Ascertainment of Cancer Cases in the German National Cohort (NAKO)

Methods and Initial Results

Background: The German National Cohort (NAKO Gesundheitsstudie) is a prospective cohort study with 205 053 participants. Its goal is to investigate risk factors for chronic diseases, including cancer.

Methods: We describe the methods of ascertaining cancer cases in NAKO (linkage with cancer registries and self-reporting), the incidence and prevalence figures obtained so far, and the ratio of observed to expected (O/E) case numbers.

Results: Altogether, 2774 prevalent (diagnosed within the 5 years prior to study enrollment) and 4295 incident (diagnosed up to 5 years after enrollment) cancer cases were identified. Breast, prostate, lung, and colorectal cancers made up 55% of the incident cases. Fewer incident cases were found than would have been expected on the basis of incidence rates in the general population (O/E ratio 0.80, 95% confidence interval [0.76;0.83] for the first 2 years of prospective follow-up). The O/E ratio varied, however, across tumor sites (breast 1.02, prostate 1.12, lung 0.38, colorectal 0.62). Possible explanations include bias caused by the fact that most study participants were healthy, delayed tumor reporting, incomplete data linkage, and incomplete self-reporting.

Conclusion: Rising case numbers for the most commonly occurring types of cancer are expected in the next few years. This will enable comprehensive epidemiological analysis of the risk factors for cancer.

Cite this as: Katzke V, Arndt V, Greiser KH, Karpa F, Zoller D, Fortner R, Steindorf K, Hoffmeister M, Brenner H, Schirmacher P, Leitzmann M, Nimptsch K, Haug U, Hofmann K, Stallmann C, Mueller U, Werdecker A, Pigeot I, Swart E, Ahrens W, Meyer M, Luttmann S, Nennecke A, von Rüsten A, Kieschke J, Hartz T, Stang A, Hoffmann W, Holleczek B, Fabian HG, Peters A, Willich SN, Krist L, Schulze MB, Guenther K, Zeeb H, Ebert N, Schmidt B, Köttgen A, Kluttig A, Mikolajczyk R, Harth V, Obi N, Klett-Tammen C, Harries M, Lieb W, Becher H, Minnerup H, Karch A, Meinke-Franze C, Baurecht H, Mons U, Koch-Gallenkamp L, Berger K, Völzke H, Pischon T, Katalinic A, Kaaks R: Ascertainment of cancer cases in the German National Cohort (NAKO): Methods and initial results. Dtsch Arztebl Int 2026; 123: 439–46. DOI: 10.3238/arztebl.m2026.0061

Prospective population-based cohorts provide an optimal observational epidemiological study design to investigate risk factors and mechanisms underlying the development of common chronic diseases. The German National Cohort (NAKO Gesundheitsstudie) aims to provide the epidemiological resources needed to elucidate causes and courses of chronic disease development, including cancer, cardiovascular disease, rheumatic disease, chronic infectious diseases, diabetes, and dementia (1, 2, 3). For the NAKO study, 205 053 participants randomly drawn from German population registries were recruited between 2014 and 2019. They were examined and interviewed across 18 study centers. In epidemiological analyses, it is important to achieve high validity and completeness in prospective disease ascertainment. Depending on disease type, case documentation can be achieved by record linkage with specialized disease registries, mortality registries, health insurance data, and medical rehabilitation data, or by self-reports from study participants. Such self-reports are subsequently verified by the treating physicians and the medical records reviewed.

For cancer, NAKO uses a combination of ascertainment strategies, with linkage to cancer registries as the primary approach. Several studies have demonstrated the feasibility of linking external cohorts with cancer registries in Germany (4, 5, 6); NAKO is, however, the first epidemiological study that requires individual-record linkages for a large number of individuals across nearly all German federal states.

Here, we describe the processes for cancer ascertainment among NAKO participants, with a focus on cancer registry linkage. We explain the linkage and matching procedures of NAKO data with registry data and report preliminary data on the incidence and prevalence of cancers within NAKO, ascertained up to October 2024. Further, we compare the incidence rates observed in NAKO with those expected, based on age- and sex-specific incidence rates from the German population. Finally, we provide an outlook on the future availability of data for scientific analyses.

