Skip to content

Red flags and safety concerns

TL;DR — Acute danger cannot be ruled out by normal-range BMI, apparent calm, or a single normal laboratory panel. Escalation is driven by physiological compromise, rapid trajectory, electrolyte/glucose disturbance, organ dysfunction, syncope, inability to maintain intake, purging, acute suicide risk and the monitoring capacity of the setting. Refeeding adds a separate early risk window. This page summarizes research and guideline signals, not individualized emergency advice.

Immediate concern domains

Domain Concerning finding Why it matters
Circulation Marked/worsening bradycardia, hypotension, orthostasis, syncope Low output, falls, arrhythmia/instability
Temperature Hypothermia Severe energy deficit/physiological compromise
Electrolytes Potassium, phosphate or magnesium abnormality Arrhythmia, weakness, organ dysfunction, refeeding risk
Glucose Hypoglycaemia Neurocognitive injury and acute deterioration
Hydration/renal Severe dehydration, oliguria, renal dysfunction Shock/electrolyte concentration and clearance
Cardiac electrical Prolonged QT or arrhythmia Medication/electrolyte interaction and sudden harm
Nutrition Near-total intake refusal or rapid continuing loss Imminent deterioration
Psychiatric Acute suicidal intent, severe self-harm risk, inability to maintain safety Independent mortality pathway
Refeeding Oedema, dyspnoea, weakness, confusion, falling phosphate after nutrition starts Possible refeeding syndrome

The Society for Adolescent Health and Medicine position paper emphasizes physiological and contextual admission criteria across restrictive eating disorders (PMID 36058805). NICE NG69 recommendation 1.10.3 advises acute medical care, including emergency admission, for people with an eating disorder who have severe electrolyte imbalance, severe malnutrition, severe dehydration or signs of incipient organ failure; recommendation 1.11.2 states that no absolute weight or BMI threshold should be used when deciding on day-patient or inpatient admission, and 1.2.8 rules out single measures such as BMI or illness duration as gates on treatment altogether (NICE NG69, recommendation text retrieved and read 2026-09-02).

Atypical AN safety failure

The systematic review found that atypical AN carries many physiological complications seen in AN despite absence of the low-weight criterion, at lower frequency for some outcomes (Walsh 2023, PMID 36508318; Lee 2026, PMID 42557659). Randomized inpatient data make the point more sharply than any cross-sectional comparison can: among 111 hospitalized adolescents and young adults, the 43% with atypical AN took longer to restore a normal heart rate (8.7±4.0 vs 6.5±3.9 days, p=0.008), gained less weight (3.1±5.9 vs 5.4±2.9 %mBMI, p<0.001), and had roughly triple the odds of hypomagnesaemia (29% vs 11%, OR 3.29) — because a fixed kcal/day prescription delivered them about 25% less energy per kilogram (Garber 2024, PMID 38179719). Instability in atypical AN is not merely present; in a monitored setting it resolved more slowly. Historical maximum, growth curve, magnitude/rate of loss, recent intake and measured physiology belong in risk assessment, and so does weight-indexed nutritional dosing.

Refeeding window

Risk does not end when nutrition begins. Phosphate, potassium, magnesium, fluid balance, thiamine status and cardiorespiratory/neurologic symptoms require structured monitoring according to baseline risk and protocol. Higher-calorie evidence came from monitored programs and cannot be translated into unsupervised rapid intake (Garber 2021, PMID 33074282; Garber 2016, PMID 26661289).

