Reply.

We thank Gao and Xu for their interest in our study. We would like to address each of their comments.

Gao and Xu first propose performing multiple models to prove the robustness of the result in our multivariate regression analysis. We agree that this is an insightful comment for our study. To demonstrate the robustness of our conclusion, we performed the leave-one-out analysis. The association between nonalcoholic fatty liver disease (NAFLD) and severe coronavirus disease 2019 (COVID-19) persisted to be mostly insignificant (P > .05). However, the models without body mass index (BMI) or cirrhosis were both significant (P < .05). The conclusion drawn from this sensitivity analysis was consistent with that in our article.

We believe that the missing decimal points in Supplementary Table 9 are typos induced during the press production procedure. We will work closely with the production team and editors to correct these errors.

In addition, Gao and Xu recommend multivariable Mendelian randomization (MVMR) analysis. The problem of MVMR is its limited statistical power. Nevertheless, we performed MVMR to assess the causal effect of NAFLD on severe COVID-19 while controlling for BMI and C-reactive protein (CRP). Similar to 2-sample Mendelian randomization, the result showed no evidence of a causal effect of NAFLD (odds ratio, 1.02; 95% confidence interval,  0.93–1.11; P = .68). In fact, the F statistics for BMI, CRP, and NAFLD decreased from 293.47, 580.61, and 16572.45 in the single-variate model to 20.71, 89.47, and 9.49 in the multiple-variate model, respectively. Therefore, we disagree that the current study would benefit from an MVMR analysis.

Finally, elevated CRP level could be a result of severe COVID-19, rather than a cause. The provided references observed the association, not the causality. Our article aimed to discover causal factors, rather than sequelae. Of course, it is also possible that our finding is only limited to the sample sets we studied.

Again, we appreciate the authors for pointing out the problems and divergences in our article and giving us this opportunity to make clarifications.