ArticlePDF Available

Abstract and Figures

While previous studies have validated vaccine hesitancy scales with uptake behavior at the individual level, the conditions under which aggregated survey data are useful are less clear. We show that vaccine public opinion data aggregated at the subnational level can serve as a valid indicator of aggregate vaccine behaviour. We use a public opinion survey (Eurobarometer EB 91.2) with data on vaccine hesitancy for the EU in 2019. We link this information to (subnational) regional immunization coverage rates for childhood vaccines – DTP3, MCV1, and MCV2 -- obtained from the WHO for 2019. We conduct multilevel regression analyses with data for 177 regions in 20 countries. Given the variation in vaccine hesitancy and immunization rates between countries and within countries, we affirm the valuable role that surveys can play as a public health surveillance tool when it comes to vaccine behavior. We find statistically significantly lower regional vaccine immunization rates in regions where vaccine hesitancy is more pronounced. Our results suggest that different uptake rates across subnational regions are due, at least in part, to differences in attitudes towards vaccines and vaccination. The results are robust to several alternative specifications.
Content may be subject to copyright.
Association of vaccine hesitancy and immunization coverage rates
in the European Union
Florian Stoeckel, University of Exeter
Charlie Carter, London School of Economics
Ben Lyons, University of Utah
Jason Reifler, University of Exeter
Key words:
Vaccine hesitancy, immunization coverage, European Union,
vaccination coverage, public health policy, surveys
While previous studies have validated vaccine hesitancy scales with uptake behavior at the
individual level, the conditions under which aggregated survey data are useful are less clear.
We show that vaccine public opinion data aggregated at the subnational level can serve as a
valid indicator of aggregate vaccine behaviour. We use a public opinion survey (Eurobarometer
EB 91.2 ) with data on vaccine hesitancy for the EU in 2019. We link this information to
(subnational) regional immunization coverage rates for childhood vaccines DTP3, MCV1,
and MCV2 -- obtained from the WHO for 2019. We conduct multilevel regression analyses
with data for 177 regions in 20 countries. Given the variation in vaccine hesitancy and
immunization rates between countries and within countries, we affirm the valuable role that
surveys can play as a public health surveillance tool when it comes to vaccine behavior. We
find statistically significantly lower regional vaccine immunization rates in regions where
vaccine hesitancy is more pronounced. Our results suggest that different uptake rates across
subnational regions are due, at least in part, to differences in attitudes towards vaccines and
vaccination. The results are robust to several checks.
Forthcoming Vaccine 39 (2021) pp. 3935-3939
https://doi.org/10.1016/j.vaccine.2021.05.062
Acknowledgements:
Florian Stoeckel reports funding from the British Academy (SRG20\200348). Jason Reifler
received funding from the European Research Council (ERC) under the European Union’s
Horizon 2020 research and innovation program (grant agreement No 682758).
Corresponding author:
Dr Florian Stoeckel
University of Exeter
f.stoeckel@exeter.ac.uk
!
1
Background
Scholars have given increasing attention to “vaccine hesitancy” (Larson et al., 2014).
Owing to the potential dangers of vaccine hesitancy, the Strategic Advisory Group of Experts
[SAGE] on Immunization established a Working Group dealing with vaccine hesitancy in
March 2012 (SAGE, 2012), and due to hesitancy’s continued acceleration, the Lancet
recently announced a Commission on Vaccine Refusal, Acceptance, and Demand in the USA
(Hotez et al., 2021). Concerns about vaccines may then translate into refusal or delay of some
vaccines, total refusal, or even total acceptance despite concern. Previous research has found
that there is a link between aggregate levels of vaccine hesitancy and vaccine uptake in
different contexts and for different vaccines. We extend this line of inquiry to subnational
levels in the European context.
A number of systematic reviews have summarized the research on determinants of
and potential solutions to vaccine hesitancy (e.g., Larson et al., 2014; Jarret et al., 2015; Dube
et al., 2015; Salmon et al., 2015). Further, a number of scales have been developed to
measure hesitancy in surveys. These include the Vaccine Confidence Scale (VCS) (Gilkey et
al., 2014); Parental Attitudes about Childhood Vaccines (PACV) (Opel et al., 2011); Parental
Perspectives Regarding Vaccines Scale (Freed et al., 2010; Nyhan et al. 2014); the
Vaccination Attitudes Examination (VAX) Scale (Martin et al., 2017); the Vaccine Attitude
Scale (Horne et al., 2015); and the SAGE Working Group’s Vaccine Hesitancy Scale (Larson
et al., 2015), among others.
Validation at individual level
Do vaccine hesitancy scales measure an attitude that is related to behaviour? At the
individual level, this appears to be the case. Opel et al. (2013) validated the PACV among
437 new parents in Seattle, WA in 2010, examining children’s immunization status using
!
2
GHC electronic immunization records cross-referenced with the state’s central immunization
registry. They concluded that scores on the PACV predict childhood immunization status and
have high reliability. This group also validated the PACV in Seattle in 2013 (Strelitz et al.,
2015) using medical records, focusing on influenza vaccine refusal among 152 parents,
finding higher hesitancy scores associated with refusal. Roberts et al. (2015) used medical
records to validate a version of the PACV modified for the adolescent setting, surveying 363
parents in Oklahoma and South Carolina. They found that several individual items were
associated with vaccine uptake, but the modified PACV scale itself was not associated with
vaccination status. Finally, Gilkey et al. (2016) validated the VCS among a population-based
sample of 9,354 U.S. parents who completed the 2011 National Immunization Survey
household survey. This survey included self-reported vaccination status that was
subsequently verified by healthcare providers in the 2011 NIS provider survey using medical
records. Vaccination confidence was consistently associated with early childhood vaccination
behavior across multiple vaccine types. Overall then, validation studies conducted at the
individual level suggest hesitancy and vaccination behaviour are likely related, though not
every study finds this is the case (e.g., Roberts et al., 2015).
Use at aggregate levels
Hesitancy has also been included on cross-national surveys such as the
Eurobarometer, and may be used in essence as a “surveillance” or “monitoring” system (de
Figueiredo et al., 2020) for detecting national-level trends regarding vaccination (e.g., in the
EU see Larson et al., 2018; see also Larson et al., 2016; Lunz, Trujillo & Motta, 2020).
