Global Vaccine Safety

Global Advisory Committee on Vaccine Safety, report of meeting held 6-7 June 2012

Published in the WHO Weekly Epidemiological Record on 27 July 2012

The Global Advisory Committee on Vaccine Safety (GACVS), an expert clinical and scientific advisory body, was established by WHO to provide independent, scientifically rigorous advice on vaccine safety issues of potential global importance.1 GACVS held its 26th meeting in Geneva, Switzerland, on 6–7 June 2012.2 The Committee reviewed the following specific topics:

  • the safety of thiomersal;
  • the safety of aluminium adjuvants;
  • the safety profile of influenza vaccines during pregnancy.

It also reviewed 3 general issues for vaccine pharmacovigilance:

  • safety of immunization during pregnancy and lactation;
  • causality assessment for serious individual cases of adverse events following immunization (AEFI);
  • core variables for AEFI monitoring.

Thiomersal in vaccines

In 1999, concerns were raised in the United States of America (USA) regarding exposure to mercury following immunization with thiomersal-containing vaccines. This was based on the calculation that the cumulative amount of mercury in primary infant immunization schedules in the USA potentially exceeded the recommended threshold set by its Environmental Protection Agency for methyl mercury. Hence, the policy decision in the USA to use only vaccines without thiomersal was based on a precautionary principle founded on the presumption of equal pharmacokinetics of ethyl mercury and methyl mercury, despite the fact that thiomersal contains only ethyl mercury.

Between 2002 and 2008, GAVCS reviewed several pharmacokinetic and epidemiological studies concerning thiomersal. Pharmacokinetic data in human infants, including premature and low birth-weight infants, established that the half-life of ethyl mercury is 3–7 days, and that ethyl mercury is efficiently excreted in the stools and does not accumulate over the long-term in blood, since levels returned to baseline within 30 days of vaccination.

At the June 2012 meeting, GACVS reviewed the most recently available information concerning the safety of thiomersal since it last reviewed this topic in 2008. A comprehensive review identified 28 publications that addressed mercury blood levels in the short and long term following vaccine administration, and epidemiological studies that examined the relation between thiomersal receipt and several health outcomes. Three ecological studies suggesting an association between thiomersal and neurodevelopmental disorders were found to be fraught with methodological flaws. In addition, the continuous increase in the number of cases of autism diagnosed in the USA despite removal of thiomersal from most vaccines strongly argues against a causal association (fulfilling the exposure and removal criteria). All other studies reviewed, which were conducted with more robust epidemiological designs and in different countries, failed to identify any association with neurodevelopmental disorders.

Recently published studies confirm that in all populations studied, including pre-term and low birth-weight babies, the half-life of ethyl mercury in blood is between 3 and 7 days. A quantitative risk assessment model for cumulative toxicity of thiomersal in humans by US Federal Drug Administration (FDA) was also reviewed. This methodology is based on a pharmacokinetic model of ethyl mercury and provides a framework for interpreting studies in animals and humans that evaluate linkages among dose, blood and brain levels, and toxicity. Using this framework the GACVS concluded that animal or human toxicity studies suggest that the levels of ethyl mercury attained in the blood and brain from cumulative doses of vaccines do not reach toxic levels, making biologically implausible any relation between thiomersal in vaccines and neurological toxicity.

Based on the current evidence, GACVS considers that no additional studies of the safety of thiomersal in vaccines are warranted and that available evidence strongly supports the safety of the use of thiomersal as a preservative for inactivated vaccines. GACVS believes that consideration of additional evidence suggestive of the contrary should be based on studies using the same high standards of epidemiological and causal inference needed for scientific research. Thiomersal allows millions of people worldwide to have access to life-saving vaccines and to date, no other safer and equally efficacious alternative has been identified for many vaccines.

Aluminium adjuvants

The GACVS reviewed 2 published papers alleging that aluminium in vaccines is associated with autism spectrum disorders3, 4 and the evidence generated from quantitative risk assessment by a US FDA pharmacokinetic model of aluminium-containing vaccines.

GACVS considers that these 2 studies3, 4 are seriously flawed. The core argument made in these studies is based on ecological comparisons of aluminium content in vaccines and rates of autism spectrum disorders in several countries. In general, ecological studies cannot be used to assert a causal association because they do not link exposure to outcome in individuals, and only make correlations of exposure and outcomes on population averages. Therefore their value is primarily for hypothesis generation. However, there are additional concerns with those studies that limit any potential value for hypothesis generation. These include: incorrect assumptions about known associations of aluminium with neurological disease, uncertainty of the accuracy of the autism spectrum disorder prevalence rates in different countries, and accuracy of vaccination schedules and resulting calculations of aluminium doses in different countries.

