If you’re planning research with German statutory health insurance data, there’s a critical structural break in 2019 that could invalidate your entire study design. Data Model 3 introduced significantly higher granularity, but only for recent years. Here’s what changed and why it matters for longitudinal analyses.
Key Takeaways
- Data Model 3, introduced within Germany’s FDZ Gesundheit for data from 2019 onwards, delivers meaningfully higher granularity in statutory health insurance (SHI) claims data – expanding what researchers can study without redesigning the entire dataset.
- The Digital Healthcare Act (DVG), passed in late 2019, was a pivotal regulatory enabler that made richer, pseudonymized claims data available for research at a national scale.
- Longitudinal studies spanning pre- and post-2019 data must account for structural differences between data models, requiring consistency checks before methods are finalized.
- Known limitations persist – particularly around the granularity and completeness of pharmaceutical prescription and incapacity-for-work data – and planning around these early saves significant redesign effort later.
- The FDZ access framework continues to develop, with the 2025 Health Data Lab launch having broadened access further, and planned linkages to cancer registries and electronic patient records on the way.
Germany’s statutory health insurance data has long been one of the most valuable assets for real-world evidence research in Europe – covering roughly 74 million individuals, or approximately 90% of the population. Data Model 3 marks a clear step forward in what that data can actually tell researchers. Understanding precisely what changed, where the boundaries are, and how to plan studies accordingly is what separates well-designed research from avoidable rework.
The value of the additional granularity depends on the research question. For rare disease studies, however, another consideration can be even more important: the size of the available patient population. See Rare Disease Research Using German Claims Data.
More Detail in German SHI Data Since 2019
The FDZ Gesundheit – Germany’s Research Data Center for Health, housed at the Federal Institute for Drugs and Medical Devices (BfArM) – provides centralized, secure access to pseudonymized SHI claims data stretching from 2009 through 2024. Within that longitudinal archive, not all years are created equal. Data Model 3 applies specifically to data from 2019 onwards, introducing a noticeably finer level of detail compared to earlier structures.
For researchers working in epidemiology, health economics and outcomes research (HEOR), patient pathway analysis, or comparative effectiveness research, this distinction matters. More granular data means more precise cohort definitions, richer descriptions of care utilization, and a better ability to trace treatment sequences across time. MEDDDICAL’s overview of German health data providers and claims data offers useful context on how Data Model 3 fits within the broader landscape of data options available to researchers in Germany.
What Data Model 3 Actually Changes
Higher Granularity, Not a Full Redesign
Data Model 3 is not a ground-up rebuild of the FDZ data infrastructure. The governance framework, the secure server-based analytical environment, the disclosure review process, and the requirement to work within approved research projects – none of that changes. What changes is the level of detail available within the dataset itself.
German SHI claims data already encodes a great deal: diagnoses coded in ICD-10, inpatient procedure codes via the Operation and Procedure Classification System (OPS), outpatient procedures via the Uniform Value Scale (EBM), and prescribed medications. Data Model 3 builds on this foundation by increasing the granularity at which these elements are captured, enabling more precise observation of healthcare events.
Research Capabilities That Broadly Expand
The granularity gains from Data Model 3 support a wider range of study designs. Research areas that stand to benefit most include:
- Detailed patient cohort definitions – finer data allows more specific inclusion and exclusion criteria
- Richer longitudinal analyses – greater event-level detail over time improves trajectory modeling
- Improved characterization of healthcare utilization – captures nuances in care-seeking behavior that coarser data would obscure
- More precise treatment pathway observation – useful for both epidemiology and comparative effectiveness work
- Expanded health services research – particularly for questions about care delivery patterns across regions or provider types
Whether a given project actually needs this level of detail depends on the research question. Broad population trend analyses or national service evaluations may find that earlier data structures are entirely sufficient. The deciding factor is the analytical requirement – not simply which model is newest.
