Descriptive analysis on disproportionate medication errors and associated patient characteristics in the Food and Drug Administration's Adverse Event Reporting System

Medication errors (MEs) are a major public health concern which can cause harm and financial burden within the healthcare system. Characterizing MEs is crucial to develop strategies to mitigate MEs in the future.


| INTRODUCTION
The National Coordinating Council for Medication Error Reporting and Prevention (NCCMERP) defines a medication error (ME) as "any preventable event that may cause or lead to inappropriate medication use or patient harm while the medication is in the control of the healthcare professional, patient, or consumer." 1 This definition is also adopted by the Food and Drug Administration (FDA). 2 In addition, this definition excludes "the deliberate or intentional use (e.g., abuse, misuse, off label use) of a drug product in a manner that is inconsistent with FDA-required labeling." 3The World Health Organization initiated the third Global Patient Safety Challenge in 2017 titled "Medication Without Harm" for the purpose of promoting solutions to address problems concerning safety of medication practices, which includes dealing with MEs. 4 In 2012 it has been estimated that the worldwide annual costs of avoidable MEs were 42 billion dollars. 5A systematic review from 2018 showed that the prevalence of MEs ranged from 2% to 94% among adults in various study populations around the world. 6Reported prevalence numbers of MEs vary greatly due to heterogeneity in studied populations and variations in ME definitions. 6,7A recent 2020 study estimated that annually 237 million MEs occurred in England alone. 8Of these MEs, the researchers estimated that 28% were potentially clinically significant.Furthermore, the same study estimated that the adverse drug events (ADEs) stemming from MEs have led to an additional annual consumption of 181 626 hospital-bed-days and contributed up to 1708 annual deaths.
Another research estimated that annually 7150 ME-related deaths are preventable for inpatients in the United States of America (USA) with a life-expectancy greater than 3 months. 9Besides the economic burden and the healthcare impact, MEs also contribute to a lack of trust in the healthcare system and to patient dissatisfaction. 10Hence, it is of high importance to gain more insights into reported MEs in order to prevent MEs in the future.
However, this study provided limited information on patient characteristics and clinical outcomes of MEs.Furthermore, those studies did not touch upon the topic of data quality or completeness of the Individual Case Safety Reports (ICSRs), while research on the data quality within FAERS appears scarce. 13e objective of this paper was to provide an overview of ME reports from FAERS with emphasis on ME characteristics, associated patient characteristics, clinical outcomes, and SDRs of DMEs.

| Data source
FAERS is a database that contains spontaneous reports related to medication or medical devices, submitted to the FDA on adverse events (AEs), product quality complaints resulting in AEs, and MEs. 14yone can freely send in a report through FDAs MedWatch Online Voluntary Reporting Form. 15The Medical Dictionary for Regulatory Activities ® (MedDRA ® ) is used to code AEs and MEs.This study uses MedDRA ® version 23.1.The Legacy Adverse Event Reporting System (LAERS) is the predecessor of FAERS and it embodies earlier data from January 2004 up to September 2012. 16,17The database files from the newer FAERS database are published since October 2012.
The publicly available database files from LAERS and FAERS were the primary data source for this study.In the remainder of the manuscript, we refer to the combined LAERS and FAERS database as "the FAERS database."

| Data extraction and processing
The Adverse Event Open Learning through Universal Standardization (AEOLUS) system developed by Banda et al. was the cornerstone for extraction and processing of FAERS data. 18The main functions of AEOLUS include standardization of FAERS data, deduplication of case reports, and generating SDRs.AEOLUS is available as a GitHub repository. 19For this research, a forked version was used. 20The original structured query language (SQL) scripts from AEOLUS were adjusted in such a way that all data files up to the end of 2020 could be downloaded.AIOLI, a drug mapping tool developed within the Medical Informatics department, 21 allowed conversion of drug names to Anatomical Therapeutic Chemical (ATC) classification.Various age units encountered in the data were converted to years.Ages between 0 and 120 years were considered valid.The age variable was considered unknown in case of missing age, missing time format, or negative age.

