ANAPHYLAXIS: CAN WE TELL WHO IS AT RISK OF A FATAL REACTION?
[Curr Opin Allergy Clin Immunol 4(4):285-290, 2004. © 2004 Lippincott
Williams & Wilkins]
Anaphylaxis is frightening and patients commonly fear their next reaction
will be fatal. This review looks at the characteristics of fatal reactions
to find if a fatal recurrence is predictable.
Recent Findings: Most publications on fatal anaphylaxis are case
reports that do not help predict risks. Most epidemiological studies focus
on non-fatal reactions. The UK fatal anaphylaxis register demonstrates
that over two-thirds of those dying from sting reactions and over
four-fifths dying from drug anaphylaxis had no previous indication of
their allergy, whereas those dying from food allergy had usually had
previous reactions but these were typically not severe. Recent reports of
anaphylaxis epidemiology based on diagnostic coding or attendance for
treatment may be biased by differences in health service resource
utilization according to the cause and course of the reaction.
Summary: Most fatal anaphylactic reactions are unpredictable. The
appropriate management after recovery from a severe reaction may be
protective against a fatal recurrence. An accurate identification of the
cause and effective avoidance is a crucial part of this management,
together with effective treatment of asthma for those with food allergy,
immunotherapy for sting allergy, the avoidance of drugs that potentiate
anaphylaxis, and effective training in self-treatment.
For anyone who
has had an anaphylactic reaction, the prospect of a fatal recurrence is
daunting, yet there has been little research into the chances of this
happening. Since 1992, I have tried to identify and analyse every
anaphylactic death in the UK.[1,2**] Taken together with data
collected systematically from over 1500 patients with non-fatal
anaphylaxis seen in our allergy clinics over the same interval, it becomes
possible to analyse the risk factors for fatal reactions. This review
relates these findings to the other recent epidemiology of anaphylaxis.
Incidence and Causes
20 deaths have been identified each year (one each year for each 3 million
of population). There will have been additional undetected fatal
reactions, such as those to antibiotics taken at home, in which autopsy
findings are commonly limited to the infection (often with asthma) for
which the antibiotic was taken.
lists the UK anaphylactic deaths over an approximately 10-year period.
one in five patients referred to our clinic for anaphylactic reactions
attributed to food allergy have negative specific IgE, skin tests and
challenge tests for the supposed cause. Up to a quarter of sudden deaths
blamed on foods are of a similar pattern and may have been caused by
non-allergic anaphylaxis, angioedema or asthma.
Epidemiological studies of
anaphylaxis have reported 8-20% idiopathic reactions,[4**,5]
and 'food anaphylaxis' deaths may include some of these. Of the remainder,
70% were attributed to nut allergy with approximately equal numbers of
peanuts, tree nuts and mixed or unidentified nuts, and a peak incidence
age of 15-24 years of 2.3 deaths in each 100 000 with nut allergy each
year. [Calculated from 17 nut-allergy deaths in 10 years in
15-24-year-olds: UK population aged 15-24 years=7.2 million, assuming 1%
have nut allergy. A UK survey gave a lower prevalence, and
using the unpublished raw data from this, the lifetime risk of dying from
their nut allergy was calculated as R0.001
for females and 0.0005 for males (where
= risk of dying from nut allergy in each interval i)].
Table 1. Suspected Causes
allergy fatalities included 14% with cows' milk allergy, with the rest
caused by allergy to fish, shellfish, chickpea, hens' egg and banana (Table
No other studies have reported complete national statistics for fatal
anaphylaxis, but there is much helpful information in case reports and
small series. In general, the data in such papers is concordant with the
UK study, with the following exceptions: in contrast to the study of
Foucard and Malmheden Yman, no cases of fatal
soy-ingestion-related asthma have been identified in the UK (allergy to
soy protein is rarely seen in our clinics), nor has fatal asthma caused by
sulphites in wine been identified. Exercise
was not an obvious precipitant in any UK fatality (apart from a single
recent possible exception), although it may be a common co-factor for
Studies of Non-Fatal Anaphylaxis in the UK
of anaphylaxis detected will differ according to the research method used.
