What we’re looking at
Eating disorders (EDs)—anorexia nervosa (AN), bulimia nervosa (BN), binge eating disorder (BED), and related conditions—affect millions worldwide, with rates rising sharply after the COVID-19 pandemic [PMID 42083839]. These are not lifestyle choices or phases; they are serious psychiatric illnesses with medical, psychological, and social consequences. Treatment outcomes have historically been modest, guidelines vary widely across countries, and the field has long struggled with questions like: Which therapies work best for which patients? Do medications help? And how long does recovery actually take?
To answer these questions, we synthesized eight recent systematic reviews and meta-analyses representing hundreds of randomized controlled trials and cohort studies involving tens of thousands of patients. The goal: distill what we know—and don’t know—about treating eating disorders in 2024.
The evidence — what studies actually found
Family-based therapy for younger patients
Across multiple meta-analyses, family-based therapy (FBT) consistently outperformed active control treatments in adolescents with anorexia nervosa and bulimia nervosa [PMID 36084848]. FBT treats the family as the primary agent of change, empowering parents to restore their child’s weight and interrupt ED behaviors. One massive umbrella review of meta-analyses found that FBT had the broadest support in youth with AN, and also showed efficacy in adolescent BN [PMID 36084848]. A later transdiagnostic review confirmed that FBT was associated with higher recovery rates in AN, alongside cognitive-behavioral therapy (CBT), psychodynamic therapy, and nutritional interventions [PMID 38214616].
Cognitive-behavioral therapy for adults
For adults with bulimia nervosa, individual CBT specifically adapted for eating disorders (CBT-ED) demonstrated the largest effect sizes versus active controls [PMID 36084848]. CBT-ED targets the distorted thoughts and rigid rules around food, weight, and shape that maintain binge-purge cycles. The evidence base is robust: CBT-ED also showed efficacy in binge eating disorder and in other specified feeding or eating disorders (OSFED) [PMID 38214616]. Self-help versions of CBT, dialectical behavior therapy (DBT), and psychodynamic therapy also helped reduce symptoms in BN, though CBT-ED had the broadest replication [PMID 36084848, PMID 38214616].
Medications: modest benefits, specific contexts
Pharmacological treatment plays a supportive—but limited—role. In anorexia nervosa, the atypical antipsychotic olanzapine modestly increased body mass index (BMI) compared to placebo (Hedges’ g = 0.283), while fluoxetine (an SSRI) did not significantly improve weight [PMID 37393954]. Interestingly, pharmacological treatment in AN was associated with lower recovery rates in one large review, possibly reflecting its use in more severe or treatment-resistant cases [PMID 38214616].
In bulimia nervosa, fluoxetine (typically at higher doses, 60 mg/day) reduced binge eating episodes (Hedges’ g = 0.203) and showed a trend toward reducing purging, though the effect on purging did not reach statistical significance [PMID 37393954]. Antidepressants as a class outperformed placebo for BN but had lower acceptability due to side effects [PMID 36084848].
For binge eating disorder, lisdexamfetamine—a stimulant approved for ADHD—reduced both weight (Hedges’ g = 0.259) and binge frequency (Hedges’ g = 0.571) [PMID 37393954]. Other medications (stimulants, antidepressants) also beat placebo in short-term trials but had poorer tolerability [PMID 36084848]. Notably, medication effects in BED were generally short-lived; psychotherapy showed more durable benefits [PMID 36084848].
Recovery rates: the long view
One of the most comprehensive syntheses to date pooled 415 studies (N = 88,372) and found that 46% of patients with eating disorders achieved overall recovery at a mean follow-up of 3.7 years [PMID 38214616]. Recovery rates climbed steadily with time: 42% at under 2 years, 54% at 4–6 years, and 67% at 10+ years. Chronicity (persistent illness) affected about 25% overall, peaking at 40% in the 2–4 year window but dropping to 18% at 10+ years [PMID 38214616]. These figures held across AN, BN, BED, and OSFED with no significant differences by diagnosis.
Mortality was low in the pooled trials (0.4% overall), but observational cohort studies—which follow patients longer—reported 5.2 deaths per 1,000 person-years, with the highest rate in mixed ED samples (8.2) and the lowest in BN (3.4) [PMID 38214616]. Hospitalization occurred in 26% of patients, ranging from 32% in AN to just 4% in BN [PMID 38214616].
