Chapter 29 — Key Takeaways
One page. ★ Closes Part V.
⚠️ Nine Days
Ivor Petrakis, 80, fractured hip. ⚠️ ~350 kcal/day for nine days. 78 → 73 kg. Lost the ability to stand from a chair. Rehabilitation bed instead of home.
Day 1 nil (theatre) · Day 2 nil (list bumped) · Day 3 ~300 · Day 4 ~500 (nausea, tray removed) · Day 5 ~400 (at X-ray over lunch) · Day 6 ~600 (denture at home) · Day 7 nil until 18:00 (possible procedure that didn't happen) · Day 8–9 ~650–700.
⚠️ Not one of those decisions was wrong. Nobody decided to starve him. Nine individually reasonable decisions did it, and no single person was positioned to see the total.
He was screened on admission. He scored medium risk. Nothing was triggered by it.
⚠️ Every fix was free: feed him when the list is confirmed full · treat the nausea before the meal · replace a meal missed for an X-ray · phone about the denture · feed him at 16:00 when the decision changed · OPEN THE PACKAGING · total the chart daily.
⚠️ His line: "Nobody was unkind. Not one person. I just didn't eat for a week and it took five weeks to get back."
§29.1–29.2 — Scale and Screening
Prevalence on admission commonly a fifth to a half. Associated with longer stay, more complications, higher readmission, higher mortality, ⚠️ and changed discharge destination.
GLIM criteria: ⚠️ phenotypic (weight loss, low BMI, reduced muscle mass) plus aetiologic (reduced intake/assimilation, disease burden/inflammation).
Tools: MUST · NRS-2002 · MNA · SGA. ⚠️ Why screening fails: WEIGHT ISN'T MEASURED (the commonest failure point) · forms compete · nobody owns the result · and nutrition has no monitor — nothing alarms.
⚠️ §29.3 — The Nil-by-Mouth Cascade
Prolonged pre-procedure fasting · cancelled and deferred procedures with no re-feed · absent at mealtimes · trays delivered and removed · no help to eat · ⚠️ PACKAGING · dentures · positioning · untreated nausea, pain, constipation.
Fixes: protected mealtimes · red tray systems · charts someone reads · ⚠️ RE-FEED WHEN A PROCEDURE IS CANCELLED · open the packaging · get the dentures · treat the nausea.
⚠️ The single highest-yield intervention: when a procedure is cancelled, the patient eats. No equipment, no budget, no training — and the one most often missed.
⚠️ §29.3b — Lying Still
Bed rest causes measurable muscle loss within DAYS, faster in older adults, accelerated by inflammation, compounded by anabolic resistance (Ch 8 §8.6) and catabolic drugs.
⚠️ Nutrition and mobilization are the SAME intervention. Protein without loading does much less; loading without protein has nothing to build with. Ivor got neither for six days.
§29.4–29.6 — The Routes
⚠️ IF THE GUT WORKS, USE IT. The least invasive adequate route wins.
1. Food, optimized and fortified · 2. ONS 🟢–✅ (between meals, as an ADDITION — "left on the locker, untouched, for three days" is the standard failure) · 3. Enteral · 4. ⚠️ Parenteral, last.
⚠️ NG TUBE MISPLACEMENT INTO THE LUNG IS A NEVER EVENT AND HAS KILLED PEOPLE. Confirm by aspirate pH (≤5.5) and/or radiography interpreted competently. ⚠️ The whoosh test and bubbling in water are NOT valid.
⚠️ Diarrhoea on enteral feeding is more often caused by medications (sorbitol preparations, magnesium, antibiotics) or C. difficile than by the feed.
❌ ⚠️ PEG in advanced dementia does NOT prolong survival, prevent aspiration, improve pressure ulcers, function or comfort — because aspiration is largely of saliva and refluxed gastric contents. Careful hand feeding instead.
Parenteral: narrow indications (non-functioning or inaccessible gut). ⚠️ Catheter-related bloodstream infection, thrombosis, liver disease, refeeding, cost, specialist team. Historically over-used.
⚠️ §29.7 — REFEEDING SYNDROME
Prolonged starvation depletes INTRACELLULAR phosphate, potassium and magnesium while serum may look normal. Thiamine stores deplete in weeks.
| Feed carbohydrate | ⚠️ Insulin surges |
|---|---|
| ⚠️ PO₄, K, Mg driven into cells | ⚠️ Serum crashes |
| Thiamine consumed by carbohydrate metabolism | ⚠️ Wernicke's |
| Sodium and water retained | Fluid overload |
⚠️ Arrhythmia, cardiac failure, respiratory failure, rhabdomyolysis, seizures, death — typically within 72 hours. Hypophosphataemia is the hallmark. ⚠️ Described in liberated prisoners of war and famine survivors: people survived starvation and were killed by food.
HIGH RISK if ONE of: BMI <16 · loss >15% in 3–6 months · little intake >10 days · ⚠️ low pre-feeding K, PO₄ or Mg. OR TWO of: BMI <18.5 · loss >10% · little intake >5 days · ⚠️ alcohol misuse, insulin, chemotherapy, antacids, diuretics.
⚠️ 1. THIAMINE BEFORE FEEDING — not with it. ⚠️ 2. START LOW, GO SLOW — around 5–10 kcal/kg/day in the highest risk. ⚠️ 3. CHECK PO₄, K, Mg BEFORE and DAILY AFTER; replace aggressively. ⚠️ 4. DO NOT wait for normal levels to start.
