Non-alcoholic fatty liver disease (NAFLD) and COVID-19 outcomes: A systematic review, meta-analysis, and meta-regression

Abstract

It is important to identify risk factors for poor outcomes of coronavirus disease 2019 (COVID-19) patients. Currently, the correlation between non-alcoholic fatty liver disease (NAFLD) and COVID-19 outcomes has not been established. This study was conducted to determine the association between NAFLD and in-hospital outcomes of COVID-19 patients. The systematic searches were conducted by using PubMed and the Europe PMC databases and particular keywords were used as of December 10, 2020. Further searches were conducted up to 2022. All articles that include data about COVID-19 and fatty liver disease were collected. Statistical analysis was performed by using Review Manager 5.4 and Comprehensive Meta-Analysis version 3 software. A total of 7,210 COVID-19 patients from 18 studies were included in the final analysis. Meta-analysis revealed that NAFLD increased the risk of developing poor in-hospital outcome (pooled both severe disease and death) in COVID-19 patients (RR 1.42; 95%CI: 1.17–1.73, p<0.001, I²=84%, random-effect modeling). Subgroup analysis however found that having NAFLD only increased the chance of getting severe COVID-19 (RR 1.67; 95%CI: 1.32–2.13, p<0.001, I²=86%, random-effect modeling) and not mortality (RR 1.00; 95%CI: 0.68–1.47, p=0.98, I²=80%, random-effect modeling). Meta-regression suggested that age (p=0.001) and diabetes (p=0.029) were significantly influenced the relationship between NAFLD and in-hospital outcomes of COVID-19 (pooled both severe disease and mortality). The weaker association of NAFLD and in-hospital outcomes of COVID-19 was found for studies with median age ≥45 years old (RR 1.29) when compared to studies with median age <45 years old (RR 2.96). In addition, studies with the prevalence of diabetes ≥25% (RR 1.29) had a weaker association with in-hospital outcomes when compared to studies with diabetes prevalence <25% (RR 1.85). In conclusion, NAFLD increased the risk of chance of getting severe COVID-19 and therefore it should be evaluated closely to reduce the chance of getting severe COVID-19.

Introduction

The coronavirus disease 2019 (COVID-19) is a significant health problem for almost every country around the world. The disease has impacted several aspects of human live such as health, society, and economics. Therefore, identification of the risk factor for poor outcomes from this disease is utmost important, enabling the risk stratifications and optimization of the allocations for hospital resources. The manifestation of this disease is diverse from mild respiratory symptoms to serious life-compromising infirmities such as respiratory failure, shock, arrhythmia, sepsis, and multi-organ failure [1, 2]. Up until now, comorbid conditions such as hypertension, diabetes, dyslipidemia, HIV infection, cardiovascular diseases, dementia, and thyroid diseases are correlated with the development of poor outcomes from COVID-19 [3-9]. A previous study [10] found that non-alcoholic fatty liver disease (NAFLD) is associated with higher 30-day all-cause mortality in patients with community-acquired pneumonia (CAP). Nonetheless, the correlation between NAFLD and COVID-19 is yet to be established. The aim of this study was to determine the potential association between NAFLD and in-hospital outcomes of COVID-19 patient through a systematic review and meta-analysis.

Methods

Eligibility criteria

Articles incorporated in this study were selected if complied with the population, intervention, comparison and outcomes (PICO) framework (P: COVID-19 positive patients; I: non-alcoholic fatty liver disease (NAFLD) as patients' comorbidity; C: patients who do not have NAFLD; and O: in-hospital outcomes (severe COVID-19 and mortality). All randomized clinical trial, observational studies (cohort or case-control), and cross-over studies were considered eligible. Any studies besides original research (e.g., review articles, letters, or commentaries); case reports; studies reported other than in the English language; studies focusing on the populations of young age (below 18 years old) and women during their pregnancy were excluded.

