Commit to an answer before revealing the explanation. Use the section menu to navigate; each question and answer has its own link. Five-option evidence updates ask for the most defensible direction of change; discuss a justified adjacent response.
Save and export your first responses
First choices and optional confidence stay in this browser and are not sent to the website. Opening an explanation first is recorded. Revisiting an unchanged case preserves the first response; revised cases receive a new version. Website visit analytics are separate. On a shared device, export and clear the history when finished.
Sections in this case
Question and answer links
- W1: A 70-year-old ICU patient has Na 138 mEq/L, Cl 104 mEq/L, HCO₃ 18 mEq/L and albumin 2.0 g/dL. Your lab's normal anion ga · Answer
- W2: A 55-year-old woman's creatinine rises from 0.8 mg/dL to 1.7 mg/dL over 36 hours. Urine output is 0.8 mL/kg/h. The lab p · Answer
- W3: stop while he is ill? · Answer
- Q4: If you were thinking diabetic ketoacidosis in a man on empagliflozin with three days of vomiting and fast, deep breathin · Answer
- Q5: Using the ABG and BMP above (pH 7.29, pCO₂ 30, HCO₃ 14, Na 134, Cl 86, albumin 4.0; your lab's normal gap is 8–12 and no · Answer
- Q6: Lactate is 1.6 mmol/L and the osmolal gap is 4 mOsm/kg. He takes empagliflozin and metformin. The resident says, "glucos · Answer
- Q7: Creatinine 1.9 mg/dL (baseline 1.0 three months ago), BUN 46, lab eGFR 40, specific gravity 1.028, hyaline casts only, 4 · Answer
- Q8: It is hour 2. Potassium is 3.1 mEq/L, magnesium 1.8, β-hydroxybutyrate 5.8, glucose 188. He is on a cardiac monitor. The · Answer
- Q9: Hour 14: 5 L of 0.9 % saline in, chloride 104 (from 86), bicarbonate 15, anion gap 18, delta-delta 0.67, glucose 164, po · Answer
- Q10: Day 3: Na 138, K 4.1, Cl 110, HCO₃ 17, anion gap 11, creatinine 1.1, β-hydroxybutyrate 0.3. Urine Na 60, K 30, Cl 115 (u · Answer
Session S1 · 2026-10-22 · in-session · 37 min Anchor: Fluids, electrolytes and acid-base — anion-gap, mixed and non-gap acidosis with the RTA table · Integrates: SGLT2 inhibitor euglycemic DKA, metformin and the sick-day rules; prerenal AKI and KDIGO staging from the trend; IV fluid choice; potassium and phosphate shifts on insulin One-line hook: A man on a kidney-protective drug does exactly what the pharmacy label told him — never miss a dose — and arrives breathing like a bellows with a glucose of 206. The ED says "not DKA." Read the BMP in order before you agree.
Author: Andrew Bland, MD, FACP, FAAP — Medical Associates Dept of Nephrology | UICOMP | UDPA | Butler COM Patient: fictional. Any resemblance to a real person is coincidence.
Learning Objectives
Answers for this section: W1 · W2 · W3 · Q4 · Q5 · Q6 · Q7 · Q8 · Q9 · Q10
By the end of this case the student can:
- Read a BMP and blood gas in a fixed order, compute the anion gap (with albumin correction), apply Winter's formula, and compute the delta-delta ratio to name a mixed disorder.
- Recognize SGLT2 inhibitor euglycemic ketoacidosis from the gap and the ketones at a near-normal glucose, order the one test that settles it, and separate a saline- and ketone-loss non-gap acidosis from renal tubular acidosis with the urine anion gap, reciting the RTA table as a differential to recognize, not chase.
- Stage AKI from the creatinine trend using KDIGO 2012, explain why the printed eGFR is not valid while creatinine is moving, and sort this AKI as prerenal from history, exam and urine.
- Sequence potassium, insulin, dextrose and fluid choice in ketoacidosis (when to replace phosphate is discussed, not tested).
- Apply the SADMANS sick-day hold list and recognize medicines requiring an individualized plan or specialist advice; restart criteria and the contrast and surgery rules are discussed, not tested.
Warm-Up (individual commit before the case)
Answers for this section: W1 · W2 · W3
Three readiness items. These check prerequisite knowledge only; no reasoning is graded from them.
W1The gap and the albumin
A 70-year-old ICU patient has Na 138 mEq/L, Cl 104 mEq/L, HCO₃ 18 mEq/L and albumin 2.0 g/dL. Your lab's normal anion gap is 8–12 mEq/L. How do you read the gap?
AG 16 — borderline; this is a non-gap acidosis
AG 16 — the albumin does not change the gap
Corrected AG 21 — a high-gap acidosis
Corrected AG 11 — the low albumin lowers the gap, so it is normal
Show answer and explanation · Link to this question
Correct answer: C
Learning point: AG = Na − (Cl + HCO₃) = 138 − (104 + 18) = 16. Albumin is the main unmeasured anion that normally fills the gap. When albumin is low, the gap you measure is artificially small, so you add it back: 2.5 mEq/L for every 1 g/dL of albumin below 4.0. Here that is 2.5 × 2 = 5, giving a corrected gap of 21 — a real high-gap acidosis that the uncorrected number would have called borderline. Why not the others: A misreads 16 as near-normal and misses the correction; B ignores the albumin rule entirely; D corrects in the wrong direction (a low albumin hides a gap, it does not create one). Course reading: Evidence for this case
W2Stage it
A 55-year-old woman's creatinine rises from 0.8 mg/dL to 1.7 mg/dL over 36 hours. Urine output is 0.8 mL/kg/h. The lab prints eGFR 35. Her KDIGO AKI stage is:
Stage 1 — the rise is at least 0.3 mg/dL within 48 hours
Stage 2 — creatinine is 2.1 times baseline
Stage 3 — creatinine has more than doubled in under 48 hours
Not AKI — an eGFR of 35 is CKD G3b
Show answer and explanation · Link to this question
Correct answer: B
Learning point: KDIGO 2012 stages by whichever criterion gives the higher stage. 1.7 ÷ 0.8 = 2.1 times baseline, which is stage 2 (2.0–2.9×). The ≥0.3 mg/dL rule also fires, but it only reaches stage 1, and the higher stage wins. Urine output is above 0.5 mL/kg/h, so it adds nothing here. Why not the others: A stops at the first criterion that fires; C needs ≥3.0× baseline, a creatinine ≥4.0 mg/dL with an acute rise, or dialysis; D reads an eGFR that is not valid — the equation assumes a steady creatinine, and hers is climbing; a CKD stage needs three months of stability, not 36 hours. Course reading: Evidence for this case
W3The sick-day call
A 68-year-old man on lisinopril, metformin, empagliflozin, metoprolol and over-the-counter naproxen calls the clinic with two days of vomiting and diarrhea and says he can keep down only sips. Which medication should he NOT stop while he is ill?
