Seizure Prophylaxis After Severe TBI: Levetiracetam or Phenytoin in Acute Kidney Injury
A single patient, eighteen hours into severe traumatic brain injury care. Both agents are reasonable seizure prophylaxis by the same guideline — the disagreement is about which one is reasonable for a kidney that is worsening in real time.
D.O., a thirty-four-year-old man, is three years into working bridge-resurfacing crews as a rigger, and had just started telling coworkers about a wedding date set for October when a section of scaffolding gave way under him and dropped him roughly twenty feet onto a concrete apron. EMS found him unresponsive at the scene, Glasgow Coma Scale 6, and intubated him before the twelve-minute transport to the trauma center. He is previously healthy, with no seizure history and no anticoagulant use.
Head CT in the emergency department showed bilateral frontal contusions and a thin right-sided subdural hematoma; neurosurgery elected non-operative management and placed an intracranial pressure monitor rather than proceeding to craniotomy, given his exam and the hematoma's modest size. A section of the scaffolding also came down across his right thigh and pelvis, and what looked at first like a straightforward crush injury has turned into something the ICU team is now actively managing: his creatine kinase, drawn on arrival at just over 4,000, was above 15,000 by the twelve-hour mark, and his creatinine has climbed from a normal 0.9 to 2.1 over the same window despite aggressive isotonic fluid resuscitation — a real, ongoing rhabdomyolysis-driven acute kidney injury, not yet at whatever peak it's headed for.
Eighteen hours in, he is sedated, GCS 7T, ICP holding within goal, no clinical seizure activity observed, and continuous EEG not yet in place. Bifrontal contusion plus subdural hematoma put him squarely inside the injury pattern Temkin's 1990 phenytoin trial enrolled — serious head trauma with structural injury on imaging — so nobody on the team is debating whether he needs an anticonvulsant for the first week; early post-traumatic seizures are common enough in this exact pattern, and costly enough when they raise ICP in an already swollen brain, that skipping prophylaxis isn't seriously on the table. Where he leaves that trial population is not his brain but his kidney, and the departure is still in motion: a 0.9 that became a 2.1 in twelve hours has not found its floor, so any clearance calculated from it is already out of date by the time the dose is hung. That is what makes this an argument about which drug's error would be visible — one of them fails as a number on tomorrow's chemistry panel, the other as a change in the only neurologic exam anyone has.
Eighteen hours in, choosing an anticonvulsant
Start him on levetiracetam, dose-adjusted for what his kidney is actually doing right now, not switched to phenytoin because the kidney is inconvenient. The randomized comparison we actually have — a single-center trial by Szaflarski and colleagues in 2010 that randomized 52 patients with severe TBI or subarachnoid hemorrhage two-to-one to levetiracetam or phenytoin — found no difference in seizures between them, though I'd be careful how much weight I put on that: eight seizures across fifty-two patients cannot demonstrate equivalence, only fail to detect a difference. What it did find, controlling for baseline severity, was better long-term outcomes on levetiracetam — a lower Disability Rating Scale score at three months and a higher Glasgow Outcome Scale score at six. What levetiracetam avoids, and what matters enormously in this patient specifically, is phenytoin's own real cost: free-level monitoring that only means something when you trust the albumin it's corrected against, a drug-interaction profile that complicates every other medication on his chart, and an IV loading dose that carries real hypotension and arrhythmia risk in a patient whose hemodynamics are already being managed around his crush injury. His creatinine climbing is a real problem, but it's a dosing problem — halve the maintenance dose against his current CrCl, watch the trend, and every one of those other advantages holds.
If his kidney function were already at a stable, known baseline rather than actively worsening, I wouldn't call this a close call at all — levetiracetam dose-adjusted to a fixed clearance is straightforward. The argument that follows is entirely about the fact that his creatinine is a moving target today, not a fixed number.
You're right that the monitoring burden is real — I'd rather not be correcting phenytoin levels against an albumin that's dropping from his own resuscitation. But a moving target is exactly the problem with dosing a renally cleared drug against a kidney that hasn't found its floor yet. His creatinine went from 0.9 to 2.1 in twelve hours and is still climbing; whatever dose reduction we pick today is calculated against a clearance that will likely be lower tomorrow. Levetiracetam accumulation is more forgiving than phenytoin toxicity in most patients, I'll grant that, but what it produces is sedation and behavioral change — in a patient whose only real window into his brain injury right now is a serial neuro exam, because his ICP monitor tells us pressure, not function. If he starts trending sedated three days from now, I have no clean way to tell whether that's his TBI, his ICU delirium, or his levetiracetam level climbing with his creatinine.
The Szaflarski comparison you're citing wasn't run in patients with evolving acute kidney injury — it doesn't actually answer whether the two drugs stay comparable once one of them can no longer clear the way it did in that trial's population.
Both of you are treating this as the decision that matters, and I think the decision that actually matters is the one nobody's put on the calendar. Temkin's original 1990 trial — the one that established anticonvulsant prophylaxis for early post-traumatic seizures in the first place — never showed a benefit past the first week, and no trial since has shown either drug prevents late post-traumatic epilepsy. The guideline duration is seven days, full stop, regardless of which agent we're debating right now. Whatever we start today, if it's still running on hospital day fourteen because nobody actively stopped it, that's a bigger, more common failure than picking the marginally wrong agent for one renally impaired week. I'd start levetiracetam, dose-reduced and reassessed daily against his creatinine trend, because the sedation risk is at least something a bedside exam can catch — but only if we write the day-seven stop order into the chart today, not as a task for whoever's covering him next week.
Agreed within the hour: levetiracetam started at a reduced maintenance dose, reassessed daily against his creatinine trend, with a hard stop written into the chart for hospital day seven regardless of who is covering him then. Continuous EEG was also ordered, mainly so any late accumulation-related change in his exam could be distinguished from a subclinical seizure rather than guessed at.
Not agreed: how much weight to give the accumulation risk if his creatinine hasn't turned around by hospital day three. The neurointensivist wants a standing plan to switch to phenytoin at that point, protein-binding monitoring notwithstanding, rather than wait for a sedated exam to force the question. The clinical pharmacologist would rather adjust the levetiracetam dose again first, arguing that switching drugs mid-course reintroduces exactly the interaction and infusion risks the original choice was made to avoid. Neither position was overruled — the plan for hospital day three was left as an open decision point, not a default.