Electrolyte Replacement
Replacing electrolytes is easy. Working out why they are low, and whether replacing them is safe, is the part that separates a doctor from a protocol.
What to check before you replace anything Read this first
This is the section that matters most and the one that gets skipped. Four questions, before you write for potassium at two in the morning.
The kidney is what protects the patient from your replacement. A creatinine of 3.4, a urine output of 15 mL an hour or a dialysis schedule changes every dose here, and potassium and magnesium go from routine to dangerous. Look at the urine output, not just the creatinine.
Loop and thiazide diuretics waste potassium and magnesium. Proton pump inhibitors cause chronic hypomagnesemia. Amphotericin, aminoglycosides and cisplatin waste both. Insulin and beta agonists drive potassium into cells. If a drug you can stop is the cause, replacement alone is a treadmill.
A potassium of 2.9 in diabetic ketoacidosis sits on a massive total body deficit and will fall further with insulin. The same 2.9 after a beta agonist nebulizer or in an acute alkalosis is a shift and comes back on its own. Same number, very different orders.
Hemolysis falsely raises potassium and phosphate. A difficult draw, a small needle, a long tourniquet or a clenched fist will do it. A potassium of 6.9 with a normal EKG and normal renal function is a redraw, not a code. In the other direction, blood drawn above a running IV line is diluted and reports everything low.
Potassium
Serum potassium is a poor proxy for total body potassium, since 98 percent of it is intracellular. The teaching rule is that every 10 mEq of potassium chloride raises the serum by about 0.1 mEq/L with normal renal function. Getting from 3.0 to 4.0 takes on the order of 100 mEq, while the whole body deficit at 3.0 is commonly 200 to 400 mEq and far more at 2.5. That is why the level drifts back down over two days, and why replacing 40 mEq and declaring victory does not work.
Oral versus intravenous
If the gut works and the patient is not vomiting, oral is better. It is safer, cheaper, raises the level at least as fast as an IV rider, and does not hurt. Typically 40 mEq of potassium chloride by mouth, repeated and spaced out, because larger doses cause nausea. The liquid tastes bad enough that patients hide it, so ask whether it was swallowed. Reserve IV for the patient who cannot take oral, has an ileus, or is symptomatic now.
Rate limits, and why they exist
Peripheral line: up to about 10 mEq per hour, conventionally 10 mEq in 100 mL. Concentrated potassium is caustic to veins and it hurts. Patients turn off their own pumps because of it.
Central line: up to about 20 mEq per hour with continuous cardiac monitoring. Here the limit is cardiac, not venous. A rapid rise in the potassium arriving at the right atrium can cause arrhythmia, which is what the monitor is for.
Faster protocols exist in intensive care. They belong to that setting and to your local policy, not to a floor patient at 3 am.
Never let concentrated potassium hang without a pump.
The magnesium rule
If a potassium will not come up despite adequate replacement, or a calcium will not come up at all, check a magnesium. It is the highest yield habit on this page.
The mechanisms explain what you are watching. Renal potassium wasting: intracellular magnesium normally blocks the ROMK channel in the distal nephron from the inside. Deplete it and the block is released, the channel secretes freely, and the kidney dumps whatever potassium you give. You can pour it in all night and watch it appear in the urine.
Impaired parathyroid hormone: magnesium is required both for parathyroid hormone secretion and for the end organ response to it. Low magnesium produces a functional hypoparathyroidism, and the hypocalcemia that follows is refractory to calcium until the magnesium is fixed.
So in any patient with hypokalemia or hypocalcemia, send a magnesium on the same draw and replace it in parallel rather than waiting to see whether the potassium moves.
Replacing magnesium
Typical replacement is 2 g of magnesium sulfate intravenously over 1 to 2 hours for a mild deficit and 4 g for a larger one, usually as repeated 2 g doses. Two things surprise people:
- Much of an IV dose is excreted. The acute rise in serum magnesium suppresses tubular reabsorption, so a large share of a rapid infusion is in the urine within hours. That is why the level looks better at hour two and worse the next morning, and why real repletion takes days.
- Slower is better. Infusing over 2 to 4 hours rather than 30 minutes improves retention. The exception is an emergency such as torsades de pointes or severe asthma, where you push it and accept the loss.
In renal impairment magnesium accumulates, and the consequences are respiratory depression, hypotension and cardiac arrest. Reduce the dose substantially, infuse over longer, recheck earlier, and follow deep tendon reflexes, which go before anything worse happens.
Phosphate
Phosphate is ignored until it causes a problem, and then the problem is serious: diaphragmatic weakness and failure to wean, rhabdomyolysis, confusion and seizures. It matters in four settings.
- Refeeding syndrome. Anyone who has not eaten for a week. Feeding drives insulin, insulin drives phosphate into cells, and the level collapses within 72 hours. Anticipate it, check daily, feed slowly.
- Recovery from diabetic ketoacidosis, where insulin drives phosphate intracellularly exactly as it does potassium.
