The panel of clinical toxicologists was made that comprised the research society for emergency antidotes stock and delivery system in Korea consisted of a diverse group of 13 professionals who represented various perspectives (Table 1). The principal investigator, who served as the chairperson, selected these experts on the basis of their clinical toxicology researches and professional experiences. The experts were first assigned particular antidotes to generate evidence-based summaries of the medical literature. They measured the usefulness of emergency antidotes for Korean poisoning patients. Each antidote was classified on the basis of its efficacy and the urgency with which it must be available when a case of acute poisoning presents at ED. The efficacy of a particular antidote was graded on the basis of its strength of evidence into one of three classes from class I, the strongest to class III, the weakest evidence. In terms of the urgency of availability, the antidotes were classified into three levels (A to C) on the basis of a WHO classification system (Table 2) (10). These panel members participated three times of Delphi survey for consensus of antidote classification among experts between August and October, 2009. Subsequently, a primary reviewer was allocated to each antidote. Each primary reviewer independently reviewed and revised the original antidote summary and created a standard summary to reassess each antidote. The primary reviewer could add articles to the original summary and provided a provisional antidote classification. The revised literature summary and the recommended classification were then presented to the entire panel. The panel deliberated in October, 2009. An iterative process was used to reach consensus regarding the efficacy and urgency of availability each antidote. Thus, after presentation of an antidote by the primary reviewer and discussion by the entire panel, a vote was taken to determine consensus. An antidote was recommended to be stocked if the panel consensus was affirmative for all questions. A consensus was defined as the full agreement of the eligible panel members. Each member could vote in one of three ways: agreement, disagreement, or strong disagreement. If one or more panel members expressed strong disagreement, the discussion was continued and final consensus was reached through the panel discussion.

In the absence of a Korean epidemiological data about nationwide acute poisoning, estimates of antidote utilization frequency were based primarily on a multicenter epidemiological study. We performed a multicenter epidemiological study to find the antidote utilization for acute intoxication patients in EDs between June, 2010 and May, 2011. A clinical toxicologist in each hospital was responsible for the data collection at each ED. To identify cases of acute poisoning, 228 ICD-10 codes that define acute intoxication or poisoning were used. Inclusion criteria of the patient was over 16 yr of age, had been admitted to the ED during the study period, the diagnosis was acute poisoning. The following cases were excluded: adverse reactions and secondary drug effects; chronic poisoning; insect bites; and inert foreign body ingestions. A specific report form was designed to record the following data: personal data (age and sex); time and day of poisoning; type of intoxication (suicide, intentional, illicit drug abuse, and other); type of toxic compound(s); type of medical assistance before admittance to the ED; severity of symptoms on admittance; treatment; and patient's outcome and condition on discharge. Clinical severity was measured by using the Poisoning Severity Score, which allows the simple stratification of patients on the basis of the clinical effects of poisoning (11). This score grades severity as follows: none (score 0), no symptoms or signs related to poisoning; minor (score 1), mild, transient, and spontaneously resolving symptoms; moderate (score 2), pronounced or prolonged symptoms; severe (score 3), severe or life-threatening symptoms; and fatal (score 4), causing death. The needed case of emergency antidotes were defined the patients who showed moderate-to-severe symptoms after acute poisoning at ED.

The quantities of antidotes that should be stocked at the EDs in Korea were estimated on the basis of the number of poisoned patients whose data were entered into NEDIS. It was assumed that the poisoned patients identified in the nationwide multicenter epidemiological study were an appropriate sample of the population. NEDIS is a nationwide information and standardization system that records emergency treatment data from EDs in Korea. The NEDIS gathered the clinical data of emergency patients at the most of moderate-to-large sized hospitals in Korea. The data of these patients are collected from the EDs via the web and entered into the central server of the national emergency medical center. The minimum quantities of the emergency antidotes that would meet the immediate needs of acute intoxication patients and that should be stocked at ED were calculated on the basis of frequency of use, as indicated by the epidemiological study. The minimum stocking amounts was estimated on the basis of the initial antidote amount that is needed for the dose of treating an adult patient.

The chi-square test was used to compare proportions. Standard descriptive statistics were reported as means and standard deviations. Means were compared by using the 2-tailed Student's t-test and the Mann-Whitney U-test. All calculations were performed by using the statistical software package SPSS 20.0 (Statistical Package for the Social Sciences, Chicago, IL, USA). Differences were considered to be statistically significant if the P value was less than 0.05.