Lisinopril
Metformin
Naproxen
Metoprolol
Show answer and explanation · Link to this question
Correct answer: D
Learning point: The sick-day hold list (SADMANS — sulfonylureas, ACE inhibitors, diuretics, metformin, ARBs and ARNI, NSAIDs, SGLT2 inhibitors; mineralocorticoid antagonists are held too) covers drugs that lower kidney perfusion, accumulate when GFR falls, or add volume loss during an illness that is already draining him. Beta-blockers are the opposite case: stopping one abruptly causes rebound tachycardia and hypertension. He continues metoprolol unless he is frankly hypotensive or bradycardic. Why not the others: Lisinopril dilates the efferent arteriole and drops GFR when he is volume-depleted; metformin accumulates if GFR falls and raises the lactic acidosis risk; naproxen constricts the afferent arteriole and should be swapped for acetaminophen. Course reading: Evidence for this case
Part 1 — Presentation
Answers for this section: Q4 · Q5 · Q6
ED, Thursday 14:10. Mr. Dale Rourke is a 61-year-old retired diesel mechanic with type 2 diabetes for nine years, hypertension, and a body mass index of 31. His wife brings him in because "he's breathing like he just ran up the stairs, and he hasn't moved off the couch in two days."
Three days ago the whole family came home from a church potluck. By midnight he was vomiting; by morning he had watery diarrhea. His wife and two grandchildren had the same thing for a day and recovered. He did not. He has vomited six to eight times a day, had five or six loose stools a day, and has kept down "a few sips of ginger ale and some crackers." No fever, no blood in the stool, no abdominal surgery. He stopped making much urine yesterday, then this morning "went a lot but it was dark."
He took every pill, every day. "The pharmacist told me never to miss the kidney pill, so I didn't."
He does not drink alcohol, does not smoke, and takes no over-the-counter medicines — he specifically denies ibuprofen or naproxen. No insulin at home. His last clinic visit was three months ago; his wife has the after-visit summary on her phone.
Vitals: T 37.1 °C · heart rate 112 · BP 104/62 seated, 88/54 standing at 1 min (heart rate 124 standing) · respiratory rate 28, deep and regular · SpO₂ 98 % on room air · Weight 94 kg (98 kg on the home scale last week, his wife says).
Exam: Alert, oriented, speaking in full sentences between deep breaths. Breath has a sweet, fruity odor. Dry oral mucosa, cracked lips. Neck veins flat lying at 30°. Lungs clear. Heart tachycardic, regular, no murmur. Abdomen soft with mild diffuse tenderness, no guarding, no rebound. No edema. Skin cool, turgor reduced over the sternum.
The deep, regular, air-hungry breathing is Kussmaul respiration — the lungs blowing off carbon dioxide to compensate for an acid load. It is a sign to read before any lab returns.
Medications (from the pharmacy list):
| Drug | Dose | Route | Frequency | Indication |
|---|---|---|---|---|
| Empagliflozin | 10 mg | PO | once daily | Type 2 diabetes with albuminuria (UACR 180 mg/g, A2) — kidney and heart protection; started 8 months ago |
| Lisinopril | 20 mg | PO | once daily | Hypertension; albuminuria |
| Metformin | 1,000 mg | PO | twice daily | Type 2 diabetes |
| Atorvastatin | 40 mg | PO | nightly | Atherosclerotic cardiovascular risk reduction |
He took all four this morning, with the ginger ale.
The clinic baseline (three months ago, 2026-07-14): Na 139, K 4.3, Cl 103, HCO₃ 25, BUN 16, creatinine 1.0 mg/dL, eGFR 86 mL/min/1.73 m² (CKD-EPI 2021), glucose 142, HbA1c 7.4 %, UACR 180 mg/g.
The ED resident's note at 14:40, before the gas is back: "61M, 3 days viral gastroenteritis, dehydrated, tachycardic, orthostatic. Glucose 206 — not DKA. Plan: 2 L normal saline bolus, ondansetron 4 mg IV, recheck BMP."
The first labs return at 15:05.
| Test | Result | Reference range |
|---|---|---|
| Sodium | 134 mEq/L | 136–145 |
| Potassium | 3.4 mEq/L | 3.5–5.0 |
| Chloride | 86 mEq/L | 98–107 |
| Bicarbonate (CO₂) | 14 mEq/L | 22–26 |
| Anion gap (lab-calculated) | 34 mEq/L | 8–12 |
| BUN | 46 mg/dL | 7–20 |
| Creatinine | 1.9 mg/dL | 0.7–1.3 |
| eGFR (lab-printed) | 40 mL/min/1.73 m² | not valid — creatinine is moving |
| Glucose | 206 mg/dL | 70–99 fasting |
| Calcium | 9.0 mg/dL | 8.5–10.5 |
| Magnesium | 1.9 mg/dL | 1.7–2.2 |
| Phosphorus | 4.1 mg/dL | 2.5–4.5 |
| Albumin | 4.0 g/dL | 3.5–5.0 |
| Lactate | 1.6 mmol/L | 0.5–2.0 |
| Serum osmolality, measured | 300 mOsm/kg | 275–295 |
| Serum osmolality, calculated | 296 mOsm/kg | 2 × Na + glucose/18 + BUN/2.8 = 268 + 11 + 16 |
| Osmolal gap | 4 mOsm/kg | < 10 |
| β-hydroxybutyrate | pending | 0.0–0.4 mmol/L |
| WBC | 13.2 × 10³/µL | 4.0–11.0 |
| Hemoglobin | 15.8 g/dL | 13.5–17.5 |
| Platelets | 240 × 10³/µL | 150–400 |
Arterial blood gas, room air (15:10): pH 7.29 (7.35–7.45) · pCO₂ 30 mmHg (35–45) · pO₂ 98 mmHg · HCO₃ 14 mEq/L (calculated).
