How to Help Someone With an Eating Disorder

If you have just worked out that someone you love has an eating disorder, three things matter before anything else on this page.

Know what counts as an emergency. Fainting or collapse. Chest pain, palpitations, an irregular heartbeat. Confusion or disorientation. A seizure. Being unable to stand up from squatting or lying down without pushing with the arms. Refusing all fluids. Any of those means an ambulance, not a GP appointment.

And the part that British emergency guidance puts hardest, because it is what kills people: someone can be in serious danger while looking physically well and having normal blood tests. Normal results are not reassurance. A person who seems fine can be close to a cardiac event.

Do not start a confrontation about food. Not tonight. The first conversation should be private, calm, short, and anchored to something you have actually noticed rather than to an accusation.

Say the two things that help. That it is not their fault. That people recover from this. Both are true and both are recommended by every source. Do not reassure them about their body, which is covered below and is the one thing people get wrong.

Beat opens its guidance for families with two lines. Recognise that you are not to blame. And tell your loved one that they are not to blame either.

That ordering is deliberate, and it is a correction to something.

For most of the twentieth century families were treated as the cause. Charcot described parents as a pernicious influence. Gull called them the worst attendants. In the 1970s the psychosomatic family model taught that enmeshment, rigidity and conflict-avoidance produced the illness, and treatment sometimes meant removing the child from the family altogether, a practice with a name: parentectomy.

None of it replicated. Families of people with eating disorders turn out to be about as varied as families in general. The model that replaced it takes no position on cause at all and treats the family as the main resource for getting someone well.

But the residue is still in the room, which is why the national guideline instructs clinicians to expect that families will feel guilty and responsible. They usually do.

Worth being precise about what the modern position actually claims, because it is narrower than it sounds. It is not that family environment never matters. It is that blanket family-causation was never demonstrated, and that blaming parents made treatment worse by removing the people best placed to help.

Carer guidance is unusually specific about sentences. Here are some that are recommended, more or less verbatim.

“I might not understand, but I love you and I am here to listen.”

“I have noticed you have not been eating with us any more.” Anchored to an observed behaviour rather than a diagnosis, and framed as something you saw rather than something they did wrong.

At a difficult meal, short: “I am here to keep you safe.” “I know this is hard. You do not have to like this.” “We both hate this eating disorder, but I know you can do it.”

Any version of just eat. The National Alliance for Eating Disorders answers it in one line: if it were that simple, they would not have an eating disorder.

Also why can’t you eat normally, you are being ridiculous, I am disappointed in you, and think what this is doing to your family. Sentences that begin with you and describe a failure produce defensiveness. They also tend to repeat, word for word, what the illness has been telling the person already.

And anything about how they look. Including the kind things.

This is the single most universal piece of advice in the entire literature, and unlike most of it the reason comes from patients rather than clinicians. They report hearing you look healthier and you look well as you have gained weight. The illness does the rest.

Clinicians often separate the illness from the person, talking about what the anorexia wants rather than what the patient wants. It is a real technique and it is in the guidelines.

It also backfires when a family member does it. Qualitative research found externalising works when the person leads it in their own words, and fails when it is imposed on them, leaving them feeling unseen as a person. Carers who tried attributing behaviour to “the anorexic voice” reported arguments and disconnection. One study’s advice to families was to hold the idea privately rather than say it out loud.

There is a word for reorganising your household around someone’s illness, and there is a scale that measures it.

It looks like cooking a separate meal. Buying only the safe foods. Letting rules about which plate, which cutlery, which time of day quietly govern what everyone else does. Answering the same anxious question again, and again. Not booking the restaurant. Not going on the holiday. Seeing something and deciding not to mention it.

It grows in proportion to how much time you spend with the person. And it arrives gradually enough that families almost never notice it arriving, which is the whole problem with it.

The honest caveat: accommodation correlates with worse outcomes and carer skills programmes reduce it, but whether reducing it causes the person to get better has not been established. It might be that sicker illness produces more accommodation rather than the other way round. The advice to stop accommodating is sound and the mechanism behind it is not proven.

If you are supporting someone through meals, this is a distinct practical skill with published guidance behind it, and the advice is consistent across sources.

Agree the plan beforehand so that nothing is negotiated at the table. Have everything ready before you sit down, because a delay is time for the illness to argue. Keep the conversation light and away from food, weight and bodies. Eat the same meal alongside them.