The GACVS also reviewed the US FDA risk assessment model of aluminium in vaccines. The FDA calculations incorporate the most recently published aluminium risk assessments by adjusting for gastrointestinal absorption and uptake from the site of injection. The FDA analysis indicates that the body burden of aluminium following injections of aluminium-containing vaccines never exceeds safe US regulatory thresholds based on orally ingested aluminium even for low birth-weight infants. GACVS concludes that this comprehensive risk assessment further supports the clinical trial and epidemiological evidence of the safety of aluminium in vaccines. Current research on pharmacokinetics of aluminium in vaccines is ongoing and should be encouraged as a means of further validating and improving this model.

Vaccine safety in pregnancy and lactation

Several available vaccines have the potential to reduce maternal and fetal morbidity and mortality from preventable diseases. Thus, optimal protection against preventable diseases that pose a higher risk for disease and death in pregnant woman and their offspring should be balanced against the risk of malformations, abortions, stillbirth or other adverse outcomes that theoretically could affect the fetus as a result of vaccination in pregnancy. Maternal antibodies induced by vaccination during pregnancy are actively transferred to the fetus and confer passive protection in the infant after birth. GACVS recently established a subgroup to review the safety profile of several important vaccines for pregnant and lactating women. In addition to the review of available data on influenza vaccines described below, the committee also reviewed the accumulated safety data for rubella-containing vaccines when inadvertently administered to pregnant women to complement the review conducted in June 2008. GACVS concludes that the data remain very reassuring for the use of vaccines during pregnancy, with no evidence of adverse fetal outcomes identified. Protection of mothers at risk and their young infants will be critical to attain the reduction of morbidity and mortality due to infections that affect many populations around the world.

Use of influenza vaccines during pregnancy

As the risk of influenza disease is increasingly recognized among pregnant women and a growing body of evidence supports the benefits to infants of maternal vaccination, the committee reviewed the safety data available for influenza vaccines derived from clinical trials, observational studies, and spontaneous reporting. The data confirm the safety of non-adjuvanted trivalent inactivated seasonal influenza vaccines in pregnancy. For example in the USA outcomes of pregnancy were assessed in 3719 vaccinated pregnant women compared with 45 866 controls in the Vaccine Safety Datalink during the period 1997–2002. From 1990–2009, an estimated 11.8 million pregnant women were vaccinated in the USA. In addition, a review of spontaneous reports found no maternal deaths, no unexpected pattern of adverse pregnancy events or fetal outcomes, and no increased risk of adverse pregnancy outcomes when compared to background rates. Extensive evaluation during the 2009 influenza A(H1N1)pdm09 pandemic supported the safety of adjuvanted and non-adjuvanted influenza vaccines when used in pregnant women; overall, the safety profile was comparable to seasonal influenza vaccine in non-pregnant adults, and there was no evidence of teratogenicity or any other negative impact on pregnancy outcomes. In addition, preliminary data from a few studies of influenza vaccine in pregnant women have confirmed not only the benefit of providing protection in this vulnerable population, but positive effects in their infants, including the reduction of low birth weights, and a significant decrease in influenza-related pneumonia in young children.

Causality assessment of Adverse Events following Immunization

Attribution of causality to AEFI, especially those considered severe, of public importance, and programmatically disruptive, are critical for ensuring vaccine safety. In 2005, WHO published an aide-mémoire to a systematic, standardized causality assessment process for serious AEFI (including clusters), providing a method for individual causality assessment to be used by staff of national immunization programmes, regulatory authorities and pharmacovigilance or surveillance departments.5 After 7 years, several limitations had been identified during its use in the field, including: the need for more detailed guidance on the elements required to perform the assessment of causality, confusion over the terms used to classify the likelihood of association of the event to the vaccine, and the incomplete use of parameters for establishing causal association.

Following the GACVS decision to review the causality assessment system in December 2010, a working group was established to review the aide-mémoire and develop a method that would be simple, objective, adaptable and evidence-based when used by countries with different resources and capabilities. After concluding a thorough review of the most innovative methods available for determining causation for drugs and biologicals, an algorithmic scheme that incorporates additional elements of causation was designed. The guide was harmonized after the Clinical Immunization Safety Assessment (CISA) network’s newly developed algorithm which is available in the USA6 and the new definition of AEFI proposed by the Council for International Organizations of Medical Sciences (CIOMS).7

The new WHO proposed method allows the National Committees for AEFI case review and causality assessment to screen serious cases reported by their surveillance system for completeness and quality of information, ensuring the objectiveness of the assessment. Cases deemed incomplete are directed towards additional case investigation and review. A checklist containing the elements of causality assessment was included to guide the committee or the assessor to gather the evidence needed for case review, and when completed allows the application of an algorithm that helps determine if the AEFI could be consistent or inconsistent with an association with the immunization, or is deemed indeterminate due to lack of evidence. A repository of all AEFI cases sorted through this new document is considered critical and recommended to allow for future signal detection and determining the need for additional epidemiological studies.