Medication and interaction risks

Context Safety issue
Bradycardia/QT Additive QT-prolonging drugs and correction-formula interpretation
Electrolyte disturbance Arrhythmia and seizure risks can change rapidly
Dehydration/renal or hepatic dysfunction Altered exposure and adverse effects
Sedating/orthostatic drugs Falls, impaired monitoring, hypotension
Purging/laxative/diuretic use Hidden potassium and volume abnormalities

Suicide risk

One in five deaths in the 2011 AN mortality meta-analysis was suicide (Arcelus 2011, PMID 21727255), and two independent 2026 syntheses reproduce that proportion and quantify the relative risk: a suicide mortality risk ratio of 9.86 (95% CI 5.63–17.27) for AN versus the general population across 83 studies, and suicide accounting for 21% of deaths across 30 AN cohorts (Semchishen 2026, PMID 41536100; Lai 2026, PMID 41277145). The same AN-specific meta-analysis identifies cardiac causes as a further 19% of deaths; these cause proportions do not establish that the underlying risks or prevention strategies are equivalent (Lai 2026, PMID 41277145).

Two risk modifiers are now quantified rather than assumed. All-cause mortality rises as mean BMI falls, and male sex plus comorbid psychiatric, substance-use, alcohol-use, mood or personality disorders are each associated with higher mortality (Semchishen 2026, PMID 41536100). Involuntary treatment marks a group at especially high risk — adjusted suicide hazard ratio 5.30 (95% CI 2.07–13.54) versus those never treated involuntarily — which identifies a population needing sustained follow-up rather than an effect of compulsion itself (Bager 2026, PMID 42383339). Risk assessment must be direct and repeated; neither weight restoration nor denial of intent alone establishes safety. Trial and service outcomes should separate suicide deaths from medical mortality.

Two organ-system reviews anchor the domains above: no body system is spared as AN becomes more severe and chronic (Westmoreland 2016, PMID 26169883), and the renal/electrolyte picture specifically includes hyponatraemia, hypokalaemia and metabolic alkalosis, hypokalaemic nephropathy progressing to dialysis-requiring chronic kidney disease after prolonged purging, and Pseudo-Bartter's syndrome — oedema and rapid weight gain on stopping purging, which can be misread as treatment progress (Puckett 2023, PMID 36803805).

Neurological emergencies that are missed

Wernicke's encephalopathy. Thiamine deficiency in AN is uncommon but under-recognized, and the pattern of recognition is itself the warning. In a systematic review of 12 published cases of Wernicke's encephalopathy following AN, 8 of 12 presented with the full triad of mental status change, ocular signs and ataxia — against roughly 16% presenting with the full triad in alcohol-related Wernicke's. Many also had additional features (vertigo, diplopia, consequences of refeeding syndrome). The authors read the excess of full-triad presentations as evidence of late recognition, since the complete triad is a late finding, and identify rapid weight loss as the vulnerability marker; they recommend prophylactic thiamine checks and treatment, with parenteral thiamine where Wernicke's is suspected (Oudman 2018, PMID 29984541). The confound they name is that thiamine-deficiency symptoms overlap with the encephalopathy they precede.

Seizures. In a retrospective study of 1,664 eating-disorder charts, 75 patients (4.5%) had documented seizures. The causes were mostly not what a malnutrition-first model would predict: psychogenic non-epileptic seizures accounted for 29.3%, substance abuse or withdrawal 18.7%, primary seizure disorder 12%, electrolyte abnormality or hypoglycaemia 10.7%, and presumed Wernicke's encephalopathy 4% (Gibson 2023, PMID 37092766). A seizure in AN should therefore trigger a full differential — including psychogenic non-epileptic seizures and substance withdrawal — rather than being attributed to electrolytes by default.

Exercise as a risk variable

Maladaptive exercise is not a peripheral feature. In 31,671 people with a lifetime eating disorder recruited across the United States, Australia, New Zealand and Sweden (Eating Disorders Genetics Initiative, NCT04378101), pooled lifetime prevalence of maladaptive exercise was 88%; current prevalence was 40% for any driven exercise, 35% for compulsive exercise and 12% for regular driven exercise, highest in AN. Compulsive Exercise Test phenotypes correlated most strongly with eating-disorder symptoms, while EDE-Q "regular driven exercise" was most strongly associated with BMI < 18.5, and risk was higher in younger birth cohorts. Prevalence and severity varied substantially by which instrument was used, and the authors recommend assessing both current and lifetime maladaptive exercise in every patient (Watson 2026, PMID 41115789). For risk assessment this matters twice: exercise adds energy expenditure to an existing deficit, and it is systematically under-captured unless asked about with a specific instrument.

