Associations of Unhealthy Lifestyle and Nonalcoholic Fatty Liver Disease With Cardiovascular Healthy Outcomes

Abstract

Background

The role of nonalcoholic fatty liver disease (NAFLD) as a mediator in the association between various unhealthy lifestyles and major adverse cardiovascular events and all‐cause death remains unclear.

Methods and Results

This study used data from the UK Biobank, with follow‐up until the end of 2021. It involved the calculation of unweighted and weighted lifestyle scores using the Cox model to classify participants on the basis of these scores. Additionally, the research assessed the mediation effect proportion of NAFLD using the difference method and examined the interaction and joint effects of lifestyle and NAFLD on health outcomes. Among the 134 616 enrolled participants, 4024 had records of major adverse cardiovascular events, while among the 130 144 participants included in the analysis of all‐cause death, 6697 deaths occurred. The proportions of the association between overall lifestyle and major adverse cardiovascular events mediated by NAFLD were 19.4% and 21.7% (95% CI, 16.2–22.6 and 17.8–25.7) for scores 1 and 2, respectively, and those for all‐cause death were 14.1% and 10.1% (95% CI, 11.3–17.1 and 7.9–12.2). After fully adjusting for traditional cardiovascular risk factors, the mediating effects declined across both outcomes. The associations between overall lifestyle and outcomes were stronger among those of the non‐NAFLD group, and significant interactions were observed between overall lifestyle and NAFLD status. The joint analysis revealed that patients with NAFLD with unhealthy lifestyle had the highest risk of major adverse cardiovascular events and all‐cause death.

Conclusions

Improving lifestyle and addressing metabolic risk factors are essential for cardiovascular risk management in patients with NAFLD.

Nonstandard Abbreviations and Acronyms

FLI

fatty liver index

MACEs

major adverse cardiovascular events

NAFLD

nonalcoholic fatty liver disease

TDI

Townsend deprivation index

UKB

UK Biobank

Clinical Perspective.

What Is New?

• Nonalcoholic fatty liver disease (NAFLD) acts as a mediating factor in the associations between various lifestyles (diet habits, physical activity habits, sleep habits, and overall lifestyle) and major adverse cardiovascular events as well as all‐cause death; the mediating effect of NAFLD primarily stems from traditional cardiovascular risk factors, such as obesity, type 2 diabetes, hypertension, dyslipidemia, smoking, and low socioeconomic status.

• Improvement of lifestyle contributes to the reduction of risks associated with major adverse cardiovascular events and all‐cause death, both in individuals with NAFLD and individuals without NAFLD.

• Furthermore, the enhancement of sleep habits yields greater benefits, specifically within NAFLD individuals.

What Are the Clinical Implications?

• NAFLD should be integrated as a risk factor into cardiovascular disease control and prevention strategies.

• To reduce cardiovascular risks among individuals with NAFLD, the emphasis lies in managing traditional cardiovascular risk factors such as hypertension, diabetes, and dyslipidemia; improving lifestyle holds positive implications for cardiovascular disease and all‐cause death prevention across the general population, with particular significance for patients with NAFLD with unfavorable lifestyles.

Cardiovascular disease (CVD) remains the leading disease burden in the world, and the burden from CVD attributable to modifiable risk factors continues to increase globally. ¹ Numerous studies support the notion that preventable unhealthy lifestyle choices, such as insufficient physical activity, sedentary behavior, smoking, unhealthy diet, and poor sleep quality, are closely associated with CVD and represent potential risk factors for various CVDs. ² , ³

Worldwide lifestyle changes in recent decades are making higher intakes of calorie‐dense foods and reduced physical activity and sleep more common, a trend that has also driven the prevalence of other metabolic diseases. Nonalcoholic fatty liver disease (NAFLD) has received wider attention due to its high prevalence, severe disease burden, ⁴ and association with increased all‐cause death. ⁵ NAFLD is an emerging global epidemic of a chronic metabolic disease, which is considered as the target organ manifestation of the liver part of metabolic syndrome and shares cardiometabolic risk factors with CVD. ⁶ Multiple studies believe that NAFLD is a risk factor for CVD and is closely associated with different stages of CVD. ⁷ , ⁸ , ⁹ However, important gaps in knowledge remain. First, the nature and strength of the association between NAFLD and CVD has not been clearly determined, and it is unclear whether the increased disease risk associated with NAFLD is derived from metabolic risk factors associated with CVD or from factors related to liver disease, with several observational studies providing different conclusions. ¹⁰ , ¹¹ Considering that healthy lifestyle choices that are beneficial for the prognosis of CVD are also helpful for NAFLD prevention and improvement, whether NAFLD is a mediator factor connecting lifestyle and CVD outcomes needs to be further clarified. Second, previous studies mostly focused on the impact of single lifestyle factors on CVD and NAFLD, while the combination of lifestyle factors may further reflect the synergistic effect on the occurrence and prognosis of disease, and this has not been properly investigated. ¹² Third, it is unclear whether NAFLD and lifestyle have a combined effect on CVD and death.

Obtaining a better understanding of the associations among lifestyle choices, NAFLD, and cardiovascular outcomes is particularly important for the inclusion of NAFLD in the assessment and monitoring of cardiovascular risk factors for further intervention and for targeting individual and population‐level public health interventions. The present study used a large population data cohort from the UK Biobank (UKB) to evaluate the complex relationships of lifestyles and NAFLD with all‐cause death and CVD.

