Introduction
Antimicrobial resistance is now recognised as one of the most serious long term threats to the effectiveness of Australian health care. Every course of therapy exerts selection pressure on the bacteria carried by the person treated and by those who share their environment, which makes settings of dense, prolonged co-residence particularly consequential. Residential aged care facilities are exactly such settings. More than 190,000 Australians used permanent residential aged care during a single year, most living with multiple chronic conditions, cognitive impairment and polypharmacy (Australian Institute of Health and Welfare [AIHW], 2024). Residents are therefore both the group most exposed to antimicrobial harm and the group least able to tolerate it.
Surveillance through the Aged Care National Antimicrobial Prescribing Survey, reported in the national AURA program, shows that antimicrobial use in Australian residential aged care is high and frequently poorly documented. On any given day, approximately one in five residents is prescribed an antimicrobial, roughly one third of prescriptions are for prophylaxis rather than a confirmed infection, and a substantial proportion of orders carry neither a documented indication nor a review or stop date (Australian Commission on Safety and Quality in Health Care [ACSQHC], 2023). General practice, which supplies most antimicrobial orders to residents, shows a similar pattern of persistent excess relative to guideline expectations (NPS MedicineWise, 2020).
Urinary tract infection is the clearest example of the problem. Non-specific changes such as confusion, falls, malodorous urine or reduced oral intake are commonly interpreted as urinary infection in older residents, prompting a urine culture that then grows organisms reflecting asymptomatic bacteriuria rather than disease. International guidance is unambiguous that asymptomatic bacteriuria should not be treated in older adults, including residents of aged care facilities, because treatment confers no benefit while increasing adverse drug events, Clostridioides difficile infection and resistance (Nicolle et al., 2019). Continuous antimicrobial prophylaxis for recurrent urinary infection, often commenced during an acute episode and never reviewed, compounds the exposure.
Problem statement
Australia has strong policy scaffolding for antimicrobial stewardship in aged care but weak evidence about what actually changes prescribing in that setting. The Antimicrobial Stewardship Clinical Care Standard sets expectations for indication documentation, guideline concordant selection and review of therapy across all care settings (ACSQHC, 2020), and Standard 3 of the Aged Care Quality Standards obliges approved providers to minimise infection related risks and to promote appropriate antibiotic prescribing and use (Aged Care Quality and Safety Commission, 2023). The Royal Commission into Aged Care Quality and Safety (2021) nonetheless found clinical governance in the sector to be inconsistent and under resourced. Most published stewardship evaluations in Australian aged care are single site pre-post studies without a control condition, so it remains unclear whether observed reductions reflect the intervention, secular trends, or seasonal variation. Facilities are being asked to implement stewardship without robust local evidence about which components work, for whom, and through what mechanisms.
Aim and research questions
This project aims to evaluate the effectiveness and implementation of a tailored, multi-component antimicrobial stewardship bundle in eight Australian residential aged care facilities, using a stepped wedge cluster randomised design. Three research questions structure the study:
- Does the stewardship bundle reduce total systemic antimicrobial prescribing, measured as prescriptions per 1,000 resident days, relative to usual care?
- Does the bundle improve concordance with Therapeutic Guidelines: Antibiotic, particularly by reducing empirical and prophylactic prescribing for suspected urinary tract infection?
- Which barriers and enablers, and which implementation processes, explain variation in effect between participating facilities?
Literature Review
Antimicrobial use in Australian residential aged care
National surveillance provides a consistent picture of overuse, with AURA reporting prophylactic courses that continue well beyond six months without recorded review (ACSQHC, 2023). Surveillance definitions matter here: when Australian facility data are re-analysed against revised McGeer criteria, a large share of episodes recorded as urinary tract infection fail to meet a case definition, which indicates that the diagnostic threshold applied in practice sits well below the evidentiary one (Bennett et al., 2016). Overuse in this setting is therefore not primarily a problem of agent selection; it begins with the decision that an infection exists at all.