A recent analysis of diagnostic codes in a UK general practice research
database estimated an incidence of 8.4 per 100 000
person-years, with insect stings and oral medicines the predominant
causes. A different, more likely diagnosis is found in as many as half the
cases referred from general practice to our clinics with suspected
anaphylaxis, and compared with that report, we found a higher ratio of
food-induced acute allergic reactions to those caused by stings and drugs.
Patients seek help from different parts of the health service according to
their preference, and the circumstances and nature of their reactions
(Fig. 1). This will also be reflected in the statistics collected from
Figure 1. Each reaction of patients attending clinic was
graded for severity by the method described by Pumphrey and Stanworth.
This chart analyses the relationship between the severity of reaction
and where the patients went for emergency medical help. There were
also time trends, such as the marked fall in those seeking treatment
from general practitioners over the past few years. All of those
treated by paramedics and some of those first treated by general
practitioners then went to accident and emergency departments for
further treatment of observation.
see patients with intellectual, cognitive and memory impairment after
cerebral anoxia during anaphylactic reactions or bolus intravenous
epinephrine (adrenaline) given to conscious patients. The rate of such
persistent sequelae must be higher than for fatality, but I can find no
are not admitted to hospital after emergency medicine treatment for
anaphylaxis, so an analysis of hospital discharge codes carries a risk of
bias to the more slowly evolving forms of 'reaction' for which an
alternative diagnostic code might have been more accurate. With this
caveat in mind, the studies of Sheikh and Alves[12,13] showed a
rapid increase in the rate of anaphylaxis between 1991 and 1995, and a
predominance of females from rural, southern and affluent areas. However,
I have found no significant change in the number of deaths each year
attributed to anaphylaxis in the UK, the past 12 months reflecting the
pattern for the previous 11 years. The steady UK death rate from
food-induced asthma differs from the marked decrease in fatal asthma
reported by Formgren et al.
Mode of Death
In contrast to survivable allergic reactions, symptoms in fatal
anaphylaxis are commonly limited to one system: deaths were divided
equally between shock and respiratory arrest. Shock in some (mostly
younger people with healthy hearts) was caused by vasodilation with volume
redistribution, leading to pulseless electrical activity when no blood
returned to the heart. Death in this group was associated with an upright
posture.[15*] In other cases (mostly older people with
pre-existing myocardial pathology) shock was caused by dysrhythmia,
possibly associated with the local release of anaphylactic mediators in
the myocardium. Electrocardiogram changes of myocardial
ischaemia are typically found in anaphylactic shock, leading to diagnostic
Respiratory arrest was either caused by severe bronchospasm (often in
those taking daily treatment for asthma) or upper airway angioedema; for
some, there was evidence of both these processes. Compared with stings and
drug-related anaphylaxis, fatal food allergy was much more commonly caused
by asthma and much less often caused by shock; the association of asthma
with severe reactions to nuts is well recognized. Asphyxia
caused by upper airway swelling was more common in food and sting
reactions than in those caused by drugs (Table
Table 2. Mode of Death
Rate of Reaction
The interval between eating the food and fatal collapse has usually been
25-35 min (range 10 min-6 h, the longest being a case with vigorous early
treatment including epinephrine for mild symptoms who had a sudden rapid
deterioration at 6 h). This is longer than for drugs (median 5 min) or
stings (median 12 min). Some were resuscitated after this collapse but
died later (median 60 h); such resuscitation was most common after
iatrogenic reactions (35%) and least common with sting deaths (16%),
presumably a direct consequence of accessibility to medical management of
Genetic, environmental and ontogenic factors influence the nature and
severity of reactions.