Search strategy and study selection

The articles were explored systemically and obtained from PubMed and Europe PMC database. Search terms applied were: ("fatty liver disease" OR "clinical characteristic" OR "risk factor") AND ("coronavirus disease 2019" OR "COVID-19"); restricted from 2019 until December 10, 2020 and written in English. The titles and abstracts were first assessed on all articles obtained through literature searching process. The process was then followed by full text screening to determine the suitability with our eligibility criteria. Additional searchers for potential eligible articles were conducted by analyzing the papers cited by authors of all identified studies. The search strategy was presented in the PRISMA diagram [11].

Data extraction and quality assessment

The extraction process was conducted by two authors and an extraction form was developed to list the essential information on authors, year, study design, number of participants, age, gender, diabetes, NAFLD, dyslipidemia, body mass index (BMI), severe COVID-19, and mortality from COVID-19. The outcome of interest was in-hospital results, which comprised of severe COVID-19 and mortality. The clinical characteristic criteria (at the time of, or after, the admission) for severe COVID-19 followed the WHO criteria; if the patient had any of the followings: (1) respiratory distress (≥30 breaths per min); (2) oxygen saturation ≤93% at rest; (3) proportion of the partial pressure of arterial oxygen (PaO2) to a fractional concentration of oxygen inspired air (FiO2) ≤300 mmHg; or (4) critical complication (respiratory failure, septic shock, and or multiple organ dysfunction/failure) or intensive care unit (ICU) admission based on WHO criteria. Mortality was established by the proportion of patients' death caused by COVID-19.

The studies' quality was evaluated independently using the Newcastle–Ottawa Scale (NOS) [12]. Two authors marked scores regarding selection, comparability, and exposure in each study from zero to nine with a good-quality cut off of 7.

Statistical analysis

A meta-analysis was conducted using Review Manager 5.4 (Cochrane Collaboration) and Comprehensive Meta-Analysis version 3 software. To calculate dichotomous variables, Mantel-Haenszel's formula with random-effects models was applied. Heterogeneity assessment implies low, moderate, and high degrees by employing I² statistics with the content of <25%, 26-50%, and >50% [13]. The risk ratio (RR) along with its 95% confidence intervals (CIs) was used to assess the association between fatty liver disease and (1) disease severity of COVID-19; (2) mortality of COVID-19; and (3) poor COVID-19 outcome (both severity and mortality) in the meta-analysis. The results were considered significant when the two-tailed p-value was ≤0.05. Random-effects meta-regression was achieved using a restricted-maximum likelihood toward pre-specified variables including age, diabetes, and dyslipidemia. The qualitative risk of publication bias was assessed with Begg's funnel plot analysis.

Results

Study selection and characteristics

There were 4,574 records were obtained by systematic electronic searches and 3,358 remained after removal 1216 duplicates between databases. After reading the titles or abstracts, a total of 3,310 articles were excluded, leaving 48 full texts for eligibility. Absence of interest (severe COVID-19 and mortality), having no a control/comparison group and written in non-English with a total of 17, 10, and 3 sequentially were excluded. Ultimately, 7,210 COVID-19 patients within 18 studies [14-31] were included in the meta-analysis (Figure 1). Within these studies, 17 were retrospective cohorts [14-21, 23-31] and one was a case-control study [22]. The fundamental aspects of the studies are compiled in Table 1.

Figure 1.

Preferred reporting items for systematic reviews and meta-analyses (PRISMA) flowchart of the study

Table 1. Characteristics of included studies (n=18).

