Utility of the Standardized Infection Ratio (SIR) for reporting healthcare-associated infections in Australia: benefits and limitations using a Victorian dataset

Ms. Sandra Johnson1, Dr. Nabeel Imam1, Professor Allen Cheng2, Associate Profressor Leon Worth1

1VICNISS Coordinating Centre, Melbourne, Australia

2School of Public Health and Preventive Medicine, Monash University, Melbourne, Australia


Introduction: Appropriate data reporting is essential for effectual healthcare-associated infection (HAI) surveillance programs. Since 2010, the US National Health and Safety Network have reported surgical site infections (SSIs) using standardised infection ratios (SIRs), a summary statistic comparing observed with expected events. The objective of this study was to apply SIR methodology to SSI surveillance data from Victorian hospitals.

Method: SSI data submitted to the Victorian Healthcare Associated Infection Surveillance System (VICNISS) were extracted for caesarean section, cardiac bypass and colorectal surgeries. Expected infection numbers were calculated using logistic regression modelling of statistically and/or clinically-significant risk factors using 2015 data. Procedure-specific SIRs (observed infections/expected) were calculated for participating hospitals using rolling 6-month data submitted during 2016-17. Flags were applied for any 6-month period significantly greater than baseline.

Results: Modelled risk factors included age, sex, BMI, duration and other procedure-specific factors (e.g. emergency caesarean sections). The number of hospitals where SIRs could be calculated for the entire surveillance period for caesarean, cardiac bypass and colorectal surgery was 13/21 (65%), 6/10 (60%) and 14/19 (73.7%), respectively. For caesarean sections, SIR values spanned 0-2.6 (with 2 flags generated); for cardiac bypass procedures, range was 0.2-3.2 (3 flags); for colorectal procedures, range was 0-2.0 (1 flag).

Conclusions: SIRs provide a simplified measure of performance and case-mix adjustment. However, SIRs cannot be calculated for low-volume or low-risk procedures (i.e. expected number <1), limiting the value for a statewide surveillance network. Current reporting of rates should be retained, with SIR reporting used as an adjunctive measure only.


Sandra Johnson is an Epidemiologist with VICNISS. Sandra has a science background with public health and epidemiology experience. She has worked in infectious disease epidemiology, outbreak investigations and surveillance across multiple disease areas including healthcare associated infections, influenza and sexually transmissible infections. She previously worked at the Health Protection Agency (now Public Health England) in the United Kingdom and at a regional office of the Victorian Department of Health, and has a keen interest in evidence-based public health policy.

Recent Comments