Santibanez et al. (2020) have validated this approach at the subnational level in the US with
their finding that states with higher state-level estimates of parental vaccine hesitancy have
lower state-level child influenza vaccination coverage. We extend this approach to the
!
3
European context by aggregating individual level data from national surveys to produce sub-
national estimates of vaccine hesitancy and examine their association with vaccine uptake at
the same sub-national levels. Vaccine confidence is particularly low in Europe compared to
other continents (de Figueiredo et al., 2020; Larson, 2018; Larson et al., 2018), and there is
considerable heterogeneity within Europe, with confidence declining in Czech Republic,
Finland, Poland, and Sweden but rising in France, Greece, Italy, and Slovenia since 2015
(Larson et al., 2018).
Data and Method
Study design and population
We conduct an ecological analysis that links aggregated survey data on vaccine
hesitancy to subnational vaccine uptake rates. Data on immunization coverage rates comes
from the WHO and is publicly available on request at
https://www.who.int/immunization/monitoring_surveillance/data/subnational/en/. We
obtained subnational immunization coverage data for DTP3, MCV1, and MCV2 for the year
of our survey data (2019). A robustness check with uptake data for 2018 can be found in the
appendix (Tables A7 and A8), but hesitancy data is only available in the Eurobarometer EB
91.2 from 2019 (which is why we focus on 2019 uptake data). DTP3 is the third dose of the
diphtheria, tetanus toxoids, and pertussis–containing vaccine. MCV1 is the first dose of the
measles-containing vaccine, MCV2 is the second dose. We focus on MCV1, MCV2, and
DTP3 because they are widely used in related vaccine coverage validation and vaccine
hesitancy studies (e.g., Anand et al., 2007; Arsenault et al., 2017; Bechini et al., 2019;
Murray et al., 2003) and they are also among the vaccines for which the WHO possesses the
most complete data. Subnational divisions in the data are based on the EU’s nomenclature of
territorial units for statistics (NUTS), which is used to subdivide member states’ territory
!
4
from larger (NUTS 1) to smaller (NUTS 3) territorial units. Not all countries have every level
of division, depending on their size. In Ireland, the WHO data does not correspond to NUTS
regions but instead to Local Health Office (LHO) areas. We obtained information from
Health Service Executive Ireland to align the data. Subnational coverage rates for these
vaccines serve as our outcome variables.
Data on vaccine hesitancy comes from the Eurobarometer survey of Spring 2019 (EB
91.2). The survey was conducted face to face with probability samples from each EU
member state in March 2019. It includes about 1,000 respondents from each country except
for Luxembourg, Cyprus, and Malta, where about 500 individuals were interviewed. This
yields a total sample size of 27,524 respondents. Our analysis is carried out at the lowest
regional level for which we have data for our outcome variable vaccine uptake, albeit the data
is available at a lower NUTS level in some countries than in others. While it would be
preferable for all data to be at an equally low level (e.g. NUTS3), we pool different NUTS
levels to retain as many within country observations as possible (for similar approaches, see
e.g. Mohl and Hagen, 2010). Results for a robustness check in which we aggregate data up to
the same level (NUTS2) can be found in the appendix. Our analysis of the relationship
between hesitancy and DTP3 uptake includes 176 regions from 20 countries, it includes 177
regions in 20 countries for MCV1 and 142 regions in 15 countries for MCV2.
Scale description and measures
Vaccine coverage rates are calculated by dividing the number of vaccine doses that
were administered in a district by the target population of the district (i.e., the number of
children in case of MCV1, MCV2, and DTP3). While the literature agrees on the importance
of measuring vaccine hesitancy, the specific items to do so differ widely. The Eurobarometer
survey does not rely on items from the scales reviewed above, but does include a range of
!
5
questions that allows us to build an index with five items, following suggestions from the
literature for multiple-item scales (Martin & Petrie 2017, Horne et al., 2015, Opel et al.,
2011). Each of five vaccine hesitancy items offers four response options on Likert style
response scales, where the least vaccine hesitant answer is coded as 0 and the most vaccine
hesitant answer is coded as 3. The vaccine hesitancy score for each respondent is their
average across the five items. To calculate the regional vaccine hesitancy score, we take the
average of the individual level vaccine hesitancy score for all respondents in a region. The
results we present are restricted to parents, as is typical for studies of childhood
immunization, e.g., Santibanez et al. (2020). (Parents are defined as individuals who live
together with their children in one household.) Additional analyses in the appendix include
analyses based on the full sample rather than just parents. We drop respondents who did not
select one of the four substantive response categories. We apply the weights provided by the
Eurobarometer when aggregating scores to produce regional hesitancy measures.
The MCV1, MCV2, and DTP3 vaccinations are mandatory in some of the countries in
our sample (see Table 2). We use a control variable to account for this difference in all
models. Results for models without this control and just for countries without mandate are in
the appendix (see appendix Tables A2 and A8).
< Table 1: Question wording >
Statistical Analyses
We use multilevel models to analyse the relationship between hesitancy and vaccine uptake
(with a linear link function) because they account for the particular structure of our data:
subnational observations being nested within countries. These models are also well suited to
a situation in which variation is between countries and within countries. In our case, half of
!
6
the variation in vaccine uptake is between countries while the remainder of the variation
results from variation within countries.
Regional hesitancy scores are based on aggregations from samples with an unequal
number of observations. We do several robustness checks that are common in the literature to
take into account that this might affect the results. The first approach is based on the fact that
we have more confidence in regional hesitancy scores derived from large samples than in
those coming from small samples. To account for this we follow previous research (Carle,
2009, Charron et al., 2016) by weighting observations by the sample-size based uncertainty
around regional hesitancy estimates. (We include the inverse of the standard errors of
regional hesitancy means as weights in the multilevel models.) Our second robustness check
corrects for the chance that the composition of the regional samples might differ from the
actual population of a region. (This problem should be expected as samples that are
nationally representative may not be perfectly representative at subnational levels.)
