The Fever That Might Be the Antibiotic Itself
A patient doing well clinically on the one antibiotic that actually covers his infection develops a new, otherwise-unexplained fever — the debate is whether that fever is reason enough to give up the only drug that's working.
Roland K., a 61-year-old man, is nine days into a planned six-week course of intravenous cefazolin for MSSA infective endocarditis of his native mitral valve — caught early, no perivalvular extension on echocardiogram, no embolic events, and until yesterday, an entirely unremarkable admission. He is a retired postal carrier who still walks his old route most mornings out of habit, and had been asking daily when he could go home to finish the course as an outpatient. Overnight his temperature rose to 38.7°C for the first time since hospital day two, with no new symptoms he can identify — no chills, no rash, no joint pain, no change in the murmur, no new shortness of breath.
His exam is otherwise unremarkable, and that absence is doing real work here: no rash anywhere, normal mental status, no new murmur suggesting valve progression. His labs show a new eosinophilia — 640 cells/µL, up from a normal differential on admission — and a heart rate of 74 that has not risen proportionally to his fever, a relative bradycardia some clinicians read as a classic drug-fever clue, though Mackowiak and LeMaistre's critical review of 148 episodes found it in only a minority of them — present often enough to be worth noting when it appears, far too often absent to be leaned on. Two repeat blood cultures drawn this morning are pending. He has no prior drug allergies on record, has never previously been exposed to a cephalosporin, and his renal function — stable at a baseline creatinine of 1.0 through the admission — has not shifted with the new fever, arguing against a renally-mediated accumulation explanation. Nothing about his valve, his infection control, or his overall trajectory has changed in the other direction either: his most recent troponin remains undetectable, his repeat echocardiogram from two days ago is unchanged from admission, and he has had no new embolic phenomena on serial neurologic and skin exams. The only new fact, arriving into an otherwise steadily improving picture, is the fever itself, nine days into the one drug his endocarditis actually needs — a timing that is entirely compatible with drug fever without being evidence for it, since the same review found the lag between starting a drug and the fever appearing to be highly variable rather than clustered in any usable window. The clock, in other words, cannot settle this either way.
Day nine of therapy, a new fever with no other explanation
Any new, otherwise-unexplained finding on a drug is grounds to stop the drug first and sort out the cause after — that's the safer default, especially with a hypersensitivity reaction, which can look mild on day one of the fever and progress before its full picture is visible.
I'm not saying this is DRESS or hypersensitivity myocarditis today. I'm saying we don't get to know that it isn't just by watching — the picture we'd want to catch it early looks exactly like this one.
I take the caution seriously, but look at what's actually here against what a drug reaction needing discontinuation actually looks like: isolated fever, mild eosinophilia, no rash, no organ dysfunction, otherwise clinically improving. That's close to the textbook picture of uncomplicated drug fever, not an evolving hypersensitivity syndrome.
You're right that hypersensitivity reactions can start this quietly — but stopping cefazolin here means switching to vancomycin for a valve infection where cefazolin is the better drug on cure rates, and doing that on a single fever spike without giving the diagnosis a chance to declare itself trades a real efficacy cost for a risk we haven't actually confirmed yet.
Neither of you has to be right on belief alone — there's a direct test here. Hold cefazolin for one dosing interval, watch the curve. Isolated drug fever characteristically defervesces within roughly forty-eight to seventy-two hours of stopping the culprit; an evolving infection or a coincidental process typically does not.
That gives us an answer built on his actual physiology rather than either of your priors. The real constraint is how long we can safely hold the drug given how early we still are in treating his endocarditis — one interval, with cultures and exam reassessed daily during it, is short enough that I don't think it meaningfully jeopardizes valve control either way.
Agreed: cefazolin held for one dosing interval, repeat blood cultures already in hand, temperature and exam reassessed every four hours through the window — the hospitalist's caution and the infectious disease physician's reluctance to switch both folded into the clinical pharmacologist's dechallenge plan rather than either being overruled outright.
Not agreed: what happens if the fever curve is genuinely ambiguous — neither clearly resolving nor clearly persisting by the end of the interval. The hospitalist would restart on a different agent regardless; the infectious disease physician would want one more interval before conceding the drug. That branch was named explicitly rather than smoothed over, to be decided only if the curve doesn't give a clean answer.