Uncertainty and heterogeneity in network meta-analysis with small subgroups and few studies

Project summary

Individual patient data meta-analysis can utilize patient-level characteristics to investigate different treatment effects in potentially small subgroups. Yet, when the evidence base is small due to small subgroups and/or a small number of studies, treatment effect estimates are associated with large uncertainty. We will particularly focus on network meta-analysis, which accommodates the need to simultaneously synthesize evidence on multiple treatments. Specifically, we will develop a framework to explore subgroup treatment effects and treatment effects conditional on post-randomization characteristics. This will provide methods to 1) identify patient characteristics with strong predictive properties regarding the outcome of interest 2) estimate treatment effects within the selected subgroups, and 3) provide recommendations for adding data via new studies.

Our methods

  • Knowledge-driven modeling
  • Neural networks
  • Local perspective

Principal investigator 1

Doctoral researcher position 1

(supervised by PI Nikolakopoulou)


Perform original statistical research: development of statistical methods for prediction models, quantifying heterogeneity and develop methods for making concrete recommendations for future research.


  • Master’s degree or equivalent in statistics
  • Experience in programming in a high-level language, such as R or Python

Principal investigator 2

Doctoral researcher position 2

(supervised by PI Schramm)


Developing a protocol for selecting patient characteristics in the context of persistent depression, dealing with directionality in bias in meta-analysis, and interpret evidence based on clinical judgements. Further tasks comprise administration of study documents and materials, literature search, and evaluation of study data.


  • Master’s degree or equivalent in psychology
  • Very good statistical knowledge and preferably experience in quantitative data analysis (e.g., with SPSS, R)
  • Good command of the German language

Administrative Manager

Marc Schumacher

Institute of Medical Biometry and Statistics,
Faculty of Medicine and Medical Center –
University of Freiburg