The Onkonet database: taking stock of an Internet-based, multi-centre database on surgical prostate cancer treatment

Abstract

The Onkonet database has been developed and coordinated by the Berliner Tumorzentrum e. V. (http://www.prostata-ca.net) and contains data on pre-, peri- and postoperative parameters of radical prostatectomy documented since January 2005. With its user-friendly interface and its integrated benchmarking tool, the main goal of Onkonet was to outline and improve the surgical care of prostate cancer patients in Germany. This study aimed to analyse all Onkonet data documented from the beginning of the project until June 2018. We focused on the completeness and plausibility of data to investigate and define the possibilities and limits of further analyses. All patients who underwent radical prostatectomy in one of the urological clinics participating in this project until June 2018 were included in this retrospective study. The completeness of all documented patient data was analysed using Excel 2013. The statistical analysis was descriptive. A total of 21 474 patients were documented in Onkonet. 58,6 % (12 591) of them had a complete dataset including date of birth, date of surgery, dates of hospitalisation and discharge, initial PSA value, Gleason score of the biopsy, clinical T stage, pathological T stage, pathological Gleason score, as well as information on the surgical technique. Mean completeness of pre-operative parameters was 26,8 %, of hospitalisation parameters 64,5 %, and of pathological parameters 58,1 %. Amongst these, the documentation of the pathological T stage was complete in 80,1 %, documentation of N stage in 78,8 %, of M stage in 74,8 %, of pathological Gleason Score in 78,7 %, and of R1 status in 78,7 %. Completeness of follow-up data was 8,1 %, with PSA data being available in 27,2 %, continence data in 23,0 %, and potency data in 13,9 %. Comprising 21 474 documented patients and over 200 parameters, Onkonet is one of the most comprehensive clinical registers for the documentation of prostate cancer patients in Germany. The data analysis showed that the limitations of such a database are mainly due to the high number of parameters and the high susceptibility to errors due to manual data submission.