Study (Year 2020)	Sample size	Design	Overall age (median)	Male (%)	Diabetes (%)	NAFLD (%)	Dyslipidemia (%)	BMI (mean)
Cai et al. [14]	298	Retrospective cohort	47.5 (62.5 vs 41)	48.6 (67.2 vs 44.1)	6 (13.7 vs 4.1)	4.7 (10.3 vs 3.3)	N/A	23.2 (24.5 vs 22.9)
Chen et al. [15]	342	Retrospective cohort	63 (58.5 vs 66.5)	53.5 (50 vs 57.3)	43.3 (48.3 vs 37.8)	52 (56.7 vs 51.2)	47 (46.6 vs 47.6)	30 (34.7 vs 26.6)
Forlano et al. [16]	193	Retrospective cohort	64.2 (61 vs 70.5)	62.6 (60 vs 64)	39.3 (47 vs 35)	31.6 (18 vs 20)	23.8 (23 vs 24)	28.4 (30.6 vs 27.1)
Gao et al. [17]	130	Retrospective cohort	46 (47 vs 46)	63.1 (67.1 vs 50.3)	N/A	50 (77.2 vs 44.4)	64.6 (75.4 vs 53.8)	25 (26.2 vs 23.7)
Hashemi et al. [18]	363	Retrospective cohort	63.4 (64.8 vs 63)	55.4 (53.6 vs 55.8)	32.2 (40.6 vs 30.4)	15.1 (16.4 vs 13.2)	46.5 (46.4 vs 46.6)	30.3 (32 vs 29.9)
Huang et al. [20]	280	Retrospective cohort	43 (43.5 vs 42.5)	52.1 (58.1 vs 49.5)	7.5 (11.6 vs 5.7)	30.7 (14 vs 8.2)	N/A	24.2 (27.1 vs 23.1)
Ji et al. [20]	202	Retrospective cohort	44.5 (55.1 vs 42.9)	55.9 (69.2 vs 52.8)	N/A	37.6 (87.2 vs 25.8)	N/A	24 (26.6 vs 23.4)
Kim et al. [21]	867	Retrospective cohort	56.9 (59.8 vs 52.1)	54.7 (57.6 vs 49.5)	42.9 (48.4 vs 34.2)	52.6 (47.9 vs 61.8)	38.6 (40.8 vs 35.1)	N/A
Mahamid et al. [22]	71	Case-control	51 (53.7 vs 56.2)	27.2 (31.8 vs 32.7)	29.5 (36.3 vs 38.8)	30.9 (36.3 vs 10.2)	25.3 (40.1 vs 18.3)	25 (29.2 vs 26.1)
Mushtaq et al. [23]	589	Retrospective cohort	45.8 (47.7 vs 44.5)	84.7 (80.8 vs 89.6)	39.8 (50.3 vs 27.7)	54.3 (60.6 vs 39.3)	N/A	28.4 (30.7 vs 25.6)
Rabiee et al. [24]	112	Retrospective cohort	61 (63.4 vs 59.7)	54.5 (43.3 vs 33.3)	45.5 (60 vs 27.1)	14.3 (33.3 vs 6.1)	20.5 (30 vs 14.8)	N/A
Sun et al. [25]	63	Retrospective cohort	47 (59 vs 47)	58.7 (60 vs 52.8)	7.9 (15.7 vs 4.5)	1.5 (5.2 vs 0)	N/A	24.4 (25.9 vs 24)
Targher et al. [26]	310	Retrospective cohort	48.3 (59.9 vs 45.9)	54.8 (57.1 vs 43.5)	12.5 (28.6 vs 7.4)	30.3 (50 vs 50)	N/A	24 (26.1 vs 23)
Wang et al. [27]	45	Retrospective cohort	39 (43 vs 38)	51.1 (60 vs 48.6)	9.0 (20 vs 5.7)	6.7 (0 vs 8.6)	N/A	N/A
Wargny et al. [28]	2796	Retrospective cohort	69.7 (69.6 vs 69.7)	63.7 (70.2 vs 61.1)	100 (28.6 vs 71.3)	8.3 (8.7 vs 8.1)	46.8 (49 vs 45.9)	28.4 (29.1 vs 28.1)
Wu et al. [29]	299	Retrospective cohort	50 (62 vs 43)	45.8 (59.2 vs 41.7)	12.3 (25.4 vs 8.3)	13.3 (16.9 vs 12.3)	5.3 (7 vs 4.8)	N/A
Zhang et al. [30]	140	Retrospective cohort	57 (64 vs 51.5)	50.7 (56.9 vs 46.3)	12.1 (13.8 vs 11)	5.7 (6.9 vs 5)	5 (3.4 vs 6.1)	N/A
Zhou et al. [31]	110	Retrospective cohort	42.1 (43.4 vs 40.9)	74.5 (81.8 vs 67.3)	11.8 (20 vs 3.6)	50 (73.9 vs 43.6)	61.8 (81.8 vs 41.8)	25.6 (26.1 vs 25)