Eating disorder with type 1 diabetes

Insulin restriction as a weight-control behaviour carries a mortality signal larger than most in this page. In an 11-year follow-up of 234 women with type 1 diabetes (60% of the original cohort; mean age 45, mean diabetes duration 28 years), 71 (30%) reported insulin restriction at baseline and 26 died during follow-up. Insulin restriction conveyed a threefold increased risk of mortality in multivariate Cox regression adjusted for baseline age, BMI and HbA1c, and mean age at death was 45 versus 58 years (p < 0.01). Restrictors reported more nephropathy and foot problems; women who died had reported more frequent insulin restriction and more eating-disorder symptoms at baseline (Goebel-Fabbri 2008, PMID 18070998). The behaviour is invisible on a weight chart and on a food diary, and the authors propose a routine screening question in diabetes care.

Recognition at the front door

Undetected instability can surface in emergency departments, and paediatric eating-disorder presentations rose during the pandemic (see epidemiology). Structured emergency-department approaches to these presentations — history, examination findings, investigations and admission decision — have been published as practical guidance rather than as evidence syntheses (Rothwell 2023, PMID 35790339). A live PubMed re-search on 2026-09-02 found case reports and recognition guidance but no study quantifying the miss rate for eating-disorder-related instability.

Two under-used bedside signals

Palm temperature. Bradycardia in outpatients is easy to miss in short appointments. Continuous 72-hour heart-rate monitoring with resting energy expenditure and temperature measured at 12 body sites in 58 patients with AN and 29 with bulimia nervosa found that palm temperature explained 18% of the variance in bradycardia (p < 0.001) — capturing it more efficiently than BMI did (Duriez 2022, PMID 35037343). The authors propose palm temperature as a warning sign for bradycardia; it has not been prospectively validated as a triage measure, and 18% of variance is a signal rather than a test.

Oral findings. A systematic review and meta-analysis of 33 articles found that people with eating disorders had higher dental caries and gingival inflammation severity and higher prevalence of dental erosion, myofascial pain, malocclusion, oral lesions and dentine sensitivity than controls, along with lower salivary flow and pH; bruxism did not differ and oral hygiene and periodontal findings were inconclusive. Certainty of evidence was graded very low throughout (Chan 2024, PMID 39433151). The reviewers note that dental professionals may detect signs during routine appointments and may provide an early referral route.

Missed alternative or coexisting disease

Persistent focal symptoms, inflammatory features, neurological signs, unexpected laboratory patterns or a course discordant with the eating-disorder formulation warrant re-evaluation. AN can coexist with gastrointestinal, endocrine or other disease; diagnostic closure is unsafe (Harrington 2015, PMID 25591200).

