Abgleich einer großen Patientinnen-Kohorte aus der klinischen Praxis mit dem Krebsregister Schleswig-Holstein

Translated title of the contribution: Record linkage of a large clinical practice patient cohort with the cancer registry Schleswig-Holstein

N. Obi*, A. Waldmann, V. Babaev, A. Katalinic

*Corresponding author for this work
7 Citations (Scopus)

Abstract

Introduction: A precondition for the evaluation of outcomes in cohort studies and screening programmes is the availability of follow-up data. In Germany, established cancer registries provide such data for incident primary cancer diseases and mortality. To utilise these cancer registry data a person's identifying code has to be correctly linked to study or programme records, a procedure which, up to date, has been only rarely used in Germany. Exemplarily, the feasibility and validity of record linkage of a cohort of 173050 patients from the Quality-assured Mamma Diagnostic programme (QuaMaDi) to the cancer registry Schleswig-Holstein was assessed by the accuracy of the classified outcome. Methods: Name, date of birth and address of the QuaMaDi cohort members were coded in the confidential administration center of the registry. These codes were passed by the codes of 129455 female cancer registry records. Datasets were synchronised for each match, so that QuaMaDi participants could be identified in the registry file. In a next step epidemiological registry records were linked to the QuaMaDi study records. The accuracy of classifying outcome was assessed by agreement measures, i.e., Cohen's kappa. In cases of disagreement, a questionnaire has been sent to QuaMaDi patients gynaecologists to validate the final diagnosis. Results: Synchronisation of both cohorts resulted in 18689 one to one matches with any kind of malignant tumour, therein 8449 breast cancers (ICD-10 C50, D05). Absolute agreement between files according to diagnosed or suspected breast cancer was 97.6% with a kappa value of 0.79. When suspicious BIRADS 4 cases from QuaMaDi were excluded, agreement and kappa rose to 99.5% and 0.948, respectively. After correction of the final diagnosis according to the physician's responses, agreement measures slightly improved in both groups of ascertained diagnosis including and excluding the suspected cases. Conclusion: Within QuaMaDi the diagnosed breast cancer cases were predominantly notified in the cancer registry. Discordant matches (false negatives and false positives) may have resulted due to various causes, thereof a very low percentage of record linkages from different persons. In conclusion, synchronisation of study cohort files to registry files using pseudonymous personal data is feasible and valid. The generated combined datasets can be used for comparative analysis of several objectives. One of them will be the evaluation of screening programmes in the near future.

Translated title of the contributionRecord linkage of a large clinical practice patient cohort with the cancer registry Schleswig-Holstein
Original languageGerman
JournalGesundheitswesen
Volume73
Issue number7
Pages (from-to)452-458
Number of pages7
ISSN0941-3790
DOIs
Publication statusPublished - 22.07.2011

Research Areas and Centers

  • Research Area: Center for Population Medicine and Public Health (ZBV)

DFG Research Classification Scheme

  • 205-02 Public Health, Health Services Research and Social Medicine
  • 205-14 Haematology, Oncology

Fingerprint

Dive into the research topics of 'Record linkage of a large clinical practice patient cohort with the cancer registry Schleswig-Holstein'. Together they form a unique fingerprint.

Cite this