Hyponatraemia, hypernatraemia and the dialysis connection — a water problem solved with arithmetic, not intuition
Based on Koyner, Handbook of Critical Care Nephrology (2021) · NTUH Yunlin Branch
Learning objectives
Koyner Ch 19–20, 48 · Ronco Ch 56, 62, 133 · Spasovski G et al. Eur J Endocrinol 2014;170:G1
The serum sodium tells you about total body water; it tells you almost nothing about total body sodium.
ADH and thirst change total body water, not total body cation. A hyponatraemic patient is a patient with too much water for the cations they have — which is why the treatment is almost never "give salt".
Repleting a potassium of 2.6 will raise the serum sodium exactly as if you had given the same millimoles of sodium. This is one reason hypokalaemia is an independent risk factor for osmotic demyelination — the correction runs away while you are treating something else.
Edelman IS et al. J Clin Invest. 1958;37:1236–1256 · Koyner Ch 19 · Ronco Ch 56 · Potassium repletion strategy: Lecture 06
Rohrscheib M et al. Am J Med Sci. 2015;349:537–544 · Koyner Ch 19 · Ronco Ch 56
Take a patient who cannot feel thirst (sedated, intubated, delirious), give them a continuous non-osmotic ADH stimulus (pain, nausea, ventilation, hypovolaemia), then infuse hypotonic maintenance fluid and dissolve their drugs in dextrose. Hyponatraemia is not a complication of that plan — it is the arithmetic of it.
Koyner Ch 19 · Ronco Ch 56, 62 · Overgaard-Steensen C, Ring T. Crit Care. 2013;17:206 · Fluid tonicity in prescribing: Lecture 03
Na 118 · urine 1.0 L/day · U_Na 110 · U_K 40
EFWC = 1.0 × [1 − 150/118] = −0.27 L/day
The kidney is adding 270 mL of pure water a day. No achievable fluid restriction corrects this patient — you must change the solute load or block the V2 receptor.
Na 128 · urine 0.9 L/day · U_Na 12 · U_K 28
EFWC = 0.9 × [1 − 40/128] = +0.62 L/day
Concentrated urine, but electrolyte-poor: it removes 620 mL of free water a day. Restrict intake below that and the sodium rises.
Koyner Ch 19 (Eq 19.2, Fig 19.1) · Ronco Ch 56 (Eq 3) · Rose BD. Am J Med. 1986;81:1033
Three questions in order — is the sodium real, is the serum hypotonic, and is ADH on — before any question about volume status.
Under-treating acute hyponatraemia kills by cerebral oedema within hours. Over-treating chronic hyponatraemia disables by osmotic demyelination within days. There is no other electrolyte where both errors are catastrophic and in opposite directions.
Spasovski G et al. Eur J Endocrinol. 2014;170:G1 · Liamis G et al. Am J Med. 2013;126:256 · Renneboog B et al. Am J Med. 2006;119:71.e1 · Hoorn EJ et al. J Bone Miner Res. 2011;26:1822 · Koyner Ch 19 · Ronco Ch 62
Indirect ISE (main-lab autoanalyser) dilutes the sample and reports sodium per litre of plasma. When the solid phase expands — triglycerides > 1000 mg/dL, IgG paraproteinaemia, IV immunoglobulin — the reported sodium falls although plasma-water sodium is normal. Direct ISE (blood-gas analyser) measures activity in plasma water and is immune.
Free bedside test: send the same blood to the gas analyser. Lab Na 121, gas Na 139 — you have your answer in five minutes.
Glucose, mannitol or glycine draws water out of cells and dilutes the sodium. The patient is hypertonic and must never be given hypertonic saline for the number. Mannitol cannot be corrected for, because serum mannitol is not routinely measurable — read the osmolal gap instead.
Koyner Ch 19 · Ronco Ch 56 · Rohrscheib M et al. Am J Med Sci. 2015;349:537
Worked: measured Na 122, glucose 900 mg/dL (50 mmol/L).
