We thank the authors for their work [1], which attempted to answer a fundamental question in the current management of surgical patients worldwide, and to quantify the risk of deciding to perform surgery on a patient previously infected with SARS-CoV-2. While it is fairly clear from a risk-benefit perspective that urgent and cancer surgery should be performed promptly whenever possible despite the current pandemic, a more difficult question is how to deal with patients requiring surgery that can be deferred. This question is becoming increasingly common as countries lift their restriction policies regarding planned surgery while the pandemic is brought under control. This study strongly suggests that non-essential surgical procedures should be postponed in patients with recent SARS-CoV-2 infection, including those without symptoms, in the interests of patient safety and not just because of a lack of healthcare resources.

However, we would like to point out a potential bias that does not seem to have been clearly controlled nor discussed in this study. Among the included patients with a recent infection (between 0 and 6 weeks, precisely when adjusted mortalities were highest), there was a majority of patients from low- and middle-income countries (from 58.6% to 65.5%), whereas the inverse was observed for patients without infection or with an older infection (from 34.3% to 42.1%), which could have led to some excess mortality in recently infected patients. This might be supported by the observation that living in a low- and middle-income country was significantly associated with higher mortality in the unadjusted analysis. Although the authors used the country income as a covariate in the logistic regression models to adjust for mortality, it would have been more appropriate to use a mixed model to separate the random effects of country income levels from the fixed effects related to patients’ conditions and their surgical procedures. This would have also provided control over possible interactions between the effects of some pre-existing conditions on mortality and the national income level, since it can be hypothesised that some factors, such as age, may influence mortality differently depending on the country income. This is of particular concern as there appears to be an 'ecological fallacy' when looking at the aggregate data for the COVID-19 pandemic; while high-income countries seem to have a higher case fatality rate than low- and middle-income countries at first sight [2], the individual data suggest the opposite, with higher case fatality rate among lower-income people [3, 4]. Therefore, it does not seem appropriate to use country income as a characteristic of an individual to be used for fixed effect. It would be interesting to know whether the effect observed by the authors was consistent across country income levels, by providing a sensitivity analysis using this covariate.

我们感谢作者的工作 [ 1]，它试图回答当前全球外科患者管理中的一个基本问题，并量化决定对先前感染 SARS-CoV-2 的患者进行手术的风险。尽管从风险收益的角度来看，尽管当前大流行，但应尽可能及时地进行紧急和癌症手术是相当清楚的，但更困难的问题是如何处理需要可以推迟手术的患者。随着各国在大流行得到控制的情况下取消对计划手术的限制政策，这个问题变得越来越普遍。这项研究强烈建议，对于近期感染 SARS-CoV-2 的患者，包括没有症状的患者，应推迟非必要的外科手术，

然而，我们想指出一个潜在的偏见，在本研究中似乎没有得到明确的控制或讨论。在近期感染的患者中（0 至 6 周，恰好是调整后死亡率最高的时间），大多数患者来自低收入和中等收入国家（从 58.6% 到 65.5%），而相反的是在未感染或感染较旧的患者（从 34.3% 到 42.1%）中观察到，这可能导致最近感染患者的死亡率过高。这可能得到以下观察结果的支持：在未经调整的分析中，生活在低收入和中等收入国家与较高的死亡率显着相关。尽管作者在逻辑回归模型中使用国家收入作为协变量来调整死亡率，使用混合模型将国家收入水平的随机效应与与患者状况及其外科手术相关的固定效应分开会更合适。这也可以控制某些预先存在的条件对死亡率和国民收入水平的影响之间可能存在的相互作用，因为可以假设某些因素，例如年龄，可能会根据国家收入不同地影响死亡率。这尤其令人担忧，因为在查看 COVID-19 大流行的汇总数据时似乎存在“生态谬误”；虽然乍一看，高收入国家的病死率似乎高于低收入和中等收入国家[ 这也可以控制某些预先存在的条件对死亡率和国民收入水平的影响之间可能存在的相互作用，因为可以假设某些因素，例如年龄，可能会根据国家收入不同地影响死亡率。这尤其令人担忧，因为在查看 COVID-19 大流行的汇总数据时似乎存在“生态谬误”；虽然乍一看，高收入国家的病死率似乎高于低收入和中等收入国家[ 这也可以控制某些预先存在的条件对死亡率和国民收入水平的影响之间可能存在的相互作用，因为可以假设某些因素，例如年龄，可能会根据国家收入不同地影响死亡率。这尤其令人担忧，因为在查看 COVID-19 大流行的汇总数据时似乎存在“生态谬误”；虽然乍一看，高收入国家的病死率似乎高于低收入和中等收入国家[ 这尤其令人担忧，因为在查看 COVID-19 大流行的汇总数据时似乎存在“生态谬误”；虽然乍一看，高收入国家的病死率似乎高于低收入和中等收入国家[ 这尤其令人担忧，因为在查看 COVID-19 大流行的汇总数据时似乎存在“生态谬误”；虽然乍一看，高收入国家的病死率似乎高于低收入和中等收入国家[2 ]，个人数据表明相反，低收入人群的病死率较高 [ 3, 4 ]。因此，将国家收入作为个人特征用于固定效应似乎并不合适。通过使用该协变量进行敏感性分析，了解作者观察到的效果在不同国家收入水平上是否一致将会很有趣。