Normal Pressure Hydrocephalus: A Medical Bridge Before Shunt Surgery
A single patient with the classic NPH triad and a six-week wait before she's a safe surgical candidate. The question is whether a carbonic anhydrase inhibitor can actually bridge that gap, or only sounds like it should.
V.S., a 74-year-old woman, moved to a house three blocks from her daughter two years ago specifically so they could see each other most days — a decision that's paid off in an unexpected way, since her daughter is the one who's spent the last four months noticing what V.S. herself kept explaining away: a gait she now describes as “walking like my shoes are nailed to a magnet,” short, shuffling, wide-based steps that make turning around take several extra shuffles rather than a pivot. Her daughter says the bigger change is that she used to run the neighborhood book club from memory and now needs the same reminder note left out twice a week, and there have been two incontinence episodes in the past month that mortify her more than she'll admit outright.
The triad — gait, cognition, incontinence — combined with an MRI showing ventricles disproportionately enlarged relative to her degree of cortical atrophy, is consistent enough with normal pressure hydrocephalus that neurosurgery has scheduled a large-volume lumbar puncture to test whether draining CSF actually improves her gait, the single best predictor available of whether a permanent shunt will help her. Her Evans' index of 0.37 clears the 0.30 threshold for ventriculomegaly, and at 18.4 seconds her 10-metre walk puts gait, not cognition, at the center of her triad. The complication is timing: she has a mechanical mitral valve and has been anticoagulated on warfarin for eleven years, and any shunt placement needs a structured anticoagulation bridge her cardiologist estimates will take five to six weeks to arrange safely. The tap test can happen this week. The shunt, if it's indicated, cannot.
Her daughter has read that acetazolamide is sometimes used for “water on the brain” and is asking whether it can hold things over until surgery is safe. The mechanistic answer complicates that hope: the roughly 50% reduction in CSF flow usually quoted for acetazolamide comes from Vogh and Maren's rabbit work, and appeared only above 99.5% inhibition of choroid-plexus carbonic anhydrase — and NPH is a disorder of absorption, not overproduction. But the argument no longer has to rest on mechanism. The DRAIN trial (Virhammar and colleagues, 2026) randomized idiopathic NPH patients to low-dose acetazolamide or placebo, treating until shunt admission or up to nine months. Gait did not improve: adjusted between-group difference 0.09 units (95% CI −3.61 to 3.79, p=0.96), with 9 of 25 treated patients stopping for side effects against 2 of 25 on placebo. Her own wait is five to six weeks — comfortably inside the window that trial covered, which is why “it's only a few weeks, what's the harm” is the argument the evidence has closed off.
Pre-tap-test planning visit
I understand the appeal of a bridge, but acetazolamide's mechanism doesn't match this disease. It inhibits carbonic anhydrase at the choroid plexus, and the 50-to-60% reduction in CSF flow people quote comes from Vogh and Maren's rabbit work, at inhibition above 99.5% — genuinely useful in idiopathic intracranial hypertension, where overproduction and elevated pressure are the actual problem. NPH is a disorder of impaired absorption and altered venous compliance. Asking a production-reducing drug to fix an absorption problem is mechanistically backwards, and in a 74-year-old on warfarin, acetazolamide's real risks — metabolic acidosis, renal stones, hypokalemia, paresthesias — aren't free just because the drug is familiar. And I'd rather not rest on mechanism at all when we now have the trial: DRAIN answered this directly and negatively last year.
I'm not going to argue NPH is a production problem — it isn't, and DRAIN settles the part I'd otherwise have argued. A year ago I'd have said what geriatricians always say here, that “no proof of mechanism” isn't the same as “won't help,” and that a two-week trial with a baseline and follow-up metabolic panel costs little. That argument is gone. Fifty patients isn't enormous, but the point estimate is 0.09 units with a confidence interval sitting squarely across zero, and better than a third of the treated arm came off the drug — in a woman on warfarin with a mechanical valve, where an electrolyte disturbance or a bout of dehydration is not a minor event.
Where I do push back is on treating this as settled and therefore finished. She has five to six weeks of real decline ahead of her, fixed by cardiology's timeline and not by anything either of us controls, and her daughter came in having done her own reading and asked a specific question. “The trial was negative” is the correct answer and a completely inadequate visit. If we have nothing pharmacologic to offer during the wait, then the fall-prevention and home-safety work stops being an afterthought and becomes the actual plan.
You've converged on the drug question, so let me name what's left. The number that actually determines how long she waits is the anticoagulation bridge timeline, and nobody in this room has called cardiology yet. Let's use this week's already-scheduled tap test as the forcing function: get cardiology looped in today to plan the warfarin bridge in parallel rather than after a positive tap-test result comes back, which is the sequence currently costing her real time. Every week we compress off that five-to-six-week estimate is worth more than anything acetazolamide was ever going to deliver, and unlike the drug it isn't hypothetical — it's a phone call. On the conversation with her daughter, I'd give her the trial by name and the reason it matters that it exists, because “there's no good evidence” and “it was tested properly and didn't work” land very differently on someone who has been reading, and only one of them is true.
Agreed: cardiology contacted today to begin planning the anticoagulation bridge in parallel with, not after, this week's tap test; acetazolamide not started, with DRAIN named explicitly to the daughter rather than deflected as “no good evidence”; and structured fall-prevention and home-safety work started this week as the real interim intervention rather than a consolation offered after the drug was declined.
Not agreed: how hard to push cardiology on the five-to-six-week estimate. The geriatrician wants the bridge timeline treated as a negotiable number and revisited weekly, on the grounds that her decline is measurable and the estimate was given before anyone knew how fast she was moving; the neurosurgeon is unwilling to have a mechanical mitral valve bridged on a compressed schedule to buy gait weeks, and would rather she wait longer than be shunted on a rushed anticoagulation plan. Both agreed that is cardiology's call to make with the facts in front of them, and that the facts should include her current rate of change, not just her diagnosis.