will be supervised by Professor Per Kragh Andersen, University of Copenhagen, and Henrik Ravn. The PhD student’s academic focus will continuously be evaluated by CVIVA's steering committee. Research results will be published in leading international journals and in conference proceedings, and in all cases, the Center and the Danish National Research Foundation be thanked for support. The PhD student will participate in all key meetings and activities of the Center and contribute to the Center's general activities.