The study protocol was reviewed by the institutional review board of Asan Medical Center (IRB No. 2011-0372). Due to the retrospective nature of the study, informed consent was waived.

The panel summarized the effectiveness of each antidote in clinical practice and its urgency of availability with the reviews of many articles and provisional recommendations (12, 13). Antidote effectiveness in practice was classified according to the solidity of the evidence showing effectiveness (Table 2). Thus, Class I evidence of effectiveness means that the effectiveness of the antidote is well documented; typically, it is clear that the antidote reduces the lethality of the poison. Class II evidence means that while the antidote is widely used, its effectiveness is not yet universally accepted because of a lack of research data. Class III evidence means its efficacy is questionable. The antidotes were also classified according to urgency of availability: the Class A antidotes should be immediately available (within 30 min); the Class B antidotes are required within 2 hr; and the Class C antidotes are required within 6 hr (Table 2). Using this classification system, the panel made recommendations regarding 36 antidotes could be effective at hospitals in Korea (Table 3).

The emergency antidotes that should be available immediately were determined by the consensus for the statue of recommendation. There were sixteen IA antidotes, namely, antidotes with well-documented effectiveness that are needed immediately. These were fomepizole (methylpyrazole), N-acetylcysteine, amyl nitrite, atropine, beta-blockers, calcium gluconate gel, digoxin-specific Fab antibody fragments, ethanol, folinic acid, glucagon, glucose (hypertonic), hydroxocobalamin, pralidoxime, sodium bicarbonate, sodium nitrite, and sodium thiosulfate. Four of the IA antidotes are used either in combination as a kit system (amyl nitrite, sodium nitrite, and sodium thiosulfate) or alone (hydroxocobalamin) to treat cyanide intoxication. Therefore, these three antidotes in a kit were considered as one antidote. There were four IIA antidotes, namely, antidotes that are immediately required but whose research data are insufficient or involve controversies regarding the universal usage of the drug. These were activated charcoal, diazepam, flumazenil, and neostigmine. Thus, the panel generated a list of 18 emergency antidotes that should be available immediately and should be stocked at EDs in Korea (Table 4).

The eleven EDs participated which were belonged to regional emergency medical centers or university hospitals in different regions of Korea (Fig. 2). There were three hospitals in Seoul and the Gyeonggi region, two in Gangwon, two in Daejeon and Chungcheong, two in Gyeongsang, one in Jeolla, and one in Jeju. Totally 4,870 cases were identified and these cases represented 0.6%-1.3% of the each hospital admitting patients at the 11 EDs during the study period. The mean age was 42 yr (range 16-95, standard deviation±21) and the most prevalent age was the forties (18.0%). Females accounted for 57.5% of the cases. Multiple toxic substances were involved in 21.9%. Regarding the type of intoxication, 66.8% were intentional self-intoxications, 28.4% were unintentional cases, 1.0% was suspected to be due to adverse drug events, and 3.8% were unknown.

Compared to the unintentionally poisoned patients, the intentionally poisoned patients were more likely to be older (45 vs. 37 yr, P=0.01) and female (58.7% vs. 54.5%, P=0.01). Many (58.9%) of the patients were poisoned at home and 28.0% were directly transported to the ED by the national 119 rescue services. The intentionally poisoned patients were transported to the ED more frequently by the national 119 rescue services after intoxication (31.2% vs. 21.3%, P<0.001). In 92.6% of cases, the route of poison administration was through oral ingestion. Oral route was more frequent in intentional patients (96.7% vs. 82.0%, P<0.001). There were two main categories of poisoning substances, namely, pharmaceutical products (53.8%) and agricultural products (22.7%). Chemicals, household products, plant poisons, and animal bites accounted for 9.9%, 7.5%, 5.6%, and 0.1% of the cases respectively. In 0.1% of the cases, the causative substance was unknown. The drugs that were most frequently involved were benzodiazepines (42.1%), acetaminophen (13.6%), anticholinergic sleep inducers (2.2%), tricyclic antidepressants (3.1%), beta-adrenergic blockers (1.7%), opioids (1.6%), methemoglobinemia-inducing agents (1.3%), and calcium channel blockers (0.5%). In the agricultural products, the most frequently involved substances were organophosphate insecticides (19.4%), dipyridylium herbicide (13.9%), glyphosate herbicide (6.8%), and carbamate insecticides (5.6%). In the household products, the most frequently involved substance was sodium hydroxide (28.8%) in cleaning products. In the chemical agents including gases, the most frequently involved agents were carbon monoxide (49.8%), cyanide (6.9%), hydrofluoric acid (3.7%), and toxic alcohols (3.5%). In the plant poisons and animal bites, the most frequently involved agents were mushrooms (28.5%) and snake bite (20.2%). Compared to the unintentionally poisoned patients, the intentionally poisoned patients were more likely to be poisoned by pharmaceuticals (57.2% vs. 45.8%, P<0.001) and agricultural products (26.8% vs. 13.1%, P<0.001).

