Modeling Cost-Effectiveness of Universal Varicella Vaccination With Different Varicella Vaccines in the United Kingdom

To the Editor—We read with interest the article by Akpo et al [1] comparing the cost-effectiveness of varicella vaccination in the United Kingdom (Varilrix, Priorix-Tetra, GSK, Belgium [V-GSK] and Varivax, ProQuad, Merck & Co, Inc, Kenilworth, NJ, USA [V-MSD]). This is an important contribution to the literature demonstrating value of varicella vaccination; however the use of predicted efficacy inputs for 1-dose V-MSD may not accurately reflect the actual vaccine performance and cost-effectiveness, considering availability of observed efficacy and effectiveness data. Efficacy inputs are among the key drivers of the cost-effectiveness of any intervention. The authors derive 1-dose efficacy inputs of 78% for V-MSD from a methodological study using a statistical model [2] relating immunogenicity data (varicella-zoster virus antibody titer, >5 glycoprotein enzyme-linked immunosorbent assay units per milliliter, 6 weeks after vaccination) to long-term disease breakthrough. The efficacy estimate reported [2] was based on antibody titer with predicted efficacy of 94.0% for all ages and 87.2% in younger children (n = 326; median age, 13 months). However, Akpo et al [1] used predicted efficacy of 78% from sensitivity analysis that was included to illustrate the impact of a 2-fold decrease in antibody titer on efficacy (from 88% to 78%) in children who were vaccinated at age 18 months. While immunogenicity data can be used as a correlate of protection, using predicted efficacy based on antibody titers alone is a limitation given actual efficacy data is available for V-MSD. Several randomized control trials (RCTs) and observational studies have been published, demonstrating the long-term efficacy [3–6] and effectiveness of V-MSD [7–9]. Kuter et al [3], in an RCT with 10 years of follow-up, showed that 1-dose efficacy of V-MSD was 94.4% (95% confidence interval, 92.9%–95.7%), and 2-dose efficacy was 98.3% (97.3%–99.0%). Akpo et al [1] did not include the data showing higher efficacy of 1-dose V-MSD [3], with the rationale that this RCT was conducted in children aged 12 months to 12 years and noting that older children may experience a lower risk of infection. However, the average age in this RCT was 4.43 years, supporting efficacy in younger children. Another RCT showed that the seroconversion rates for V-MSD by age groups were comparable—98% for age 12–15 months, 97% for 16–23 months and 2–4 years, and 95% for 5–12 years—with efficacy of 86% for all ages [4]. Two other RCTs (average age of children, 3.6 years and 15 months) showed 1-dose efficacy of V-MSD of was 88.5% and 90.5%, respectively [5, 6]. Similarly, literature reviews, meta-analyses and surveillance studies with up to 14 years of follow-up have shown 1-dose effectiveness for V-MSD ranging from 81% to 100%, depending on disease severity [7–9] (Table 1). The incremental cost-utility ratios reported in the publication showed marginal differences (at most 15%) between the 2 vaccines across all scenarios and time horizons. Given the sensitivity of incremental cost-utility ratio estimates to small changes in utility gains, results regarding the relative cost-effectiveness of different vaccines need to be interpreted with caution. Sensitivity analyses of relevant data sources for efficacy parameters are warranted to comprehensively test the performance and cost-effectiveness of these vaccines.


Modeling Cost-Effectiveness of Universal Varicella Vaccination With Different Varicella Vaccines in the United Kingdom
To the Editor-We read with interest the article by Akpo et al [1] comparing the cost-effectiveness of varicella vaccination in the United Kingdom (Varilrix, Priorix-Tetra, GSK, Belgium [V-GSK] and Varivax, ProQuad, Merck & Co, Inc, Kenilworth, NJ, USA [V-MSD]). This is an important contribution to the literature demonstrating value of varicella vaccination; however the use of predicted efficacy inputs for 1-dose V-MSD may not accurately reflect the actual vaccine performance and cost-effectiveness, considering availability of observed efficacy and effectiveness data.
