Statistics from Altmetric.com
We thank Colver, MacDougall, and Cant for their response to our paper.1 The message underpinning our paper was that severe allergic reactions to foods are not as uncommon as MacDougall et al suggested.2 However, we are not comparing like with like and the problem lies in the definition of a severe food reaction. MacDougall et al only identified the extreme end of the clinical spectrum (children who have suffered cardio-respiratory arrest, received at least two doses of epinephrine (adrenaline), a fluid bolus, inotropic support, or at least two doses of nebulised bronchodilator—all unvalidated and unreferenced outcome measures). Having set the threshold for inclusion so high, it is hardly surprising that the incidence of this extreme outcome is so low. What was the authors’ purpose in identifying this extreme population? If it was reassurance about the low incidence of life threatening reactions, then it should be appreciated that in a recent American series,3 8/10 children who died of food allergy had previously reacted to the food that caused the fatality. Also, in a UK survey, 21/92 (23%) of peanut allergic patients with a history of a mild reaction subsequently had a severe reaction to peanut.4 MacDougall’s data should not be used to dilute care given to children with less severe food allergy.
MacDougall et al have failed to identify the much larger population, of the most interest to paediatricians, allergists, and parents: children who have suffered a less extreme but still severe food allergy reaction (or anaphylaxis) characterised by airway narrowing.5 It is important to identify and study this large group as there is no consensus on how they should be managed. We have shown that intervention with a structured management plan, incorporating assessment of severity, avoidance advice and tailored medication for self treatment, together with a patient held treatment plan, reduces the risk and severity of further reactions, in both adults and children.6
It is clear that the British Paediatric Surveillance Unit (BPSU) survey missed children referred to allergy centres, not run by paediatricians. MacDougall et al studied only inpatient admissions and not outpatient referrals. It is erroneous to assume that children referred to an allergy clinic would have been seen by a paediatrician. We look after over 790 outpatient referrals of children with nut allergy; 218 (29%) have had reactions involving respiratory symptoms and 58 (8%) severe dyspnoea and collapse. This number would be much larger if all foods were considered, and this effect is repeated for other allergy centres around the country. These children with severe reactions are derived from approximately 1/30th of the UK population. The majority were referred directly from their general practitioner and have had no contact with a paediatrician, and therefore would not have been notified to the BPSU. The fact that the number of inpatient admissions in MacDougall’s paper matched that from a previous survey is irrelevant, as neither considers the large number of children who present to allergy outpatient clinics via their general practitioner.
We present data to estimate the incidence of those severe reactions that were missed by MacDougall’s paper, using a different and more appropriate definition of severity. We compare them to MacDougall’s estimate of 0.19 severe food reactions per 100 000 children per year. We have data from 791 children with nut allergy, 58 of whom had at least one severe nut induced reaction; the data were collected over six years, giving an incidence of at least 1.93 severe reactions per 100 000 children per year for nut allergy alone: a figure 10-fold higher than MacDougall’s data for all food allergy. This estimate is a minimum, as each child has had at least one severe reaction (possibly more) and is for nut allergy alone, rather than all food allergies.
The estimate of the incidence of deaths due to food allergy produced by MacDougall et al is an underestimate, as the authors did not consider asthmatic deaths, where no allergenic cause was suspected and we thank Colver for acknowledging this. There are 60-fold more deaths due to asthma, than those identified as due to a severe food reaction by MacDougall et al. If only a few of these were due to food allergy, it would alter the figures dramatically.
Colver has citied Bock’s important series of fatal food reactions,3 which showed 80% (8/10) of deaths in under 16 year olds were due to peanuts or nuts, the youngest being 2 years old. It is not clear, however, why we should be reassured by MacDougall’s finding of no peanut allergy induced deaths in children under 13 years. Do the authors not consider tree nut allergy induced deaths in 2 year olds, or peanut allergy induced deaths in 14–16 year olds relevant?
Finally, we agree that the issues surrounding provision of adrenaline autoinjectors require good data on the clinical features and natural history of nut allergy. However, to consider adrenaline autoinjectors in isolation is missing the point. The provision of adrenaline autoinjectors should be seen in the broader context of a comprehensive management plan for all children with food allergy. Children with mild food allergy reactions can go on to have more severe reactions.4 We have shown that enrolment of children with nut allergy of all severities in such a management plan, consisting of repeated specialist advice on allergen avoidance, together with tailored medication including oral antihistamine with or without inhaled adrenaline or injectable adrenaline, results in a reduced number and severity of further reactions.6 Colver has emphasised the efficacy of this treatment plan in children, by highlighting that only three adults (aged 27–41 years) and no children had further severe reactions.