Methods

The supporting data for the findings of present study can be obtained from the corresponding author upon reasonable request.

Data Collection and Patient Selection

The present study included 502 492 participants from the UKB program aged 40 to 69 years, who were followed up until the end of 2021. Informed consent was obtained from all participants in the UKB, and the UKB project was approved by the research ethics committee (reference 11/NW/0382). The data used in the present study are approved by the UKB (application number 62017) and accessible to researchers through the UKB Resource (http://www.ukbiobank.ac.uk/about‐biobank‐uk/) by submitting an application. More details about UKB have been widely used in previous studies. ¹³ , ¹⁴ To avoid confounding NAFLD with other liver diseases, participants were excluded from the analysis if they were diagnosed with alcoholic liver disease, toxic liver injury, or viral hepatitis, or if they had corresponding outcomes at baseline. In addition, participants who could not have their fatty liver index (FLI) calculated and those with missing lifestyle data or other covariates were excluded. Considering the representativeness of the results, the unhealthy lifestyle screening was based on the entire database sample. This study was conducted in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology Statement: guidelines for reporting observational studies. ¹⁵

NAFLD Assessment

NAFLD was diagnosed using the FLI. The FLI is an index that uses body mass index (BMI), waist circumference, triglycerides, and γ‐glutamyl transpeptidase to evaluate liver status, and FLI ≥60 indicates NAFLD. Compared with liver biopsy, which is an expensive and invasive procedure, FLI is inexpensive and easy to perform, and is widely used to screen large‐scale populations. The formula for FLI is as follows:

$$\mathrm{FLI}=\frac{e^{0.953\times \ln \left(\mathit{TG}\right)+0.139\times \mathrm{BMI}+0.718\times \ln \left(\mathit{GGT}\right)+0.053\times \text{Waist}-15.745}}{1+e^{0.953\times \ln \left(\mathit{TG}\right)+0.139\times \mathrm{BMI}+0.718\times \ln \left(\mathit{GGT}\right)+0.053\times \text{Waist}-15.745}}\times 100$$

Ascertainment of Outcomes

The date and cause of death were identified by referring to death registries of the National Health Service Information Centre for participants in the UKB study from England and Wales, and the National Health Service Central Register of Scotland for those from Scotland. ¹⁴ Outcome data were classified according to the International Classification of Diseases, Tenth Revision (ICD‐10). The outcomes of this study were major adverse cardiovascular events (MACEs) and all‐cause death. MACE was defined as the first myocardial infarction (ICD‐10 codes I21–I25 and self‐reported), cerebral infarction (ICD‐10 codes I60, I61, I63, and I64, and self‐reported), or cardiovascular death (ICD‐10 codes I05–I89) of a participant.

Ascertainment of Lifestyle Scores and Other Covariables

Constructing the lifestyle scores is a 2‐step process. First, we used the Cox model to separately analyze the association of diet, physical activity, and sleep‐related variables and health outcomes. Subsequently, scoring was performed using 2 different methods on the basis of the model results. ¹⁶ , ¹⁷ , ¹⁸ The aim of constructing lifestyle scores is to categorize participants' lifestyles.

Variables related to diet were obtained through questionnaires in the UKB project, which included consumption (yes/no) of vegetables, fruit, coffee, tea, milk, cheese, fish, poultry, pork, beef, lamb, processed meat, whole‐grain foods, sugary drinks, and salty foods. The variables associated with physical activity included the metabolic equivalent of physical activity, which was calculated on the basis of the International Physical Activity Questionnaire, and sedentary time, which was calculated by adding the amount of time spent driving, using computers, and watching television. Sleep duration per night, chronotype, snoring, insomnia, nap times, and morning difficulty were used to assess the sleep habits of participants.

The specific scoring process was as follows: model variables related to diet, physical activity, and sleep habits and outcomes. For score 1, each variable was scored according to whether the hazard ratio (HR) of each variable group was significant, with a significant risk group scoring 1 point and a nonsignificant risk group scoring 0 points. The scores for all variables were then added. ¹⁹ , ²⁰ For score 2, the point estimates (effect values) of all variable groups were directly added to obtain the scores for diet, physical activity, and sleep habits for each participant. It is a weighted method that avoids errors caused by different weights of variables in score 1 (the correlation coefficients of all behavioral habit variables in this study were <0.5, which can support the hypothesis of variable independence). ¹⁹ , ²⁰ For scores 1 and 2, the groups were classified according to the scores from low to high, to obtain good, moderate, and poor score groups. In both scores, higher scores meant worse lifestyles. An overall unhealthy lifestyle was established on the basis of the following criteria: Unhealthy diet, physical activity, and sleep habits were scored, and then the participant was assigned to the high‐score group (poor lifestyle group, 3 points), middle‐score group (moderate lifestyle group, 2 points), or low‐score group (healthy lifestyle group, 1 point). The scores of the 3 living habits were summed, with 3 or 4 points for the healthy lifestyle group, 5 to 7 points for the moderate lifestyle group, and 8 or 9 points for the poor lifestyle group.

