RR:C19 Evidence Scale rating by reviewer:
Reliable. The main study claims are generally justified by its methods and data. The results and conclusions are likely to be similar to the hypothetical ideal study. There are some minor caveats or limitations, but they would/do not change the major claims of the study. The study provides sufficient strength of evidence on its own that its main claims should be considered actionable, with some room for future revision.
In recent weeks, the SARS-CoV-2 pandemic intensified with many countries around the world reporting increasing number of infected cases and deaths, with epidemiological evidence of growth observed in some jurisdictions, suggestive of a ”third wave” of pandemic. This latest stage of the pandemic is characterized by the increased burden of infections caused by variants of concerns (VOC) as well as the start of vaccination campaigns in some countries. Despite multiple vaccines receiving authorizations by the respective regulatory authorities around the world and the relative success of early vaccination phases in some countries, the roll-out of vaccination, however, continues to be dependent upon limited supplies of the vaccines. Most authorized vaccines require administration of 2 doses separated by 3-4 weeks based on the data from randomized control trials. However, there are instances when public health authorities modify vaccination regimens (either by changing the intervals between doses or reducing the number of doses administered), based on evidence emerging through post-marketing studies.
In this paper, the authors examined the population health outcomes of two approaches to vaccination strategy: the one – with public health authorities adhering to the standard interval between 2 doses of mRNA SARS-CoV-2 vaccines tested in RCTs, and the other – where the second dose of such vaccines is delayed to allow for a larger proportion of population to receive at least one dose of the vaccine. The rationale for the latter strategy is that it would allow to achieve herd immunity in a shorter period of time and to stop or significantly slow down the exponential growth of the third way of the pandemic.
The authors in this paper used agent-based mathematical modelling to explore several important scenarios for the delayed second dose immunization strategy.
The strength of the study is that a range of vaccine effectiveness after the first dose as well as the speed of the roll-out of vaccination campaign are examined. The implementation of the age-dependent phases of vaccination is consistent with vaccination recommendations in many countries. The impact of two vaccination strategies described above was evaluated based on mortality and cumulative infections in a synthetic 100,000 population. The infection dynamic was simulated in the model through household, occupational and random networks. The authors also included a simulation of a delayed vaccination strategy excluding individuals over 65 years of age. The rationale for this was based on the need for maximum possible protection for elderly among whom the highest mortality is observed. Even though not explicitly stated, this modelling approach would address emerging concerns from immunogenicity studies that one dose of mRNA vaccine may not induce sufficient immune response in elderly or immunocompromised. Furthermore, the impact where vaccination does not have any effect on transmission of infection was simulated.
The model used initial conditions corresponding to achieving 1% cumulative infection rate observed in the US and Europe prior to simulating the effects of vaccination strategies. The assumptions of vaccine effectiveness after the first dose (70%, 80% and 90% after 12 days post-first dose) were derived primarily from adjustment of data from BNT162b2 vaccine trial. These assumptions are reasonable and, in fact, can be further strengthened by including several post-marketing vaccine effectiveness studies from Israel and the United Kingdom, which showed similar ranges of vaccine effectiveness for the first dose. The examination of three different daily vaccination rates is a strong point for this paper. The assumptions used (0.1%, 0.3% and 1% of population) were based on calculation of daily vaccination rate in the US.
The results from this study demonstrated the reduction in mortality, cumulative infections, and hospitalizations for the delayed second dose strategy optimal for the first dose vaccine effectiveness at 80% or greater and the daily vaccination rate at 0.3% or lower. The results are consistent with the early analysis of data from the UK which implemented a delayed second dose of Covid-19 vaccine. The National Advisory Committee on Immunization (NACI) in Canada also recently recommended delaying the second dose of Covid-19 vaccines.
Of a particular interest to the readers and policy makers is that the beneficial effect of the delayed second dose of vaccination in this study is dependent upon daily vaccination rate, with a more rapid vaccination campaign limiting the benefits of a delayed second dose strategy. However, this needs to be viewed in the context of available vaccine supply, which can lead to unpredictable fluctuations of the daily vaccination rates. Under the scenario where individuals over 65 years of age were excluded from a delayed second dose strategy, the reduction in mortality was comparable to standard strategy at 1% daily vaccination rate. This suggests that the option of having a delayed second dose strategy only among younger individuals is a safer option but may still need to balanced against supply constraints. The fact that the model was not sensitive to sterilising versus non-sterilising immunity is somewhat surprising. Intuitively one would expect a better performance of a delayed second dose strategy under the sterilising immunity conditions. This requires further examination of the underlying model transmission dynamic.
The limitation of this study includes lack of description of calibration and/or validation of the model. As with many mathematical modelling studies, the validity of results produced by the model depends on whether the model simulated output reproduces real-world data in a given jurisdiction and time. This is particularly so for parameters which cannot be derived from other epidemiological or clinical studies. The readers should be able to draw their own conclusions on the validity metrics used to depict the transmission dynamic in the model, rather than relying on provided reference. Specifically, the quoted model (reference 15) was calibrated based on the early stages of pandemic (March - July 2020). Is it still appropriate in the current context? The other limitations are that the model did not investigate the effect of possible waning of immunity after the first dose, which may result in breakthrough infections and can affect conclusions, as well as the effect of different durations of delays after the first dose (e.g., 12 or 16 weeks). In the model, the authors assumed the persistence of the first-dose effectiveness for 180 days, the entire period which was simulated. Authors quoted emerging evidence for the AstraZeneca vaccine that the effectiveness of a second dose is higher after a longer interval after the first dose, which may warrant examining. Furthermore, any epidemiological shift where a larger number of younger individuals are hospitalized (which may occur with VOC), can possibly make a delayed second dose strategy more compelling. It would also be interesting to see if the results will be scalable to a larger population.
Overall, this study does provide policy makers with robust analysis on a range of scenarios, which they can apply or extrapolate to their respective jurisdictions to assess whether a delayed second dose strategy is appropriate for their epidemiological situation.