The public response to vaccine hesitancy – Joseph Robertson | AgeWage: Making your money work as hard as you do

Vaccines: Coverage, hesitancy, and public health response

By Josephine Robertson COVID-ARG.COM

COVID-19 Actuaries Response Group – Learn. Share. Educate. Influence.

Josephine Robertson

Summary

Herd immunity thresholds, though interesting, fail to recognise the heterogeneity in our population in our daily behaviour and in vaccine hesitancy. There are ways to tailor the vaccination roll-out to allow for this diversity, and to engage with the reasons for vaccine hesitancy. Health policy should reflect a deep knowledge of the population to tailor vaccination programmes, surveillance and outbreak responses accordingly if infection is to be managed as non-pharmaceutical interventions are slackened.

Herd immunity and vaccine coverage

Herd immunity is a term often used but can convey different meanings. When considered in line with vaccination benefits a) and c) shown below, it can be considered as “a particular threshold proportion of immune individuals that should lead to a decline in incidence of infections”[i]. Vaccinations can reduce the risk of infection for the individual (benefit a) – the direct effect. They can also reduce the risk of transmission (benefit c) – the indirect effect. Herd immunity is afforded to those with vaccination contraindications due to this indirect effect. For a vaccine to provide herd immunity, the vaccine must provide a sufficient indirect effect and the vaccine coverage must be at a level to overcome the natural propensity of the infections to spread.

At a population level, vaccine coverage (V_c) is a conceptual threshold: it is the proportion of immune individuals required to enable infections to decline in lieu of non-pharmaceutical interventions (NPIs). It relies upon the reproduction rate (R₀) and vaccine real-world effectiveness (E). Where E reduces with time, higher coverage and boosters would be required for the reduced risk of infection to be maintained.

Real-world effectiveness and vaccination take-up statistics are now emerging. Though pleasing to see a higher take-up than initially estimated (seeing 90% vs 75%)[ii], we can assess the theoretical population level vaccine coverage required for a decline in infections in the hypothetical world where NPIs are not implemented. Exploring theoretical coverage levels highlights major contingencies on maintaining reduced infection as NPIs are slackened including: vaccine effectiveness to reduce infection risk, an adequate coverage throughout the population (rather than overall), and actions to protect vaccine effectiveness (booster vaccines, management of virus variant risk).

We assume here that R₀ is 2.63 without NPIs[iii] as it relates to the virus the vaccines were developed in line with. RCT studies provided vaccine efficacy on disease risk. The EAVE II study provides real-world effectiveness for the AstraZeneca and Pfizer vaccines on disease risk[iv]. The more vaccinated people in a population, the lower the disease risk. However, providing a vaccination that does not confer reduced infection risk does not create herd immunity, so the grey cells below are for reference only. Herd immunity requires effectiveness against infection risk[1]. The SIREN study provides evidence on real-world effectiveness of the Pfizer vaccine on risk of infection[v], indicating that a 72% vaccine coverage would be required to see infection remain low and stable. Thresholds are useful to provide scenarios as shown below, .e.g. an alternative R₀ of 4 paired with a 90% coverage require a vaccine effectiveness of 83%.

Table 1: Vaccine Coverage Scenarios

		Vaccine – Dose (Study)	Vaccine Effectiveness	Vaccine Coverage
	R	AZ = AstraZeneca, P=Pfizer, M = Moderna	E	Vc
Efficacy (disease risk)	2.63	AZ – 1 dose (RCT)[vi]	64%	97%
		AZ – 2 dose (RCT)	70%	89%
		P – 1 dose (RCT)[vii]	52%	119%
		P – 2 doses (RCT)	95%	65%
		M – 1 dose (RCT)[viii]	80%	77%
		M – 2 doses (RCT)	95%	66%
Effectiveness (disease risk)	2.63	P – 1 dose (EAVE II) – at 28-34 days[ix]	85%	73%
		AZ – 1 dose (EAVE II) – at 28-34 days	94%	66%
		P&AZ – 1 dose (EAVE II) – at 28-34 days	84%	74%
		P&AZ – 1 dose (EAVEII) 18-64 age – at 28-34 days	85%	73%
		P&AZ – 1 dose (EAVEII) 65-79 age – at 28-34 days	79%	78%
		P&AZ – 1 dose (EAVEII) 80+age – at 28-34 days	81%	77%
Effectiveness (infection risk)	2.63	P – 1 dose (SIREN) <65age – at 21 days[x]	72%	86%
Effectiveness (infection risk)	2.63	P – 2 dose (SIREN) <65age – at 7 days	86%	72%
Scenarios	2.63	Full coverage requires an effectiveness of	62%	100%
	2.63	Fully effective vaccine requires coverage of	100%	62%
	4.00	R₀ = 4 and 90% coverage requires effectiveness of	83%	90%