Multilevel regression with post-stratification (MRP) is a common tool for this correction
(Downs & Carlin, 2019, Loux et al., 2019). We collected information on the age and gender
composition of the regions in our analysis and use MRP to re-estimate regional vaccine
hesitancy that takes this composition into account. In a third robustness check (appendix
Tables A3), we exclude all regions in which vaccine hesitancy scores were calculated based
on fewer than 25 respondents. (To account for the loss of statistical power from removing
regions for this analysis, we pool the three vaccines into a single model and account for the
resulting clustering.) The results of these robustness checks reinforce our findings.
Results
Vaccine uptake differs both between countries and within countries (see Table 2). For
instance, average country level DTP3 uptake varies between 86.26 percent (in Romania) and
!
7
99.87 percent in Hungary. Country averages conceal a considerable amount of variation
within countries. While MCV1 uptake is relatively similar across regions in Hungary (which
must be the case given the extremely high reported uptake rate), there is much variation in
other countries (e.g., uptake in Croatia ranges from 73.24 percent to 98.38 percent).
We also find that vaccine hesitancy varies considerably across countries and sub-
nationally (see the appendix for details). Average country level vaccine hesitancy is lowest in
Denmark and highest in Latvia. In fact, the least vaccine hesitant sub-national region in
Latvia is more vaccinate hesitant than the most vaccine-hesitant region in Denmark.
< Table 2: Descriptive data >
We find a relationship between vaccine hesitancy scores and uptake rates of DTP3, MCV1,
and MCV2 , though the results differ between specifications (see Figure 1 and Table 1 in the
appendix). We find vaccine hesitancy to be associated with DTP3 uptake (95% CI -3.658, -
0.035) , MCV1 uptake (95% CI -5.495, -0.779) , and MCV2 uptake (95% CI -5.706, -0.264).
When taking uncertainty around regional estimates into account, we still find hesitancy to be
related with DTP3 uptake (90% CI -3.139, -0.100) , MCV1 uptake (95% CI -4.933, -0.186) ,
and MCV2 uptake (95% CI -6.069, -0.520). The results hold when hesitancy scores were
calculated using MRP (DTP3: 95% CI -6.714, -0.074 ; MCV1: 95% CI -9.528, -1.169 ;
MCV2: 95% CI -10.832, -0.282 ).
< Figure 1: Regression results for the association of vaccine hesitancy and vaccine uptake >
We probe the nature of the relationship in several additional ways. First, our main model
pools countries with and without vaccination mandates as there is variation in vaccine uptake
!
8
even where mandates exist. The results also hold for DTP3 and MCV1 when estimated just
with the sample of countries that have no mandate in place (appendix Table A8; note that this
leads to a much smaller number of observations). Second, our main model pools observations
from different NUTS levels in order for us to use the fine-grained data that is available and to
avoid losing precision from aggregating up lower level data. We also conduct a robustness
check in which we aggregate up lower level data in order for the analysis to be based on the
same level (NUTS2). The effects are consistent with the ones we find in the main model, but
the number of observations is much smaller and standard errors are larger (hence fewer
effects are statistically significant; appendix Table A4). Finally, we find that the results hold
when we include all respondents rather than just parents (appendix Table A6). This suggests
that in some cases, even survey data that lacks measures of parenthood may be useful in
monitoring hesitancy at the regional level.
Discussion
Our findings have several implications. First, public immunization rates are related to
observed levels of vaccine hesitancy. Uptake rates are lower in regions where hesitancy is
more pronounced. The upshot of this finding is that as a surveillance tool, public opinion
surveys can be used to understand where vaccines are more likely to be rejected and who
should be the target of information campaigns. Our findings complement those showing
negative associations of subnational estimates of parental vaccine hesitancy with child
vaccination coverage in the U.S (Santibanez et al., 2020). We extend the scope of this finding
in several ways, including geographic (cross-nationally in the E.U.), the vaccinations of
concern (DTP3, MCV1, and MCV2), and the potential sample of interest (showing
association both among parents and within the general population). By looking at the
association between vaccine hesitancy and vaccine uptake at the regional level, it may be
!
9
possible to identify areas that perform extremely well in terms of vaccination relative to
hesitancy levels (perhaps drawing lessons of best practice that could be applied elsewhere) as
well as identifying areas that are underperforming (where adopting best practices or using
other interventions may be especially likely to be successful).
A second implication of our findings is that we do not presently have ideal data on
vaccine coverage rates and vaccine hesitancy, even though such data seems valuable to
monitor public health issues. For instance, the WHO does not have coverage rates for six EU
member states for the vaccines that are included in our analysis. Publicly available influenza
vaccine uptake data is even more incomplete: at present, uptake data is only available for 19
countries in 2019, and regional data is unavailable. Regarding vaccine hesitancy, it is
valuable that the Eurobarometer 91.2 included such questions, as it is rare that a large-scale
cross-country dataset does so. Such efforts could be improved by including question items
from the most commonly used hesitancy scales, which would also foster the comparability
across studies. Finally, an important share of variation in vaccine hesitancy is at the regional
level within countries, rather than simply at the cross-national level. Consequently, we
encourage larger and more representative samples at the regional level. (Representative here
may mean of the general population or appropriate target populations, such as parents.) We
recognize that resources for surveillance surveys may be finite; while we accept the tradeoff
for larger and more representative samples at the regional level, others may be less willing to
do so. Along these lines, we encourage future work to consider the extent to which vaccine
hesitancy measured in surveys is better as a leading or trailing indicator of vaccination
uptake, and at what time lag.
The WHO is careful to list the limitations of the vaccine uptake data and these
limitations apply to our analysis of this data
(https://www.who.int/immunization/monitoring_surveillance/data/limitations.pdf?ua=1). One
!
10
of the limitations is that uptake data can be imprecise and uptake rates can be above one
hundred percent if more children get vaccinated in a region than are registered there (which
sets the target): this might occur for instance when many more children get vaccinated in a
region where they are not registered.
We use a survey that is designed to represent populations at the national level, but we
use it to gauge regional vaccine hesitancy. While we probe the robustness of this procedure, it
is less accurate than a survey which is devised for this purpose, with large and equally sized
regional samples that are designed to be representative of regional populations.
Conclusion
We found that across the E.U., greater rates of vaccine hesitancy as measured by the
Eurobarometer were associated with lower uptake of DTP3, MCV1, and MCV2 vaccines.
