Launch an inpatient hyponatremia playbook: classification, safe correction rules,…
- User
Launch an inpatient hyponatremia playbook: classification, safe correction rules, monitoring ladders, and relapse prevention.
- Assistant
Inpatient Hyponatremia Playbook
Sherlock AI Protocol | Trailblazer Labs
Rapid reference for inpatient management. Confirm diagnosis (Na⁺ <135 mEq/L). Assess symptoms (none/mild: headache, N/V; severe: seizures, coma). Always evaluate acuity (acute <48h vs. chronic >48h) and volume status first. Consult nephrology/endocrinology early.1. Classification
Classify by severity, acuity, and volume status to guide therapy.
Severity Na⁺ (mEq/L) Symptoms Mild 130–134 Asymptomatic or mild (fatigue, N/V) Moderate 125–129 Moderate (headache, confusion) Severe <125 Severe (seizures, coma, resp arrest) Volume Status (clinical + labs: urine Na/Osms, serum Osms):
- Hypovolemic (10–15% cases): Dehydration (GI losses, diuretics, hemorrhage). Urine Na⁺ <20, FeNa <1%.
- Euvolemic (75%): SIADH, hypothyroidism, adrenal insuff. Urine Osms >100, urine Na⁺ >40.
- Hypervolemic (15%): CHF, cirrhosis, nephrotic. Edema, low effective volume.
Etiology Ladder (common inpatient):
- SIADH (drugs: SSRI, carbamazepine; pulmonary/CNS disease)
- Hypervolemic (heart/liver/kidney failure)
- Hypovolemic (GI/renal losses)
- Pseudohyponatremia (hyperlipidemia/hyperglycemia)
- Other: Beer potomania, exercise, MDMA
2. Safe Correction Rules
Core Principle: Prevent osmotic demyelination (ODS). Limit ΔNa⁺:
- Acute symptomatic: 4–6 mEq/L in 4–6h.
- Chronic/asymptomatic: <8 mEq/L in 24h; <18 mEq/L in 48h.
- High-risk ODS (malnourished, alcoholics, hypokalemia): <6 mEq/L/24h.
- Stop/reverse if overcorrection (>10 mEq/L/24h): D5W 6 mL/kg bolus + desmopressin 2–4 mcg IV.
Therapy Ladder (titrate based on class/symptoms):
Scenario Initial Therapy Dose/Rate Target ΔNa⁺ Severe symptoms (seizure/coma) All types 3% saline IV bolus 100–150 mL over 10–20 min (2–3 mL/kg); repeat ×2 if needed +4–6 mEq/L in 1–4h Moderate symptoms 3% saline infusion 1–2 mL/kg/h ×4–6h +4–6 mEq/L in 6h Mild/asymptomatic hypovolemic Isotonic saline (0.9% NS) 1–2 L bolus, then maintenance Match losses Mild/asymptomatic euvolemic/hypervolemic Fluid restriction 800–1000 mL/day <0.5 mEq/L/h Refractory SIADH Tolvaptan (Vasostrict) or urea 15 mg PO BID; taper <8/24h Hyperacute (<24h) Hypertonic saline + loop diuretic 3% saline + furosemide 20–40 mg IV +6–8 mEq/L/24h OK Formulas:
- Infusion rate (3% saline): (Target ΔNa × TBW × 0.6) / 1 (for 513 mEq/L saline). TBW = 0.6 × wt (men), 0.5 × wt (women).
- Adrogue-Madias: Change = [Infusate Na – Serum Na] / (TBW +1).
3. Monitoring Ladders
q1–2h initially → step down based on stability. Neuro checks q4h.
Phase Labs (Na⁺, K⁺, Osms) Vitals/Clinical Duration Acute Phase (symptomatic, 3% saline) q1–2h ×6h, then q4–6h Neuro q1h, I/Os strict, daily wt Until stable Stabilization (ΔNa <6 mEq/L) q6–8h ×24h, then q12h Neuro q4h, I/Os 24–48h Maintenance (asymptomatic) q12–24h Neuro q shift, I/Os permissive Until discharge High-Risk (chronic + risk factors) q4h ×48h Daily neuro exam 48h min - Red Flags: ΔNa >8/24h → hold therapy, D5W bolus.
- Telemetry if arrhythmia risk (hypoK common).
4. Relapse Prevention
80% relapse risk without intervention. Discharge planning:
Strategy Indications Details Treat Cause All Stop inciting drugs (thiazides, SSRIs); Rx hypothyroidism/adrenal insuff. Fluid Restriction SIADH/euvolemic 1–1.5 L/day lifelong; educate pt/family. Salt Tab + Loop Diuretic SIADH relapse NaCl 3–6g/day + furosemide 20 mg/d. Vaptans Refractory SIADH Tolvaptan 15–30 mg/d; monitor LFTs, Na q24h. Urea Mild SIADH 15–30g PO BID; cheap, safe alt. Hypervolemic CHF/cirrhosis Optimize GDMT (diuretics, ACEi); salt/fluid restrict. Discharge Criteria:
- Na stable >130 ×24h.
- Asymptomatic.
- Etiology addressed.
- F/U plan: Nephro/endo in 1–2wk; home Na check if high-risk.
Evidence Base: Derived from US/EU expert consensus (2013–2023): Spasovski, Verbalis, Sterns. Tailor to pt; this is not personalized medical advice—verify with EMR/chart.
Playbook launched. Query for case-specific tweaks?
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.