Refeeding and nutritional rehabilitation

TL;DR — Nutritional rehabilitation is essential, but refeeding can precipitate electrolyte and fluid shifts, particularly hypophosphataemia. In the StRONG trial (n=120 enrolled, 111 in modified intention-to-treat, aged 12–24, at ≥60% median BMI; NCT02488109), starting at 2,000 kcal/day with 200-kcal daily increases restored medical stability sooner than 1,400 kcal/day with increases every other day (hazard ratio 1.67, 95% CI 1.10–2.53; p=0.01), shortened stay by 4.0 days (95% CI −6.1 to −1.9) and saved $19,056 per participant, with no excess of electrolyte abnormalities or other adverse events (Garber 2021, PMID 33074282). The trial's own 1-year follow-up found that this short-term advantage did not translate into higher clinical remission or lower rehospitalization (Golden 2021, PMID 33753542) — the faster protocol is better value, not better recovery. A secondary analysis further shows that the same kcal/day prescription systematically underfeeds patients with atypical AN because of their greater body weight (Garber 2024, PMID 38179719). Risk stratification, serial examination and electrolyte monitoring are inseparable from the calorie prescription.

Pathophysiology

After prolonged undernutrition, carbohydrate-driven insulin secretion shifts phosphate, potassium and magnesium intracellularly and promotes sodium/fluid retention. Thiamine demand rises. Clinical refeeding syndrome is more than a low phosphate value: it denotes organ dysfunction attributable to these shifts. Hypophosphataemia, hypokalaemia and hypomagnesaemia are the characteristic derangements of the refeeding period specifically, as distinct from the hyponatraemia, hypokalaemia and metabolic alkalosis associated with purging; a further trap is Pseudo-Bartter's syndrome, in which stopping purging produces oedema and rapid weight gain that can be misread as nutritional progress (Puckett 2023, PMID 36803805).

The empirical frequency of the syndrome itself is low in the modern monitored literature. Across 20 studies and 2,191 hospitalized children and adolescents published between 2010 and February 2023, only one study reported a true clinical case of refeeding syndrome, and the reviewers found no evidence that higher-calorie refeeding increased that risk, including at very low BMI (Mosuka 2023, PMID 37351245). The monitoring framework below exists to keep that frequency low, not because the syndrome is common under it.

Evidence table

Evidence	Population/design	Protocol contrast	Main inference
Garber 2021 (StRONG, short-term)	RCT, n=120 enrolled / 111 mITT, age 12–24, AN or atypical AN, ≥60% mBMI, 2 tertiary programs	2,000 kcal +200/day vs 1,400 +200 every other day	Medical stability restored earlier (HR 1.67, 95% CI 1.10–2.53; p=0.01); stay 4.0 days shorter; $19,056 saved; no excess adverse events (PMID 33074282)
Golden 2021 (StRONG, 1-year)	Same cohort, 111 mITT (60 higher / 51 lower), followed to 12 months post-discharge	Same	No group difference in clinical remission over 1 year (p=0.42); medical rehospitalization 32.8% (19/58) vs 35.4% (17/48), p=0.84; days rehospitalized 6.0 vs 5.1, p=0.81 (PMID 33753542)
Garber 2024 (StRONG, atypical AN)	Same cohort; 43% had atypical AN (%mBMI > 85)	Diagnosis and caloric dose in kcal/kg	Atypical AN had slower heart-rate restoration (8.7±4.0 vs 6.5±3.9 days, p=0.008), less weight gain (3.1±5.9 vs 5.4±2.9 %mBMI, p<0.001) and more hypomagnesaemia (29% vs 11%, OR 3.29) — explained by receiving only 32.4±6.9 vs 43.4±9.8 kcal/kg (PMID 38179719)
Garber 2016	27-study systematic review; 26 (96%) observational, hospital-based	Multiple settings/protocols	Low-calorie approaches too conservative in mild/moderate malnutrition; insufficient evidence to change standard of care in severe malnutrition; parenteral nutrition not recommended (PMID 26661289)
Mosuka 2023	Systematic review, 20 articles, 2,191 children/adolescents, 2010–Feb 2023	Higher vs lower calorie	Only 1 of 20 studies reported a true clinical case of refeeding syndrome; no evidence that higher-calorie refeeding increased refeeding-syndrome risk, including at very low BMI (PMID 37351245)
Norris 2016	Editorial commentary, not primary data	Empiric phosphate vs watch-and-wait	Frames the practice divergence; does not resolve it (PMID 27210006)
Bargiacchi 2019	Narrative review	Conservative vs higher-calorie care plans	Argues for the switch away from conservative refeeding and names the counter-risk — "underfeeding syndrome" and maintained weight suppression (PMID 30483963)
Le Grange 2013	Editorial	Current adolescent practice	Called for re-examination of conservative protocols before the randomized evidence existed (PMID 24138761)