BMI: body mass index; NA: not available; NAFLD: non-alcoholic fatty liver disease; vs: NAFLD group vs non-NAFLD group

Quality of study assessment

We assess the observational studies’ quality by using the Newcastle Ottawa Scales (NOS) for cohort and case-control design (Table 2). All studies were classified as 'good' and therefore included in the meta-analysis.

Table 2. Newcastle-Ottawa quality assessment of observational studies (n=18).

First author (2020)	Study design	Selection	Comparability	Outcome	Total score	Result
Cai et al. [14]	Retrospective cohort	***	**	***	8	Good
Chen et al. [15]	Retrospective cohort	***	**	***	8	Good
Forlano et al. [16]	Retrospective cohort	***	**	***	8	Good
Gao et al. [17]	Retrospective cohort	***	**	**	7	Good
Hashemi et al. [18]	Retrospective cohort	***	**	***	8	Good
Huang et al. [20]	Retrospective cohort	***	**	***	8	Good
Ji et al. [20]	Retrospective cohort	**	**	***	7	Good
Kim et al. [21]	Retrospective cohort	***	**	****	9	Good
Mahamid et al. [22]	Case-control	***	**	***	8	Good
Mushtaq et al. [23]	Retrospective cohort	**	**	***	7	Good
Rabiee et al. [24]	Retrospective cohort	***	**	***	8	Good
Sun et al. [25]	Retrospective cohort	**	**	***	7	Good
Targher et al. [26]	Retrospective cohort	**	**	***	7	Good
Wang et al. [27]	Retrospective cohort	***	**	***	8	Good
Wargny et al. [28]	Retrospective cohort	****	**	***	9	Good
Wu et al. [29]	Retrospective cohort	****	**	***	9	Good
Zhang et al. [30]	Retrospective cohort	***	**	***	8	Good
Zhou et al. [31]	Retrospective cohort	***	**	***	8	Good

NAFLD and COVID-19 outcomes

The pooled analysis showed a significant association between NAFLD and poor in-hospital outcome of COVID-19 (severe disease and mortality), with high heterogeneity (RR 1.42; 95%CI: 1.17–1.73, p<0.001, I²=84%, random-effect modeling) (Figure 2). On the subgroup analysis, patients with NAFLD had a higher chance of getting severe COVID-19 (RR 1.67; 95%CI: 1.32–2.13, p<0.001, I²=86%, random-effect model) but not mortality (RR 1.00; 95%CI: 0.68–1.47, p=0.98, I²=80%, random-effect model).

Figure 2.

Forest plot demonstrates the association of the non-alcoholic fatty liver disease (NAFLD) with in-hospital outcomes of coronavirus disease 2019 (COVID-19), which comprised of severe COVID-19 and mortality.

Meta-regression

Our meta-regression found that the association of NAFLD and in-hospital outcomes (severity and mortality) of COVID-19 patients was affected by age (p=0.001) (Figure 3A) and diabetes (p=0.029) (Figure 3B). Dyslipidemia (p=0.530) did not affect the association of NAFLD and in-hospital outcomes (Figure 3C).

Figure 3.

Bubble-plot for Meta-regression and Begg's funnel-plot analysis. Meta-regression analysis showed that the association between non-alcoholic fatty liver disease (NAFLD) and in-hospital outcomes of coronavirus disease 2019 (COVID-19) was affected by age (A) and diabetes (B), but not by dyslipidemia (C). Begg's funnel-plot analysis showed a qualitatively symmetrical inverted funnel-plot for the association between NAFLD and in-hospital outcomes of COVID-19 (severity and mortality) (D).