Open questions

  • Which multivariable rule predicts deterioration better than consensus thresholds?
  • How should rapid loss modify risk at non-low BMI (Walsh 2023, PMID 36508318)?
  • Can standardized post-discharge monitoring reduce early readmission and death?
  • How often is eating-disorder-related medical instability missed at first emergency presentation? Practical recognition guidance exists but no study located in this pass quantifies the miss rate (Rothwell 2023, PMID 35790339).
  • Should thiamine be given prophylactically to everyone refed at low BMI? The case literature shows late recognition — 8 of 12 published AN cases presented with the full Wernicke triad — but the evidence base is 12 case reports (Oudman 2018, PMID 29984541).
  • Which seizure in AN is metabolic? Fewer than 11% of documented seizures in a 1,664-chart eating-disorder cohort were attributable to electrolyte disturbance or hypoglycaemia (Gibson 2023, PMID 37092766).
  • Should insulin restriction be screened for routinely in type 1 diabetes, given a threefold adjusted mortality risk and a 13-year lower mean age at death (Goebel-Fabbri 2008, PMID 18070998)?
  • Which measure of maladaptive exercise should be used for risk assessment, given that prevalence estimates vary from 12% to 40% across instruments in the same 31,671-person sample (Watson 2026, PMID 41115789)?
  • Could palm temperature serve as an outpatient triage signal for bradycardia, given that it outperformed BMI in explaining bradycardia variance but has never been prospectively validated (Duriez 2022, PMID 35037343)?
  • Should dental services be a formal referral route? Oral findings differ from controls across many measures, but at very low certainty of evidence and with no guideline pathway (Chan 2024, PMID 39433151).

References

  1. Society for Adolescent Health and Medicine. Medical management of restrictive eating disorders in adolescents and young adults. J Adolesc Health. 2022. PMID 36058805.
  2. Walsh BT, et al. A systematic review comparing atypical anorexia nervosa and anorexia nervosa. Int J Eat Disord. 2023. PMID 36508318.
  3. Garber AK, et al. Short-term outcomes of higher- versus lower-calorie refeeding. JAMA Pediatr. 2021. PMID 33074282.
  4. Garber AK, et al. A systematic review of approaches to refeeding. Int J Eat Disord. 2016. PMID 26661289.
  5. Puckett L, et al. Renal and electrolyte complications in eating disorders. J Eat Disord. 2023. PMID 36803805.
  6. Westmoreland P, et al. Medical complications of anorexia nervosa and bulimia. Am J Med. 2016. PMID 26169883.
  7. Arcelus J, et al. Mortality rates in anorexia nervosa and other eating disorders. Arch Gen Psychiatry. 2011. PMID 21727255.
  8. Harrington BC, et al. Initial evaluation, diagnosis, and treatment of anorexia nervosa and bulimia nervosa. Am Fam Physician. 2015. PMID 25591200.
  9. Semchishen SN, et al. All-cause and cause-specific mortality risk in individuals with eating disorders. World Psychiatry. 2026;25:125-140. PMID 41536100.
  10. Lai ET, et al. Systematic review and meta-analysis of mortality in patients with anorexia nervosa. Int J Eat Disord. 2026;59:424-449. PMID 41277145.
  11. Bager L, et al. Excess mortality among patients with anorexia nervosa treated involuntarily. Int J Eat Disord. 2026. PMID 42383339.
  12. 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.
  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. Oudman E, et al. Preventing Wernicke's encephalopathy in anorexia nervosa: a systematic review. Psychiatry Clin Neurosci. 2018;72:774-779. PMID 29984541.
  15. Gibson D, et al. Seizures in eating disorders. Int J Eat Disord. 2023;56:1650-1660. PMID 37092766.
  16. Watson HJ, et al. Maladaptive exercise in people with a lifetime history of eating disorders: a multicountry observational study. Int J Eat Disord. 2026;59:190-204. PMID 41115789. (NCT04378101)
  17. Goebel-Fabbri AE, et al. Insulin restriction and associated morbidity and mortality in women with type 1 diabetes. Diabetes Care. 2008;31:415-419. PMID 18070998.
  18. Rothwell L, et al. Fifteen-minute consultation: recognition and management of eating disorders presenting to the emergency department. Arch Dis Child Educ Pract Ed. 2023;108:330-334. PMID 35790339.
  19. Duriez P, et al. Clinical meaning of body temperatures in anorexia nervosa and bulimia nervosa. Eur Eat Disord Rev. 2022;30:124-134. PMID 35037343.
  20. Chan CCK, et al. Oral health status of individuals with eating disorders: systematic review and meta-analysis. J Dent. 2024;151:105427. PMID 39433151.