Other variables were demographic variables (age, race, sex), physical and biochemical indicators (BMI, waist circumference, fasting plasma glucose, glycosylated hemoglobin, high‐density lipoprotein, low‐density lipoprotein, total cholesterol, triglycerides, systolic blood pressure, diastolic blood pressure, alanine aminotransferase, aspartate aminotransferase, direct bilirubin, γ‐glutamyl transpeptidase, serum creatinine, and C‐reactive protein), comorbidities (hypertension and diabetes), drug use (statins), smoking status (never/former or current), and socioeconomic status (Townsend deprivation index [TDI]). The TDI is an indicator that reflects socioeconomic status on the basis of household ownership of vehicles, housing overcrowding, housing availability, and employment. ²¹ A larger TDI represents a lower socioeconomic level in the local area.

Definition of Traditional Cardiovascular Risk Factors

Cardiovascular risk factor assessment has been regarded as a vital part of CVD prevention and treatment for >20 years, and this prevention strategy has been widely used in the CVD prevention and treatment guidelines of various countries. An early and influential cardiovascular risk factor assessment study was the Framingham Heart Study in the United States. ²² Like other early epidemiological studies, that study established a series of major traditional cardiovascular risk factors that included age, sex, dyslipidemia, elevated blood pressure, antihypertensive therapy, obesity, diabetes, smoking status, family history of premature CVD, and low socioeconomic status. Among them, age, sex, blood pressure, blood lipid level, and smoking status had independent predictive effects on CVD risk, and so have been used in almost all algorithms for CVD prognosis evaluation models, including Chinese, American, and European guidelines for CVD prevention and treatment. ²³ , ²⁴ The present study focused on the mediating role of NAFLD on CVD, which differs from that of traditional cardiovascular risk factors. The adjustment of confounding factors, mediation analysis, subgroup analysis, and joint effect in the present study were therefore based on the above‐mentioned traditional cardiovascular risk factors.

Statistical Analysis

Categorical variables are expressed as quantities and percentage values. Continuous variables that fit a normal distribution are expressed as mean and SD, and continuous variables that do not conform to a normal distribution are expressed as median and interquartile range. The chi‐square test, ANOVA, and Wilcoxon rank‐sum test were used to analyze the clinicopathological characteristics of the NAFLD group and non‐NAFLD group. The causal mediation analysis of survival data is used to assess the mediating effects, and we used the “regmedint” package in R to complete this analysis process (R Foundation for Statistical Computing, Vienna, Austria). ²⁵ , ²⁶ It uses the Cox model as the fundamental survival analysis model. Model 1 was adjusted for age, sex, race, and TDI (some traditional cardiovascular risk factors). Model 2 additionally adjusted for smoking status, BMI, waist circumference, total cholesterol, triglycerides, hypertension, diabetes, liver function (aspartate aminotransferase, alanine aminotransferase, and bilirubin), kidney function (serum creatinine), and statin use (all traditional cardiovascular risk factors). A “difference method” was used to calculate the mediation proportion by the mediator for the association between lifestyle and each outcome—that is, comparing estimates from models with and without the hypothesized mediator. ²⁷

A subgroup analysis based on participants' NAFLD status was conducted to explore the differences in the effect of lifestyle on health outcomes across different subgroups of NAFLD. In each subgroup, the reference group was set as the participants with the least healthy lifestyle.

To assess the joint effect of lifestyle and NAFLD, we divided the participants into 6 groups based on lifestyle (most healthy, moderate healthy, and least healthy) and NAFLD status (with and without). ¹⁹ , ²⁸ , ²⁹ , ³⁰ The reference group consisted of individuals with the most healthy lifestyle and without NAFLD. We calculated the HRs for MACEs and all‐cause death outcomes in the other groups. All statistical analyses were performed using R software version 3.6.3. All statistical tests were 2‐sided, and a probability value of P<0.05 was considered statistically significant.

Results

Baseline Characteristics

Table 1 lists the main characteristics of the enrolled participants. The 134 616 participants (mean age, 55.78±8.28 years; 61.93% women) comprised 29 874 in the NAFLD group and 104 742 in the non‐NAFLD group. Participants in the NAFLD group had a higher proportion of men, current smokers, hypertension, diabetes, and statin users, as well as higher levels of TDI, BMI, waist circumference, low‐density lipoprotein, triglycerides, aspartate aminotransferase, alanine aminotransferase, glycosylated hemoglobin, γ‐glutamyl transpeptidase, serum creatinine, and C‐reactive protein.

Table 1.

