Background: Procedure codes in the Danish National Patient Registry are used for administrative purposes and are a potentially valuable resource for epidemiologic research. To our knowledge, the validity of antineoplastic procedure codes has only been evaluated in one study.
Methods: We randomly extracted a sample of 420 patients in the Southern Region of Denmark with a diagnosis of colorectal cancer and an oncology contact during 2016-2018. Using the medical record as gold standard, we computed the positive predictive value (PPV) and sensitivity of antineoplastic procedure codes recorded in the Danish National Patient Registry.
Results: We identified 2,243 codes for antineoplastic treatments in the registry and 2,299 in the medical records. We confirmed that 213 of 214 patients with registered therapies in the Danish National Patient Registry received therapy, corresponding to a PPV of "any registration" of 1.00 (95% confidence interval [CI] = 0.97, 1.00). Considering single registrations, the overall PPV was 0.95 (95% CI = 0.94, 0.95), and the overall sensitivity was 0.90 (95% CI = 0.89, 0.91). Number of recorded treatments and treatments administered were strongly correlated. Considering the most frequent single antineoplastic regimens, PPV ranged from 0.90 (95% CI = 0.87, 0.92) for capecitabine to 0.98 (95% CI = 0.95, 1.00) for cetuximab, whereas sensitivity ranged from 0.81 (95% CI = 0.75, 0.87) for 5-fluorouracil and irinotecan (FOLFIRI) regimen to 0.97 (95% CI = 0.94, 0.99) for bevacizumab. Analysis per hospital showed the highest validity of registrations at the University Hospital.
Conclusion: The validity of antineoplastic procedure codes in the Danish National Patient Registry is generally high and thus usable for epidemiologic research.