Treating Before the Scan: Antibiotic Timing When a Seizure Triggers a Pre-LP CT
A new seizure makes a head CT appropriate before the lumbar puncture — the team has to decide, in real time, whether that scan is allowed to become the reason antibiotics and dexamethasone wait.
D.R., a 29-year-old graduate student, lives with two roommates and has no significant medical history; she is up to date on her routine vaccinations, though she cannot recall whether she ever received a meningococcal booster. She developed a sudden, severe headache over about an hour this evening, followed by fever, photophobia, and neck stiffness over the next several hours. Her roommate brought her in after finding her unusually slow to respond and repeating questions she'd already answered. In the emergency department her temperature is 39.4°C, heart rate 118, she appears acutely ill, and Kernig's sign is present. While she is being positioned for a lumbar puncture, the triage nurse witnesses a brief, roughly thirty-second episode of focal twitching in her right hand, followed by confusion that is clearly worse than her baseline. A new-onset seizure and already-altered mental status are both real, guideline-named reasons to obtain a CT scan before attempting the LP, out of concern for a mass lesion or elevated intracranial pressure that could make the tap itself dangerous.
That correctly triggered scan now adds real minutes — transport, imaging, a radiologist's read — to a workup where the delay literature is directionally consistent but softer than it is usually quoted. All of it is observational. Proulx and colleagues (2005) found excess mortality only once door-to-antibiotic time passed six hours; Koster-Rasmussen (2008) found roughly nine percent worse odds of an unfavorable outcome per hour of delay, but reported no mortality effect; Bodilsen's Danish cohort (2016) put in-hospital mortality at 1.1 per hour, with a confidence interval running from 0.8 to 1.5 that does not exclude no effect at all. Eisen's pooled patient-level analysis (2022) found the risk of death roughly doubled past two hours, while noting heterogeneity across the source studies severe enough to make conventional meta-analysis untenable. The second clock is narrower and far better anchored: in the European Dexamethasone Study (de Gans and van de Beek, 2002), dexamethasone given fifteen to twenty minutes before or with the first antibiotic dose cut unfavorable outcomes from twenty-five percent to fifteen, with the effect concentrated in pneumococcal disease. D.R. has no Gram stain yet, so nobody in this room knows whether she belongs to the subgroup that benefit was demonstrated in — which is precisely the argument for giving the steroid on the trial's schedule rather than the scanner's, since by the time her organism is known the only window ever studied has already closed.
In the resuscitation bay, before the scanner
Cultures, then treatment, right here, before anyone moves her toward the scanner. I'll say plainly that the per-hour delay figures are weaker than they usually get quoted as — Bodilsen's mortality estimate crosses one, and Eisen's pooled analysis had to abandon meta-analysis outright because the source studies disagreed so badly. What survives all of that is the direction and the tail: Proulx's excess mortality past six hours, Eisen's doubling past two. I don't need a precise hazard per hour to know which side of that I want her on. She needs a CT because of the seizure, and that's a correctly triggered scan, but nothing about needing a scan changes when the antibiotics start.
I'm not disputing that timeline for the antibiotics themselves — I'm flagging something narrower that gets lost exactly when a room moves this fast. De Gans and van de Beek gave dexamethasone fifteen to twenty minutes before or with the first antibiotic dose, and that is the only schedule on which the benefit has ever been demonstrated. What they showed was unfavorable outcome falling from twenty-five percent to fifteen, concentrated in pneumococcal disease; the hearing-loss reduction people usually quote comes mostly from the pooled corticosteroid data across all ages, not from that adult trial by itself. Either way the timing isn't a detail of the finding, it's the condition of it.
Racing to “just get antibiotics running” is exactly the kind of urgency that gets dexamethasone pushed ten minutes late as an afterthought, and by then a meaningful part of its benefit is already gone — this isn't a slower version of your point, it's a different clock that a fast antibiotic push can actually break if nobody is watching it specifically.
Then we don't sequence any of it around each other or around the CT. Cultures drawn in under a minute, dexamethasone and the first doses of ceftriaxone and vancomycin given together right here in this bay, in that order, and she goes to the scanner after all three are already running, not before. Her seizure is a real reason to image her — it isn't a reason to wait on any of this.
One thing worth saying out loud, because it is the price of doing it this way: her lumbar puncture now happens after antibiotics, and that costs us the CSF culture. Kanegaye's series found meningococcus sterilized within two hours of a third-generation cephalosporin, as early as fifteen minutes, and the first negative pneumococcal culture at just over four. The blood cultures we are drawing right now are our best chance at an organism; Gram stain and CSF PCR hold up after pretreatment in a way culture does not. I would make the same trade again. I just don't want anyone surprised in two days when the CSF grows nothing and we are arguing about what to de-escalate.
Agreed and carried out within minutes: blood cultures drawn, dexamethasone and the first doses of ceftriaxone and vancomycin given together in the resuscitation bay, and only then did she go to CT for the seizure workup. The lumbar puncture was performed once the scan confirmed no mass effect.
Not agreed, and left open: whether vancomycin can be stopped if an organism is recovered at all and proves fully cephalosporin-susceptible — a question that may end up resting on the blood cultures rather than the CSF, given when the first doses went in. The infectious disease physician was comfortable de-escalating promptly on that result, consistent with routine antimicrobial stewardship. The neurologist was more cautious, wanting to see a second confirmatory susceptibility read given real regional variability in resistance patterns before dropping coverage. Neither position was overruled before the team moved on to reassessing her mental status.