Stewardship interventions and audit and feedback
Antimicrobial stewardship is best understood as a coherent set of activities that promote responsible use rather than as a single restrictive control, and effective programs in residential settings combine education, decision support, prescriber feedback and structural changes to ordering. Audit and feedback is the most extensively tested component, and its effects are reliable but modest, with performance strongly moderated by design features: feedback works better when it comes from a trusted source, is delivered close to the behaviour, includes an explicit comparator, and is paired with an action plan rather than presented as a bare statistic (Brehaut et al., 2016). Much of the disappointing variance in stewardship trials plausibly reflects feedback that satisfies none of these conditions.
Prescriber and nurse behaviour
Qualitative Australian work locates the decision to prescribe in the structure of aged care rather than in individual knowledge deficits. General practitioners commonly prescribe by telephone, after hours and on the basis of a nurse’s verbal report, without examining the resident; nursing staff act as the effective gatekeepers of that report; and both groups describe family expectation and fear of a deteriorating resident as powerful drivers of precautionary treatment (Lim et al., 2015). Prescribing is consequently a distributed act. Interventions aimed only at prescribers address the final signature while leaving the assessment, the framing and the request untouched, which helps explain the weak effects of education only programs.
Nurse-initiated protocols and the evidence gap
Structured nurse-initiated pathways for suspected urinary tract infection respond directly to that distributed structure. Such protocols require documented assessment against explicit criteria, a defined period of increased hydration and observation for residents without systemic features, and restriction of urine culture to residents who meet the criteria, before the prescriber is contacted. Evaluations report reduced culture ordering and reduced treatment of asymptomatic bacteriuria, but the Australian trials are few, small and rarely controlled, and almost none report implementation fidelity or the mechanisms by which change occurred. This study addresses that gap by combining a controlled effectiveness design with a theory driven process evaluation.
Conceptual Framework
The intervention is designed using the Theoretical Domains Framework, which organises the psychological and organisational determinants of clinical behaviour into domains that can be linked systematically to intervention functions and behaviour change techniques (Atkins et al., 2017). Baseline interviews will identify which domains are dominant in each facility, and the bundle will be tailored accordingly rather than delivered as a fixed package. Figure 1 illustrates the mapping from barrier domains to intervention components to measured outcomes, together with the balancing measures monitored throughout.
Methodology
Design and setting
A stepped wedge cluster randomised trial will be conducted across eight residential aged care facilities in metropolitan and regional New South Wales, each with 80-110 beds. In this design all clusters begin in the control condition and cross over to the intervention at randomly allocated times until every cluster is exposed, which suits an intervention that cannot be withheld indefinitely and permits both within cluster and between cluster comparison (Hemming et al., 2015). An independent statistician will randomise the facilities, using a computer generated sequence, to four crossover steps of two facilities each. Data collection runs for 12 consecutive months: two baseline months with all facilities in control, four two-month steps, and two final months with all facilities exposed. The trial will be prospectively registered with the Australian New Zealand Clinical Trials Registry and reported against the CONSORT extension for stepped wedge trials.
Intervention
The bundle comprises three tailored components delivered by a visiting pharmacist and a nurse champion at each site: case based education for nursing staff and attending general practitioners anchored in Therapeutic Guidelines: Antibiotic (Therapeutic Guidelines Limited, 2024); monthly audit and feedback reports presenting facility prescribing rates benchmarked against the study cohort, with a documented action plan agreed at a short governance meeting; and a nurse-initiated protocol for suspected urinary tract infection requiring structured assessment, a defined hydration and observation period, and restriction of urine culture to residents meeting explicit criteria before a prescriber is contacted. Usual care during control periods comprises existing infection prevention practice and any pre-existing provider education.
Measures
The primary outcome is the rate of systemic antimicrobial prescriptions per 1,000 resident days. Secondary outcomes, process measures and balancing measures are set out in Table 1. Appropriateness will be judged by two blinded assessors against Therapeutic Guidelines: Antibiotic criteria for indication, agent, dose and duration, with disagreements resolved by a third assessor and inter-rater reliability reported as Cohen’s kappa.
