Introduction

Anaphylaxis is the most severe manifestation of immediate hypersensitivity with an incidence of 0.05 to 2% in general population [25]. Anaphylaxis concerns 4.5 to 18 admissions on 10,000 in pediatric emergency departments (EDs), with a trend to increase during the last years [14, 15]. In France, the anaphylaxis-related mortality rate in children is 0.83 per million per year [19]. A rapid clinical diagnosis, based on Sampson’s criteria [23], is required for administering epinephrine as quickly as possible in accordance with the different recommendations [16, 26]. However, many studies underline constant under-use of epinephrine in children with anaphylaxis not only in pre-hospital but also in ED management [6, 11, 21, 24, 30]. One possible explanation for this gap between recommendations and practice could be a kind of “intuitive” adaptation of epinephrine delivery according to the severity of symptoms. This study aimed to analyze pediatricians’ epinephrine prescription as a function of clinical severity during a 6-year period in children admitted for anaphylaxis in the two pediatric EDs of the University Hospitals of Marseille, France.

Methods

Patients

Design and setting

This was a retrospective study performed in the two pediatric EDs of the Assistance Publique-Hôpitaux de Marseille, France, between January 2010 and December 2015. These EDs received patients up to 18 years.

Data source and inclusion criteria

In France, for each patient admitted to an ED, an Electronic Emergency Department Abstract (EEDA), directly available from the patients’ computerized medical files, is anonymously transmitted to the French Institute of Public Health Surveillance. EEDAs report on date and hour of ED admission, age, sex, disposition after ED visit (home discharge, hospitalization, death), and final diagnosis according to the national standardized thesaurus based on the International Classification of Diseases, Tenth Revision (ICD-10), codes were collected. We also analyzed the clinical severity appreciated by the triage nurse scale upon admission (high levels of severity were levels 1–2), which was therefore not specific for anaphylaxis.

Visits related to “anaphylaxis” (“anaphylactic shock” or “Quincke edema”), “urticaria,” and “other allergic reaction” were defined using grouping of ICD-10 codes. The medical files of all patients with an ICD-code related to anaphylaxis were reviewed. Patients who met Sampson’s criteria were considered to have confirmed anaphylaxis. Patients were excluded if Sampson’s criteria were not fulfilled. The anaphylaxis severity was defined by the Ring and Messmer classification (grade I: mucocutaneous symptoms; grade II: moderate multivisceral failure; grade III: severe mono/multivisceral failure; grade IV: cardiac arrest) [20]. Medical history, anaphylaxis features, and management (biological tests, treatments, planned allergist visit, and auto-injectable epinephrine prescription) were noted for each patient.

Statistical analysis

Visits related to anaphylaxis, urticaria, and other allergic reaction were compared for age, sex, severity, and hospitalization rate. A descriptive analysis of the admissions for anaphylaxis according to Sampson’s criteria was performed. Comparison of the data according to anaphylaxis severity rank was realized. Variables were described as percent, and mean (standard deviation). Associations were tested with simple logistic regression for qualitative variables and quantified by unadjusted odd ratio (OR) and their confidence interval. Quantitative variables were compared with a Student or Fisher test depending on the number of data. Analysis was performed with SPSS version 20.0® (SPSS Inc., Chicago, IL, USA). A p value of 0.05 or less was considered as statistically significant.

Ethical considerations

This study was approved by the Ethics Committee of the French Society of Pediatrics (Société Française de Pédiatrie, No. 2016_018_2) and the National Commission of Informatics and Liberty (Commission Nationale Informatique et Libertés, No. 2016-16).

Results

Admissions with a diagnosis related to allergy

During the study period, 422,483 admissions took place (Fig. 1). The diagnosis discharge was missing for 18,029 (4.2%). Overall, 4141 visits were coded urticaria (102.13/10,000 admissions), 1358 other allergic reaction (33.49/10,000), and 235 anaphylaxis (5.76/10,000).

Fig. 1
figure 1

Flowchart of the study evaluating anaphylaxis episodes in two French pediatric emergency departments during the 2010–2015 period

Full size image

Compared with visits coded urticaria and to other allergic reaction, visits coded anaphylaxis concerned older children, more severe patients (according to the triage scale), and had a higher hospitalization rate (Table 1). Under 12 years, the rate of admissions for anaphylaxis was higher in boys than in girls (5.27/10,000 vs 3.25/10,000, p = 0.006). Conversely, above 12 years old, this rate was higher in girls (16.25/10,000 vs 9.25/10,000, p = 0.011).