What counts as refeeding syndrome

Much of the apparent disagreement in this literature is definitional. Until 2020 there was no agreed clinical definition, and incidence figures were therefore not comparable. The ASPEN consensus proposes that refeeding syndrome be diagnosed when serum phosphorus, potassium and/or magnesium fall — any one, two or three of them — within 5 days of reintroducing calories, graded as mild (10–20% decrease), moderate (20–30%) or severe (>30%, and/or organ dysfunction attributable to those falls or to thiamine deficiency) (da Silva 2020, PMID 32115791).

Applying that definition does not tidy the literature. A systematic review of 35 observational studies found reported refeeding-syndrome incidence ranging from 0% to 62%, and refeeding hypophosphataemia from 7% to 62%, with "no substantial change in the originally reported incidence" when ASPEN criteria were applied retrospectively. Incidence was highest in intensive-care inpatients and in those started above 20 kcal/kg/day (pooled 44%, 95% CI 36–52% for refeeding syndrome; 27%, 95% CI 21–34% for hypophosphataemia), and the reviewers state that because of heterogeneity "summary incidence measures are meaningless" (Cioffi 2021, PMID 34134001). Two numbers in this wiki — Mosuka's near-absence of clinical refeeding syndrome in monitored adolescents (PMID 37351245) and Cioffi's up-to-62% — are not in conflict; they measure different things in different populations under different definitions.

The severe end is real but rare. A retrospective survey of French intensive-care units identified only 68 AN admissions across 11 participating units over two years (mean BMI 12 ± 3 kg/m², 21 mechanically ventilated). Seven developed refeeding syndrome, and their day-one calorie intake was significantly higher than those who did not (23.2 ± 5 vs 14.1 ± 3 kcal/kg/day, p = 0.02). Seven patients died — a crude mortality of about 10% (Vignaud 2010, PMID 20920160). This is the population in which caution about initial rate is best supported, and it is precisely the population excluded from the randomized higher-calorie trials.

Electrolyte prophylaxis

Practice diverges between prophylactic phosphate/magnesium/potassium supplementation from the start of feeding and supplementation triggered by monitoring. A review of the available evidence found three studies in which prophylactic potassium, magnesium and/or phosphate supplementation — alongside routine thiamine and multivitamin — prevented refeeding syndrome or refeeding hypophosphataemia, but the refeeding methods, doses, durations and populations varied so much that no specific recommendation could be derived, and the authors call for randomized trials using comparable protocols (Gallagher 2022, PMID 34648201). Indirect support comes from the nasogastric literature: of six studies providing prophylactic phosphate, all reported refeeding hypophosphataemia in under 1% of patients (Rizzo 2019, PMID 30070730). Neither body of evidence is randomized, and the observed low rates are confounded with the monitoring intensity that accompanies protocolized feeding.

Route: nasogastric versus oral

A systematic review of ten studies (eight retrospective chart reviews, one prospective cohort, one randomized trial; nine inpatient) found nasogastric refeeding produced average weight gain above 1 kg/week in eight studies, and in four of five studies with an oral-only comparison, weekly weight gain and caloric intake were significantly higher in tube-fed patients. Nasogastric feeding was not associated with increased adverse outcomes and was judged safe and well tolerated — with the explicit caveat that study designs were weak and that standardized, evidence-based protocols do not yet exist (Rizzo 2019, PMID 30070730). The systematic review of refeeding approaches reaches the compatible conclusion that meal-based and combined nasogastric-plus-meal approaches can both deliver higher calories, while parenteral nutrition is not recommended (Garber 2016, PMID 26661289). Route is therefore a tolerability and delivery question, not an efficacy one — with the important qualification that the comparison is nowhere randomized at adequate scale.

