2.1 EudraVigilance

ADRs are defined as noxious and unintended responses to medicinal products which can occur within or outside (e.g., off-label use) the terms of the marketing authorization [16]. Suspected cases of ADRs can be reported by healthcare professionals (HCPs) (e.g., pharmacists, physicians, nursing staff) as well as by non-healthcare professionals (non-HCPs) (e.g., consumers, lawyers) [17, 18]. Physicians and pharmacists, for example, are obliged to report suspected cases of ADRs according to their professional code of conduct [2, 18].

The ADR database EudraVigilance (EV), which is operated by the European Medicines Agency (EMA), contains ADR reports from all member states of the European Economic Area [17]. In EV drugs are coded with the EudraVigilance medicinal product dictionary and ADRs are coded in accordance with the MedDRA® (Medical Dictionary for Regulatory Activities) terminology [17, 19].

MedDRA® is the international medical terminology developed under the auspices of the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH). The MedDRA® trademark is owned by the International Federation of Pharmaceutical Manufacturers and Associations (IFPMA) on behalf of the ICH.

Our analysis refers to spontaneous reports, these are unsolicited reports spontaneously reported in everyday practice, which are not collected in any organized data collection scheme such as clinical trials. Spontaneously submitted ADR reports have to include certain information to fulfil the minimum criteria (e.g., identifiable reporter and patient, at least one suspected drug and suspected ADR) [4]. Spontaneous ADR reports may contain structured and unstructured (free-text) information [20].

The primary source qualification (in the following “reporter type”) describes if the person who reported the ADR was an HCP or a non-HCP together with a more detailed specification (e.g., physician, consumer). Theoretically, one ADR report can contain several reporter types, for example, if information to the same case was reported by a physician and a consumer [4]. In order to analyse if there are differences between different reporter types, we restricted our analysis to those reports listing only one reporter type within the detailed specification. Note that despite that restriction, one ADR report may have been reported by more than one person of the same reporter type. We assume that the reporter type “pharmacists” also covers reports from other pharmacy staff such as pharmaceutical technical assistants (PTA). Further information regarding the reporting channels has been published elsewhere [2].

The designation of reports as “serious” or “non-serious” corresponds to the legal definition of the seriousness of the ADR reports, but not to its clinical severity. Therefore, all reports, including at least one ADR that was life-threatening, conferred lasting or significant disability or incapacity, led to death, required or prolonged hospitalization, led to congenital abnormalities, respectively birth defects, or endangered the patient and required medical or surgical intervention/treatment (“other”) are considered as serious [4].

2.2 Identification of Reports from Physicians, Pharmacists or Consumers

All spontaneous ADR reports from Germany received between 1 January 2018 and 31 December 2021 were extracted (n = 548,347) (see Fig. 1) from EV. In order to exclude the influence of reports referring to vaccines, all reports listing any vaccine as suspected/interacting drug were excluded (n = 276,353). As one example, COVID19 vaccines could have been subject to an overstimulated reporting, because they were subject to intense public debates [21]. Besides that, we excluded all reports containing hyposensitising solutions as suspected/interacting drugs (n = 4579) as well as duplicates, identified via equal case report identifiers (n = 154).

Fig. 1

Flowchart. Identification of adverse drug reaction (ADR) reports exclusively reported by physicians, pharmacists or consumers as reporter type in EudraVigilance (EV)

Only reports whose reporter type was exclusively specified as “physician” or “pharmacist” or “consumer” were further investigated (n = 233,516). Within these, 69,976 reports (30.0%) referred to physicians’, 42,396 (18.2%) to pharmacists’, and 121,144 (51.9%) to consumers.

2.3 Descriptive Analysis

All reports identified for physicians, pharmacists or consumers were analysed descriptively regarding their superordinate classification of seriousness (serious yes/no), year of receipt, and distribution of sex and age of the patients. Means, together with standard deviations (± sd), and medians including their interquartile ranges (IQRs) were calculated for patients’ age and the quality of documentation (see below). For all other results, frequencies and percentages were calculated.

2.4 Quality of Documentation

An overall estimation of the quality of documentation was performed by applying a computer-based algorithm according to a published score (vigiGrade) [22]. This score, ranging from 0.07 to 1, was originally created to assess the provided information in the structured format of ADR reports in the global ADR database of the World Health Organization (WHO)—VigiBase [23]. According to the original publication, a report with a vigiScore > 0.8 is defined as “well documented”. For the assessment of the ADR reports in EV, the score was slightly adapted.

2.5 Consideration of Relevant Information

All sub-datasets were screened for the presence of certain information in the structured format of each ADR report. Please note that the quality of the information provided was not assessed.

The information provided was assigned to four categories: (1) report-related information, (2) patient-related information, (3) drug-related information, and (4) ADR-related information.

Sub-analyses of serious and non-serious ADR reports were performed.

2.6 Report-Related Information

The country of origin, report type, seriousness, and the narrative were categorized as report-related information. The country of origin (Germany), the report type (spontaneous report), and if the report was classified as serious (yes or no) was provided in each ADR report. In the narrative, the reporter has the option of supplying further information regarding the ADR in an unstructured free text format. In this case, we analysed if the respective reporter provided any further information as free text.

2.7 Patient-Related Information

Sex, age, height, weight and medical histories of the patients were categorized as patient-related information. Sex can be coded as female, male and not specified. Age, height and weight are numerical values or reported as not available. The medical history contains co-morbidities and previous conditions of the patients or is stated as not available. In this case, “not available” does not necessarily mean that the patient has no co-morbidities or previous conditions, it can also indicate that the information was not reported. In our analyses, not specified in case of sex or not available in case of age, height, weight and medical history were considered as information not provided.

