RR:C19 Evidence Scale rating by reviewer:
Not informative. The flaws in the data and methods in this study are sufficiently serious that they do not substantially justify the claims made. It is not possible to say whether the results and conclusions would match that of the hypothetical ideal study. The study should not be considered as evidence by decision-makers.
This analysis used a stochastic SARS-CoV-2 transmission model to investigate how stringent non-pharmaceutical interventions (NPIs) must be, and for how long they must be implemented as vaccination is rolled out, to main control of COVID-19. The authors parameterised the model to use China as a case study and concluded that NPIs will need to remain in place for at least a year after the start of vaccination in order to prevent major, widespread epidemics.
This is evidently an important research question, but I don’t believe the structure of the model employed is sufficiently nuanced to produce reliable results. The model is age-structured, but there is no spatial stratification. In a mass action SIR model, the recipients of infection transmission are randomly chosen from the population, as if the entire population of China were mixing freely. This produces the extraordinary rise in infections shown in Figure 1A and C, but this timescale does not seem reasonable. This paper is investigating whether a vaccination campaign can keep up, and overtake, the pace of infection spread, and so these timescales are vital. Furthermore, zero vaccination at time t=0 in a totally susceptible population, as we are informed by authors is the situation in China, would surely lead to some kind of ring vaccination policy – vaccinating those in the locality of the outbreaks. This cannot be investigated with the current model structure.
I would also question the use of an SIR model for SARS-CoV-2 as there is plenty of evidence of waning immunity (but this may make little difference in the current analysis, where the timescale under evaluation is so short).
I would encourage the authors to provide more information on their model in the main text and cut down the Discussion section. It is more important for the reader to understand how and why the model is producing the displayed output and more clarification is needed (for example, it appears that vaccine efficacy refers to efficacy in preventing infections, but this should be made clear, and whether there is additional efficacy in terms of reducing disease severity). For journal publication, the Introduction could be updated as the situation has rapidly changed since the manuscript was posted.