The two factors differ by 6 mEq/L here — enough to change a diagnosis but not a treatment. Use 2.4 at high glucose, but treat the number as a prediction of where the sodium will land once insulin works, not as a target. A patient who corrects to 141 does not need saline; a patient who corrects to 128 has genuine hyponatraemia waiting underneath the diabetic emergency.
Katz MA. N Engl J Med. 1973;289:843 · Hillier TA et al. Am J Med. 1999;106:399 · Koyner Ch 19 (Eq 19.4) · Ronco Ch 56 · DKA management: Lecture 07
Serum osmolalityLow = true hypotonic hyponatraemia. Normal or high sends you back to pseudo- or translocational causes.
Urine osmolality<100 = ADH suppressed → primary polydipsia, low-solute intake, advanced kidney failure. >100 = ADH is driving it.
Volume statusHypovolaemic, euvolaemic or hypervolaemic — history, weight trend, fluid chart, POCUS, not the skin turgor.
Urine sodium + uric acidU_Na <20–30 with high urate = hypo- or hypervolaemic. U_Na >40 with low urate = SIADH.
Diuretics raise urine sodium in a genuinely hypovolaemic patient and abolish the U_Na discriminator. CKD fixes urine osmolality near isotonic and floors the concentrating and diluting range, so neither extreme is interpretable. Recent fluid administration — saline given in the emergency department before the urine was sent — rewrites both numbers. And the volume examination itself is unreliable: in the classic Chung series, experienced clinicians correctly classified hypovolaemic hyponatraemia in only about half of cases.
Koyner Ch 19 (Fig 19.3) · Chung HM et al. Am J Med. 1987;83:905 · Spasovski G et al. Eur J Endocrinol. 2014;170:G1
Random cortisol above 700 nmol/L (25 µg/dL) in a critically ill patient effectively excludes adrenal deficiency; below that, do a short Synacthen test rather than assume.
CNS: subarachnoid and intracranial haemorrhage, traumatic brain injury, meningitis, encephalitis, stroke, hydrocephalus
Pulmonary: bacterial and viral pneumonia, empyema, tuberculosis, and mechanical ventilation itself
Malignant: small-cell lung carcinoma above all; oropharyngeal, GI, urogenital, thymoma, lymphoma
Situational: pain, nausea, the perioperative state, delirium tremens
Split SIADH into self-limiting (pain, nausea, pneumonia, post-operative) and persistent (paraneoplastic, reset osmostat). Self-limiting SIADH is the classic overcorrection trap: the moment the stimulus stops the patient aquareses and the sodium climbs on its own.
Ronco Ch 56 (Tables 56.1–56.2) · Koyner Ch 19 (Table 19.4) · Spasovski G et al. Eur J Endocrinol. 2014;170:G1
| Class | Agents | Mechanism / note |
|---|---|---|
| Thiazides & thiazide-like | Hydrochlorothiazide, indapamide, metolazone | Desalination at the distal tubule with preserved medullary gradient; recurrence on rechallenge is near-certain — never re-expose |
| Antidepressants | SSRIs, venlafaxine, duloxetine, mirtazapine, tricyclics | SIADH, usually in the first 2–4 weeks; elderly women at highest risk |
| Anti-seizure drugs | Carbamazepine, oxcarbazepine, valproate | Direct V2 sensitisation; oxcarbazepine is the more potent of the pair |
| Antipsychotics | Haloperidol, phenothiazines, risperidone, clozapine | SIADH plus, in chronic psychiatric patients, coexisting primary polydipsia |
| Antineoplastics | Cyclophosphamide (high dose), ifosfamide, vincristine, cisplatin | Cyclophosphamide protocols mandate a water load — a documented cause of fatal water intoxication |
| V2 agonists | Desmopressin, vasopressin, terlipressin, oxytocin | Pharmacological SIADH — the commonest iatrogenic cause on a critical care unit |
| Miscellaneous | NSAIDs, PPIs, trimethoprim, voriconazole, MDMA, nicotine | MDMA combines SIADH with compulsive water drinking — a lethal pairing in young patients |
Liamis G et al. Am J Kidney Dis. 2008;52:144 · Ronco Ch 56 (Table 56.1) · Friedman E et al. Ann Intern Med. 1989;110:24 (thiazide rechallenge) · Koyner Ch 19
The overwhelming majority of hyponatraemia after subarachnoid haemorrhage, trauma or neurosurgery is SIADH, not CSW. The error is expensive in both directions: fluid-restricting a genuinely hypovolaemic CSW patient risks delayed cerebral ischaemia after SAH, while volume-loading an SIADH patient with saline can drive the sodium down by desalination. When you cannot decide, treat both safely — give hypertonic saline, not isotonic, and do not restrict fluid in a patient at risk of vasospasm. Fludrocortisone 0.2–0.4 mg/day is the adjunct in confirmed CSW.