Table 1: Outcome, process and balancing measures
| Measure | Type | Definition | Data source |
|---|---|---|---|
| Antimicrobial prescribing rate | Primary outcome | Systemic antimicrobial courses per 1,000 resident days | Electronic medication administration records |
| Urinary prophylaxis prevalence | Secondary outcome | Residents on continuous urinary prophylaxis for 28 days or more, per 100 residents | Monthly medication chart census |
| Guideline concordance | Secondary outcome | Proportion of new courses concordant on indication, agent, dose and duration | Blinded audit, 20 courses per facility per month |
| Documented indication and review date | Process | Proportion of orders recording an indication and a stop or review date at initiation | Medication chart audit |
| Urine culture ordering | Process | Urine cultures per 1,000 resident days | Pathology provider extract |
| Feedback fidelity | Process | Proportion of scheduled feedback reports delivered within 10 working days | Study implementation log |
| Protocol uptake | Process | Proportion of eligible suspected urinary episodes managed via the protocol before prescriber contact | Protocol record forms |
| Infection related hospital transfer | Balancing | Transfers with a primary infection diagnosis per 1,000 resident days | Linked admitted patient data |
| Bacteraemia and sepsis | Balancing | Confirmed episodes per 1,000 resident days | Linked pathology and admissions data |
| All cause mortality | Balancing | Deaths per 1,000 resident days | Facility records and death registrations |
Sample size and power
Each facility contributes approximately 2,700 resident days per month (90 occupied beds x 30 days), or 32,400 resident days across the 12 measurement months, giving 259,200 resident days in total and roughly 2,074 expected prescriptions at the assumed baseline rate. The study is powered to detect a reduction from 8.0 to 6.0 prescriptions per 1,000 resident days, a 25 per cent relative reduction consistent with published stewardship effects. Using the standard formula for comparing rates between clusters, with 90 per cent power, a two sided alpha of 0.05 and a between facility coefficient of variation of 0.20:
c = 1 + (1.96 + 1.28)2 x [(8.0 + 6.0) / 32.4 + 0.202 x (8.02 + 6.02)] / (8.0 – 6.0)2
c = 1 + 10.50 x [0.432 + 4.000] / 4.000
c = 1 + 10.50 x 1.108 = 12.6, that is 13 clusters per arm under a parallel design.
Because every facility contributes both control and intervention periods, the stepped wedge design is substantially more efficient. Applying a design efficiency factor of 0.32 for a five period, four sequence configuration with an assumed intracluster correlation of 0.05 (Hemming et al., 2015) gives 13 x 0.32 = 4.2, or approximately four clusters per arm equivalent, which the eight recruited facilities satisfy. Sensitivity analysis indicates that if the coefficient of variation is 0.25 rather than 0.20, power falls to approximately 80 per cent, which remains acceptable.
Statistical analysis
Analysis will follow intention to treat at the cluster period level. The primary outcome will be modelled using a generalised linear mixed model with a Poisson distribution, log link and an offset for the logarithm of resident days, including a fixed effect for intervention exposure, fixed effects for categorical calendar period to adjust for secular trend, and a random intercept for facility with a random facility by period term. Results will be reported as incidence rate ratios with 95 per cent confidence intervals. Given only eight clusters, small sample corrections to standard errors and degrees of freedom will be applied, and a negative binomial specification fitted if overdispersion is detected. Binary secondary outcomes such as guideline concordance will use mixed effects logistic regression, and pre-specified sensitivity analyses will exclude months affected by declared respiratory or gastrointestinal outbreaks.
Process evaluation
A concurrent process evaluation will conduct 32 semi-structured interviews with registered nurses, care staff, attending general practitioners and facility managers at baseline and after crossover. Transcripts will be analysed using the framework approach with the Theoretical Domains Framework as the coding structure, and quantitative and qualitative results will be integrated in a joint display that positions each facility’s effect size against its dominant barriers and implementation fidelity.