Fatal reactions caused by cows' milk allergy were mostly in children
(youngest 5.5 months, median age 8 years), by peanuts in young adults
(youngest 13 years, median 21), and by tree nuts in adults (youngest 18
years, median 27). The remaining food allergies affected adults (18-53
years, median 25); there were too few to analyse each allergen
individually. Overall, fatal food allergy affected younger people than did
fatal sting or drug allergy (median 21 years compared with 56 or 61 years,
Seventy-five per cent of UK nut-allergic fatalities were female, but for
other foods there was an equal number of males and females. Recent figures
from the USA have suggested an equal male: female ratio for fatal nut
allergy. Fatal contrast media reactions and sting reactions
were more common in males, concordant with previous studies.[19,20]
A variety of racial groups was represented in the cases studied. The
numbers were too low to assess whether the proportions differed from
racial prevalence in the UK. An unexpectedly high proportion of UK milk
allergy fatalities were boys of Asian or mixed race.
Cases occurred in all regions of the UK, in proportion to the population,
apart from a lower incidence in Scotland, probably reflecting the
different system of reporting deaths in that country. Diagnostic coding
suggested a higher incidence of non-fatal reactions in the south of
England and rural areas.
A majority of fatal reactions attributed to food allergy were principally
asthmatic reactions in those taking daily treatment for asthma. Data for
atopy were not so good for sting and iatrogenic deaths as for food
allergy. It is probable that there was no strong correlation between atopy
and fatal reactions for these other causes. Specific IgE to foods and wasp
venom were in some cases low (e.g. 6 KUA/l to peanut, with a
total IgE of 70 KIU/l) and in some high (e.g. >500 KUA/l to
milk, with a total IgE of 4000 KIU/l). There were too few measurements to
detect any pattern.
What Made the
combination of factors may have conspired to make the reaction fatal,
including the severity of the allergy, a high dose of potent allergen,
underlying genetic tendencies for angioedema, hypotension
and bronchospasm (recognized by previous idiopathic or allergen-triggered
events), and concurrent disease or medication
synergizing with these genetic tendencies. The recorded treatment was
optimal in some cases, suboptimal in others, and some were untreated.
Although the accuracy of the records was a point of discussion during
individual case reviews, it is probable that even with optimal management
some cases of anaphylaxis will be fatal.
have been reported to cause anaphylaxis, but far fewer have been
associated with fatal reactions. It seems the latter may be intrinsically
more dangerous; for example, nuts seem to be the most dangerous food and
amino-beta-lactams such as amoxicillin and cefaclor the most dangerous
antibiotics. The probable quantity ingested of the suspected food varied
from trace contamination to over 100 g. The distribution was approximately
log normal, with modal values of 1 g for nuts and 10 g for other foods
(Fig. 2). Most sting deaths followed single stings, and medication was
usually the standard therapeutic dose; however, death was so rapid in many
of these that it is likely that a much smaller dose might have been fatal.
The threshold for IgE-mediated reactions to protein allergens is log
normally distributed over 5-7 orders of magnitude.[24**] It
seems possible that the threshold dose for anaphylaxis may be similarly
distributed, with a similar but higher distribution of thresholds for
fatal reactions. The distribution of rates for fatal reactions may reflect
the amount by which the dose given exceeded the threshold for anaphylaxis.
Figure 2. Estimated amount eaten (dry weight) of the food
ingredient supposed to have caused the fatal allergic reaction
Asthma has been
a major component of fatal reactions in children, and for most fatal
reactions to foods. Daily treatment for asthma in some cases has been
suboptimal. The evidence suggests that children with mild reactions before
developing asthma had fatal reactions not because their allergy had
increased, but because their airways were affected by asthma.
had recurrent angioedema symptoms, thought to be idiopathic until their
fatal attack of upper airways angioedema attributed to food allergy.