Efficacy inputs are among the key drivers of the cost-effectiveness of any intervention. The authors derive 1-dose efficacy inputs of 78% for V-MSD from a methodological study using a statistical model [2] relating immunogenicity data (varicella-zoster virus antibody titer, >5 glycoprotein enzyme-linked immunosorbent assay units per milliliter, 6 weeks after vaccination) to long-term disease breakthrough. The efficacy estimate reported [2] was based on antibody titer with predicted efficacy of 94.0% for all ages and 87.2% in younger children (n = 326; median age, 13 months). However, Akpo et al [1] used predicted efficacy of 78% from sensitivity analysis that was included to illustrate the impact of a 2-fold decrease in antibody titer on efficacy (from 88% to 78%) in children who were vaccinated at age 18 months.
Akpo et al [1] did not include the data showing higher efficacy of 1-dose V-MSD [3], with the rationale that this RCT was conducted in children aged 12 months to 12 years and noting that older children may experience a lower risk of infection. However, the average age in this RCT was 4.43 years, supporting efficacy in younger children. Another RCT showed that the seroconversion rates for V-MSD by age groups were comparable-98% for age 12-15 months, 97% for 16-23 months and 2-4 years, and 95% for 5-12 years-with efficacy of 86% for all ages [4]. Two other RCTs (average age of children, 3.6 years and 15 months) showed 1-dose efficacy of V-MSD of was 88.5% and 90.5%, respectively [5,6]. Similarly, literature reviews, meta-analyses and surveillance studies with up to 14 years of follow-up have shown 1-dose effectiveness for V-MSD ranging from 81% to 100%, depending on disease severity [7][8][9] (Table 1).
The incremental cost-utility ratios reported in the publication showed marginal differences (at most 15%) between the 2 vaccines across all scenarios and time horizons. Given the sensitivity of incremental cost-utility ratio estimates to small changes in utility gains, results regarding the relative cost-effectiveness of different vaccines need to be interpreted with caution. Sensitivity analyses of relevant data sources for efficacy parameters are warranted to comprehensively test the performance and cost-effectiveness of these vaccines.   [2] for the OKA/Merck vaccine. We wish to clarify why the 10-year OKA/Merck VE of 94.4% [3] was considered inappropriate, with emphasis on vaccination age, dose level (plaque-forming units [PFU]) and effectiveness studies.
The Kuter et al. [3] study was a 10-year follow-up of Weibel et al. [4], in which subjects aged 1-12 years (mean age, 4.43 years) received a 17,430 PFU-containing formulation. In the study by Povey et al. [1], children aged 12-22 months (mean age, 14.2 months) received the OKA/RIT vaccine with a potency of 1,995 PFU.
Studies by GSK and MSD suggest that older age at vaccination leads to a lower risk of varicella and a higher VE. Varis and Vesikari [5] demonstrated a lower VE with OKA/RIT vaccinees aged 10-18 months (64%) versus vaccinees aged 19-24 months (82%). Chan et al. [2] showed that at 5gp enzyme-linked immunosorbent assay, the risk of varicella infection decreased by ~ 80% in children aged 5.5 years versus children aged 1.5 years. Comparisons at equivalent titers indicated that the varicella infection risk decreased by ~ 73% in children aged 4.43 years versus children aged 14 months.
VE differences resulting from varying dose levels need to be highlighted as higher doses (10,000-17,000 PFU) are associated with better protection than lower doses (1,000 PFU) [5,6]. This is illustrated by a crude comparison of the 100% OKA/Merck VE after 9 months of follow-up in Weibel et al. [4] with the 86% VE at 1 year in White et al. [7], in which the OKA/Merck dose ranged between 1,000 and 1,625 PFU among enrollees with a mean age of 3.98 years. Similarly, Kuter et al. [8], in a 7-year follow-up of Weibel et al. [4], with enrollees aged 4.7 years on average reported that 95% of vaccinees remained varicella-free following household exposure. This VE rate could be compared with Vessey [2], for the reason previously reported, acknowledging limitations inherent to the absence of head-tohead efficacy studies across similar age groups and dose levels. A meta-analysis of observational studies by Marin et al.
[10] reported a pooled 1-dose VE of 81% (95% confidence interval, 78%-84%) against any varicella with no differences by vaccine, in agreement with our conclusion on predicted similar effectiveness between GSK and MSD varicella-containing vaccines.
Conclusively, we believe that the most accurate VE estimate was used for the OKA/Merck vaccine. Importantly, both vaccines effectively reduce the varicella burden, with GSK varicella-containing vaccines potentially being more cost-effective.