Read the BMP in order — the skeleton key. Sodium 134: a touch low, read it with the glucose (glucose-corrected Na = 134 + 1.6 × (206 − 100)/100 ≈ 136 — essentially normal; the 1.6 factor is the standard correction). Potassium 3.4: low, and about to matter. Chloride 86: strikingly low for a sodium of 134 — the chloride has gone somewhere (vomit is hydrochloric acid). Bicarbonate 14: a metabolic acidosis. Anion gap 34: unmeasured anions are filling the space. BUN 46 and creatinine 1.9 against a baseline of 1.0: an AKI, with a BUN:creatinine ratio of 24:1. Glucose 206: mildly high — hold that number; Q4 asks what it means. Name the pattern before the diagnosis: a high-gap metabolic acidosis with a chloride that is too low, in a volume-depleted man with AKI.
Q4 is the evidence-update item — individual cards, no team step Say the hypothesis aloud, then the finding, then cards up on the five-point scale. Read the spread back. Ask "what else could this finding mean?" before you open the answer. Two rules every time: Unchanged means the finding does not move the hypothesis — it does not mean "I'm not sure" — and this is practice in updating, not a graded concordance test; there is no expert panel behind the key, the teaching explanation.
Q4Evidence update
If you were thinking diabetic ketoacidosis in a man on empagliflozin with three days of vomiting and fast, deep breathing, and you then learn that his glucose is 206 mg/dL, this hypothesis becomes:
Much less likely
Less likely
Unchanged
More likely
Much more likely
Show answer and explanation · Link to this question
Correct answer: C
Learning point: On an SGLT2 inhibitor the glucose is the one finding whose meaning is conditional. The drug holds the blood glucose near normal by dumping it in the urine while ketoacid production runs unopposed, so a glucose of 206 carries no information about whether ketoacidosis is present — it is the expected glucose in SGLT2 inhibitor ketoacidosis. The hypothesis stays exactly where the vomiting, the fruity breath and the Kussmaul breathing put it; what moves it are the gap (already 34) and a ketone measurement — Q6 asks which one. Off an SGLT2 inhibitor the same glucose would be reassuring; that is the circumstance the student must notice, and the one the resident missed. What else could the finding mean? A near-normal glucose during illness can also mean he has not eaten for three days, or that his own insulin secretion is still adequate — both true and both irrelevant to whether ketoacids are accumulating. The only way to know is to measure them. Course reading: Evidence for this case
Pause — commit before the reveal Q5 is the key-feature decision of Part 1. Cards up individually, then teams.
Q5Name the acid-base picture
Using the ABG and BMP above (pH 7.29, pCO₂ 30, HCO₃ 14, Na 134, Cl 86, albumin 4.0; your lab's normal gap is 8–12 and normal HCO₃ is 24), which statement best describes his acid-base status?
A single high-gap metabolic acidosis with appropriate respiratory compensation
A high-gap metabolic acidosis plus a metabolic alkalosis, with appropriate respiratory compensation
A high-gap metabolic acidosis plus a respiratory acidosis, with inadequate respiratory compensation
A high-gap metabolic acidosis plus a non-gap metabolic acidosis, with appropriate respiratory compensation
Show answer and explanation · Link to this question
Correct answer: B
Learning point: Three lines. Gap: 134 − (86 + 14) = 34, and albumin 4.0 needs no correction. Winter's formula — the pCO₂ the lungs should produce for this bicarbonate — is 1.5 × 14 + 8 = 29 ± 2, so the actual 30 sits inside the window: compensation is appropriate and there is no respiratory disorder. Delta-delta: ΔAG = 34 − 12 = 22 against ΔHCO₃ = 24 − 14 = 10, ratio 2.2 — above 2 means something is holding the bicarbonate up, and three days of vomiting hydrochloric acid (the chloride of 86 is its fingerprint) is a metabolic alkalosis hidden inside the acidosis. Without it his bicarbonate would be about 24 − 22 = 2: he is sicker than 14 looks. Why not the others: A stops after Winter's and never asks whether the bicarbonate fall fits the gap; C misreads Winter's — 30 is inside the 27–31 window, so compensation is appropriate; D runs the delta-delta in the wrong direction — a ratio below 1 means an added non-gap acidosis, above 2 means an added alkalosis. Course reading: Evidence for this case
Q6What is filling the gap?
Lactate is 1.6 mmol/L and the osmolal gap is 4 mOsm/kg. He takes empagliflozin and metformin. The resident says, "glucose is 206 — this isn't DKA." Which single additional test most directly settles what is filling the anion gap?
