LAUSR

The manuscript by Hoshi et al. provides the first economic evaluation of vaccination against Herpes Zoster (HZ) in older adults in Japan, and as such fills an important gap. The authors utilize a well-structured Markov model with health states representing HZ disease progression, reflecting Japan-specific input data, and supported by both deterministic and probabilistic sensitivity analyses. Hoshi et al. based their model inputs for vaccine effectiveness and associated waning, on statistical modelling presented in the paper published by Li et al. [1] Li et al. developed their statistical model using only data from the Shingles Prevention Study (SPS) and the Short Term Persistence Study (STPS), an extension of the original SPS, which assessed the vaccine efficacy (VE) of Zoster Vaccine Live (ZVL) produced by an overseas manufacturer. A Long Term Persistence Study (LTPS) was also carried out and provides a more complete follow-up of subjects in the SPS study [2]. The model by Li et al. failed to include data from the LTPS study citing a lack of a concurrent placebo control group as the reason for exclusion. As articulated in a letter by Le and Rothberg [3], the modelling results of Li et al. are not consistent with the observed VE estimates in the LTPS study. For example, the LTPS study provided a VE estimate at year 11, of 1.7%, suggesting that in subjects aged 60 years old VE for ZVL had waned to zero by year 11. This is in direct contrast to the results from the modelling exercise of Li et al. which suggest that VE persist for much longer, e.g. 15–20 years. Furthermore, based on the LTPS study, the Center for Diseases Control and Prevention developed a model where VE against HZ for ZVL waned to zero over 10 years in all the age groups [4]. Importantly, in a recent study using data from Kaiser Permanente Southern California, the observed effectiveness of ZVL fell from 68.7% in year 1 to 4.2% in year 8 [5], an even more rapid waning of VE for ZVL than reported in the clinical trials, again suggesting that the VE results in Li et al. are optimistic. Furthermore, it is also important to note that VE estimates in the SPS study, where subjects were followed for a median of 3.1 years, were 41% and 18% for subjects aged 70–79 and 80 years old, respectively [6]. However, the estimated values identified from the Li et al. model and reported in table 1 of Hoshi et al. range from 49.0% to 58.9% for subjects aged 70 and 75 and 42.8% to 45.7% in subjects aged 80 years old during the first 3 years. In summary, considering alternative VE and waning scenarios as discussed above, based on both clinical and real-world long term evaluations, may generate relevant outputs to assess the merits of HZ vaccination in Japan.