According to a recent definition, adverse drug reactions are harmful and unintended responses to medicines.1,2 The main negative consequences of adverse drug reactions are increased morbidity and mortality, prolonged hospital stay and increased healthcare costs.3,4 On the other hand, adverse drug reactions are an important cause of medicine-related emergency department visits and many of these are preventable.5,6 A fundamental step in the prevention of harm caused by adverse drug reactions is the continuous investigation and characterization of these reactions, in order to be able to implement appropriate intervention measures.7 One model of action focused on minimizing adverse drug reactions is the use of the Beers criteria to reduce potentially inappropriate prescriptions in elderly patients.8

The emergency department constitute a communication system between the hospital environment and the community, offering a valuable source of information for the study of adverse drug reactions.9 Numerous studies have been published with the aim of describing adverse drug reactions leading to emergency department visits, although data concerning risk factors contributing to hospitalization are less frequent.10–12

The primary objective of our study was to determine the prevalence of emergency department visits for adverse drug reactions and to describe their characteristics. The secondary objective was to determine the predictor variables of hospitalization associated with emergency department visits.

During the study period, 10,799 patients visited the emergency department. Initially, 243 patients with suspected adverse drug reactions were recorded by clinicians. Finally, after imputability analysis, 216 (2%) patients were included. The mean age was 70 ± 17.5 (18-98) years and 47.7% of the patients were male. Of the patients with adverse drug reactions, 54.6% (n = 118) required hospital admission. The demographic and clinical characteristics of the patients were described in Table 1.

The total number of drugs involved in the adverse drug reactions was 315, with 149 different drugs and a mean number of drugs involved per patient of 1.5 ± 0.75 (range=1-5). The pharmacological group corresponding to the nervous system (group N) constituted the most representative group (n = 81) (see Table 2). On the other hand, antithrombotic agents (n = 53) were the most common subgroup (23 vitamin K antagonists, 14 direct acting anticoagulants, 12 antiplatelet agents, and 4 low molecular weight heparins) followed by antineoplastics (n = 48) (11 monoclonal antibodies, 9 platinums, 8 taxanes, 7 antimetabolites, 5 alkylating agents, 3 vinca alkaloids and others). Vaccines accounted only for 13 adverse drug reactions; 7 directed against COVID 19, of which 2 required hospital admission. The drug most frequently associated with adverse drug reactions was acenocoumarol (n = 20), followed by edoxaban (n = 7), dexamethasone (n = 7), furosemide (n = 7), quetiapine (n = 6), paclitaxel (n = 6) and fentanyl (n = 6). Regarding hospital admission, the groups of drugs with the highest participation in order were: antithrombotic agents (n = 36), antineoplastics (n = 34), systemic corticosteroids (n = 13) and immunosuppressants (n = 11); this last with an admission rate of 100%. Acenocoumarol (n = 12), followed by dexamethasone (n = 6), edoxaban (n = 5), fentanyl (n = 5), pembrolizumab (n = 4), and tacrolimus (n = 4) constituted the drugs most frequently associated with hospitalizations.

The adverse drug reactions collected and catalogued according to the SOC are shown in Table 3. Gastrointestinal (n = 62) and neurological (n = 40) disorders were the major groups of adverse reactions. Diarrhea (n = 16) represented the most frequent, while gastrointestinal haemorrhage (n = 13) caused the highest number of hospitalizations.

Three patients had a fatal outcome directly attributed to adverse drug reactions. One case because of severe neutropenic enterocolitis related to the administration of cisplatin, etoposide and bleomycin; a severe anemia secondary to gastrointestinal bleeding related to edoxaban in a frail patient; and finally a subdural haemorrhage due to acenocoumarol with an INR of 12.7, in a 79-year-old patient.

Charlson comorbidity index behaved as a risk factor related to hospitalization for adverse drug reactions according to adjusted logistic regression analysis (aOR 3.24 95% CI:1.47-7.13, in Charlson comorbidity index 4-6; and aOR 20.07; 95% CI:6.87-58.64 6,87-58,64 in Charlson comorbidity index ≥ 10) (see Table 4).

