VigiBase, the World Health Organization’s global database of individual case safety reports (ICSRs) for medicines and vaccines, is maintained by the Uppsala Monitoring Centre (UMC) in Sweden. As of 31 December 2024, VigiBase contained over 40 million unique case reports from more than 160 countries, making it the largest and most comprehensive pharmacovigilance database worldwide.
This article provides a detailed medical and scientific overview of VigiBase, based on the 2026 resource profile update published by UMC researchers. It covers the database’s history, data collection and processing workflows, coding standards (WHODrug Global and MedDRA), analytical tools (including vigiMatch, vigiGrade, vigiRank, and the VigiBase pregnancy algorithm), and key descriptive statistics. Special attention is given to global patterns and trends in adverse event reporting, including the increasing contribution from Asia and non‑physician reporters, the rise in reports of medication errors and off‑label use, and the growing prominence of antineoplastic and immunosuppressant drugs.
The article also discusses the strengths and limitations of VigiBase as a tool for signal detection, scientific research, and public health decision‑making. Understanding these features is essential for pharmacovigilance professionals, researchers, and regulators who rely on VigiBase for safety monitoring and signal management.
1. Introduction: The Genesis and Mission of VigiBase
The thalidomide tragedy of the early 1960s exposed the catastrophic consequences of inadequate post‑marketing drug safety monitoring. In response, the World Health Organization launched the WHO Programme for International Drug Monitoring (PIDM) in 1968, creating a collaborative network of national pharmacovigilance centres. To support this programme, the Uppsala Monitoring Centre (UMC) in Sweden was entrusted with the development and maintenance of a global database of adverse event reports – VigiBase.
Over the decades, VigiBase has grown from a modest collection of paper forms to an electronic repository containing tens of millions of individual case safety reports. As of 31 December 2024, the database had processed over 56 million reports, representing approximately 40 million unique cases after deduplication and follow‑up adjustments. Roughly 80% of these reports concern medicines, 14.4% COVID‑19 vaccines, and 5.1% non‑COVID‑19 vaccines; only 0.3% involve more than one of these product categories.
The primary purpose of VigiBase is to support signal management – the detection, assessment, and understanding of potential causal relationships between medicinal products and adverse events. Its vast scale enables the study of rare adverse events that would be impossible to detect in national databases alone, and its global coverage allows cross‑regional comparisons that strengthen the evidence for safety signals. Notable examples of signals first identified or confirmed using VigiBase include:
- Neonatal withdrawal syndrome following maternal use of selective serotonin reuptake inhibitors (SSRIs) during pregnancy
- Myocardial infarction associated with the antiretroviral abacavir
- Tachycardia with the prokinetic agent cisapride
- Drug‑drug interactions, such as those detected using the Omega disproportionality measure
Beyond signal detection, VigiBase data are increasingly used to characterise the temporality, dose‑response, reversibility and seriousness of adverse reactions, thereby informing clinical practice and the design of follow‑up epidemiological studies.
2. Data Capture and Management
2.1 Sources of Reports
VigiBase aggregates ICSRs from national pharmacovigilance databases of WHO PIDM members, which currently include more than 160 countries. Key sources are:
- The US FDA Adverse Event Reporting System (FAERS) for medicines
- The US Vaccine Adverse Event Reporting System (VAERS)
- EudraVigilance, the European centralised pharmacovigilance database (from 2017 onwards)
- National databases from Asia, Africa, Latin America and the Caribbean, and Oceania
The database is continually updated; most members share reports at least monthly, with some (e.g., EudraVigilance) providing daily updates. However, there is considerable variation in reporting practices: some centres accept only serious adverse events, others collect any event regardless of causality, and the classification of a product as “suspected”, “interacting” or “concomitant” differs between reporters and organisations.
2.2 Reporting Formats and Data Harmonisation
Historically, reports were submitted using the INTDIS (International Drug Information System) format. Today, most member organisations use the ICH E2B format (versions R2 and R3). To standardise analyses, reports in INTDIS and E2B R3 are currently converted to a version resembling E2B R2. However, a complete upgrade to E2B R3 is planned for 2026.
Standardisation is also achieved through the use of WHODrug Global (UMC’s drug dictionary, which assigns Anatomical Therapeutic Chemical (ATC) codes and includes the Herbal ATC system) and MedDRA (the Medical Dictionary for Regulatory Activities, used for coding adverse events and indications). Where automatic coding fails, human experts intervene. Historical terminologies such as WHO‑ART and ICD‑10 are mapped to MedDRA.