2.8 Drug-Related Information

Drugs reported as suspected/interacting, action taken with the drug, applied dose of the drug, indication of the drug, reporting of suspected/interacting drugs classified as black triangle drugs and concomitant drugs were categorized as drug-related information. In accordance with the minimum criteria of ADR reports, at least one drug has to be reported as suspected/interacting. Among others, action taken with the drug can be reported as withdrawn, continued, dose reduced, or information not available. The applied dose of the drug can be supplied as a numerical value with its unit name (e.g., milligram), as the number of applied dosage forms (e.g., one dose administered) or set to “not available”. The indication describes the underlying disease or condition for which the drug was taken by the patient or is reported as “not available”. The drugs classified as concomitant were not considered to be causative for the described ADR(s) by the reporter. In this category, all drugs taken concomitantly can be reported or set to “not available”. In this case, our data did not allow distinguishing whether the patient did not receive any concomitant medication or if the medication was not mentioned in the report.

If more than one drug was reported as suspected/interacting per ADR report, information regarding action taken with the drug, applied dose, and indication of the drug can be reported for each drug. Here, information was classified as available if action taken with drug, applied dose, or indication was reported for at least one drug.

Further on, we determined the number of drugs reported as suspected/interacting and reported as concomitant per ADR report.

In order to identify reports with at least one black triangle drug reported as suspected/interacting, the “list of medicinal products under additional monitoring” was used [24]. All active ingredients with a date of inclusion in 2022 were removed. Drugs labelled with a black triangle in the package leaflet or product information leaflet are subject to additional surveillance (e.g., new active ingredients, biologicals or drugs with conditional approvals).

2.9 Adverse Drug Reaction (ADR)-Related Information

The reported ADR(s), the outcome of the ADR(s), the time to onset of the ADR(s), as well as if the ADR is known or designated as a medically important event are categorized as ADR-related information. In accordance with the minimum criteria of ADR reports, at least one ADR has to be reported. The outcome describes, among others, if the ADR is recovered/resolved, fatal or not recovered/not resolved or unknown. The time to onset describes the time between the first intake of a drug and the occurrence of the ADR. It can be calculated based on the start date of the drug reported as suspected/interacting and the start date of the ADR. If multiple ADRs per report were mentioned, an outcome and a time to onset could be reported or calculated for each ADR. Information was counted as available if at least one outcome apart from unknown was reported, respectively, one time to onset could be calculated for one ADR. If the calculated time to onset was negative, the time to onset was classified as not available (caused by, e.g., coding errors or if the suspected/interacting drug was taken after the ADR occurred).

The number of reported ADRs per ADR report was analysed.

All reports were analysed to determine if at least one ADR contained in the “designated medical event list” was reported. The list of designated medical events describes drug-related serious events defined by the EMA and is publicly available on the EMA’s website [25].

We determined whether at least one of the reported drug-ADR combinations per ADR report was already listed as known in SIDER. SIDER lists more than 140,000 known drug-ADR combinations based, for example, on information from package inserts. Further information on the content, data collection, and structure of SIDER can be found elsewhere [26, 27].

2.9.1 Logistic Regression Analyses

Logistic regression analyses were performed for the comparisons of physician versus consumer reports, physician versus pharmacist reports, pharmacist versus consumer reports, and serious versus non-serious reports of all reporter types. To this end, the reporter type, respectively, the classification serious/non-serious, was set as the outcome variable and all other variables, which were analysed in the consideration of relevant information as covariates. Country of origin, report type, and seriousness were not included as variables in the logistic regression models since all reports contained information on these three aspects. The results of the logistic regression analyses are presented as odds ratios (ORs) with 95% confidence intervals (CI) in square brackets [lower CI–upper CI].

All results of the logistic regression analyses are presented by mentioning the reference group last (e.g., in case of “physicians versus pharmacists”, pharmacists are considered as reference). Concerning the analyses of non-serious and serious reports within each reporter type, serious reports were always considered as the reference.

2.9.2 Reports per Physician, Pharmacist and Consumer

The number of reports from physicians and pharmacists was considered in relation to the number of practising physicians respectively pharmacists per year in Germany. For this purpose, the numbers of practising physicians in outpatient and hospital settings for each year were extracted from the “Ärztestatistik zum 31. Dezember 2021” provided by the Bundesärztekammer [28]. The numbers of practising pharmacists represents the number of pharmacists, pharmacists in internship (PhiP), pharmaceutical technical assistants (PTAs), pharmacy technicians and pharmacist assistants in public pharmacies, and pharmacists working in hospital pharmacies, which were extracted for each year from “Die Apotheke - Zahlen Daten Fakten 2021/2022" published by the Bundesvereinigung deutscher Apothekerverbände [29, 30]. In order to consider the number of reports from consumers in relation to the number of potentially drug-exposed inhabitants in Germany, we extracted the number of German inhabitants from statista “Bevölkerung—Einwohnerzahl von Deutschland 1990–2021” and subtracted the number of practising pharmacists and practicing physicians, as reports from these professionals were to be included in the respective categories [31]. The resulting number was multiplied with the proportion of Germans expected to take any medication according to a German survey [32].

Note that this is only an approximation since one pharmacist or physician or consumer can have submitted more than one ADR report and one ADR report can be reported by more than one pharmacist or physician.