Methods

NAKO examinations and follow-up

A total of 103 471 women and 101 582 men aged 19–74 years were recruited to NAKO between 2014 and 2019. The participants were examined at 18 study centers across Germany, using a standardized protocol for baseline assessments of exposures and functional health examinations. Further details on the NAKO cohort have been published previously (1, 2, 3, 7, 8), and a summary can be found in the eMethods. Participants were and are invited for re-examination approximately every 5 years. Additionally, questionnaires are sent out to reassess exposures and collect information on incident disease occurrence, including cancer. Additionally, passive record linkages to cancer registries, health insurance funds, and German pension funds are conducted every 2 years.

Cancer diagnoses: cancer registry linkage and self-reports

Cancer ascertainment in NAKO relies primarily on linkage to cancer registry data. In Germany, the registration of all incident cancer cases is mandatory by law and organized at federal state level through population-based cancer registries (9, 10, 11). For NAKO, procedures for separate cohort linkages with each of the relevant federal state registries have been established on a contractual basis (for details, see eMethods, eFigure 1). It is possible that not all incident cancer cases are reported through linkage with cancer registries, e.g., due to incomplete registry coverage (< 5%) (12) or lack of consent in relation to NAKO (8% at baseline). NAKO therefore also uses additional data sources to identify incident cancer cases, such as self-reports by the participants and linkage with death registers (see eMethods) or with health insurance and pension funds data (not in this article). Linkage with data from health insurance and pension funds falls outside the remit of this article.

eFigure 1

Chart showing the data flow between the NAKO and the state cancer registries

Data processing and statistical methods

After linkage procedures, individual-level data from the registries were combined and inconsistencies resolved. In addition, self-reported incident cancers were coded using ICD-10 and merged with cancer registry data. Matching of self-reported cases to registry cases was based on:

• ICD-10 code
• Age at diagnosis
• Topography
• Morphology

The matching process started with exact agreements, and subsequently the criteria were relaxed. Furthermore, data from local mortality registries were added. Non-melanoma skin cancer (C44) was excluded from all analyses.

Prevalence of cancers derived from cancer registry data is shown only for cancers diagnosed up to 5 years prior to recruitment; data over a longer retrospective period were less complete because the various cancer registries were founded at different times. The incidence of cancer sites was derived from cancer registry data, from self-reports, and from mortality registries, in terms of total numbers as well as stratified by sex and year of follow-up.

We calculated for each cancer site the expected number of new cancer cases and the ratio of observed to expected cases (O/E) with the corresponding 95% confidence interval (95% CI). To this end, we multiplied age- and sex-specific incidence rates from the German cancer registries covered by the Center for Cancer Registry Data (ZfKD) by the number of NAKO participants in identical sex and age-group strata, using ZfKD data on cancers diagnosed in the period 2014–2021 (13). The O/E ratios were calculated only for the first 2 years of prospective follow-up, because at the time this analysis was conducted, case ascertainment for the longer follow-up times was still incomplete.

Results

Based on two rounds of linkage per cancer registry, 4295 new cases of invasive cancer were recorded up to a maximum of 5 years after recruitment. The most common were cancer of the female breast (N = 882; 44% of cancers in women), prostate cancer (N = 880; 38% of cancers in men), colorectal cancer (N = 343), lung cancer (N = 271), and malignant melanoma of the skin (N = 274; Table 1, Figure 1 a and b). As shown in Table 1, annual numbers of incident cases declined after the first years of follow-up. Within the first 2 years of follow-up, the following numbers of cases were reported through cancer registries:

Figure

The 10 most frequently occurring prevalent (a) and incident (b) cancers (C00–C96 without C44) in the German National Cohort (NAKO) according to cancer registry linkage. The diagnoses presented are those established in the 5 years immediately before NAKO recruitment in the period 2014–2019 (prevalent) and in the 5 years thereafter (incident). NHL, Non-Hodgkin lymphoma

Table 1

Frequencies of selected incident cancers (C00–C96 without C44)*1 in men and women of the German National Cohort (NAKO)

• 149 colorectal cancers
• 110 lung cancers
• 135 cutaneous malignant melanomas
• 353 prostate cancers
• 404 female breast cancers

In addition to invasive cancers, 706 in situ tumors (484 in women, of which 77% affected the breast and genital organs), 106 benign neoplasms, and 123 tumors of uncertain behavior were recorded (not shown in the table).