Why 2019 Is the Dividing Line
DVG’s Role in Enabling Richer Data
The 2019 cutoff reflects a direct legislative trigger. The Digital Healthcare Act (DVG), approved in late 2019, enabled the collection of invoicing data in pseudonymized form at a designated research data center – laying the groundwork for richer data availability for science. Data Model 3 is a direct expression of that expanded mandate.
Germany’s regulatory trajectory has continued in the same direction since then. The Health Data Use Act (GDNG), which came into effect on March 26, 2024, further facilitates the analysis and linkage of health data across sources, and explicitly supports the use of health data for testing and training AI applications. The FDZ Gesundheit’s operating model is also being designed in alignment with the forthcoming European Health Data Space (EHDS), which aims to simplify cross-border health data exchange for secondary use across EU member states. Each of these developments builds on the foundation DVG established in 2019.
Longitudinal Studies Across Data Model Boundaries
Earlier Years Use Different Structures
The FDZ Gesundheit’s longitudinal reach – data going back to 2009 – is one of its most compelling attributes for researchers studying disease trajectories, long-term treatment outcomes, or population-level trends over time. Studies spanning more than a decade need to account for a structural reality: years prior to 2019 are organized under earlier data models with different variable availability and granularity levels.
This requires deliberate planning. A study that treats pre-2019 and post-2019 data as structurally equivalent risks introducing inconsistencies that undermine the analysis.
Consistency Checks Before Finalizing Methods
Before locking in an analytical approach for any longitudinal study crossing the 2019 boundary, building in an explicit consistency review is worthwhile. Key questions to work through:
- Are the variables central to the analysis available and comparably defined across all reporting years?
- Does the shift in granularity affect how the exposure or outcome is measured in earlier versus later periods?
- If certain variables are only available from 2019 onwards, can the study still answer its core question using a truncated observation window?
Catching these issues at the design stage is far less costly than discovering them mid-analysis.
Known Limitations Researchers Must Plan Around
Data Model 3 is a meaningful improvement – but it does not close every gap. Known limitations persist around the granularity and completeness of pharmaceutical prescription data and incapacity-for-work data. For projects where these variables are central to the research question, assessing whether the available data adequately supports the analytical objectives before finalizing study design is essential.
The improvements in Data Model 3 can create an expectation of completeness that is not fully warranted. Framing the study around what the data can reliably support – rather than retrofitting the analysis to work around gaps discovered late – leads to stronger, more defensible research.
The FDZ Access Framework Is Actively Developing
Remote-SUF and the 2025 Health Data Lab Launch
Accessing FDZ Gesundheit data happens within a tightly governed environment. Researchers work inside a secure, virtual, access-controlled server – raw data cannot be downloaded, and all outputs go through a disclosure review before release. The access infrastructure itself, however, is not static.
The FDZ currently offers a Remote Scientific Use File (Remote-SUF) model, and the 2025 Health Data Lab launch has already broadened access further. The roadmap also includes gradual enrichment of the dataset through linkages with cancer registries and electronic patient records (ePA) – additions that would provide a considerably more complete picture of patient health across the care continuum.
For researchers planning multi-year projects, monitoring these developments is worthwhile. The data environment available at project completion may be meaningfully richer than what exists at the design stage.
Data Model 3 Raises the Bar for German RWE Research
Data Model 3 represents a genuine upgrade in what is analytically possible with Germany’s national SHI data. By providing more granular claims data from 2019 onwards, it expands the range of research questions that can be answered within the FDZ Gesundheit – without altering the governance structures that protect data integrity and patient privacy.
Researchers who align their study design with the strengths and known limits of Data Model 3 are better positioned to produce credible, publication-ready findings. That means matching the research question to the right data model, planning carefully across any pre/post-2019 boundary, and building gap assessments into the early stages of study design rather than treating them as an afterthought.
Germany’s health data infrastructure is heading in a clear direction – toward richer, more linkable, more accessible data under frameworks like the GDNG and EHDS. Data Model 3 is an important waypoint on that trajectory, not the final destination.
For researchers working with Germany’s complex health data landscape, MEDDDICAL provides expert guidance on accessing and working with SHI claims data for real-world evidence and HEOR research.