Our work provides valuable information on the prevalence of emergency department visits due to adverse drug reactions, the profile of patients at greatest risk of hospitalization and the main therapeutic groups involved. The continued development of pharmacovigilance studies is essential to understand the health burden of adverse drug reactions and to design interventional measures to prevent them.17 The epidemiology of adverse drug reactions visits previously published in the literature varies from 1.3% to 41.3%, in the general population.18 The prevalence shown in our study is 2%, lower than other studies performed in our country.19–20 This variability can be partly explained by the different inclusion criteria for adverse drug reactions in each study and the differences between prospective and retrospective methodology. Prospective studies involve a more active and interventional search for adverse drug reactions by a team of investigators, whereas retrospective studies are limited to a registry usually conducted by a clinician, underestimating adverse reactions.21–22 Therefore, a significant proportion of emergency department visits related to drugs are not considered by emergency physicians and may go unnoticed,23 especially when the drug is implicated in a multifactorial pathological process.24

Our study population was characterised by advanced age, with 70% of patients over 65 years of age and marked dependency (a quarter of patients had some degree of functional dependency), both factors related to adverse drug reactions.25 In addition, polymedication and comorbidity played an important role, with a patient profile very similar to the national multicentre registry of adverse drug events in the emergency department, recently conducted by the Working Group on Pharmaceutical Care in the Emergency Department of the Spanish Society of Hospital Pharmacy (REDFASTER).26 These circumstances may partly justify a slightly higher rate of hospitalization for adverse drug reactions compared to other previously published registries, represented by a less frail and aging population.10,11

Charlson comorbidity index was the only risk factor for hospitalization that was statistically significant in the multivariate analysis, confirming that comorbidity may be associated with the presence of severe adverse drug reactions.27 On the other side, age and comorbidity are variables that go hand in hand, with polypharmacy being a consequence of both.28,29 In our study, polypharmacy and age did not behave as independent variables of hospitalization, possibly due to the smaller sample size compared to previous studies.8,9 Mortality associated with adverse drug reactions obtained from pharmacovigilance registries is around 1%, and antineoplastics played an important role.30

In addition, drugs classified as high-risk drugs according to ISMP-Spain,16 those most likely to cause serious harm when used incorrectly, were involved in almost 60% and more than 40% of patients, respectively. Finally, one out of every 3 patients included in the study was readmitted to the emergency department 30 days after discharge. These circumstances highlight the importance of implementing action measures in all the processes of the drug utilization chain, from prescription to dispensing, administration, treatment follow-up and education for both patients and caregivers.

Coinciding with the evidence of studies carried out in our country, the pharmacological group belonging to the nervous system (group N according to the ATC classification) was the group most frequently associated with adverse drug reactions.19,20 On the other hand, according to the subgroup category, antithrombotic agents, including the new oral anticoagulants, caused the majority of visits and hospitalizations, with acenocoumarol being the main drug involved. Very similar data were shown by the national multicentre registry of Ruiz-Ramos et al, previously mentioned,26 in addition to international studies.5,11,27 It is worth noting the high presence of adverse reactions caused by cytostatics, a consequence of a population marked by a notable existence of neoplasms. This fact may justify that almost half of the drugs involved were considered as high-risk drugs, in addition to a notable presence of the parenteral route.

With respect to the SOC classification of adverse drug reactions, gastrointestinal and nervous system disorders were among the main adverse reaction observed, while gastrointestinal bleeding caused by antithrombotic agents were the adverse reactions with the highest number of hospitalizations in our study. These results were very similar to those shown by 2 multicentre pharmacovigilance registries carried out in Italy.10,11

Our study has some limitations. Firstly, the study was conducted in a single centre and during winter, with consequent seasonality biases. Furthermore, our work is retrospective in nature and could be influenced by selection bias, where the recording of adverse drug reactions may be underestimated, due to the difficulty for clinicians to recognize reactions masked by multifactorial processes, where rigorous diagnostic evaluation would be necessary. In addition, this bias may explain the high hospitalization observed in our study, as the most severe adverse drug reactions are usually recorded. Finally, the principal investigators did not classify adverse drug reactions according to causal category and did not use any algorithm to determine their avoidability.

In conclusion, although the observed prevalence of emergency department visits for adverse drug reactions is low compared to other studies, it remains a non-negligible health problem. High-risk drugs such as antithrombotic agents are the main therapeutic subgroup involved. In addition, Charlson comorbidity index is an independent predictor of hospitalization for adverse drug reactions, while gastrointestinal bleeding is the adverse reaction with the highest number of hospital admissions. More pharmacovigilance studies are needed to provide healthcare professionals with knowledge about adverse drug reactions to design and implement appropriate and up-to-date intervention and prevention measures.

The institutional review board granted a waiver of informed consent due to retrospective data collection.

The authors received no funding.