Antihypertensive drugs such as angiotensin-converting enzyme inhibitors or
beta-blockers may be a co-factor for anaphylaxis, particularly noted
during insect venom hyposensitization. Other series have
reported the use of beta-blocker antihypertensive drugs and underlying
coronary artery pathology in those dying from sting reactions,
but this pattern was not seen in UK sting deaths. Fatal food allergy
mostly affects young people, so only two out of 70 patients had been
taking such drugs. Angiotensin-converting enzyme inhibitors may by
themselves cause life-threatening angioedema affecting the upper airway;
in three UK cases this was fatal. Under the new European Academy of
Allergy and Clinical Immunology's revised definition of anaphylaxis,
this would be classified as non-allergic anaphylaxis.
was present in one sting death and one fatal anaesthetic reaction. The
association between mastocytosis, raised basal mast cell tryptase levels
and reactions to hymenoptera venom is well documented.[22,27]
The severity of reactions to wasp venom but not bee venom was higher in
the group with raised basal tryptase in one study, but this was not
reported in the other.
may synergize with muscle relaxants in anaesthetic reactions: antibodies
to each cross-react and solid-phase-bound morphine can be used to detect
IgE antibodies to suxamethonium. Alcohol or aspirin and
exercise or concurrent infection may synergize with food allergy to
increase the severity of the reaction. Some reactions are catamenial.
Food-allergy-induced shock has only been fatal when patients have been
made to stand up.[15*]
epinephrine treatment is needed to prevent asphyxia from upper airway
oedema; tracheostomy has been overlooked as a treatment option when
epinephrine was used too late. Some of the deceased had been provided with
an epinephrine auto-injector but failed to use it, often because it had
been left elsewhere. Others used their epinephrine without benefit.
Epinephrine was given to eight of the UK food allergy fatalities within 30
min of ingestion but they still died, in contrast to the series reported
by Sampson et al.. Death followed inappropriate
treatment given for minor reactions in three cases when the effects of
epinephrine overdose were misinterpreted as a worsening of the reaction,
leading to further over-treatment.
reviewed above demonstrates the difficulty in identifying who may be at
risk of a fatal reaction. Different strategies are required for the
prevention of fatalities in those who have demonstrated their allergy and
those who so far have not.
reaction was thought to be the first in the case of 23 out of 34 sting, 17
out of 24 antibiotic, 17 out of 20 muscle relaxant, four out of six
non-steroidal anti-inflammatory drug, 13 out of 13 other drug, and 10 out
of 10 contrast media-related anaphylactic deaths, although most patients
had been stung or exposed to the causative drug previously without
reaction. There are no tests with adequate sensitivity and specificity to
indicate who might be at risk of a fatal reaction; because of the
million-fold range of threshold dose and the sometimes steep escalation in
the severity of reaction with even small increases of dose above the
threshold, challenge testing with a reduced dose is unlikely to warn of
anaphylaxis without causing it.
drug-induced fatal reactions received immediate expert management but
because of their rapidity, this management was restricted to attempted
resuscitation rather than prevention of the progression of the reaction;
some first-time fatalities are therefore unavoidable. For other cases, the
treatment was not optimal; it is essential that all those carrying out
interventions associated with anaphylaxis are aware of the correct
immediate management of anaphylactic reactions.
commonly think their reactions will be more severe each time they are
exposed; we could find no evidence that this was the case.
In a series of reactions, the most severe was as commonly the first, an
intermediate one or the most recent. Mullins[4**] reported that
only a quarter of those with a recurrence had a reaction more severe than
previously. However, severe previous reactions were associated with a
severe recurrence, one in six needing epinephrine or emergency medical
care. Previous studies of severe venom reactions could find no
characteristics that would allow the prediction of severe recurrence.
Those dying from food allergy had usually had previous reactions, but
these were typically not severe, indicating that the severity of previous
reactions is not a risk factor for a fatal reaction to nuts (Fig. 3).
