1. Introduction
Cephalosporins are one of the most commonly used antibiotics in clinical practice [
1]. Cephalosporins are generally considered safe; but allergic reactions, renal dysfunction, hepatic dysfunction, and seizures have occurred as cephalosporin-class adverse reactions and are listed on the drug labels. These adverse drug reactions (ADRs) are responsible for morbidity and mortality and represent a significant burden to both the affected patient and to the health care system [
2,
3].
ADRs are usually detected during clinical trials. However, short testing periods and small numbers of subjects in clinical trials hinder identification of all ADRs before marketing [
4]. For this reason, investigation of post-marketing adverse events (AEs) is necessary to ensure the safety of the drug. In particular, rare (1 in 1000) or very rare (1 in 10,000) serious ADRs can only be observed after a large number of patients have been administered the drug [
5]. AE reporting systems are widely used and cost-effective methods for collecting post-marketing AEs [
5]. Spontaneous AE reports from pharmaceutical companies, national and international pharmacovigilance centers, or regulatory authorities are collected in national pharmacovigilance databases [
5]. The Korea Institute of Drug Safety & Risk Management (KIDS) developed the Korea Adverse Events Reporting System (KAERS) database (KIDS-KD) to collect and manage AE reports [
6]. Appropriate statistical analysis of AE reports that were collected in the database enables detection of signals of unknown ADRs [
7]. A signal is defined as the information implying a known or unknown association among drugs and AEs that warrants further investigation [
8].
Large-scale studies have investigated AEs of cephalosporins from hospital data or AE reporting systems [
2,
3,
9,
10]. However, these studies only descriptively analyzed the frequency of AEs and did not detect signals using statistical analysis. Some studies analyzed the cephalosporin-class adverse reactions (anaphylaxis, renal dysfunction, hepatic dysfunction, and Clostridium difficile infection) that have already been well-documented [
11,
12,
13,
14,
15].
The aim of this study was to detect signals that are not currently listed on drug labels of cephalosporins through analyzing AE reports in the KIDS-KD.
3. Discussion
In this study, AE reports concerning cephalosporins were investigated and signals were detected through disproportionality analysis. As a result, 114 signals and six serious ADRs that were not included on the drug labels were found. To our knowledge, this is the first study that detected new signals through disproportionality analysis of a national pharmacovigilance database.
Allergic reactions were previously the most commonly reported cephalosporin-related AE with a frequency of approximately 1 to 5% [
16,
17]. These reactions are mainly manifested as skin symptoms such as rash, pruritus, urticaria, and angioedema [
16,
17,
18,
19]. In addition, severe cutaneous adverse reactions (SCARs), such as acute generalized exanthematous pustulosis (AGEP), Stevens-Johnson syndrome (SJS)/toxic epidermal necrolysis (TEN), and drug rash with eosinophilia and systemic symptoms (DRESS) can also be caused by cephalosporin [
20,
21,
22]. Skin symptoms accounted for approximately one-third of the cephalosporin-induced AEs in the present study. The results that the organs most commonly affected were the skin and appendages were consistent with those of other studies [
3,
9,
23].
A national pharmacovigilance database enables detection of unknown ADRs in addition to known ADRs. The KIDS-KD has been successfully used to detect signals of several drugs including imipenem and quinolone antibiotics [
24,
25]. In this study, we were able to detect signals of cephalosporins. For example, cefatrizine, a first-generation cephalosporin used in East and Southeast Asia, was found to be associated with corneal ADRs. Cefatrizine-induced corneal ADRs have never been reported in the literature. This confirms the usefulness of a national pharmacovigilance database for detecting unidentified ADRs.
Most ADRs of cephalosporins are mild, self-limiting, and resolve with discontinuation of the drug [
16]. However, serious ADRs can affect morbidity and mortality. Thus, this study focused on serious AEs. Some of the reported serious AEs such as anaphylactic shock are well-documented ADRs of cephalosporins. The present study identified six serious AEs that were not listed on the drug labels. Among the AEs, cefaclor-induced stupor is the most commonly reported. Among cephalosporins, stupor is listed only on the drug label of cefepime. Cefepime, a fourth-generation cephalosporin, can access the cerebral spinal fluid [
26]. Cefepime has been associated with the induction of seizures and encephalopathy. The mechanism proposed for this induction is the competitive inhibition of γ-aminobutyric acid (GABA) A receptors [
27]. Other cephalosporin antibiotics can also cause serious central nervous system toxicities [
10]. The risk is particularly high if the drug accumulates in the body due to renal failure [
10,
28]. Although stupor is not listed as an ADR of cefaclor, caution is needed.
