Antimicrobial stewardship: principles and interventions

At a glance

• Stewardship means using the right drug, dose, route and duration, and no antibiotic when none is needed.
• Antibiotics are the only drugs whose use in one patient reduces their effect in others. Each course selects resistant flora in the gut and skin, which then spread.
• In Australia, every hospital must run an AMS program under NSQHS Standard 3 (Preventing and Controlling Infections). The AMS Clinical Care Standard (2020) sets eight quality statements.
• National monitoring: NAPS (prescribing appropriateness), NAUSP (hospital use in DDD) and PBS data, reported in AURA.
• Paper 1 tests the evidence behind interventions, the measures of use, and applied scenarios where stewardship and patient safety meet.

Why it matters

• Selection pressure. Treatment kills susceptible flora and leaves resistant organisms to expand. Carbapenem use selects for CPE, cephalosporins for VRE and C. difficile, fluoroquinolones for MRSA and resistant Gram-negatives. Mechanisms are in mechanisms of antimicrobial resistance.
• Collateral harm to the patient. C. difficile infection, allergy, QT prolongation, tendon rupture, nephrotoxicity and drug interactions. See acute infectious diarrhoea.
• Australian data (AURA 2026, covering 2022 to 2024):
  • 37.1% of Australians had at least one PBS antimicrobial dispensed in a year.
  • 88.1% of restricted last-line antibiotic prescriptions in hospitals were appropriate.
  • Only 45.0% of antibiotics given after surgery were appropriate, usually because prophylaxis ran past 24 hours.

Interventions

Type	Examples	Evidence and notes
Restrictive	Formulary restriction, preauthorisation for carbapenems, newer beta-lactam combinations, IV vancomycin	Fast effect on use. Must not delay the first dose in sepsis
Persuasive	Prospective audit and feedback by an ID physician and pharmacist	Strong IDSA/SHEA recommendation. Builds prescriber learning
Enabling	Local guidelines, TG Antibiotic, electronic decision support, order-set durations	Underpins appropriate empiric choice
Structural	Rapid diagnostics (PCR panels, MALDI-TOF), procalcitonin, penicillin allergy delabelling	Diagnostics help only if someone acts on the result
Education	Prescriber teaching, patient information	Weak alone. Combine with the above

Bundled behaviours on the ward. The Clinical Care Standard asks that antibiotics are given immediately for life-threatening infection, chosen from current guidelines, documented with indication and review date, guided by microbiological testing, reviewed and de-escalated, given for the shortest effective duration, discussed with the patient, and that surgical prophylaxis follows guidelines. Penicillin allergy history is part of the decision.

Penicillin allergy delabelling

Around one in five Australian inpatients carries an allergy label to any antibiotic, and penicillin labels alone affect about 10 to 20%. Most labels are wrong. Labelled patients receive more vancomycin, fluoroquinolones and carbapenems, and have more MRSA, VRE and C. difficile. Direct oral amoxicillin challenge in low-risk patients (PEN-FAST under 3) is a stewardship intervention in its own right. See drug allergy.

Measuring use

Worked example: DDD per 1,000 occupied bed-days

The WHO defined daily dose for parenteral ceftriaxone is 2 g. A medical ward uses 600 g of ceftriaxone in a month with 900 occupied bed-days.

• DDDs = 600 g / 2 g = 300 DDD
• Rate = 300 / 900 × 1,000 = 333 DDD per 1,000 occupied bed-days

If the ward's usual dose is 1 g daily, DDDs understate the number of patient-days treated by half. DDDs also misrepresent paediatric and renally adjusted doses. Days of therapy (DOT) counts each day a patient gets a drug regardless of dose, and is preferred where electronic prescribing allows.

• NAPS measures quality, not quantity: was the drug indicated, guideline-compliant, correctly dosed and of appropriate duration?
• The WHO AWaRe classification groups antibiotics into Access, Watch and Reserve. The 2024 UN political declaration set a target of at least 70% Access use by 2030.

Duration: shorter is usually enough

Infection	Evidence-based duration
Community-acquired pneumonia, improving	5 days
Uncomplicated cystitis in women	3 days (trimethoprim), 5 days (nitrofurantoin)
Uncomplicated Gram-negative or other non-S. aureus bacteraemia	7 days (BALANCE, NEJM 2025)
Cellulitis, improving	5 days
Surgical prophylaxis	Single dose, under 24 hours

BALANCE randomised 3,608 patients with bloodstream infection to 7 or 14 days. Ninety-day mortality was 14.5% with 7 days and 16.1% with 14 days, meeting non-inferiority. It enrolled hospitalised patients, including ICU patients, and excluded S. aureus, Candida, severe immunosuppression and infections needing prolonged therapy, which still need longer courses. See bacteraemia.

• Procalcitonin can support stopping antibiotics in lower respiratory infection and sepsis once the patient is improving. It should not delay starting them.
• IV to oral switch once the patient is improving, can absorb, and an oral agent with good bioavailability exists (fluoroquinolones, doxycycline, linezolid, trimethoprim-sulfamethoxazole and metronidazole have near-complete bioavailability).

Stewardship and the sick patient

Stewardship never means withholding treatment from a septic patient. Surviving Sepsis 2026 recommends antibiotics within 1 hour for probable sepsis or shock, and suggests against empiric MRO cover in low-risk patients. The stewardship step comes at 48 to 72 hours: narrow, switch or stop. Choosing the first agent is covered in principles of empiric antimicrobial therapy.

How it's tested

• Picking the most effective stewardship intervention for a scenario (audit and feedback, restriction, prophylaxis duration).
• Calculating or interpreting DDD per 1,000 occupied bed-days, and knowing its limits.
• Applying trial evidence on duration, including who BALANCE excluded.
• Deciding to stop, narrow or switch after culture results, often with a C. difficile or MRO consequence in the stem.

Pitfalls

• Pitfall: restriction that delays the first dose of an appropriate drug in septic shock causes harm. Most programs allow the first dose before approval.
• Pitfall: surgical prophylaxis continued "until the drain is out" adds C. difficile and resistance without reducing wound infection.
• Pitfall: applying 7-day bacteraemia courses to S. aureus. It needs at least 14 days IV, longer if complicated.
• Pitfall: DDD falls can reflect lower doses rather than fewer patients treated.
• Pitfall: treating asymptomatic bacteriuria in older or catheterised patients is a leading source of inappropriate prescribing, except in pregnancy and before urological procedures.