T5.1/5.2 Use Cases “Nutritional Epidemiology” and “Epidemiology of chronic diseases”

The Use Cases “Nutritional epidemiology” and “Epidemiology of chronic diseases” were chosen to showcase complex exposition and outcome data from the epidemiological research area. The complexity in nutritional data mainly results from a variety of dietary assessment instruments, which were chosen depending on the research questions in the studies. This is resulting in varying methods within and between studies and in distinct granularity of detail for the assessed research data (e.g. frequencies vs. intake amount; covered observation periods). Additionally, different dimensions of nutritional data can be analysed: either in more aggregated forms like meal- and dietary patterns, on the level of food groups and food items, or on the level of macro- and micronutrients. For the research data on chronic diseases, similar complexity is introduced by different assessment methods (self-report to medical diagnosis) and the use of different versions regarding the disease classification system by the WHO (ICD-10 vs. ICD-11).

The epidemiological studies in Germany are also characterised by this heterogeneity in the research data, such as the German National Cohort (GNC), EPIC-Potsdam, EPIC-Heidelberg, KORA and SHIP. The efforts to standardise and harmonise the assessed research data and the immediate implementation of developed services in NFDI4Health will provide a huge advantage for the overall research field in Germany.

Based on typical research questions in the respective research areas, scientists of the Use Cases 5.1 “Nutritional Epidemiology” and 5.2 “Epidemiology of chronic diseases” will work on the development of services regarding data standardisation and harmonisation, which will be provided later for the whole research community.

The following pilot studies for data standardisation and harmonisation in epidemiological studies are ongoing:

1. Systematic investigation of methodological limitations in the derivation of dietary patterns 

Contact: Dr. Franziska Jannasch (DIfE)

Even though numerous exploratorily derived dietary patterns have been generated in the last two decades, there has been no comprehensive investigation of methodological constraints, such as the influence of different energy adjustment techniques or different levels of granularity, when condensing food items into food groups, on the resulting pattern composition. Furthermore, for simplified dietary patterns, which can be useful in multi-centre analyses, the systematic investigation of different cut-offs for factor loadings on pattern composition and the comparison with the original dietary patterns are useful analyses to conduct.

2. Association of dietary sugar intake, sugar sweetened beverages and related foods with prospective changes in body fatness and chronic disease risk

Contact: Tracy Bonsu Osei (MDC) and Dr. Ines Perrar (Uni Bonn)

In Germany, about 89% of the disease burden (quantified as disability-adjusted life years, DALYs) is attributed to chronic illnesses including type 2 diabetes (T2D) and cardiovascular diseases (CVD), which are mainly driven by overweight and obesity. Although an inadequate diet is a major risk factor for the development of obesity and chronic diseases, there are large research gaps regarding the association of sugar intake with body fatness and chronic disease risk. Furthermore, most of the evidence related to the intake of total sugar or sugar-rich foods such as sugar-sweetened beverage (SSB) in relation to chronic disease risk stems from studies in non-German populations. By re-using existing data from German cohort studies in a federated data analysis, this study aims to examine the associations of dietary sugar intake, SSB and related foods with changes in body fatness and chronic disease risk and mortality.

In NFDI4Health, data standardisation and harmonisation will be performed based on the Maelstrom harmonisation procedures, which have been adapted to the current pilot projects.

First, the metadata is collected at the study and resource description level via the NFDI4Health German Central Health Study Hub. The collection of the metadata descriptive of research data required for our pilot studies is carried out using the Maelstrom metadata schema (step 1 of the Maelstrom Harmonization Guidelines). At the same time, the installation and configuration of OPAL/DataSHIELD is carried out. DataSHIELD is an infrastructure that enables non-disclosive analysis of sensitive research data without the research data having to leave the data holders' servers. A Standard Operating Procedure (SOP) for the installation and configuration of Opal/DataSHIELD for the NFDI4Health consortium is published here: Github-opal-datashield-sop. Once all the necessary metadata has been collected and the DataSHIELD infrastructure has been installed, harmonization can begin. NFDI4Health provides a central R server to conduct federated data analysis within DataSHIELD.

To support the participating studies, a harmonization protocol was created and published on Github: Github-data-harmonisation-protocol. After harmonization, the corresponding research questions are analysed. The harmonised metadata will then be made available for re-use in future research projects.

Contact