Characteristics of the Participants in the Cohort

Characteristics	Total	Non‐NAFLD	NAFLD	P value
	(N=134 616)	(N=104 742)	(N=29 874)
Age, y	55.78 (8.28)	55.62 (8.36)	56.34 (7.95)	<0.001
Sex, female (%)	83 366 (61.93)	70 059 (66.89)	13 308 (44.55)	<0.001
Race (%)
White	124 516 (92.50)	96 673 (92.30)	27 853 (93.24)	<0.001
Asian	3467 (2.58)	2812 (2.68)	667 (2.23)
Black	2535 (1.88)	1952 (1.87)	583 (1.95)
Mixed	1366 (1.01)	1083 (1.03)	272 (0.91)
Chinese and other ethnic group	2732 (2.03)	2222 (2.12)	499 (1.67)
TDI	−1.36 (3.02)	−1.47 (2.96)	−0.97 (3.18)	<0.001
BMI (kg/m2)	27.53 (4.92)	25.77 (3.29)	33.71 (4.69)	<0.001
Waist circumference (cm)	89.44 (13.52)	84.55 (9.93)	106.61 (9.96)	<0.001
Glucose (mmol/L)	5.10 (1.24)	4.98 (0.95)	5.48 (1.89)	<0.001
Systolic blood pressure (mm Hg)	138.00 (19.43)	136.35 (19.47)	143.80 (18.13)	<0.001
Diastolic blood pressure (mm Hg)	81.44 (10.55)	80.10 (10.30)	86.14 (10.06)	<0.001
Total cholesterol (mmol/L)	5.69 (1.13)	5.67 (1.10)	5.71 (1.22)	0.478
HDL (mmol/L)	1.40 (0.35)	1.47 (0.35)	1.18 (0.26)	<0.001
LDL (mmol/L)	3.57 (0.86)	3.56 (0.84)	3.63 (0.91)	<0.001
Triglycerides (mmol/L)	1.74 (1.00)	1.50 (0.74)	2.61 (1.26)	<0.001
Smoking status (%)
Never	83 217 (61.82)	67 105 (64.07)	16 120 (53.96)	<0.001
Former	39 441 (29.30)	28 646 (27.35)	10 795 (36.14)
Current	11 958 (8.88)	8991 (8.58)	2959 (9.90)
Hypertension (%)	25 039 (18.60)	16 444 (15.70)	8634 (28.90)	<0.001
Diabetes (%)	11 981 (8.90)	5237 (5.00)	6692 (22.40)	<0.001
AST (U/L)	25.56 (9.05)	24.69 (8.06)	28.64 (11.37)	<0.001
ALT (U/L)	22.79 (13.38)	20.36 (10.57)	31.31 (17.95)	<0.001
HbA1c (mmol/mol)	36.19 (6.77)	35.33 (5.33)	39.18 (9.76)	<0.001
Bilirubin (μmol/L)	8.86 (4.36)	8.92 (4.41)	8.66 (4.14)	<0.001
γ‐GT (U/L)	23.7 (17.2 to 35.8)	21.1 (16.0 to 29.6)	38.9 (28.0 to 58.3)	<0.001
Serum creatinine (μmol/L)	71.22 (17.45)	70.15 (16.64)	74.95 (19.59)	<0.001
CRP (mg/L)	1.36 (0.67 to 2.81)	1.12 (0.57 to 2.24)	2.63 (1.42 to 4.95)	<0.001
Statin therapy (%)	19 115 (14.20)	11 836 (11.30)	7289 (24.40)	<0.001
Fatty liver index	34.31 (28.59)	21.56 (16.77)	79.00 (11.42)	<0.001

Values are mean±SD, n (%), or median (interquartile range). ALT indicates alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; CRP, C‐reactive protein; γ‐GT, γ‐glutamyl transpeptidase; HbA_1c, glycosylated hemoglobin; HDL, high‐density lipoprotein; LDL, low‐density lipoprotein; NAFLD, nonalcoholic fatty liver disease; and TDI, Townsend deprivation index.

Mediation Analysis of Lifestyle on Associations of NAFLD With MACEs and All‐Cause Death

Of the 134 616 participants who were included in the MACE outcome analysis, with a median follow‐up time of 11.65 years, 4204 had MACE outcomes recorded. In both model 1 and model 2, after adjusting for FLI, the HRs of MACEs decreased. In model 1, after adjusting for FLI and other covariates, the HRs of MACEs were 1.68 (95% CI, 1.52–1.86; P<0.001) for score 1 and 1.55 (95% CI, 1.39–1.73; P<0.001) for score 2 in the least healthy overall lifestyle group compared with the most healthy overall lifestyle group as the reference, and the proportions of the mediation effect of NAFLD were 19.4% and 21.7% (95% CI, 16.2–22.6 and 17.8–25.7) for score 1 and score 2, respectively. Model 2 showed the results of the model adjusted for all traditional cardiovascular risk factors and FLI, the HRs of MACEs were 1.59 (95% CI, 1.44–1.75; P<0.001) for score 1 and 1.48 (95% CI, 1.32–1.65; P<0.001) for score 2 in the least healthy overall lifestyle group compared with the most healthy overall lifestyle group. At this point, for score 1 and score 2, the proportions of the mediation effect of NAFLD on MACEs were reduced to 2.5% (95% CI, 1.4–3.6) and 3.5% (95% CI, 1.9–5.1), respectively (Table 2).

Table 2.

Associations of Lifestyle With MACEs and All‐Cause Death and Mediation Proportion of Different Lifestyle in Health Attributed to NAFLD