Ethical Considerations
The study will be submitted to a certified human research ethics committee and conducted in accordance with the National statement on ethical conduct in human research (National Health and Medical Research Council [NHMRC], 2018), with site specific governance authorisation from each approved provider. Because the intervention operates at facility level and targets clinician behaviour, individual resident consent for the effectiveness component is not sought; a waiver of consent will be requested for the use of de-identified routinely collected medication, pathology and administrative data, on the grounds that the research carries no more than low risk to residents, that consent is impracticable across a mobile and cognitively impaired population, and that no resident is denied treatment. Residents and families will receive plain language information at admission and by facility notice, with an opt out mechanism for the record audit.
Written consent will be obtained from all interview participants. Where a resident lacks decision making capacity and their individual record is examined outside the routine audit, consent will be sought from the person responsible under the relevant guardianship legislation, and any indication of dissent from the resident will be treated as declining, consistent with the National Statement’s position on participants with impaired capacity. Data linkage to admitted patient and death registration collections will be undertaken by an accredited state linkage authority using a separation principle, so that the research team receives only de-identified files. Because reduced prescribing carries a theoretical risk of undertreatment, an independent safety monitoring group will review balancing measures quarterly and may recommend suspension at any site.
Project Timeline
Table 2 sets out the 18 month schedule, aligning approvals, the stepped crossover sequence and dissemination.
Table 2: Project timeline and milestones
| Phase | Months | Key activities | Milestone |
|---|---|---|---|
| 1. Approvals and set-up | 1-3 | Ethics submission, site governance, linkage application, facility agreements, randomisation of crossover sequence | Ethics and linkage approved; trial registered |
| 2. Baseline and tailoring | 4-6 | Barrier interviews, data feed testing, two months of baseline measurement across all eight facilities | Baseline dataset; tailored bundle finalised |
| 3. Step 1 crossover | 7-8 | Facilities 1 and 2 receive education, first feedback cycle, protocol launch | Sequence A active; fidelity audit 1 |
| 4. Step 2 crossover | 9-10 | Facilities 3 and 4 cross over; continued feedback at earlier sites | Sequence B active; interim safety review |
| 5. Step 3 crossover | 11-12 | Facilities 5 and 6 cross over; mid-study process interviews | Sequence C active; fidelity audit 2 |
| 6. Step 4 crossover | 13-14 | Facilities 7 and 8 cross over; all sites exposed | Full exposure achieved |
| 7. Sustainment measurement | 15-16 | Final two measurement months, post-crossover interviews, data cleaning and linkage delivery | Locked analytical dataset |
| 8. Analysis and dissemination | 17-18 | Mixed model analysis, joint display integration, provider reports, manuscript and conference submission | Final report and two manuscripts |
Significance
The project addresses a policy expectation that currently outruns its evidence base. Providers are already accountable for antimicrobial stewardship under the Aged Care Quality Standards and the Antimicrobial Stewardship Clinical Care Standard, yet they have little controlled Australian evidence to guide investment (ACSQHC, 2020; Aged Care Quality and Safety Commission, 2023). By combining a design that controls for secular trend with a theory driven process evaluation, the study will report not only whether a bundle works but which components carry the effect and under what facility conditions, which is the knowledge providers and the sector regulator need. Because the intervention uses existing pharmacist and nurse roles rather than new specialist positions, a demonstrated effect would be scalable across a sector under acute workforce constraint. The appropriateness indicators also mirror those used in AURA reporting (ACSQHC, 2023), allowing participating facilities to benchmark themselves nationally after the trial concludes.