Koyner Ch 19 · Ronco Ch 56 · Sterns RH, Silver SM. J Am Soc Nephrol. 2008;19:194
GI losses, diuretics, third-spacing, adrenal insufficiency, burns. Baroreceptor ADH plus hypotonic replacement. U_Na usually <20–30 — unless a diuretic or vomiting-induced bicarbonaturia is in play. Treatment is isotonic crystalloid; expect a brisk rise once volume is restored.
Heart failure, cirrhosis, nephrotic syndrome, advanced CKD. Low effective arterial blood volume → ADH and RAAS on together → concentrated, sodium-poor urine. Hypertonic saline worsens congestion and, in cirrhosis, ascites. Treatment is decongestion plus fluid restriction, and tolerating a sodium of 128–132 is usually correct.
Beer potomania, tea-and-toast, anorexia. Maximum water excretion = daily solute load ÷ minimum urine osmolality. At 700 mOsm/day and U_Osm 50 that is 14 L; at 100 mOsm/day it is 2 L. Primary polydipsia instead exceeds the ~18 L ceiling from the intake side. Both autocorrect explosively.
In hypovolaemia, isotonic saline raises both the serum sodium and the urine sodium. In SIADH it raises the urine sodium while the serum sodium falls: the infused sodium is excreted in a small volume of concentrated urine and the water is retained. That failure to respond is the practical bedside test. If the starting sodium is already below 120, run the challenge with 3% rather than 0.9% saline so a wrong guess does not cost 3 more mEq/L.
Koyner Ch 19 · Ronco Ch 56 · Spasovski G et al. Eur J Endocrinol. 2014;170:G1
Symptoms set the urgency, risk factors set the ceiling, and a sodium measured every two hours beats every formula ever published.
| Severity | Features | First move |
|---|---|---|
| Severe | Seizure, coma (GCS ≤ 8), obtundation, vomiting, cardiorespiratory distress | 3% saline bolus now, before any diagnostic workup is complete |
| Moderately severe | Nausea without vomiting, confusion, headache, abnormal somnolence | One 150 mL bolus of 3%, then cause-specific therapy with a hard 24-hour ceiling |
| Absent | No attributable neurological signs | No hypertonic saline. Find the cause, remove it, and defend the ceiling |
You cannot document absent confusion in a patient on propofol. Two rules follow. First, adjudicate causality: if the hyponatraemia is mild and the symptoms severe, look for another diagnosis; if the symptoms persist after the sodium has risen 5 mEq/L, they were never the sodium. Second, there is no threshold sodium below which symptoms appear — it is the rate and relative size of the fall that matters. A drop from 144 to 122 causes as much cerebral oedema as a drop from 122 to 104, and in a patient with pre-existing raised intracranial pressure a fall of only 4–6 mEq/L can be critical.
Spasovski G et al. Eur J Endocrinol. 2014;170:G1 (Table 1) · Ronco Ch 56 (Box 56.1) · Koyner Ch 19 (Table 19.1)
Bolus100–150 mL of 3% NaCl (or 2 mL/kg) IV or intraosseously over 10–20 min. Peripheral line is acceptable for a bolus.
ReassessSymptoms at 5–20 min; sodium at the end of the bolus. Ask only one question: better or not?
Repeat ≤ 3×Stop when the sodium has risen 4–6 mEq/L or symptoms resolve — whichever comes first.
