Table 1.

Elements of antibiotic prescribing in hospitals relevant for evaluating appropriateness [9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29].

Prescribing Elements (Potential Audit Variables)	Comments	Selected for Audit
START SMART
No antibiotic if not indicated (no reasonable evidence of infection)	Unnecessary antibiotic exposure selects for avoidable resistance [9,10,11].	✓
Indication documented	Good practice for continuity of care but of uncertain relevance to resistance.	✓
Appropriate specimens taken for microscopy, culture, and sensitivity (MC&S)—blood cultures and suspected site of infection	Important for establishing evidence of infection and for targeting appropriate therapy but requires manual audit and >50% of cultures are negative [12,13].	✓
No allergy or contra-indication to treatments	Important patient safety consideration but not relevant for resistance.	✕
Prompt administration of first dose	Important patient safety consideration in cases of severe sepsis but of uncertain relevance to resistance. Already captured by national sepsis audits.	✕
Treatment regimen adequate to cover most likely pathogens	Meta-analysis of RCTs reports increased risk of mortality if initial regimen inadequate [14]. Relevance to resistance uncertain.	✓ *
Treatment regimen not unnecessarily broad spectrum	Indiscriminate use of critical broad-spectrum agents unnecessarily selects for resistance [15,16,17].	✓ *
No redundant agents in treatment regimen	Unnecessary antibiotic exposure selects for avoidable resistance [9,10,11].	✓
Treatment regimen compliant with local/national guideline or justified deviation	Validity dependent upon quality of local guideline. Relevance to resistance uncertain.	✕
Treatment regimen cost-effective	Not relevant to resistance.	✕
No underdosing	Limited evidence from modeling suggests that low doses may select resistance in pneumococci [18] but underdosing unlikely to be a problem in NHS hospitals due to pharmacist and nurse intervention.	✕
No overdosing	Important patient safety consideration but likely to reduce rather than increase risk of selecting resistance [19,20,21,22,23,24].	✕
Correct route of administration	Relevant for efficacy, length of stay, and risk of line infection but of uncertain relevance to resistance.	✕
Prompt appropriate source control	Subjective assessment. Of uncertain relevance to resistance.	✕
No missed doses or delayed doses	Of uncertain relevance to selection of resistance.	✕
Therapeutic drug monitoring (TDM) for narrow therapeutic index drugs	Important primarily for patient safety (but also for efficacy); of uncertain relevance to resistance.	✕
THEN FOCUS
Prompt discontinuation of antibiotics if alternative diagnosis established and infection excluded	There is RCT evidence that unnecessary continuation selects for multi-resistant organisms [25,26,27].	✓
Appropriate broadening of spectrum in response to MC&S results	This may necessitate an increase in broad-spectrum agent use if indicated by MC&S results. Failure to adjust ineffective treatment to MC&S results is associated with a higher risk of mortality [27].	✓ *
Appropriate narrowing of spectrum in response to MC&S results	Evidence largely from observational studies suggests that de-escalation to narrow-spectrum agents is safe when patients are improving clinically and a plausible pathogen has been identified [28].	✓ *
Prompt referral to outpatient parenteral antibiotic therapy OPAT services for suitable patients	Relevant for length of stay and risk of healthcare-associated infection (HCAI) but of uncertain relevance to resistance.	✕
Prompt switch from IV to oral route of administration when safe and effective	Relevant for length of stay and risk of line infection but of uncertain relevance to resistance.	✕
Antibiotic plan documented in the notes	Good practice for continuity of care but of uncertain relevance to resistance.	✕
No unjustified prolonged duration of treatment	There is evidence from RCTs and observational studies that unnecessarily prolonged duration selects for multi-resistant organisms [25,26,29]. Can only be audited at the end of therapy.	✓

* Prescribing elements relating to antibiotic spectrum; deprioritised for audit tool prototype