	Model 1			Model 2
	Hazard ratio (95% CI)		Mediation proportion (%) (95% CI)	Hazard ratio (95% CI)		Mediation proportion (%) (95% CI)
	Unadjusted for FLI	Adjusted for FLI		Unadjusted for FLI	Adjusted for FLI
MACEs
Score 1
Most healthy lifestyle	1 (reference)	1 (reference)	…	1 (reference)	1 (reference)	…
Moderate healthy lifestyle	1.40 (1.28–1.53)*	1.33 (1.22–1.45)*	19.0 (14.2–23.7)	1.30 (1.19–1.41)*	1.29 (1.18–1.41)*	1.9 (0.4–3.4)
Least healthy lifestyle	1.89 (1.71–2.08)*	1.68 (1.52–1.86)*	19.4 (16.2–22.6)	1.60 (1.45–1.77)*	1.59 (1.44–1.75)*	2.5 (1.4–3.6)
Score 2
Most healthy lifestyle	1 (reference)	1 (reference)	…	1 (reference)	1 (reference)	…
Moderate healthy lifestyle	1.33 (1.19–1.49)*	1.27 (1.14–1.42)*	23.8 (19.1–28.7)	1.26 (1.13–1.41)*	1.25 (1.12–1.40)*	3.5 (1.6–5.3)
Least healthy lifestyle	1.73 (1.55–1.93)*	1.55 (1.39–1.73)*	21.7 (17.8–25.7)	1.49 (1.34–1.67)*	1.48 (1.32–1.65)*	3.5 (1.9–5.1)
All‐cause death
Score 1
Most healthy lifestyle	1 (reference)	1 (reference)	…	1 (reference)	1 (reference)	…
Moderate healthy lifestyle	1.19 (1.09–1.30)*	1.16 (1.06–1.26)†	14.7 (10.6–18.8)	1.15 (1.05–1.25)†	1.14 (1.04–1.24)†	4.5 (2.3–6.7)
Least healthy lifestyle	1.63 (1.49–1.79)*	1.52 (1.39–1.67)*	14.1 (11.3–17.1)	1.47 (1.35–1.62)*	1.45 (1.32–1.59)†	4.9 (3.1–6.7)
Score 2
Most healthy lifestyle	1 (reference)	1 (reference)	…	1 (reference)	1 (reference)	…
Moderate healthy lifestyle	1.23 (1.13–1.34)*	1.21 (1.11–1.31)*	11.9 (9.1–14.7)	1.20 (1.10–1.31)*	1.19 (1.10–1.30)*	3.7 (2.1–5.4)
Least healthy lifestyle	1.73 (1.59–1.88)*	1.64 (1.50–1.78)*	10.1 (7.9–12.2)	1.58 (1.46–1.72)*	1.56 (1.44–1.70)*	3.5 (2.1–4.9)

Model 1: adjusted for age, sex, race, and TDI. Model 2: adjusted for age, sex, race, TDI, smoking status, BMI, waist circumference, total cholesterol, triglycerides, hypertension, diabetes, AST, ALT, bilirubin, serum creatinine, and statin use. ALT indicates alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index, FLI, fatty liver index; MACEs, major adverse cardiovascular events; and TDI, Townsend deprivation index.

^* P<0.001.

^† P<0.01.

The statistical analysis of all‐cause death included 130 144 participants, of which 6697 died. The median follow‐up time was 11.54 years. For all‐cause death, the HRs without adjustment for FLI were also larger. When adjusting for only some cardiovascular risk factors (age, sex, race, and TDI) and FLI, the HRs of the least healthy overall lifestyle group were 1.52 and 1.64 (95% CI, 1.39–1.67 and 1.50–1.78; all P<0.001) compared with the most healthy overall lifestyle group. The proportions of the mediation effect of NAFLD were 14.1% and 10.1% (95% CI, 11.3–17.1 and 7.9–12.2) for score 1 and score 2, respectively. After adjusting for FLI and all traditional cardiovascular risk factors, the HRs of the least healthy overall lifestyle group were 1.45 and 1.56 (95% CI, 1.32–1.59 and 1.44–1.70; all P<0.001) compared with the most healthy overall lifestyle group. The proportions of the mediation effect of NAFLD were 4.9% and 3.5% (95% CI, 3.1–6.7 and 2.1–4.9) for score 1 and score 2, respectively (Table 2). The above results indicated that in different lifestyle groups, the association of lifestyle and healthy outcomes were mediated by NAFLD, but the mediating effect was significantly reduced after considering traditional cardiovascular risk factors, and the mediating effect on MACE outcome was further reduced. Similar patterns were found in the analysis of diet, physical activity, and sleep habits (Table S1 through S3).

Effect of Lifestyle With MACEs and All‐Cause Death in NAFLD and Non‐NAFLD Groups

Compared with those in the least healthy overall lifestyle group, individuals with and without NAFLD in the most healthy overall lifestyle group had lower risk of MACEs, with the HRs of 0.68 (95% CI, 0.57–0.81; P<0.001) and 0.60 (95% CI, 0.53–0.68; P<0.001), respectively, in score 1, and 0.69 (95% CI, 0.61–0.78; P<0.001) and 0.64 (95% CI, 0.51–0.80; P<0.001) in score 2 (Figure 1A). For all‐cause death, individuals both with and without NAFLD with the most overall healthy lifestyle had a lower risk of all‐cause death compared with those with the least healthy overall lifestyle, with HRs of 0.78 (95% CI, 0.65–0.94; P<0.001) and 0.66 (95% CI, 0.59–0.73; P<0.001), respectively, in score 1, and 0.66 (95% CI, 0.55–0.79; P<0.001) and 0.63 (95% CI, 0.57–0.70; P<0.001) in score 2 (Figure 1B). Regarding the risks of MACEs and all‐cause death, a healthy lifestyle was more beneficial to the participants without NAFLD (all P for interaction <0.05). Further analysis indicated that improvements in diet, physical activity, and sleep habits were all beneficial in reducing the risks of MACEs and all‐cause death. Improvements in diet (P for interaction <0.01) and physical activity (P for interaction <0.05) were more beneficial for participants without NAFLD, while improvements in sleep habits (P for interaction <0.01) were more beneficial for participants with NAFLD (Figure S1 through S3).

Figure 1. Associations of lifestyle with MACEs and all‐cause death stratified by NAFLD status.