This finding suggests vaccine hesitancy as measured at aggregated levels can be useful for
monitoring potentially problematic regions and guiding interventions. Strategies to improve
large national and cross-national public opinion surveys would further improve the usefulness
of vaccine hesitancy as a monitoring tool for health professionals.
!
11
References
Anand, S., & Bärnighausen, T. (2007). Health workers and vaccination coverage in
developing countries: an econometric analysis. The Lancet, 369(9569), 1277-1285.
Arsenault, C., Harper, S., Nandi, A., Rodríguez, J. M. M., Hansen, P. M., & Johri, M. (2017).
Monitoring equity in vaccination coverage: a systematic analysis of demographic and
health surveys from 45 Gavi-supported countries. Vaccine, 35(6), 951-959.
Bechini, A., Boccalini, S., Ninci, A., Zanobini, P., Sartor, G., Bonaccorsi, G., ... & Bonanni,
P. (2019). Childhood vaccination coverage in Europe: impact of different public health
policies. Expert review of vaccines, 18(7), 693-701.
Brown, C., Clayton-Boswell, H., Chaves, S. S., Prill, M. M., Iwane, M. K., Szilagyi, P. G., ...
& Hall, C. B. (2011). Validity of parental report of influenza vaccination in young
children seeking medical care. Vaccine, 29(51), 9488-9492.
Carle, A.C., (2009). Fitting multilevel models in complex survey data with design weights:
Recommendations. BMC Medical Research Methodology, 9(1), 49.
Charron, N., Dahlström, C., Fazekas, M. and Lapuente, V., (2017). Careers, Connections, and
Corruption Risks: Investigating the impact of bureaucratic meritocracy on public
procurement processes. The Journal of Politics, 79(1), 89-104.
de Figueiredo, A., Simas, C., Karafillakis, E., Paterson, P., & Larson, H. J. (2020). Mapping
global trends in vaccine confidence and investigating barriers to vaccine uptake: a
large-scale retrospective temporal modelling study. The Lancet, 396(10255), 898-908.
Downes, M., & Carlin, J. B. (2020). Multilevel regression and poststratification as a
modeling approach for estimating population quantities in large population health
studies: A simulation study. Biometrical Journal, 62(2), 479-491.
!
12
Dubé, E., Gagnon, D., & MacDonald, N. E. (2015). Strategies intended to address vaccine
hesitancy: Review of published reviews. Vaccine, 33(34), 4191-4203.
Freed, G. L., Clark, S. J., Butchart, A. T., Singer, D. C., & Davis, M. M. (2010). Parental
vaccine safety concerns in 2009. Pediatrics,125(4), 654–659.
Gilkey, M. B., Magnus, B. E., Reiter, P. L., McRee, A. L., Dempsey, A. F., & Brewer, N. T.
(2014). The Vaccination Confidence Scale: a brief measure of parents’ vaccination
beliefs. Vaccine, 32(47), 6259-6265.
Gilkey, M. B., McRee, A. L., Magnus, B. E., Reiter, P. L., Dempsey, A. F., & Brewer, N. T.
(2016). Vaccination confidence and parental refusal/delay of early childhood vaccines.
PloS one, 11(7), e0159087.
Hotez, P. J., Cooney, R. E., Benjamin, R. M., Brewer, N. T., Buttenheim, A. M., Callaghan,
T., ... & Omer, S. B. (2021). Announcing the Lancet Commission on Vaccine Refusal,
Acceptance, and Demand in the USA. The Lancet, 397(10280), 1165-1167.
Horne, Z., Powell, D., Hummel, J. E., & Holyoak, K. J. (2015). Countering antivaccination
attitudes. Proceedings of the National Academy of Sciences, 112(33), 10321-10324.
Jarrett, C., Wilson, R., O’Leary, M., Eckersberger, E., & Larson, H. J. (2015). Strategies for
addressing vaccine hesitancy–A systematic review. Vaccine, 33(34), 4180-4190.
Larson, H. J. (2018). The state of vaccine confidence. The Lancet, 392(10161), 2244-2246.
Larson, H. J. et al. (2018) State of Vaccine Confidence in the EU 2018, Luxembourg:
Publications Office of the European Union. doi: 10.2875/241099.
Larson, H. J., Jarrett, C., Eckersberger, E., Smith, D. M., & Paterson, P. (2014).
Understanding vaccine hesitancy around vaccines and vaccination from a global
perspective: a systematic review of published literature, 2007–2012. Vaccine, 32(19),
2150-2159.
!
13
Larson, H. J., Jarrett, C., Schulz, W. S., Chaudhuri, M., Zhou, Y., Dube, E., ... & Wilson, R.
(2015). Measuring vaccine hesitancy: the development of a survey tool. Vaccine,
33(34), 4165-4175.
Larson, H. J., De Figueiredo, A., Xiahong, Z., Schulz, W. S., Verger, P., Johnston, I. G., ... &
Jones, N. S. (2016). The state of vaccine confidence 2016: global insights through a 67-
country survey. EBioMedicine, 12, 295-301.
Larson, H., de Figueiredo, A., Karafillakis, E., & Rawal, M. (2018). State of vaccine
confidence in the EU 2018. Luxembourg: Publications Office of the European Union,
10, 241099.
Loux, T., Nelson, E. J., Arnold, L. D., Shacham, E., & Schootman, M. (2019). Using
multilevel regression with poststratification to obtain regional health estimates from a
Facebook-recruited sample. Annals of Epidemiology, 39, 15-20.
Lunz Trujillo, K, and Motta, M. (2020). Why Are Wealthier Countries More Vaccine
Skeptical?: How Internet Access Drives Global Vaccine Skepticism. APSA Preprints.
doi: 10.33774/apsa-2020-bbpld-v2.
Martin, L. R., & Petrie, K. J. (2017). Understanding the dimensions of anti-vaccination
attitudes: The vaccination attitudes examination (VAX) scale. Annals of Behavioral
Medicine, 51(5), 652-660.
Mohl, P., & Hagen, T. (2010). Do EU structural funds promote regional growth? New
evidence from various panel data approaches. Regional Science and Urban Economics,
40(5), 353-365.
Murray, C. J. L., Shengella, B., Gupta, N., Moussavi, S., Tandon, A., & Thieren, M. (2003).