Serum β-hydroxybutyrate
Urine ketones by dipstick
A repeat serum lactate in two hours
Serum salicylate level
Show answer and explanation · Link to this question
Correct answer: A
Learning point: Run the gap differential (MUDPILES — methanol, uremia, DKA/ketones, propylene glycol, isoniazid/iron, lactate, ethylene glycol, salicylates) against what you already have: lactate 1.6 removes lactic acidosis, including metformin-associated lactic acidosis from the other drug on his list; osmolal gap 4 does not suggest a large unmetabolized alcohol load, but cannot exclude a late toxic-alcohol exposure; creatinine 1.9 is unlikely to explain this large gap by uremia alone. The SGLT2 exposure, fasting and symptoms make blood ketone measurement the best next test. The dominant ketoacid in DKA is β-hydroxybutyrate, measured in blood; the urine dipstick's nitroprusside reaction detects acetoacetate (and weakly acetone) and does not react with β-hydroxybutyrate at all, so the dipstick reads falsely low early in the illness. A serum β-hydroxybutyrate of 3.0 mmol/L or more is diagnostic-range ketonemia regardless of the glucose. On an SGLT2 inhibitor the drug dumps glucose into the urine, so the blood glucose stays near normal while ketoacid production runs unchecked — euglycemic diabetic ketoacidosis (the classic definition is a ketoacidosis at a glucose under 250 mg/dL; the 2024 consensus report draws the line at 200 — either way, the glucose is not the test). Why not the others: B is the insensitive test — his dipstick will read 1+ while his blood level is in the diagnostic range; C re-measures a value already normal and delays the diagnosis two hours; D chases a MUDPILES item with no history of aspirin and a normal osmolal gap, and salicylate does not explain a chloride of 86. Course reading: Evidence for this case
The reveal, 16:20. β-hydroxybutyrate 5.8 mmol/L. Urine dipstick ketones 1+. The diagnosis is SGLT2 inhibitor–associated ketoacidosis at a near-normal glucose — "euglycemic DKA" by the classic cutoff of 250 mg/dL (the 2024 consensus report draws the line at 200; either way, the glucose was never the test) — in a man whose vomiting stacked a metabolic alkalosis on top and whose kidneys are running on fumes. The resident re-reads his own note and crosses out "not DKA."
The mechanism in one breath: empagliflozin blocks glucose reabsorption in the proximal tubule, so glucose leaves in the urine. Blood glucose falls, insulin secretion falls with it, glucagon rises, and the liver switches to burning fat — ketoacids. Three days of not eating, vomiting and volume loss pushed the same switch harder. The glucose stayed near 200 because the kidney kept throwing it away. The acid did not.
Clinical Pearl — The gap decides, not the glucose In anyone on an SGLT2 inhibitor who is nauseated, vomiting, or breathing fast, a near-normal glucose — under 250 mg/dL by the classic definition, under 200 by the 2024 consensus — does not exclude ketoacidosis. Calculate the gap and send a serum β-hydroxybutyrate. The urine dipstick is the wrong test.
Part 2 — Data
Answers for this section: Q7 · Q8 · Q9
The 2 L of 0.9 % saline the ED ordered are in by 16:30. Empagliflozin, lisinopril and metformin are held on admission — the first order after the fluids. Standing blood pressure is now 98/60. Repeat labs are drawn at 16:30 (hour 2), before any insulin.
| Test | Hour 0 (14:10) | Hour 2 (16:30) | Reference range |
|---|---|---|---|
| Sodium | 134 | 135 mEq/L | 136–145 |
| Potassium | 3.4 | 3.1 mEq/L | 3.5–5.0 |
| Chloride | 86 | 92 mEq/L | 98–107 |
| Bicarbonate | 14 | 13 mEq/L | 22–26 |
| Anion gap | 34 | 30 mEq/L | 8–12 |
| BUN | 46 | 42 mg/dL | 7–20 |
| Creatinine | 1.9 | 1.8 mg/dL | 0.7–1.3 |
| Glucose | 206 | 188 mg/dL | 70–99 fasting |
| Magnesium | 1.9 | 1.8 mg/dL | 1.7–2.2 |
| Phosphorus | 4.1 | 3.9 mg/dL | 2.5–4.5 |
| β-hydroxybutyrate | — | 5.8 mmol/L | 0.0–0.4 |
Urinalysis (hour 0 sample):
| Dipstick | Result | Reference |
|---|---|---|
| Specific gravity | 1.028 | 1.005–1.030 |
| pH | 5.0 | 4.5–8.0 |
| Glucose | 4+ (≥1,000 mg/dL) | negative |
| Ketones | 1+ (15 mg/dL) | negative |
| Protein | trace | negative |
| Blood | negative | negative |
| Leukocyte esterase / nitrite | negative / negative | negative |
| Microscopy | Result | Reference |
|---|---|---|
| RBC | 0–2 /hpf | 0–2 /hpf |
| WBC | 0–2 /hpf | 0–4 /hpf |
| Casts | few hyaline casts; no granular, no muddy brown, no cellular casts | none or rare hyaline |
| Crystals / bacteria | none | none |
ECG: sinus tachycardia 112, PR 150 ms, QRS 88 ms, QTc 438 ms, no U waves, no peaked T waves.
Read the kidney next. Creatinine 1.9 against a baseline of 1.0 is 1.9 times baseline — KDIGO stage 1 (1.5–1.9×). Not stage 2, though it is one decimal away; the stage can change on the next draw, which is the point of tracking the trend. The printed eGFR of 40 is meaningless: the CKD-EPI equation assumes creatinine production equals excretion, a steady state, and his creatinine was climbing until two hours ago. Write "eGFR not valid — creatinine is moving" and follow the creatinine.
Which compartment? Prerenal — the kidney is underperfused but structurally intact. Four kilograms below last week's home-scale weight, orthostatic drop of 16 mmHg, flat neck veins, dry mucosa; BUN:creatinine 24:1 (the kidney is reabsorbing urea avidly, which is what a hypoperfused kidney does); specific gravity 1.028 (concentrated urine — read it with a grain of salt, because 4+ glucose from the empagliflozin inflates the specific gravity a little); a bland sediment with hyaline casts and nothing else. No granular casts, no tubular cells, no blood. We do not send a FENa: he is not oliguric, and the SGLT2 inhibitor pushes sodium into the urine anyway, so the number would mislead.
Pause — commit before the reveal Q7 and Q8 are the key-feature decisions of Part 2.
Q7Stage, compartment, first move
Creatinine 1.9 mg/dL (baseline 1.0 three months ago), BUN 46, lab eGFR 40, specific gravity 1.028, hyaline casts only, 4 kg weight loss in a week and orthostasis. Which statement is correct?
