Portal de Periódicos da CAPES

Safety of medicines and vaccines – building next generation capability

2021; Elsevier BV; Volume: 42; Issue: 12 Linguagem: Inglês

10.1016/j.tips.2021.09.007

1873-3735

Andrew Bate, Jens‐Ulrich Stegmann,

Pharmaceutical studies and practices

Pharmacovigilance relies on a variety of data sources, which continue to diversify in terms of their nature and extent of usage in the field.Improvement of safety data should focus not only on volume, but also on quality and availability/accessibility.Scientific and technological advances such as automation and machine learning are not fully integrated and exploited in pharmacovigilance.Patient engagement has increased considerably, along with expectations for transparency on the way safety data are collected and used, but can be further improved.Phenotypic profiling following advances in pharmacogenomics and related fields can enhance safety data analysis and progress us towards more personalised safety profiles. The systematic safety surveillance of real-world use of medicinal products and related activities (pharmacovigilance) started in earnest as a scientific field only in the 1960s. While developments have occurred over the past 50 years, adding to its complexity and sophistication, the extent to which some of these advances have positively impacted the capability for ensuring patient safety is questionable. We review how the conduct of safety surveillance has changed, highlight recent scientific advances, and argue how they need to be harnessed to enhance pharmacovigilance in the future. Specifically, we describe five changes that we believe should and will need to happen globally in the coming years: (i) better, more diverse data used for safety; (ii) the switch from manual activities to automation; (iii) removal of limited value, extraneous transactional activities and replacement with sharpened focus on scientific efforts to improve patient safety; (iv) patient-involved and focussed safety; and (v) personalised safety. The systematic safety surveillance of real-world use of medicinal products and related activities (pharmacovigilance) started in earnest as a scientific field only in the 1960s. While developments have occurred over the past 50 years, adding to its complexity and sophistication, the extent to which some of these advances have positively impacted the capability for ensuring patient safety is questionable. We review how the conduct of safety surveillance has changed, highlight recent scientific advances, and argue how they need to be harnessed to enhance pharmacovigilance in the future. Specifically, we describe five changes that we believe should and will need to happen globally in the coming years: (i) better, more diverse data used for safety; (ii) the switch from manual activities to automation; (iii) removal of limited value, extraneous transactional activities and replacement with sharpened focus on scientific efforts to improve patient safety; (iv) patient-involved and focussed safety; and (v) personalised safety. The potential for medicines/vaccines to have adverse effects has been known as long as they have been used [1.Jones J.K. Kingery E. History of Pharmacovigilance.3rd edn. John Wiley & Sons, 2014Crossref Scopus (6) Google Scholar,2.Stephens M.D.B. The Dawn of Drug Safety. George Mann, 2010Google Scholar]. However it was not until the 1960s, when the association between thalidomide and congenital malformations was detected, that systematic national and international safety surveillance systems were introduced [3.Van Grootheest K. The dawn of pharmacovigilance.Int. J. Pharm. Med. 2003; 17: 195-200Crossref Scopus (17) Google Scholar]. In the aftermath of this tragedy, there was an acute need for a better system that would enable effective, rapid, and widespread sharing of information on potential adverse events (AEs; see Glossary), and that could lead to timely identification of safety issues and trigger any necessary changes in immunisation or prescribing practices. Pharmacovigilance as a scientific discipline was therefore born. The identification of safety signals is core to pharmacovigilance. Safety signals are suspected adverse drug reactions (ADRs) or adverse events following immunisation (AEFIs) after administration of an