Validation of the endometriosis diagnosis code in the Danish National Patient Registry from 1995 to 2018

PURPOSE: Data from the Danish National Patient Register (DNPR) is widely used in research. Although overall data quality is high, studies reveal large fluctuations in code accuracy, emphasizing the importance of evaluating code validity when used for research. In this study we examine the validity of the ICD-10 endometriosis diagnosis codes used in the DNPR from 1995 to 2018. METHODS: We assessed the positive predictive value (PPV) with 95% confidence intervals (CI) in a random sample of 300 patients, registered with at least one endometriosis code at a gynaecologic and obstetric department in the Region of Southern Denmark. Each code was assessed on two levels; disease category (endometriosis: yes vs. no) and anatomical subcategory. Medical records was used as the gold standard. RESULTS: Endometriosis was verified in 279 women, resulting in a PPV of 94.9% (95% CI 91.7-97.1). The PPV's of the anatomical subcategory codes showed large fluctuations. CONCLUSION: The ICD-10 endometriosis codes in the DNPR are highly valid for identifying patients with endometriosis (yes vs. no). If codes are used to distinguish by location, special attention is required. We discuss key insights to consider when using the DNPR to assess endometriosis status.