Table 1 Comparison of the coded admissions characteristics for anaphylaxis, urticaria, other allergic reactions, and other diagnosis in two French pediatric emergency departments during the 2010–2015 period
Full size table

Admissions with confirmed anaphylaxis

History and clinical presentation

Among visits related to anaphylaxis, 204 (86.8%) fulfilled Sampson’s criteria (Fig. 1 and Table 2), underlying a high predictive value of our coding for correct diagnosis. The mean age was 7.9 years (2 months to 17.7 years) with a sex ratio of 1.27 (n = 114 boys, 55.9%). A history of allergy was reported in 124 files (60.8%). When a previous episode of anaphylaxis was reported (n = 75; 36.8%), an epinephrine auto-injector was available at home. Asthma was reported in 61 files (35.1%), with a prophylactic daily treatment in 24 (39.3%). A total of 174 different children were concerned.

Table 2 Main characteristics of 204 pediatric admissions to two French emergency departments (EDs) for confirmed anaphylaxis (clinical severity was classified according to the Ring and Messmer criteria)
Full size table

The current anaphylactic episode mainly occurred in private homes (n = 63/111, 56.7%) or schools (n = 32/111, 28.8%), during or after food intake (n = 122/155, 78.8%). Initial symptoms were mucocutaneous (n = 200/204, 98.5%), respiratory (n = 154/204, 75.7%), gastrointestinal (n = 47/204, 23%), cardiovascular (n = 23/204, 11.3%), or neurological (n = 21/204, 10.3%). No cardiac arrest or death was noted.

In 67.2% of admissions, patients self-presented to the ED. At ED admission, 170 cases (83.3%) were classified grade II according to the Ring and Messmer classification and 34 grade III (16.7%). A causal allergen was suspected in 129 cases (63.2%): food allergen (n = 101, 78.3%, with nuts n = 42, 32.5%; fruits n = 17, 16.8%; milk n = 15, 14.9%; egg n = 6, 5.9%; others n = 21, 16.3%), drugs (n = 11, 8.5%), insect venom (n = 6, 4.7%), and other allergens (n = 11, 8.5%). Specific IgEs were measured in 28 cases (13.7%) and tryptase in 42 cases (20.6%).

Treatment

Data concerning treatments administered before arrival at the EDs were available for 201 episodes. Intramuscular epinephrine was administered in 23 cases (11.4%). Admissions with recurrent anaphylaxis received epinephrine fourfold more often than others (16/75 vs 7/126, 21.3% vs 5.6%). Epinephrine was mainly injected by family or school staff (14/23, 60.9%). Other treatments included antihistamines (n = 93/201, 46.3%), steroids (n = 83/201, 41.3%), inhaled beta-2 agonists (n = 26/201, 12.9%), or inhaled epinephrine (n = 5/201, 2.5%).

Data on ED treatments were available for 202 episodes. Intramuscular epinephrine was administered in 45 cases (22.3%). Overall, 65/199 (32.7%) children received epinephrine before arrival or in the ED. The epinephrine injection rate was higher for grade III than for grade II anaphylaxis (84.8% vs 22.3%, p < 0.001; OR = 19.05 [7.05–54.10]). Epinephrine was also more often prescribed when the patients were classified as urgent by triage nurses (level 1 and 2), but in a less important proportion (47.8% vs 24.6%; p < 0.001; OR = 2.81 [1.51–5.21]). Other treatments included antihistamines (n = 156/202, 90.7%), steroids (n = 126/202, 73.3%), inhaled beta-2 agonists (n = 25/202, 14.5%), inhaled epinephrine (n = 10/202, 5.8%), oxygen (n = 9/202, 5.2%), and vascular filling (n = 8/202, 4.7%). The length of stay in the ED was longer (378 min versus 197 min; p < 0.001), and the hospitalization rate higher (76.2% versus 34.1%, p < 0.001) in case of epinephrine administration. To note, all hospitalized children did not received epinephrine (rate of administration 52.7%; n = 48/91). When the child was discharged at home (n = 104/204), an epinephrine auto-injector was prescribed in 33 (31.7%) cases, and an allergist visit planned in 46 (44.2%) cases, solely (Table 2).

Discussion

To our knowledge, this is the first study reporting the link between epinephrine administration and the severity of pediatric anaphylaxis. Our study also shows that epinephrine is underused during pediatric anaphylaxis management. Among the ED admissions that met Sampson’s anaphylaxis criteria, only one third received epinephrine before or during ED management.

Our study has several limitations due to its retrospective design. First, inclusion criteria (Sampson’s criteria) and Ring and Messmer classification are not, in daily practice, pre-registered in the medical record, but they have been deducted by investigators from the medical data. However, in our institution, Sampson’s criteria and Ring and Messmer classification are specified in the institutional protocol for anaphylaxis management, are well-known to pediatricians, and are well reported by physicians in the medical files. Also because of the retrospective design of the study, many data that would have been relevant to explain the underutilization of epinephrine, such as the specific delay between the onset of symptoms and ED visit, are not available. From this point of view, a qualitative approach exploring the reasons given by pediatricians to justify the absence of adrenaline prescription would also be relevant.

