Authors’ response

Sir,

We thank the author of the letter for keen interest in our article1. The author notes a ‘discrepancy’ concerning the ‘N’ for vaccination status in Table 1 of the article. This disparity arises from the fact that out of the 577 cases reported having received any dose of the COVID-19 vaccine, only 288 had valid vaccination dates, enabling us to calculate the interval between the most recent vaccination and the date of death. As per the definition given in the article, the date of vaccination was crucial to categorize the interval between the date of vaccination and the date of death (for cases and controls). We direct the author to paragraph three of the limitations section where we elaborate on the same.

In our study, we explored various exposure factors, encompassing COVID-19 vaccination, hospitalization for COVID-19 and behavioural/lifestyle factors for unexplained sudden deaths. While acknowledging the superior status of cohort designs in the hierarchy of evidence, we opted for case–control design due to the unavailability of datasets to construct the cohorts of exposed and unexposed individuals, especially for COVID-19 vaccination and COVID-19 hospitalization, and the absence of outcomes in terms of sudden deaths in these cohorts. We believe that the case–control design was the most appropriate and practical and at the same time, a valid scientific approach for answering the question in hand, examining the multiple exposures for a rare outcome. Randomized controlled trial, as pointed out by the author, is scientifically and ethically inappropriate for the research question addressed in our article.

The coverage of COVID-19 vaccination in India during the study reference period (October 2021-March 2023) varied across the months; As of March 2023, the vaccination coverage was around 95 per cent for two doses (https://ourworldindata.org/covid-vaccinations) and not 100 per cent as alluded by the author. COVID-19 vaccination status of the cases and their matched controls (before the death of cases) corresponded to any time during October 2021 and March 2023 and not at the end of March 2023 as presumed by the author. Further, individual-level exposures for a public health measure such as vaccination are unlikely to be universal.

For the selection of controls, we considered neighbourhood controls based on the following reasons: (a) comparable opportunities with that of the case for being exposed to the risk factors for sudden death influenced by the locality of residence, including vaccination (b) similar probability for controls to be identified as cases due to comparable health-seeking behaviour for serious medical emergency owing to their residence in the same locality (c) operational ease of recruitment and (d) gaining statistical efficiency. These are the basic principles for the selection of controls2. We do not believe that this design would have resulted in overmatching. In fact, we found a significant negative effect of the vaccination on the occurrence of unexplained sudden death.