Antimicrobial stewardship in the emergency department — empirical therapy, de-escalation and an elderly stewardship pitfall

(a) Immediate empirical antimicrobial regimen (6 marks).

Treat this as sepsis from a urinary source, recognise the qSOFA 2 (respiratory rate, confusion) and the lactate above 2 mmol per litre, and apply the Surviving Sepsis Campaign Hour-1 bundle: antibiotics within one hour of recognition.^[2] Kumar and colleagues established that every hour of delay in effective antimicrobial therapy in septic shock is an independent mortality multiplier, and Seymour and colleagues showed at scale that each hour of delay in antibiotic administration raised risk-adjusted in-hospital mortality (odds ratio 1.04 per hour) — the antibiotic is given now, not after the cultures.^[1][6]

• Empirical agent: ceftriaxone 2 g intravenously as a single daily dose. Ceftriaxone covers the likely community urinary pathogens (E. coli, Klebsiella, Proteus) and the common Gram-positive uro-sepsis organisms (pneumococcus, group B streptococcus; enterococcus is intrinsically resistant and not covered), and is renal-safe at this clearance. No dose adjustment is required at a creatinine clearance of 28 mL per minute (ceftriaxone is biliary-excreted), which removes one source of error.
• Add metronidazole 500 mg intravenously every 8 hours if any intra-abdominal or pelvic source is possible (the nursing-home patient with confusion may have an occult abdominal source). Ceftriaxone has no anaerobic cover.
• Escalate to piperacillin–tazobactam 4.5 g intravenously only if healthcare-associated infection, recent antibiotics, recent instrumentation, or known prior resistant organism — not in this patient unless the local antibiogram demands it.
• Send blood cultures from two separate sites, a urine culture, and a sputum culture before the antibiotic dose when feasible within the one-hour window; the urine is the most likely source and the most likely to return positive, providing the basis for de-escalation.
• Source control — assess for an obstructed urinary tract (the renal ultrasound at the bedside, the catheter check, the consideration of a bladder scan); Reitz and colleagues showed source control within 6 hours of sepsis onset associated with lower risk-adjusted 90-day mortality (adjusted odds ratio 0.71), and the source must be sought in parallel with the antibiotic.

Justification of each excluded option: trimethoprim–sulfamethoxazole is wrong (potentiates warfarin — INR already 3.4; precipitates hyperkalaemia and acute kidney injury with ramipril in the renally impaired elderly); nitrofurantoin is wrong (creatinine clearance below 30 mL per minute, inadequate urinary concentration, and ineffective for pyelonephritis and bacteraemia); ciprofloxacin is wrong (QT prolongation, delirium in the elderly, C. difficile risk, and potentiation of warfarin); gentamicin requires renal-adjusted dosing and monitoring and is reserved for resistant organisms or beta-lactam allergy.^[4][5]

(b) Antimicrobial stewardship plan (5 marks).

• Cultures before the dose, within the one-hour window: two sets of blood cultures from separate sites, a midstream or catheter-specimen urine for culture and susceptibility, and a sputum culture if she can produce. Document the time of antibiotic administration relative to culture collection.
• The antibiotic time-out at 48 to 72 hours, when the culture and susceptibility data return and the clinical response is assessed. Four questions: is this still an infection? what is the organism and its susceptibilities? can the regimen be narrowed? can the route be switched to oral?
• Narrow to the narrowest-spectrum active agent. If the urine grows E. coli susceptible to trimethoprim and the patient is stable, switch to trimethoprim 300 mg orally daily for 3 days (the warfarin interaction is now manageable because the antibiotic course is short, the INR is monitored, and the patient is an inpatient). If the urine grows an ESBL-producing E. coli, switch to nitrofurantoin for cystitis (if creatinine clearance has improved and the isolate is susceptible) or to a susceptibility-guided oral agent, or continue intravenous therapy guided by infectious diseases — per the 2024 IDSA guidance on resistant Gram-negative infections.
• Stop if the working diagnosis is overturned (a sterile urine and a normal inflammatory pattern, with an alternative explanation for the confusion — for example, opiate toxicity, hyponatraemia, or a new stroke).
• Intravenous-to-oral switch criteria: the patient is haemodynamically stable, afebrile for 24 to 48 hours, has a functioning gastrointestinal tract, has an oral agent with bioavailability greater than 90 per cent available (levofloxacin, trimethoprim–sulfamethoxazole qualify), and is improving clinically. Document the switch and the planned total duration at the time of the switch.
• Planned duration: pyelonephritis 7 days with prompt response, 5 days for cystitis if it was a true cystitis. The duration is written on the prescription at the moment of ordering, not deferred to the ward.

(c) Three antimicrobial stewardship pitfalls (4 marks).

1. Long-term nitrofurantoin for urinary suppression is inappropriate. Both Beers 2023 and STOPP/START version 2 flag long-term nitrofurantoin for pulmonary toxicity (acute pneumonitis and chronic interstitial fibrosis), hepatic toxicity, and peripheral neuropathy, particularly in the elderly and the renally impaired. At a creatinine clearance of 28 mL per minute, the long-term nitrofurantoin is also ineffective (inadequate urinary concentration).^[4][5] It is deprescribed on this admission, in consultation with her general practitioner, and the recurrent urinary tract infection is managed by non-antibiotic strategies (hydration, vaginal oestrogen if appropriate, review of the catheter if present).
2. The trimethoprim–sulfamethoxazole–warfarin–ramipril interaction. Trimethoprim–sulfamethoxazole inhibits cytochrome P450 2C9 (potentiating warfarin within days) and acts as a potassium-sparing agent like amiloride, precipitating hyperkalaemia and acute kidney injury when combined with an ACE inhibitor or an angiotensin receptor blocker in the elderly. Both Beers and STOPP flag it; the INR of 3.4 on admission is a warning. If trimethoprim is genuinely the narrowest choice at de-escalation, the warfarin dose is reduced, the INR is monitored daily, and the ramipril is held — or the choice is changed.
3. Asymptomatic bacteriuria is over-diagnosed and over-treated in the nursing-home elderly patient. The Nicolle 2019 IDSA guideline is explicit: asymptomatic bacteriuria in the catheterised, the institutionalised, or the confused elderly patient is not treated, because treatment does not prevent symptomatic infection, does not improve delirium or continence, and selects resistant organisms.^[3] This patient has symptoms (fever, hypotension, confusion with a systemic inflammatory response) and the positive urine is consistent with symptomatic infection — but the chronic asymptomatic colonisation that drove the long-term nitrofurantoin in the first place is not treated on discharge.