Stop and thinkSymptoms persisting after +10 mEq/L are not the sodium. Look for another cause.
A rise of about 5 mmol/L reduces intracranial pressure by roughly 50% within an hour and reliably aborts seizures. Brain volume responds to the first few millimoles; everything beyond that buys no neuroprotection and buys demyelination risk. This is also why 0.9% saline has no role in the emergency: in SIADH it can lower the sodium further, and its effect is unpredictable in every phenotype. Vaptans have no role in the acute phase either — they cannot be titrated over minutes.
Spasovski G et al. Eur J Endocrinol. 2014;170:G1 · Koenig MA et al. Neurology. 2008;70:1023 · Ronco Ch 56 (Fig 56.3) · Koyner Ch 19
58 kg woman, post-operative, serum Na 110, generalised seizure.
In two independent retrospective series the equation substantially underestimated the achieved rise, because it cannot model the water diuresis that starts when the ADH stimulus is removed. Mohmand found inadvertent overcorrection in a large minority of patients treated with hypertonic saline by formula.
Use the formula to choose a starting dose, then abandon it. Sodium every 2 hours until the trend is predictable, then every 4–6 hours. Read the trend, not the last value.
Adrogué HJ, Madias NE. N Engl J Med. 2000;342:1581 · Mohmand HK et al. Clin J Am Soc Nephrol. 2007;2:1110 · Hanna RM et al. Clin Kidney J. 2016;9:530 · Koyner Ch 19 (Eq 19.3)
| Patient | Target rise / 24 h | Absolute ceiling | Trigger to relower |
|---|---|---|---|
| Standard risk, chronic or unknown duration | 4–8 mEq/L | 10 mEq/L in the first 24 h, then 8 mEq/L per subsequent 24 h | >8 in 6 h, or >10 in 24 h |
| High ODS risk — Na <105, K <3.0, alcohol use disorder, malnutrition, advanced liver disease or liver transplant | 4–6 mEq/L | 8 mEq/L in any 24 h | >6 in any 24 h |
| Documented acute (<48 h) hyponatraemia with severe symptoms | Enough to abort symptoms, usually 4–6 | Still 10 in 24 h — acute onset is not a licence | >10 in 24 h |
Write the ceiling as a number in the notes — "do not exceed Na 118 by 08:00" — because the night team will not reconstruct your reasoning. And remember that osmotic demyelination has been reported at guideline-compliant correction rates: the ceiling reduces risk, it does not abolish it.
Spasovski G et al. Eur J Endocrinol. 2014;170:G1 · Verbalis JG et al. Am J Med. 2013;126(10 Suppl 1):S1 · Ronco Ch 56 · Koyner Ch 19 (Table 19.2)
The commonest cause of overcorrection is not too much hypertonic saline. It is the water diuresis that begins when the ADH stimulus disappears: volume repleted, nausea settled, thiazide stopped, cortisol replaced, solute given to a beer-potomania patient. Urine output jumps to 300–500 mL/h of dilute urine and the sodium climbs 2 mEq/L per hour with no infusion running at all.
Koyner Ch 19 · Ronco Ch 56 · Sterns RH et al. Semin Nephrol. 2009;29:282
70 kg man, Na 108 at midnight, 121 by 14:00 — a 13 mEq/L rise in 14 h against a ceiling of 8.
Re-induction of hyponatraemia after overcorrection reduces mortality in the rat model and has been done successfully in human case series. Desmopressin reliably prevents and reverses inadvertent overcorrection. There is no randomised evidence and there never will be — the effect size in the animal work and the irreversibility of the alternative settle the argument.
Dexamethasone has been suggested as an adjunct; the human and animal data are thin and it is not standard care.
Perianayagam A et al. Clin J Am Soc Nephrol. 2008;3:331 · Gankam Kengne F et al. Kidney Int. 2009;76:614 · Soupart A et al. Clin Nephrol. 1999;51:383 · Koyner Ch 19 · Ronco Ch 56
Stop all maintenance fluid — you are about to control water balance deliberately; you cannot do that with a background infusion running.