| ME case identification and characterization
ICSRs of MEs within FAERS were identified by using the preferred terms (PTs) from the narrow standardised MedDRA ® Query (SMQ) for the key topic on MEs. 22,23(See Table S1 for a complete list of PTs.) Characterization was performed on variables such as age, sex, drug use, indications and procedures, reporter type, country of origin, and clinical health outcomes.The clinical health outcomes were reported as "death," "life-threatening," "hospitalization (initial or prolonged)," "disability," "congenital anomaly," "required intervention to prevent permanent impairment or damage," and "other serious (important medical events)."The health outcome "other serious (important medical events)" is suggested by the FDA to be used in case that "the event [medication error] does not fit the other outcomes [i.e., death, life-threatening, etc.], but the event may jeopardize the patient and may require medical or surgical intervention (treatment) to prevent one of the other outcomes [i.e., death, lifethreatening, etc.]." 24

| Disproportionality analysis
To assess the association between reported drug use and MEs, a disproportionality analysis was performed.DME-counts were performed by AEOLUS and shown in Table 1. 18e reporting odds ratio (ROR) was calculated with the equation: The respective 95% confidence intervals (CIs) were calculated with the equation: Only drugs that were marked as primary suspects in the ICSRs were considered for analysis.Thresholds for displaying relevant SDRs were applied according to the EMA: a lower bound of the 95% CI >1 and number of individual cases ≥3. 25 Stratified disproportionality analyses were also performed based on sex and age.

| RESULTS
The selection and the processing of ICSRs is displayed in Figure 1.A total of 14 919 552 reports were retrieved of which 10 651 341 (71%) were from the latest FAERS system.The AEOLUS system found   showed some overlap, but overall there was a wide variety regarding the top 10 strongest SDRs.The stratified results are available in Tables S2-S9.

| DISCUSSION
In this study, we investigated the frequency of MEs, it's characteristics, and signals of disproportionality.We observed an increase in reporting of MEs over time, almost half of the MEs were associated with serious health outcomes, and MEs among antineoplastic and immunomodulating agents were reported most frequently.
The observed increased reporting rate in FAERS can be related to increased healthcare access, utilization due to the Affordable Care Act enacted in 2010, [26][27][28] and the introduction of the reporting form FDA3500B, released in 2013, which aided consumers in voluntary reporting in addition to online reporting via MedWatch. 29Also, public health campaigns on medication safety might have contributed to awareness on the necessity of reporting and therefore increased reporting. 30,31The fact that FAERS ICSRs were mostly submitted by consumers rather than HCPs resonates with literature emphasizing many limiting factors for HCPs, such as lack of time, difficulties determining the cause of ADE, poor reporting systems and uncertainty about the reporting procedures. 32Furthermore, other research reported that consumers were motivated to self-report if they were under the impression that HCPs did not (have the time to) report an ADR, did not take their ADR seriously, or if HCPs asked consumers to report themselves. 33Also, we assume that MEs might not be The drug count forthcoming from a total count of 512 933 drug-medication error combinations found in the 488 470 medication error-related individual case safety reports represented on level 1 of the Anatomical Therapeutic Chemical classification.recognized or considered as AEs by the HCP, because the patient never mentions them or does not seek HCP guidance.Interestingly, the proportion and the amount of ME-related ICSRs also increased over time.[39] Most described demographic attributes correspond with what is observed among the general population of health consumers.
Most subjects in ME ICSRs were women, who receive more drug prescriptions and have a higher healthcare consumption than men. 40,41Ageing coincides with an increased risk of disease development and thus increased drug use. 39Despite lower amount of reports among the pediatric population in this study, the risk of MEs in children should not be underestimated; a systematic review reported that MEs occurred in up to 27% of treated children. 42e observed ME health outcomes from this study as described in Table 2 show similarities with especially the moderate and severe health outcomes of MEs from a systematic review focusing on MEs among nursing home residents, mostly in the USA. 43It seems that polypharmacy individuals were underrepresented in our study as less than one-third of the ICSRs reported individuals using 2 or more unique drugs at the time of the ME, while literature suggests that half of the medication users in the USA use 2 or more drugs concomitantly. 37terature on MEs related to drug use is limited.5][46] The overall proportions of drugs on ATC level 1 related to MEs from the narrow SMQ in our study were largely similar to findings from Carnovale et al. 12 Findings by Newbould et al. based on EudraVigilance data showed that MEs were mainly reported for drugs belonging to the "nervous system" drug group, similar to our findings in FAERS. 11Vaccines (ATC: J07) were most commonly reported in EudraVigilance, but rarely in FAERS data, as vaccine-related reports are primarily submitted to the Vaccine Adverse Event Reporting System (VAERS) in the USA.Findings from spontaneous reporting systems do not necessarily reflect findings from research focused on active identification of MEs, 47,48  correspond with our findings, although the exact frequencies or proportions were not provided in the paper of Carnovale et al. 11,12 Not all of the most frequent and strongest SDRs observed in this study could be confirmed by literature.For instance, there is no focused literature on the association between adapalene use and the risk or causes of MEs.While there are studies published on adapalene adherence, there is no information regarding MEs. 49,50Narratives of received ICSRs, provided by the FDA, show how patients have overdosed on over-the-counter (OTC) adapalene, without providing further background information on what was the root-cause of overdosing.
For the drug pegfilgrastim, MEs were mostly related to issues with (handling of) the device.2][53][54][55] However, it is unclear whether this represents MEs or product quality issues.Received narratives from the FDA show a variety of circumstances under which pegfilgrastim-related MEs occurred, emphasizing the necessary caution when dealing with such drugs, especially in complex healthcare settings.
For triprolidine, which had the strongest SDR (but with low number of reports), there was no literature confirming this association.Limited literature was available on the second strongest SDR forthcoming from sodium hypochlorite in relation to MEs. [56][57][58][59] Albiglutide was overall found to be a relatively frequent and strong signal in terms of narrow PTs within FAERS, however, scientific literature on albiglutide was limited, 60 and the drug has been discontinued for commercial grounds since 2018. 61The inhalation medication salbutamol produced a strong SDR, backed with a high frequency of ME occurrence.It is a commonly used drug among patients with a respiratory disease and it is known that MEs occur frequently among patients using inhaler devices, while sometimes HCPs can be also at fault for inhaler related MEs. [62][63][64][65]

