Refeeding Syndrome: How Slowly Is Slowly Enough
A severely malnourished trauma patient finally has a working gut. The randomized evidence for feeding her slowly is real — but it was built on patients who had already crashed their phosphate, and hers hasn't moved yet.
S.A., a 39-year-old woman, spent six days without meaningful oral intake before her emergency bowel resection for a strangulated hernia, on top of what her family describes as months of poor appetite and unintentional weight loss she'd attributed to stress before the surgery forced the issue. Her gut is finally working again on post-operative day three, feeding tube in place and ready to start, and her admission labs already show a phosphate of 2.1 and magnesium of 1.3 — low enough, even before any feeding has started, to flag her as high-risk for refeeding syndrome once nutrition genuinely resumes.
Refeeding syndrome is driven by insulin's own physiology: reintroducing carbohydrate after prolonged starvation triggers an insulin surge that drives phosphate, potassium, and magnesium sharply into cells, and the resulting extracellular deficits — not the calories themselves — are what can precipitate cardiac arrhythmia, respiratory failure, and death within the first days of refeeding. The traditional NICE-derived protocol responds to this by starting caloric advancement very slowly (5-10 kcal/kg/day) with pre-emptive electrolyte correction ahead of any real caloric load. The one randomized trial run in this exact population points the opposite way from the intuition that slow feeding is merely cautious bookkeeping. Doig and colleagues enrolled critically ill adults whose phosphate had already fallen below 0.65 mmol/L after feeding began, and found protocolised caloric restriction improved survival — 91 percent alive at day 60 against 78 percent — with fewer respiratory infections rather than more. What that trial cannot settle is the question S.A. actually poses, because it enrolled patients after the fall in phosphate, and hers is 2.1 mg/dL, about 0.68 mmol/L, before a single calorie has gone in. She sits just above their entry threshold and has not yet been fed, which places her in the group the trial was designed to create rather than the group it studied.
The months of poor appetite before her surgery were never formally worked up — she'd attributed them herself to a stressful year at work, and her primary care visits in that window focused on unrelated complaints rather than the slow weight loss now visible in her hospital gown fitting loosely at the shoulders. That gap matters clinically as much as narratively: it means her true duration of inadequate intake, and therefore her true refeeding risk, is almost certainly underestimated by the six hospital days alone that show up cleanly in her chart.
Planning nutrition, post-operative day 3
I'd follow the conservative NICE-derived protocol here — start at 5 to 10 kcal/kg/day, correct her already-low phosphate and magnesium pre-emptively before any real caloric load lands. Her degree of malnutrition and prolonged poor intake aren't borderline; she's squarely in the population this protocol was built to protect.
I'd want to name the real cost of that approach honestly: I'd push on your warrant rather than your rate. Doig's trial — the only randomized evidence in this population — found protocolised caloric restriction improved 60-day survival, 91 percent against 78, with fewer respiratory infections. That is positive evidence for restriction, not merely an absence of harm, so I'm not arguing you're being over-cautious. I'm arguing you can't claim her. Doig enrolled patients whose phosphate had already dropped below 0.65 after feeding started; hers is 0.68 and nothing has started. Saying she's "squarely in the population this protocol was built to protect" runs a treatment trial's result as a prevention protocol, and that extrapolation should be named rather than folded into "NICE says so.
Her risk factors are real, but they're the same kind of risk factors the newer literature's own high-risk cohorts had, and that literature still didn't find a clear complication difference favoring the ultra-conservative rate.
Whichever caloric strategy you land on, thiamine goes in before any carbohydrate load, full stop — that's a distinct, cheap, easily missed step, and unrecognized thiamine depletion driving Wernicke's encephalopathy is preventable regardless of how this caloric-rate disagreement resolves. I'd rather that be explicit than assumed folded into either plan.
Agreed: thiamine given before feeding started, baseline electrolytes corrected to normal, and a restricted start at 10 kcal/kg/day with electrolyte panels checked every eight hours for the first 72 hours rather than daily. Her phosphate fell to 0.61 mmol/L on day two of feeding — below the threshold Doig's trial used to enrol — caught on the eight-hourly panel and corrected without any clinical event.
NOT agreed: whether her day-two phosphate fall vindicated the restricted start or merely showed she would have declared herself either way. The critical care physician read it as the protocol working as designed. The clinical pharmacologist noted that her crossing below 0.65 on day two is exactly the point at which she would have become eligible for the trial the team had been reasoning from, and that this makes the case for restriction stronger retrospectively than it was prospectively — which is not the same as having been right at the time.