Diagnostic migration and comorbidity
Eating disorders are not static. A striking 8% of people initially diagnosed with AN later met criteria for BN, and 16% moved to OSFED; among those with BN, 19% shifted to OSFED over time [PMID 38214616]. This diagnostic flux underscores that EDs exist on a spectrum and that symptoms evolve.
Comorbidity is the rule, not the exception. Eating disorders commonly co-occur with anxiety, depression, obsessive-compulsive disorder, and personality disorders. They also overlap significantly with attention-deficit/hyperactivity disorder (ADHD): people with ADHD face a 3.8-fold increased risk of developing any ED, with similar elevations across AN (OR = 4.28), BN (OR = 5.71), and BED (OR = 4.13) [PMID 27859581]. Conversely, those with EDs have 2.6 times the odds of having ADHD, rising to 5.8-fold in patients with binge eating specifically [PMID 27859581]. The mechanisms may involve shared impulsivity, emotion dysregulation, and reward-system abnormalities.
Another emerging concept is “food addiction,” assessed via criteria adapted from substance use disorder. One meta-analysis found the prevalence of food addiction highest in BED, followed closely by BN; it was substantially elevated even in non-clinical obese samples and detectable in the general population [PMID 35041154]. This suggests food addiction may represent a separate, transdiagnostic phenomenon that worsens prognosis when present alongside an ED.
Perfectionism: a pervasive risk factor
Perfectionism—both the self-oriented striving for flawlessness and the maladaptive concern over mistakes and criticism—correlates robustly with ED symptoms. A meta-analysis of 95 studies (N = 32,840) found that perfectionistic concerns (worry about errors, doubts about actions) had a medium-sized association with ED pathology (r = 0.33 overall, r = 0.40 in clinical samples), while perfectionistic strivings (high personal standards) also correlated significantly (r = 0.20 overall, r = 0.35 in clinical samples) [PMID 37327637]. These findings reinforce that targeting perfectionistic cognitions—common in CBT protocols—may be therapeutically valuable.
How it might work — biological mechanisms
Neurobiology of reward and control
Eating disorders involve disruptions in brain circuits governing reward, habit formation, and cognitive control. In anorexia nervosa, hyperactivity in the dorsal striatum and prefrontal cortex may reflect over-engagement of cognitive control systems that override hunger signals. Restricted eating can paradoxically activate reward pathways, reinforcing the behavior. Medications like olanzapine, which block dopamine D2 and serotonin 5-HT2A receptors, may dampen some of this compulsive restriction and reduce anxiety around eating, explaining the modest weight gains observed [PMID 37393954].
In bulimia nervosa and binge eating disorder, altered dopamine and serotonin signaling in the ventral striatum (the brain’s reward hub) may drive binge episodes. SSRIs like fluoxetine increase serotonin availability, potentially stabilizing mood and reducing the urge to binge or purge [PMID 37393954]. Lisdexamfetamine, a prodrug of dextroamphetamine, enhances dopamine and norepinephrine transmission, which may reduce impulsivity and improve inhibitory control over eating [PMID 37393954].
Psychological processes
CBT-ED works by interrupting the cognitive-behavioral loops that maintain EDs: rigid dietary rules lead to deprivation, which triggers binge eating, followed by compensatory behaviors (purging, exercise) and shame, reinforcing the cycle. By normalizing eating patterns, challenging distorted beliefs (e.g., “I must be thin to be worthy”), and teaching emotion-regulation skills, CBT-ED addresses the core maintaining mechanisms [PMID 36084848].
Family-based therapy leverages a different mechanism: it removes blame, externalizes the illness, and mobilizes the family’s resources to disrupt ED behaviors before they become entrenched. This is especially powerful in adolescents, whose neural circuits for self-regulation are still developing and who remain embedded in family systems [PMID 36084848].
Developmental timing
Eating disorders typically emerge in adolescence and early adulthood. A global meta-analysis of age at onset found that the median age for AN and BN is 18 years, with 48% of cases starting before age 18 [PMID 34079068]. Onset during the critical neurodevelopmental window of adolescence—when identity formation, peer influence, and biological maturation converge—may explain why family involvement is so effective in younger patients. It also highlights the importance of early intervention: children and adolescents had significantly better outcomes than adults across multiple reviews [PMID 38214616].
Where the evidence is strong (and weak)
Strong evidence
- Family-based therapy for adolescent AN and BN has been replicated across multiple meta-analyses and is considered first-line [PMID 36084848, PMID 38214616].
