From Model to Tool: The Klinik-Erreichbarkeits-Tool

Background

After the first post on hospital accessibility in Germany went online¹, the collaboration with the Science Media Center (SMC) kept going, and it grew into something bigger. What started in 2023 and 2024 as a series of individual analyses on birth clinics, care for extremely premature infants, and stroke units has since turned into a standalone, interactive research tool: the SMC’s Klinik-Erreichbarkeits-Tool, or Clinic Accessibility Tool.²

Together with several people from journalism, above all the SMC data team, we built a valid and nationally consistent dataset of existing hospital locations. It draws on additional parameters from hospital quality reports, the InEK facility registry, and a few other sources. I continue to provide the underlying accessibility calculations, based on my traffic network model.

Unlike the earlier one-off analyses, the tool is not tied to a single fixed scenario. Journalists can now put together their own scenarios, for example the hypothetical closure of one or more hospitals or specific departments, and see the effect on accessibility as a map right away.

Technical Background

At its core, the tool still relies on the traffic network model described in Part 1 and on the travel times it produces from every inhabited 100-meter grid cell, now based on the 2022 census rather than 2011, to the relevant hospital locations. Compared to the first version, though, there was one significant methodological improvement: how grid cells get connected to the classified road network.

In the original version, each grid cell was linked to the network using a fairly simple rule. As described in Part 1, this occasionally produced distortions caused by network errors, such as mislabeled one-way streets or poorly chosen connection points. For the new tool, this step was refined considerably.

For every grid cell, several candidate connection points are now identified, separated by road category. From these, a matrix of connection travel times is built, and the connection with the lowest overall travel time is selected. Individual network errors or unfavorable local conditions tend to cancel each other out this way, without requiring anyone to check every single cell by hand.

The selection of eligible roads was also restricted. Only roads that could realistically serve as access to a property are used for connections. Motorways and other high-speed roads are excluded from consideration, which means a grid cell can no longer be misclassified as sitting directly next to a motorway just because that motorway happens to be geometrically close. This makes the resulting travel times more plausible, particularly in rural areas and on the edges of cities, where motorways often run close to residential buildings without actually offering any legal access from there.

This refinement was necessary to give the tool a property that matters a great deal for journalistic research. The results need to be consistent and traceable across the entire country, no matter which region is being examined.

Results and Application

The tool lets users compare two scenarios for any selection of hospital locations, whether that is every obstetrics department or every hospital with a stroke unit. One scenario represents the status quo, and the other represents a changed state, for instance after a location has closed. The output includes not just the accessibility maps for each scenario individually, but also a difference map that shows the additional travel time in the second scenario at a glance.

The potential of the tool is especially visible in obstetrics. The map below shows scenario 2 for birth clinics from the interactive accessibility tool.

One real case examined with the tool is the closure of the delivery room at Sana Hospital in Templin in April 2024. Before the closure, the average travel time to the nearest birth clinic in the municipality of Templin was around six minutes, compared to roughly 27 minutes in the surrounding municipalities. After the closure, travel time in Templin increased by about 45 minutes. Expectant parents there now need a little over 50 minutes to reach the nearest hospital with a delivery room.

Numbers like these are meaningful on their own, but their real value comes from the consistency of the underlying methodology. Because the same calculation logic is applied everywhere in Germany, the results are directly comparable across regions. A municipality in Brandenburg can be compared to one in Bavaria without the comparison being skewed by different data sources or different calculation methods.

What really sets the tool apart, though, is not just the quality of individual analyses but its flexibility. Starting from the status quo, almost any scenario can be assembled, saved as a link, and shared with colleagues or readers. A newsroom no longer has to send us a request and wait for a custom analysis. It can run the investigation itself, quickly and independently.

For deeper analyses, such as exact values at the grid cell level or complete datasets for a publication, the SMC data team continues to provide code and data on request.

Collaboration and New Stakeholders

What began as a collaboration with Julius Tröger at ZeitOnline and Lars Koppers at the SMC has since grown into a much wider network. Alongside Lars Koppers, Simon Essink now leads data analysis at the SMC Lab, while Philipp Jacobs oversees editorial content on medicine and life sciences. This data foundation has already supported several nationwide analyses, covering obstetrics in general, care for extremely premature infants, stroke care, and, through a dedicated dashboard, the hospital reform in North Rhine-Westphalia.

The decisive step, however, was turning these individual analyses into a tool that no longer depends on us as a point of contact. The tool is aimed primarily at journalists doing regional or local research, the people tracking changes in hospital care in their own area, who want answers to their own questions using their own scenarios, without having to wait for a custom analysis from someone else.

Closing Remarks

The core message of this second phase of the project can be summed up fairly simply. A detailed, nationally consistent, and transparent calculation of hospital accessibility provides a solid foundation for both journalistic and professional research. The fact that the tool is also completely free to use, and covers the whole of Germany, makes it a strong basis for regional studies and analyses that go well beyond its original use case.

I want to thank Lars Koppers, Simon Essink, and Philipp Jacobs, and everyone else at the SMC who helped turn a series of individual analyses into a tool that is now freely accessible to anyone.

Publications

• Science Media Center, Interaktives Recherche-Tool zur Erreichbarkeit von Kliniken., April 21, 2026.
• Science Media Center, Geburtshilfe in Deutschland: Fahrzeiten, Fallzahlen, Versorgungslevel.
• Science Media Center, Auswirkungen der neuen Mindestmenge für die Behandlung von Frühchen unter 1250 Gramm.
• Krankenhausplanung NRW dashboard: krankenhausplanung-nrw.smc.report

Citation

If you cite this post, please use:

@online{holthaus_20260421_From_Model_to_Tool_The_Klinik_Erreichbarkeits_Tool,
  title   = {From Model to Tool: The Klinik-Erreichbarkeits-Tool},
  author  = {Holthaus, Tim},
  year    = {2026},
  month   = {04},
  day     = {21},
  url     = {https://me.timholthaus.com/posts/stories/20260421_from_model_to_tool_the_klinik_erreichbarkeits_tool/}
}