Figure 3. The most severe reaction of patients attending
clinic and the most severe pre-fatal reaction of those who
subsequently died from their food allergy were graded for severity by
the method described by Pumphrey and Stanworth. There
is a tendency for older patients to have experienced a more severe
worst reaction than young patients, but the severity of the deceased's
worst pre-fatal reaction gave no clue that they would later die from
None of the 11
UK sting fatalities with a known history of previous reactions had been
given venom immunotherapy.
avoidance failed for a variety of reasons: a few were tragic accidents
when a patient with a known drug allergy was given the fatal dose. It
should be noted, however, that such accidental administration of
penicillins to those with 'known penicillin allergy' very rarely leads to
a reaction. Recurrent reactions were commonest among those with food
allergy. Very few had been given expert advice about avoidance, and
although there is a view that recurrence is unavoidable,
others consider that good advice can greatly reduce patients' risk of
recurrence. Commercial catering caused 68% of nut-related
reactions. Asking for a meal without nuts proved an unsuccessful avoidance
strategy because neither the person serving nor in some instances the
caterer realized that the food contained nuts. Peanut flour in Indian
commercial catering has caused confusion, as those preparing the food may
not recognize this as 'nuts'. For other foods, the source was distributed
more evenly between whole foods, domestic cooking, commercial cooking and
with steroid therapy and the overuse of inhaled short-acting beta-2
agonists is associated with fatal asthma; this is true also for
food-allergy induced asthma. The high daily use of a short-acting beta-2
agonist may lead to epinephrine resistance, making it difficult to treat
reactions; optimal asthma control is important for those
with asthma and food allergy. Asthmatic female students living away from
home seem to be at particular risk from their nut allergy.
The author is
grateful to Colin Summers for help in extracting statistics from this
of the clinic database has been supported partly by the Food Standards
Authority grant ZT0702.
Can We Tell Who is at Risk of a Fatal Reaction?
Allergy Clin Immunol 4(4):285-290, 2004. © 2004 Lippincott Williams &
particular interest, published within the annual period of review, have
been highlighted as:
*of special interest
**of outstanding interest
Pumphrey RSH. Lessons for management of anaphylaxis from a study of fatal
reactions. Clin Exp Allergy 2000; 30:1144-1150.
Pumphrey RSH. Fatal anaphylaxis in the UK, 1992-2001. 2004
Anaphylaxis. Chichester: Wiley (Novartis Foundation Symposium 257); 2004.
** A large population-based study of fatal anaphylaxis with suggestions
for improved management.
Pumphrey RS, Roberts IS. Postmortem findings after fatal anaphylactic
reactions. J Clin Pathol 2000; 53:273-276.
Mullins RJ. Anaphylaxis: risk factors for recurrence. Clin Exp
Allergy 2003; 33:1033-1040.
** A useful population-based study of anaphylactic reactions with
Yocum MW, Khan DA. Assessment of patients who have experienced
anaphylaxis: a 3-year survey.
Mayo Clin Proc 1994; 69:16-23.
Emmett SE, Angus FJ, Fry JS, Lee PN. Perceived prevalence of peanut
allergy in Great Britain and its association with other atopic conditions
and with peanut allergy in other household members.
Allergy 1999; 54:380-385.
Foucard T, Malmheden-Yman I. A study of severe food reactions in Sweden -
is soy protein an underestimated cause of food anaphylaxis?
Allergy 1999; 54:261-265.
Tsevat J, Gross GN, Dowling GP. Fatal asthma after ingestion of sulfite-containing
wine. Ann Intern Med
T, Yoshizawa I, Ogawa N, et al.
dependent exercise-induced anaphylaxis.
Asian Pac J Allergy Immunol
Peng MM, Jick H. A population-based study of the incidence, cause and
severity of anaphylaxis in the United Kingdom.
Arch Intern Med 2004;
Pumphrey RS, Stanworth SJ. The clinical spectrum of anaphylaxis in
north-west England. Clin
Exp Allergy 1996; 26:1364-1370.
Sheikh A, Alves B. Hospital admissions for acute anaphylaxis: time trend
study. BMJ 2000;
Sheikh A, Alves B. Age, sex, geographical and socioeconomic variations in
admissions for anaphylaxis: analysis of four years of English hospital
data. Clin Exp Allergy
Formgren H, Bergstrom SE, Boman G, et al.
deaths in young people be predicted and prevented?
[in Swedish]. Lakartidningen
2001; 98:5314-5315; 5318-5321.
Pumphrey RSH. Fatal posture in anaphylactic shock. J Allergy Clin
Immunol 2003; 112:451-452.
