Use of multiple epinephrine doses in anaphylaxis: A systematic review and meta-analysis

Background Regulatory bodies recommend that all patients at risk of anaphylaxis be prescribed 2 epinephrine autoinjectors, which they should carry at all times. This is in contrast to some guidelines. The proportion of anaphylaxis reactions that are treated with multiple doses of epinephrine has not been systematically evaluated. Objective Our aim was to undertake a systematic review and meta-analysis of published studies reporting epinephrine treatment for anaphylaxis in which data relating to the number of doses administered were available. Methods We searched the Medline, Embase, and Cochrane databases for relevant studies reporting at least 10 anaphylaxis events (due to food or venom) from 1946 until January 2020. Data were extracted in duplicate for the meta-analysis, and the risk of bias was assessed. The study was registered under the PROSPERO identifier CRD42017069109. Results A total of 86 studies (36,557 anaphylaxis events) met the inclusion criteria (20 of the studies [23%] were prospective studies; 64 [74%] reported reactions in the community, and 22 [26%] included food challenge data). Risk of bias was assessed as low in 50 studies. Overall, 7.7% of anaphylaxis events from any cause (95% CI = 6.4-9.1) were treated with multiple doses of epinephrine. When only epinephrine-treated reactions for which subsequent doses were administered by a health care professional were considered, 11.1% of food-induced reactions (95% CI = 9.4-13.2) and 17.1% of venom-induced reactions (95% CI = 11.3-25.0) were treated with more than 1 epinephrine dose. Heterogeneity was moderate to high in the meta-analyses, but at sensitivity analysis this estimate was not affected by study design or anaphylaxis definition. Conclusion Around 1 in 10 anaphylaxis reactions are treated with more than 1 dose of epinephrine.


Search Strategy and Eligibility/Inclusion Criteria
We searched MEDLINE, EMBASE, Cochrane Controlled Register of Trials, from launch to July 2019 for primary records referring to anaphylaxis reactions to food or venom triggers with data on the use of epinephrine. The full list of search terms for individual search databases with the records noted at each stage are listed in Table S1-S3. The search terms and strategies were reviewed by Imperial College London university library services prior to being run. The search was updated on 18 Jan 2021.
Reference lists were checked for additional publications where relevant. All primary data-records were included where at least 10 cases of anaphylaxis were reported. Where abstracts from peerreviewed journals met eligibility criteria, data were included. Original studies from case series, registries, food challenge datasets and surveys (where health-professional review of the anaphylaxis diagnosis was apparent) were eligible. Reviews (non-original) data were excluded, although reference lists were cross-checked to include any relevant records not found via the search strategy. Where multiple publications were available with overlapping datasets, a single publication was included and authors were contacted to provide outcomes for the overall dataset where needed. Where extrapolative modelling had been performed on a sample dataset, studies were excluded unless complete data were also available on the sample, unmodelled data. Datasets exclusively focussed on cases of fatal anaphylaxis were excluded.

Participants/Population included:
Inclusion criteria: All age ranges, genders, ethnicities, country of study.

Intervention/Exposures
Inclusion criteria: We included studies where food or venom (hymenoptera) were listed (not exclusively) as known triggers in any case series. This included datasets where other triggers (eg. drug, unknown) were also included. We decided to focus on food and venom reactions as they form the key indications for epinephrine auto-injector prescription to patients.

Exclusion criteria:
We excluded studies where there was no evidence that any food or venom anaphylaxis events were included in the study dataset. Datasets pertaining to immunotherapyrelated reactions were excluded.

Outcomes
Data was collected on the following reference and study characteristics: publication year, country of study, duration of data collection, prospective/retrospective, study design and patient characteristics, definition of anaphylaxis used.
Data was extracted on the following study outcomes: By any trigger, food-only reactions, venomonly reactions): Numbers of allergic reactions, number of anaphylaxis reactions as defined by study authors, number of cardiorespiratory reactions, epinephrine use, use of multiple epinephrine doses, use of multiple epinephrine doses where any repeat doses were administered by a health-care professional, admissions to hospital, intensive care and fatalities and salbutamol use.   The search terms and records returned in the initial and updated search are listed below.
"Adrenalin" in Title Abstract Keyword OR "adrenaline" in Title Abstract Keyword OR "anaphylaxis" in Title Abstract Keyword -with Cochrane Library publication date Between Jan 1900 and May 2019 (Word variations have been searched). Generates: Initial search: 34 Cochrane reviews 5260 trials 1 Cochrane protocol Additional records in updated search: 5 Cochrane reviews 370 trials 0 Cochrane protocol

B. Number of Datasets and Events included in each Metanalysis
The number of datasets and number of allergic reactions (denominator) in each Metanalysis are listed below.    Tables   For all included references the table below lists the risk of bias assessment.   Below are the corresponding forest plots for the meta-analyses listed in Table S7.  Above are funnel plots for each of the analyses listed in Table S7. Below are the results of the corresponding Egger Tests.

G. Summary of Pooled Estimates for Sensitivity Analyses
The pooled estimates of each of the metanalyses conducted for the listed sensitivity analyses are summarised below. Data are presented by trigger and anaphylaxis definition as % (95% CI).

H. Record Flow: Excluded Records at Full-Text screening
The references reviewed at during the full-text screening, along with outcome and reason are listed below.