Desmopressin 2 µg IV every 8 hours — deliberately fix the urine osmolality high so the kidney can no longer generate a surprise water diuresis.
3% NaCl infusion 1–1.5 mL/kg/h over 6 h — now the only input of tonicity is the one you are writing, and the rise becomes linear and predictable.
Fluid restrict to 1.2 L/day and check sodium every 2 h; once the slope is stable, every 4–6 h.
Titrate the 3% rate to <8 mEq/L/day, adjusting no more often than every 6 h, and reading the trend rather than the latest value.
Continue until Na 125–130, then stop both together. Not for volume-overloaded patients, and not as first-line in severe symptomatic hyponatraemia — that patient gets boluses first.
Koyner Ch 19 (Table 19.3) · Sood L et al. Am J Kidney Dis. 2013;61:571 · Perianayagam A et al. Clin J Am Soc Nephrol. 2008;3:331
Osmotic demyelination is also described after a rapid rise in tonicity from any cause — rapidly developing hypernatraemia, salt poisoning, and abrupt correction of hyperglycaemia. Any large, fast change in effective osmolality is the hazard; hyponatraemia is simply where we generate it most often.
It has also been reported after rapid correction of an acute (<48 h) hyponatraemia, so "it was acute" is not a defence for going fast.
King JD, Rosner MH. Am J Med Sci. 2010;339:561 · Shah MK et al. Am J Kidney Dis. 2018;71:436 · Ronco Ch 56 (Fig 56.2) · Koyner Ch 19
Treat the cause and stop the drug — pneumonia, pain, nausea, the SSRI, the oxcarbazepine. In self-limiting SIADH this is the whole treatment, and the risk shifts to overcorrection.
Fluid restriction below urine output, typically 800–1000 mL/day. It fails whenever (U_Na + U_K)/serum Na > 1 — check that ratio before committing three uncomfortable days to it.
Salt tablets plus a loop diuretic — the loop blunts the medullary gradient so the solute you give is excreted in more water. Watch potassium and magnesium.
Urea 15–30 g/day. MW 60, so 15 g = 250 mOsm of freely filtered, rapidly cleared solute. At a fixed urine osmolality of 500 that is an extra 500 mL of urine per day. The same osmolar load from salt would take seven 1 g NaCl tablets (17 mmol NaCl per gram → 34 mOsm). Retrospective inpatient data show it safe and effective in SIADH, heart failure and cirrhosis; palatability is the only real obstacle.
Tolvaptan — reserved, monitored, and never in the acute symptomatic patient.
Oral tolvaptan vs placebo in euvolaemic and hypervolaemic hyponatraemia: significantly higher serum sodium at day 4 and day 30, with the benefit lost after withdrawal. It corrects the number; no trial has shown it improves survival. Cautions: overly rapid correction (start in hospital with sodium checks), FDA restriction to ≤30 days for dose-dependent hepatotoxicity, no role in hypovolaemic hyponatraemia, cost, and in cirrhosis the added hazards of conivaptan raising portal flow through V1a blockade.
Schrier RW et al. N Engl J Med. 2006;355:2099 · Rondon-Berrios H et al. Clin J Am Soc Nephrol. 2018;13:1627 · Sterns RH et al. Kidney Int. 2015;87:268 · Konstam MA et al. JAMA. 2007;297:1319 (EVEREST) · Ronco Ch 62
63 F, alcohol use disorder, admitted with vomiting. Presenting Na 104, K 2.8, drowsy but rousable, no seizure. U_Osm 90, U_Na 12. She received 2 L of 0.9% saline and 40 mmol KCl in the emergency department. Four hours later: Na 113, urine output 400 mL/h and dilute.
A. Continue saline — she is correcting nicely · B. Give 3% saline; 113 is still dangerously low · C. Stop the saline, give D5W and desmopressin, and bring the sodium back down · D. Start tolvaptan for the persistent hyponatraemia
Hands up — then name her ODS risk factors out loud
Always a failure to drink — and in a sedated, intubated patient that failure is ours, which is why it reads as a marker of care quality.