| Strengths and limitations
As for all observational studies, our study has strengths and limitations.Despite the increased reporting of MEs over time, findings in literature suggest that barriers remain in healthcare practices for reporting MEs which lead to (i) underreporting as well as selective reporting. 66,67It is already known that FAERS in general has an underreporting problem similar to all other spontaneous reporting systems. 68(ii) Vaccine-related-ADRs are not included in the database (as these are not part of FAERS) and unmapped RxNorm names could not be included in the disproportionality analysis, which might bias the results and also make the results less comparable with findings from other spontaneous reporting systems. 11,12(iii) MedDRA ® PTs are by itself debatable in the sense that some describe similar concepts or seem generally too vague to interpret.On top of that, Med-DRA ® PTs might not have been properly chosen for ICSRs which indeed has already been described in practice. 69(iv) Despite all efforts of the code-developers for the deduplication of the FAERS data, there might be some duplicate cases present in the data.(v) Missing data in FAERS limits the interpretability of some results and might prohibit stratified analyses.(vi) Finally, the disproportionality analysis only emphasizes the disproportionality of a given DME rather than true causality.Further causality analysis is necessary for an in-depth interpretation of a SDR.The strengths of our study are that we provided details on ICSRs for MEs in terms of patient characteristics, health outcomes, and reporter details.Also, this study provides more information on unreported data within the ME population which is important to consider when conducting subsequent research on MEs.
Finally, this study contributes to the body of knowledge on reported MEs, providing insights on which MEs were commonly reported and what future focus areas should be for prevention of MEs.

| CONCLUSION
This study offers a first of its kind characterization of MEs as reported to FAERS.MEs were increasingly being reported, mostly by consumers.Most reported MEs were associated with adults and females.
About 1 out of 3 of the reported MEs resulted in a serious health outcome.A wide variety of DMEs were found, some not or rarely reported in literature, most often involving parenteral drugs.This FAERS data provides insights on ME prevention and offers possibilities for additional in-depth analyses.
data processing.A total of 11 096 667 (89%) ICSRs had a valid ATC code for the primary suspect drugs.Using the narrow SMQ for MEs, 488 470 (4%) ICSRs were identified to be related to MEs.A total of 1659 distinct DMEs could be classified under the narrow ME SMQ.About 25% of the DMEs met the criteria for disproportionality to be considered a SDR.Figure2shows how the number of spontaneous reports received by FAERS increased over the years from 204 538 in 2004 up to 1 588 017 in 2020, with a slight decrease in 2018.The proportion of ME reports followed a trend in which the proportion of MEs declined from 3.9% in 2004 to 1.7% in 2013, but increased up to 7.5% in 2020.
Newbould et al. studied MEs within (DMEs), and no results on associated clinical event outcomes or on signals of disproportionate reporting (SDRs).Carnovale et al. found 468 677 case reports on MEs within the FDA's Adverse Event Reporting System (FAERS) between 2004 and 2017, showing more detailed results on DMEs in multiple age strata with emphasis on SDRs.
Contingency table representing the necessary counts needed to perform a disproportionality analysis.

