Causal theory for settings with limited resources

Questions about limited treatment resources are ubiquitous in practice. For example, healthcare systems have experienced shortages of vaccines, ventilators, protective equipment and personnel during the current coronavirus pandemic. Similarly, patients with organ failure are allocated to waiting lists because the number of organ transplants is scarce. When treatment resources are limited, decision makers have to decide on a treatment allocation strategy. Yet, conceptual and methodological gaps exist for causal inference in these settings due to the complex dependence structures and longitudinal event processes.The objective of this project is to develop a formal theory and methodology for causal inference in limited resource settings. This is important because it will considerably extend the scope of the causal questions we can answer. This is non-trivial because factual and counterfactual resource constraints require us to consider new and complex interventions, which motivate new estimands, identifiability conditions and estimators. The development of this theory will build on a counterfactual framework for causal inference and modern semi-parametric estimation theory. Specifically, the first aim of this project is to give formal definitions, along with theory for interpretation and identification, of new causal estimands that are relevant for real-life decision making in settings with limited resources. The second aim of this project is to apply and extend existing methods for estimation of causal effects in both point-treatment settings and complex time-varying settings with resource constraints. The third aim is to develop software and to apply the new methodology to real and synthetic data.The expected outcome of this project is a transparent and actionable framework for causal inference when resources are limited, which will improve the design, analysis and interpretation of studies that aim to guide future policies, for example in medicine.