Stage 2 AKI — the creatinine nearly doubled — give isotonic crystalloid and call nephrology today
CKD G3b — eGFR 40 — add CKD to the problem list and arrange outpatient nephrology referral
Stage 1 AKI, prerenal — isotonic crystalloid, recheck creatinine in 12–24 h, ignore the printed eGFR
Acute tubular necrosis (intrinsic AKI) — send a FENa to confirm before giving more fluid
Show answer and explanation · Link to this question
Correct answer: C
Learning point: Stage by the ratio: 1.9 ÷ 1.0 = 1.9, inside the stage 1 band (1.5–1.9×); stage 2 starts at 2.0×. The compartment is prerenal on every line — losses, orthostasis, a 4 kg drop, BUN:creatinine over 20:1 with nothing else raising urea (no GI bleed, no steroids), concentrated urine, bland sediment with hyaline casts. The treatment of prerenal AKI is to restore perfusion with isotonic crystalloid and then look again — creatinine, urine output, orthostatics — in 12–24 hours. The hold-list drugs (lisinopril, metformin, empagliflozin) stay off while the kidney recovers. Why not the others: A over-stages (1.9× is not 2.0×) and calls nephrology for a stage 1 prerenal AKI that fluids will fix; B reads the eGFR as if it were steady state — it is an AKI, not CKD, and chronicity needs three months; D invents tubular injury the sediment does not show (no granular casts) and sends a FENa in a non-oliguric patient on a natriuretic drug — the one setting where it is uninterpretable. Course reading: Evidence for this case
Q8Insulin, or potassium first?
It is hour 2. Potassium is 3.1 mEq/L, magnesium 1.8, β-hydroxybutyrate 5.8, glucose 188. He is on a cardiac monitor. The resident wants to start an insulin infusion now. The best next step is:
Start insulin at 0.1 units/kg/h now and add KCl 20 mEq to each liter of the IV fluid
Give oral potassium alone and wait until morning to reconsider insulin, despite ongoing ketoacidosis
Start insulin at 0.1 units/kg/h without potassium — 3.1 is mild and the acidosis comes first
Hold insulin; infuse potassium under the monitored protocol until K >3.5, then start insulin with dextrose
Show answer and explanation · Link to this question
Correct answer: D
Learning point: At K 3.1, delay insulin and replace potassium with monitoring until K is above 3.5 mmol/L. The 2024 consensus describes starting replacement at 10 mmol/h; a higher rate requires the local monitored protocol and appropriate access. Once safe, give insulin with dextrose because glucose is already below 250 mg/dL. Recheck potassium promptly. Why not the others: A and C start insulin before potassium is safe. B under-treats ongoing ketoacidosis and postpones reassessment. D sequences potassium correction before definitive insulin treatment. Course reading: Evidence for this case
17:30. After an hour of KCl at 20 mEq/h on the monitor — 10 mEq/h through each of his two peripheral IVs, because a single peripheral line takes no more than 10 mEq/h and 20 mEq/h through one line needs central access — potassium is 3.6. Insulin starts at 0.05 units/kg/h, approximately 4.7 units/h for 94 kg because glucose is already below 250 mg/dL, with dextrose 5 % in the maintenance fluid and KCl 20–30 mEq in every liter. The targets while the gap closes: glucose 150–200 mg/dL, potassium 4.0 or higher (magnesium 1.8 is acceptable, and it is checked because potassium will not hold without it), and resolution requires blood β-hydroxybutyrate below 0.6 mmol/L plus venous pH at least 7.3 or bicarbonate at least 18 mEq/L. Anion-gap closure alone is not the stopping rule; overlap subcutaneous insulin before stopping the infusion when ongoing insulin is indicated. Potassium is rechecked every 2 hours. He is admitted to the step-down unit on 0.9 % saline at 250 mL/h.
The night. By 04:00 Friday (hour 14) he has received 5 L of 0.9 % saline in total. He is making urine, 90–120 mL/h. Orthostatics have resolved. He says he is "starving," which is the best sign of the night.
| Test | Hour 2 | Hour 14 (04:00 Fri) | Reference range |
|---|---|---|---|
| Sodium | 135 | 137 mEq/L | 136–145 |
| Potassium | 3.1 | 3.9 mEq/L | 3.5–5.0 |
| Chloride | 92 | 104 mEq/L | 98–107 |
| Bicarbonate | 13 | 15 mEq/L | 22–26 |
| Anion gap | 30 | 18 mEq/L | 8–12 |
| BUN | 42 | 30 mg/dL | 7–20 |
| Creatinine | 1.8 | 1.4 mg/dL | 0.7–1.3 |
| Glucose | 188 | 164 mg/dL | 70–99 fasting |
| Phosphorus | 3.9 | 1.6 mg/dL | 2.5–4.5 |
| Magnesium | 1.8 | 1.9 mg/dL | 1.7–2.2 |
| β-hydroxybutyrate | 5.8 | 2.1 mmol/L | 0.0–0.4 |
| Arterial blood gas | — | pH 7.31 · pCO₂ 31 mmHg · HCO₃ 15 mEq/L | pH 7.35–7.45 · pCO₂ 35–45 mmHg · HCO₃ 22–26 mEq/L |
Read the trend. The gap is closing (34 → 18) and the ketones are falling — the insulin is working. The creatinine is falling (1.9 → 1.4) — it was prerenal. Winter's: 1.5 × 15 + 8 = 30.5 ± 2 → actual 31, still appropriate. But the chloride has climbed 18 points (86 → 104) while the sodium climbed 3, and the bicarbonate has barely moved. Run the delta-delta again: ΔAG = 18 − 12 = 6; ΔHCO₃ = 24 − 15 = 9; ratio 0.67. Below 1. The alkalosis from vomiting is gone (he stopped vomiting, and we repleted his chloride), and a non-gap acidosis is now being added. Two things are adding it: five liters of 0.9 % saline (154 mEq/L of chloride, far above plasma) and the ketoanions he lost in the urine overnight — each ketoanion excreted is a bicarbonate the liver can no longer regenerate when the ketone is metabolized.