approved product, which were not/could not be identified during the clinical development phase and/or which were not predicted from the drug's mechanism of action. Some of the many examples of ADRs/AEFIs identified over the years are heart rate and rhythm disorders for nonsedating antihistamine terfenadine, heart valve disorders for fenfluramine–phentermine, intussusception for rotavirus vaccine, rhabdomyolysis for cerivastatin, and cough for angiotensin-converting enzyme inhibitors [4.Andrews E.B. Moore N. Mann's Pharmacovigilance.3rd edn. Wiley Blackwell, 2014Crossref Scopus (11) Google Scholar]. While many safety issues can be and are detected during the product's clinical development, premarketing assessment of safety cannot allow identification of all possible ADRs/AEFIs, which is why postmarketing surveillance is critical in monitoring the evolving of a new drug/vaccine [5.Schick A. et al.Evaluation of pre-marketing factors to predict post-marketing boxed warnings and safety withdrawals.Drug Saf. 2017; 40: 497-503Crossref PubMed Scopus (19) Google Scholar] and ensuring that a favourable benefit–risk profile is maintained (Box 1). The quality of information recorded is paramount for case assessment of spontaneous safety reports [6.Edwards I.R. et al.Quality criteria for early signals of possible adverse drug reactions.Lancet. 1990; 336: 156-158Abstract PubMed Scopus (105) Google Scholar]. A key challenge in pharmacovigilance is that while some ADRs/AEFIs are predictable and dose dependent, others are idiosyncratic. There is enormous heterogeneity in, for example, the time to onset of an AE, how abrupt or insidious its occurrence is, the rate of AE occurrence in an untreated population, or diagnostic difficulty [7.Aronson J.K. Ferner R.E. Joining the DoTS: new approach to classifying adverse drug reactions.BMJ. 2003; 327: 1222-1225Crossref PubMed Scopus (179) Google Scholar]. To enable effective analyses, efforts are needed to ensure that the data imparted are accurately collected and structured, in a timely fashion, and that relevant information is as complete as possible.Box 1Understanding human safety across the medicinal product's lifecycle both on an individual and population level•Helps inform medicinal knowledge and therefore future drug/vaccine prioritisation and development.•Supports better ability to predict safety profiles of medicines under development and to contextualise emerging data on developed products, leading to anticipation of future benefit–risk profile.•Helps addressing questions such as: what is the likelihood of an ADR occurring? Under what circumstances does likelihood or nature (e.g., severity) of an ADR vary? How does this affect the benefit risk profile of a medicine? If an ADR occurs, how should the ADR be best managed? What risk mitigation should be considered and if implemented, how will the impact be measured?•Helps understand any difference in the likelihood, clinical presentation, severity, duration, and risk factors of any adverse effect, and the options in terms of preventability (such as evolving therapeutic options), as healthcare provision and other environmental factors evolve. •Helps inform medicinal knowledge and therefore future drug/vaccine prioritisation and development.•Supports better ability to predict safety profiles of medicines under development and to contextualise emerging data on developed products, leading to anticipation of future benefit–risk profile.•Helps addressing questions such as: what is the likelihood of an ADR occurring? Under what circumstances does likelihood or nature (e.g., severity) of an ADR vary? How does this affect the benefit risk profile of a medicine? If an ADR occurs, how should the ADR be best managed? What risk mitigation should be considered and if implemented, how will the impact be measured?•Helps understand any difference in the likelihood, clinical presentation, severity, duration, and risk factors of any adverse effect, and the options in terms of preventability (such as evolving therapeutic options), as healthcare provision and other environmental factors evolve. Through surveillance, safety issues associated with a therapeutic intervention/immunisation can be detected and understood as early as possible. These issues are contextualised as more data accrues in routine real-world use, effective actions are taken as needed to mitigate risk, and their impact is measured [8.van Hunsel F. et al.Measuring the impact of pharmacovigilance activities, challenging but important.Br. J. Clin. Pharmacol. 