Our analysis demonstrates that vaccination with two doses of HZ/su would result in a substantially higher number of HZ cases prevented and number of QALYs gained as compared to ZVL (single dose or single dose + booster). This was explained by the relatively higher efficacy of HZ/su. However, whether vaccination with HZ/su would be cost-effective as compared to no vaccination or to ZVL depends largely on the vaccine cost per series. We found that especially for the 60-year-old cohort, there are pricing scenarios in which ZVL could potentially be the most cost-effective alternative. Adding a ZVL booster after 10 years was expected to reduce the effectiveness gap between ZVL and HZ/su but required a significant decrease of the vaccine cost per dose to be the most cost-effective alternative. The inclusion of additional efficacy against PHN or the use of recent post-licensure effectiveness data of ZVL was found to improve the cost-effectiveness of ZVL among ≥ 70-year-olds, increasing the competition between the two vaccines in these age cohorts.
Implications and possible explanations for findings
A threshold of €20,000 per QALY gained is the general Dutch cost-effectiveness threshold for healthcare interventions such as vaccination. At optimum vaccination age, i.e. 70 years for HZ/su and 60 years for ZVL, the maximum vaccine costs to remain cost-effective were estimated at €109 per series and €51 per dose, respectively. These estimates are considerably lower than the currently available vaccine costs of ZVL and HZ/su in the private sector. For instance, the private sector cost of two doses of HZ/su would be €226 (US$280) according to the price list of the Centers for Disease Control and Prevention, and the private sector cost per dose of ZVL in the Netherlands is currently €145 [
53,
54]. Our model shows that HZ vaccination would not be cost-effective with either HZ/su or ZVL when the above-mentioned vaccine costs are used.
The optimum age of vaccination with HZ/su would be in the range of 60 to 80 years, as the efficacy is relatively constant with vaccination age. However, the cost-effectiveness and the optimum age of vaccination with HZ/su were highly sensitive to the duration of protection. Currently, the waning rate had to be estimated using 4 years of follow-up data from the trial, showing a decrease in efficacy over time. However, a closer look to the trial data by arm shows that the waning of efficacy within the first 4 years may be caused by a decrease of HZ incidence in the placebo group rather than an increase of HZ incidence in the vaccinated group [
20]. Such a decrease of HZ incidence over time in the placebo group is unexpected because the risk of HZ usually increases with age. Our sensitivity analysis demonstrated that a lower waning rate would have a substantial impact on decision-making, as the optimum age of vaccination changed from 60 to 80 years to 50–60 years when a lifetime time horizon was used.
Based on the trial data, the optimum vaccination age of ZVL was 60 years and decreased rapidly with increasing age. However, sensitivity analyses of ZVL demonstrated that the cost-effectiveness of ZVL improves substantially for ≥ 70-year-olds when an additional efficacy against PHN was taken into account. Post-licensure studies also confirmed that ZVL was likely to provide additional protection against PHN as well as to other severe outcomes like ophthalmic HZ and HZ-related hospitalizations [
14,
17,
50,
55]. Moreover, effectiveness studies in the US and UK found a relatively stable effectiveness of ZVL with increasing age, explaining why in our sensitivity analysis the cost-effectiveness decreases for 60-year-olds but improves considerably for 80-year-olds when this data was explored in our model. This implies that based on trial data only, the impact of ZVL might be underestimated for the eldest and that ZVL could be a competitor for HZ/su across all age groups. However, it should be noted that although effectiveness studies are more representative for real-life conditions and have more statistical power due to larger sample sizes, they can be affected by uncontrolled confounders. For instance, observational studies rely on the healthcare registries and are therefore more likely to include severe cases as compared to clinical trials that use active surveillance.