Hyponatraemia usually resolves during treatment and overshoots; hypernatraemia almost never resolves spontaneously, so under-treatment is far commoner than over-treatment. Undercorrection is itself associated with excess mortality. The clinical failure mode is a patient who sits at 158 for four days because everyone is being careful.
Palevsky PM et al. Ann Intern Med. 1996;124:197 · Lindner G et al. Am J Kidney Dis. 2007;50:952 · Darmon M et al. Nephrol Dial Transplant. 2010;25:2510 · Polderman KH et al. Crit Care Med. 1999;27:1105 · Bataille S et al. BMC Nephrol. 2014;15:37 · Koyner Ch 20
Fever (insensible loss rises with every degree), high ambient temperature, burns, large open abdominal wounds, diarrhoea including laxative and lactulose use, vomiting, sweating, and ventilation with inadequately humidified gas.
Loop diuretics, osmotic diuresis (hyperglycaemia, mannitol, high-protein feed generating urea diuresis, SGLT2 inhibitors), central and nephrogenic diabetes insipidus, and the polyuric phase of recovering ATN or post-obstructive diuresis.
A 50 mL ampoule of 8.4% sodium bicarbonate is 1000 mmol/L. Ticarcillin carries 5 mmol of sodium per gram — nearly 70 mmol/day at 3.375 g q6h. Ciprofloxacin carries 78 mmol per gram. Add 0.9% saline flushes, hypertonic saline for intracranial pressure, and drug diluents.
Recommended daily sodium intake is about 100 mmol — the content of 0.7 L of 0.9% saline. Critically ill patients routinely receive four to ten times that from resuscitation and drug carriers alone. Weigh the patient: weight loss with hypernatraemia means dehydration and needs water; weight gain with hypernatraemia means sodium excess and needs a negative cation balance, with thiazides or, rarely, dialysis. The two treatments are opposite.
Koyner Ch 20 · Ronco Ch 56 (Fig 56.4) · Overgaard-Steensen C, Ring T. Crit Care. 2013;17:206
68 y woman, 60 kg, ICU day 9, febrile, on nasogastric feed. Serum Na 162, glucose normal.
It is a snapshot of the deficit at this instant. It contains no ongoing losses, no continuing sodium input, and no allowance for the fever that caused the problem. Prescribing the deficit alone is the single commonest reason a hypernatraemic patient does not correct.
Koyner Ch 20 (Eq 20.1) · Adrogué HJ, Madias NE. N Engl J Med. 2000;342:1493 · Ronco Ch 56
Same patient, urine 2.2 L/day → ignore 1 L, replace half of the next 1.2 L = 0.6 L.
Urine 2.2 L · U_Na 45 · U_K 25 · serum Na 162
EFWC = 2.2 × [1 − 70/162] = 2.2 × 0.568 = 1.25 L/day
The kidney is losing 1.25 L of pure water a day — twice what the rule of thumb suggested.
Koyner Ch 20 (Eq 20.2) · Ronco Ch 56 (Eq 3) · Overgaard-Steensen C, Ring T. Crit Care. 2013;17:206
A large 2019 critical-care cohort found no excess mortality, seizure or neurological morbidity in adults corrected faster than 0.5 mmol/L/h, while a separate series found undercorrection frequent and associated with death. The paediatric seizure data that generated the rule do not transfer cleanly to adults. Interpret this as permission to be decisive with a genuinely dehydrated adult — not as licence to abolish monitoring. And watch the glucose: hypernatraemia impairs insulin sensitivity, so an aggressive D5W prescription can generate hyperglycaemia, then osmotic diuresis, then more hypernatraemia.