This study has some limitations similar to those of other studies using a national pharmacovigilance database. First, since not all AEs are reported, the occurrence of AEs may be underestimated due to underreporting. Second, unlike information-rich hospital data, data from the KIDS-KD omitted information about the clinical status of patients, for example kidney and liver function. This limits the adjustment to the risk factors of ADRs. Third, some information about concomitant drugs that can affect specific AEs was also missing. The possibility of under-reporting of concomitant drugs is inevitable in studies with KIDS-KD. Fourth, the preferred term (PT) code may not be stated explicitly on the drug label. The difference in ADR terms was a barrier in evaluating whether a signal is recorded on the drug label. Moreover, an unlisted signal can be a symptom of a listed ADR. Last, some signals such as bronchitis and pharyngitis may be incorrectly detected. These cases may be due to deterioration caused by underlying diseases due to drug ineffectiveness. Therefore, the results of this study should be carefully interpreted.
4. Materials and Methods
4.1. Data Collection and Processing
This study is a case-non-case study using data from a national pharmacovigilance database. From January 2009 to December 2018, AE reports concerning antibacterial drugs for systemic use (anatomical therapeutic chemical (ATC) code J01) in the KIDS-KD were examined. The study protocol was exempted from review by the institutional review board of Ehwa Womans University (ewha-202004-0034-01).
The data consist of ‘patient information’, ‘drug information’, ‘AE information’, ‘seriousness of AE’, ‘reporter information’, ‘causality assessment’, and ‘past medical history’. All drugs were coded using the ATC code. AE information was coded using the PTs of the World Health Organization-Adverse Reactions Terminology (WHO-ART) version 092 [
29]. Serious AEs were defined as a fatal or life-threatening AE; an AE resulting in persistent or significant disability, congenital abnormality or birth defect; an AE requiring hospitalization or prolongation of ongoing hospitalization; or other medically important conditions [
30].
The data manager of KIDS extracted all reports, including ATC code requested by the researcher, and provided them in a form that can be downloaded by the researcher. First, the completeness of the data was reviewed through detection of input or logical errors. In order to eliminate duplicate cases, only the initial report was included in the study; follow-up reports were excluded from the analysis. Reports recorded as ’unlikely’ in the causal assessment item were excluded. Reports on post-operative pain, drug prescribing error, and medical device complications were also excluded. Some reports contained more than one AE or more than one drug. For analysis, one-to-one combinations of drug-AEs were made.
4.2. Statistical Analysis
Characteristics of AE reports about cephalosporins and all other systemic antibiotics (all other J01) were analyzed descriptively. These characteristics include patient demographics (sex and age), type of report, reporter, and seriousness of AE. The type of AE was analyzed at both PTs levels and SOC levels.
Disproportionality analysis, which compares the observed count for a study drug–AE combination with an expected count, is a primary tool to detect signals [
31]. There were three main indices, PRR, ROR, and IC, used in this study. A two-by-two contingency table was used to calculate these indices (
Table 5) [
32,
33]. The subtle differences in chemical structure and pharmacokinetics between cephalosporin antibiotics can affect AEs [
16]. Therefore, disproportionality analysis was conducted on individual cephalosporin antibiotics. Comparator was defined as all other J01. In this study, a drug–AE combination that met the criteria of all three indices was defined as a signal (
Table 6). Among the signals, we focused on serious adverse events that were clinically meaningful.
In addition, ROR was calculated for major SOCs to evaluate differences in AEs by cephalosporin generations: body as a whole, general disorders (WHO-ART: 1810); cardiovascular disorders, general (WHO-ART: 1010); central and peripheral nervous system disorders (WHO-ART: 0410); and respiratory system disorders (WHO-ART: 1100).
The signals were compared with ADRs listed on the drug labels from the United States (USA), United Kingdom (U.K.), Japan, and South Korea. A signal that is not currently included on the drug labels was defined as a new signal. The label for each cephalosporin antibiotic was found through searching websites that are operated by drug regulatory authorities: ‘
https://www.accessdata.fda.gov/scripts/cder/daf/’ (accessed on 10 May 2020) for the USA Food and Drug Administration (FDA), ‘
https://products.mhra.gov.uk/’ (accessed on 10 May 2020) for U.K. Medicines and Healthcare Products Regulatory Agency (MHRA), ‘
https://www.pmda.go.jp/index.html’ (accessed on 10 May 2020) for Japan Pharmaceuticals and Medical Devices Agency (PMDA), and ‘
https://nedrug.mfds.go.kr/searchDrug’ (accessed on 10 May 2020) for Korea Ministry of Food and Drug Safety (MFDS). Drug databases such as DailyMed (
https://dailymed.nlm.nih.gov/dailymed/; accessed on 10 May 2020) and Drugs.com (
https://www.drugs.com/; accessed on 10 May 2020) were also used.
All statistical analyses were performed using SAS® University Edition (SAS Institute, Cary, NC, USA) and Microsoft Excel 2019 (Microsoft, Redmond, WA, USA). A p-value less than 0.05 was considered statistically significant.