[Garbage in - garbage out? Validity of coded diagnoses from GP claims records]

Gesundheitswesen. 2009 Dec;71(12):823-31. doi: 10.1055/s-0029-1214399. Epub 2009 Apr 22.
[Article in German]

Abstract

Context: ICD-10-coded diagnoses from claims records are frequently used as morbidity indicators for research as well as for risk adjustment purposes in quality management and remuneration. A requirement for this application is the high validity of the diagnoses. In GP practices in particular, it is questionable whether claims-based diagnoses realistically reflect the health problems of patients treated over a one year period.

Methods: In a retrospective cross-sectional study of a random sample of 250 patients from 10 GP practices we examined whether, on the basis of the patients' medical records, health problems treated in the year 2003 matched claims-based diagnoses within the same time period.

Results: In spite of a high mean of 6.1 claims-based diagnoses per patient, health problems treated within the study period were under-reported in 30% of the cases, mainly relating to non-severe diagnoses frequently encountered in GP practice, chronic conditions not requiring medication, and diagnoses justifying a screening test. An over-reporting for diseases not treated within the study period was observed in 19% of the cases, most often in the case of permanent chronic conditions. In 11% of cases the ICD-10 codes of claims-based diagnoses and the diagnoses in the medical records did not match ("erroneous codes"). For six of the diagnoses most common in GP practice (hypertension, diabetes, hyperlipoproteinemia, cardiovascular disease, back pain, and acute respiratory tract infections) correctness at 71-93% was higher than completeness (56-86%).

Conclusion: The low validity of ICD-10-coded diagnoses from GP claims records calls their usefulness as morbidity indicators into question.

Publication types

  • Validation Study

MeSH terms

  • Cross-Sectional Studies
  • Diagnostic Errors / statistics & numerical data*
  • Germany / epidemiology
  • Humans
  • Insurance Claim Reporting / statistics & numerical data*
  • Insurance Claim Review
  • International Classification of Diseases / statistics & numerical data*
  • Medical Records / statistics & numerical data*
  • Physicians, Family / statistics & numerical data*
  • Reproducibility of Results
  • Retrospective Studies
  • Sensitivity and Specificity