EAACI task force anaphylaxis criteria (2007, from Sampson): any one of –

1. acute onset (up to several hours) skin/mucosa reaction (eg generalized hives, pruritus or flushing, swollen lips/tongue/uvula) plus respiratory or cardiovascular compromise eg dyspnoea, bronchospasm, stridor, hypoxia, hypotension, collapse
2. acute onset after exposure to likely allergen of at least 2 of: skin/mucosa changes, resp, cardio compromise, persistent GI (eg crampy abdo pain, vomiting)
3. hypotension after exposure to known allergen for that individual

Note that this definition talks about “compromise”, and the examples given are mostly signs that only a medical professional could identify!  This doesn’t help patients/parents.  It also allows for anaphylaxis with just skin and persistent GI symptoms (where likely allergen exposure).

In Sampson’s original 2003 criteria, he grades anaphylaxis into 5 types.  1 and 2 would not be considered anaphylaxis these days at all!  

WAO 2020 revised definition, based on input from 15 different allergy societies – “highly likely when any 1 of

1. acute onset skin/mucosa plus one of airway/breathing compromise, circulation, severe GI, or
2. acute hypotension, laryngeal involvement, bronchospasm [specifically] after likely exposure, even without skin symptoms.

BSACI’s allergy plan includes as severe symptoms hoarseness, dysphagia but also:

• persistent cough,
• noisy breathing
• tongue swelling
• persistent dizziness, pale/floppy, suddenly sleepy

AAP also has cough, hoarse, PLUS severe vomiting and diarrhoea, “many hives”(!), agitation!

On logistic regression, confusion and incontinence were strongly associated with hypotension and hypoxia. Dizziness, vomiting, abdominal pain, dyspnea and chest/throat tightness had weaker, albeit significant, associations.  Pre-existing lung disease was associated with an increased risk of hypoxia. [Journal of Allergy and Clinical Immunology Volume 114, 2004, 371-376]

In a survey of kids with anaphylaxis, the mean latent period was 15.4 (SD27.5) minutes, ie 95% will react within 90 minutes of exposure. The type of allergy does not predict the latency well; however, age is inversely related, with younger children having more gradual onset. GI and cardiovascular symptoms tended to come later than skin/respiratory. 60% of anaphylactic reactions occurred in the home, and 10% happened in health care environments. Males predominate, particularly with regards exercise induced and insect venom anaphylaxis [PEDIATRICS Vol. 101 No. 4 April 1998].

Biphasic anaphylaxis

About 20% to 30% of food-induced anaphylactic events have a biphasic or recurrent response, although only half of those severe. 90% of recurrent reactions within 12 hours. Delay in giving adrenaline increases risk of recurrence! Some anaphylactic reactions are persistent, going on for hours.

Recognition

Recognition of anaphylaxis can be poor by non-experts. It is sometimes assumed that anaphylaxis must be life threatening, and that it must be progressive to the point of shock/collapse. Well recognised that anaphylaxis can resolve without treatment in some cases. Some have called for anaphylaxis grading, such as Sampson’s, but this then undermines the basic distinction.

Equally, people will sometimes diagnosis anaphylaxis on the basis of severe facial swelling, or widespread rash, neither of which are criteria.

Part of the problem of course is that what parents report is not the medical terminology included under the definitions. Infants and toddlers are a particular problem, since they cannot describe what they feel, and hypotension/syncope are extremely late signs. In US study, 48% of caregivers recognized 1 or more less obvious symptoms of anaphylaxis only in retrospect [but that includes skin/tongue/GI symptoms: looking at table II, 15% recognised sudden behaviour change in retrospect, 11% cough/wheeze, 6% wobbly/lethargy, difficult to rouse – can’t add up though]– that lack of recognition may affect management by medical team.

US 2024 study (Handorf) proposes modified NIAID/FAAN criteria – swollen tongue/uvula/pharynx becomes respiratory, not mucocutaneous; adds cough, drooling, hoarse cry/voice to respiratory too; specifies gagging, spitting up, diarrhoea, back arching as GI; pallor, mottling, obtunded/lethargy, altered mental state as CVS. By these criteria, 52% of all ED visits in children under 5 then became “anaphylaxis” (includes those given that diagnosis by attendings plus some ambiguous cases determined by clinicians to be anaphylaxis). Attendings diagnosed anaphylaxis in 68% of cases, but falls to 59% in infants. Original NIAD/FAAN criteria have 85% sensitivity, but falls to 77% infants. Modified criteria 100% sensitive for infants; 96-98% of older children. Compared with NIAD/FAAN, modified criteria picked up 43% more CVS signs, and 32% respiratory. Only 5% increase in GI. No hypotension/syncope in infants/toddlers, as expected.

