Protection afforded by post-infection SARS-CoV-2 vaccine doses: a cohort study in Shanghai

Reviewer #1 (Public Review):

Summary:

Zheng and colleagues assessed the real-world efficacy of SARS-CoV-2 vaccination against re-infection following the large omicron wave in Shanghai in April 2022. The study was performed among previously vaccinated individuals. The study successfully documents a small but real added protective benefit of re-vaccination, though this diminishes in previously boosted individuals. Unsurprisingly, vaccine preventative efficacy was higher if the vaccine was given in the month before the 2nd large wave in Shanghai. The re-infection rate of 24% suggests that long-term anti-COVID immunity is very difficult to achieve. The conclusions are largely supported by the analyses. These results may be useful for planning the timing of subsequent vaccine rollouts.

Strengths:

The strengths of the study are a very large and unique cohort based on synchronously timed single infection among individuals with well-documented vaccine histories. Statistical analyses seem appropriate. As with any cohort study, there are potential confounders and the possibility of misclassification and the authors outline limitations nicely in the discussion.

Weaknesses:

(1) Partially and fully vaccinated are never defined and it is difficult to understand how this differs from single, and double, booster vaccines. The figures including all of these groups are a bit confusing for this reason.

(2) Figure 3 is a bit challenging to interpret because it is a bit atypical to compare each group to a different baseline (ie 2V-I-V vs 2V-I). I would label the y-axis 2V-I-V vs 2V-I (change all of the labels) to make this easier to understand.

(3) A 15% reduction in infection is quite low. It would be helpful to discuss if any quantitative or qualitative signals suggest at least a reduction in severe outcomes such as death, hospitalization, ER visits, or long COVID. I am not sure that a 15% reduction in cases supports extra vaccination without some other evidence of added benefit.

(4) Why exclude the 74962 unvaccinated from the analysis. it would be interesting to see if getting vaccinated post-infection provides benefits to this group

(5) Pudong should be defined for those who do not live in China.

(6) The discussion about healthcare utilization bias is welcomed and well done. It would be great to speculate on whether this bias might favor the null or alternative hypothesis.

Reviewer #2 (Public Review):

Summary:

This paper evaluates the effect of COVID-19 booster vaccination on reinfection in Shanghai, China among individuals who received primary COVID-19 vaccination followed by initial infection, during an Omicron wave.

Strengths:

A large database is collated from electronic vaccination and infection records. Nearly 200,000 individuals are included in the analysis and 24% became reinfected.

Weaknesses:

The article is difficult to follow in terms of the objectives and individuals included in various analyses. There appear to be important gaps in the analysis. The electronic data are limited in their ability to draw causal conclusions.

More detailed comments:

In multiple places (abstract, introduction), the authors frame the work in terms of understanding the benefit of booster vaccination among individuals with hybrid immunity (vaccination + infection). However, their analysis population does not completely align with this framing. As best as I can tell, only individuals who first received COVID-19 vaccination, and subsequently experienced infection, were included. Why the analysis does not also consider individuals who were infected and then vaccinated is not clear.

In vaccine effectiveness analyses, why was time since initial infection not examined as a modifier of the booster effect? Time since the onset of the Omicron wave is only loosely tied to the immune status of the individual.

The effect of booster vaccination on preventing symptomatic vs. asymptomatic reinfection does not appear to have been evaluated; this is a key gap in the analysis and it would seem the data would support it.

In lines 105-108, the demographic description of the analysis population is incomplete. Is sex or gender identity being described? Are any individuals non-binary? What is the age distribution? (Only the proportions 20-39 and under 6 are stated.)

Figure 1 consort diagram is confusing. In the last row, are the two boxes independent or overlapping sets of individuals? Are all included in secondary analyses?