- CBT-ED for adult BN and BED shows the largest effect sizes and is the most studied psychotherapy [PMID 36084848, PMID 38214616].
- Fluoxetine for BN has consistent, if modest, benefits [PMID 37393954].
- Lisdexamfetamine for BED is effective short-term but requires monitoring for cardiovascular and abuse potential [PMID 37393954].
- Recovery improves with time: the majority of patients improve or recover within 4–10 years, though a substantial minority remain chronically ill [PMID 38214616].
Weak or inconsistent evidence
- Medications for AN: Olanzapine shows modest weight gain, but overall pharmacotherapy has limited support and may even correlate with worse outcomes in some analyses [PMID 37393954, PMID 38214616]. More research is needed on adjunctive medication in AN.
- Relapse prevention: Few studies track patients beyond acute treatment. Relapse rates, maintenance treatments, and long-term medication use remain poorly characterized.
- Treatment for OSFED: While CBT and psychodynamic therapy show promise, the evidence base is thinner than for AN, BN, or BED [PMID 38214616].
- Tailoring to comorbidity: Despite high rates of ADHD, mood, and anxiety disorders, there is scant RCT evidence on how to adapt ED treatment when these conditions co-occur.
Heterogeneity and bias
Most meta-analyses reported moderate to high heterogeneity and some evidence of publication bias [PMID 37393954, PMID 37327637]. Small sample sizes, short follow-up durations, and reliance on sponsored trials (especially for medications) limit generalizability. Many “active control” conditions are heterogeneous (ranging from supportive counseling to treatment-as-usual), making it hard to isolate active ingredients.
Practical takeaways
For clinicians:
- Start with evidence-based psychotherapy. Refer adolescents with AN or BN to family-based therapy if available; offer CBT-ED to adults with BN or BED.
- Consider medications as adjuncts, not replacements. Fluoxetine (60 mg/day) can complement therapy in BN. Lisdexamfetamine may help in moderate-to-severe BED, especially when binge frequency is high, but monitor closely.
- Olanzapine may aid weight restoration in AN, particularly in patients with severe anxiety or agitation, but combine it with nutritional rehabilitation and psychotherapy.
- Screen for comorbidities. Ask about ADHD symptoms, perfectionism, self-injury, mood, and anxiety. Comorbid conditions predict poorer outcomes and may require integrated treatment.
- Be patient. Recovery often takes years. Chronicity at 2–4 years does not mean permanent illness; many patients improve between 4–10 years.
For patients and families:
- Early intervention matters. The sooner treatment starts, the better the prognosis, especially for adolescents.
- Family involvement helps. Parents are not to blame, and their active participation in treatment (in FBT) is one of the most effective strategies for young people.
- CBT and related therapies work. Structured, ED-focused psychotherapy targets the thoughts and behaviors keeping you stuck.
- Medications are tools, not cures. They may reduce symptoms and support other treatments, but psychotherapy remains central.
- Relapse and diagnostic shifts are common. Recovery is rarely linear. Symptoms may wax and wane; ongoing support is key.
For researchers and policymakers:
- Fund longer-term studies. Most trials last <6 months; we need data on maintenance treatment, relapse prevention, and 5–10 year outcomes.
- Investigate treatment sequencing. What should happen when first-line therapy fails? How do we integrate treatment for comorbid ADHD, trauma, or substance use?
- Explore transdiagnostic mechanisms. Perfectionism, emotion dysregulation, and food addiction cut across diagnoses and may be valuable treatment targets.
- Increase access to specialized care. Despite strong evidence for FBT and CBT-ED, these treatments remain unavailable in many regions. Training and dissemination are critical.
Limitations of this review
This synthesis relies on published meta-analyses, which aggregate randomized controlled trials and observational studies. Several limitations apply:
- Heterogeneity in definitions. Diagnostic criteria, outcome measures (recovery, remission, improvement), and treatment protocols varied across primary studies, contributing to statistical heterogeneity.
- Publication bias. Positive findings are more likely to be published, potentially inflating effect sizes, especially for newer or industry-sponsored medications.
- Short follow-up. Many RCTs followed patients for only 3–6 months, limiting conclusions about durability of treatment effects.
- Underrepresentation. Most studies were conducted in high-income Western countries; less is known about EDs and treatment response in Africa, Asia, and low-resource settings. Males, older adults, and ethnic minorities are also underrepresented in the literature.