Chauhan K et al. Clin J Am Soc Nephrol. 2019;14:656 · Sterns RH. Clin J Am Soc Nephrol. 2019;14:645 · Bataille S et al. BMC Nephrol. 2014;15:37 · Koyner Ch 20 · Ronco Ch 56
| Condition | Baseline urine osmolality | Response to desmopressin | ICU causes |
|---|---|---|---|
| Normal ADH response | Up to 1200 mOsm/kg (lower in CKD) | No further rise — already at maximal ADH effect | — |
| Complete central DI | <290 mOsm/kg | Rises by >200 mOsm/kg, usually above 500 | Neurosurgery (transsphenoidal, skull base), TBI, aneurysmal SAH, hypoxic-ischaemic injury, brain death, tumour, infiltrative disease |
| Partial central DI | 400–500 mOsm/kg | Rises by ~200 mOsm/kg | Same list, incomplete pituitary injury |
| Nephrogenic DI | Inappropriately low for the tonicity | No change | Lithium (nephrogenic DI in ~55% of long-term users), hypercalcaemia, hypokalaemia, tolvaptan, demeclocycline, foscarnet, amphotericin B, obstruction relief, recovering ATN |
Thiazide plus low-sodium diet (induces mild volume contraction and proximal reabsorption), NSAIDs where the kidney tolerates them, amiloride for lithium specifically, and acetazolamide, which has good reported efficacy in lithium-induced nephrogenic DI. Desmopressin does nothing — the receptor is the problem.
Koyner Ch 20 (Tables 20.1–20.2) · Sands JM, Bichet DG. Ann Intern Med. 2006;144:186 · Garofeanu CG et al. Am J Kidney Dis. 2005;45:626 · Gordon CE et al. N Engl J Med. 2016;375:2008 · Ronco Ch 56 (Table 56.3)
IV: 1–2 µg twice daily — the ICU route
Oral: 100 µg at night, titrated to 200 µg; twice or three times daily if polyuria breaks through
Nasal: 10 µg per spray, 1–4 sprays divided over three doses
Phase 1 — DI with polyuria and rising sodium. Phase 2 — the injured pituitary releases stored ADH: an SIADH picture with falling sodium, typically days 5–10. Phase 3 — permanent central DI. A standing desmopressin order written in phase 1 causes severe hyponatraemia in phase 2. Re-evaluate daily; much post-neurosurgical DI is transient.
Koyner Ch 20 · Ronco Ch 56 (Box 56.2) · Fenske W et al. N Engl J Med. 2018;379:428 · Garrahy A et al. Clin Endocrinol. 2019;90:23
Dialysis is a tonicity intervention whether or not you prescribed it as one — and at the extremes the machine, not the physician, sets the correction rate.
The reverse urea hypothesis: urea is cleared from plasma in minutes but leaves brain cells over 12–24 hours, because uraemia downregulates urea transporters and upregulates aquaporins. Water follows the transient gradient inward roughly twenty times faster than urea can follow it out.
If encephalopathy develops mid-session: stop dialysis immediately, secure the airway, give hypertonic saline or mannitol, and image the head.
Koyner Ch 48 (Tables 48.2–48.3) · Mistry K. Int J Nephrol Renovasc Dis. 2019;12:69 · Patel N et al. Semin Dial. 2008;21:493 · Ronco Ch 133
Blood flow 50–200 mL/min and a small surface-area dialyser — clearance is the hazard, so buy less of it.
Short first session (≈2 h), repeated daily. Two gentle sessions beat one adequate one; nobody has ever been harmed by an under-dosed first treatment.
Raise dialysate sodium to 143–146 mEq/L or use sodium modelling — deliberately maintain plasma tonicity while the urea falls.
Mannitol during the second hour to hold an osmotic gradient in the extracellular space, and consider reducing dialysate bicarbonate.
Cool the dialysate to ~35 °C, protect the blood pressure, and use isolated ultrafiltration for volume so that fluid removal and solute removal are not competing.
Or simply choose CRRT — in a patient with acute brain injury, raised intracranial pressure or extreme uraemia, continuous therapy avoids the osmolar swing rather than mitigating it.
Koyner Ch 48 (Table 48.3) · Mistry K. Int J Nephrol Renovasc Dis. 2019;12:69 · Ronco Ch 133 · Modality choice in detail: Lecture 08
Standard replacement/dialysate fluid is ~140 mEq/L. For a patient at Na 175, running 140 against them drops the sodium far faster than 10 mEq/L/day. Add 3% NaCl (513 mEq/L) to the bag:
To take a 5 L bag from 140 → 165: add 360 mL of 3% NaCl
(700 + 513 × 0.36) / 5.36 = 884 / 5.36 = 165 mEq/L
Cost: everything else is diluted ~7% — bicarbonate 32 → 30 mmol/L. Step the target down by 8–10 mEq/L per day with fresh bags.