Within the first 2 years of follow-up, 56% (N = 1273) of self-reported incident cancers could be matched with cancer registry data. The remaining self-reports, without corresponding cancer registry records, were predominantly for prostate cancer, breast cancer, and skin cancer (eFigure 2). Overall, 82.7% (N = 1890) of confirmed cancer cases were derived from cancer registries and 15.6% (N = 357) from self-reports. A further 1.7% (N = 39) were documented exclusively from death certificates (Table 2). Participants’ self-reports played a substantial part in the documentation of some cancer types, such as bladder cancer (N = 21; 35%) or melanoma (N = 52; 28%), whereas additional DCO cases were most relevant for cancers of the pancreas (N = 6; 12%) and lung (N = 7; 6%).

eFigure 2

The cancers most often diagnosed in the first 2 years of the NAKO study DCO, Death certificate only

Compared with the expected numbers of new cancers, the observed numbers, based on registry linkage, were lower overall in the first 2 years of follow-up (O/E = 0.80, [0.76; 0.83]; Table 2). However, the ratio of observed to expected cases varied by cancer site, ranging from 0.38 [0.31; 0.46] for lung cancer to 1.18 [0.94; 1.45] for renal carcinoma. Within these limits, the ratios of observed to expected cases were 0.62 [0.52; 0.73] for colorectal, 1.02 [0.93; 1.13] for breast, and 1.12 [1.01; 1.25] for prostate cancer. The combination of registry data, self-reports, and documentation on death certificates only resulted in higher ratios of observed to expected cases than registry data alone. The ratios were as follows:

Table 2

Incident cancers (ICD-10 diagnoses C00–C96 without C44) and ratios of observed to expected cases among NAKO participants in the first 2 years of follow-up*1

• Overall: 0.96 [0.92; 1.00]
• Lung cancer: 0.44 [0.36; 0.52]
• Colorectal cancer: 0.68 [0.58; 0.80]
• Breast cancer: 1.11 [1.01; 1.22]
• Prostate cancer 1.31 [1.18; 1.44]

In total, 2774 cancer registry-derived cases (excluding ICD-10 diagnostic code C44) were diagnosed within the 5 years immediately preceding NAKO recruitment (1.5% of participants, Figure 1a and eTable 1). These prevalent cancer cases were predominantly neoplasms with high long-term survival probabilities (e.g., prostate cancer N = 454 [35% of prevalent cancers in male participants] and female breast cancer N = 758 [51% of prevalent cancers in female participants]). Only small numbers of these prevalent cancer cases were characterized by low survival probabilities, e.g., cancer of the pancreas (N = 14), liver (N = 4), or ovaries (N = 32) (14).

eTable 1

Selected data for the 5-year prevalence of cancer in NAKO (diagnoses C00–C96, without C44)

eTable 2

Self-reported cancer information in NAKO

Selected variables, e.g., topography and morphology, are shown in Table 3. They serve as criteria for data quality and are important for the majority of future epidemiological analyses, Complete data were obtained from all cancer registries on the primary tumor (ICD-10) and date of diagnosis, while morphology data were available for 96% of incident cases, and most diagnoses were based on histology or cytology (94%). By contrast, the data for tumor, lymph nodes, and metastases (TNM) were complete for a lower proportion of incident cancers (70–78%).