A, MACEs; B, All‐cause death. “NAFLD group” is defined as participants with FLI ≥60. “Least healthy overall lifestyle,” “moderate healthy overall lifestyle,” and “most healthy overall lifestyle” corresponded to the highest, middle, and lowest tertiles of lifestyle score, respectively. Hazard ratios were adjusted for age, sex, race, TDI, smoking status, BMI, waist circumference, total cholesterol, triglycerides, hypertension, diabetes, AST, ALT, bilirubin, serum creatinine, and statin use. Individuals in the non‐NAFLD group with the least healthy overall lifestyle were used as the reference group for score 1 (#). Individuals in the NAFLD group with the least healthy overall lifestyle were used as the reference group for score 1 (*). Individuals in the non‐NAFLD group with the least healthy overall lifestyle were used as the reference group for score 2 (§). Individuals in the NAFLD group with the least healthy overall lifestyle were used as the reference group for score 2 (¶). ALT indicates alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; FLI, fatty liver index; HR, hazard ratio; MACEs, major adverse cardiovascular events; NAFLD, nonalcoholic fatty liver disease; and TDI, Townsend deprivation index.

Joint Analysis of Lifestyle and NAFLD With MACEs and All‐Cause Death

Table 3 lists the incidence and per 1000 person‐years of MACEs and all‐cause death events in different participant groups. With the aggravation of unhealthy lifestyle (least healthy overall lifestyle group) and the occurrence of NAFLD, the proportion of new events and incidence rate per 1000 person‐years had an increasing relationship. Figures 2 and 3 show the joint association of lifestyle and NAFLD on the MACEs and all‐cause death. Aggravation of an unhealthy lifestyle (least healthy overall lifestyle group) and the occurrence of NAFLD increased the risk of MACEs. The HRs of MACEs were 1.73 in score 1 (95% CI, 1.49–2.00; P<0.001) and 1.60 in score 2 (95% CI, 1.38–1.86; P<0.001) among the participants with NAFLD with the least healthy lifestyle compared with those without NAFLD with the most healthy lifestyle. Attributable risk had an increasing trend. Participants with NAFLD and the least healthy lifestyle had attributable risks of MACEs that were 5.21% (score 1) and 4.64% (score 2) higher than those without NAFLD and the most healthy lifestyle (Figure 2A and 2B). For all‐cause death, compared with the participants without NAFLD with the most healthy lifestyle, participants with NAFLD and the least healthy lifestyle had HRs of 1.69 in score 1 (95% CI, 1.49–1.91; P<0.001) and 1.73 in score 2 (95% CI, 1.54–1.95; P<0.001). The attributable risk also showed an increasing trend in all‐cause death. The participants with NAFLD and the least healthy lifestyle had 6.45% (score 1) and 6.57% (score 2) higher risk of all‐cause death than the participants without NAFLD and the most healthy lifestyles (Figure 3A and 3B).

Table 3.

Incidence of Outcome Events by Unhealthy Lifestyle and NAFLD Subgroups

	Non‐NAFLD group			NAFLD group
	Most healthy lifestyle	Moderate healthy lifestyle	Least healthy lifestyle	Most healthy lifestyle	Moderate healthy lifestyle	Least healthy lifestyle
MACEs
Score 1
No. of participants	28 474	58 526	17 010	4926	16 962	8718
No. of events (%)	470 (1.65)	1526 (2.61)	644 (3.79)	176 (3.57)	826 (4.87)	562 (6.45)
IR per 1000 person‐years	1.45	2.31	3.39	3.18	4.38	5.89
Person years	324696.30	661123.90	189877.80	55406.68	188624.10	95357.92
Score 2
No. of participants	19 383	46 356	37 026	2855	11 050	15 405
No. of events (%)	307 (1.58)	1069 (2.31)	1205 (3.25)	85 (2.98)	481 (4.35)	907 (5.89)
IR per 1000 person‐years	1.39	2.04	2.9	2.64	3.9	5.34
Person years	221050.10	525218.60	415639.80	32188.63	123409.10	169 723
All‐cause death
Score 1
No. of participants	13 530	56 593	29 731	2232	14 416	13 642
No. of events (%)	445 (3.29)	2314 (4.09)	1684 (5.66)	126 (5.65)	905 (6.28)	1223 (8.96)
IR per 1000 person‐years	2.87	3.58	4.98	4.98	5.53	7.99
Person years	154884.10	646873.80	338019.90	25311.70	163673.70	153039.50
Score 2
No. of participants	17 701	45 811	35 119	3015	11 667	14 280
No. of events (%)	539 (3.05)	1790 (3.91)	2015 (5.74)	139 (4.61)	687 (5.89)	1264 (8.85)
IR per 1000 person‐years	2.66	3.41	5.05	4.05	5.18	7.88
Person‐years	202829.40	524225.60	398979.10	34344.75	132628.90	160405.50

IR indicates incidence rate; MACEs, major adverse cardiovascular events; and NAFLD, nonalcoholic fatty liver disease.

Figure 2. Joint effect of lifestyle and NAFLD on MACEs.

A, MACEs for score 1; B, MACEs for score 2. AR% showed an increasing trend in the remaining 5 groups compared with the most healthy overall lifestyle with the individuals without NAFLD. Hazard ratios were adjusted for age, sex, race, TDI, smoking status, BMI, waist circumference, total cholesterol, triglycerides, hypertension, diabetes, AST, ALT, bilirubin, serum creatinine, and statin use. Individuals with non‐NAFLD status and most healthy overall lifestyle were used as the reference group. P‐interaction describes the interaction between lifestyle and NAFLD with MACEs. *P<0.001. ^† P<0.01. ^‡ P<0.05. ALT indicates alanine aminotransferase; AR, attributable risk; AST, aspartate aminotransferase; BMI, body mass index; HR, hazard ratio; MACEs, major adverse cardiovascular events; NAFLD, nonalcoholic fatty liver disease; and TDI, Townsend deprivation index.