For a patient at Na 104, a 140 mEq/L bath will correct them by 20–30 mEq/L in hours. Dilute the bag with sterile water:
To take a 5 L bag from 140 → 118: add 930 mL of sterile water
700 / 5.93 = 118 mEq/L
Cost: a 16% dilution of everything — bicarbonate 32 → 27, potassium 4.0 → 3.4 mmol/L. Replace both separately and check gases hourly.
Machine dialysate sodium is typically adjustable only over about 130–150 mEq/L, so at a serum sodium below ~115 or above ~165 the machine cannot be matched to the patient, and a 4-hour treatment delivers the entire gradient at once. CRRT gives you two things IHD cannot: a bath you can compound to any sodium you like, and a correction spread over 24 hours so an error is caught by the next blood gas rather than by a CT scan. Where local policy forbids compounding bags, the same control is achieved by running standard fluid and titrating a separate D5W or 3% saline infusion into the patient against a sodium checked every 2–4 hours.
Koyner Ch 19–20, 48 · Ronco Ch 56, 133 · Yessayan L et al. Am J Kidney Dis. 2014;64:305 · CRRT prescription: Lectures 08–09
62 F, 58 kg, otherwise well. Post-operative nausea, patient-controlled oxycodone, 2 L/day of 5% dextrose with 0.18% saline since surgery. Witnessed generalised seizure on the ward, now post-ictal in the ICU.
What goes in over the next 20 minutes, and what number do you write as tomorrow morning's ceiling?
Think 60 seconds · answer on the next slide
Treat3% NaCl 150 mL over 20 min. Repeat up to twice more, stopping at +5 mEq/L or when the seizure and obtundation resolve.
Stop the causeDiscontinue the 5% dextrose/0.18% saline, treat the nausea, review the opioid dose. This is a self-limiting SIADH.
AnticipateOnce the pain and nausea settle, she will aquarese. Watch hourly urine output; >150 mL/h of dilute urine is the alarm.
CapSodium every 2 h. Ceiling 6 mEq/L in 24 h given K 3.2 — write the target sodium, not the target rate.
Repleting her potassium from 3.2 to 4.0 delivers cations that raise the serum sodium exactly as sodium would — Edelman does not care which cation you gave. Give the potassium (she needs it, and hypokalaemia is an independent ODS risk factor), but subtract its expected effect from your hypertonic saline plan rather than discovering it on the 06:00 gas.
Spasovski G et al. Eur J Endocrinol. 2014;170:G1 · Koyner Ch 19 · Adrogué HJ, Madias NE. N Engl J Med. 2000;342:1581
74 M, small-cell lung carcinoma, ventilated for a post-obstructive pneumonia. Day 5 of a 1 L/day fluid restriction. Euvolaemic on examination and on POCUS; no diuretic; TSH normal; random cortisol 690 nmol/L.
Do the arithmetic that tells you, in ten seconds, that this restriction was never going to work — then prescribe.
Think 60 seconds · one calculation, then a plan
55 M, 70 kg, severe traumatic brain injury. Sedated, ventilated, ICP monitor in situ reading 12 mmHg. Urine output has climbed to 300 mL/h over the last six hours; fluid balance −3.5 L; weight down 3 kg since admission.
Name the diagnosis, then write today's water prescription in millilitres per hour and state your 24-hour ceiling.
Think 90 seconds · full arithmetic on the next slide
Koyner Ch 20 (Eq 20.1–20.2, Table 20.2) · Ronco Ch 56 (Fig 56.4, Box 56.2) · Garrahy A et al. Clin Endocrinol. 2019;90:23
References & further reading
Questions & discussion — bring me a sodium trend and a urine electrolyte panel, and we can solve almost anything.
Next: Lecture 06 — Potassium, Calcium, Phosphorus & Magnesium