Then stay.

The hour after the meal is harder than the meal. That is the part families do not plan for, and it is where structured support matters most: company, distraction, something to do. The three words that recur across every set of guidance are calm, consistent, compassionate.

Nearly all published guidance assumes you are the parent of a teenager. If you are not, some of it does not transfer.

One more thing that changes the picture. If the person is male, expect services and screening tools built around women, more stigma, and a later start. Research has found stigma specifically reduces help-seeking in men, and specifically for the symptoms coded as feminine. Research on the people caring for them barely exists.

Most of this page is consensus. This part is not.

Caring for someone with an eating disorder produces measurably more burden than caring for someone with depression or with schizophrenia. That comes from a study comparing 251 eating disorder carers against carers in both other groups. Worrying was the most common consequence, and being a mother or a partner predicted the heaviest load.

In the sample used to validate the accommodation scale, 61% of carers scored above the clinical threshold for anxiety. Not distressed. Above the threshold.

Carer skills courses have been properly trialled, which almost nothing else here has. They reduce carer distress, burden, criticism and accommodation, and they work whether delivered in a workshop or online.

Then the part charity guidance rarely mentions, which needs stating carefully because it is easy to get wrong in either direction.

In the largest trial, the carer benefits were clear at six months and had diminished by twelve. Families who got the package used fewer inpatient bed days in months seven to twelve after discharge, a real service-level effect. At two years the improvements ran in the right direction across every outcome measured, for patients and carers alike, with small to moderate effect sizes, and the gains in weight and in time spent caregiving had held. But none of the differences between the two groups reached statistical significance. A separate Spanish trial found the structured skills course no better than straightforward psychoeducation.

So the courses reliably help you, and the direction of travel for the ill person is encouraging without being demonstrated. That is a more honest summary than either “proven to work” or “faded to nothing”, and both have been written about this trial.

Someone can be seriously ill and decline all of it. This is common and it is not a failure of your approach.

What you can do: escalate to the GP or the specialist service, urgently, and describe what you have observed rather than what you suspect. Go to emergency care if they are physically unsafe. Keep a record of the signs.

What you cannot do: detain anyone. In the UK, anorexia is a mental disorder under the Mental Health Act and providing nutrition counts legally as treatment for it, for adults as well as children. But only clinicians can invoke that, through a statutory process. Families cannot. In the United States the rules are state by state and generally require danger to self or grave disability, with the decision resting with clinicians or a court.

That is an outline, not a map. Scotland operates under separate legislation with an additional impaired-decision-making test, thresholds and procedures vary considerably between US states, and the position for a minor differs again. Anyone actually facing this needs advice on their own jurisdiction rather than a paragraph on a website.

On suicide, which is a real and substantially elevated risk in this illness: asking someone directly whether they are having suicidal thoughts does not increase the danger. Guidance is explicit. Ask. Then involve the treatment team and make the environment safer, which national guidance says should be done with the family rather than around them.

What is not known is whether raising the subject again after a refusal helps or hardens it. Nobody has studied that, and every family faces it.

Both of these cut against what the rest of this page tells you, and leaving them out would make it tidier and less true.

Being less involved may be wrong for this illness. The concept of expressed emotion, which covers criticism and emotional over-involvement, came from schizophrenia research, where high expressed emotion reliably predicts relapse. It has been imported into eating disorders and it does not behave the same way. One eighteen-month study of adolescent anorexia found that high maternal emotional over-involvement was associated with better outcomes: better nutritional status, less rehospitalisation. Criticism looks harmful. Over-involvement may not be, and the two are usually bundled together in advice.

People who were forced into treatment do not agree about it afterwards. In a study of thirty adults looking back, those who were now doing well described involuntary treatment as necessary. Those still unwell described it as harm. Earlier work found that about a third of hospitalised patients initially denied needing to be there, and many changed their minds within a fortnight. Most patients accept that coercion can be justified to save a life while doubting it in less urgent situations, and those with repeated involuntary admissions describe it most bitterly: one summary of their view was that it saves lives and also kills spirits.

The same intervention, remembered two opposite ways, and which way appears to depend on how the story turned out. If you are the person weighing this decision for someone else, that is the honest state of the evidence, and it does not tell you what to do.