Figure 3. Joint effect of lifestyle and NAFLD on all‐cause death.

A, All‐cause death for score 1; B, all‐cause death for score 2. AR% showed an increasing trend in the remaining 5 groups compared with the most healthy overall lifestyle with the individuals without NAFLD. HRs were adjusted for age, sex, race, TDI, smoking status, BMI, waist circumference, total cholesterol, triglycerides, hypertension, diabetes, AST, ALT, bilirubin, serum creatinine, and statin use. Individuals with non‐NAFLD status and most healthy overall lifestyle were used as the reference group. P‐interaction describes the interaction between lifestyle and NAFLD with all‐cause death. *P<0.001. ^† P<0.01. ALT indicates alanine aminotransferase; AR, attributable risk; AST, aspartate aminotransferase; BMI, body mass index; HR, hazard ratio; MACEs, major adverse cardiovascular events; NAFLD, nonalcoholic fatty liver disease; and TDI, Townsend deprivation index.

Discussion

The present large prospective UK cohort study found that NAFLD played a mediating role in the association between unhealthy lifestyle and MACEs as well as all‐cause death. Without adjusting for traditional cardiovascular risk factors, NAFLD can mediate about 20% of the association between lifestyle and MACEs, but after adjusting for traditional cardiovascular risk factors, the mediating effect decreases rapidly to having a maximum of 3.5%. For all‐cause death, without adjusting for traditional CVD risk factors, NAFLD can exert a proportional mediating effect of 15% to 20% on physical activity lifestyle choices, but of <10% on other scores. After adjusting for traditional cardiovascular risk factors, the proportional risk mediated by NAFLD was around 4%, with an effect of up to 6.6% on physical activity lifestyle choices. There was an additive effect of unhealthy lifestyle and NAFLD on MACEs and death, with the risks of MACEs and death being highest in those with the worst lifestyles and NAFLD.

Metabolic syndrome is recognized as an important risk factor for CVD. Each component of metabolic syndrome is an independent risk factor for CVD, and the combination of these risk factors increases the incidence and severity of CVD. ³¹ Considering that NAFLD is strongly associated with cardiometabolic risk factors included in metabolic syndrome, such as obesity, elevated blood glucose, dyslipidemia, and elevated blood pressure, it is not surprising that NAFLD is considered a potential risk factor for CVD. ³² , ³³ However, although some studies have observed that the presence of liver disease is a predictor of cardiovascular events, the relationship with traditional cardiovascular risk factors has been overlooked. ³⁴ The key question of whether liver disease poses any additional cardiovascular risk remains unclear. In the present study, we found that NAFLD was an independent mediator of lifestyle effects on CVD, but a more interesting finding was that most of its mediating effect acted via traditional cardiovascular risk factors. The mediating effect of NAFLD on all‐cause death was smaller than on the MACEs outcome, while the effect of traditional cardiovascular risk factors was also smaller (the proportion of mediating effect was reduced less after adjusting for traditional cardiovascular risk factors in all‐cause death). The difference in the mediating effect of the 2 outcomes also confirmed the conclusion that cardiovascular death was the most important cause of death in patients with NAFLD, which has also been found in other observational studies. ³² , ³⁵

Moreover, the mediating effect analysis in our study also indicated that NAFLD mediated the association between unhealthy lifestyle and increased CVD risk due to traditional cardiovascular risk factors, but a small contribution was attributable to other nonmetabolic factors, such as liver disease. This finding seems to explain some of the controversial aspects of previous research. A study that used electronic records from 4 large European primary health care databases did not find any significant associations between NAFLD diagnosis records and the risk of developing myocardial infarction and stroke after adjusting for traditional cardiovascular risk factors. ³⁶ A cohort study that collected information on 285 adults in the United States found an association between advanced liver fibrosis and coronary heart disease that persisted after adjusting for traditional cardiovascular risk factors. ³⁷ Similar conclusions were drawn on the basis of a multinational cohort study of 458 participants, which supported the results of our study. ³⁸

