Background
Varicella zoster, commonly referred to as chickenpox, is a respiratory infectious disease that causes a characteristic red rash and pox on the skin surface [
1]. It is caused by the varicella zoster virus (VZV) which also causes shingles, often referred to as herpes zoster. Chickenpox symptoms typically arise 1–3 weeks after exposure to an infected individual, and a newly infected individual is infectious for around a week starting 1–2 days prior to the onset of symptoms. Symptoms last approximately 2 weeks, when the virus then retreats to the nerve ganglia in the spine [
2,
3]. By the age of 15, nearly all children have antibodies to VZV, whether from a natural infection or vaccine [
4,
5]. In 10–30% of adults, the latent varicella virus will reactivate and manifest as shingles, typically in adults aged over 60 [
6,
7]. The VZV vaccine prevents infection with VZV in children while a booster dose, referred to as a shingles vaccine, later in life suppresses VZV reactivation in adults [
4].
The chickenpox vaccine was first approved for use in the USA in 1995. Despite its successful use for a quarter century, global VZV vaccination policy remains a topic of much debate. It is a live-attenuated vaccine administered in two doses during childhood. The shingles vaccine is either a single (Zostovax) or double (Shingrix) dose recombinant vaccine given later-in-life to suppresses reactivation [
4,
8,
9]. Chickenpox vaccination is only used in a limited number of countries, and there are multiple reasons why most countries have not yet implemented vaccination. First, complications from chickenpox are rare, with less than 1% of infected individuals experiencing severe illness [
10]. Second, childhood immunization against VZV has been hypothesized to reduce natural VZV exposure in adults. This reduction of VZV exposure could then reduce VZV immune-boosting, which would cause additional reactivation [
11‐
16]. Third, low and intermediate levels of chickenpox vaccine coverage would shift the age distribution of chickenpox infection onto older age groups who would carry a higher burden of disease [
17]. Severe complications become more common in individuals who contract chickenpox at an older age [
18,
19]. Fourth, as with other vaccine-preventable diseases, it is likely that natural infection provides longer immunity than a vaccine dose [
20,
21]. Despite the complexities surrounding VZV vaccination, some countries have chosen to vaccinate. Although the percent of infections resulting in serious illness is low, when endemic VZV infects the majority, vaccination prevents a substantial number of serious cases (e.g., hospitalizations). For instance, vaccination prevents an estimated 4 million cases each year in the United States [
22], and 1% of those averted cases (approximately 40,000) would have been serious illness.
The primary reason the VZV vaccine has not been implemented worldwide is the presumed reduction of immune-boosting in adults. Theoretical models have predicted an increase in shingles incidence with the inclusion of the chickenpox vaccine on a countries childhood immunizations schedule. [
23,
24]. Importantly, these models are theoretical and have not been fit to, or challenged with, data. Out of necessity they have been developed in the face of many unknowns regarding viral latency. Immunity boosting has been used as a general term for the reinforcement of VZV-specific immunity, which is most likely dominated by cellular immunity. It is this T cell-mediated immunity that protects from VZV reactivation [
15]. However, empirical evidence from surveillance programs in locations that vaccinate against chickenpox have been inconclusive about the impact of chickenpox vaccination on shingles. There have been both observed increases [
25‐
27] and no change [
28,
29] in shingles incidence.
It is difficult to discern whether increases in shingles are due to the vaccine or improved reporting. Prior to the introduction of the chickenpox vaccine, cases of shingles had been increasing in Canada, [
30,
31], the UK [
30], and the USA [
32]. In Spain, where chickenpox vaccination occurs in a limited geographic areas (e.g., Madrid but nowhere else on the mainland [
33]), shingles incidence has been steadily increasing due to demographic changes [
34]. With the global availability of the VZV vaccine, and other herpesvirus vaccines in development [
35‐
37], it is vital to understand the long-term impacts of any vaccine introduction. Data from Thailand were selected for this simulation study due to the availability of population level chickenpox and shingles data. Multiple immunization scenarios were examined in Thailand, which does not vaccinate, to interpret the long-term dynamics of chickenpox and shingles.
Discussion
In this study, mathematical models fit to chickenpox and shingles data from Thailand were used to simulate various vaccination scenarios to (i) reveal that any introduction of a chickenpox vaccine would drastically reduce chickenpox incidence; (ii) identify a non-linear relationship in chickenpox coverage and reduction of chickenpox cases; (iii) demonstrate that any introduction of a shingles vaccine with realistic coverage levels (\(\le\) 50%), in combination with realistic chickenpox coverage levels (\(\ge\) 35%) would increase shingles incidence, unless the immunity provided from shingles vaccination was lifelong; and (iv) uncover a trade-off in chickenpox and shingles vaccine coverage on shingles incidence. The lack of population-level shingles data had previously limited VZV vaccination policy research to theoretical models or models based on small-sample sizes. Here, the dynamical implications of different vaccination scenarios were examined using models fit to population-level chickenpox and shingles data.
Simulations of chickenpox vaccination in Thailand demonstrated the potential for up to a
\(91.6\%\) drop in chickenpox cases during the initial 8-year study period (Fig.