2.3 Processing and Data Cleaning
Upon receipt, each report is checked against the minimum criteria for a valid ICSR: a reporting country, a national case identification number, at least one adverse event term, and at least one medicinal product term. Additional useful data elements (e.g., patient age/sex, dose, start/stop dates, indication, seriousness, outcome, dechallenge/rechallenge) are encouraged but not mandatory.
Subsequent processing steps include:
- Identification of foreign reports (reported by a country other than where the event occurred) – these are generally excluded because they are likely to be duplicates.
- Duplicate detection using the vigiMatch algorithm, which applies probabilistic pattern matching to identify suspiciously similar report pairs.
- Exclusion of reports where all products are listed as concomitant only.
- Calculation of time‑to‑onset when start and event dates are provided.
- Application of the vigiGrade completeness score, which helps users assess the amount of clinically relevant information in each report.
3. Analytical Tools and Methods
The UMC has developed a suite of algorithms and methods that are integrated into VigiBase analysis tools (see Table 1). Some are available to external users through VigiLyze or custom search services.
| Method / Algorithm | Description |
|---|---|
| Information Component (IC) | Pairwise disproportionality measure using a shrinkage observed‑to‑expected ratio; widely used for signal detection. |
| Omega interaction measure | Disproportionality method specifically for drug‑drug interactions. |
| vigiGrade | Assigns a completeness score (0‑1) to each report based on availability of key data fields. |
| vigiMatch | Probabilistic duplicate detection algorithm. |
| vigiRank | Multi‑feature predictive model for ranking emerging safety signals; used by UMC for internal signal detection since 2015. |
| vigiPoint | Identifies differences in report characteristics between data subsets, accounting for absolute frequency and strength of association. |
| vigiGroup | Consensus clustering to identify reports with similar adverse event profiles. |
| vigiVec | Semantic vector representations of adverse events and drugs, derived from co‑reporting patterns. |
| VigiBase unmasking algorithm | Detects influential outliers that may hide true associations. |
| VigiBase pregnancy algorithm | Rule‑based tool to identify reports involving pregnancy exposure, even in the absence of a pregnancy‑related adverse event. |
All these methods have been published in peer‑reviewed journals, ensuring transparency and reproducibility. Access to VigiBase is governed by the VigiBase Data Access Conditions; different levels of access are available to PIDM members (via VigiLyze), marketing authorisation holders, contract research organisations, academia (via VigiBase Search Services), and the general public (via VigiAccess, which provides high‑level summaries by active ingredient).
4. Descriptive Statistics of VigiBase (as of 31 December 2024)
4.1 Overall Volume and Growth
- Unique cases: ~39.9 million (after exclusions)
- Reports from medicines: 80.2%
- Reports from COVID‑19 vaccines: 14.4%
- Reports from non‑COVID vaccines: 5.1%
- Reports involving multiple product categories: 0.3%
Approximately 70% of all reports have been received in the past 10 years, and 40% in the past 5 years. This growth reflects expansion of the PIDM, digital reporting tools (including mobile apps), legislative changes, patient involvement, and intensive reporting during the COVID‑19 vaccination campaigns.
4.2 Patient Demographics
- Adults (≥18 years): 90% of reports
- Children and adolescents (0‑17 years): 10%
- Female sex: 60.7% of all reports (consistent across product types)
- Male sex: 39.3%
For non‑COVID vaccines, about 60% of reports involve individuals under 18 years, aligning with childhood immunisation schedules.
4.3 Geographic Distribution
| Region | Percentage of all reports |
|---|---|
| USA & Canada | 43.7% |
| Europe | 24.7% |
| Asia | 23.2% |
| Latin America & Caribbean | 4.2% |
| Oceania | 2.0% |
| Africa | 2.1% |
Using World Bank income groups: high‑income countries contribute 82% of reports, upper‑middle‑income 13%, lower‑middle‑income 4.5%, and low‑income 0.5%. This imbalance reflects differences in healthcare infrastructure, medicine access, and pharmacovigilance maturity.
4.4 Reporter Characteristics
- Spontaneous reports: 86.4% (96% for vaccines)
- Reports from studies: 12.3%
- Marketing authorisation holder (MAH) reports: 50.3% (56% for medicines, 25% for vaccines)
Reporter qualifications (all reports):
- Physicians: 31.4%
- Pharmacists: 10.4%
- Other healthcare professionals: 18.0%
- Consumers / non‑HCPs: 38.4%
For COVID‑19 vaccines, consumers/non‑HCPs were the most common reporter category (55.3%).