What else StRONG measured

Three secondary analyses of the same randomized cohort extend the trial beyond calories and stability.

Analysis	Question	Finding
Accurso 2023, PMID 36919264	Does higher-calorie refeeding increase mealtime distress?	No. In 111 randomized participants (45 with momentary ratings), treatment assignment was unrelated to food refusal, mealtime distress or affective state. Food refusal rose over the course of refeeding in both arms (p = .018) regardless of allocation. Notably, 55% of participants had preferred the lower-calorie arm before randomization — a preference the data do not support
Downey 2022, PMID 35705423	How common is renal impairment at admission?	33% of 111 participants had baseline eGFR < 90 mL/min/1.73 m². More rapid weight loss and more severe bradycardia predicted low admission eGFR. eGFR improved during refeeding, but the rate of improvement did not differ by treatment arm (95% CI −1.61 to 0.15; p = .095)
Downey 2025, PMID 39945690	Does pre-admission energy balance predict refeeding needs?	In 82 participants (mean age 16.5 ± 2.5 y, 85 ± 10 %mBMI), mean acute energy balance before admission was −898 ± 678 kcal. More negative acute energy balance predicted more days (−0.18 days per unit, 95% CI −0.35 to −0.02, p = .03) and more kilocalories to restore medical stability, and was a more important predictor than weight suppression in dominance analysis

Taken together these point away from BMI as the organizing admission variable: rate of loss, bradycardia severity and recent energy deficit each carry information that current weight does not.

Monitoring framework

Phase	Measure	Why it matters
Before initiation	Vitals/orthostasis, hydration, ECG when indicated, phosphate, potassium, magnesium, glucose, renal/liver indices	Baseline risk and correctable abnormalities
Early refeeding	Serial electrolytes, fluid balance, oedema, cardiorespiratory/neurologic symptoms	Detect intracellular shifts and overload
Weight restoration	Standardized weights, intake completion, GI tolerance, activity	Distinguish prescribed from delivered energy
Transition	Relapse plan, outpatient monitoring, food access and support	Hospital weight gain is not durable recovery

What the target weight should be

Calorie prescription is only half the dose question; the other half is where refeeding is aimed. In 201 adolescents with restrictive eating disorders showing features of AN but spanning a wide BMI range at presentation, treated in a family-based programme with defined early interventions and followed for a year, recovery was 65% by EDE-Q criterion (<2.0) and 53% by clinical interview. Independent predictors of recovery were lower EDE-Q at presentation, higher weight gain after 3 months, and lower weight suppression at follow-up — weight suppression defined as the difference between premorbid and current BMI (Swenne 2017, PMID 28915806). The authors draw the operational conclusion directly: high weight suppression at follow-up predicts poor prognosis, so premorbid BMI has to be taken into account when setting weight targets. A patient restored to a population-normal BMI who remains far below their own premorbid trajectory has not been refed to target.

This is the same logic that makes fixed kcal/day prescribing wrong for atypical AN (Garber 2024, PMID 38179719), applied to the endpoint rather than the rate: both errors come from using a population reference where an individual one is needed.

Route escalation is partly a nursing question

Nasogastric feeding rates are not fixed properties of illness severity. In a before-and-after chart review of 21 hospitalized adolescents (mean age 15.1 ± 1.9), introducing structured meal support therapy reduced nasogastric feeding from 67% to 11% (p < .02), with no difference in length of stay, weight change or readmission (Couturier 2009, PMID 19548149). The sample is small and the design uncontrolled, but the direction matters for how tube-feeding rates should be read: a unit's nasogastric rate reflects its mealtime staffing model as much as its patients.

