Lithium Toxicity: When the Serum Level Lags the Exam
An older man on stable long-term lithium develops ataxia and a witnessed myoclonic jerk after a new medication and a few days of poor intake. His lithium level looks only mildly elevated — his exam doesn't.
Walter B., a 71-year-old retired machinist, has taught teenagers to build birdhouses at the community woodworking shop most Thursday afternoons, and skipped the last two: the first because his hands wouldn't hold a chisel steady, the second because the walk back to his car had started to feel unreliable in a way he didn't have a word for. He put both down to getting older. He has taken lithium for bipolar I disorder for fifteen years, at a stable dose, without a toxicity episode in that entire time. Ten days ago his primary care physician started him on lisinopril for newly diagnosed hypertension. Three days ago he came down with a stomach bug that left him eating and drinking a fraction of his usual intake, something he mentioned to no one because, in his own words, "it was just a bug."
His lithium level, drawn today, comes back at 2.0 mEq/L — clearly elevated above his usual therapeutic range, but well under the 4.0 mEq/L at which the EXTRIP workgroup recommends dialysis for a patient whose kidney function is impaired. His exam tells a different story than the level does: dysarthria, a wide-based, unsteady gait his coordinator first noticed as tremor, and today, witnessed by the triage nurse, one brief myoclonic jerk of his right arm. His creatinine has climbed from a baseline of 1.0 to 1.6 — impaired kidney function, which puts him inside one half of that EXTRIP criterion while his level sits at half the other — consistent with the volume depletion from three days of poor intake compounded by lisinopril's own effect on renal sodium handling — ACE inhibitors reduce proximal tubular sodium reabsorption, and because lithium is reabsorbed alongside sodium in that same segment, less sodium reabsorption means more lithium reabsorption, quietly raising his lithium level over the past week and a half without anyone measuring it until today. The number and the exam disagree about how sick he is, and the gap between them is the actual clinical question: in chronic lithium toxicity, serum level lags CNS tissue burden, since lithium's brain equilibration is slow in both directions — his 2.0 may represent a serum compartment already declining relative to a central nervous system compartment still loaded well past it, which would mean the level in front of the team is a trailing indicator, not a current one. Where his findings land in EXTRIP's scheme is the argument in miniature: the recommendation to dialyze irrespective of level names depressed consciousness, seizure, or life-threatening dysrhythmia, and he has none of the three. His mild confusion sits one tier down, in the weaker suggestion.
In the ED, a level that undersells the exam
I'd start dialysis now, not wait to see if fluids alone turn this around. I'll concede the EXTRIP workgroup's 2015 wording doesn't hand me this outright: its irrespective-of-level recommendation names depressed consciousness, seizures and life-threatening dysrhythmias, and his confusion is mild enough to sit in the suggestion tier. What it also says is that dialysis is recommended in severe lithium poisoning, and the picture EXTRIP describes as severe — myoclonus, depressed consciousness, seizures — is one he has already entered at one end. The 4.0 cutoff everyone remembers is paired with impaired kidney function, and it was derived overwhelmingly from acute ingestions, where serum level and CNS burden haven't yet decoupled the way they do in a chronic exposure like his. His level of 2.0 may be genuinely misleading about how loaded his brain actually is right now. I don't want to find out in six hours that his myoclonus was the leading edge of something worse.
I've managed him for twelve of his fifteen years on lithium, stable the entire time, and I don't take that history lightly when I'm weighing an invasive procedure in a 71-year-old whose blood pressure is already sitting at 108 over 64. We have two clearly identifiable, reversible precipitants — three days of poor intake and a brand-new ACE inhibitor doing exactly what ACE inhibitors are known to do to renal lithium handling. Stop the lisinopril, run aggressive isotonic fluids, and give that a real trial before we put a dialysis catheter in a volume-depleted man whose pressure could drop further the moment we start pulling fluid off him.
Fifteen years without a prior episode is real, but it's not evidence about how he'll do with this episode — his baseline tolerance tells us nothing about how far along toward SILENT he already is today.
I don't think the actual disagreement between you two is whether severity matters — it's where the threshold sits, and I'd draw it around the myoclonus specifically rather than around his exam as one undifferentiated picture. Tremor and ataxia alone, in a chronic-toxicity patient with reversible precipitants, would make me comfortable trying fluids and holding the lisinopril first. A witnessed myoclonic jerk changes that for me — myoclonus is named in the cluster EXTRIP calls severe poisoning, alongside the seizures and depressed consciousness that trigger its strong recommendation, and I don't want to bank six hours on a conservative trial working. Set the trial anyway, but with explicit stopping rules named right now, not decided in the moment: any seizure activity, any further decline in his mental status, or no measurable neurologic improvement at six hours mandates dialysis without another round of debate. Start the fluids and stop the lisinopril immediately either way — nobody here is arguing against that part.
Agreed: lisinopril stopped immediately, aggressive isotonic fluid resuscitation started, his home lithium held, and a structured six-hour reassessment scheduled with explicit, pre-agreed triggers for dialysis — any seizure activity, any further decline in mental status, or no measurable neurologic improvement — rather than leaving the decision to whoever happens to be at the bedside when his condition changes.
Not agreed: whether the six-hour window itself is the right length. The nephrologist would have preferred starting dialysis now rather than accepting any trial period at all, and said so plainly before agreeing to the compromise; the psychiatrist thought even six hours might be too aggressive a trigger given how quickly reversible precipitants like his often resolve with fluids alone. Both deferred to the pharmacologist's proposed interval as a reasonable middle position rather than either fully agreeing it was correct — the actual number was accepted more than it was independently endorsed.