- Alcohol use disorder, with poor intake and renal wasting at once.
- Continuous renal replacement therapy, which clears it steadily.
Oral replacement is preferred when tolerated, using a sodium or potassium phosphate preparation, and the main side effect is diarrhea. Reserve IV for the severely low or symptomatic patient, typically 15 to 30 mmol over 4 to 6 hours.
Do not run intravenous phosphate through the same line as calcium gluconate, calcium chloride or a calcium containing fluid such as Lactated Ringer's. The precipitate is real and it embolizes. Flush the line, use another lumen, or space the doses.
And potassium phosphate carries a potassium load. Write 30 mmol of it and you have also given roughly 44 mEq of potassium. Count it.
Calcium
First make sure the number is real. About 40 percent of serum calcium is bound to albumin and is not physiologically active, so a low albumin drops the total without dropping the ionized fraction. Correct by adding 0.8 mg/dL for every 1 g/dL the albumin sits below 4. The corrected calcium calculator does it and explains where it fails.
In critical illness, massive transfusion, pancreatitis or any case where the answer matters, send an ionized calcium instead, because the formula performs poorly in exactly those patients. Alkalosis increases albumin binding and lowers ionized calcium with an unchanged total, which is why the hyperventilating patient gets perioral tingling with a normal calcium.
Gluconate versus chloride
One gram contains roughly 93 mg of elemental calcium. Well tolerated peripherally, and the default on the floor. Typically 1 to 2 g intravenously over 10 to 20 minutes, faster in tetany, seizure or arrhythmia.
One gram contains roughly 273 mg of elemental calcium, three times as much, which is why it is the code cart drug. It is also a vesicant that causes tissue necrosis on extravasation, so it needs central access outside an arrest. Reserve it for cardiac arrest, severe hyperkalemia with EKG changes, and profound hypocalcemia in someone with a line.
And before you chase a low calcium a third time, check the magnesium.
Sodium, in both directions
Sodium is a water problem, not a salt problem, and the one electrolyte where the speed of correction matters more than the destination.
Hyponatremia. Raise the sodium too fast in someone who has been low for days and you get osmotic demyelination. Aim for a rise of about 4 to 6 mEq/L in 24 hours and treat 8 as a ceiling, lower in the malnourished, the alcoholic and those with liver disease. Overcorrection can be reversed with free water and desmopressin, so recheck every few hours rather than once a day.
Hypernatremia. Lower it too fast and the brain, which has generated protective intracellular osmoles, takes on water. That is cerebral edema. Do not drop it faster than about 10 to 12 mEq/L in 24 hours.
Symptomatic hyponatremia with seizures or obtundation is the exception, treated urgently with small boluses of hypertonic saline aiming for a rise of 4 to 6 mEq/L to stop the seizure, then stopping. The 24 hour limit still applies.
For the arithmetic, use the free water deficit calculator, and the corrected sodium calculator to check whether a hyperglycemic patient's sodium is real before treating it. A glucose of 900 makes a sodium of 145 look normal when the corrected value is nearer 158.
Replacement table
| Electrolyte | Threshold to treat | Typical replacement | Recheck |
|---|---|---|---|
| Potassium | Below 3.5, or below 4.0 in cardiac disease or on digoxin | 40 mEq PO, repeated. IV 10 mEq/hour peripheral, 20 central with monitoring | 1 to 2 hours after the rider, or with morning labs |
| Magnesium | Below 1.8, or refractory hypokalemia or hypocalcemia | 2 g magnesium sulfate IV over 1 to 2 hours, 4 g if severe | 6 to 12 hours, and daily while replacing |
| Phosphate | Below 2.0, or weakness or ventilator dependence | Oral if tolerated. IV 15 to 30 mmol over 4 to 6 hours if severe | 6 to 12 hours after IV, daily in refeeding |
| Calcium | Symptoms, or ionized calcium clearly low | Gluconate 1 to 2 g IV over 10 to 20 minutes. Chloride only centrally | 1 to 4 hours, ionized where it matters |
| Sodium, low | Symptoms at any level, or a falling trend | Treat the cause. Hypertonic saline only for severe symptoms | Every 2 to 4 hours during active correction |
| Sodium, high | Above 145 with a deficit, or any rising trend | Enteral water where possible, otherwise D5W, no faster than 10 to 12 mEq/L per day | Every 4 to 6 hours during correction |
Thresholds are conventions, not laws. A potassium of 3.4 in a healthy 30 year old is a glass of orange juice. The same number on digoxin two days after a myocardial infarction gets replaced now.
Order the follow up lab at the same time you order the replacement. The commonest failure here is not the wrong dose, it is a correctly replaced electrolyte that nobody rechecked.
For the overnight versions of these problems, see common night calls.
This is not medical advice. It is a teaching outline for clinicians and clinicians in training. Every dose and threshold below is illustrative, there to teach the shape of the decision. Follow your institution's protocols, verify every drug and dose against a current reference, and use your own judgment.