⚠️ The instinct — "this person is starving, feed them properly" — is the instinct that causes the harm. "Eat more" needs a RATE attached, exactly as "eat less" needed a dose.
⚠️ §29.8 — Critical Illness: the intuitions were wrong
| Early full parenteral supplementation | ⚠️ EPaNIC — LATE initiation associated with better recovery |
|---|---|
| Full vs trophic enteral | ⚠️ EDEN — trophic no worse in ARDS |
| More calories | ⚠️ TARGET — no mortality benefit |
| More protein | ⚠️ EFFORT Protein — no benefit, possible harm in some subgroups incl. AKI |
Proposed explanations: ⚠️ endogenous energy production (feeding on top produces overfeeding) · ⚠️ suppression of autophagy (a much better-supported use of that mechanism than Ch 21's fasting literature made) · hyperglycaemia.
⚠️ Consensus: start enteral EARLY but LOW, advance gradually across the first week, don't chase targets in the acute phase, avoid overfeeding. NOT an argument for starving anyone, and it does not extend past the acute phase.
⚠️ The forgotten part: the RECOVERY phase, where requirements are high and sustained for months and delivery is worst — because by then they've left the unit that was counting.
§29.9–29.10 — Surgery and Drugs
⚠️ "Nil by mouth from midnight" is not supported — clear fluids to ~2 h, light meal to ~6 h. 🟢 preoperative carbohydrate loading (ERAS) · 🟢 early postoperative feeding · 🟢 prehabilitation · 🟡 immunonutrition · ✅ preoperative optimization in malnourished patients — ⚠️ which means identifying them in clinic, not on the ward.
⚠️ Warfarin: CONSISTENCY, not avoidance — the row most often got wrong in both directions. Levothyroxine and calcium/iron/coffee · tetracyclines and quinolones with divalent cations · grapefruit and CYP3A4 · metformin and B12 · PPIs · ⚠️ enteral feeds binding phenytoin, levothyroxine and ciprofloxacin.
⚠️ §29.11 — THE INVERSIONS, ALL AT ONCE
| This book said | ⚠️ In the acutely unwell |
|---|---|
| Aim for a healthy weight | ⚠️ Unintended loss is a complication |
| Energy density down | ⚠️ UP — fortify everything |
| More fibre | Sometimes; sometimes low-residue |
| Reduce added sugar | ⚠️ Irrelevant |
| Avoid ultra-processed food | ⚠️ ONS ARE ultra-processed, and they save lives |
| Eat enough | ⚠️ Yes — except the first 72 h of refeeding, where it kills |
| "Eat less" | ⚠️ THE MOST DANGEROUS SENTENCE IN THE BOOK |
⚠️ Almost all population dietary advice is calibrated for a person whose problem is EXCESS. A substantial number of people in a hospital have the opposite problem.
⚠️ §29.12 — The Restriction Pattern, Closed
Five instances: cancer · CKD · IBS · IgG panels · and hospital diet orders.
⚠️ THE GENERAL RULE: a person who is unwell, motivated and looking for something to control is unusually likely to be handed a restriction — by a clinician, a test, a protocol, or themselves.
⚠️ The question is almost never "what should be removed?" It is "are they getting enough, and of what?" Any restriction should be specific, time-limited, and justified against what it costs.
§29.12b — Nobody Owns the Total
⚠️ Every professional is doing their job correctly. Nobody's job is to notice 350 kcal a day for nine days. Helps: nutrition support teams · screening with an automatic referral trigger · protected mealtimes · ⚠️ nutrition on the ward round. ⚠️ And "improve hospital food" is the wrong target — a nutritionally excellent meal removed untouched because the patient was at X-ray has failed for reasons unrelated to the menu.
⚠️ §29.13 — When Feeding Should Stop
⚠️ A dying person is not dying because they stopped eating. They stopped eating because they are dying.
Artificial nutrition at end of life generally does not prolong life or improve comfort and can cause harm. Comfort feeding, mouth care, sips of what's wanted. ⚠️ And the family's distress deserves addressing directly: feeding someone is what you do when you love them, and being told to stop feels like being told to stop caring.
🧾 Economics — running backwards
⚠️ Free or near-free: ward scales (once) · protected mealtimes · re-feeding after cancellation · opening packaging · food fortification · THIAMINE BEFORE FEEDING. Expensive: parenteral nutrition · complications · ⚠️ an extra hospital day · ⚠️ Ivor's five-week rehabilitation bed.
⚠️ The expensive end is substantially downstream of the free end not being done.
⚠️ §29.16 — Part V Verdict Re-Audit
Affirmative share: 15.4% (Ch 16) → 25.3% (Ch 22) → ~42% (Ch 28).
⚠️ Honest, because the QUESTIONS changed, not the standard. Part IV adjudicated contested public arguments — contested precisely because the evidence is ambiguous. Part V asked applied clinical questions, and Part V's ✅s are largely clinical-medicine claims (adrenaline, coeliac testing, tube position, resistance training) where randomized evidence is far more available than in dietary epidemiology. ⚠️ And the 🟡 count grew too, 20 → 60 — which is what you would NOT see if the standard had loosened.
⚠️ Caution: a reader who concludes "nutrition science is settled" has drawn the wrong lesson from a chapter about intensive care. Part IV's thirteen-entry 🟡 column remains the honest picture of the field's core.
One Thing to Remember
⚠️ When a procedure is cancelled, the patient should eat.
No equipment, no budget, no training — and it is the intervention most often missed.