2019; 85: 2235-2237Crossref PubMed Scopus (9) Google Scholar]. While pharmacovigilance has evolved, it falls short of the dramatic scientific and technological advances made elsewhere [9.Manyika J. et al.Disruptive Technologies: Advances That Will Transform Life, Business, and the Global Economy. McKinsey Global Institute, 2013Google Scholar], although it would be erroneous to imply that no impact has been observed. We describe avenues of particularly active or promising research (Figure 1, Key figure) that we believe are currently underappreciated (see Outstanding questions) but could be implemented globally, as they have the potential for fundamentally advancing routine pharmacovigilance. Data from individual case safety reports (ICSRs) remain the cornerstone of pharmacovigilance, although the extent to which an individual report does/might further understanding of the safety of a medicine varies (Figure 2A ). Improving insights gleaned from ICSRs has mostly focused on three aspects: (i) attempting to increase the overall reporting volume; (ii) enhancing the quality of the data submitted in reports; and (iii) improving methods to effectively and rapidly gain insights from available data. Tools, methods, and processes continue to evolve and support increased volume of reporting, and the ultimate intent of improving safety signal detection. Reports are now routinely collected from newer sources [e.g., text/audio/video discussions with chatbots or healthcare providers (HCPs)] and criteria for reporting have been widened (e.g., by including solicited and marketing programme-based AEs). Analytics capabilities have advanced to leverage these new data streams with increases in input data, such as the expanding use of natural language processing and similar technologies, for example, for computerised extraction of valuable information from clinical narratives [10.Walker A.M. et al.Computer-assisted expert case definition in electronic health records.Int. J. Med. Inform. 2016; 86: 62-70Crossref PubMed Scopus (27) Google Scholar], quantitative signal detection [11.Bate A. Evans S.J. Quantitative signal detection using spontaneous ADR reporting.Pharmacoepidemiol. Drug Saf. 2009; 18: 427-436Crossref PubMed Scopus (397) Google Scholar], and overall tools for holistic signal management [12.Almenoff J.S. et al.Online signal management: a systems-based approach that delivers new analytical capabilities and operational efficiency to the practice of pharmacovigilance.Drug Inf. J. 2007; 41: 779-789Crossref Scopus (4) Google Scholar]. Increased reporting has not directly translated into equivalent improvements in signal detection capability, leading sometimes to masking of safety issues [13.Jokinen J.D. et al.Pooling different safety data sources: impact of combining solicited and spontaneous reports on signal detection in pharmacovigilance.Drug Saf. 2019; 42: 1191-1198Crossref PubMed Scopus (5) Google Scholar]. An ICSR's pharmacovigilance utility is given by its novelty and the informativeness of specific elements recorded (e.g., onset and treatment dates); however, this can vary depending on the specific product exposure, as well as the AE. While an increased volume of ICSRs can improve the effectiveness of quantitative methods in detecting statistical alerts [14.Bate A. et al.Knowledge finding in IMS disease analyser Mediplus UK database - effective data mining in longitudinal patient safety data.Drug Saf. 2004; 27: 917-918Google Scholar], it is the number of actionable high-quality cases potentially affecting benefit–risk (Figure 2A) that most impacts the overall safety signal detection capability [15.Munoz M.A. et al.Towards automating adverse event review: a prediction model for case report utility.Drug Saf. 2020; 43: 329-338Crossref PubMed Scopus (4) Google Scholar]. Missing data in ICSRs is inherent to voluntary safety reporting, but sometimes the elements omitted may contain the very information essential in evaluating a report [16.Meyboom R.H.B. et al.Causal or casual? The role of causality assessment