Hospital-coded infections in patients with cancer: evaluating disease burden and outcomes in Australian cancer patients

Mr Jake Valentine1,2,3, Associate Professor Lisa Hall2,4,5, Professor Karin Verspoor2,6,7, Dr Tim Spelman2,8,9,10, Associate Professor Leon Worth1,2,10,11,12

1Sir Peter MacCallum Department of Oncology, University of Melbourne, Melbourne, Australia,
2NHMRC National Centre for Infections in Cancer, Melbourne, Australia,
3Paediatric Integrated Cancer Service, Parkville, Australia,
4School of Public Health, University of Queensland, Brisbane, Australia,
5Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, Australia,
6Health and Biomedical Informatics Centre, University of Melbourne, Melbourne, Australia,
7School of Computing and Information Systems, University of Melbourne, Melbourne, Australia,
8Department of Clinical Neuroscience, Karolinska Institute, Stockholm, Sweden,
9Burnet Institute for Medical Research and Public Health, Melbourne, Australia,
10Victorian Healthcare Associated Infection Surveillance System Coordinating Centre, Peter Doherty Institute for Infection and Immunity, Melbourne, Australia,
11Department of Infectious Diseases, Peter MacCallum Cancer Centre, Melbourne, Australia,
12Department of Medicine, University of Melbourne, Melbourne, Australia


Patients with cancer are at risk for a range of opportunistic and healthcare-associated infections. Administratively-coded data (ICD-10-AM) are an accessible and standardised source of data for retrospective analysis of disease burden in large populations. The objective of this study was to use hospital-level ICD-10-AM data to estimate relative infection burden and outcomes in immunocompromised cancer patients.


Episodes of infection complicating hospitalisation in adult cancer patients were identified from ICD-10-AM between 1/01/2007 and 31/12/2017 at the Peter MacCallum Cancer Centre. Patients were categorised as having haematological (HM) or solid tumour neoplasms (STN). Patients with coded infections were compared with disease-matched controls. Kaplan-Meier and regression modelling were used.


Overall, 45,116 inpatient-episodes were analysed among 3,033 HM, 18,372 STN patients and 953 autologous haematopoietic stem cell transplantation recipients, of which 67%, 29% and 88% were diagnosed with ≥1 infection, respectively. Bloodstream infections were most frequently reported in patients with Hodgkin lymphoma (123/10,000 occupied bed-days [OBD]) and myelodysplastic syndrome (109/10,000 OBDs). Risk of in-hospital mortality was higher in infected cases than uninfected controls (adjusted hazard ratio: 1.26, p=0.231). Risk of increased in-hospital length-of-stay was highest for genitourinary tract infection in HM-patients (adjusted incidence rate ratio: 2.40, p<0.001).


Infection burden among cancer patients is high, with increased infection events reported in high-risk patients (HM) compared to lower-risk populations (STN) and a significant association with poorer outcomes. Validation of administratively-coded data is required to determine if such data can be used as a reliable measure of hospital infection prevention quality improvement programmes.


Jake Valentine is a Ph.D. Candidate at the NHMRC National Centre for Infections in Cancer and the University of Melbourne. He graduated with a Bachelor of Biomedical Science (Honours) at Monash University and the Central Clinical School, Alfred Health in 2017. Jake has been actively engaged as an undergraduate research assistant at the Monash Biomedicine Discovery Institute in the fields of infectious diseases and microbiology. He has an interest in enhancing the utility of hospital discharge coding data for infectious disease detection in high-risk patients with cancer. Jake works closely with relevant stakeholders to ensure that Australia’s capitation Activity Based Funding model is sustainable and equitable for all Australian hospitals managing healthcare-associated infections in hospitalised cancer patients.

Recent Comments