4.5 Product and Event Complexity
- Single suspected/interacting active ingredient: 88.7% of reports
- 2‑5 active ingredients: 10.8%
- >5 active ingredients: 0.5%
- Single MedDRA Preferred Term (PT): 47.5%
- 2‑5 PTs: 44.7%
- >5 PTs: 7.8%
Vaccine reports are more likely to list multiple PTs.
4.6 Seriousness and Data Completeness
| Seriousness criterion | All reports (%) | Medicines (%) | Non‑COVID vaccines (%) | COVID‑19 vaccines (%) |
|---|---|---|---|---|
| Death | 3.5 | 4.1 | 0.7 | 1.2 |
| Life‑threatening | 1.5 | 1.6 | 1.2 | 1.4 |
| Hospitalisation | 11.1 | 12.0 | 8.1 | 7.1 |
| Disabling | 1.3 | 1.0 | 1.7 | 2.8 |
| Congenital anomaly | 0.1 | 0.1 | 0.0 | 0.0 |
| Other medically important | 16.9 | 18.9 | 5.8 | 10.2 |
Data availability:
- Time‑to‑onset (TTO): 59.1% overall (85% for vaccines, 53% for medicines)
- Dose information: 34.3% overall (40% for medicines, 15% for vaccines)
- Indication: 58.9% overall (65% for medicines, 30% for vaccines)
5. Recent Trends in Reporting (2015‑2024 vs 1968‑2014)
To identify meaningful changes in reporting patterns, the authors compared the ~10 million reports submitted before 1 January 2015 with the ~24 million reports submitted thereafter, excluding COVID‑19 vaccine reports to avoid distorting trends.
5.1 Geographic Expansion
- Asia: increased from 13% of reports to 29%
- Latin America & Caribbean: more than doubled (still small absolute proportion)
- Africa: also more than doubled
These changes reflect the progressive expansion of the WHO PIDM and improved reporting infrastructure.
5.2 Rise of Non‑Physician Reporters
| Reporter type | Pre‑2015 | 2015‑2024 |
|---|---|---|
| Physicians | ~45% | ~28% |
| Pharmacists | 8% | 13% |
| Other HCPs | 11% | 22% |
| Consumers / non‑HCPs | 28% | 42% |
This shift is driven by legislative changes that allow and promote direct patient reporting, enable pharmacists and nurses to act as independent reporters, and increased patient engagement in pharmacovigilance networks.
5.3 Changes in Reported Medicinal Products
Using vigiPoint, the study identified ATC therapeutic subgroups with increased relative representation after 2015:
- Antineoplastic agents: from 6.3% → 11.6% of reports
- Immunosuppressants: from 9.0% → 14.2%
Conversely, reports on cardiovascular, nervous system, musculoskeletal and genitourinary drugs declined in relative share. These trends align with regulatory approval data showing a strong pipeline of anti‑cancer and immunomodulatory drugs, while approvals for neurological and cardiovascular conditions have slowed.
5.4 Changes in Reported Adverse Events (MedDRA HLGTs)
The most striking global trends are:
- Medication errors and other product use errors/issues: increased more than three‑fold (from 2.5% → 8.7% of reports)
- Off‑label uses and intentional product misuses/use issues: increased five‑fold (from 0.6% → 3.0%)
These rises reflect the gradual incorporation of medication errors and off‑label use into pharmacovigilance legislation and regulatory frameworks.
Other notable increases:
- Viral infection disorders (driven by COVID‑19 and related testing)
- Device issues (partly due to EU medical device regulation changes and specific product concerns, e.g., levonorgestrel‑releasing IUDs)
Relative declines were observed in cardiac disorders (e.g., arrhythmias), vascular disorders (e.g., hypotension), nervous system disorders (e.g., parkinsonism), and skin disorders (e.g., urticaria). These remain common but their proportional share has decreased because of the rapid growth of other event categories.
6. Strengths and Limitations of VigiBase
6.1 Strengths
- Global scale: >40 million reports from >160 countries enable detection of very rare events and analysis of small subgroups.
- Continual data collection: Real‑time accumulation allows prompt detection of new safety signals.
- Broad coverage: Includes all medicinal products, off‑label uses, traditional medicines, and combination therapies.
- Diverse reporter perspectives: Healthcare professionals report serious events and drug‑related issues; patients provide detailed descriptions of symptom impact and circumstances.