* This brief letter suggests a mechanism for pulseless electrical activity
and myocardial ischaemia in non-iatrogenic anaphylactic shock.
Marone G, de Crescenzo G, Adt M, et al. Immunological
characterization and functional importance of human heart mast cells.
Sicherer SH, Furlong TJ, Munoz-Furlong A, et al. A voluntary
registry for peanut and tree nut allergy: characteristics of the first
5149 registrants. J
Allergy Clin Immunol 2001; 108:128-132.
SA, Munoz-Furlong A, Sampson HA, et al.
to anaphylactic reactions to foods.
J Allergy Clin Immunol 2001; 107:191-193.
Lantner R, Reisman RE. Clinical and immunologic features and subsequent
course of patients with severe insect-sting anaphylaxis.
J Allergy Clin Immunol 1989;
Lenler-Petersen P, Hansen D, Andersen M, et al. Drug-related fatal
anaphylactic shock in Denmark 1968-1990. A study based on notifications to
the Committee on Adverse Drug Reactions.
J Clin Epidemiol 1995;
Hermann K, Ring J. The renin-angiotensin system in patients with repeated
anaphylactic reactions during hymenoptera venom hyposensitization and
sting challenge. J Int Arch Allergy Immunol 1997; 112:251-256.
Haeberli G, Bronnimann M, Hunziker T, Muller U. Elevated basal serum
tryptase and hymenoptera venom allergy: relation to severity of sting
reactions and to safety and efficacy of venom immunotherapy.
Clin Exp Allergy 2003;
Ober AI, MacLean JA, Hannaway PJ. Life-threatening anaphylaxis to venom
immunotherapy in a patient taking an angiotensin-converting enzyme
inhibitor. J Allergy Clin Immunol 2003; 112:1008-1009.
Bindslev-Jensen C, Briggs D, Osterballe M. Can we determine a
threshold level for allergenic foods by statistical analysis of published
data in the literature? Allergy 2002; 57:741-746.
**The concept introduced in this paper of a log-normal population
distribution of threshold dose for reactions,generally applicable for
protein allergens,extending over 5 -7 orders of magnitude,is of
Sasvary T, Müller U. Fatalities from insect stings in Switzerland
1978-1987 [in German]. Schweiz Med Wochenschr 1994; 124:1887-1894.
Johansson SG, Hourihane JO, Bousquet J, et al. Revised nomenclature
for allergy. An EAACI position statement from the EAACI nomenclature task
force. Allergy 2001;
Ludolph-Hauser D, Rueff F, Fries C, et al. Constitutively raised
serum concentrations of mast-cell tryptase and severe anaphylactic
reactions to Hymenoptera stings.
Lancet 2001; 357:361-362.
Baldo BA, Pham NH, Zhao Z. Chemistry of drug allergenicity.
Curr Opin Allergy Clin Immunol 2001; 1:327-335.
Snyder JL, Krishnaswamy G. Autoimmune progesterone dermatitis and its
manifestation as anaphylaxis: a case report and literature review.
Ann Allergy Asthma Immunol 2003;
Sampson, Mendelson L, Rosen JP, et al.
near-fatal anaphylactic reactions to food in children and adolescents.
N Engl J Med 1992;
Sicherer SH, Burks AW, Sampson HA. Clinical features of acute allergic
reactions to peanut and tree nuts in children.
Pediatrics 1998; 102:6.
Available at: http://www.pediatrics.org/cgi/content/full/102/1/e6
Vickers DW, Maynard L, Ewan PW. Management of children with potential
anaphylactic reactions in the community: a training package and proposal
for good practice. Clin Exp Allergy 1997; 27:898-903.
Pumphrey RS, Nicholls JM. Epinephrine-resistant food anaphylaxis.
Lancet 2000; 355:1099.
You may want to print this webpage for
future use or you may want to provide ithis information to others.
By clicking on "Print"below a new window
wil open and a pdf file will open.
you will need to have Acrobar Reader® installed on your computer to view
and print he file. if you don't have it you can get it