Subgroup analysis

The RR of COVID-19 in-hospital outcomes in the studies with median age of ≥45 years was lower compared to those studies with median age of <45 years (RR 1.29; 95%CI: 1.07–1.55), p=0.009, I²=82%, random-effect model vs RR 2.96; 95%CI: 1.03–8.57, p=0.004, I²=77%, random-effect model, respectively). Moreover, a lower RR has also been shown by studies with the prevalence of diabetes ≥25% when compared to studies with the prevalence of diabetes <25% (RR 1.29; 95%CI: 1.03–1.61, p=0.02, I²=87%, random-effect model vs RR 1.85; 95%CI: 1.45–2.37, p<0.001, I²=0%, random-effect modeling, respectively).

Publication bias

Funnel-plot analysis shows symmetrical inverted relation qualitatively linking NAFLD and COVID-19 in-hospital outcomes, noting no implication of publication bias (Figure 3D).

Discussion

This paper does not solely analyze the identification between NAFLD and in-hospital outcomes of COVID-19 but further develops the importance of confounding factors for instance age, gender, along comorbid conditions. These factors may also ascertain the connection to severe outcomes. Previously published studies indistinctly mention fatty liver disease's impact on the outcomes. It only shows in a form of an article review wherein liver injury in COVID-19 patients. This result was barely according to a single observational study, which shows conflicting results [32, 33].

These 18 comprehensive meta-analysis studies show the relation of NAFLD with poor in-hospital outcomes of COVID-19. There are some explanations between this relation. First, NAFLD is related to extra-hepatic signs of metabolic syndrome while obesity, hypertension, diabetes, and dyslipidemia included in metabolic syndrome itself are linked to higher severity and mortality rate of COVID-19 [34]. Concerning the severity of COVID-19 in patients with NAFLD may be caused by the relation to metabolic syndrome as mentioned before. Second, in normal conditions, the cholangiocytes and hepatocytes expressed ACE2 in low level. Nevertheless, in chronic liver damage, including NAFLD, the level of ACE2 has been shown to increase [32, 35]. Thus, increase levels of ACE2 in NAFLD patients also increase the infectivity of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and predispose them to develop severe outcomes. Finally, a lower percentage of Bacteroides and higher levels of Prevotella and Porphyromonas species leads to the development of gut dysbiosis in patients with NAFLD [34]. Some experimental and clinical studies proposed that through the gut-lung axis, gut microbiota represents an essential role in sepsis and ARDS pathogenesis. Furthermore, gut microbiota also operates the regulation of immune response, preferably through T-regulatory cells [36, 37]. Accordingly, gut dysbiosis will develop into severe outcomes such as sepsis and ARDS.

To increase the relevance of the results from our study, we have performed additional searching process for studies which analyzed the relationship between NAFLD and COVID-19 outcomes (severity and mortality). We have found seven additional studies that published between 2021–2022 (Table 3) [38-44]. Out of seven studies, four of them were retrospective cohort, two were prospective cohort, and the remaining one study was case-control study. These studies showed conflicting results regarding the relationship between NAFLD and severity or mortality outcomes in patients with COVID-19. From a total of six studies that reported the severe COVID-19 outcome, three studies did not show any significant difference between patients with NAFLD and those without NAFLD, while three other studies showed that NAFLD was associated with higher severity of COVID-19 when compared to non-NAFLD group. In terms of mortality outcome, all four studies that reported this outcome showed no significant difference in the proportion of patients who were died during the follow-up period between NAFLD and non-NAFLD group. These conflicting results indicate that further studies are still required to confirm the relationship between NAFLD and COVID-19 outcomes.

Table 3. Additional studies that analyzed relationship between non-alcoholic fatty liver disease (NAFLD) and COVID-19 outcomes (n=7).