in pharmacovigilance.Drug Saf. 1997; 17: 374-389Crossref PubMed Scopus (200) Google Scholar]. Explicit legislative requirements apply to timely attempts to collect follow-up, but guidance is broad, rendering the follow-up of reports itself difficult, incomplete, time consuming, and most of the time unsuccessful [16.Meyboom R.H.B. et al.Causal or casual? The role of causality assessment in pharmacovigilance.Drug Saf. 1997; 17: 374-389Crossref PubMed Scopus (200) Google Scholar]. Near real-time follow-up (Figure 2B), focussed on ICSRs with the greatest pharmacovigilance utility and targeting only relevant information, might lead to a more complete and hence much stronger data on potential index cases and thus, to timely and more effective detection of safety issues. Quantitative methods are now routinely used to complement clinical review of ICSRs [11.Bate A. Evans S.J. Quantitative signal detection using spontaneous ADR reporting.Pharmacoepidemiol. Drug Saf. 2009; 18: 427-436Crossref PubMed Scopus (397) Google Scholar] and other safety data sources. Analysis has mainly focused on identifying drug/vaccine–event pairs across entire populations, with additional interest directed areas such as drug–drug interactions, clustering, and subgroup identification. The relative lack of progress is at least in part due to the nonrandom, often sparse data in spontaneous reports. This does not mean that further method improvement is not possible (and necessary) in the way ICSRs are currently analysed. Several directions are possible: advances in the ability to highlight drug/vaccine–event pairs or individual cases for clinical analysis, clustering of related reports or concepts, and highlighting of duplicates (all areas of active research [15.Munoz M.A. et al.Towards automating adverse event review: a prediction model for case report utility.Drug Saf. 2020; 43: 329-338Crossref PubMed Scopus (4) Google Scholar,17.Caster O. et al.Improved statistical signal detection in pharmacovigilance by combining multiple strength-of-evidence aspects in vigiRank.Drug Saf. 2014; 37: 617-628Crossref PubMed Scopus (57) Google Scholar, 18.Scholl J.H.G. et al.A prediction model-based algorithm for computer-assisted database screening of adverse drug reactions in the Netherlands.Pharmacoepidemiol. Drug Saf. 2018; 27: 199-205Crossref PubMed Scopus (9) Google Scholar, 19.Van Holle L. et al.Using time-to-onset for detecting safety signals in spontaneous reports of adverse events following immunization: a proof of concept study.Pharmacoepidemiol. Drug Saf. 2012; 21: 603-610Crossref PubMed Scopus (27) Google Scholar, 20.Erlanson N. Improving the Speed and Quality of an Adverse Event Cluster Analysis with Stepwise Expectation Maximization and Community Detection. Uppsala University, 2020Google Scholar, 21.Markatou M. Ball R. A pattern discovery framework for adverse event evaluation and inference in spontaneous reporting systems.Stat. Anal. Data Min. 2014; 7: 352-367Crossref Scopus (5) Google Scholar]); or development of more sophisticated analyses to group and learn from clinical similarity across different terms used to capture outcomes in hierarchical terminologies [22.Brown E.G. Methods and pitfalls in searching drug safety databases utilising the Medical Dictionary for Regulatory Activities (MedDRA).Drug Saf. 2003; 26: 145-158Crossref PubMed Scopus (54) Google Scholar, 23.Lagerlund O. et al.WHODrug: a global, validated and updated dictionary for medicinal information.Ther. Innov. Regul. Sci. 2020; 54: 1116-1122Crossref PubMed Scopus (11) Google Scholar, 24.Bate A. et al.Terminological challenges in safety surveillance.Drug Saf. 2012; 35: 79-84Crossref PubMed Scopus (10) Google Scholar, 25.Crooks C.J. et al.Identifying adverse events of vaccines using a Bayesian method of medically guided information sharing.Drug Saf. 2012; 35: 61-78Crossref PubMed Scopus (9) Google Scholar, 26.Gattepaille L.M. Using the WHO database of spontaneous reports to build joint vector representations of drugs and adverse drug reactions, a promising avenue for pharmacovigilance.in: 2019 IEEE International Conference on Healthcare Informatics (ICHI), 10–13 June 2019. IEEE, Xi'an, China2019: 1-6Crossref Scopus (3) Google Scholar, 27.Natsiavas P. et al.Computational advances in drug safety: Systematic and mapping review of knowledge engineering based approaches.Front. Pharmacol. 