The phosphorus. The fall from 4.1 to 1.6 mg/dL during treatment warrants repeat measurement and clinical assessment. Routine phosphate replacement in DKA has not improved outcomes. The 2024 consensus considers replacement when muscle weakness or cardiorespiratory compromise accompanies phosphate below 1.0 mmol/L (about 3.1 mg/dL); that unit is not 1.0 mg/dL. Severe hypophosphatemia and local replacement protocols require individual review. He is currently asymptomatic; monitor and reassess rather than prescribe automatic prophylaxis. If IV phosphate is used, account for kidney function, potassium and calcium.
Q9The next liter
Hour 14: 5 L of 0.9 % saline in, chloride 104 (from 86), bicarbonate 15, anion gap 18, delta-delta 0.67, glucose 164, potassium 3.9, sodium 137. The insulin infusion continues. Which fluid plan is best for the next liters?
Switch to a balanced crystalloid with dextrose and potassium
Continue 0.9 % saline at 250 mL/h with dextrose — the gap is still open
Change to 0.45 % saline to cut the chloride load in half
Give sodium bicarbonate 100 mEq IV for the bicarbonate of 15
Show answer and explanation · Link to this question
Correct answer: A
Learning point: The chloride tells you what the saline did. 0.9 % saline carries 154 mEq/L of chloride against a plasma chloride of about 100; after several liters the excess chloride displaces bicarbonate and produces a hyperchloremic (non-gap) metabolic acidosis — the falling delta-delta is the arithmetic of exactly that. Balanced crystalloids (lactated Ringer's, Plasma-Lyte) carry about 100–110 mEq/L of chloride plus a buffer (lactate or acetate) the liver converts to bicarbonate; in large volumes they avoid the chloride load, and the current DKA consensus prefers them. He still needs volume, dextrose (glucose is 164 and insulin is running) and potassium (target ≥4.0), so those ride in the new fluid. Why not the others: B continues to add the chloride that is now the problem — the open gap is an argument for continuing insulin, not for continuing saline; C omits the dextrose and potassium he needs now (glucose 164 on an insulin infusion; potassium 3.9); half-normal saline with dextrose was the 2009 algorithm's step once the corrected sodium was normal, but the current consensus prefers a balanced crystalloid, which cuts the chloride load while staying close to isotonic; D treats a number — bicarbonate is reserved for severe acidemia in ketoacidosis (pH below 6.9 in the 2009 ADA consensus, below 7.0 in the 2024 report), his pH is 7.31, and bicarbonate would worsen the hypokalemia and add sodium. Course reading: Evidence for this case
Clinical Pearl — Every fluid is a drug with a dose Write the chloride content of the bag next to the plasma chloride. 0.9 % saline: 154. Lactated Ringer's: 109. Plasma-Lyte: 98. Plasma: about 100. The gap between the bag and the plasma, times the liters you give, is the acidosis you are ordering.
Part 3 — Course and Decisions
Answers for this section: Q10
Saturday 08:00 (hour 42). At hour 20, β-hydroxybutyrate was 0.5 mmol/L and venous pH 7.31, meeting DKA resolution criteria despite residual hyperchloremic acidosis; the insulin infusion was stopped at hour 22 after a two-hour overlap with subcutaneous insulin, and he has been eating since Friday noon. He is off IV fluids. Creatinine is 1.1. He feels well and wants to go home.
The intern on rounds is worried. "His bicarbonate is still 17 and his chloride is 110. The gap is 11. He's on lisinopril and he's diabetic — I think he has a type 4 RTA." He has already sent an arterial gas and a spot urine.
| Test | Hour 14 | Day 3 (Sat 08:00) | Reference range |
|---|---|---|---|
| Sodium | 137 | 138 mEq/L | 136–145 |
| Potassium | 3.9 | 4.1 mEq/L | 3.5–5.0 |
| Chloride | 104 | 110 mEq/L | 98–107 |
| Bicarbonate | 15 | 17 mEq/L | 22–26 |
| Anion gap | 18 | 11 mEq/L | 8–12 |
| BUN | 30 | 18 mg/dL | 7–20 |
| Creatinine | 1.4 | 1.1 mg/dL | 0.7–1.3 |
| Glucose | 164 | 148 mg/dL | 70–99 fasting |
| Phosphorus | 1.6 | 2.6 mg/dL | 2.5–4.5 |
| Magnesium | 1.9 | 2.0 mg/dL | 1.7–2.2 |
| β-hydroxybutyrate | 2.1 | 0.3 mmol/L | 0.0–0.4 |
| Arterial blood gas | — | pH 7.33 · pCO₂ 33 mmHg · HCO₃ 17 mEq/L | pH 7.35–7.45 · pCO₂ 35–45 mmHg · HCO₃ 22–26 mEq/L |
| Urine (spot, Day 3) | Result | Reference |
|---|---|---|
| Urine sodium | 60 mEq/L | varies with intake (mEq/L) |
| Urine potassium | 30 mEq/L | varies with intake (mEq/L) |
| Urine chloride | 115 mEq/L | varies with intake (mEq/L) |
| Urine pH | 5.2 | 4.5–8.0; below 5.5 expected during acidemia |
| Urine anion gap | (60 + 30) − 115 = −25 mEq/L | negative = gut or bag (kidney excreting ammonium); positive = RTA screen |
Your read. Calculate the anion gap and expected respiratory compensation from the new values. Compare the ketones and chloride with admission, then use the urine results to judge whether impaired renal acid excretion is likely. Commit to Q10 before opening the interpretation.
Pause — commit before the reveal Q10 is the key-feature decision of Part 3: a differential the student must recognize and must not chase.
