Objective: We estimated the accuracy of ICD-10 diagnosis of ovarian cancer in a Danish discharge registry (HDR) by comparing it with Cancer Registry data (DCR).
Study design and setting: Patients (N=489) living in North Jutland County, Denmark with ovarian cancer or borderline tumour registered in the HDR or the DCR. We estimated the completeness and positive predictive value (PPV) of ovarian cancer discharge diagnosis. Mortality rates were constructed for both registries.
Results: The completeness in the HDR for ovarian cancer was 96% (95% confidence interval [CI]: 94%-98%) and PPV was 87% (95% CI: 85%-90%). 87 (18%) of the patients coded with ovarian cancer in the HDR had borderline tumours. When borderline tumours were excluded from the DCR, the PPV declined to 69% and the completeness did not change. The mortality rate ratio for ovarian cancer registered in the HDR compared to the DCR was 1.08 (95% CI: 0.90-1.29).
Conclusion: The discharge data (ICD-10) had some misclassification, but can be a valuable tool in assessment of the prognosis of ovarian cancer.