How reliable are surgeon-reported data? A comparison of the British Association of Urological Surgeons radical prostatectomy audit with the National Prostate Cancer Audit Hospital Episode Statistics-linked database.

OBJECTIVES: To evaluate the accuracy and completeness of surgeon-reported radical prostatectomy outcome data across a national health system by comparison with a national dataset gathered independently from clinicians directly involved in patient care. PATIENTS AND METHODS: Data submitted by surgeons to the British Association of Urological Surgeons (BAUS) radical prostatectomy audit for all men undergoing radical prostatectomy between 2015 and 2016 were assessed by cross linkage to the National Prostate Cancer Audit (NPCA) database. Specific data items collected in both databases were selected for comparison analysis. Data completeness and agreement were assessed by percentages and Cohen's kappa statistic. RESULTS: Data from 4707 men in the BAUS and NPCA databases were matched for comparison. Compared with the NPCA, dataset completeness was higher in the BAUS dataset for type of nerve-sparing procedure (92% vs 42%) and postoperative margin status (89% vs 48%) but lower for readmission (87% vs 100%) and Charlson score (80% vs 100%). For all other variables assessed completeness was comparable. Agreement and data reliability were high for most variables. However, despite good agreement, the inter-cohort reliability was poor for readmission, M stage and Charlson score (κ < 0.30). CONCLUSIONS: For the first time in urology we show that surgeon-reported data from the BAUS radical prostatectomy audit can reliably be used to benchmark peri-operative radical prostatectomy outcomes. For comorbidity data, to assist with risk analysis, and longer-term outcomes, NPCA routinely collected data provide a more comprehensive source.