- Resource for low‑income countries: Access to global data supports smart, cost‑efficient local pharmacovigilance policies (WHO Global Smart PV Strategy).
- Methodological innovation: UMC’s openly published algorithms (IC, vigiMatch, vigiRank, etc.) provide advanced tools for signal detection and data quality assessment.
6.2 Limitations
- Geographic imbalance: Most reports still originate from high‑income Western countries (USA, Canada, Europe), despite recent progress in Asia and Africa. This limits generalisability to low‑ and middle‑income settings.
- Heterogeneity: Variations in product use, healthcare access, reporting practices, coding conventions, and legal frameworks make global data interpretation challenging. Subgroup analyses by country or region are essential.
- Data quality and completeness: Many reports lack crucial information (e.g., dose, time‑to‑onset, indication). Completeness scores (vigiGrade) help, but structural improvements in reporting systems are needed.
- Duplicate reports and foreign reports: Despite vigiMatch, duplicates and foreign reports are imperfectly identified; they are generally excluded from analyses, but some true duplicates may remain.
- Reporting biases: Spontaneous reporting is subject to under‑reporting, stimulated reporting, and notoriety bias. These biases can affect signal detection and must be considered during evaluation.
- Causality assessment: VigiBase contains reports of “suspected” adverse events; the database itself does not establish causality. Signals identified must be followed up with more rigorous epidemiological studies.
7. Practical Implications for Pharmacovigilance Professionals
7.1 Using VigiBase for Signal Management
- Signal detection: Use vigiRank, IC, or other disproportionality measures to generate hypotheses. However, always interpret results in the context of the database’s limitations (e.g., geographic and reporter biases).
- Signal assessment: Supplement statistical outputs with case‑level review using VigiLyze. Pay attention to time‑to‑onset, dechallenge/rechallenge, and alternative explanations (concomitant drugs, medical history).
- Comparison across regions: Consistent patterns across multiple countries strengthen the plausibility of a signal. Use vigiPoint to explore differences between subsets.
7.2 Improving Reporting Quality
- Encourage complete reports: advocate for inclusion of patient age/sex, dose, dates, indication, outcome, and narrative description.
- Train reporters on the importance of distinguishing suspected, interacting and concomitant products.
- Promote use of structured reporting tools (e.g., VigiFlow, VigiMobile) that facilitate complete data capture.
7.3 Accessing VigiBase Data
- PIDM members: Full access through VigiLyze.
- Marketing authorisation holders and researchers: VigiBase Search Services (aggregate or case‑level data) subject to data access conditions.
- Public: VigiAccess for high‑level summaries by active ingredient.
All users must adhere to the VigiBase Caveat document, which states restrictions and limitations on data use.
8. Future Directions
The paper highlights several ongoing and planned developments:
- Full migration to E2B R3 format (planned for 2026) to take advantage of improved data elements.
- Integration of ISO IDMP standards for richer, more precise medicinal product information (route of administration, strength, etc.).
- Artificial intelligence (AI) for automated free‑text processing to extract information from narratives, including multilingual data.
- Continued expansion of WHO PIDM to increase coverage from Africa and Latin America.
- WHO Global Smart Pharmacovigilance Strategy to promote reliance on global data for broad insights while focusing local monitoring on high‑priority products.
9. Conclusion
VigiBase is the cornerstone of global pharmacovigilance. Its unprecedented size and scope allow the detection of rare adverse events, the analysis of drug‑drug interactions, and the identification of safety signals that might otherwise remain hidden. The 2026 resource profile update provides a wealth of information on the database’s structure, processing, analytical tools, and evolving patterns of reporting.
Key takeaways for pharmacovigilance professionals:
- VigiBase has grown dramatically, especially from Asia and non‑physician reporters.
- Medication errors, off‑label use, and antineoplastic/immunosuppressant drugs are the fastest‑growing areas of reporting.
- Data completeness and geographic heterogeneity remain major limitations; always perform stratified analyses and use completeness scores.
- UMC’s algorithms (vigiMatch, vigiGrade, vigiRank, etc.) are valuable resources for signal detection and data cleaning.
- Proper analysis and interpretation of VigiBase data require awareness of its strengths and limitations – a point emphasised by the authors throughout the paper.
As the global pharmacovigilance community continues to expand and new technologies emerge, VigiBase will remain an indispensable tool for protecting public health. Understanding its features, trends, and caveats is essential for anyone who relies on spontaneous reporting data for safety monitoring or research.
