SHIP-CODES - Supervised Coding Of Causes of DEath in the SHIP Cohorts

• Applicants: Dr. Adrian Richter; Professor Dr. Henry Völzke
• Subject Area: Epidemiology and Medical Biometry/Statistics
• Term: since 2026
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 579103854

The decline in public willingness to participate in scientific studies is increasingly evident which has impact on the generalizability of research findings. In the Study of Health in Pomerania (SHIP), for example, response has declined considerably: from 68.8% (1997-2001) in SHIP-START, to 50.1% in SHIP-TREND (2008-2012), and to around 30% in SHIP-NEXT (started 2021). Such declines affect study results in population-based research, as responders often represent a subpopulation with higher health literacy, while underrepresenting individuals have lower education, income, and more health issues. Non-responders also show higher mortality; recent Scandinavian studies revealed a 2.0 to 2.5-fold elevated mortality risk among non-responders of population-based studies. In order to address this bias, it is necessary to identify and analyze the underlying selection mechanisms. Typically, non-responders are surveyed with brief questionnaires, yet response rates remain very low. A promising approach for reducing non-response bias leverages information from deceased non-responders’ death certificates, which offer relatively accurate cause-of-death data and underlying diseases. However, a standardized approach to coding information from death certificates is required for consistent processing and use of this data. The internationally used Iris software, following WHO standards, automates coding and achieves over 90% accuracy when using regular expressions. Support from the senior developer of the German version of Iris strengthens this project, and Destatis (Federal Statistical Office) has filed a Letter-of-Intent, underscoring the project’s potential. This technical approach is now being introduced into population-based research in Germany and into a novel utilization of data from non-responders. Previously, GDPR restrictions limited data collection from non-responders due to lacking informed consent. However, in January 2024, the Federal State Mecklenburg-Vorpommern legally recognized the public interest in such data, enabling its use in SHIP non-responder research. The University Medical Center Greifswald’s ethics committee has approved this studies’ protocol. Scientifically, this project aligns with the SHIP's objective of analyzing the elevated mortality rates in Mecklenburg-Vorpommern. The project enables refinement of present mortality studies by focusing on cause-specific mortality and selection mechanisms of study participation. The latter will lead to an adjustment of weighting procedures for the participants in the SHIP studies and significantly reduce the non-response bias. The project will employ modern techniques for the automated and supervised coding of free text information from death certificates. This will enhance the representativeness and accuracy of epidemiological studies and facilitate a more nuanced understanding of the factors influencing mortality.

• DFG Programme: Research Grants
• Co-Investigator: Privatdozent Dr. Till Ittermann