Table 3

Completeness of selected tumor characteristics captured in the cancer registry data for the 5 years before (prevalent) and the 5 years after (incident) NAKO recruitment

Discussion

The primary—and thus most important—source of high-quality cancer data for the NAKO study, with its 205 053 participants, are the population-based cancer registries of the German federal states. Based on two rounds of registry linkage, almost 2800 existing and 4300 newly occurring cases of cancer were documented in the 5 years immediately before and the 5 years after recruitment, respectively. Overall, the observed number of incident cancer cases was lower than expected on the basis of nationwide cancer registry data, although the ratio of observed to expected cases varied by tumor site. NAKO supplemented the cancer registry data with participants’ self-reported cancer diagnoses. This approach to ascertainment resulted in additional potential incident cancer cases, which, however, require validation via medical record review.

The reasons why fewer cases than expected were observed for some kinds of cancer are likely multifactorial. They may include healthy volunteer bias (e.g., fewer smokers in the general population: lung cancer O/E = 0.44), incomplete linkage, and time lags in cancer registration. Higher than average rates of participation in cancer screening may explain ratios above 1.0 for prostate cancer (E/O = 1.31), malignant melanoma (E/O = 1.53), and breast cancer (O/E = 1.11), and well as the ratio below 1.0 for colorectal cancer (E/O = 0.68); however, this needs to be investigated further. Discrepancies in observed versus expected case numbers probably do not affect internal validity and relative risk estimates, but may hamper external validity, especially for the estimation of absolute risks and attributable/preventable fractions when extrapolating to the general German population. To date, scientific use of NAKO data is affected not so much by imperfect ratios of observed to expected cases as by limited case numbers for some cancer sites. All these aspects need to be carefully and clearly considered for studies within NAKO.

NAKO is the first large-scale population-based study in Germany using extensive nationwide linkage of individual-level data from a number of registries for outcome ascertainment. Following the first two rounds of linkage, this is now a well-established process for all participating registries. Establishment of associations between characteristics and behavior patterns of NAKO participants prior to disease onset with occurrence of incident cancers is the key approach for identifying possible causes of disease development.

The key aspects of data quality with respect to cancer registration data, as described in detail by Bray & Parkin (15, 16) and by Arndt et al. (9), comprise:

• Comparability
• Validity
• Timeliness
• Completeness

In Germany, physicians are required to report case between 4 weeks and 6 months after diagnosis, after which cancer registries process and verify the cases within 6 weeks. In practice, the time taken for reporting and verification can vary between registries, and lag times between the date of cancer diagnosis and the transfer of data to NAKO via linkage may be as long as 2 years. In general, data quality—and especially the timely, correct, and complete reporting of cancers—depends crucially on having motivated and continuously trained physicians. Evaluations by the ZfKD indicate that population-based cancer registries in Germany capture over 95% of overall cancer incidence (12), which is comparable with the findings from cancer registries in other countries (17). As our project demonstrates, most cancer registries provide highly complete information on ICD-10-based variables, whereas stage information is less complete and varies between registries, cancer sites, and years of diagnosis. The impact of incomplete data items—e.g., TNM, ranging from 70.1% to 78.3% completeness—is currently a limitation but is expected to be resolved over the next few years, as item completeness in registries is rapidly increasing. This lack of completeness was due to the fact that some registries were recently founded or reorganized after the introduction of new federal legislation mandating clinical cancer registration. Prospectively, we expect data completeness for newly occurring tumors to be sufficient for most scientific evaluations in NAKO that require TNM data.

Complementing cancer registry data in NAKO with the participant‘s self-reports increased the overall incidence in the first 2 years by 16% and the overall ratio of observed to expected cases to 0.96. The quality of self-reported cancer data is low relative to cancer registry data, as the self-reports are limited to cancer site and year of diagnosis, with a degree of imprecision in both items. Hence, self-reports require extensive clinical validation with medical records—not only for verification purposes, but also to increase the precision of those cancer notifications (18).

