More efforts to explore the association between cirrhosis and COVID-19 mortality, and the association between NAFLD and severe COVID-19

To the Editor:

Financial support

None.

Conflict of interest

The authors declare no conflicts of interest.

Authors' contributions

Conception and design, data acquisition, analysis, and interpretation: Zheng Li, Yue Hu, and Qiang Li. Zheng Li and Yue Hu wrote the initial draft. Qiang Li revised the manuscript. All authors reviewed and approved the final manuscript.

Data availability statement

The data that support the findings of this study are all included in this article and the supplementary material.

With great interest, we contrastively read the two meta-analyses by Wang et al. ^{1 , 2} published in the Journal of Hepatology in September 2022 and October 2022, respectively. The authors conducted two congeneric meta-analyses to investigate the association between cirrhosis and COVID-19 mortality,¹ and the association between NAFLD and severe COVID-19,² respectively. One meta-analysis is consistent with the findings of the study by Marjot et al.,^{1 , 3} while the other meta-analysis is contrary to the findings of the study by Marjot et al..^{2 , 4} Therefore, the methodology quality of the meta-analysis should be further discussion. The work by Wang et al. ^{1 , 2} is significant to settle the controversies in the novel subjects noted in the EASL position paper by Marjot et al. ⁵ because of the advantages of meta-analysis.^{6 , 7} However, some methodological flaws in the two meta-analyses^{1 , 2} should be noted, especially for the meta-analysis contrary to the results of the study by Marjot et al..^{2 , 4} The comparative analysis of the two meta-analyses^{1 , 2} indicated that the inherent deficiencies deviating from methodological norms^{6 , 7} would result in the same potential biases, until these issues could be taken seriously. Addressing these deficiencies is necessary to avoid the general queries by the readers and promote wide citations of the article.

First, although the authors stated that their meta-analyses were performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines in the second paragraph, there is no statement regarding protocol and registration (item 5 of PRISMA) throughout both of the two articles.^{1 , 2 , 7} The protocol registration is a necessary reporting item for meta-analysis according to the PRISMA statement because the prospective registration could promote transparency and minimize the potential of bias.^{6 , 7}

Second, the subject heading (e.g. Medical Subject Headings (MeSH)) is a necessary approach to improve both sensitivity and precision of the search results (Table 1 ).⁶ A model of search strategy including the MeSH approach in PubMed is exhibited in Table 1. However, the authors used free-text only in their search strategy displayed in the second paragraph of each meta-analysis.^{1 , 2} In fact, some eligible and crucial studies were missed for the meta-analysis (such as the study by Bajaj et al. published in Gut),^{8 , 9} which has already resulted in a significant bias (Supplementary material).^{6 , 7 , 10} Furthermore, although the authors reported the test of publication bias in the fourth paragraph,^{1 , 2} the results for detecting publication bias are inaccurate and invalid because of the incomplete data by missing eligible included studies.^{6 , 8 , 9}

Table 1.

A model of search strategy including the MeSH approach in PubMed.

Step	The detailed search terms, approach and logical calculus	Category
#1	(a) The search for the first meta-analysis (https://doi.org/10.1016/j.jhep.2022.09.015:) "Liver Cirrhosis"[Mesh]	MeSH search
	(b) The search for the second meta-analysis (https://doi.org/10.1016/j.jhep.2022.10.009:) "Non-alcoholic Fatty Liver Disease"[Mesh]
#2	(a) The search for the first meta-analysis (https://doi.org/10.1016/j.jhep.2022.09.015:) cirrhosis[All Fields] OR “liver fibrosis”[All Fields] OR “Hepatic fibrosis”[All Fields]	Free-text search
	(b) The search for the second meta-analysis (https://doi.org/10.1016/j.jhep.2022.10.009:) “non-alcoholic fatty liver disease”[All Fields] OR NAFLD[All Fields] OR “metabolic associated fatty liver disease”[All Fields] OR MAFLD[All Fields] OR “Nonalcoholic Steatohepatitis”[All Fields] OR “Nonalcoholic Steatohepatitides”[All Fields]
#3	#1 OR #2	Logical OR
#4	"COVID-19"[Mesh] OR "SARS-CoV-2"[Mesh]	MeSH search
#5	“severe acute respiratory syndrome coronavirus 2”[Title/Abstract] OR SARS-CoV-2[Title/Abstract] OR “coronavirus disease 2019”[Title/Abstract] OR COVID-19[Title/Abstract] OR “2019 novel coronavirus”[Title/Abstract] OR 2019-nCoV[Title/Abstract]	Free-text search
#6	#4 OR #5	Logical OR
#7	(a) The search for the first meta-analysis (https://doi.org/10.1016/j.jhep.2022.09.015:) "Mortality"[Mesh] OR "Survival Analysis"[Mesh] OR "Survival Rate"[Mesh]	MeSH search
#8	(a) The search for the first meta-analysis (https://doi.org/10.1016/j.jhep.2022.09.015:) mortality[All Fields] OR death[All Fields] OR dead[All Fields] OR survival[All Fields]	Free-text search
#9	#7 OR #8	Logical OR
#10	#3 AND #6 AND #9 (a, Cirrhosis); #3 AND #6 (b, NAFLD)	Logical AND

MeSH, medical subject headings; NAFLD, non-alcoholic fatty liver disease; MAFLD, metabolic associated fatty liver disease.

Third, the seventeenth item (study selection) and the eighteenth item (study characteristics) of the PRISMA statement require the meta-analysis to present a detailed flow diagram of the studies selection, characteristics (such as sample size, PICOS, follow-up period) and citations of each included study.⁷ However, all of these necessary items could not be found in each meta-analysis.^{1 , 2 , 7} The authors should present these items in the supplementary material. At least, the included studies for meta-analysis should be listed in the form of references in the supplementary material because it is difficult for the readers to identify the included studies just by the author names in Figure 1 in the two articles.^{1 , 2}

Fourth, the risk of bias in individual studies (the items 12 and 19 of the PRISMA statement) is necessary for the meta-analysis reporting.⁷ However, there is no corresponding analysis in each meta-analysis.^{1 , 2 , 7} The authors should present detailed assessment data for within-study biases in the supplementary material. And a sensitivity analysis should be conducted based on the identified studies at lower risk of bias to test the robustness and reliability of the meta-analysis findings.^{6 , 7}

Fifth, the authors stated that “We included 29 articles including data on 6,872,587 individuals with COVID-19.”, and “22,056 cases” in the fourth paragraph.^{1 , 2} Actually, the real number of eligibility for the final meta-analysis is more valuable than the rough sampled population.^{6 , 10} Therefore, the authors should specify whether the huge number “6,872,587” and “22,056” (especially for “6,872,587”) are the total sample sizes for the actual meta-analyses, and list the individual sample size (involved in the meta-analysis) of each included study.^{6 , 10}

Sixth, the authors should specify which effect measure was chosen for data synthesized calculation in one meta-analysis¹ (https://doi.org/10.1016/j.jhep.2022.09.015) according to the item 13 (summary measures) of the PRISMA statement, just like their reporting of odds ratio (OR) in the other meta-analysis.^{1 , 2 , 7}

We congratulate Wang et al. for their breakthrough in the association between cirrhosis and COVID-19 mortality,¹ and the association between NAFLD and severe COVID-19.² Meanwhile, it is necessary to address the aforementioned deficiencies to provide more convincing evidence. As a key remediation, we have listed all the references for the included studies in our Supplementary material.

Appendix A. Supplementary data

The following is/are the supplementary data to this article: