Gram-Negative Bacteremia: Seven Days or Fourteen
A pyelonephritis-source bacteremia is clearing fast on antibiotics. The disagreement isn’t about whether she’s responding — it’s about whether a stone still sitting in her ureter means the seven-day evidence doesn’t actually apply to her yet.
Renata S., a 58-year-old woman who has run the cafeteria at the same elementary school for over twenty years, came to the emergency department three days ago with a fever of 39.4°C, flank pain, and shaking chills that started the evening before — the kind of onset, she told the team, that felt nothing like the bladder infections she has had twice before and treated at home with a few days of antibiotics from her primary doctor. Blood and urine cultures both grew Escherichia coli, and a CT scan ordered for the flank pain found the actual explanation: a 6mm stone lodged in the mid-ureter, partially obstructing but not fully blocking, with mild hydronephrosis behind it. Urology was consulted the same day and elected observation rather than emergent stenting, judging the partial obstruction unlikely to worsen on IV antibiotics alone, with elective stone removal scheduled for next week rather than this admission.
She has had type 2 diabetes for eleven years, managed on metformin with an A1c that has run in the low 7s — not tightly controlled, not neglected, and not itself the reason she is in the hospital, though it is part of why the team is being careful rather than reflexive about how long to treat her. By hospital day three she is afebrile, her white count has fallen from 16,200 to 9,100, and a repeat blood culture drawn on day two has already returned negative. The question dividing the team is not whether she is responding — she plainly is — but whether a stone still sitting in her ureter, not yet removed, means she does not actually resemble the patients who made up the trial the team would otherwise be citing to justify stopping early.
BALANCE, the trial the team keeps returning to, randomized nearly 3,600 patients with gram-negative bloodstream infection to seven or fourteen days and found the two arms statistically indistinguishable at 90 days, but the trial reached that result having excluded at enrollment exactly the sources it could not assume were controlled — an undrained abscess, an unremoved infected prosthesis — categories a stone still sitting where it was found resembles more than it resembles a drained collection or a pulled line. Reading the trial’s own enrollment logic against her chart is what makes this a genuine question rather than a settled one: her fever curve and her cultures look exactly like a BALANCE responder, but her retained stone looks like exactly the kind of source the trial’s own criteria would have screened out before randomization ever happened.
On rounds, day three of admission
Seven days is the right course here, not fourteen. BALANCE randomized nearly 3,600 patients with gram-negative bacteremia to 7 versus 14 days and found noninferior 90-day mortality, with the shorter arm carrying less C. difficile and less resistance pressure for the same outcome. She is afebrile, her count has normalized, and her second blood culture is already negative — this is exactly the clinical trajectory the trial describes.
I understand the stone is still in place, but nothing about seven days assumes the source is surgically removed — it assumes the source is controlled, and a non-obstructing collection draining adequately around a partial stone, with clearing cultures to prove it, is control by any functional definition.
I don’t dispute the noninferiority numbers, and I’m not arguing for fourteen days as a reflex. What I’m not comfortable with is treating "functional control" as equivalent to what BALANCE actually enrolled. The trial excluded at randomization exactly the source categories it couldn’t assume were controlled — an undrained abscess, an unremoved infected prosthesis. A stone still sitting there, however well she’s doing today, resembles those excluded categories more than it resembles a drained collection or a pulled line.
The rebuttal isn’t that seven days is wrong. It’s that citing a trial’s result for a patient the trial’s own enrollment criteria would likely have excluded is borrowing evidence it hasn’t earned yet.
Both of you are answering different questions as if they were the same one. The duration question is easy once the source-control question is settled — it just isn’t settled yet. Urology can tell us within 48 hours, on repeat imaging, whether that stone is still obstructing meaningfully or has effectively become a non-issue while she’s on therapy.
So: start the oral step-down today, count day seven as the earliest possible stop date rather than the committed one, and let the day-five imaging answer the actual disagreement instead of either of you pre-deciding it from the chart.
Agreed: continue definitive therapy, transition to oral ciprofloxacin today, and obtain repeat renal ultrasound on hospital day 5 specifically to answer whether the stone is meaningfully obstructing.
Not agreed, and left as the actual open branch point rather than smoothed into a single default: whether a normal day-5 ultrasound justifies stopping at seven days outright, or whether the still-retained stone — imaging notwithstanding — warrants extending to ten days as a middle position neither original side had proposed.