GACVS recognizes the boundaries of the newly developed method, mainly the limitations in the ability to associate novel, previously unknown AEFIs potentially associated with immunizations, and restrictions due to insufficient information available for individual cases. However, the new AEFI causality assessment system will provide a standardized and transparent method that allows stakeholders to understand the nature of the decision-making process, and pave the way for future evaluation of the guide to refine its effectiveness. GACVS has recommended that this new WHO AEFI causality assessment approach should be made public as soon as it is finalized, and that complementary materials and simple software be developed for use in countries to enable immunization staff to field-test the algorithm. Of the next steps deemed most important is the development of a booklet to codify the algorithm and train countries in its use. The committee encouraged the subgroup to further develop the product and endorsed the work process.

Core variables for AEFI monitoring

Collection of harmonized data on AEFI allows for better comparison and pooled analysis with findings from vaccine safety surveillance systems. In collaboration with a network of countries and independent experts, a preliminary list of core variables had been proposed. This list was subsequently compared with the reporting forms from the WHO Programme for International Drug Monitoring (Uppsala Monitoring Centre) to verify which variables are captured by the current reporting forms. Through this exercise, it became apparent that vaccine safety monitoring needs tools which are more specific to the type of variables required for proper AEFI surveillance and that the current web-based interface developed for reporting of suspected drug reactions (VigiFlow) should be adapted for AEFI reporting. To address these issues, the GACVS in December 2011 suggested developing a simpler and vaccine-specific user interface to enter AEFI data. A subgroup of GACVS was tasked to address those issues and presented the status of ongoing activities at the June 2012 meeting.

Collection of basic and advanced AEFI information

It is recognized that for the purpose of signal detection, data collection tools should remain as simple as possible. However, when signals are detected, or in cases of serious AEFI, additional data are essential to allow inferences to be drawn on the association with vaccines and to assess the need for further investigation and action. The subcommittee presented GACVS with 22 core variables that should be collected for any AEFI (basic information) and an additional 33 variables of interest for a more detailed case review (advanced information). Basic information collected needs to be prioritized because the AEFI data collection, collation, transmission, analysis and feedback systems in different countries are heterogeneous. In addition, quantitative and qualitative aspects of data need to be considered. The suggested approach proposes a basic minimum of 22 variables with 10 identified as critical. This simple structure is expected to encourage countries that do not yet have an AEFI surveillance system in place to develop one. It is proposed that the reporting tool include the WHO-ART dictionary in order to standardize the terminology used to record signs, symptoms or a diagnosis, as well as a vaccine dictionary that will include details pertaining to all of the vaccines suspected. For the advanced information, details on the nature and frequency of reporting for events such as in campaigns or in routine immunization programmes, breast or bottle feeding, status of previous vaccination are proposed.


“VacciFlow” will be developed as the adaptation of drug-specific VigiFlow 4.2 to facilitate the entry of vaccine-related AEFI data including immunization programme errors. Ideally “VacciFlow” will be used by both the national regulatory authority and the immunization programme staff. The possibility of incorporating this new interface with minimal computer capabilities and mobile phone technology was encouraged by GACVS. There will be 3 flexible levels created in “VacciFlow” enabling national and subnational level users to analyse and use the data available for action at each level. Automatic feedback to reporters on the status of the report will be built in. Adapting (modifying) existing AEFI reporting systems to adjust to the data proposed in this core set of variables will require an educational and dissemination effort in many countries. It is expected that the upcoming “VacciFlow” will be sufficiently simple and user-friendly to allow tailor-made adjustment for locally collected information.

  • See No. 41, 1999, pp. 337–338.
  • GACVS invited additional experts to present and discuss evidence related to particular topics. These experts included persons affiliated with: Bambino Gesù Hospital, Rome, Italy; Center for Biologics Evaluation and Research (U.S. F.D.A), Rockville MD, USA; John Hopkins Bloomberg School of Public Health, Baltimore MD, USA; Program for Applied Technologies in Health, Seattle, USA; Rochester General Hospital Research Institute, Rochester NY, USA; Shantha Biotechnics Limited, Hyderabad, India; University of California, Los Angeles CA, USA; University of Washington, Seattle WA, USA; Uppsala Monitoring Centre, Uppsala, Sweden.
  • Tomljenovic L, Shaw CA. Do aluminum vaccine adjuvants contribute to the rising prevalence of autism? Journal of Inorganic Biochemistry, 2011; 105: 1489–1499.
  • Tomljenovic L, Shaw CA. Aluminum vaccine adjuvants: are they safe? Current Medicinal Chemistry, 2011; 18(17):2630–2637.
  • Aide-mémoire: Adverse events following immunization (AEFI): causality assessment. Geneva, World Health Organization, 2005. Available from; accessed July 2012.
  • Halsey NA et al. Algorithm to assess causality after individual adverse events following immunizations. Vaccine, 2012. Available online at, accessed July 2012.
  • Definitions and application of terms for vaccine pharmacovigilance. Geneva, World Health Organization/ Council for International Organizations of Medical Sciences, 2012. Available at, accessed July 2012.