These simple population thresholds assume random vaccination delivery and random population mixing. However, the vaccination delivery in the UK is non-random; the priority groups are set predominately by age, delivery pace may vary, and take-up rates are likely to vary by region. In addition, populations mix in a non-random fashion. Given this heterogeneity, the breaks in onward transmission due to vaccination may vary in ‘pockets’ of the population, which implies pockets with varying rates of outbreak risk. Where there is increased risk, there is less herd immunity.

Components of V_cmay vary by population pockets. Vaccine effectiveness is likely to vary by age. A degradation of vaccine effectiveness is possible through variants which may be more likely to arise in some pockets (e.g. hospitality staff, tourist locations). Variations in vaccine roll-out and take-up rates create a different ratio of susceptible to immune by area or within communities who regularly mix, affecting the onward reproduction rate. In addition, some variants may not significantly reduce vaccine effectiveness but can increase R as seen with more recent variants. Modelling in this area is complex given the array of parameters involved including spatial and generational mixing.

Outbreak risk: Take-up and non-random vaccination

The priority groups were set with reference to reducing disease risk starting with the most vulnerable. If sufficient evidence emerges on the effectiveness of vaccines to reduce infection risk, there may be reason to reconsider priority groups with higher transmission risk to better achieve herd immunity. However, it is likely both groups would be vaccinated as it may not be ethical to leave people susceptible to disease risk. Higher overall take-up works to better afford those with vaccine contraindications herd immunity. The take-up rates so far for England[xi] are shown[2]. There is variation by region, though overall good vaccine coverage in the initial priority age groups.

Looking closer at a region with a consistently lower take-up across the age bands, London shows variation within the region by NHS Integrated Care Systems (ICSs). During the roll-out as the priority groups are invited for vaccination there will be improved take-up, but less movement occurs over time for the priority groups already called.

The ICS regions can be cross referenced with a map of deprivation.

It can be appreciated how lower vaccine take-up may create geographical outbreak risk.

In addition to geographical risk, the population can be thought of in terms of pockets who undertake non-random mixing. For example, the congregation of sports fans, faith groups, professional conferences, or cultural and festive celebrations. OpenSAFELY data provides insight up to February 24^th 2021 of vaccine coverage by deprivation and ethnicity for the over 80s[xiv]. There appears to be variation in vaccination take-up by Index of Multiple Deprivation and ethnicity. However, these variables are both linked to regional population differences.

Studies suggest take-up is lower in areas with a higher proportion of minority ethnic groups based on early COVID-19 (and other) vaccinations, though this is not always the case[i]. Surveys of COVID-19 vaccination acceptability suggest greater hesitancy in Black /Black British, Pakistani/ Bangladeshi, and Other White background (including Eastern Europeans) groups[ii]. As the ethnicity varies by age band, we expect take-up to be lower in younger priority groups.

We do, however, know that clusters of unvaccinated individuals are vulnerable to infection outbreaks. Where outbreak risk is correlated to other factors associated with health inequalities in society, this can work to deepen inequalities. Mortality statistics show COVID-19 has had a greater impact in the most deprived areas[xviii]. In addition, outbreaks provide an environment for the virus to mutate and so risk degrading the overall vaccine effectiveness.Studies suggest take-up is lower in areas with a higher proportion of minority ethnic groups based on early COVID-19 (and other) vaccinations, though this is not always the case[xvi]. Surveys of COVID-19 vaccination acceptability suggest greater hesitancy in Black /Black British, Pakistani/ Bangladeshi, and Other White background (including Eastern Europeans) groups[xvii]. As the ethnicity varies by age band, we expect take-up to be lower in younger priority groups.