Validity of reported vaccine coverage in 45 countries. Health systems performance
assessment. Debates, methods and empiricism, 265-271.
!
14
Nyhan, B., Reifler, J., Richey, S., & Freed, G. L. (2014). Effective messages in vaccine
promotion: A randomized trial. Pediatrics, 133(4), e835–e842.
Opel, D. J., Mangione-Smith, R., Taylor, J. A., Korfiatis, C., Wiese, C., Catz, S., & Martin,
D. P. (2011). Development of a survey to identify vaccine-hesitant parents: the parent
attitudes about childhood vaccines survey. Human vaccines, 7(4), 419-425.
Opel, D. J., Taylor, J. A., Zhou, C., Catz, S., Myaing, M., & Mangione-Smith, R. (2013). The
relationship between parent attitudes about childhood vaccines survey scores and future
child immunization status: a validation study. JAMA pediatrics, 167(11), 1065-1071.
Roberts, J. R., Thompson, D., Rogacki, B., Hale, J. J., Jacobson, R. M., Opel, D. J., &
Darden, P. M. (2015). Vaccine hesitancy among parents of adolescents and its
association with vaccine uptake. Vaccine, 33(14), 1748-1755.
SAGE (2012). SAGE Working Group on Vaccine Hesitancy
http://www.who.int/immunization/sage/sage_wg_vaccine_hesitancy_apr12/en/
Salmon, D. A., Dudley, M. Z., Glanz, J. M., & Omer, S. B. (2015). Vaccine hesitancy:
causes, consequences, and a call to action. Vaccine, 33, D66-D71.
Santibanez, T. A., Grohskopf, L. A., Zhai, Y., & Kahn, K. E. (2016). Complete influenza
vaccination trends for children six to twenty-three months. Pediatrics, 137(3).
Santibanez, T. A., Nguyen, K. H., Greby, S. M., Fisher, A., Scanlon, P., Bhatt, A., ... &
Singleton, J. A. (2020). Parental vaccine hesitancy and childhood influenza
vaccination. Pediatrics, 146(6).
Strelitz, B., Gritton, J., Klein, E. J., Bradford, M. C., Follmer, K., Zerr, D. M., ... & Opel, D.
J. (2015). Parental vaccine hesitancy and acceptance of seasonal influenza vaccine in
the pediatric emergency department. Vaccine, 33(15), 1802-1807.
Table 1: Question wording
min max mean SD
1
It is important for everybody to have routine vacci-
nations (0= totally agree, 1=tend to agree, 2=tend
to disagree, 3=totally disagree)
0 3 0.59 0.77
2
Not getting vaccinated can lead to serious health
issues (0= totally agree, 1=tend to agree, 2=tend to
disagree, 3=totally disagree)
0 3 0.65 0.79
3
Vaccines are important to protect not only yourself
but also others (0= totally agree, 1=tend to agree,
2=tend to disagree, 3=totally disagree)
0 3 0.50 0.70
4
Vaccination of other people is important to protect
those that cannot be vaccinated (e.g. newborn chil-
dren, immune depressed or very sick people) (0=
totally agree, 1=tend to agree, 2=tend to disagree,
3=totally disagree)
0 3 0.53 0.71
5
[List of deseases in previous question] All the diseases
mentioned earlier are infectious diseases and can be
prevented. Do you think that vaccines can be eec-
tive in preventing them? (0= Yes, definitely, 1=Yes,
probably, 2=No, probably not, 3=No, not at atll)
0 3 0.57 0.74
1
Table 2: Country mean and within country variation (range) of vaccine uptake by country for each vaccine. All figures are percentages (share of
children who were vaccinated in a region out of all children registered in a region), data for 2019 obtained from the WHO
Vaccine Hesitancy DTP3 MCV1 MCV2
Country Mean Range Mean Range Mandatory Mean Range Mandatory Mean Range Mandatory
BE 0.64 0.48-0.80 98.20 97.00-98.90 95.30 94.10-96.20 81.47 75.50-93.40
BG 0.63 0.48-0.76 93.14 85.10-95.80 394.84 90.48-98.53 392.79 87.59-96.19 3
CY 0.45 0.40-0.50 95.86 92.80-99.20 85.94 81.20-92.60
CZ 0.59 0.47-0.69 96.69 92.87-98.94 392.90 91.40-94.50 3—— 3
DE 0.47 0.19-0.85 93.29 88.60-96.80 97.47 95.20-98.60 93.29 89.80-96.00
DK 0.23 0.11-0.48 96.69 95.44-97.00 95.06 93.91-96.00 90.32 87.27-92.00
EE 0.57 0.44-0.67 91.67 90.07-95.50 88.21 84.66-91.90 89.46 88.10-91.81
ES 0.40 0.00-0.85 95.94 92.52-98.26 97.75 95.18-99.96 93.81 85.52-99.32
HR 0.73 0.19-1.38 95.04 84.91-98.86 393.34 73.24-98.38 395.24 84.52-98.97 3
HU 0.52 0.32-0.78 99.86 99.79-99.90 399.87 99.78-99.93 399.80 99.63-99.90 3
IE 0.48 0.25-0.59 93.88 91.48-96.52 91.41 88.44-95.44
IT 0.67 0.26-1.14 94.83 88.11-97.40 393.89 85.60-96.11 387.54 78.62-93.70 3
LT 0.64 0.51-0.88 93.01 89.70-96.70 94.21 90.90-96.70 94.90 90.40-98.30
LV 0.90 0.75-1.01 98.71 91.36-107.42 98.00 87.20-112.00 95.52 81.20-104.00
NL 0.24 0.10-0.40 93.44 87.62-95.35 93.37 87.56-95.28 90.19 84.79-95.02
PT 0.37 0.13-0.62 98.64 98.10-99.40 98.68 97.80-99.40 97.20 95.10-98.50
RO 0.84 0.55-1.13 87.67 83.50-92.02 88.89 79.82-92.26
SE 0.31 0.08-0.68 97.40 96.64-97.85 97.20 96.39-97.74
SK 0.67 0.35-1.14 96.75 96.52-97.15 395.80 94.81-96.66 397.63 96.92-98.17 3
UK 0.46 0.05-0.80 93.33 87.00-96.00 91.92 83.00-95.00 88.67 76.00-92.00
2
10 8642 0
DTP3
10 8642 0
MCV1
10 8642 0
MCV2
Model Type Multilevel Weighted Multilevel MRP
10 8642 0
DTP3
10 8642 0
MCV1
10 8642 0
MCV2
Model Type Multilevel Weighted Multilevel MRP
Figure 1: Regression results for the association of vaccine hesitancy and vaccine uptake. Plots show the
regression coecients for regional level vaccine hesitancy on regional level vaccine uptake. Values below
zero indicated a negative relationship between vaccine hesitancy and vaccine uptake; as vaccine hesitancy
increases in a region, vaccine uptake decreases. Each subfigure reports multiple model specifications to show
robustness. Thin lines: 95 percent CI, thick lines 90 percent CI. Models control for mandatory vaccination.
