Skip to content

Medical complications

TL;DR — AN can affect every organ system through energy deficit, weight loss, purging and refeeding. Bradycardia and endocrine adaptation may be expected consequences of malnutrition yet still coexist with acute danger; severity depends on the whole physiological picture. Bone deficits may persist after weight recovery: recovery improves many hormonal abnormalities but "deficits in bone accrual can persist", and transdermal oestrogen raises BMD in adolescents although "catch-up is incomplete" (Misra 2014, PMID 24731664). The randomized evidence for that oestrogen effect is Misra 2011, which showed increased spine and hip BMD Z-scores versus placebo over 18 months (PMID 21698665). The clinical endpoint that matters is now measured directly: in a matched cohort of 7,332 people hospitalized with AN and 73,215 controls followed up to 34 years, fracture hospitalization rates were 47.6 vs 21.0 per 10,000 person-years in women and 74.0 vs 39.9 in men, with an adjusted hazard ratio of 7.50 (95% CI 5.62–10.01) for osteoporotic fractures (Cyrenne-Dussault 2026, PMID 41109616). Atypical AN can produce important medical instability despite non-low BMI (Walsh 2023, PMID 36508318).

Complication matrix

System Findings Measurement priorities Evidence
Cardiovascular Sinus bradycardia, hypotension, orthostasis, reduced mass/output, arrhythmia risk Vitals, orthostasis, ECG, electrolytes Westmoreland 2016, PMID 26169883
Renal/electrolyte Hyponatraemia, hypokalaemia, metabolic alkalosis; acute kidney injury through to dialysis-requiring CKD; hypokalaemic nephropathy from chronic purging; Pseudo-Bartter's syndrome causing oedema and rapid weight gain on cessation of purging; refeeding hypophosphataemia, hypokalaemia, hypomagnesaemia Intake/output, creatinine, electrolytes, acid-base Puckett 2023, PMID 36803805
Endocrine Hypogonadotropic hypogonadism with oestrogen/androgen deficiency, growth-hormone resistance with low IGF-1, hypercortisolaemia, non-thyroidal illness syndrome, hyponatraemia, hypo-oxytocinaemia; leptin suppressed, ghrelin and (paradoxically) peptide YY elevated Growth/puberty, menstrual history, targeted hormones Misra 2014, PMID 24731664; Schorr 2017, PMID 27811940
Bone Low BMD, impaired microarchitecture, strength and accrual; clinical fractures DXA when indicated; fracture history Misra 2014, PMID 24731664; Cyrenne-Dussault 2026, PMID 41109616
Gastrointestinal Delayed gastric emptying, constipation, pain, bloating, liver/pancreatic abnormalities Clinical evaluation; avoid reflexively attributing all symptoms Westmoreland 2016, PMID 26169883
Hematologic/immune Cytopenias, altered infection presentation Full blood count and clinical context Baenas 2024, PMID 37598049
Neurologic/muscular Weakness, cognitive slowing, neuropathy in severe deficiency Examination and nutrition history Westmoreland 2016, PMID 26169883

Cardiovascular interpretation

Sinus bradycardia can reflect adaptation to low metabolic demand, but profound bradycardia, hypotension, syncope, conduction abnormalities, electrolyte disturbance or rapid deterioration changes risk. QT interpretation must account for correction formula, heart rate, drugs and electrolytes. The research literature does not support a single BMI that safely excludes cardiovascular compromise (Society for Adolescent Health and Medicine 2022, PMID 36058805).

Endocrine adaptation and fertility

Endocrine changes include hypothalamic amenorrhoea, acquired growth-hormone resistance with low IGF-1, relative hypercortisolaemia, low leptin/insulin/amylin/incretins and increased ghrelin, peptide YY and adiponectin (Misra 2014, PMID 24731664). A parallel review adds non-thyroidal illness syndrome, hyponatraemia and hypo-oxytocinaemia to that list and stresses that most — though explicitly not all — of these disturbances are adaptive to chronic starvation and reverse with treatment, while several contribute directly to impaired skeletal integrity and to neuropsychiatric comorbidity (Schorr 2017, PMID 27811940). Many improve with nutritional recovery, but bone accrual deficits can persist (Misra 2014, PMID 24731664). Amenorrhoea is neither required for diagnosis nor a reliable severity gate.