Calories are not the only variable

Protocols differ in oral meals, supplements, nasogastric use, macronutrient composition, phosphate/thiamine policy, fluid/sodium management and escalation thresholds. Trials of “higher” versus “lower” calories therefore compare care packages, not calories in isolation, and the review that assembled that evidence also concluded that meal-based and combined nasogastric-plus-meal approaches can both deliver higher calories, that nutrient compositions within recommended ranges are appropriate, and that parenteral nutrition is not recommended (Garber 2016, PMID 26661289).

The StRONG secondary analysis makes the point quantitatively. Prescribing by absolute kcal/day rather than by body weight produced a dose gradient across diagnoses, and dose — not diagnosis — carried the outcome: for every 10 kcal/kg increase, heart rate was restored 1.7 days faster (95% CI 1.0–2.5, p<0.001), weight gain was 1.6 %mBMI greater (0.8–2.4, p<0.001) and the odds of hypomagnesaemia fell 70% (12–128%, p=0.02) (Garber 2024, PMID 38179719). A "higher-calorie protocol" is therefore not a fixed thing; it is a dose that has to be indexed to body weight.

Atypical AN and risk

Garber's randomized population included atypical AN — 43% of the modified intention-to-treat sample — which is why it can speak to this group at all (Garber 2021, PMID 33074282; Garber 2024, PMID 38179719). The finding is not the reassuring one. Patients with atypical AN did worse on every refeeding outcome measured: slower heart-rate restoration, less weight gain, and roughly triple the odds of hypomagnesaemia. The mechanism was iatrogenic rather than biological — a fixed kcal/day start delivered about 25% less energy per kilogram to a heavier group — and the authors conclude directly that these patients need more calories, not fewer (Garber 2024, PMID 38179719). Weight-loss trajectory and recent intake matter independently of current BMI, and so does weight-indexed dosing.

Open questions

• What protocols optimize safety and time to recovery in extreme malnutrition? StRONG excluded patients below 60% median BMI, and the 2016 systematic review found the evidence insufficient to change the standard of care in severe malnutrition; as of September 2026 no randomized trial has enrolled patients below 60% median BMI (Garber 2016, PMID 26661289; Garber 2021, PMID 33074282).
• Which patients benefit from prophylactic phosphate rather than surveillance-triggered replacement? The question was framed editorially in 2016 and no randomized comparison of empiric versus monitoring-triggered phosphate replacement has been published as of September 2026 (Norris 2016, PMID 27210006).
• Do faster inpatient gains improve one-year remission and readmission, not only length of stay? Answered, and negatively: at 12 months post-discharge, remission trajectories did not differ (p=0.42), nor did rehospitalization rate (32.8% vs 35.4%), number of rehospitalizations or days rehospitalized (Golden 2021, PMID 33753542). The case for higher-calorie refeeding rests on earlier stability, shorter stay and cost — not on better one-year recovery.
• Does weight-indexed (kcal/kg) rather than fixed (kcal/day) prescribing eliminate the atypical-AN outcome gap? The dose–response relationship is established observationally within StRONG but has never been randomized (Garber 2024, PMID 38179719).
• Which definition of refeeding syndrome should incidence be reported against? Applying ASPEN criteria retrospectively did not reconcile a 0–62% range across 35 studies, and the reviewers judged pooled incidence meaningless (da Silva 2020, PMID 32115791; Cioffi 2021, PMID 34134001).
• Does prophylactic phosphate/magnesium/potassium supplementation outperform monitoring-triggered replacement? Three non-randomized studies favour prophylaxis and the nasogastric literature reports <1% hypophosphataemia under it, but no randomized comparison has been published as of September 2026 (Gallagher 2022, PMID 34648201; Rizzo 2019, PMID 30070730).
• Should admission triage use acute energy balance and rate of loss rather than BMI, given that pre-admission energy deficit outperformed weight suppression in predicting refeeding intensity (Downey 2025, PMID 39945690)?
• What is the correct starting rate in intensive care, where the only series available shows refeeding syndrome associated with a day-one intake of 23.2 vs 14.1 kcal/kg and ~10% crude mortality — a population that no randomized refeeding trial has enrolled (Vignaud 2010, PMID 20920160)?
• Should refeeding targets be set against premorbid BMI rather than a population reference, given that residual weight suppression at one year predicts poor outcome (Swenne 2017, PMID 28915806)?
• How much of the between-unit variation in nasogastric feeding rates is explained by mealtime support models rather than by patient severity (Couturier 2009, PMID 19548149)?