3
... Despite this, reasons for vaccine hesitancy may vary according to country, socio-economic factors, one's confidence towards the vaccine, and other factors [152,382]. Moreover, a lack of proper vaccination campaigning, ignorance of vaccination, and media communication have a substantial effects on people [148,205,[383][384][385]. ...
Article
Full-text available
To prevent the coronavirus disease 2019 (COVID-19) pandemic and aid restoration to pre-pandemic normality, global mass vaccination is urgently needed. Inducing herd immunity through mass vaccination has proven to be a highly effective strategy for preventing the spread of many infectious diseases, which protects the most vulnerable population groups that are unable to develop immunity, such as people with immunodeficiencies or weakened immune systems due to underlying medical or debilitating conditions. In achieving global outreach, the maintenance of the vaccine potency, transportation, and needle waste generation become major issues. Moreover, needle phobia and vaccine hesitancy act as hurdles to successful mass vaccination. The use of dissolv-able microneedles for COVID-19 vaccination could act as a major paradigm shift in attaining the desired goal to vaccinate billions in the shortest time possible. In addressing these points, we discuss the potential of the use of dissolvable microneedles for COVID-19 vaccination based on the current literature.
... Other work already establishes that people with higher vaccine hesitancy and lower risk tolerance may need to be targeted differently by vaccination campaigns than those with higher vaccine favorability and risk tolerance [17]. Furthermore, vaccine hesitancy and risk aversion are associated with real world vaccine behavior [18][19][20][21]. Therefore, we evaluate how individual differences in vaccine hesitancy and risk preferences condition the effect of vaccine attributes on vaccine acceptance. ...
Article
Full-text available
Why do people prefer one particular COVID-19 vaccine over another? We conducted a pre-registered conjoint experiment (n = 5,432) in France, Germany, and Sweden in which respondents rated the favorability of and chose between pairs of hypothetical COVID-19 vaccines. Differences in effectiveness and the prevalence of side-effects had the largest effects on vaccine preferences. Factors with smaller effects include country of origin (respondents are less favorable to vaccines of Chinese and Russian origin), and vaccine technology (respondents exhibited a small preference for hypothetical mRNA vaccines). The general public also exhibits sensitivity to additional factors (e.g. how expensive the vaccines are). Our data show that vaccine attributes are more important for vaccine preferences among those with higher vaccine favorability and higher risk tolerance. In our conjoint design, vaccine attributes–including effectiveness and side-effect prevalence–appear to have more muted effects among the most vaccine hesitant respondents. The prevalence of side-effects , effectiveness , country of origin and vaccine technology (e.g., mRNA vaccines) determine vaccine acceptance, but they matter little among the vaccine hesitant. Vaccine hesitant people do not find a vaccine more attractive even if it has the most favorable attributes. While the communication of vaccine attributes is important, it is unlikely to convince those who are most vaccine hesitant to get vaccinated.
... Despite evidence demonstrating the benefits of vaccination for individuals, populations, health systems and society in general, 1 vaccine hesitancy still remains a key factor influencing vaccine uptake, sometimes leading to outbreaks of vaccine-preventable diseases. [2][3][4] Factors influencing public confidence in vaccination have been extensively studied and summarised in literature reviews, [5][6][7][8][9][10] revealing the global scale of vaccine hesitancy and its impact on vaccination acceptance. Health-care professionals (HCPs) are some of the most trusted source of information around vaccination and therefore play an important role in motivating, reassuring and convincing individuals to get vaccinated. ...
Article
Full-text available
Health-care professionals (HCPs) have a fundamental role in vaccination, their own beliefs and attitudes affecting both their uptake and recommendation of vaccines. This literature review (n = 89) summarises evidence on HCPs' perceptions of the risks and benefits of vaccination, trust, and perceptions of mandatory vaccination in Europe. HCPs across studies believed that vaccination is important to protect themselves and their patients. However, beliefs that some diseases such as influenza are less risky were reported by some HCPs as a reason for not getting vaccinated. Concerns about both short- and long-term side effects were identified among HCPs in most studies, such as those affecting the immune or neurological system. Mistrust toward health authorities and pharmaceutical industry was reported in some studies. The question of mandatory vaccination revealed mixed opinions, with some favoring self-determination and others viewing vaccination as a duty. This review highlights key factors influencing HCPs' confidence in vaccination in Europe.