Previous studies have found that lifestyles represented by diet and physical activity are associated with cardiovascular and all‐cause death in patients with NAFLD, and unhealthy diet and sedentary behavior have adverse effects on all‐cause and cardiovascular death in patients with NAFLD. ⁵ , ³⁹ Further research suggested that in most cases, the intake of anti‐inflammatory and antioxidant‐rich diets and engaging in sufficient physical activity to control BMI or obesity play a crucial role in reducing cardiovascular and all‐cause death. ⁴⁰ , ⁴¹ , ⁴² Therefore, improvements in diet and physical activity are considered the most cost‐effective ways to reduce cardiovascular risk and are also generally recommended as first‐line treatments for NAFLD in the absence of drug therapy. ⁴³ The present study further confirmed the protective effect of a healthy diet, physical activity, sleep habits, and overall lifestyle on cardiovascular events and all‐cause death in the population with NAFLD. We found that improvements in overall lifestyle, dietary habits, and physical activity all showed a positive synergistic effect with individuals without NAFLD; that is, implementing lifestyle improvements in individuals without NAFLD can reduce the occurrence of MACEs and all‐cause death to a greater extent. Conversely, for those with existing NAFLD, the effects of enhancing overall lifestyle, dietary habits, and physical activity may not be as anticipated. This does not imply that the health effects brought about by the above interventions are necessarily greater in the non‐NAFLD population; it is merely a manifestation of the overall effect within the larger population. As NAFLD progresses, the challenge of reversing its status through interventions becomes more challenging, accompanied by changes in body function and organic changes. ⁴⁴ , ⁴⁵ The healthy benefits brought about by the mentioned interventions might be overshadowed. However, it is pivotal to emphasize the importance of early lifestyle improvements in reducing the risks of MACEs and all‐cause death, whether the individual has NAFLD or not. On the other hand, the protective effect of healthy sleep habits on the individuals with NAFLD was stronger than that in the non‐NAFLD group, which indicated that improving sleep quality has greater health benefits for the individuals with NAFLD. Although sleep is often a neglected lifestyle habit in CVD and NAFLD research, more evidence appears to be improving the situation. ⁴⁶ To promote the concept of health from disease treatment to early health intervention, the American Heart Association proposed 7 elements of cardiovascular/life health in 2010 and classified the 7 elements into health behaviors and health factors. Modifiable health behaviors were significantly and substantially associated with longer life span and CVD‐free survival and included 3 components: smoking, physical activity, and diet—specifically, less exposure to cigarettes; greater intensity or duration physical activity; and more fruits and vegetables, legumes, and grains and less salt, red meat, processed meat, and sugar‐sweetened beverages are better for cardiovascular health. ⁴⁷ In 2022, the agency updated the definition of healthy behavior. Sleep was added as a new healthy behavior; 7 to 9 hours of sleep is considered healthy sleep, while too long or too short sleep can increase the risks of coronary heart disease and all‐cause death. ⁴⁸ Similarly, the 2021 European Society of Cardiology also highlighted the role of improving and optimizing sleep habits in preventing cardiovascular disease. ²⁴ These guidelines also support the conclusions of the present study. Improving sleep quality in those with NAFLD could be a new target for preventing MACEs and all‐cause death.

Overall, the available evidence not only supports the association between NAFLD and both MACEs and all‐cause death but also supports the perspective that the joint effect of NAFLD and unhealthy lifestyle could further elevate the risks of MACEs and all‐cause death. These findings may have important implications for future decision‐making in both public health and clinical practice and underscore the importance of cardiovascular risk monitoring, lifestyle modification, and liver disease treatment in patients with NAFLD. A large contribution to the risk of an unhealthy lifestyle combined with NAFLD comes via traditional cardiovascular risk factors, and lifestyle improvements can also improve NAFLD‐related glycemia, lipids, blood pressure, and obesity. A higher‐quality diet (eg, Mediterranean diet), guideline‐recommended levels of physical activity (general health guidelines recommend at least 150 minutes of moderate‐intensity leisure‐time physical activity per week or 10 000 steps per day for primary prevention of CVD), reduced risk of metabolic disease, and adequate sleep (sufficient time and higher quality) will help reduce the risks of CVD and all‐cause mortality through improving NAFLD. ⁴⁹ , ⁵⁰ , ⁵¹ Given that lifestyle improvement has become a crucial target in preventing and controlling CVD and related adverse events, our findings offer evidence from a large‐scale population to support the “environment‐sleep‐emotion‐exercise‐diet” intervention model proposed by previous researchers and the Life's Essential 8 advocated by the American Heart Association. Additionally, our study extends the applicability of these conclusions to the NAFLD population, highlighting their relevance in this specific context. ⁴⁸ , ⁵² At the same time, we strongly recommend that adult patients with NAFLD should at least be included in the primary prevention strategies for CVD to address the global NAFLD epidemic.

It is undeniable that this study also had some limitations. First, FLI is not the preferred criterion for evaluating NAFLD. However, considering that the large‐scale implementation of liver biopsies is difficult to achieve, and the accuracy of ultrasound is greatly affected by doctors and equipment, FLI is currently the best choice for studying NAFLD based on large‐scale populations. Second, although the correlation coefficient matrix of all behavioral variables (the coefficients of the study variables were all <0.5) supports the assumption of variable independence, the interactions among the lifestyle variables were not fully considered. Third, although prospective cohort studies are a research method that produces higher levels of evidence, they still cannot fully prove causality as in randomized controlled trials, and there may be uncontrolled confounding factors. Finally, our conclusions may need to be further validated in other large prospective cohort studies in the future, and evidence from the laboratory is also necessary. However, we currently do not have data from other similar large‐scale prospective cohort studies like UKB to validate this. The conclusions of this study still need to be further demonstrated in a research design that can produce a higher level of evidence.

Conclusions

NAFLD played a mediating role in the effects of diet, physical activity, sleep, and overall lifestyle habits on CVD and all‐cause death risks. Most of the mediating effects were derived from traditional cardiovascular risk factors, but there was still an independent mediating effect. Active adjustment of metabolic risk factors coexisting with CVD should therefore be considered in the management of patients with NAFLD. The attributable risks of CVD and all‐cause death and the person‐year incidence rate increased gradually after patients with NAFLD were classified by the degree of unhealthy lifestyle, which highlights the importance of lifestyle improvement for the whole population, especially in patients with NAFLD and an unhealthy lifestyle.

Sources of Funding

None.

Disclosures

None.

Supporting information

Tables S1–S3

Figures S1–S3