Background: The National Death Index (NDI) is a centralized database containing information from death certificates that are frequently referenced by health and medical investigators to ascertain vital statistics. Yet, it commonly includes misclassified causes of death. Since the NDI is frequently relied upon in studies that evaluate outcomes following radical prostatectomy (RP) for prostate cancer (PC), we evaluated its validity by referencing mortality data from the Shared Equal Access Regional Cancer Hospital (SEARCH) database which is a prospectively managed database of 5009 Veterans who underwent a RP at eight Veterans Affairs medical centers between 1982 and 2016.
Methods: We compared vital status, cause of death and date of death from the SEARCH database with the NDI.
Results: A total of 1312 men in SEARCH were deceased, yet the NDI reported 17% (219) of those men as still alive. Among the 1093 men who had concordant vital status in both SEARCH and NDI, the date of death was an exact match within one day, a week, or 31 days in 94%, 97%, 99%, and 100%, respectively. Of those men coded as dying from prostate cancer in the SEARCH database (n = 105), 12% were coded as having died from non-PC causes in the NDI. Meanwhile, among patients coded by the NDI as having died of PC (n = 139), 34% were coded in SEARCH as having died of non-PC causes.
Conclusions: These findings demonstrate that the NDI provides accurate dates of death, but frequently misclassifies whether a death was due to prostate cancer. Studies that rely upon death certificates, as captured in the NDI, may be unreliable to report prostate cancer-specific mortality rates after prostatectomy.