Shift work and fatty liver disease linked to metabolism problems: a review of observational studies

The systematic review was conducted and reported according to recommendations from the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement (Page et al. 2021). The systematic review protocol was registered within the International Prospective Register of Systematic Reviews (PROSPERO) on 22 November 2024 (Registration number: CRD42024612749).

We searched published articles using PubMed, Scopus, and Web of Science from the inception of each database to November 25, 2024. Exposure was identified using a range of terms related to shift and night work, while outcome terms included various descriptors of fatty liver disease, including MASLD, MASH, NAFLD, and NASH. Both MASLD and NAFLD were included as search terms, as evidence suggests that 99% of individuals diagnosed with NAFLD also meet the criteria for MASLD (Hagström et al. 2024). The hierarchical Medical Subject Headings (MeSH) terms "Non-alcoholic Fatty Liver Disease" and "Shift Work Schedule" in the PubMed database were used to optimize the inclusion of study exposure and outcome. The full search strategy is shown in Supplementary Table 1. Articles retrieved according to the search strategies were then imported to Covidence to remove duplicates, conduct independent screening, and resolve discrepancies.

Two independent authors scanned titles, abstracts, and full texts and included the studies that met the following inclusion criteria: (1) was a cohort, case–control, or cross-sectional study; (2) reported any type of shift work (e.g., regular night/morning work, 2-shift work, or other shift work schedule) as the exposure compared with the regular daytime work as the reference group, or any type of shift work in comparison with participants without that type of work; (3) reported MASLD as the study outcome. We excluded all animal studies, editorials, letters, comments, abstracts, posters, meta-analyses, and reviews. Studies reporting the impact of shift work on liver function without specifying MASLD as an outcome were also excluded.

We extracted information on the author, publication year, study design, sample size, follow-up time, participants' age and gender, occupational group, measurements of shift work and MASLD, covariates, and study results. Both the crude and adjusted odds ratio (OR), relative risk (RR), incidence rate ratio (IRR), hazard ratio (HR), and 95% confidence interval (CI) were included as the risk estimates. The aforementioned data elements were extracted in a pre-designed Excel spreadsheet. The National Institutes of Health (NIH) quality assessment tool was used in evaluating the quality of included studies. The NIH quality assessment tool is a 14-item scale covering the clarity of the research question and study population, participation rate and application of inclusion and exclusion criteria, sample size justification, exposure and outcome measurement, sufficiency of timeframe, follow-up rate, and confounding adjustment (Ma et al. 2020; Stang 2010). Both data extraction and quality assessment were conducted by two authors independently. Agreement on key variables extraction was assessed, yielding a percent agreement of 94.4%, indicating high consistency. Interrater reliability for the quality assessment was measured using Cohen's kappa, which was 0.59, indicating moderate agreement. All disagreements were resolved by consensus.

Most studies collected shiftwork information through questionnaires. Zhang et al. (2020) collected the information through face-to-face personal interviews and further verified it with the company's records, which, to some degree, avoided the recall bias and increased the study's validity.

Shift work exposure has been classified using various methods across studies, reflecting schedule, duration, and frequency differences. Three studies categorized the type of shift work based on whether it involved fixed/regular shifts or rotating/irregular shifts (Balakrishnan et al. 2017; Lee and Lee 2024; Maidstone et al. 2024). Taechasan and Jiamjarasrangsi (2024) classified participants into three groups: non-shift workers (those with no experience of night shift work or who worked fewer than four night shifts per month), former shift workers (those who previously worked more than four night shifts per month but were not engaged in shift work at the study's start), and current shift workers (those currently engaged in fixed or rotating night shift work for at least four nights per month). Several studies focused on metrics related to the duration and frequency of shift work. For example, Balakrishnan et al. (2017) classified shift workers by weekly hours worked (< 35, 36–40, or > 40 h). Che et al. (2024) and Taechasan and Jiamjarasrangsi (2024) categorized night shift frequency based on the number of night shifts per month. Additionally, Taechasan and Jiamjarasrangsi (2024), along with Kim et al. (2022), classified the total duration of shift work exposure by years, such as less than 10 years, 10–20 years, or more than 20 years. Xu et al. (2023) grouped participants based on the number of night shifts during the past year at baseline into seldom (< 1 time per week), occasional (1–3 times per week), and frequent (> 3 times per week) shift work categories. Huang et al. (2023) assessed exposure through multiple metrics, including duration (years of night shift work), frequency (average shifts per month), length (hours per shift), and consecutive shifts within a night shift period. Zhang et al. (2020) developed a comprehensive framework incorporating various metrics, including current/ever shift work, duration (years), cumulative number of night shifts (nights), cumulative length (hours), average frequency (nights/month), and average length (hours/night). The diverse classification systems used across studies underscore the complexity of measuring shift work exposure and highlight the need for standardized approaches to improve comparability and synthesis in future research.