2019; 10: 415Crossref PubMed Scopus (5) Google Scholar]. In addition to the use of established registries, real-world evidence is derived from secondary analysis of healthcare databases, electronic health records, and transactional insurance claims databases. Real-world data (RWD) are commonly used in epidemiological studies testing specific hypotheses and for other applications across the entire lifecycle of a product [28.Bate A. et al.Designing and incorporating a real world data approach to international drug development and use: what the UK offers.Drug Discov. Today. 2016; 21: 400-405Crossref PubMed Scopus (20) Google Scholar], although its use remains more exploratory for signal detection [29.Bate A. et al.Hypothesis-free signal detection in healthcare databases: finding its value for pharmacovigilance.Ther. Adv. Drug Saf. 2019; 10 (2042098619864744)Crossref PubMed Scopus (5) Google Scholar]. Rapidly increasing coverage and accessibility of healthcare data has led to rapid cycle analysis; a process in which RWD of relevance for pharmacovigilance are used to readily contextualise information and allow near real-time detection of AEs [30.Mahaux O. et al.Pharmacoepidemiological considerations in observed-to-expected analyses for vaccines.Pharmacoepidemiol. Drug Saf. 2016; 25: 215-222Crossref PubMed Scopus (15) Google Scholar,31.Platt R. et al.The FDA Sentinel initiative – an evolving national resource.N. Engl. J. Med. 2018; 379: 2091-2093Crossref PubMed Scopus (82) Google Scholar]. Technological advances such as distributed networks [32.Curtis L.H. et al.Design considerations, architecture, and use of the Mini-Sentinel distributed data system.Pharmacoepidemiol. Drug Saf. 2012; 21: 23-31Crossref PubMed Scopus (160) Google Scholar], where analyses are conducted without requiring central pooling of patient-level data, have enabled rapid generation of evidence across large RWD networks. RWD are still not always present and available internationally, or even linked widely across all forms of care (e.g., speciality medicine usage in UK hospitals [28.Bate A. et al.Designing and incorporating a real world data approach to international drug development and use: what the UK offers.Drug Discov. Today. 2016; 21: 400-405Crossref PubMed Scopus (20) Google Scholar]). Also, while accumulating at an unprecedented pace, RWD remain necessarily imperfect given their secondary use for pharmacovigilance and limited access. The ability to link these data is often also restricted by privacy regulations and/or other concerns. However, linking across data sources and enriching RWD by further primary data collection can sometimes mitigate their limitations, albeit not uniformly [33.Alshammari T.M. et al.National pharmacovigilance programs in Arab countries: a quantitative assessment study.Pharmacoepidemiol. Drug Saf. 2020; 29: 1001-1010Crossref PubMed Scopus (6) Google Scholar, 34.Jusot V. et al.Enhancing pharmacovigilance in Sub-Saharan Africa through training and mentoring: a GSK pilot initiative in Malawi.Drug Saf. 2020; 43: 583-593Crossref PubMed Scopus (2) Google Scholar, 35.Pisaniello H.L. Dixon W.G. What does digitalization hold for the creation of real-world evidence?.Rheumatology (Oxford). 2020; 59: 39-45Crossref PubMed Scopus (3) Google Scholar, 36.Dreyer N.A. et al.Primary Data Collection for Pharmacoepidemiology.6th edn. John Wiley & Sons, 2020Google Scholar]. One can anticipate that with the wider use of clinical support systems, the integration of spontaneous reporting capabilities embedded into tools [37.Linder J.A. et al.Secondary use of electronic health record data: spontaneous triggered adverse drug event reporting.Pharmacoepidemiol. Drug Saf. 2010; 19: 1211-1215Crossref PubMed Scopus (45) Google Scholar] will eventually lead to the often merging or at least closer alignment of the two distinct data streams. Spontaneous ICSRs from HCPs can be collected and reported from healthcare databases [37.Linder J.A. et al.Secondary use of electronic health record data: spontaneous triggered adverse drug event reporting.Pharmacoepidemiol. Drug Saf. 2010; 19: 1211-1215Crossref PubMed Scopus (45) Google Scholar], linked to routinely collected data on the rationale for clinical