Preprint
Full-text available
Background Vaccination is key to successful prevention of COVID-19 particularly nosocomial acquired infection in health care workers (HCWs). 'Vaccine hesitancy' is common in the population and in HCWs, and like COVID-19 itself, hesitancy is more frequent in ethnic minority groups. UK-REACH (United Kingdom Research study into Ethnicity and COVID-19 outcomes) is a large-scale study of COVID-19 in UK HCWs from diverse ethnic backgrounds, which includes measures of vaccine hesitancy. The present study explores predictors of vaccine hesitancy using a 'phenomic approach', considering several hundred questionnaire-based measures. Methods UK-REACH includes a questionnaire study encompassing 12,431 HCWs who were recruited from December 2020 to March 2021 and completed a lengthy online questionnaire (785 raw items; 392 derived measures; 260 final measures). Ethnicity was classified using the Office for National Statistics' five (ONS5) and eighteen (ONS18) categories. Missing data were handled by multiple imputation. Variable selection used the islasso package in R, which provides standard errors so that results from imputations could be combined using Rubin's rules. The data were modelled using path analysis, so that predictors, and predictors of predictors could be assessed. Significance testing used the Bayesian approach of Kass and Raftery, a 'very strong' Bayes Factor of 150, N=12,431, and a Bonferroni correction giving a criterion of p<4.02 x 10-8 for the main regression, and p<3.11 x 10-10 for variables in the path analysis. Results At the first step of the phenomic analysis, six variables were direct predictors of greater vaccine hesitancy: Lower pro-vaccination attitudes; no flu vaccination in 2019-20; pregnancy; higher COVID-19 conspiracy beliefs; younger age; and lower optimism the roll-out of population vaccination. Overall 44 lower variables in total were direct or indirect predictors of hesitancy, with the remaining 215 variables in the phenomic analysis not independently predicting vaccine hesitancy. Key variables for predicting hesitancy were belief in conspiracy theories of COVID-19 infection, and a low belief in vaccines in general. Conspiracy beliefs had two main sets of influences: i) Higher Fatalism, which was influenced a) by high external and chance locus of control and higher need for closure, which in turn were associated with neuroticism, conscientiousness, extraversion and agreeableness; and b) by religion being important in everyday life, and being Muslim. ii) receiving information via social media, not having higher education, and perceiving greater risks to self, the latter being influenced by higher concerns about spreading COVID, greater exposure to COVID-19, and financial concerns. There were indirect effects of ethnicity, mediated by religion. Religion was more important for Pakistani and African HCWs, and less important for White and Chinese groups. Lower age had a direct effect on hesitancy, and age and female sex also had several indirect effects on hesitancy. Conclusions The phenomic approach, coupled with a path analysis revealed a complex network of social, cognitive, and behavioural influences on SARS-Cov-2 vaccine hesitancy from 44 measures, 6 direct and 38 indirect, with the remaining 215 measures not having direct or indirect effects on hesitancy. It is likely that issues of trust underpin many associations with hesitancy. Understanding such a network of influences may help in tailoring interventions to address vaccine concerns and facilitate uptake in more hesistant groups.
Article
Full-text available
Background: There is growing evidence of vaccine delays or refusals due to a lack of trust in the importance, safety, or effectiveness of vaccines, alongside persisting access issues. Although immunisation coverage is reported administratively across the world, no similarly robust monitoring system exists for vaccine confidence. In this study, vaccine confidence was mapped across 149 countries between 2015 and 2019. Methods: In this large-scale retrospective data-driven analysis, we examined global trends in vaccine confidence using data from 290 surveys done between September, 2015, and December, 2019, across 149 countries, and including 284 381 individuals. We used a Bayesian multinomial logit Gaussian process model to produce estimates of public perceptions towards the safety, importance, and effectiveness of vaccines. Associations between vaccine uptake and a large range of putative drivers of uptake, including vaccine confidence, socioeconomic status, and sources of trust, were determined using univariate Bayesian logistic regressions. Gibbs sampling was used for Bayesian model inference, with 95% Bayesian highest posterior density intervals used to capture uncertainty. Findings: Between November, 2015, and December, 2019, we estimate that confidence in the importance, safety, and effectiveness of vaccines fell in Afghanistan, Indonesia, Pakistan, the Philippines, and South Korea. We found significant increases in respondents strongly disagreeing that vaccines are safe between 2015 and 2019 in six countries: Afghanistan, Azerbaijan, Indonesia, Nigeria, Pakistan, and Serbia. We find signs that confidence has improved between 2018 and 2019 in some EU member states, including Finland, France, Ireland, and Italy, with recent losses detected in Poland. Confidence in the importance of vaccines (rather than in their safety or effectiveness) had the strongest univariate association with vaccine uptake compared with other determinants considered. When a link was found between individuals' religious beliefs and uptake, findings indicated that minority religious groups tended to have lower probabilities of uptake. Interpretation: To our knowledge, this is the largest study of global vaccine confidence to date, allowing for cross-country comparisons and changes over time. Our findings highlight the importance of regular monitoring to detect emerging trends to prompt interventions to build and sustain vaccine confidence. Funding: European Commission, Wellcome, and Engineering and Physical Sciences Research Council.
Preprint
Full-text available
Vaccine skepticism interferes with governments’ abilities to maintain public safety. However, vaccine skepticism positively predicts country wealth rather than negatively. One explanation is that higher internet access could help spread anti-vaccine misinformation throughout society, particularly for those lower in scientific and medical expert trust (Online Accessibility hypothesis). An alternative explanation is that citizens in richer countries are less aware of vaccine-preventable disease risks because they are rarely experienced directly (Out of Sight hypothesis). To test these hypotheses, we merge country-level data with nationally-representative survey data (N = 149,014) from 144 countries. We find evidence for the Online Accessibility hypothesis; people in countries with greater internet access are significantly more likely to be vaccine skeptical. Further, we examine Americans’ vaccine attitudes during the COVID-19 outbreak - where a communicable disease is very ‘in-sight. ’We find anti-vaccine attitudes are similar or higher than pre-outbreak levels, which counters the Out of Sight Hypothesis.
Article
Full-text available
Background High confidence in vaccination programmes is crucial for maintaining high coverage rates. Across the European Union (EU), however, vaccine delays and refusals are contributing to declining immunisation rates in a number of countries and are leading to increases in disease outbreaks. Methods We assessed the overall state of confidence in vaccines among the public in all 28 EU member states and among general practitioners (GP) in ten EU member states, conducting the largest ever study on attitudes to vaccines and vaccination in the EU, eliciting the views of approximately 28,000 respondents across the 28 EU member states. Results We found that a number of member states (including France, Greece, Italy, and Slovenia) have become more confident in the safety of vaccines since 2015, but that the Czech Republic, Finland, Poland, and Sweden have become less confident. While GPs generally hold higher levels of vaccine confidence than the public, the survey found that 36% of GPs surveyed in Czech Republic and 25% in Slovakia do not agree that the MMR vaccine is safe and 29% and 19% respectively do not believe it is important. Countries whose GPs hold higher confidence in vaccines tend to have a larger proportion of the public expressing positive vaccination beliefs. Conclusions Even countries with well-established vaccination programmes and high levels of confidence are not immune to rising vaccine hesitancy. There is a need for continuous monitoring, preparedness and response plans to maintain and increase confidence in the importance, effectiveness and safety of vaccines, among both the public and health professionals.