1). These results were derived from realistic scenarios reflecting previous immunization efforts in Thailand, and are in-line with the US experience with VZV vaccination, which had a 67–84% reduction in chickenpox cases after only 5 years of immunization [
40]. As expected, increased chickenpox vaccine coverage reduced the total number of chickenpox cases; additionally, even a gradual vaccine roll-out would drastically reduce chickenpox morbidity if higher coverage levels were not feasible, which is evidence of herd immunity [
41,
42]. Simulations also revealed that low-to-medium chickenpox vaccination efforts would have larger than expected impacts on reducing chickenpox incidence, while higher vaccination efforts would have reduced effects on chickenpox incidence (Fig. S
1). There were minimal differences between the three chickenpox roll-out scenarios over the course of the 100-year simulation; in all scenarios chickenpox cases fell below 2% of the null vaccination scenario after 50 years. Our model assumed heterogeneous mixing, and halfway through the 100-year simulation there were less than 50 cases of chickenpox annually across the country under all vaccination scenarios. It is likely that chickenpox would stochastically die off well before this point due to low numbers and herd immunity, with occasional re-introductions to pockets of under-vaccinated populations [
43].
Under realistic scenarios of shingles vaccination (up to 50% coverage with 5 years of protective immunity), simulations that included slow, moderate, or aggressive chickenpox vaccine roll-out increased shingles incidence (Fig. S
2), though some scenarios of lower chickenpox coverage resulted in slight shingles case reductions (Fig.
2). This was because immunity from chickenpox vaccination only provided 20 years of protection, which increased the number of individuals available for shingles reactivation. The only simulations where vaccination had a noticeable long-term impact on shingles reduction compared to no vaccination occurred when immunity from shingles coverage was vaccination was lifelong or when immunity was 5 years and shingles coverage was extremely high. Even with lifetime immunity from shingles vaccination, there were no scenarios where realistic shingles coverage equated to an equal rate of case reduction when chickenpox vaccination was also included (Tables S
2 and S
3). Only when chickenpox vaccination was removed and shingles immunity was lifelong did vaccinating half (UK estimate) or a third (US estimate) of the population reduce the total shingles cases by 50% or 33% (Fig.
4). While scenarios with lower chickenpox coverage/coverage and high shingles coverage with lifetime immunity performed best, we observed an interesting trade-off between chickenpox uptake and shingles coverage, which existed in both the 5 year and lifetime shingles immunity simulations (Figs.
2 and S
2).
Under realistic vaccination scenarios observed in countries that currently vaccinate against both chickenpox and shingles, where the shingles vaccine provides 5 years of immunity and high chickenpox coverage (
\(\ge\) 50%) exists, anything less than 80% shingles coverage led to an increase in shingles incidence compared to the null model at some point in the simulation (Figs.
2 and
3). This high level of shingles coverage would be nearly impossible to achieve at the population level, as routine adult immunizations are uncommon and the highest cited level of national shingles coverage is 50% [
44‐
46]. Under both the 5 year and lifetime immunity from shingles vaccination simulations, mid-level chickenpox coverage resulted in a greater reduction in shingles incidence than low or high chickenpox coverage. This ‘C’ shape can be seen in Figs.
2 and
3 (seen as a “U”). If chickenpox vaccine immunity was extended to 40 years, this pattern remained, though the increase in shingles cases at higher chickenpox vaccination levels was reduced (Fig. S
3). These counter-intuitive results reveal that to achieve the greatest reduction in both chickenpox and shingles policy makers should strive for mid-level chickenpox coverage and focus their efforts on maximizing shingles coverage.
An important limitation of this work, which examined chickenpox and shingles dynamics under various vaccination scenarios in Thailand, was that it did not include infection complications, including VZV caused death in the model. Any low- to mid-level chickenpox coverage would lead to an increase in the mean age of chickenpox infection, which could lead to more serious chickenpox complications in unvaccinated individuals if herd immunity were not rapidly achieved and sustained. However, while chickenpox vaccination would increase shingles incidence, previous work has demonstrated that chickenpox vaccination would reduce the mean age of shingles reactivation [
47], potentially curtailing serious side-effects of shingles, as younger individuals tend to have milder symptoms [
48‐
51].
Furthermore all simulations omitted the potential impact of chickenpox vaccination reducing population-level VZV exogenous boosting in adults. Any reduction of this boosting (via chickenpox vaccination) would likely increase the number of shingles cases, particularly at higher chickenpox vaccination levels seen in Figs.
2,
3,
4, and S
3. We previously attempted to fit exogenous boosting to these data from Thailand [
38], and models that included boosting did not perform better. These are vital next steps in VZV vaccination research. There have been a host of theoretical models attempting to understand the impact of such boosting, however population-level studies remain sparse, primarily because shingles is not a notifiable disease in most countries. Important questions regarding exogenous boosting include; What is the relationship between chickenpox vaccination and exogenous VZV boosting? At what levels of population-level chickenpox vaccination coverage do we start to see an impact (decrease) in population-level VZV immunity? Is this relationship linear or exponential? Does chickenpox vaccination impact the length of immunity from VZV reactivation differently from individuals with a natural infection?
Conclusions
The complicated nature of VZV makes it impossible to select a single vaccination scenario as universal policy. Some countries may wish to minimize total VZV cases, while others may prefer to focus on chickenpox or shingles individually. Strategies focused on reducing both chickenpox and shingles incidence, but prioritizing the latter, should concentrate on raising awareness for shingles vaccination [
45] and maximize efforts towards shingles vaccination, while slowly incorporating chickenpox vaccination. The observed non-linear relationship between chickenpox coverage and the number of cases prevented could be exploited to minimize both chickenpox and shingles incidence. Low and high chickenpox vaccine coverage performed similarly in preventing chickenpox cases during the first few years of the simulation (Fig.
1), and were nearly identical in the long term (Table S
2 and S
3) while lower chickenpox coverage also prevented excess shingles cases (Fig. S
2). Alternatively, countries may wish to minimize VZV complications of both chickenpox and shingles, which would lead to maximizing vaccination across both chickenpox and shingles. Balancing the consequences of vaccination to overall health impacts, including understanding the impact of an altered mean age of infection for both chickenpox and shingles, would need to be considered prior to any vaccine introduction.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.