To minimize some bias, and in accordance with other studies [7, 12, 13, 22], we have carefully selected our cases of anaphylaxis following Sampson’s criteria. First, we have compared the visits for possible allergic reactions (urticaria and “nonspecific allergy”) to those for anaphylaxis, and we have shown that visits for anaphylaxis clearly concerned more severe patients as assessed by the clinical score value and the hospitalization rate. Second, when analyzing all the files coded as anaphylaxis, we have observed that anaphylaxis was correctly coded in 86.8% of the cases, better than the 63.1% figure reported by Walsh in the same context [29]. In another study in Australian children, ED diagnosis of anaphylaxis had a sensitivity of 43.2% and specificity of 97.9% [27].

Admissions with confirmed anaphylaxis had the same clinical characteristics as these described in the most recently published studies [10, 21, 24, 30]: age around 6–7 years, boys, mainly mucocutaneous symptoms, and food trigger. Our results also confirm the female predominance of anaphylaxis in children older than 12 years [1]. In keeping with the European Anaphylaxis Registry [11], a previous anaphylactic episode was found in about one third of current admissions for anaphylaxis. The utility of assessing tryptase and specific IgE is still a matter of debate [8], and these laboratory tests were seldom prescribed during the ED stay. Our rate of hospitalization was in agreement with the literature, between 36 and 54% [9, 15].

We have retrospectively classified anaphylaxis episodes according to the Ring and Messmer severity classification. Grade III cases (16.7% of our patients) presented well-known risk factors of severe anaphylaxis, such as asthma or allergy [3], more frequently than grade II cases. They also had higher blood tryptase levels and rate of hospitalization. Others have defined the anaphylaxis severity on the basis of anaphylaxis outcome [10, 14], estimating at 13% to 34.3% the part of severe anaphylaxis. Using this definition, severe anaphylaxis has been linked to drug allergy [10], or fewer administrations of epinephrine [14].

Current treatment recommendations for anaphylaxis highlight a prompt intramuscular epinephrine injection as the gold standard allowing to reduce morbidity, mortality, and hospitalization [16, 26]. However, the use of epinephrine is still insufficient in almost all pediatric studies. Prehospital administration of epinephrine varies in the literature from 36 to 62% [2, 6, 21, 28]. Anaphylaxis occurring at school and a one-organ system involvement would be favoring factors for use [21]. Our rate of ED administration of epinephrine is similar to that described in other European pediatric EDs that has nevertheless doubled between 2011 and 2014 [11]. In USA, EDs’ epinephrine prescription concerns about one admission among three for the first episode [24, 30], increasing to two admissions among three in case of recurrence [17]. In Asia, epinephrine is given in about 90% of the admissions for anaphylaxis [10, 14]. Such gaps between medical epinephrine prescriptions in EDs may reflect inclusion of different phenotypes of anaphylaxis and/or different prehospital management attitudes among countries.

Our results suggest that pediatricians had an intuitive approach of severity and reserved epinephrine for the most severe admitted children (grade III). This medical behavior ascertainment may be explained by the delay between the ED admission/management and the episode onset, and maybe also by a relative fear of the epinephrine side effects, even if its safety has been clearly demonstrated (21.6% of usually mild and transient side effects, such as tremors, palpitations, anxiety, more frequent among adults) [5]. Indeed, if anaphylaxis signs lowered or disappeared before ED admission, practitioners might have considered that the epinephrine administration was useless. The absence of hemodynamic or respiratory dysfunctions might also contribute to under-prescription as it has been previously shown on the management of a 2-year-old child with peanut-induced anaphylaxis where one in five interviewed pediatricians reported not using epinephrine because of this reason [18]. It also could explain why 15% of grade III anaphylaxis episodes had not received epinephrine in our study. Our data do not allow us to determine whether the behavior of our pediatricians induces more hospitalization, but a lack of epinephrine prescription in nearly half of the hospitalized patients is clearly amazing and questioning. Currently, to improve epinephrine prescription in EDs, the use of auto-injectors should be encouraged. Easy to use, they could also favor family education and encourage a too low medical prescription upon home discharge, as also found in other studies [24, 30]. However, auto-injectors have therefore some limitations such as a too high dose (0.15 mg) for the lightest children, especially < 7.5 kg [26]. In addition, auto-injectors with self-retracted needle should be clearly preferred in order to reduce the risk of lacerations and embedded needles [4]. Finally, better medical education on allergy is clearly required. Indeed, recent attendance of continuing medical education on food allergy induces significantly more epinephrine prescription [18].

Conclusion

This study shows that epinephrine was underused in EDs face to children with anaphylaxis, but that its prescription was intuitively adapted to the clinical severity. Auto-injector devices may enhance the epinephrine use in EDs, as well as family education.