The endocrine literature is also sex-skewed in a way that constrains what can be said: 5–15% of patients with AN are men, but only limited data exist on the endocrine impact of the illness in adolescent boys and men (Schorr 2017, PMID 27811940).

Bone evidence

Intervention/evidence Population Finding Boundary
Weight/nutritional recovery Adolescents/adults Foundational; improves endocrine environment Catch-up may remain incomplete
Physiologic transdermal estrogen Adolescent girls Increased BMD versus placebo Does not replace nutrition; population-specific (Misra 2011, PMID 21698665)
Transdermal estradiol ± rhIGF-1 75 women aged 14–22 randomized, 33 completed; 12 months Negative for the addition: lumbar areal BMD increased more on oestrogen + placebo than on oestrogen + rhIGF-1 (p=0.004); no group difference in geometry, microarchitecture, volumetric BMD or strength Adding rhIGF-1 conferred no benefit; oestrogen adherence was variable (Singhal 2021, PMID 33693703)
Teriparatide 21 women with AN and T-score ≤ −2.5, randomized 10:11; 6 months No change in bone microarchitecture; non-significant trends in intertrochanteric cortical thickness (+13%, p=0.075) and buckling ratio (−9%, p=0.076); osteocalcin +216% (p=0.01) Very small, short, trend-level — not a treatment recommendation (Amorim 2026, PMID 41591404)
Pharmacologic systematic review 19 studies, 1,119 participants, 10 double-blind RCTs Bisphosphonates increased BMD in adult women; transdermal oestrogen increased BMD in mature adolescents; oral contraceptives did not increase BMD in randomized or controlled trials All conclusions are BMD-based; the review's own stated aim — which treatments reduce lifetime fracture risk — remained unanswered (Robinson 2017, PMID 28554377)

No approved therapy for bone loss in AN exists (Amorim 2026, PMID 41591404). The evidence chain that would justify one is now partly assembled: the fracture excess is established at population scale (Cyrenne-Dussault 2026, PMID 41109616) and several agents move BMD (Robinson 2017, PMID 28554377), but as of September 2026 no trial in anorexia nervosa has used incident fracture as a primary endpoint — every randomized bone study located in this audit reports densitometric or microarchitectural surrogates in samples of 21 to 110 participants.

What actually occurs at extreme malnutrition

The complication matrix above lists what can happen. A single-centre series of 354 consecutive adults admitted to a French clinical-nutrition eating-disorder referral unit between 1997 and 2014 (mean age 28.7 ± 10.7 years, mean illness duration 9.5 ± 9 years, admission BMI 12.2 ± 1.6 kg/m², 79.3% previously hospitalized elsewhere) quantifies what does:

Complication during hospitalization Prevalence
Anaemia 79%
Neutropenia 53.9%
Hypertransaminasaemia 53.7%
Osteoporosis 46.3%
Hypokalaemia 39.5%
Hypophosphataemia 26%
Infectious complications 24.3%
Hypoglycaemia 13.8%
Cardiac dysfunction 7.1%
Biopsy-proven gelatinous bone-marrow transformation 6.5%

Ten per cent were transferred to intensive care and five patients died; enteral nutrition was used in 85.9%; mean weight gain was 4.1 ± 3.9 kg over 36.9 ± 30.5 days (Guinhut 2021, PMID 33023762). Subtype patterned the complications: hypokalaemia was more frequent in the binge-eating/purging subtype, while lympho-neutropenia and hypertransaminasaemia were more frequent in the restricting subtype. These are referral-centre figures for the most severely malnourished tail of the illness and should not be read as prevalences in general AN populations — but they are the best available answer to "what does severe AN look like on the ward".

Cardiovascular: what the data do and do not support

Sudden cardiac death is repeatedly named as a mechanism of the excess mortality in AN. The systematic review that examined this across structural, repolarization/conduction, haemodynamic and peripheral-vascular domains reached a blunt conclusion: "despite the prevalent theory that malignant arrhythmias are implicated as a cause of sudden death in this disorder, data to support this causal relationship are lacking" (Sachs 2016, PMID 26710932). Its practical recommendations follow from that — routine ECG and orthostatic vitals in everyone presenting with AN, echocardiography only when clinical signs prompt it, and telemetry admission for severe sinus bradycardia or junctional rhythm, markedly prolonged QTc, or syncope.

