gms | German Medical Science

Background: Very high completeness of endpoint data is the key to bias-free evaluations of risk associations with incident diseases in prospective cohort studies, such as the German National Cohort (NAKO). Optimally, completeness can be achieved by a combination of several methods, such as linkage with registries, health insurance companies, clinics directly, with mortality registries, or actively through collection of self-reports through study participants or next-of-kin and their optional verification through medical records viewing. Of all strategies applied in the NAKO, cancer registries are obviously the primary method of choice, the most important and also reliable resource, with robust, highly complete, and quality-controlled data and numerous variables to describe and characterize cancer phenotypes.

Methods: We will describe the latest efforts to ascertain cancer in the NAKO based on several strategies, with a focus on federal state cancer registry linkages. For the first time, we will also detail on the methodological semi-automatic approach to combine two consecutive registry-based datasets per federal state cancer registry. In addition, comparisons of data received from registries with those from health insurance companies will be presented. Derived 5-year prevalence and up to 5-year incidence as well as timelines for scientific data use will be shown.

Results: The 18 NAKO centers are distributed across Germany and are covered by 14 out of 17 federal state cancer registries (population-based and/or clinical). Based on results of our NAKO linkages, most of the population-based cancer registries have highly satisfactory information on diagnoses and vital status, whereas stage information varies greatly not only between registries but also between cancer types and years of diagnosis. With a cancer registry consent as high as 92% and 95% coverage of cancer registry linked NAKO participants at least twice since 2019, we yielded an absolute coverage of 88% of all study participants. Based on this, more than 3,000 5-year-prevalent and almost 2,700 5-year-incident cancer cases were diagnosed. Most frequent incidence tumours were those of the breast, prostate, colorectum and lung. Results of the semi-automatic linkage approach and the link with health insurance data is currently under investigation but will be ready by September this year.

Discussion: Compared to German Cancer Statistics NAKO participants present less often with neoplasms preceding baseline examination that are characterised by very low cancer survival rates, such as those of the pancreas, liver, or lung. Although cancer registry date present with high data quality and completeness, supplemental addition of further ascertainment methods are deemed relevant in the NAKO.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.