Epidemiology

UK hospital admissions for food anaphylaxis have increased by 6.6% each year between 1998 and 2018 for under 15s (about 3x increase over whole period).  For other ages much lower increase.  Possibly due to 4hr waiting time rules for emergency departments? Or NICE recommendation on observation after suspected anaphylaxis? Case fatality rate has gradually fallen over time.  At least 46% of all deaths triggered by peanuts or tree nuts.  Cow’s milk responsible for 26% of deaths in school aged children (and rising), which shows how serious it can be when you don’t grow out of it in infancy.  About a quarter trigger unknown (no comment – presumably no details in coding, rather than actually unknown), peanut and tree nuts together about a third.  AAI prescriptions over the same period have increased by 336%.

Commentary by authors talks about effects on family of food allergy diagnosis but then goes straight into reasons for overdiagnosis, either through self reporting, commercial pressures, even campaigning by allergy charities.  They say it is a “familiar pattern” with sharp increases in softer indicators of allergy, but no increase in markers of severe disease, and that this is the case in other countries such as the US and Australia.  They conclude that we live in an era of increasing concern and awareness, but not a food allergy epidemic.

Whether the data are reliable, is a whole other question.  Anaphylaxis is not well recognized, treated or documented, and only cases that were admitted were included, so these cases are probably just a fraction of what is actually coming to emergency departments.  That doesn’t matter so much if that fraction stays the same and you’re only interested in trends.  But even the interpretation of their own data seems curiously sceptical – if it is true that allergy is no more common than it was at any time in the last 30 years, are we saying families in the past just didn’t bother bringing their children to hospital when they had anaphylaxis? Or just more likely to admit (as NICE guidance from 2011 encourages observation for at least 6 hours, although the authors themselves say this probably only produced a minor increase in cases)?

Falling rates in peanut/nut, better awareness esp industry?  Whereas milk awareness low? Although milk allergy common in young children, 5% of deaths in adults still due to milk (and prob harder to spot than nuts in food).

Higher rates of food anaphylaxis admissions in boys before puberty (male-to-female ratio 1.6:1), but reverses from age 15 years onwards.  Highest rates of admission and death in teenagers, but in this study this risk continues through into mid adulthood, so supports the idea that it’s not specifically “teenage behaviour” that increases risk, rather biological vulnerability.  In children under 5, deaths rare even though highest rate of admission.

[Conrado, BMJ 2021]

Risk Factors

Fatal anaphylaxis rate is about 1 in ¼ million.  Risk of hospital admission with anaphylaxis 1 in 10 000.  More likely to die on way to hospital appointment!? Doesn’t mean we avoid driving.  Important to see that there can be an “acceptable risk” of living with food allergy.

Although there are a range of associations, most of these are very weak, which makes them unhelpful, or even misleading.

There is also a difference between sensitivity and severity.  Some children during food challenges may only start reacting after a relatively high threshold dose – but their reactions tend to be worse, not unexpectedly.

Exercise is an important factor esp in teenagers, can even (rarely) be the sole identifiable trigger! See Exercise-Induced Anaphylaxis.

Statistically, females have worse reactions.  In mice, the difference appears to be due to effect of oestradiol on increased tissue expression of eNOS (one of the NO synthases).

Medication – eg beta blockers, NSAIDS may increase risk.

Sleep deprivation reduced threshold in UK TRACE peanut study!

Most food anaphylaxis related to nuts (including peanut) and milk, but this is partly just how frequent these allergies are and how frequently the foods are encountered.

Deaths

Mostly in children over 5, despite the fact that food allergies are more common under 5 and usually lessen with time.  Families often assume risk is higher in young children, which is incorrect.  Nearly all are in children with asthma. Of the 8 deaths between 1990 and 2000, 4 were due to milk, 2 peanut, 1 egg and 1 mixed (but see below for bigger study). Both the peanut deaths were in children over 13yrs. Children are over 250x more likely to die in road traffic accident, although this is hardly a fair comparison.

Fatal Anaphylaxis Registry – set up in 1994 by Dr Richard Pumphrey based at Manchester University NHS Foundation Trust. This is now evolving into a European wide anaphylaxis registry.

In 80% of fatal anaphylaxis cases [mixed causes], adrenaline not given before arrest, which suggests delayed administration could be a factor.  Same study shows that adrenaline does not prevent death even when given before arrest.

Most cases were in people with no previous history of anaphylaxis [but how many nut+asthma, where higher risk recognised?]

Survival better if adrenaline given within 30 mins [1992 Sampson article NEJM].

1 death at 6 hours, despite repeated adrenaline. 1 death after intentional consumption, despite immediate epipen administration.

So avoidance and asthma management are arguably more important than providing adrenaline autoinjectors! [Pumphrey]