Actions to counter this can be reactive or proactive. Reactive responses may include regional NPIs such as tiered lockdowns or the cancellation of gatherings linked to transmission risk (e.g. school, music festivals). This may cause or deepen social and economic disparity in society. Proactive responses may include near real-time surveillance (symptom, infection, or disease) and public health/field epidemiology (test, track, and trace).

Overall, simplistic thresholds should be used with caution especially when slackening NPIs. Given the complexity in understanding the ways in which clusters of unvaccinated individuals, and so risk of outbreaks, could emerge it may be better to aim for 100% coverage to enable public health practice to proactively reduce outbreak risk. In the meantime, where coverage is known to be lower, there should be increased timely regional surveillance and a robust ‘field epidemiology’ response.

Public Health: Tailoring the vaccination roll-out to improve take-up

The ONS undertake online surveys providing insight to the changing acceptability and so fluidity of vaccine hesitancy in the UK. Though survey responses may overstate actual behaviour[xix], the proportions ‘unlikely to accept’ are much lower than other countries[xx]. Of those unlikely to accept a vaccination (a group which is reducing in line with diffusion theory over time – see Bulletin 101), the main reasons provided are shown in Table 2.

Table 2: Acceptability of Vaccinations

Time Period	ONS[xxi] 2020 to 2021	10-13 Dec	22 Dec 3 Jan	7-10 Jan	13-17 Jan	20-24 Jan	27-31 Jan	3-7 Feb	10-14 Feb	17-21 Feb	24-28 Feb
Offered a vaccine	Vaccinated	–	–	–	7%	12%	16%	22%	26%	35%	39%
	Waiting	–	–	–	5%	6%	6%	5%	4%	5%	5%
	Refused	–	–	–	1%	<1%	<1%	<1%	1%	1%	1%
Not yet offered a vaccine	Likely to accept	78%	85%	86%	89%	88%	91%	89%	88%	90%	90%
Not yet offered a vaccine	Unlikely to accept	10%	7%	5%	5%	5%	4%	4%	5%	4%	4%
Reasons provided by those unlikely to accept or declined when offered Reasons over 20%	Side effect worries	52%	51%	43%	42%	50%	45%	38%	31%	37%	25%
	Long-term health effect worries	46%	51%	46%	43%	41%	44%	43%	41%	43%	45%
	Want to wait to see how well it works	52%	47%	35%	40%	42%	35%	43%	27%	34%	39%
	Do not believe it is safe	29%	38%	28%	26%	25%	20%	26%	20%	28%	28%
	Feel it is not a personal risk	16%	16%	17%	19%	18%	11%	16%	21%	21%	26%
	Worries around fertility	–	–	–	–	–	–	–	16%	22%	17%
	Prefer more in need receive before me	–	–	–	–	–	–	–	27%	17%	17%

Figure 5: Matrix of vaccine hesitancy determinants and intervention (taken from[xxii])

When offered a vaccine, the decision-making involves a person “in effect performing complex cost-benefit analysis based on imperfect assumptions.”[xxiii]. The benefits of vaccination are set out in Box 1. The costs can include time, inconvenience, side effects, and money including travel, child minding, and lost work[xxiv].

To improve take-up is then to improve the understanding of the associated costs and benefits, and where possible reduce associated costs. As the causes of hesitancy can be complex, the approach taken to improve take-up is tailored to its rationale, as shown here in Figure 5[xxv].

Initially refused – Revisiting the choice: Vaccine decision-making is not commonplace for many unless they are a parent or in the annual flu vaccination cycle, and so some may take time to decide. Some may, having seen others assess the cost and benefits, revisit their choice. Logistical interventions can include an easy and non-judgmental re-booking service to those who previously declined.

