Although we agree with Dr Wrate that a context to risk is important, ^{Reference Wrate1} we cannot concur with his view of the context. As healthcare professionals caring for young people with eating disorders, we do not regard seven deaths in children and adolescents as reassuring. When put into the context of more ‘traditional’ paediatric illnesses, 1.5 deaths per year are not far removed from 12 deaths from diabetic ketoacidosis and 14 deaths from invasive childhood meningococcal disease (sources available from the authors on request). Even within the 18 years since Dr Wrate's search, there has been a continuing reduction of vaccine-preventable diseases, improvement in neonatal outcomes and survival from childhood malignancies, resulting in a shift in childhood illnesses from communicable to non-communicable diseases, including eating disorders.

Death is a preventable outcome of eating disorders and may be secondary to lack of awareness, knowledge or timely treatment; yet it would seem that paediatric medical and mental health services are sometimes ill equipped to meet the challenges of young patients presenting with eating disorders, especially acutely. Studies show that 50% of children less than 13 years old are hospitalised early in their illness ^{Reference Nicholls, Lynn and Viner2} and numbers of children under 14 hospitalised for eating disorders have risen over the past 10 years. A recent survey of on-call paediatric registrars in hospitals in England and Wales revealed poor knowledge of the acute management of children with eating disorders (C. Cumby, 2012, personal communication). Most were unaware of the frequent cardiovascular complications and unable to outline the complications of refeeding syndrome.

We agree with Dr Wrate that there has been too much emphasis historically on hospitalisation rather than on good-quality out-patient treatments and early intervention; however, we believe that it is important not to negate risk in order to redress this balance. The decision to admit is a complex one; unfortunately, it is our observation that admission to hospital is often more dependent on available services and experience of healthcare professionals than clinical need.

We also wish to comment on Dr Wrate's commentary on morbidity in eating disorders in childhood, which he describes as relating only to bone mineral density. Not wanting to understate the significance of potentially irreversible low bone density as a medical complication, we must highlight the multisystem effects of eating disorders and associated malnutrition in this group. Other unique medical complications found in children with eating disorders include an impact on growth and development and changes in brain structure and function. ^{Reference Hudson and Court3,Reference Chui, Christensen, Zipursky, Richards, Hanratty and Kabani4} Such complications are treatable if recognised early and treated aggressively.

To date, there are too few data on the outcomes of children with eating disorders for us to relax about risk. In preparing the Junior MARSIPAN report, ⁵ instances of deaths were rare. Nonetheless, the cases described in the report show this is risky business indeed and examples were not hard to come by. The risk assessment framework in the Junior MARSIPAN report acknowledges the complexity of risk assessment, and is stratified to avoid overstatement. Valuable lessons learnt from improvements in survival from preventable childhood diseases such as meningococcal disease include the importance of early recognition, awareness and training. In the absence of consistency, we argue that erring on the side of caution is prudent and the MARSIPAN guidelines are important safeguards to help clinicians understand, recognise and treat eating disorders.