- Lack of mechanistic data. While we describe plausible mechanisms, few trials directly test them or link mechanism to outcome. Neurobiological and genetic research is advancing but not yet integrated into treatment guidelines.
- Comorbidity. Trials often exclude patients with severe psychiatric or medical comorbidity, yet these are precisely the patients seen in real-world practice.
- Therapist expertise. Treatment fidelity and therapist training influence outcomes but are inconsistently reported.
Finally, none of the included reviews addressed disparities in access to care, cost-effectiveness, or patient preferences—critical factors for implementing evidence in practice.
FAQ: Eating Disorders Treatment Evidence
What is the most effective treatment for anorexia nervosa in teenagers?
Family-based therapy (FBT) shows the strongest evidence for treating adolescent anorexia nervosa, according to multiple meta-analyses [PMID 36084848]. In this approach, parents take an active role in restoring their child’s weight and interrupting eating disorder behaviors, rather than the therapist working solely with the patient. FBT consistently outperformed other active treatments in adolescents with anorexia and has also demonstrated efficacy for bulimia nervosa in younger patients.
How effective is cognitive-behavioral therapy for eating disorders in adults?
Cognitive-behavioral therapy adapted specifically for eating disorders (CBT-ED) demonstrates the largest effect sizes for adult bulimia nervosa compared to active control treatments [PMID 36084848]. CBT-ED targets the distorted thoughts and rigid rules about food, weight, and body shape that maintain binge-purge cycles. The evidence base extends beyond bulimia—CBT-ED also shows efficacy for binge eating disorder and other specified feeding or eating disorders (OSFED), with robust replication across multiple studies [PMID 38214616].
Do medications help treat anorexia nervosa?
Medications play only a modest, supportive role in anorexia nervosa treatment. The atypical antipsychotic olanzapine shows small but measurable benefits in increasing body mass index (BMI) compared to placebo (effect size = 0.283), while the antidepressant fluoxetine does not significantly improve weight [PMID 37393954]. Interestingly, pharmacological treatment in anorexia was associated with lower recovery rates in one large review, possibly because medications are often prescribed for more severe or treatment-resistant cases rather than causing worse outcomes [PMID 38214616].
Why are eating disorder rates rising after COVID-19?
Eating disorder rates have risen sharply following the COVID-19 pandemic [PMID 42083839], though the exact mechanisms are still being studied. Potential contributing factors include increased social isolation, disruption of normal eating routines, reduced access to treatment and support services, heightened anxiety and depression during lockdowns, and increased social media exposure. These disorders are serious psychiatric illnesses—not lifestyle choices or temporary phases—with medical, psychological, and social consequences that require professional treatment.
How long does it take to recover from an eating disorder?
Recovery timelines vary considerably, but evidence suggests patience is essential. One meta-analysis found that nearly half of patients achieved recovery at four or more years of follow-up, regardless of their specific eating disorder diagnosis. This highlights that eating disorders are chronic conditions for many individuals, and initial treatment response does not predict final outcome. Recovery rates improve substantially with longer follow-up periods, emphasizing the importance of sustained treatment and support rather than expecting rapid resolution within weeks or months.
Studies referenced
- PMID 36084848 — Umbrella review of meta-analyses on ED treatment; found family-based therapy effective in adolescent AN/BN, CBT-ED best for adult BN, medications modest but helpful in BED.
- PMID 34079068 — Large meta-analysis on age at onset of mental disorders globally; ED median onset = 18 years, with nearly half starting before age 18.
- PMID 37393954 — Systematic review and meta-analysis of psychopharmacology for EDs; olanzapine increased BMI in AN, fluoxetine reduced binges/purges in BN, lisdexamfetamine effective in BED.
- PMID 38214616 — Transdiagnostic meta-analysis of ED outcomes; 46% recovery at ~4 years, 25% chronicity, mortality 5.2/1,000 person-years; children/adolescents fare better than adults.
- PMID 37327637 — Meta-analysis of perfectionism and EDs; perfectionistic concerns r=0.33 and strivings r=0.20 with ED symptoms, stronger in clinical samples.
- PMID 27859581 — Meta-analysis showing ADHD increases ED risk ~4-fold and vice versa, with especially high overlap in binge-eating presentations.
- PMID 35041154 — Systematic review disentangling food addiction from BED; food addiction most prevalent in BED and BN but also present in obesity and general population.
- PMID 42083839 — French-language review on ED diagnosis and treatment; emphasizes rising rates post-COVID-19 and importance of early, multidisciplinary care.