A loss of confidence in vaccine safety, due to rumours or coincidental adverse events after vaccination, is a major and ongoing issue that has resulted in many documented instances of under-vaccination potentially putting large numbers of people at risk of vaccine-preventable disease and death, as well as requiring costly outbreak response activities.^1-4 Stringent regulatory standards to ensure the safety and quality of vaccines are essential to maintain public confidence in vaccines. Swift investigation, response and communication with the public are alsoimportant to prevent a loss in public confidence following reports of adverse events following immunization (AEFI). Hepatitis B vaccination at birth is particularly prone to association with coincidental deaths because infant mortality is highest during the early neonatal period.⁵

Diverse methods have been used to investigate the impact of reported AEFIs, including modelling the impact on disease,⁴ in-depth qualitative studies⁶ and monitoring vaccination coverage. The comprehensive approach taken by Yu et al.⁷ (and presented in this edition of the journal) warrants particular consideration in investigating the impact of similar incidents in the future.

The study from China by Yu et al.⁷ compared changes over time in perceptions of vaccine safety, related to a specific incident related to neonatal deaths temporally associated with hepatitis B vaccination at birth, with trends in hepatitis B vaccine coverage and doses administered. Data for the analysis came from a series of health care provider and parent surveys as well as administrative vaccination data. Collection of survey data and effective use of administrative data in investigations of reported AEFIs can help monitor impact of events and inform direct responses to regain public confidence as quickly as possible.

Another potential method could have been a longitudinal study of a cohort of caregivers to gauge the impact of an AEFI event on caregiver attitudes towards vaccines over time. This approach has the benefit of increasing the comparability of results over time and reducing the impact of confounding compared with cross-sectional studies, but might be constrained by bias caused by repeat interviews of participants or characteristics of people who participate in longitudinal studies.

Yu and colleagues⁷ present aggregate results from several provinces, but analysis of trends in vaccination coverage and public vaccine confidence at the subnational level or by population group might also be useful in particular contexts. The effects of individual factors, such as socioeconomic status and caregivers’ education level, on vaccine hesitancy are highly context-specific. For example, outbreaks of measles due to non-vaccination have occurred among highly educated upper-income groupsin the USA,² and targeted surveys of attitudes towards vaccines in these groups might be warranted. In low-income countries such as Burkina Faso, vaccine knowledge and higher socioeconomic status might reduce vaccine hesitancy and increase vaccination coverage.⁸ Although the link between individual factors and vaccine confidence was not the focus of the Yu et al.⁷ study, such analysis could potentially identify factors associated with vaccine hesitancy to help target appropriate messages following widely publicized events temporally associated with immunizations.

Interestingly, this study found that the internet was the most common source of information regarding the AEFI event described. Other sources for monitoring vaccine confidence include electronic news and social media coverage and qualitative analyses of blog posts in response to news articles.⁹ These sources can be reviewed retrospectively and may be less resource-intensive than primary data collection.

The first survey in the study by Yu et al.⁷ was conducted in December, the same month that the AEFI event was reported. The logistics of preparing a survey methodology, obtaining ethics approval and recruiting, training and deploying data collectors within such a short time frame could prove to be challenging in other settings without capacity to organize and implement a quality survey in a very short time frame.

In addition to the careful research conducted by Yu et al.⁷, the fortuitous timing of a previously planned vaccine confidence study provided an important baseline of parental confidence. Periodic cross-sectional surveys could be used as an additional tool to monitor trends in vaccine confidence and respond early to vaccine hesitancy. The population surveyed could be chosen for timeliness or representativeness of the affected population as needed.

Overall, this study by Yu et al.⁷ has greatly contributed to the knowledge of effective methods of assessing the impact of AEFIs on vaccination coverage and vaccine hesitancy. Continued innovation in this area of research will be helpful in strengthening the investigation and response to rumours and reported AEFIs.