Table 2
Patient, reporter and data characteristics of medication error related individual case safety reports.Age characteristics of individuals from the medication error related individual case safety reports.
provides an overview of the report, reporter, and database characteristics of the 488 470 ICSRs related to MEs.About 419 931 (86%) of the cases originated from the latest FAERS database.In 269 032 (55%) cases, the sex of the user of the primary drug suspect was registered as female, 162 445 (33%) were male, and 56 993 (12%) were not reported.The median reported age was 57 years with an interquartile range of 37-70 years as presented in Table 3.More detailed age distributions of both the ME population and the entire FAERS population, additionally stratified by sex, are provided in Figures S1-S6.Consumers accounted for 286 469 (59%) of submitted ICSRs, healthcare professionals (HCPs) for 117 788 (24%), lawyers for 2021 (<1%), and for 36 769 (8%) of ICSRs the reporter type was unknown.Most of the reports, being 387 656 (79%), originated from the USA and 29 640 (6%) were from unknown origin.The health outcome for most MEs was not reported, accounting for 338 744 (63%) reports.The most commonly reported health outcome was hospitalization in 65 369 (12%) reports.Health F I G U R E 2 Proportion of medication error reports compared to the total number of reports received in FAERS between 2004 and 2020.FAERS, Food and Drug Administration's Adverse Event Reporting System.F I G U R E 1 Flowchart for the selection and analysis of individual case safety reports (ICSRs).ATC, Anatomical Therapeutic Chemical classification; DME, Drug-medication error combination; FAERS, Food and Drug Administration's Adverse Event Reporting System; LAERS, Legacy Adverse Event Reporting System; SDR, Signal of Disproportionate Reporting.outcomesdeath,life-threatening,disability,congenitalanomaly,and"requiredintervention to prevent permanent impairment or damage" were each reported in less than 5% of the cases.The majority of the ICSRs (337 640 reports; 69%) reported one drug per ICSR.From the 488 470 ICSRs on MEs, there were 302 325 (62%) ICSRs which listed an indication for the primary drug suspect.The most frequent system organ classes (SOCs) corresponding with the indications of the primary drug suspect were "Musculoskeletal and connective tissue disorders" (46 129 reports; 15%), "Metabolism and nutrition disorders" (35 151 reports; 12%), "Nervous system disorders" (25 042 reports; 8%), "Skin and subcutaneous tissue disorders" (24 689 reports; 8%), and "Surgical and medical procedures" (23 062 reports; 8%), accounting for 51% of all SOCs.More details are provided in FigureS7.A total of 512 933 DMEs were found among the 488 470 MErelated ICSRs as seen in Figure3.About 70% of all DMEs were associated with "Antineoplastic and Immunomodulating agents" (129 588 DMEs; 25%), drugs from the "Nervous system" (90 090 DMEs; 18%), drugs involved in the "Alimentary tract and metabolism" (76 351 DMEs; 15%), and drugs for the "Respiratory system" (61 941 DMEs; 12%).The most frequently reported PTs were "Incorrect dose administered" (90 903; 9%), "Wrong technique in product usage process" the route of administration for most drugs displayed in Table 5 was parenteral.Disproportionality results based on highest ROR are displayed in Table 6.Triprolidine (ATC: R01AX03) showed the strongest signal with a ROR of 48.08 (95% CI: 10.76-214.82),but the number of reports was three.Stratified results based on sex and age groups T A B L E 2 The 10 most frequent drug-medication error combination resulting in a signal of disproportionate reporting.The 10 strongest signals of disproportionate reporting related to medication errors.
which emphasizes the underreporting problem and thus the lack of a complete overview of MEs within a database with spontaneous reports on ADRs.The most frequently reported MEs by Carnovale et al. and Newbould et al.