Leveraging Big Data for the advancement of healthcare: Generation of causal evidence on intervention effectiveness from health insurance claims

• Applicant: Anna-Janina Stephan, Ph.D.
• Subject Area: Public Health, Healthcare Research, Social and Occupational Medicine ; Epidemiology and Medical Biometry/Statistics
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 532417373

Background: Many publicly funded preventative and curative health(care) interventions are insufficiently evaluated with regard to their safety and (cost-)effectiveness in the wider population. The methodological gold standard for creating causal evidence on intervention efficacy are Randomized Controlled Trials (RCTs). However, an intervention's (cost-)effectiveness under real-world conditions may differ from RCT results due to different population characteristics and real-world behaviors. This creates inefficiencies in resource allocation of public funds within the healthcare sector while simultaneously financial pressure on healthcare systems continues to rise worldwide. Analysis of administrative healthcare claims can add effectiveness evidence in large population-representative samples under real-world conditions, but comes with specific methodological challenges, especially high risk of unobserved confounding. Therefore, I propose to combine my thorough understanding of the strengths and pitfalls of health insurance claims data with a methods transfer of advanced (quasi-experimental) study designs from social and economic sciences to complement RCT results with causal real-world evidence and support sustainable and safer allocation decisions. Objective: The objective of this project is to leverage health insurance claims (a special form of Big Data) for the advancement of healthcare by generating causal evidence on real-world intervention effectiveness drawing on quasi-experimental study designs. Methods: The project will be conducted at Harvard Medical School in cooperation with world-leading experts on causal inference methods in healthcare claims data and comprises three dedicated Work Packages (WPs): An exemplary (cost-)effectiveness evaluation for a pharmaceutical intervention in US Medicare claims data using (quasi-experimental) study designs and advanced statistical methods such as instrumental variable approaches and propensity scores (WP1). In the second (optional) WP, I propose to conduct an exploratory feasibility assessment on evaluations in routine data of digital health applications as an example for non-pharmaceutical interventions, drawing on my experiences collected within WP1. With WP3, I will prepare a consecutive funding proposal with the objective of conducting (cost-)effectiveness evaluations for several types of interventions, including reimbursable digital health applications, within the German healthcare system. Outlook: The knowledge, skills and experience acquired within WP1 and 2 will allow for substantial knowledge transfer to the German healthcare context in WP3, allowing for a broad range of important applications. These will include (cost-)effectiveness evaluations for both pharmaceutical and non-pharmaceutical healthcare interventions such as surgeries, conservative therapies, health apps, and other complex interventions financed by the Statutory Health Insurance in Germany.

• DFG Programme: WBP Fellowship
• International Connection: USA
• Host: Professor Dr. Sebastian Schneeweiss