Study	Sample size	Design	Outcome Results (Severe COVID-19 and mortality outcomes)
Madan et al. [38] 2022	446	Case control	No significant difference in the mortality rate between NAFLD and non-NAFLD (13.24% vs 13.81%; p=0.866)
			No significant difference in the ICU requirements between NAFLD and non-NAFLD (32.52% vs 39.49%; p=0.752)
			No significant difference in the ventilatory support between NAFLD and non-NAFLD (9.34% vs 8.91%; p=0.385)
Nath et al. [39] 2022	3933	Prospective cohort	The mortality rate was comparable between NAFLD and non-NAFLD group (6.7% vs 6%; p=0.381).
			In the multivariate analysis, NAFLD was not associated with mortality in COVID-19 patients (OR 1.149; 95% CI:
			0.669–1.976, p=0.614).
Okuhama et al. [40] 2022	222	Retrospective cohort	Patients with severe COVID-19 had higher prevalence of fatty liver (OR 6.33; 95% CI: 3.37–12.14, p<0.001).
			In the multivariate analysis, fatty liver on CT-scan was significantly associated with severe COVID-19 (OR 6.20; 95%
			CI: 2.82–13.62, p<0.001).
Tripon et al. [41] 2022	719	Retrospective cohort	In patients with NAFLD, no significant difference was observed in the proportion of patients who experience severe
			COVID-19 compared to those who experience only mild-to-moderate COVID-19 (59.8% vs 63%, p=0.455)
			More patients with severe COVID-19 had NAFLD associated with a high fibrosis-4 index (FIB-4): 72.8% vs. 57.9%,
			respectively (p<0.001)
Vrsaljko et al. [42] 2022	216	Prospective cohort	Patients with NAFLD more frequently required high-flow nasal cannula or non-invasive ventilation than those without
			NAFLD (21.66% vs 10.42%; p=0.289)
			No significant differences in the requirement for invasive mechanical ventilation between NAFLD and non-NAFLD
			groups (5% vs 3.12%)
			No significant differences in the in-hospital mortality between NAFLD and non-NAFLD groups (6.67% vs 3.12%;
			p=0.3529)
Wang et al. [43] 2021	218	Retrospective cohort	No significant differences were observed between NAFLD and non-NAFLD patients in terms of severe COVID-19
			(22.1% vs 16.7%; p=0.316).
			The mortality rate did not differ significantly between subjects with NAFLD and those without NAFLD (0% vs 1.5%;
			p=0.251).
			In the normal BMI group, after adjusting for age, gender, hypertension, cardiovascular disease, diabetes, and chronic
			liver disease, it has been revealed that NAFLD was significantly associated with severity of COVID-19 (HR 3.26; 95%
			CI: 1.17–9.04, p=0.023).
Yoo et al. [44] 2021	74244	Retrospective cohort	In matched cohort, patients with NAFLD had a higher risk of severe COVID-19 disease progression than patients
			without NAFLD (0.8% vs 0.6%; aOR 1.41; 95% CI: 1.08–1.83).
			Progression to severe COVID-19 was more commonly observed on patients with NAFLD than those without NAFLD
			(aOR 1.35; 95% CI: 1.05–1.71).

COVID-19: coronavirus disease 2019; CT-scan: computed-tomography scan; ICU: intensive care unit; HR: hazard ratio; NAFLD: non-alcoholic fatty liver disease; OR: odds ratio

This study has some limitations. The limitation of data on the NAFLD/diabetic/dyslipidemia medications in included studies makes it unable to be analyzed. Moreover, any heterogeneity found in our study was presumably generated by diverse demographic characteristics among the studies included. However, we assume this study could still give advanced insight into additional risk stratification for COVID-19.

Conclusions

Patients with NAFLD should be encouraged to take extra forethought in reducing the risk of exposure. Patients with suspected COVID-19 infection should also be closely observed for the appropriate detection of disease progression. Certainly, the NAFLD shall be a consideration to risk stratification models for COVID-19 in the prospect.

Ethics approval

Not required

Conflict of interest

All the authors declare that there are no conflicts of interest.

Funding

This study received no external funding.

Underlying data

All data underlying the results are available from the corresponding author upon reasonable request.

How to cite

Kurniawan A, Hariyanto TI. Non-alcoholic fatty liver disease (NAFLD) and COVID-19 outcomes: A systematic review, meta-analysis, and meta-regression. Narra J 2023; 3 (1): e102 -http://doi.org/10.52225/narra.v3i1.102.