Although no study utilized the gold standard liver biopsy for outcome assessment, various validated imaging and diagnostic methods were employed. Xu et al. (2023) and Lee and Lee (2024) utilized abdominal ultrasonography performed by experienced sonographers blinded to the study aims. Taechasan and Jiamjarasrangsi (2024) utilized diagnostic scores such as the lipid accumulation product (LAP) and hepatic steatosis index (HSI), both validated for sensitivity and specificity in prior research. Cross-sectional studies by Kim et al. (2022) and Zhang et al. (2020) also employed ultrasonography, interpreted by two blinded sonographers, to classify NAFLD severity into normal, light, and moderate-to-severe categories. Similarly, Che et al. (2024) relied on abdominal B-ultrasound findings as imaging evidence. In contrast, Balakrishnan et al. (2017) indirectly assessed NAFLD using liver enzyme levels due to the lack of imaging data in the NHANES cycles analyzed. Two studies based on the UK Biobank used different criteria. Huang et al. (2023) identified NAFLD using ICD-10 codes for hospital admissions or deaths, while Maidstone et al. (2024) adopted a more comprehensive approach, combining the Dallas steatosis index (DSI)—a logit model incorporating variables such as age, diabetes, hypertension, triglycerides, and ethnicity—with ICD-10 codes and hepatic proton density fat fraction (PDFF).

Several studies also implemented exclusion criteria to remove potential secondary causes. For example, Balakrishnan et al. (2017), Zhang et al. (2020), and Kim et al. (2022) excluded individuals with hepatitis B or C, as well as those with high alcohol consumption. Che et al. (2024) further excluded cases of drug-induced or autoimmune liver disease, while Lee and Lee (2024) excluded patients with cirrhosis based on ultrasonography and those taking medications for hepatitis, cirrhosis, or hepatic steatosis. To minimize alcohol-related confounding, Xu et al. (2023) excluded individuals with current or past excessive alcohol use and conducted sensitivity analyses on participants without liver diseases. Similarly, the two UK Biobank studies adjusted for alcohol intake frequency, with Huang et al. (2023) also performing a sensitivity analysis excluding participants with excessive alcohol consumption. Applying even stricter criteria, Taechasan and Jiamjarasrangsi (2024) excluded participants with abnormal liver function tests, such as elevated Aspartate Aminotransferase (AST), Alanine Aminotransferase (ALT), lipid accumulation product (LAP), or hepatic steatosis index (HSI), ensuring that only incident NAFLD cases were included in the analysis. These exclusion criteria help strengthen the focus on MASLD by minimizing the influence of other liver diseases or confounding factors.

Notably, temporality was not established in many of the included studies, particularly among those with cross-sectional designs. For example, Kim et al. (2022) and Zhang et al. (2020) assessed both shift work and NAFLD concurrently. Balakrishnan et al. (2017) also explicitly acknowledged that their study design precluded the assessment of a temporal sequence. Similarly, Maidstone et al. (2024) and Che et al. (2024) analyzed data from large cohort studies in a cross-sectional manner, offering no clear temporal direction. In contrast, cohort studies addressed temporality more effectively through their design. Huang et al. (2023) excluded participants with liver disease at baseline. Lee and Lee (2024) and Taechasan and Jiamjarasrangsi (2024) excluded individuals with a history of liver disease. Xu et al. (2023) excluded participants diagnosed with NAFLD prior to cohort entry, which demonstrated temporal alignment to support potential causal inferences between shift work and NAFLD.

Demographic factors were commonly adjusted for across studies, with several also considering socioeconomic variables, such as education level and household income (Balakrishnan et al. 2017; Zhang et al. 2020; Huang et al. 2023; Taechasan and Jiamjarasrangsi 2024; Maidstone et al. 2024; Lee and Lee 2024). Regarding lifestyle factors, studies consistently accounted for smoking status, alcohol consumption, and physical activity, with some incorporating more specific metrics, such as the Dietary Approaches to Stop Hypertension (DASH) score (Zhang et al. 2020) and sleep duration (Maidstone et al. 2024). Adjustments for clinical and metabolic factors varied among studies. For instance, body mass index (BMI) was controlled for in most of the studies (Balakrishnan et al. 2017; Zhang et al. 2020; Kim et al. 2022; Xu et al. 2023; Taechasan and Jiamjarasrangsi 2024; Maidstone et al. 2024; Lee and Lee 2024), while some studies also adjusted for liver function tests, including serum concentrations of liver enzymes (Kim et al. 2022; Xu et al. 2023). Additionally, work-related variables, such as job type and length of the working week, were adjusted for in studies by Taechasan and Jiamjarasrangsi (2024) and Maidstone et al. (2024).

Cross-sectional studies provided mixed results. For example, Kim et al. (2022) found that shift workers had higher odds of moderate-to-severe NAFLD compared to daytime workers (OR 1.499, 95% CI 1.028–2.043), although this association did not extend to light NAFLD. This result could be due to the limitations of abdominal ultrasound, which could not accurately distinguish between normal and light NAFLD. Zhang et al. (2020) identified night shift duration, cumulative number of night shifts, cumulative length of night shifts, and the frequency and length of night shifts as significant independent risk factors for NAFLD. Che et al. (2024) further reinforced the role of shift work intensity, identifying working more than five-night shifts per month as a strong independent risk factor for NAFLD prevalence (OR 3.512, 95% CI 1.280–9.633). Maidstone et al. (2024) provided different insights, showing that hepatic PDFF was elevated in irregular-shift workers but not in permanent night-shift workers, which implies that circadian misalignment may be a key factor driving pathological liver fat accumulation in certain groups of shift workers. In contrast, Balakrishnan et al. (2017) reported no significant association between shift work and NAFLD risk (OR 1.11, 95% 0.87–1.43). The reliance on enzyme-based assessments rather than imaging for diagnosis may have contributed to this null finding, as enzyme-based measures might lack the sensitivity required to detect early or subclinical NAFLD.

