Background: The treatment of men with locally advanced prostate cancer is becoming increasingly complex with different treatment strategies available. Locally advanced prostate cancer is likely to harbour occult nodal metastases which requires local treatment of the prostate as well as treatment of the surrounding pelvic lymph nodes. The optimal management strategy is yet to be established but particular areas of interest include the use of additional pelvic lymph node irradiation and increasing the radiotherapy dose (through a brachytherapy boost). To date, there is little data about how different radiotherapy treatment strategies affect treatment-related toxicity and cancer outcomes, especially in the general population and outside the constraints of a clinical trial. This increasing complexity of prostate cancer management will also affect the organisation and delivery of cancer services in England, the impact of which is currently unknown. Routinely collected hospital data has the benefit of being able to provide ‘realworld’ data for health services research and effectiveness studies. However, issues such as missing cancer stage data and the inability to accurately measure disease progression need to be appropriately addressed for any research affected by these issues to be considered robust. This thesis aims to investigate treatments of patients with locally advanced prostate cancer using hospital data routinely collected in England’s National Health Service. Methods: Methodological development work explored ways to improve the completeness of cancer stage information and to measure disease progression (skeletal-related events) when using cancer registry, administrative hospital and radiotherapy data linked at patient level. These datasets were also used to assess the impact of cancer service centralisation on how men with locally advanced prostate cancer are managed and to investigate treatment-related toxicity and cancer outcomes after different radiotherapy treatment strategies. Toxicity was measured according to the need for a procedural intervention, as identified through administrative hospital data, and using results from the National Prostate Cancer Audit patient-reported outcome survey. This patient survey was mailed out to men at least 18 months after diagnosis and included the Expanded Prostate Cancer Index Composite 26-item version, a validated instrument to measure functional outcomes in men with prostate cancer, and the EuroQol, which describes generic health-related quality of life. Cancer outcomes included skeletal-related events, as identified in administrative hospital data and radiotherapy data (developed earlier), and prostate cancer-specific mortality, as identified from official death records. Results: The methodological development work established and validated methods to improve the completeness of prostate cancer stage data and to identify the occurrence of skeletal-related events in routinely collected hospital data. The hospital where a man was diagnosed with locally advanced prostate cancer was not associated with whether or not he received radical treatment, but the use of surgery or a brachytherapy boost was more common if that treatment modality was available locally. Additional irradiation of the pelvic lymph nodes was not associated with worse treatment-related toxicity. However, adding a brachytherapy boost to radiotherapy was shown to be detrimental for these outcomes, especially a low-dose rate brachytherapy boost which was also associated with increased gastrointestinal toxicity. A high-dose rate brachytherapy boost was found to improve cancer outcomes compared to radiotherapy only but low patient numbers prevented any definitive conclusion with respect to a low-dose rate brachytherapy boost. Conclusions: Improvements of methods for handling missing data and identifying cancer progression can overcome some of the limitations of using routinely collected hospital data. I demonstrated that routinely collected hospital data can be used to study the impact of local radiotherapy service provision on the treatment that patients receive, and to compare treatment-related toxicity and cancer outcomes in patients receiving different radiotherapy treatment strategies. I conclude that additional pelvic lymph node irradiation may not lead to worse toxicity and should be considered in locally advanced prostate cancer. In addition, if a brachytherapy boost is considered, a high-dose rate is preferable over a low-dose rate given its lower rate of gastrointestinal toxicity and its better cancer control compared to radiotherapy only. Taken together, these results can be used to help improve radiotherapy service provision and treatment selection, especially with regards to additional lymph node irradiation and a brachytherapy boost.