In treatment, 55.2% of the patients were initially treated at the ED and the remaining 44.8% were managed previously by prehospital emergency medical services or primary care hospitals. On arrival at the ED, 8.7% of the patients were asymptomatic. Less asymptomatic patients were in the intentionally poisoned patients (6.4% vs. 14.8%, P<0.001). More than half of cases (53.1%) exhibited minimal toxicity but there was no significant difference (54.1% vs. 60.7%). In 22.3% of cases, the toxicity was moderate; this was more common in intentionally poisoned cases (25.7% vs. 16.9%, P<0.001). In 11.3% of cases, the toxicity was more severe in intentionally poisoned patients (13.3% vs. 7.3%, P<0.001). In total, 0.4% of the cases died. There was no significant difference between intentionally and unintentionally poisoned patients in terms of fatality rates (0.5% vs. 0.2%).

In 71.6% of the cases, the poisoning was managed by general supportive therapies. Another 32.0% was treated by gut decontamination, 23.6% received activated charcoal, and 2.1% underwent forced renal elimination. Only 4.7% of the patients actually received antidotes. These were N-acetylcysteine (n=49), pralidoxime and atropine (n=47), flumazenil (n=29), antivenin (n =25), methylene blue (n=14), atropine (n=4), naloxone (n= 3), vitamin K1 (n=3), high fractional oxygen inhalation (n=3), and others. Of the patients who received an antidote, 64.5% arrived with moderate to severe toxicity at the hospital and 3.1% died of moribund state at hospital discharge.

Of the all patients, 48.0% were discharged without toxicity symptoms from the EDs. 18.9% were admitted to the intensive care unit, while 48.2% were hospitalized. 3.2% (157 patients) died. Compared to the unintentionally poisoned group, the intentionally poisoned cases were significantly more likely to have death as outcomes (4.0% vs. 0.8%, P<0.001).

The 4,870 acute poisoning patients were then stratified according to clinical severity by using the poisoning severity score. The 1,951 (40.1%) patients with moderate, severe, or fatal poisoning were deemed to have been potentially eligible for antidote treatment. Of these, 855 (17.6% of the total patients) would have been eligible for treatment with antidotes that matched the poisons. Of these, 719 (14.8%) would have been eligible for emergency antidotes. However, only 92 (1.9%) actually received emergency antidotes other than activated charcoal (Fig. 1).

The numbers of adult doses of the 18 emergency antidotes that should be stocked by EDs were estimated extrapolating the data from the multicenter epidemiological study to the NEDIS data (Table 5). The NEDIS is the largest and representative data system about the patients who admitted at Korean EDs. The injury surveillance report, which is generated by the Korea Centers for Disease Control and Prevention, is a comprehensive nationwide epidemiological surveillance analysis including poisoning patients. The NEDIS data showed that every year between 2,400,000 and 2,900,000 patients visit about the 124 EDs throughout Korea that participate in NEDIS. The overall frequency of poisoning cases exceeds 1.7% and more than 700,000 patients presented with acute poisoning each year. It was assumed that the cohort of poisoning patients in the epidemiological study was a good sample of the poisoning cases recorded in NEDIS. Extrapolation of these two frequencies to the 700,000 poisoning patients recorded in NEDIS data.

The ideal and minimum quantities of the 18 emergency antidotes that should be stocked by the 124 NEDIS-participating EDs to treat acute intoxication patients were calculated on the basis of the epidemiological data. It was estimated that 103,348 (14.8%) adult doses of the 18 emergency antidotes should be available for possible emergency treatment. Of these, 13,224 (1.9%) adult doses (575 of atropine, 144 of calcium gluconate or other calcium salts, 2,587 of flumazenil, 3,450 of N-acetylcysteine, 5,893 of pralidoxime, 287 of hydroxocobalamin, 144 of sodium nitrite, and 144 of sodium thiosulfate) would be needed and should be stocked at EDs in Korea for maintaining the present level of initial treatment with emergency antidotes.