QTc in particular is weaker as a severity marker than its clinical prominence suggests. In a hospitalized cohort of very-low-BMI patients (n=19, mean BMI 12.3 kg/m², 95% female) with digital ECGs measured by a single blinded electrophysiologist and Fridericia rather than Bazett correction, 4 patients (21%) had QTc prolongation and 2 (10.5%) exceeded 500 ms — each of those two with concomitant contributors, and 68% of the cohort taking QT-prolonging drugs. QTc was not significantly correlated with left-ventricular mass, LV mass index, BMI or resting energy expenditure (Krantz 2012, PMID 21917317). Rate-correction formula matters at the extremes of heart rate seen in AN, and Bazett over-corrects at bradycardic rates; a prolonged QTc in AN is more often a drug-and-electrolyte finding than a malnutrition-severity finding.

Structural abnormalities are common and mostly reversible. In 48 consecutive adults with AN versus 44 matched controls, pericardial effusion was present in 9/48 versus 0/44 (p = 0.003), left-ventricular mass was 63 ± 15 g versus 99 ± 30 g (p < 0.001), and global longitudinal strain was reduced (−18.9 ± 2.8% vs −20.2 ± 1.8%, p = 0.010); on follow-up 11 of 13 patients with the abnormal basal strain pattern recovered completely (Scheggi 2022, PMID 35597626). In 38 children admitted for physical instability, 63% met a composite definition of "AN cardiopathy" and 26% had pericardial effusion, bradycardia and mitral regurgitation together; global longitudinal strain recovered at 3–8 months, but end-diastolic diameter, E/A ratio and LV sphericity index remained impaired (Borgia 2021, PMID 34050378). The reversibility claim therefore holds for deformation but not for every geometric measure, and the follow-up is months rather than years.

Hepatic injury and hypoglycaemia: the terminal-malnutrition picture

Transaminase elevation in AN has two distinct causes with opposite management implications. Starvation hepatitis reflects hepatocyte injury and death and becomes more common as BMI falls; refeeding can separately raise enzymes through hepatic steatosis, distinguishable by fatty liver on ultrasound. Further laboratory testing, imaging or biopsy has low diagnostic yield and is not generally recommended, and starvation hepatitis usually normalizes with supervised caloric increase and weight restoration. Patients with starvation hepatitis are at increased risk of hypoglycaemia because glycogen is depleted and gluconeogenesis impaired (Rosen 2017, PMID 28932925).

Two series put numbers on that risk. Among 48 patients admitted at BMI < 13 kg/m² (median 10.51), 32 experienced blood glucose < 55 mg/dL (< 3.05 mmol/L); liver volume measured on CT was significantly reduced around hypoglycaemic episodes, hypoglycaemia was accompanied by low triglycerides and liver dysfunction, and every patient with a poor prognosis was in the hypoglycaemic group. The authors argue this is a distinct entity from refeeding syndrome — the liver being consumed as a last energy source — and that it marks terminal-stage malnutrition, so under-feeding is the danger (Matsunaga 2024, PMID 38702806). In a separate cohort of 34 patients at BMI < 14.5 kg/m² managed on an exclusively meal-based rapid-weight-gain protocol with frequent point-of-care glucose testing, hypoglycaemia was detected in 50%, was highest on the day of admission, was generally asymptomatic, and was severe (< 50 mg/dL) in 20.6%; lower admission BMI predicted both occurrence and duration (Fischer 2022, PMID 35994205). Asymptomatic hypoglycaemia at admission is thus common enough that it will be missed without deliberate testing.

Gastrointestinal and haematologic specifics

Superior mesenteric artery syndrome — duodenal compression between the aorta and the SMA as the mesenteric fat pad is lost — is a rare but potentially life-threatening cause of abdominal pain in severe malnutrition, and can precipitate acute gastric dilatation during refeeding; it is diagnosed on upper GI series or abdominal CT and resolves with weight restoration (Mascolo 2015, PMID 25639251). It is the reason new severe abdominal pain during refeeding is not assumed to be functional.

