Evaluation of causes of death among Danish men with initial non-malignant histopathology of the prostate

Background

The quality of registry-based research depends on the quality of the registered data. The registration of cause of death could be affected by medical history e.g., if cause of death is unclear in a man previously suspected of prostate cancer. Therefore, we assessed the cause of death in men with initial non-malignant prostate biopsy.

Methods

From the Danish Prostate Cancer Registry (DaPCaR) men with initial non-malignant biopsies between January 1st, 1995 to December 31st, 2016 who, according to the Danish Register of Cause of Death, subsequently died of prostate cancer (n = 964) or other causes (n = 9,362) until December 31st, 2019 were identified. All men with prostate cancer death and 1000 randomly selected men who died of other causes, were manually assessed by two experienced urologists. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated for the Cause of Death registry with the manual assessment as standard.

Results

According to manual assessment, 183 men with non-malignant biopsy and subsequent prostate cancer death had metastatic disease at time of non-malignant biopsy and were excluded for further analysis. There was a total of 428 true positives and 353 false positives for prostate cancer-specific death. Moreover, 993 true negatives and 7 false negatives were assessed for other causes of death. The NPV was 99.3% (95%CI: 98.6–99.7), the PPV 54.8% (95%CI: 51.2–58.3), the sensitivity 98.4% (95%CI: 96.7–99.4), the specificity 73.8% (95%CI: 71.3–76.1) and the accuracy 79.8% (95%CI: 77.8–81.6).

Conclusion

In men with initial non-malignant biopsy, the cause of death registry is highly reliable in defining who does not die of the disease at the cost of overestimating the disease-specific death. Thus, future registry studies in similar settings should control the cause of death to assure that disease-specific mortality is not inflated.