Q10The bicarbonate that would not come up
Day 3: Na 138, K 4.1, Cl 110, HCO₃ 17, anion gap 11, creatinine 1.1, β-hydroxybutyrate 0.3. Urine Na 60, K 30, Cl 115 (urine anion gap −25), urine pH 5.2. He received 5 L of 0.9 % saline on day 1. Which is the best explanation and plan?
Type 4 RTA from lisinopril plus diabetic hypoaldosteronism — check plasma renin and aldosterone and start fludrocortisone
Non-gap acidosis from the saline load and urinary ketoanion loss — no further saline or alkali; recheck a BMP in 48 h
Type 1 (distal) RTA unmasked by the illness — start potassium citrate and order a renal ultrasound for nephrocalcinosis
Incompletely treated ketoacidosis — restart the insulin infusion at 0.1 units/kg/h until the bicarbonate is above 18
Show answer and explanation · Link to this question
Correct answer: B
Learning point: Resolved ketonemia, chloride 110 after a large saline load, normal potassium and recovering kidney function favor transient hyperchloremic acidosis from saline and urinary ketoanion loss. A negative urine anion gap supports ammonium excretion in a suitable setting but is an imperfect surrogate; ketoanions and other unmeasured urinary ions can confound it. Urine pH 5.2 argues against classic distal RTA but is not a universal test of total acid excretion. Recheck rather than reflexively prescribe alkali or restart a DKA infusion. Why not the others: A and C are less consistent with the course and electrolyte pattern; no single UAG or potassium value excludes every RTA. D ignores the resolved blood ketones. Persistent or worsening acidosis requires reassessment rather than an assumed saline explanation. Course reading: Evidence for this case
The RTA table — the differential you recognize and do not chase. Fill one row per type, from the student handout. A PA student needs the three rows; the loading tests and fractional bicarbonate excretion are fellowship work.
| Type | Where and what fails | Urine pH during acidosis | Serum K | Classic causes | Hallmark | Treatment |
|---|---|---|---|---|---|---|
| 1 — distal | Collecting duct cannot secrete H⁺ | > 5.5 (cannot acidify) | Low, often severe | Sjögren's, SLE, amphotericin B, lithium | Calcium phosphate stones, nephrocalcinosis | Alkali — potassium citrate (replaces K, adds base, cuts stone risk) |
| 2 — proximal | Proximal tubule cannot reabsorb HCO₃ | Variable; < 5.5 once serum HCO₃ falls low enough | Low, moderate | Fanconi syndrome (a leaky proximal tubule that also spills glucose, phosphate and amino acids), myeloma, tenofovir, acetazolamide | Bicarbonate wasting; glucosuria, phosphaturia, aminoaciduria if Fanconi | Large-dose alkali; thiazide; treat the cause |
| 4 — hyperkalemic | Aldosterone low or resisted → less K⁺ and H⁺ secretion | < 5.5 (can acidify) | High | Diabetes (the diabetic kidney makes too little renin, so too little aldosterone), ACEi/ARB, NSAIDs, heparin, TMP-SMX, K-sparing diuretics, CKD | Mild acidosis (HCO₃ 17–24) with hyperkalemia | Stop the drug; low-K diet; loop diuretic; fludrocortisone in hypoaldosteronism |
His row would have been type 4 — and the potassium of 4.1 and the UAG of −25 close that door.
Discharge planning — the conversation that prevents the readmission. The question is not whether to restart his drugs but when, and what to hand him in writing.
What he should have done on Monday night, when the vomiting started. The SADMANS list, read against his bottles:
| Letter | Class | His drug | Why it is held when vomiting, having diarrhea, febrile or not eating | Restart when |
|---|---|---|---|---|
| S | Sulfonylureas | (none) | Hypoglycemia when not eating | Eating normally |
| A | ACE inhibitors | lisinopril | Efferent arteriole dilated → GFR falls when volume is low; hyperkalemia | Creatinine back to baseline and eating; BMP within 1–2 weeks |
| D | Diuretics | (none) | Compound the volume loss | Euvolemic and eating/drinking |
| M | Metformin | metformin | Accumulates when GFR falls → lactic acidosis risk | Creatinine stable at baseline and eating; full dose if eGFR ≥45 at steady state |
| A | ARBs (and ARNI) | (none) | As for ACE inhibitors | As for ACE inhibitors |
| N | NSAIDs | (none — and keep it that way) | Afferent arteriole constricted; the "triple whammy" with an ACEi and a diuretic | Use acetaminophen instead |
| S | SGLT2 inhibitors | empagliflozin | Volume loss continues through the urine; euglycemic ketoacidosis when not eating | Ordinary sick-day hold: resume after recovery according to the prescribing plan. After DKA: keep off pending specialist review |
| also | Mineralocorticoid antagonists (spironolactone, eplerenone, finerenone) | (none) | Hyperkalemia in AKI | Potassium normal and eGFR stable |
Medicines not automatically held by a sick-day mnemonic: beta-blockers, statins and thyroid replacement need individualized review. Severe hypotension, bradycardia or another drug-specific problem can change the plan. Do not abruptly stop chronic steroids; seek advice about stress dosing. Changes to transplant immunosuppression require the transplant team.
Metformin, specifically. The thresholds he should know: do not start it below an eGFR of 45; reduce the dose at 30–45; stop it below 30. Hold it during any illness with vomiting, diarrhea or poor intake, and during an AKI. Around IV iodinated contrast, the ACR approach permits continuation at stable eGFR ≥30 without AKI; with AKI or eGFR <30, hold and reassess kidney function after 48 hours before restarting. Certain arterial catheter procedures also require holding; follow the stated procedural policy. The drug label is stricter and also holds it at an eGFR of 30–60, so hospital policy varies. Metformin did not cause his acidosis — his lactate was 1.6 — but a man whose GFR fell by half while he kept taking 2,000 mg a day was one bad day from a second acidosis. His steady-state eGFR is 86, so he restarts the full dose once his creatinine has been at baseline for 48 hours and he is eating.