Barriers to access – Facilitating delivery: There may be good intentions, but money, time and inconvenience play their part. Bus fare reimbursement, mobile vaccination units or child friendly delivery centers may reduce associated costs. Some may not feel comfortable asking questions at large vaccination sites, may feel rushed, or sites might not suit a person’s disabilities or cultural beliefs. This may include where English is a second language. Patient involvement alongside tailored vaccine and information delivery may improve take-up[xxvi]. For others without a permanent address (eg traveler communities, asylum seekers) or no longer living at the GP registered address (eg students, young professionals), there may not be an invitation through the usual process. Logistical interventions may include tailored reach-out and drop-in centers.

Waiting for effectiveness and safety information: Vaccine hesitancy does not necessarily mean refusal and an individual may reconsider their choice. As more information emerges on how well it works and side effects, we can see a reduction in this rationale over time in Table 2. It is important that real-world safety and effectiveness information is portrayed clearly and with care so as not to detrimentally influence people still making their vaccine decision. Long-term health effect worries remains static, as there is less new information emerging in this area. Dialogue and information on how the vaccine works in the body may allay worries.

Other key issues in decision-making include misunderstanding of collective benefit and risk of infection[xxvii]. Not perceiving COVID-19 as a personal risk may tip the scales where prevalence or consequences of infection are not well understood. Where there is less appreciation for the collective benefits, an individual may also be less inclined to take the associated costs. In a similar vein, where take-up is close to the overall V_c threshold, individuals may find a degree of unfounded confidence in being able to ‘free-ride’ herd immunity. Dialogue and information can be provided through access to quality information on Government websites[xxviii] and through the NHS[xxix].

Mistrust: Though a person may wish to do what is best for their own health and the health of their family and community, there may be mistrust in those delivering the information[xxx]. The mistrust may not necessarily be from direct experience but shared historical experience[xxxi]. This may arise for medics, politicians, police or other ‘trusted’ members of society who promote vaccination. Some who experience mistrust, may already be marginalised in society therefore, listening and gaining trust are important to protect wellbeing[xxxii].

Culturally tailored information may be delivered by alternative sources who are trusted (e.g. celebrities, sports personalities) or from NHS workers with similar shared historical experience. Training for healthcare staff can allow for an acknowledgement of mistrust reasons and improve culturally aware dialogue. In addition, delivery can be undertaken in familiar environments to improve acceptability. For example, working with faith-based organisations or vaccinating in local community centres.

Misinformation: Dialogue and information can be provided to reduce hesitancy caused by misinformation. Information dissemination may be tailored to mediums used by groups exposed to misinformation such as social media. Misinformation may cause worries around fertility, the use of animal products and much more. Misinformation and mistrust may interact to have a multiplicative impact on take-up, so it is important to understand where misinformation is being targeted in order to counter it appropriately

Conspiracy: A study on scepticism and conspiracy beliefs surrounding coronavirus makes for an interesting read with beliefs ranging from: the government deliberately misleading the public, the virus is man-made by powerful organisations, the virus is an attempt by activists to stop climate change, to Bill Gates trying to reduce population size, 5G and radiation, and more[xxxiii]. Action here is complex but can start with listening. Though we may not agree with the views expressed, the belief is the lived reality of the individual and so an open and non-judgmental dialogue may allow us to better engage with the perceived issues.

Concluding remarks

A crucial part of improving vaccine coverage is compassion. By not blaming communities or creating a stigmatizing environment, society acknowledges it is richer for its diversity, and in acknowledging that diversity, we accept that a ‘one size fits all’ approach is not always suitable. If society applies a uniform approach and some groups do not conform, this reflects not a failure of those groups but a failure of society to adapt to its diversity.

Tailoring of the vaccine programme has started[xxxiv] but this is only part of the story. Through considering theoretical coverage levels, the major contingencies of slackening NPIs are more apparent including: the vaccine’s effectiveness in reducing infection and transmission risk, adequate coverage throughout the population (rather than overall), and actions to protect vaccine effectiveness (booster vaccines, controlling virus variant risk).

In the next bulletin, we will look at the non-compulsory nature of the vaccination programme and the ethics of alternative approaches.

9 March 2021