Author's response: I am grateful to Dr Wrate for raising the issues he has. I would point out, first, that the Management of Really Sick Patients with Anorexia Nervosa (MARSIPAN) report ¹ was intended for clinicians caring for adult patients over 18 with severe anorexia nervosa. It was clear during the preparation of MARSIPAN that a further document for children and adolescents was required. The work was done and the junior MARSIPAN report ² is the result. I think that the main issue raised by Dr Wrate, namely the appropriateness or otherwise of specialist hospital care for children and adolescents with anorexia nervosa, needs to be addressed by a child and adolescent psychiatrist such as those involved in the junior MARSIPAN report. However, I should be grateful if I could comment on some of the other issues discussed in the letter.

Assessing whether a person is at a risk high enough to warrant hospital treatment is one such problem. In adults, current opinion suggests that a body mass index (BMI) of <13 kg/m ² , electrocardiographic abnormalities, low potassium (especially <3.0 mmol) and severe anorexic myopathy constitute a serious threat to life. In one study, the patients who died from malnutrition had BMI between 9.1 and 12.9. ^{Reference Rosling, Sparén, Norring and von Knorring3} In adolescents, junior MARSIPAN recommends that a BMI<0.4th percentile indicates high (‘red’) risk. This turns out to be more conservative, as a BMI at the 0.4th percentile in a 15-year-old is 15. I hope that my child and adolescent psychiatrist or physician colleagues will take the opportunity to give a view on this. From my practice, the most reliable sign that a patient requires admission is when I feel my own heart sinking. This usually accords with the high-risk parameters in the patient, quoted in the MARSIPAN report.

Dr Wrate correctly notes that the past two decades saw a decline in death rates for anorexia nervosa, but argues that this is due to the fact that treatment is now more effective and introduced earlier, not necessarily because it is hospital based. It is uncertain whether patients presenting with very high risk would have similar survival rates outside hospital with community care. The Scottish Anorexia Nervosa Intensive Treatment Team (ANITT; www.anitt.org.uk) provides community care for adults of very low weight, but no evaluation of that or any other similar service has been published, nor are there randomised trials of care in this very high-risk group of (adult) patients.

On the question of chronicity, Dr Wrate identifies progressive loss of bone mineralisation as the only significant medical complication of adolescent anorexia nervosa. However, I am aware of many reports of serious complications such as irreversible failure of linear growth, irreversible failure of breast development, and cardiac abnormalities in this patient group. ^{Reference Katzman4} Again, the views of my colleagues treating younger patients would be appreciated.

Another interesting point raised by Dr Wrate is that with regard to young people with anorexia nervosa, risk may be ‘socially constructed’. The implication is that if a risk is socially constructed rather than medically evidenced, it is related to the needs of individuals and systems such as the family and hospitals rather than a real risk of death. This may be true in many cases, especially if the usual risk factors are not too seriously impaired. However, I think it would be dangerous to apply it to the most seriously ill, for example a patient with a BMI of 10.

Finally, there is the issue of patients resisting nasogastric feeding as opposed to treatment as such. This is a complicated matter. The act of admitting a patient to a specialist eating disorders unit may well engender fury in the patient and a determination not to gain weight. On the other hand, the admission may have been appropriate because of their dire physical state. In adult eating disorder services there is varying opinion about whether a seriously ill patient ever requires nasogastric feeding. If a patient resists eating, as may be the case, the option is to provide nutrition against their wishes, often under the Mental Health Act 1983. This might involve forcing the patient to eat by restraining them and pushing food into their mouth. This may be ineffective, or so aversive to staff that nasogastric feeding may be preferred. Some have said that skilled nursing can always result in a patient accepting food, thereby avoiding nasogastric feeding. I suspect that the situation in which a patient's life would be lost if forced feeding were not done is more commonly encountered in adults, as suggested by Dr Wrate. However, when it does occur, clinicians may be forced into more and more coercive treatment. Occasionally, such treatment may not be short lived and there are, at present, several adult patients in units around the UK receiving forced nutrition, under the Mental Health Act, by nasogastric or percutaneous endoscopic gastrostomy (PEG) feeds for periods which can run to several years. This may be very aversive to patients, staff and relatives, not to mention the enormous cost to the National Health Service (£1000 per day is not unusual in this situation), and merits audit and research.