Empagliflozin after this admission. Routine resumption after an ordinary sick-day interruption is different from restarting after SGLT2-associated DKA. Empagliflozin stays off at discharge; the diabetes and kidney teams review the indication, precipitating factors and recurrence risk before considering future use. The 2024 consensus does not routinely recommend restarting after DKA in type 2 diabetes because safety data are limited. Before elective surgery, empagliflozin, dapagliflozin and canagliflozin are generally withheld for three days and ertugliflozin for four.
His wife's question. Before discharge she asks the one that matters most: "If this happens again and I check his sugar and it's 180, that means it's not the ketone thing, right? I can just push fluids at home?" The answer is the whole case in a sentence: on empagliflozin a glucose of 180 does not exclude ketoacidosis. Check blood ketones; if they are 1.5 mmol/L or higher, or he is breathing fast, or he cannot keep fluids down or develops signs of dehydration, he comes in. That sentence goes on the plan.
The written sick-day plan. One page, in his hand before he leaves, copied to his wife's phone:
If you are vomiting, have diarrhea, have a fever, or cannot eat and drink normally: 1. STOP empagliflozin, lisinopril and metformin. Do not take ibuprofen, naproxen or any "anti-inflammatory." Acetaminophen is fine. 2. Keep drinking — small sips often. Broth, sports drinks, or water with crackers. Aim for a cup every hour you are awake. 3. Check blood ketones (the meter and strips from the pharmacy) if you feel sick to your stomach, are breathing fast, or feel short of breath — even if your sugar is normal. A normal sugar does not rule out the ketone problem on this medicine. 4. Go to the emergency department if ketones are 1.5 or higher, if you are breathing fast, if you cannot keep fluids down for 24 hours, or if you make very little urine. 5. Call before restarting. Lisinopril and metformin can resume after normal intake and kidney function have recovered according to the discharge plan. Do not restart empagliflozin yourself after this DKA admission; keep it off until the diabetes/kidney team has reviewed it. 6. Keep taking atorvastatin. Only the three pills named in step 1 stop. If you are ever prescribed a beta-blocker (metoprolol, carvedilol), do not stop it automatically; urgent advice is needed if you are faint, very hypotensive or bradycardic. Ask before changing other medicines.
Resolution and Teaching Points
Answers for this section: W1 · W2 · W3 · Q4 · Q5 · Q6 · Q7 · Q8 · Q9 · Q10
Mr. Rourke went home on day 4 (Sunday). Day-5 labs at the clinic nurse visit: Na 139, K 4.2, Cl 106, HCO₃ 22, anion gap 11, BUN 14, creatinine 1.0 — eGFR 86 at steady state, back to his baseline. Lisinopril was restarted on day 4 with a BMP at one week (creatinine 1.0, K 4.4). Metformin was restarted at full dose on day 6 once two creatinines sat at baseline. Empagliflozin remained withheld at the one-week visit pending specialist risk–benefit review; the sick-day plan was reviewed and a blood ketone meter was dispensed. He asked for the plan to be taped inside the medicine cabinet. It was.
The ED resident presented the case at the department's morbidity conference under the title "Not DKA."
Clinical Pearl — Read the BMP in order, every time Sodium (with the glucose), potassium, chloride, bicarbonate, gap, BUN:creatinine, glucose — then calcium, magnesium, phosphorus. The chloride of 86 was the clue nobody read. A chloride that is far too low for the sodium means a metabolic alkalosis is hiding somewhere, and in a vomiting patient you know where.
Clinical Pearl — Three lines, then a diagnosis Anion gap (corrected for albumin), Winter's formula, delta-delta. Every high-gap acidosis gets all three before anyone names a cause. A delta-delta above 2 is a hidden alkalosis; below 1 is a hidden non-gap acidosis; the second one usually arrives in the bag.
Clinical Pearl — On an SGLT2 inhibitor, ketones decide Euglycemic DKA can occur without marked hyperglycemia. Measure blood β-hydroxybutyrate. Treat with fluids, potassium replacement before insulin when potassium is low, and insulin with dextrose as indicated. Hold the SGLT2 inhibitor; after DKA, do not use the ordinary sick-day card as permission to restart it.
Clinical Pearl — Stage from the trend, not from the eGFR KDIGO stage is a ratio to baseline, and the printed eGFR is a steady-state number that means nothing while creatinine is moving. Write "eGFR not valid — creatinine is moving" and follow the creatinine. A prerenal AKI declares itself: losses, orthostasis, BUN:creatinine over 20, concentrated bland urine with hyaline casts. Fluids, hold the hold-list drugs, look again in 12–24 hours.
Clinical Pearl — A negative urine anion gap is a working kidney In this recovering patient, a negative urine anion gap supports ongoing ammonium excretion and makes classic impaired-acid-excretion RTA less likely. It is a surrogate, not a direct ammonium measurement: unmeasured urinary anions, low distal sodium delivery and other confounders can mislead. Interpret the gap with the drug history, potassium, urine pH and clinical course.
Pitfall — "Glucose is 206, so it's not DKA" The most expensive sentence in this case. In an SGLT2 inhibitor user, glucose does not reliably exclude ketoacidosis. The ED lost two hours and gave two liters of saline without insulin or potassium because a normal number was read as reassurance.
Pitfall — Diabetes plus an ACE inhibitor plus a low bicarbonate equals type 4 RTA Type 4 RTA usually presents with hyperkalemic non-gap acidosis. A negative UAG, normal potassium and this recovery trajectory make it less likely, but do not constitute universal exclusion criteria. Reassess if the acidosis persists rather than assuming either RTA or saline indefinitely.
Concept Links
Answers for this section: W1 · W2 · W3 · Q4 · Q5 · Q6 · Q7 · Q8 · Q9 · Q10
Assigned reading: the explanations in this case and the question-linked course evidence guide. The objective map distinguishes tested decisions from concepts taught in explanations.
Concepts: aki-01, aki-02, aki-04, aki-15, htn-cardiorenal-25, lytes-02, lytes-03, lytes-11, lytes-20, lytes-21, lytes-22, lytes-23, lytes-24, lytes-26, structural-19, structural-23.