Sir Peter MacCallum Department of Oncology - Theses
Permanent URI for this collection
Search Results
Now showing
1 - 1 of 1
-
ItemNo Preview AvailableDeveloping novel methods of infection surveillance in haematology-oncology patients and implications for health policyValentine, Jake Christopher ( 2021)Healthcare-associated and opportunistic infections are a leading cause of mortality, morbidity, and increased healthcare costs in haematology-oncology patients. Surveillance is recognised as the cornerstone of infection prevention to guide clinical decision making and to monitor quality improvement. The utility of current case ascertainment methods is poorly delineated in patients with underlying malignancy. Administrative data comprise a standardised ontology to classify disease with the potential to support large scale infection surveillance programmes; however, it is unclear if administrative data can support surveillance activities in high-risk settings. The overall aims of this thesis were to: (i) present and argue the case for novel case ascertainment methods; (ii) develop and apply a methodology using administrative data to identify infection in haematology-oncology patients, and determine the classification performance and healthcare funding implications of these data in line with current health policy; and (iii) establish and evaluate a hospital-wide linked dataset integrating multiple data sources, together with administrative data, to achieve maximal performance for automated surveillance in patients with haematological malignancies. Methods include a systematic review to describe the scope of existing surveillance methods among haematology-oncology units, analysis of continuous statewide surveillance datasets and hospital-level administrative data extracts, performance evaluation of administrative data to classify infection and simulation of pay-for-performance funding methodology in a cancer casemix, and development of a hospital-wide linked dataset to identify discrete data combination yielding highest performance for automated infection surveillance. The thesis findings demonstrate significant heterogeneity in existing infection monitoring methods among haematology-oncology patients. Estimates of the burden of disease in a predefined haematology-oncology population relative to a statewide cohort were determined, together with longitudinal trends in incidence over time. Administrative data show to be a feasible alternative to current surveillance data to enable standardised comparison of intra- and interhospital infection epidemiology in patients with underlying malignancy, however, at the expense of poor-to-moderate classification performance associated with significant shortfalls in hospital remuneration. Linkage of administrative data with microbiology, histopathology, and antimicrobial-dispensing data according to specific data combinations demonstrated improvements in classification performance for discrete opportunistic and healthcare-associated infections in patients with haematological malignancy. It was demonstrated that although administrative data enable standardised comparison of infection epidemiology, these data are an unreliable proxy for infection surveillance in Australian haematology-oncology units. Refinements to current pay-for-performance funding specifications are necessary before administrative data can reliably be used as quality improvement measures in a cancer casemix. This thesis posits data linkage as an efficient means to optimise the utility of administrative data, together with hospital-level datasets, to support an automated surveillance strategy in haematology-oncology patients. Future research agenda are outlined regarding the evaluation of electronic medical record data and other codified nomenclature to support electronic surveillance and quality improvement monitoring.