Next to cost-effectiveness, budget impact is often considered to be important to decision-makers. As most of the gains are due to the reduction of the burden of illness, vaccination would increase the total healthcare expenditure on HZ considerably. In budget impact analyses, usually a relative short time horizon of maximum 5 years is used and future costs and health effects are not discounted. Under these conditions, the average annual budget impact of vaccination of 50% of the 60-year-olds would be €6.0 million per year for ZVL (one-dose schedule) and €12.7 million per year for HZ/su (two-dose schedule), when corresponding maximum vaccine costs per series of €51 per dose for ZVL and €104 for HZ/su were used. This implies that the implementation of HZ/su would result in a twofold increase of the total healthcare expenditure on HZ as compared to ZVL and a more than fourfold increase as compared to no vaccination.
As HZ/su needs to be administered twice within 2–6 months, vaccination will result in higher healthcare utilization as compared to ZVL that is given as a single dose. Moreover, the prospect of a two-dose regimen might also result in a lower uptake as compared to a one-dose regimen. But we do also note that HZ/su is registered for immunocompromised populations as well, which might be beneficial to the overall vaccination coverage. Both vaccines can be safely combined with influenza vaccination [
56,
57], which might facilitate the implementation and save administration costs. However, influenza vaccination is only given once a year, and immunogenicity data indicates that revaccination with HZ/su after 12 months is less immunogenic as compared to 2–6 months after the first dose [
58]. Our sensitivity analysis shows that the adherence to the second dose of HZ/su impacts the cost-effectiveness considerably, as a single dose of HZ/su is expected to have a substantially lower efficacy, especially among ≥ 70-year-olds, and a higher waning rate as compared to two doses [
47].
Safety studies showed that both HZ/su and ZVL were not associated with serious adverse events among immunocompetent older adults but that HZ/su gives a substantially higher risk of grade 3 adverse events and local adverse events as compared to ZVL [
59]. On the other hand, vaccination with HZ/su can also reduce the risk of serious adverse events because post-licensure studies of ZVL indicate that immunocompromised individuals were, although its contraindication, occasionally vaccinated [
15,
17]. ZVL can cause serious adverse events in immunocompromised patients, as it may result in a symptomatic, progressive infection of the vaccine virus, causing severe rashes [
60].
During the evaluation of ZVL by the Dutch Health Council in 2016, it was concluded that vaccination against HZ did not meet the criteria to be included in the national immunization programme because it does not control VZV transmission nor does it prevent significant mortality [
13]. Vaccination against HZ might however be indicated for a public programme when the vaccine would be regarded as essential healthcare due to a substantial reduction of the individual disease burden [
61]. The Dutch Health Council considered ZVL not as essential healthcare because of its relatively low efficacy in the eldest, short duration of protection and the contraindication for immunocompromised individuals [
13]. Our results demonstrate that HZ/su is expected to have a significantly higher impact on the health economic burden of HZ as compared to ZVL (without or with a booster after 10 years), especially among ≥ 70-year-olds.
In our opinion, these results are also of interest to other countries that are reconsidering HZ vaccination. Recently, the US Advisory Committee on Immunization Practices (ACIP) decided to (i) give HZ/su a preferential status above the ZVL vaccine, (ii) extend the target group from all immunocompetent ≥ 60-year-olds to all immunocompetent ≥ 50-year-olds and (iii) revaccinate individuals that had previously been vaccinated with ZVL [
59]. The UK launched a publicly funded vaccination programme using ZVL for 70-year-olds with a catch-up for 78-year-olds in 2013 [
62] but now needs to decide whether vaccination with HZ/su should be preferred above ZVL, and if so, whether ZVL-vaccinated individuals should be revaccinated with HZ/su. Since the UK Joint Committee on Vaccination and Immunisation recently suggested a similar cost-effectiveness threshold for vaccines as compared with the Netherlands of £15,000 (€17,400) per QALY gained [
63] and the incidence of HZ tends to be similar across European countries [
64], the HZ/su threshold vaccine cost per series might be in the same range as we estimated for the Dutch setting. However, the cost-effectiveness of HZ/su in a cohort that is vaccinated with ZVL will presumably be decreased due to a remaining protection offered by ZVL. For instance, with the use of our model, the threshold vaccine cost per series of HZ/su among 70-year-olds decreased from €109.2 per dose to €80.4, €97.6 and €107.2 per series at 1, 3 and 5 years, respectively, after vaccination with ZVL, when using published vaccine effectiveness data from the UK [
50].