In conclusion, linkages to German cancer registries covering all NAKO recruitment areas have been successfully set up and performed. As of 2024, completeness of cancer data for the entire cohort over a full 5-year period has not yet been attained, nor is there satisfactory item completeness for all of the relevant data elements, e.g., information on TNM. However, regular linkages will result in complete data on all aspects of cancer in subsequent years. Incident case numbers will increase as participants age. Verification of self-reports can be used to complement the cancer registry data, and item completeness will automatically increase in the coming years, mirroring the completeness of cancer registry data. We therefore expect that the next few years will see a highly comprehensive and complete dataset with rapidly growing case numbers for the most frequent cancers. This will allow researchers to address important questions on organ-specific cancer etiology in a large, well characterized prospective cohort.

Remaining authors

Karin Halina Greiser, Florian Karpa, Dorothee Zoller, Renée Fortner, Karen Steindorf, Michael Hoffmeister, Hermann Brenner, Peter Schirmacher, Michael Leitzmann, Katharina Nimptsch, Ulrike Haug, Katharina Hofmann, Christoph Stallmann, Ulrich Mueller, Andrea Werdecker, Iris Pigeot, Enno Swart, Wolfang Ahrens, Martin Meyer, Sabine Luttmann, Alice Nennecke, Anne von Rüsten, Joachim Kieschke, Tobias Hartz, Andreas Stang, Wolfgang Hoffmann, Bernd Holleczek, Hans Gert Fabian, Annette Peters, Stefan N. Willich, Lilian Krist, Matthias B. Schulze, Kathrin Guenther, Hajo Zeeb, Nina Ebert, Börge Schmidt, Anna Köttgen, Alexander Kluttig, Rafael Mikolajczyk, Volker Harth, Nadia Obi, Carolina Klett-Tammen, Manuela Harries, Wolfgang Lieb, Heiko Becher, Heike Minnerup, André Karch, Claudia Meinke-Franze, Hansjörg Baurecht, Ute Mons, Lena Koch-Gallenkamp, Klaus Berger, Henry Völzke, Tobias Pischon, Alexander Katalinic

Acknowledgments

We thank Dr. Jutta Kneisel for sharing her in-depth knowledge on verification of self-reported cancer via medical records in the EPIC cohort (European Prospective Investigation into Cancer and Nutrition), Dr. Linda Weisser for supporting piloting studies (NAKO study site in Mannheim), and Dr. Antje Damms-Machado for establishing initial collaborations with federal state cancer registries. We also thank the federal state cancer registries in Germany for providing individual level cancer data. We are grateful to all volunteers who took part in the NAKO study and to the staff of this research initiative.

Funding

This project was conducted with data (application nos. NAKO704 and NAKO799) from the German National Cohort (NAKO) (www.nako.de). The NAKO is funded by the Federal Ministry of Research, Technology and Space (BMFTR) (project funding reference numbers: 01ER1301A/B/C, 01ER1511D, 01ER1801A/B/C/D and 01ER2301A/B/C), the German federal states of Germany and the Helmholtz Association, the participating universities, and the institutes of the Leibniz Association.

Conflict of interest statement

André Karch is president of the German Society for Epidemiology (DGEpi).

The remaining authors declare that no conflict of interest exists.

Manuscript submitted on 1 October 2025, revised version accepted on 10 April 2026

Corresponding author
Dr. sc. hum. Verena A. Katzke

References

1. Peters A, German National Cohort C, Peters A, et al.: Framework and baseline examination of the German National Cohort (NAKO). Eur J Epidemiol 2022; 37: 1107–24

2. German National Cohort C: The German National Cohort: Aims, study design and organization. Eur J Epidemiol 2014; 29: 371–82

3. Rach S, Sand M, Reineke A, et al: The baseline examinations of the German National Cohort (NAKO): Recruitment protocol, response, and weighting. Eur J Epidemiol 2025; 40: 475–89

4. Boeing H, Wahrendorf J, Becker N: EPIC-Germany—a source for studies into diet and risk of chronic diseases. European investigation into cancer and nutrition. Ann Nutr Metab 1999; 43: 195–204