Gelatinous transformation of the bone marrow — serous atrophy of marrow fat with extracellular mucopolysaccharide deposition — was biopsy-proven in 6.5% of the extreme-malnutrition series above (Guinhut 2021, PMID 33023762) and is reversible with nutritional restoration. A comprehensive review argues against attributing it to nutrient deficiency alone, proposing that malignancy, infection and inflammatory conditions also contribute — which matters clinically, because finding it in a patient with AN does not close the differential (Gunaratne 2024, PMID 39060221).

Pregnancy and neonatal outcomes

Two large population-based registry studies agree on direction and roughly on magnitude.

Outcome Swedish Medical Birth Register, 2,769 women with AN vs 1,225,321 without (Mantel 2020, PMID 31746972) Quebec, 2,134,945 pregnancies, AN requiring hospitalization (Ante 2020, PMID 32100355)
Preterm birth RR 1.6 (95% CI 1.4–1.8) RR 1.32 (95% CI 1.13–1.55)
Stillbirth not reported RR 1.99 (95% CI 1.20–3.30)
Low birth weight — RR 1.69 (95% CI 1.44–1.99)
Small for gestational age — RR 1.52 (95% CI 1.35–1.72)
Microcephaly RR 1.9 (95% CI 1.5–2.4) —
Hyperemesis RR 2.1 (95% CI 1.8–2.5) —
Anaemia (active AN) RR 2.1 (95% CI 1.3–3.2) —
Antepartum haemorrhage RR 1.6 (95% CI 1.2–2.1), stronger in active disease —

Risk was elevated for women with previous as well as active eating disorders in the Swedish cohort, and the Quebec cohort found low birth weight and small-for-gestational-age associations most prominent when hospitalization occurred during pregnancy or within two years of delivery — so recency of illness modifies but does not abolish the association. Neither study can separate the effect of the disorder from that of the treatments, comorbidities and socioeconomic circumstances that accompany it.

Atypical AN

Comparative evidence shows many of the same physiological complications in atypical and typical AN, although frequency differs for some outcomes (Walsh 2023, PMID 36508318). The 2026 update across 64 publications names the specific pattern: menstrual disturbance and reduced bone mineral density occur in atypical AN as in AN, but less frequently (Lee 2026, PMID 42557659). Rate and magnitude of weight loss, intake, purging and current physiology must accompany BMI.

Open questions

  • Which cardiac markers add prognostic value beyond direct instability measures?
  • Which bone strategy reduces fractures, not only DXA endpoints? The fracture excess is quantified (adjusted HR 7.50 for osteoporotic fracture) and BMD-active agents exist, but no fracture-endpoint trial has been run (Cyrenne-Dussault 2026, PMID 41109616; Robinson 2017, PMID 28554377).
  • Does the male fracture excess (HR 1.86, 95% CI 1.34–2.58) reflect the same bone biology studied almost entirely in girls and women (Cyrenne-Dussault 2026, PMID 41109616; Schorr 2017, PMID 27811940)?
  • What complication risks are independently predicted by weight-loss rate in atypical AN (Walsh 2023, PMID 36508318)?
  • Is sudden death in AN arrhythmic? The mechanism is widely assumed and the supporting causal data are explicitly described as lacking, with no autopsy or monitored-death series adequate to settle it (Sachs 2016, PMID 26710932).
  • Does QTc add anything to bedside risk assessment once drugs, electrolytes and rate-correction formula are accounted for (Krantz 2012, PMID 21917317)?
  • Which cardiac geometry changes fail to reverse, and do the residual abnormalities seen at 3–8 months persist or matter over years (Borgia 2021, PMID 34050378; Scheggi 2022, PMID 35597626)?
  • Should glucose be monitored systematically at admission in low-BMI AN, given that hypoglycaemia is present in 50% and usually asymptomatic (Fischer 2022, PMID 35994205; Matsunaga 2024, PMID 38702806)?
  • Are the adverse pregnancy outcomes attributable to current nutritional state, to lasting physiological consequences of past illness, or to confounding — the associations persist in women with previous rather than active disease (Mantel 2020, PMID 31746972; Ante 2020, PMID 32100355)?