Limitations
Several limitations are acknowledged. Staff cannot be blinded to allocation, so observed improvement may partly reflect measurement reactivity, although the blinded appropriateness audit and objective dispensing data reduce this risk. Many general practitioners attend residents at more than one facility, creating potential contamination that would bias the effect estimate towards the null; attending prescriber overlap will be recorded and examined in sensitivity analysis. Eight clusters is a small number for a stepped wedge design, and estimates of the intracluster correlation will be imprecise even with small sample corrections. Appropriateness assessment depends on the quality of documentation, which is itself a target of the intervention, so improved recording could inflate apparent concordance. Finally, 18 months is sufficient to detect changes in prescribing but not in facility level resistance, and results from New South Wales facilities of 80-110 beds may not transfer to small rural or remote services with different staffing models.
Conclusion
Antimicrobial prescribing in Australian residential aged care remains high, frequently undocumented and often directed at urinary findings that do not represent infection. National standards now require providers to act, but the evidence available to guide that action is dominated by uncontrolled single site studies. This proposal describes a stepped wedge cluster randomised trial in eight facilities that tests a tailored bundle of education, benchmarked audit and feedback, and a nurse-initiated protocol for suspected urinary tract infection, with rigorous outcome, process and balancing measurement and a parallel process evaluation grounded in the Theoretical Domains Framework. If effective, the bundle offers a scalable, workforce feasible model for meeting existing regulatory obligations; if it is not, the process evaluation will explain why, which is equally valuable to a sector currently investing in stewardship without a controlled Australian evidence base.
References
Aged Care Quality and Safety Commission. (2023). Guidance and resources for providers to support the Aged Care Quality Standards.
Atkins, L., Francis, J., Islam, R., O’Connor, D., Patey, A., Ivers, N., Foy, R., Duncan, E. M., Colquhoun, H., Grimshaw, J. M., Lawton, R., & Michie, S. (2017). A guide to using the Theoretical Domains Framework of behaviour change to investigate implementation problems. Implementation Science, 12, Article 77.
Australian Commission on Safety and Quality in Health Care. (2020). Antimicrobial Stewardship Clinical Care Standard.
Australian Commission on Safety and Quality in Health Care. (2023). AURA 2023: Fifth Australian report on antimicrobial use and resistance in human health.
Australian Institute of Health and Welfare. (2024). People using aged care services.
Bennett, N. J., Johnson, S. A., Richards, M. J., Smith, M. A., & Worth, L. J. (2016). Infections in Australian aged care facilities: Evaluating the impact of revised McGeer criteria for surveillance of urinary tract infections. Infection Control and Hospital Epidemiology, 37(5), 610-612.
Brehaut, J. C., Colquhoun, H. L., Eva, K. W., Carroll, K., Sales, A., Michie, S., Ivers, N., & Grimshaw, J. M. (2016). Practice feedback interventions: 15 suggestions for optimising effectiveness. Annals of Internal Medicine, 164(6), 435-441.
Hemming, K., Haines, T. P., Chilton, P. J., Girling, A. J., & Lilford, R. J. (2015). The stepped wedge cluster randomised trial: Rationale, design, analysis, and reporting. BMJ, 350, Article h391.
Lim, C. J., Kwong, M., Stuart, R. L., Buising, K. L., Friedman, N. D., Bennett, N. J., Cheng, A. C., Peleg, A. Y., Marshall, C., & Kong, D. C. M. (2015). Antibiotic prescribing practice in residential aged care facilities: Health care providers’ perspectives. Medical Journal of Australia, 202(2), 98-102.
National Health and Medical Research Council. (2018). National statement on ethical conduct in human research.
Nicolle, L. E., Gupta, K., Bradley, S. F., Colgan, R., DeMuri, G. P., & Drekonja, D. (2019). Clinical practice guideline for the management of asymptomatic bacteriuria: 2019 update by the Infectious Diseases Society of America. Clinical Infectious Diseases, 68(10), e83-e110.
NPS MedicineWise. (2020). General practice insights report July 2018 to June 2019.
Royal Commission into Aged Care Quality and Safety. (2021). Final report: Care, dignity and respect (Vol. 1).
Therapeutic Guidelines Limited. (2024). Therapeutic guidelines: Antibiotic (Version 17).