Article
Objectives: To quantify the prevalence of parental vaccine hesitancy (VH) in the United States and examine the association of VH with sociodemographics and childhood influenza vaccination coverage. Methods: A 6-question VH module was included in the 2018 and 2019 National Immunization Survey-Flu, a telephone survey of households with children age 6 months to 17 years. Results: The percentage of children having a parent reporting they were "hesitant about childhood shots" was 25.8% in 2018 and 19.5% in 2019. The prevalence of concern about the number of vaccines a child gets at one time impacting the decision to get their child vaccinated was 22.8% in 2018 and 19.1% in 2019; the prevalence of concern about serious, long-term side effects impacting the parent's decision to get their child vaccinated was 27.3% in 2018 and 21.7% in 2019. Only small differences in VH by sociodemographic variables were found, except for an 11.9 percentage point higher prevalence of "hesitant about childhood shots" and 9.9 percentage point higher prevalence of concerns about serious, long-term side effects among parents of Black compared with white children. In both seasons studied, children of parents reporting they were "hesitant about childhood shots" had 26 percentage points lower influenza vaccination coverage compared with children of parents not reporting hesitancy. Conclusions: One in 5 children in the United States have a parent who is vaccine hesitant, and hesitancy is negatively associated with childhood influenza vaccination. Monitoring VH could help inform immunization programs as they develop and target methods to increase vaccine confidence and vaccination coverage.
Article
Purpose: We assess the effectiveness of multilevel regression with poststratification (MRP) as a tool to mitigate selection bias from online surveys of small geographical regions. Methods: We collected self-reported health information from an Internet-based sample of adults residing within the St. Louis, MO, metropolitan area in 2017. We created Bayesian hierarchical models with three sets of predictor variables for each of six common health behaviors and outcomes, with results poststratified using the American Community Survey to estimate region and ZIP Code Tabulation Area-level prevalence. Results: When comparing MRP estimates with a population-based sample as a reference, we found that adjustment using MRP can reduce bias in prevalence estimates and provide estimates for local area prevalence. 14 of 18 adjusted estimates were closer to the benchmark than the unadjusted estimates and MRP using all three covariate sets resulted in better overall agreement with the benchmark compared with the unadjusted estimates. Conclusions: MRP can improve prevalence estimates from self-selected Internet-based samples, although a nonnegligible amount of bias may remain. Illustrating the utility and limitations of this method will help researchers develop relevant estimates of the local public health burden, helping local health officials better understand and reduce poor health outcomes.
Article
Introduction: Vaccination is a crucial tool for the primary prevention of infectious diseases. Thanks to widespread vaccination, in the last century smallpox has been eradicated, and the Americas and Europe have become polio-free. Areas covered: The aim of our study was to assess vaccination coverage in childhood up to booster preschool age in order to update the analysis on European trends of immunization coverage in the last nine years (2009–2017) and to assess the impact of different national policies on vaccine uptake. Expert opinion: Despite the good results, several EU countries are going through unprecedented outbreaks of vaccine-preventable diseases due to insufficient coverage. The increase of vaccine hesitancy has made it difficult to reach and maintain high rates of vaccination coverage. Enforcing mandatory vaccinations is one of the strategies adopted by some countries, like Italy and France, to increase coverage. However, each country should find the most suitable way to keep up with vaccination coverages according to own cultural and organizational background.
Article
There are now a growing number of applications of multilevel regression and poststratification (MRP) in population health and epidemiological studies. MRP uses multilevel regression to model individual survey responses as a function of demographic and geographic covariates. Estimated mean outcome values for each demographic-geographic respondent subtype are then weighted by the proportions of each subtype in the population to produce an overall population-level estimate. We recently reported an extensive case study of a large nationwide survey and found that MRP performed favorably compared to conventional survey sampling weights for the estimation of population descriptive quantities in a highly selected sample. In this study, we aimed to evaluate, by way of a simulation experiment, both the accuracy and precision of MRP versus survey sampling weights in the context of large population health studies. While much of the research into MRP has been focused on U.S. political and social science, we considered an alternative population structure of smaller size and with notably fewer geographic subsets. We explored the impact on MRP performance of sample size, model misspecification, interactions, and the addition of a geographic-level covariate. MRP was found to achieve generally superior performance in both accuracy and precision at both the national and state levels. Results were generally robust to model misspecification, and MRP performance was further improved by the inclusion of a geographic-level covariate. These findings offer further evidence that MRP provides a promising analytic approach for addressing participation bias in the estimation of population descriptive quantities from large-scale health surveys and cohort studies.
Article
Background Anti-vaccination attitudes are important predictors of vaccination behavior. Existing measures of vaccination attitudes focus on specific age groups and/or particular vaccines; a more comprehensive measure would facilitate comparisons across studies. PurposeThe aim of this study was to develop a short measure of general vaccination attitudes and establish its reliability and validity. Methods Two studies were conducted using the VAX scale. For Study 1, participants were 409 individuals (53% female), with a mean age of 34.5 years. For Study 2, participants were 92 individuals (67% female) with a mean age of 28.6. Participants answered paper-and-pencil questions about their attitudes toward vaccines, prior and expected-future vaccination behaviors, perceived sensitivity to medicines, online behavior, and basic demographic information. Exploratory and confirmatory factor analyses were conducted with correlations and t tests then used to assess the scale’s reliability and validity. ResultsFour distinct but correlated vaccine attitudes were identified: (1) mistrust of vaccine benefit, (2) worries about unforeseen future effects, (3) concerns about commercial profiteering, and (4) preference for natural immunity. These factors were significantly related to prior vaccination behavior, future intentions to obtain recommended vaccinations, perceived sensitivity to medicines, and the tendency to obtain health information online. Conclusions The VAX scale provides an efficient method for identifying those with vaccination resistance, and the four subscales enable a more nuanced understanding of the nature of those views. It should be noted, however, that the strong correlations amongst the four subscales suggest that interventions should target all four attitude areas, and it remains to be seen whether differential emphasis across the four areas is warranted.