Related pages

• Medical complications — baseline organ and electrolyte risk.
• Service models and setting — where rehabilitation occurs.
• Red flags and safety concerns — escalation signals.

References

1. Garber AK, et al. Short-term outcomes of the Study of Refeeding to Optimize Inpatient Gains. JAMA Pediatr. 2021. PMID 33074282.
2. Garber AK, et al. A systematic review of approaches to refeeding in patients with anorexia nervosa. Int J Eat Disord. 2016. PMID 26661289.
3. Mosuka EM, et al. Clinical outcomes of refeeding syndrome: high- versus low-calorie diets. Cureus. 2023. PMID 37351245.
4. Norris ML, et al. Phosphate supplementation during refeeding of hospitalized adolescents with anorexia nervosa. J Adolesc Health. 2016. PMID 27210006.
5. Le Grange D, et al. Examining refeeding protocols for adolescents with anorexia nervosa. J Adolesc Health. 2013. PMID 24138761.
6. Bargiacchi A, et al. Refeeding in anorexia nervosa. Eur J Pediatr. 2019. PMID 30483963.
7. Puckett L, et al. Renal and electrolyte complications in eating disorders. J Eat Disord. 2023. PMID 36803805.
8. Golden NH, et al. Higher-calorie refeeding in anorexia nervosa: 1-year outcomes from a randomized controlled trial. Pediatrics. 2021;147:e2020037135. PMID 33753542.
9. Garber AK, et al. Short-term outcomes of the study of refeeding to optimize inpatient gains for patients with atypical anorexia nervosa. Int J Eat Disord. 2024;57:859-868. PMID 38179719.
10. da Silva JSV, et al. ASPEN consensus recommendations for refeeding syndrome. Nutr Clin Pract. 2020;35:178-195. PMID 32115791.
11. Cioffi I, et al. The incidence of the refeeding syndrome: a systematic review and meta-analyses of literature. Clin Nutr. 2021;40:3688-3701. PMID 34134001.
12. Vignaud M, et al. Refeeding syndrome influences outcome of anorexia nervosa patients in intensive care unit: an observational study. Crit Care. 2010;14:R172. PMID 20920160.
13. Gallagher D, et al. Prophylactic supplementation of phosphate, magnesium, and potassium for the prevention of refeeding syndrome in hospitalized individuals with anorexia nervosa. Nutr Clin Pract. 2022;37:328-343. PMID 34648201.
14. Rizzo SM, et al. Enteral nutrition via nasogastric tube for refeeding patients with anorexia nervosa: a systematic review. Nutr Clin Pract. 2019;34:359-370. PMID 30070730.
15. Accurso EC, et al. Hospital-based higher calorie refeeding and mealtime distress in adolescents and young adults with anorexia nervosa or atypical anorexia nervosa. Int J Eat Disord. 2023;56:1219-1227. PMID 36919264.
16. Downey AE, et al. Renal function in patients hospitalized with anorexia nervosa undergoing refeeding: findings from the Study of Refeeding to Optimize Inpatient Gains. J Adolesc Health. 2022;71:432-437. PMID 35705423.
17. Downey AE, et al. The effect of preadmission energy balance on short-term medical outcomes: findings from the Study of Refeeding to Optimize Inpatient Gains. J Adolesc Health. 2025;76:702-709. PMID 39945690.
18. Swenne I, et al. Family-based intervention in adolescent restrictive eating disorders: early treatment response and low weight suppression is associated with favourable one-year outcome. BMC Psychiatry. 2017;17:333. PMID 28915806.
19. Couturier J, Mahmood A. Meal support therapy reduces the use of nasogastric feeding for adolescents hospitalized with anorexia nervosa. Eat Disord. 2009;17:327-332. PMID 19548149.