Regarding the longitudinal analyses, Huang et al. (2023) identified a dose–response relationship between night shift work and NAFLD risk. Workers engaging in occasional night shifts (OR 1.12, 95% CI 0.96–1.31) or permanent night shifts (OR 1.27, 95% CI 1.08–1.48) were at higher risk of developing NAFLD compared to those who never or rarely worked night shifts. Moreover, among 75,059 participants with lifetime shift work data, longer duration, higher frequency, more consecutive night shifts, and longer shift lengths were consistently associated with elevated NAFLD risk. Similarly, Xu et al. (2023) also found that frequent shift work, but not occasional shift work, was a significant risk factor for NAFLD in rail workers (HR = 1.179, 95% CI 1.059–1.312).

A growing body of research has explored how demographic, lifestyle, chronotype, genetic, and occupational factors modify the relationship between shift work and MASLD risk. For example, Lee and Lee (2024) suggested shift work significantly increased NAFLD risk in women (IRR = 1.39, 95% CI 1.05–1.84) in their 20s but not in other gender and age groups, suggesting that younger women may be more susceptible to the adverse effects of shift work. Conversely, Zhang et al. (2020) reported a significant association between night shift work and NAFLD among male workers but not among females, likely due to the small number of female workers and the low prevalence of NAFLD in this group, which may have limited the accuracy of interval estimates.

Lifestyle factors also play a crucial role. Balakrishnan et al. (2017) found a significant association between shift work and NAFLD only in individuals with normal BMI. Kim et al. (2022) observed that workers prohibited from sleeping during shifts and those who consumed food at night had a higher risk of moderate-to-severe NAFLD. Similarly, Lee and Lee (2024) found that shift work was significantly associated with an increased risk of NAFLD, especially among individuals with poor sleep quality and those in their 20s (IRR = 1.37, 95% CI 1.11–1.70). Xu et al. (2023) found that the NAFLD risk associated with occasional shift work was higher among individuals who had never smoked and maintained a balanced diet, and the NAFLD risk associated with frequent shift work was elevated among those with never smoking, a balanced dietary pattern, middle salt intake, seldom physical activity, and moderate sleep duration, although the interaction was not statistically significant. These findings highlight the complex relationships between shift work and lifestyle factors.

Two studies explored the potential effect modification by chronotype and genetic factors. Maidstone et al. (2024) examined chronotype and found that patients with extreme late chronotypes had a higher likelihood of NAFLD (OR 1.45, 95% CI 1.34–1.56 for DSI-defined NAFLD; OR 1.23, 95% CI 1.09–1.39 for ICD-10-diagnosed NAFLD). In addition, the association between shift work and DSI was stronger among morning types than among evening types (P-interaction < 0.01). Huang et al. (2023) used polygenic risk scores (PRS) to assess genetic susceptibility to NAFLD in individuals of European descent. Despite variations in genetic risk, the study found no significant interaction between genetic predisposition and the relationship between night shift work and incident NAFLD (P-interaction: 0.151–0.386), indicating that genetic risk does not significantly modify the effects of shift work on liver health.

Findings from Taechasan and Jiamjarasrangsi (2024) also reveal a relationship between shift work duration, occupation type, and NAFLD risk. Healthcare workers with 10–20 years of shift work had a higher risk of elevated lipid accumulation product (e-LAP), whereas those with over 20 years of exposure had a lower risk. In contrast, non-healthcare workers with less than 10 years of shift work had a higher risk of elevated hepatic steatosis index (e-HSI). These findings suggest that prolonged shift work may lead to adaptive changes in healthcare workers, potentially reducing NAFLD risk. The decreasing workload of shift workers in Thailand as shift work duration increases, particularly healthcare personnel, may also help explain these results.

The role of BMI and obesity in mediating the relationship between shift work and MASLD has been demonstrated in several studies. Huang et al. (2023) conducted a mediation analysis, revealing that BMI accounted for 31.7% (95% CI 14.3–56.4, P < 0.001) of the association between current night shift work and NAFLD development. Xu et al. (2023) also examined the mediating role of metabolic factors, finding that higher BMI, mean arterial pressure (MAP), fasting blood glucose (FBG), total cholesterol (TC), triglycerides (TG), ALT, and AST directly contributed to NAFLD occurrence, while the indirect effects of FBG, TC, TG, ALT, and AST were not statistically significant. Maidstone et al. (2024) obtained similar results that approximately 36% of the relationship between shift work and NAFLD was mediated by BMI (P < 0.01), underscoring the mediating role of BMI in this pathway.