Comparison with other studies
A recent study from the US found that the cost-effectiveness of vaccination of 60-year-olds against HZ would remain below a cost-effectiveness threshold of US$50,000 (€40,400) per QALY gained when the vaccine cost per series was below US$360 (€290) for HZ/su and US$350 (€282) for ZVL [
21]. These costs were, after adjusting for the higher cost-effectiveness threshold, relatively higher as compared to our findings, which can be explained by the use of a lifetime time horizon, a threefold higher healthcare costs per HZ episode and the inclusion of additional protection against PHN and burden of illness. With the same model, it was demonstrated that a HZ/su booster in individuals previously vaccinated with ZVL would only be cost-effective within 5 years after vaccination if the adherence to the second dose of HZ/su approached 100% [
65]. A public health impact study for Germany estimated a similar NNV to prevent a HZ case among ≥ 70-year-olds of 10 for HZ/su and somewhat higher NNV of 50 for ZVL [
47]. A cost-effectiveness analysis with the same model found that the ICER of vaccination with HZ/su ranged between €37,000 and €44,000 per QALY gained when the cost per series was €220 [
22]. A recent study from Italy estimated the cost-effectiveness of ZVL while taking into account the effect of demographic changes over time and an accompanying varicella vaccination programme [
5]. They found that the incidence of HZ is expected to increase over the next decades due to the ageing of the population, that varicella vaccination might cause a further increase of the incidence of HZ because of the reduction of exogenous boosting and that HZ vaccination would cost-effectively reduce this increasing burden of HZ. Finally, two earlier Dutch studies assessing the cost-effectiveness of ZVL estimated ICERs of €22,000 per QALY gained and €30,000 per QALY gained for 70-year-olds using vaccine cost per dose of €77 and €87, respectively [
66,
67]. The most important explanation for finding a relatively lower threshold cost per dose in our current study was a substantially lower QALY loss per HZ case. Some other differences were the use of updated HZ incidence rates and long-term efficacy data of ZVL.
Strengths and limitations
The main strength of our study is that we were able to combine high-quality vaccine efficacy data from large clinical trials with HZ burden estimates from national data sources. HZ incidence was obtained from a GP network that has been validated as a good representation of the general Dutch population. Cost and QALY loss estimates were obtained from a large Dutch prospective cohort study with a long-term follow-up period of 12 months after onset and using the validated EQ-5D instrument to estimate HR-QoL. Moreover, we included a vaccination alternative of ZVL with a booster in our analysis, which has not been compared with HZ/su so far and we are the first exploring post-licensure effectiveness data of ZVL in a cost-effectiveness model.
Our analysis also has its limitations. The duration of protection of HZ/su is currently unknown, which appeared to be an important parameter in the sensitivity analysis. Next, data on adherence to the second dose of HZ/su and the efficacy and waning of one dose of HZ/su is scarce, which also was shown to have an impact on the cost-effectiveness. Also, applying data from the total Dutch population on immunocompetent cohorts might have led to an overestimation of the impact of vaccination. However, we performed a sensitivity analysis by adjusting the epidemiological parameters for this aspect. The patient recruitment and data acquisition of the prospective cohort study that was mainly used for the estimation of HZ-related costs and QALY losses were partly web-based, which might have introduced a selection bias due to the inclusion of healthier subjects able to understand and fill out a web-based questionnaire. However, the target group in our model consisted of immunocompetent individuals, which might also represent a healthier cohort than the general population. Finally, rare HZ-related complications like monaural deafness and monocular blindness were not included in our model.