suspicion (suspected medicine, prior history, dietary detail, or diagnostic justification), thus potentially providing rich data for analysis. Similarly, this may occur for suspicions of AEs reported in social media. We expect the proportion of isolated spontaneous reporting relative to that linked to routine longitudinal health-related data to decrease over time. Umbrella terms like social or digital media include a wide spectrum of data sources from (potential) patients and safety outcomes, transmitted through various channels, ranging from general information available on websites to platforms dedicated to specific patient groups or diseases/illnesses. Overall use of social media is increasing [38.Nwosu A.C. et al.Social media and palliative medicine: a retrospective 2-year analysis of global Twitter data to evaluate the use of technology to communicate about issues at the end of life.BMJ Support. Palliat. Care. 2015; 5: 207-212Crossref PubMed Scopus (29) Google Scholar], so determining its best utilisation is a key ongoing challenge for pharmacovigilance [39.Bate A. et al.The hope, hype and reality of Big Data for pharmacovigilance.Ther. Adv. Drug Saf. 2018; 9: 5-11Crossref PubMed Scopus (21) Google Scholar]. While some research is underwhelming in suggesting its value and it will certainly not become a panacea [40.van Stekelenborg J. et al.Recommendations for the use of social media in pharmacovigilance: lessons from IMI WEB-RADR.Drug Saf. 2019; 42: 1393-1407Crossref PubMed Scopus (29) Google Scholar], there are specific examples where social media seem particularly promising for safety surveillance. For instance, mining data from popular and open French forums allowed the identification of unexpected misuse behaviours for methylphenidate [41.Chen X. et al.Mining patients' narratives in social media for pharmacovigilance: adverse effects and misuse of methylphenidate.Front. Pharmacol. 2018; 9: 541Crossref PubMed Scopus (19) Google Scholar]. Given the scale of internet use to discuss general issues, including wellbeing, a nuanced or at least niche role of social media seems clear for pharmacovigilance. The Internet of Things is a technological concept that encompasses connected wireless communication systems, sensor networks, and machine-to-machine communications [42.Atzori L. et al.Understanding the Internet of Things: definition, potentials, and societal role of a fast evolving paradigm.Ad Hoc Netw. 2017; 56: 122-140Crossref Scopus (282) Google Scholar]. This concept is continuously expanding and may materialise as a future wide interconnected system in information and communication technology. The ability to link at scale huge numbers of different apps/devices has led to the emergence of various data streams with potential direct or indirect value for safety surveillance. Such examples are electrocardiogram signals from mobile heart monitoring, X-ray images, video and audio data [39.Bate A. et al.The hope, hype and reality of Big Data for pharmacovigilance.Ther. Adv. Drug Saf. 2018; 9: 5-11Crossref PubMed Scopus (21) Google Scholar,43.Bate A. Hobbiger S.F. Artificial intelligence, real-world automation and the safety of medicines.Drug Saf. 2021; 44: 152-132Crossref Scopus (4) Google Scholar], and digital biomarkers in clinical trial settings [44.Hamy V. et al.Developing smartphone-based objective assessments of physical function in rheumatoid arthritis patients: the PARADE study.Digit. Biomark. 2020; 4: 26-43Crossref PubMed Scopus (6) Google Scholar]. To date, there is limited research on the impact of sensor data on safety surveillance; however, some studies have described proof-of-concept work for aggregating multiple data streams, including from personal digital devices [45.Dhruva S.S. et al.Aggregating multiple real-world data sources using a patient-centered health-data-sharing platform.NPJ Digit. Med. 2020; 3: 60Crossref PubMed Scopus (17) Google Scholar]. We expect a scale-up of such projects in the future, as general interconnectivity of devices continues to increase, and anticipate their use in pharmacovigilance. Historically, safety analyses were primarily characterised by the use of different data types at specific timepoints in the medicinal product's lifecycle: clinical trials for understanding human safety before approval, postmarketing spontaneous reporting for identifying safety signals, and epidemiological studies for assessing potential safety issues. These analyses were often conducted with data from a single database. It has now become more common to see wider usage of a single type of data for different purposes across the medicine/vaccine development lifecycle (e.g., RWD [28.Bate A. et al.Designing and incorporating a real world data approach to international drug development and use: what the UK offers.Drug Discov. Today. 