Typhoid Fever From Karachi: Outpatient Azithromycin or an Admission for Ceftriaxone
A software engineer back from three weeks in Karachi has a classic stepwise fever and suspected typhoid, in a region where the circulating strain is resistant to fluoroquinolones and third-generation cephalosporins together. The debate isn't just which drug to start — it's whether he can safely be treated as an outpatient at all before culture results are back.
R.V., a 27-year-old software engineer, spent three weeks in Karachi on a client engagement his company had been trying to land for over a year, and by his own sheepish account ate everything his local colleagues put in front of him despite the pre-travel counseling he'd technically received. He's been home five days. The fever started gradually four days before that — low-grade at first, climbing a little higher each evening in the stepwise pattern that's more typhoid's signature than any single number on the thermometer — and today it's 39.8°C. He describes constipation rather than diarrhea, a detail the resident almost didn't ask about because it runs against what most people expect from a travel-associated GI illness, and his pulse on exam is 84 against a temperature that should be driving it well past 100 — relative bradycardia, one of the few semiotic signs that still points toward typhoid before a single culture result exists.
The region he's coming from is not an incidental detail. Since 2016, Sindh province has been the site of an outbreak of extensively drug-resistant Salmonella Typhi — a single circulating H58 lineage clone, characterized by Klemm and colleagues in 2018, that combined fluoroquinolone resistance with third-generation cephalosporin resistance for the first time anywhere, closing off two of the drug classes clinicians would otherwise reach for without a second thought. Azithromycin has held up as the most reliably active oral option against that strain, which is part of why current guidance for suspected typhoid acquired in Pakistan specifically favors it empirically over the older fluoroquinolone reflex. But azithromycin resistance in the same region is not zero — still a minority finding, but a real and increasingly reported one — and a drug that fails silently in an outpatient who isn't being watched is a different risk than one that fails in a monitored bed.
Outpatient pill bottle or an admission order
Azithromycin, outpatient. The Sindh province XDR clone lost fluoroquinolone and third-generation cephalosporin activity years ago, but it has stayed reliably susceptible to azithromycin, which is exactly why current guidance for typhoid acquired in Pakistan leads with it empirically rather than falling back on ceftriaxone by habit. He has none of the markers — no altered mental status, no GI bleeding, no peritoneal signs — that would push this toward an automatic admission.
If this were a region where azithromycin resistance was common rather than a documented minority, I'd be having a different conversation. It isn't, yet, in Karachi specifically.
I'm not disputing the susceptibility data — azithromycin resistance in that clone is real but still uncommon. What I'm weighing is what happens if he's the uncommon case and he's at home instead of in a bed we're watching. Typhoid's worst complications, perforation and GI hemorrhage, cluster in the first several days of illness, and a drug that's quietly failing doesn't announce itself until one of those happens.
He looks well today. That's exactly the population where perforation catches everyone off guard, not the population where it's already obviously coming.
The perforation argument is the one I'd actually act on, and I don't think admission is how you act on it. A bed doesn't lower perforation risk; catching a failing drug early does, and the bed is only one of several ways to do that — and the least specific one, since nobody is watching his abdomen hourly on a medical floor either. Build the catching in directly instead: azithromycin today, but a structured in-person recheck at 48 hours, not a vague callback instruction, and a clear pre-agreed trigger — no defervescence or any new red-flag symptom by 72 hours — for coming straight back for IV ceftriaxone.
A safety-netted outpatient plan isn't a weaker version of admission, it's a different way of buying the same protection for a patient who currently doesn't meet any criterion for needing a hospital bed.
Agreed: azithromycin started today as an outpatient, with a scheduled in-person recheck at 48 hours rather than an open-ended callback instruction, and an explicit, written threshold for returning immediately — persistent fever past 72 hours, or any new abdominal pain, bleeding, or mental status change before then.
Not fully agreed: the hospitalist would still have preferred admission today given the regional azithromycin-resistance data, and said so plainly even after accepting the safety-netted outpatient plan as reasonable given how well R.V. currently looks. The infectious disease physician's own position, equally unmoved, was that nothing about the case actually met an admission criterion and that treating him as an inpatient purely on regional statistics would be overtreating the individual in front of them. Both signed off on the same plan; neither changed the other's underlying read of how much weight the resistance data alone should carry.