Epidemiology of castration resistant prostate cancer: a longitudinal analysis using a UK primary care database

Abstract

Background: Castration resistant prostate cancer (CRPC) is defined clinically as a failure of castration to prevent an increase in circulating hormones which are associated with worsening prostate cancer. Published information on the frequency and characteristics of CRPC patients are lacking. This may be partly because there is no specific code which doctors can use to record CRPC, making research in existing data sources such as the General Practice Research Database (GPRD) difficult.

Methods: The aim of this study was to firstly develop a method to accurately and thoroughly identify CRPC patients in the GPRD, using a combination of codes for treatments and diagnostic tests which these patients are likely to have received. Secondly, the anonymised electronic medical records were used to study the identified CRPC patient characteristics, such as age, treatments, comorbidity and expected survival.

Results: After comprehensive exploratory research, the final algorithm was selected to identify patients' assumed recording on the GPRD of either a rising prostate specific antigen (PSA) value (clinical definition of CRPC) or evidence of a switch in treatment after castration. Using the algorithm, over the 1999-2009 study period, 11600 castrated prostate cancer patients were identified. Of these, 3277 (28%) developed CRPC during the study period, with an incidence rate of 8.3 per 100 person years in castrated prostate cancer patients, and 3.8 per 100 person years in all prostate cancer patients. Mean patient age at CRPC was 76.8 years, and mean survival duration from CRPC status was 13.5 months, and from first prostate cancer diagnosis was 48.2 months. The most common comorbidities among CRPC patients were hypertension, dyspnoea, and anaemia.

Conclusions: Sensitivity analyses to test algorithms with differing inclusion criteria suggested that the primary algorithm was robust, reducing bias through missing data, while avoiding 'false positives' and retaining clinical credibility. Overall, our study has documented a reliable method for identifying CRPC patients in GPRD, and thus we have been able to characterise these patients more accurately.