References

  1. Westmoreland P, et al. Medical complications of anorexia nervosa and bulimia. Am J Med. 2016. PMID 26169883.
  2. Puckett L, et al. Renal and electrolyte complications in eating disorders. J Eat Disord. 2023. PMID 36803805.
  3. Misra M, Klibanski A. Endocrine consequences of anorexia nervosa. Lancet Diabetes Endocrinol. 2014. PMID 24731664.
  4. Schorr M, Miller KK. The endocrine manifestations of anorexia nervosa. Nat Rev Endocrinol. 2017. PMID 27811940.
  5. Misra M, et al. Physiologic estrogen replacement increases bone density in adolescent girls with anorexia nervosa. J Bone Miner Res. 2011. PMID 21698665.
  6. Singhal V, et al. Effect of transdermal estradiol and IGF-1 on bone endpoints of young women with anorexia nervosa. J Clin Endocrinol Metab. 2021. PMID 33693703.
  7. Robinson L, et al. Pharmacological treatment options for low BMD and secondary osteoporosis in anorexia nervosa. J Psychosom Res. 2017. PMID 28554377.
  8. Walsh BT, et al. A systematic review comparing atypical anorexia nervosa and anorexia nervosa. Int J Eat Disord. 2023. PMID 36508318.
  9. Society for Adolescent Health and Medicine. Medical management of restrictive eating disorders. J Adolesc Health. 2022. PMID 36058805.
  10. Baenas I, et al. Medical complications in anorexia and bulimia nervosa. Med Clin (Barc). 2024. PMID 37598049.
  11. Cyrenne-Dussault M, et al. Anorexia nervosa and risk of fractures: a matched cohort study of 80,000 men and women. Am J Med. 2026;139:333-340. PMID 41109616.
  12. Amorim T, et al. Effects of teriparatide on hip structure and bone microarchitecture in women with anorexia nervosa: a placebo-controlled randomized trial. Osteoporos Int. 2026;37:749-758. PMID 41591404.
  13. Lee V, Hagan KE. An invited updated systematic review and meta-analysis comparing atypical anorexia nervosa and anorexia nervosa. Int J Eat Disord. 2026. PMID 42557659.
  14. Guinhut M, et al. Extremely severe anorexia nervosa: hospital course of 354 adult patients in a clinical nutrition-eating disorders unit. Clin Nutr. 2021;40:1954-1965. PMID 33023762.
  15. Sachs KV, et al. Cardiovascular complications of anorexia nervosa: a systematic review. Int J Eat Disord. 2016;49:238-248. PMID 26710932.
  16. Krantz MJ, et al. Factors influencing QT prolongation in patients hospitalized with severe anorexia nervosa. Gen Hosp Psychiatry. 2012;34:173-177. PMID 21917317.
  17. Scheggi V, et al. Echocardiographic abnormalities in adults with anorexia nervosa. Am J Cardiol. 2022;175:152-157. PMID 35597626.
  18. Borgia F, et al. Anorexia nervosa-related cardiopathy in children with physical instability. Eur J Pediatr. 2021;180:3379-3389. PMID 34050378.
  19. Rosen E, et al. Hepatic complications of anorexia nervosa. Dig Dis Sci. 2017;62:2977-2981. PMID 28932925.
  20. Matsunaga H, et al. Severe hypoglycemia with reduced liver volume as an indicator of end-stage malnutrition in patients with anorexia nervosa. J Eat Disord. 2024;12:55. PMID 38702806.
  21. Fischer LK, et al. Monitoring and treating hypoglycemia during meal-based rapid nutritional rehabilitation in patients with extreme anorexia nervosa. Eat Weight Disord. 2022;27:3301-3308. PMID 35994205.
  22. Mascolo M, et al. Severe gastric dilatation due to superior mesenteric artery syndrome in anorexia nervosa. Int J Eat Disord. 2015;48:532-534. Case report. PMID 25639251.
  23. Gunaratne MDSK, et al. A comprehensive review on gelatinous transformation of the bone marrow. Expert Rev Hematol. 2024;17:547-554. PMID 39060221.
  24. Mantel Ä, et al. Association of maternal eating disorders with pregnancy and neonatal outcomes. JAMA Psychiatry. 2020;77:285-293. PMID 31746972.
  25. Ante Z, et al. Pregnancy outcomes in women with anorexia nervosa. Int J Eat Disord. 2020;53:403-412. PMID 32100355.