Search

Background: Prostate cancer is the most common cancer in men in the UK and the sixth-fastest increasing cancer in males. Within England survival rates are improving, however, these are comparatively poorer than other countries. Currently, information available on outcomes of care is scant and there is an urgent need for techniques to improve healthcare systems and processes. Aims: To provide prostate cancer pathway analysis, by applying concepts of process mining and visualisation and comparing the performance metrics against the standard pathway laid out by national guidelines. Methods: A systematic review was conducted to see how process mining has been used in healthcare. Appropriate datasets for prostate cancer were identified within Imperial College Healthcare NHS Trust London. A process model was constructed by linking and transforming cohort data from six distinct database sources. The cohort dataset was filtered to include patients who had a PSA from 2010-2015, and validated by comparing the medical patient records against a Case-note audit. Process mining techniques were applied to the data to analyse performance and conformance of the prostate cancer pathway metrics to national guideline metrics. These techniques were evaluated with stakeholders to ascertain its impact on user experience. Results: Case note audit revealed 90% match against patients found in medical records. Application of process mining techniques showed massive heterogeneity as compared to the homogenous path laid out by national guidelines. This also gave insight into bottlenecks and deviations in the pathway. Evaluation with stakeholders showed that the visualisation and technology was well accepted, high quality and recommended to be used in healthcare decision making. Conclusion: Process mining is a promising technique used to give insight into complex and flexible healthcare processes. It can map the patient journey at a local level and audit it against explicit standards of good clinical practice, which will enable us to intervene at the individual and system level to improve care.

In the last twenty-five years surgical practice has undergone radical change. Although this change has heralded improved patient outcomes it has also presented a new set of problems centred around the reduction in sensory inputs received by the modern surgeon. A potential solution to mitigate for this sensory loss is image guidance. The thesis has its heart four guiding aims: 1) To define the window of opportunity for the development of an image guidance platform for intra-abdominal MIS. 2) To establish whether there is a user need for image guidance in RAPN. 3) To better understand the fundamental safety and behavioural implications of the implementation of an IEOE. 4) To develop and validate a novel approach to image guidance in partial nephrectomy, utilising the preceding evidence base to inform this development. The first of these aims was met, through the development of a novel metric of innovation, with the growth potential of image guidance in abdominal surgical oncology subsequently examined. The second was addressed using a qualitative survey of robotic urologists. These two studies demonstrated that image guidance in intra-abdominal surgery lies in a period of rapid innovation growth, and that demand exists for image guidance in partial nephrectomy amongst the target population. In order to meet the third of the outlined aims, two potential drawbacks of augmenting the surgeon’s intraoperative view were examined, namely: inattention blindness, and the reliability and accuracy of the most commonly used method of image preparation: segmentation. Prior to the development of the platform the limitations of existing research platforms were examined and the problems needing to be addressed for efficacious image guidance were defined as part of a systematic review into image guidance in RAPN. These problems can be distilled down to issues pertaining to: image preparation, registration, deformation compensation, and display. Informed by this review and the findings of the previous thesis chapters a dual modality platform was developed capitalising on the respective strengths of pre- and intraoperative 4 imaging to deliver an image guidance solution offering significant benefit for both planning and execution phases. Overall, this thesis has systematically devised, and evidenced a first generation dual modality image guidance platform for partial nephrectomy, meeting the differing guidance needs for specific operative steps.

This thesis outlines the proposed attributes of healthcare which can define its quality and reviews existing research in this area. Specifically, the limitations of existing research into the volume-outcome relationship within surgery are highlighted, thereby addressing the ‘health outcome’ dimension of quality of care. The principles and existing application of funnel plots within surgery are reviewed alongside their ability to overcome the limitations, risks and implications of contemporary ranking of surgical performance. Methods by which future research should be conducted are proposed as a conceptual model. This model, along with the proposed ‘quality’ attributes, identifies the research themes of the empirical studies. The application of a validated methodological scoring system formally explores the limitations of existing volume-outcome research within the uro-oncological field. A cross-sectional analysis of administrative data assesses the volume-outcome relationship for radical cystectomy in England, using an improved methodology that incorporates a multilevel model to account for the relationship and influence of both the surgeon and institution and adjustment for institutional structural and process of care confounders. Subsequently, risk-adjusted funnel plot methodology is applied to the dataset to further explore provider performance. A longitudinal analysis assesses compliance with healthcare policy for radical pelvic surgery in England by exploring the patterns of service provision, in response to the existing understanding of the volume-outcome relationship for uro-oncology. This thesis demonstrates the limitations with existing volume-outcome methodology and the need for improved methodology for both the interpretation and presentation of volume-outcome research. Appropriate handling of the data in volume-outcome analysis, which recognises the hierarchical nature, is important to adequately inform future service reconfiguration. Volume-outcome research is just one component of a quality framework that will help align healthcare with quality improvement programmes and must incorporate a multidimensional approach to measuring and presenting both the clinical and patient-orientated measures of quality of care.