How High, How Fast: Medical Management or Emergent Dialysis for Severe Hyperkalemia
A single patient with severe hyperkalemia, ECG changes, and new oliguric AKI on a background of chronic ACE-inhibitor therapy. The disagreement is whether medical management has genuinely bought enough time, or whether the ECG has already answered that question.
E.P., a 76-year-old retired librarian, has managed her blood pressure and mild proteinuria for eleven years on lisinopril and, for the last two, low-dose spironolactone added by her cardiologist for a mildly reduced ejection fraction — a combination that has served her well until a bout of norovirus swept through her senior center this week and left her unable to keep fluids down for three days. Her daughter, who calls every evening without fail, knew something was wrong when the calls stopped being answered, and found her confused and weak this morning before bringing her in directly. She arrived profoundly volume-depleted, and the labs that came back within the hour turned what looked like a straightforward dehydration case into something considerably more urgent.
Her potassium is 6.9 mEq/L, and her ECG, drawn as part of the same initial workup, shows peaked T waves with a QRS complex that has begun to widen — a finding that changes this from a lab abnormality to a cardiac emergency in progress, since QRS widening reflects hyperkalemia's effect on cardiac conduction directly, not merely a number on a chemistry panel. Her creatinine is 3.1 mg/dL, up from a stable outpatient baseline of 1.0 confirmed on a recent visit, and she has made no urine since arrival three hours ago. The mechanism here is not mysterious: three days without oral intake on top of a chronic ACE inhibitor and an aldosterone antagonist — two drugs that independently reduce potassium excretion — left her with essentially no reserve once volume depletion caused a real acute kidney injury on top of both. Neither drug is new and neither dose has changed in over a year; this is not a medication error revealing itself, but two chronically appropriate drugs losing their safety margin the moment the kidney clearing them stopped working. The medical team already has insulin-dextrose and calcium gluconate running. The question dividing the room is whether that buys real time, or whether the widening QRS on the monitor has already made the decision for them.
Emergency department, thirty minutes after the first ECG
Calcium gluconate is already running to stabilize the membrane, and insulin/dextrose is shifting potassium intracellularly — both are working, in the sense that they buy time, but neither one removes potassium from her body, and a widening QRS in front of me is not something I am comfortable managing on a watch-and-wait basis. I want nephrology at bedside arranging emergent dialysis access now, not after the next ECG.
I do not disagree that dialysis is very likely where this ends, and I am already placing a temporary line while we speak. My only pushback is on sequencing: calcium has already narrowed her QRS on the repeat strip you just handed me, and insulin-dextrose takes roughly fifteen to thirty minutes to produce a measurable potassium drop, meaning we have not yet actually seen what medical management does for her — we are deciding to bypass it before its own effect window has passed. Sodium zirconium cyclosilicate binds potassium in the gut, and its FDA label reports a measurable reduction in serum potassium beginning at one hour — unlike the older resins, which need the better part of a day. That is meaningfully faster than dialysis access and setup runs in most emergency departments. I will name the limit of that citation myself: the label's onset data come from non-emergent hyperkalemia, not from patients with a QRS already widening, so it tells us when the drug begins working, not that it begins working fast enough for her.
None of this is an argument against dialysis happening — only against treating it as more urgent than the medical measures already in progress that have not yet had time to show their effect.
The repeat strip narrowing is a fair point, and I will grant that the picture is trending the right direction rather than actively worsening in front of us. I still want the dialysis access placed on the same timeline regardless — not because I think medical management has failed, but because if the next value comes back unchanged or worse, I do not want thirty more minutes lost to setting up access we could have placed already.
I am not disputing the pharmacology of SZC's faster onset; I am saying that in a patient who is already anuric with these ECG findings, having dialysis ready the moment it's needed costs nothing and buys real safety margin if the medical measures underperform.
Agreed: medical management continued and dialysis access placed in parallel rather than sequentially, with actual initiation of dialysis contingent on the next potassium value and ECG rather than automatic. Sodium zirconium cyclosilicate was added as a bridging measure.
Both physicians explicitly agreed the sequencing disagreement was really a shared-risk-tolerance question, not a disagreement about the pharmacology itself — and resolved it by doing both in parallel rather than picking one path over the other.