5. Guo F, Chen C, Holleczek B, Schottker B, Hoffmeister M, Brenner H: Strong reduction of colorectal cancer incidence and mortality after screening colonoscopy: Prospective cohort study from Germany. Am J Gastroenterol 2021; 116: 967–75

6. Flesch-Janys D, Slanger T, Mutschelknauss E, et al.: Risk of different histological types of postmenopausal breast cancer by type and regimen of menopausal hormone therapy. Int J Cancer 2008; 123: 933–41

7. Schipf S, Schöne G, Schmidt B, et al.: [The baseline assessment of the German National Cohort (NAKO Gesundheitsstudie): Participation in the examination modules, quality assurance, and the use of secondary data]. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 2020; 63: 254–66

8. The National Cohort—a prospective epidemiologic study resource for health and disease research in Germany. 2018. https://nako.de/wp-content/uploads/NAKO-Wissenschaftliches-Konzept.pdf (last accessed on 03 September 2025).

9. Arndt V, Holleczek B, Kajüter H, et al.: Data from population-based cancer registration for secondary data analysis: Methodological challenges and perspectives. Gesundheitswesen 2020; 82: S62–S71

10. Katalinic A, Halber M, Meyer M,et al.: Population-based clinical cancer registration in Germany. Cancers (Basel). 2023;15: 3934

11. Holleczek B, Katalinic A: Toward a comprehensive cancer registration in Germany. Eur J Cancer Prev 2017; 26: S132–S8

12. Robert Koch Institute – German Center for Cancer Registry Data: Registration completeness estimation. 2020. www.krebsdaten.de/Krebs/EN/Content/Methods/Registration_completeness_estimation/registration_completeness_estimation_node.html (last accessed on 03 September 2025).

13. Robert Koch Institute – German Centre for Cancer Registry Data (ZfKD): Database query with estimates on incidence, prevalence and survival rates of cancer in Germany, based on epidemiological federal state cancer registries. Mortality data provided by the Federal Statistical Office of Germany. www.krebsdaten.de/abfrage (last accessed on 03 September 2025).

14. Robert Koch-Institute (eds.) und die Gesellschaft der epidemiologischen Krebsregister in Deutschland e.V. (eds.): Krebs in Deutschland für 2019/2020. 14th issue. Berlin 2023.

15. Parkin DM, Bray F: Evaluation of data quality in the cancer registry: Principles and methods Part II. Completeness. Eur J Cancer 2009; 45: 756–64

16. Bray F, Parkin DM: Evaluation of data quality in the cancer registry: Principles and methods. Part I: Comparability, validity and timeliness. Eur J Cancer 2009; 45: 747–55

17. Arndt V, Siesling S: [Cancer registration in Europe]. Onkologie 2024; 30: 245–56

18. Cortes-Ibanez FO, van Pinxteren B, Sijtsma A, et al.: The validity of self-reported cancer in a population-based cohort compared to that in formally registered sources. Cancer Epidemiol 2022; 81: 102268

e1. Arbeitsgemeinschaft Deutscher Tumorzentren e.V. (ADT)/Gesellschaft der epidemiologischen Krebsregister in Deutschland e. V. (GEKID): Bundeseinheitlicher Onkologischer Basisdatensatz. 2021. https://basisdatensatz.de/ (last accessed on 03 September 2025).

e2. WHO. World Health Organization: International classification of diseases. 10th Revision. 2019. https://icd.who.int/browse10/2019/en#/II (last accessed on 03 September 2025).

e3. Stegmaier C, Hentschel S, Hofstaedter F, Katalinic A, Tillack A, Klinkhammer-Schalke M (eds.): Das Manual der Krebsregistrierung. Germering/München: W. Zuckschwerdt Verlag GmbH 2019.

e4. NAKO e.V.: NAKO Data Transfer Hub: https://transfer.nako.de/transfer/index (last accessed on 03 September 2025).

e5. Nimptsch K, Jaeschke L, Chang-Claude J, et al.: [Self-reported cancer in the German National Cohort (NAKO Gesundheitsstudie): Assessment methods and first results]. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 2020; 63: 385–96