2016; 21: 400-405Crossref PubMed Scopus (20) Google Scholar]), and across multiple data sets of the same data type. However, recent research has focused on the use of various types of data streams to generate a more complete picture of specific safety issues, detected in single or same-type databases. Such integrated approaches are used to contextualise specific issues like safety signals (e.g., examining cardiac outcomes for sildenafil) in randomised controlled trials, spontaneous reports, and RWD [46.Sobel R.E. Reynolds R.F. Integrating evidence from multiple sources to evaluate post-approval safety: an example of sildenafil citrate and cardiovascular events.Curr. Med. Res. Opin. 2008; 24: 1861-1868Crossref PubMed Scopus (8) Google Scholar], but further research is needed to apply them at scale for other pharmacovigilance purposes. For example, when weighing data inputs for each data stream, a nuanced approach would be needed for effective semiautomated/quantitative data fusion in pharmacovigilance, perhaps leveraging decision trees to support consistent, transparent methodologies. The relative weighing of importance of each data stream varies by type of outcome and exposure (accuracy of data capture, frequency of outcome occurrence, background untreated population incidence, etc.). Treating the data types themselves less discretely may encourage a more holistic approach in pharmacovigilance, taking advantage of the strengths and limitations of the individual data types. Safety surveillance have traditionally been focussed on centralised systems. Since the early days of pharmacovigilance, the international reach of spontaneous reporting has increased [47.Olsson S. The role of the WHO programme on International Drug Monitoring in coordinating worldwide drug safety efforts.Drug Saf. 1998; 19: 1-10Crossref PubMed Scopus (109) Google Scholar], along with the number of organisations expecting copies of such reports. The original recipient of an ICSR, in addition to managing and acting on the report, has also legislative obligations to share the ICSR with multiple partners (e.g., other companies or regulators) [48.Ghosh R. et al.Automation opportunities in pharmacovigilance: an industry survey.Pharmaceut. Med. 2020; 34: 7-18Crossref PubMed Scopus (8) Google Scholar]. In this context, for example, unidentified duplicate copies of ICSRs can readily propagate and hinder effectiveness of the system [49.Hauben M. et al.'Extreme duplication' in the US FDA Adverse Events Reporting System database.Drug Saf. 2007; 30: 551-554Crossref PubMed Scopus (45) Google Scholar] (Box 2).Box 2Increasing reporting and expanding international regulation lead to high reporting burden•An increase in reporting rates has been observed. For instance, in a survey of large biopharmaceutical companies, the median number of ICSRs was reported to have increased from 84 960 in 2007 to over 200 000 in 2017. This was largely attributable to an increase in both nonserious cases and follow-up cases [55.Stergiopoulos S. et al.Adverse drug reaction case safety practices in large biopharmaceutical organizations from 2007 to 2017: an industry survey.Pharmaceut. Med. 2019; 33: 499-510PubMed Google Scholar].•Other types of reports have also been introduced over the years, to convey information on AEs/ADRs from pre- or postmarketing experience, with varying formats and obligations (drug safety update reports, periodic safety update reports, risk-management plans, etc.).•A high number of ICSRs are often uninformative, and also provide limited or no ability to obtain necessary follow-up information [16.Meyboom R.H.B. et al.Causal or casual? The role of causality assessment in pharmacovigilance.Drug Saf. 1997; 17: 374-389Crossref PubMed Scopus (200) Google Scholar]. Thes