COVID-19 Severity and Mortality Among Chronic Liver Disease Patients

A Systematic Review and Meta-Analysis

Ramya Nagarajan, MD; Yuvaraj Krishnamoorthy, MD; Sathish Rajaa, MD; Vishnu Shankar Hariharan, MD

Disclosures

Prev Chronic Dis. 2022;19(8):e53 

In This Article

Results

We found 3,659 records through the systematic literature search and deemed 221 of those studies relevant for full-text retrieval. We also retrieved the full text for 36 articles obtained through manual searching of the bibliographies in the retrieved studies. During the second screening stage, 40 studies with 908,032 participants met the eligibility criteria and were included in the analysis (Figure 1) (8–12,15,19–52). This study was reported as per the PRISMA statement guidelines (Supplementary Table 2 available at: https://drive.google.com/drive/folders/1mVlexUbFzmHcfvi44LTFi18OmTnMZXtT?usp=sharing).

Figure 1.

RISMA flowchart showing the identification of studies for analysis of the association of chronic liver disease with severity of disease and mortality among COVID-19 patients. Abbreviation: PRISMA, Preferred Items for Systematic Reviews and Meta-Analyses.

In total, 909,831 participants were found in the included studies, with a sample size ranging from 41 to 259,110 (Table). Among the 40 studies included, 15 reported on mortality due to COVID-19, 14 reported on the severity of COVID-19, and 11 reported both on severity and mortality. All included studies were retrospective; most studies were conducted in China (n = 14) and the US (n = 10). Half (21 of 40) of the included studies were low-quality (ie, per the Newcastle-Ottawa Scale) (Supplementary Table 3 available at: https://drive.google.com/drive/folders/1mVlexUbFzmHcfvi44LTFi18OmTnMZXtT?usp=sharing).[16]

Association Between CLD and COVID-19 Outcomes

Severity. In our analysis, 25 studies reported the severity of the CLD and the non-CLD groups.[9,10,19–21,23,24,27–29,31,32,34,36,37,40–44,46,49–52] The pooled OR was 2.44 (95% CI, 1.89–3.16; I 2 = 91.3%; 95% PI, 0.79–7.55) (Figure 2), indicating that the odds of developing severe disease among COVID-19 patients with CLD were 2.44 times higher than among those without CLD. High heterogeneity was found between the studies reporting the severity outcome (I 2 = 91.3%, P < .001).

Figure 2.

Forest plot showing the difference in severity between COVID-19 patients with and without CLD (N = 27). Abbreviation: CLD, chronic liver disease.

Subgroup analysis showed that COVID-19 patients with NAFLD had the highest odds of COVID-19 severity (pooled OR = 5.60; 95% CI, 1.52–20.64), followed by MAFLD (pooled OR = 3.20; 95% CI, 1.99–5.14) and cirrhosis (pooled OR = 3.09; 95% CI, 1.95–4.89) (Supplementary Figure 1 available at: https://drive.google.com/drive/folders/1mVlexUbFzmHcfvi44LTFi18OmTnMZXtT?usp=sharing). COVID-19 patients who had viral hepatitis did not have significantly higher odds of having a severe form of COVID-19 (pooled OR = 1.29; 95% CI, 0.36–4.63). Subgroup analysis by study design showed significantly higher odds of severity in the studies following cohort design (pooled OR = 3.10; 95% CI, 2.08–6.60; P = .001) (Supplementary Figure 2 available at: https://drive.google.com/drive/folders/1mVlexUbFzmHcfvi44LTFi18OmTnMZXtT?usp=sharing).

Results of the univariable meta-regression showed that geographic region, type of CLD, quality of study, year of publication, sample size, and mean age of participants were not significantly associated with the pooled effect size and cannot explain the substantial heterogeneity in the results (Supplementary Table 4 available at: https://drive.google.com/drive/folders/1mVlexUbFzmHcfvi44LTFi18OmTnMZXtT?usp=sharing).

Publication bias was graphically checked by funnel plot and Doi plot (Supplementary Figures 3 and 4 available at: https://drive.google.com/drive/folders/1mVlexUbFzmHcfvi44LTFi18OmTnMZXtT?usp=sharing). The funnel plot showed no sign of asymmetry, and it was also statistically proved by Egger test (P = .36); the Doi plot also showed no asymmetry, with an LFK index of 0.93. Sensitivity analysis showed no significant variation in the magnitude or direction of the outcome, indicating a lack of influence of a single study on the overall pooled estimate (Supplementary Figure 5 available at: https://drive.google.com/drive/folders/1mVlexUbFzmHcfvi44LTFi18OmTnMZXtT?usp=sharing).

Mortality. In total, 26 studies reported on the mortality outcome among CLD and non-CLD patients.[8,10–12,15,19,20,22,24–26,29,30,33,35,38,39,41,44–51] The pooled OR was 2.35 (95% CI, 1.84–3.00; I 2 = 96.26%; 95% PI, 0.76–7.18) (Figure 3), indicating that COVID-19 patients with CLD had 2.35 times higher odds of dying as patients without CLD. We found substantial heterogeneity between the studies reporting the mortality outcome (I 2 = 96.3%, P < .001).

Figure 3.

Forest plot showing the difference in mortality between COVID-19 patients with and without chronic liver disease (CLD) (N = 27).

Subgroup analysis based on the type of CLD could not be done because only cirrhosis had enough studies to give a pooled estimate (all the other studies reporting mortality outcomes were conducted among CLD patients without categorizing them based on the type of CLD). We found that COVID-19 patients with cirrhosis had 3.51 times higher odds of dying as patients without cirrhosis (pooled OR = 3.51; 95% CI, 2.41–5.10) (Supplementary Figure 6 available at: https://drive.google.com/drive/folders/1mVlexUbFzmHcfvi44LTFi18OmTnMZXtT?usp=sharing). Subgroup analysis by study design showed significantly higher odds of severity among the studies conducted using a cohort (pooled OR = 2.94; 95% CI, 2.09–4.13; P < .001) and a retrospective cohort design (pooled OR = 2.19; 95% CI, 1.51–3.17; P < .001) (Supplementary Figure 7 available at: https://drive.google.com/drive/folders/1mVlexUbFzmHcfvi44LTFi18OmTnMZXtT?usp=sharing).

Univariable meta-regression showed that only the mean age of the patients had a significant association with the pooled effect size (P = .01) and explained 48.3% of the between-study variability (Supplementary Figure 8 available at: https://drive.google.com/drive/folders/1mVlexUbFzmHcfvi44LTFi18OmTnMZXtT?usp=sharing). None of the other factors were significantly associated with the pooled effect size and cannot explain the substantial heterogeneity in the results (Supplementary Table 5 available at: https://drive.google.com/drive/folders/1mVlexUbFzmHcfvi44LTFi18OmTnMZXtT?usp=sharing).

Publication bias was graphically checked by funnel plot and Doi plot (Supplementary Figures 9 and 10 available at: https://drive.google.com/drive/folders/1mVlexUbFzmHcfvi44LTFi18OmTnMZXtT?usp=sharing). The funnel plot showed signs of asymmetry, with the Egger test (P = .10) also showing signs of possible publication bias. The Doi plot showed significant asymmetry, with an LFK index of 4.47. Sensitivity analysis showed no significant variation in the magnitude or direction of the outcome, indicating a lack of influence of a single study on the overall pooled estimate (Supplementary Figure 11 available at: https://drive.google.com/drive/folders/1mVlexUbFzmHcfvi44LTFi18OmTnMZXtT?usp=sharing).

processing....