Diet Quality and its Association With Nonalcoholic Fatty Liver Disease and All-cause and Cause-specific Mortality

Eric R. Yoo; Donghee Kim; Luis M. Vazquez-Montesino; Jessica A. Escober; Andrew A. Li; Sean P. Tighe; Christopher T. Fernandes; George Cholankeril; Aijaz Ahmed


Liver International. 2020;40(4):815-824. 

In This Article


The current therapeutic approach for subjects with NAFLD has predominantly focused on lifestyle modification with weight loss by diet and physical activity. In this US population-based study, we found that higher diet quality was associated with a significant reduction in all-cause mortality and cancer-related mortality only in subjects without NAFLD. Notably, this protective association between mortality and higher diet quality was not demonstrated in subjects with pre-existing NAFLD. We also found that higher diet quality was associated with a progressively lower risk of NAFLD. In particular, higher consumption of fruits and vegetables and lower consumption of sodium were associated with significantly lower odds of NAFLD. In light of these findings, primary prevention of NAFLD with high-quality diet may be a more effective and beneficial goal for patients at risk for developing NAFLD.

The WHO Global Health Observatory data in 2016 showed that 39% of men and 39% of women aged 18 years or older were overweight (BMI ≥ 25) and 11% of men and 15% of women (BMI ≥ 30) were obese; this translates to about two billion overweight adults and over half a billion obese adults worldwide.[33] The prevalence of NAFLD among the obese population ranges from 30% to 37%, and abdominal obesity with increased waist circumference is strongly correlated with NAFLD.[34,35] Recent studies have also demonstrated an association between body weight change and incidence of NAFLD[36,37] with a modest gain of 2 kg in body weight within the normal range showing increase in the risk of developing NAFLD.[38] The demographic characteristics associated with poorer diet quality were younger age, higher BMI, waist circumference, non-Hispanic black, cigarette smoking – both active and previous, and sedentary lifestyle. Therefore, the target population for improving diet quality would be subjects at high risk of developing NAFLD and those with factors associated with poor diet quality as above. Ensuring a high-quality diet in individuals who may be at high risk for developing NAFLD may be an effective approach in reducing all-cause mortality and cancer-related mortality.

The concept that improvements in diet quality over time can decrease the risk of death and that worsening diet quality can increase the risk of death has been studied previously. A recent meta-analyses showed that higher diet quality scores measured with the Alternate HEI, Alternate Mediterranean Diet, DASH score and the HEI-2010 were associated with an estimated 17%-26% reduction in risk of all-cause mortality.[39,40] A large study by Sotos-Prieto et al found consistent associations between improved diet quality over 12 years as assessed by the Alternate HEI, Alternate Mediterranean Diet, and DASH scores and a reduced risk of death in the subsequent 12 years.[41] While prior studies have assessed diet quality and mortality, we are the first to do so in the context of NAFLD. Our study supports previous findings that higher diet quality is associated with a risk reduction in all-cause mortality in the general population without NAFLD. However, our study stresses that higher diet quality and its association with risk reduction in mortality is not applicable to those with NAFLD, underscoring the need to improve diet quality prior to developing NAFLD.

The most common cause of death in subjects with NAFLD is cardiovascular disease independent of other metabolic comorbidities, and subjects with NAFLD have increased all-cause mortality compared to matched-control populations without NAFLD.[14,42,43] Cancer-related mortality is also among the top three causes of death in those with NAFLD.[44] Our study shows that higher diet quality was associated with a lower risk for all-cause mortality in subjects without NAFLD. This protective association was not extended in those with NAFLD, regardless of their diet quality. In addition, there was an association with a lower risk for cancer-related mortality in those with high diet quality for the total population and in those without NAFLD. Again, this association was not seen in those with NAFLD. Subjects with NAFLD were older and had a higher prevalence of diabetes, hypertension and hyperlipidaemia compared to subjects without NAFLD (Table S1). It is possible that the effects of HEI may be minimal in subjects who have already developed multiple unfavourable metabolic characteristics, and in turn, high diet quality may not be protective against cancer-related and all-cause mortality in this NAFLD population. Future studies should stratify subjects with NAFLD to those with or without unfavourable metabolic characteristics. When our analysis was restricted to specific causes, there was no significant difference in cardiovascular disease-related mortality based on diet quality among the total study population, in those with and without NAFLD. This lack of association between cardiovascular disease-related mortality and diet, while still controversial, echoes findings from a recent large epidemiologic cohort study that did not find associations between total fat and types of fat with cardiovascular disease, myocardial infarction and cardiovascular disease-related mortality.[45]

The strengths of our study are the wealth of demographic and metabolic variables and the large representative sample of the US population with a median 23 years follow-up. In contrast to other cycles of NHANES which lacked hepatic ultrasonographic results, the third NHANES included ultrasonography-diagnosed hepatic steatosis. These data have allowed for a more accurate diagnosis of NAFLD compared to those in recent versions of the NHANES database, which relied on non-invasive laboratory markers. Our study has several limitations. First, the association between NAFLD and the HEI was cross-sectional, so we are unable to ascertain temporal associations. Second, the third NHANES data do not provide serial abdominal ultrasonography data; therefore, we are unable to assess longitudinal changes in NAFLD status in our subjects. Third, diagnosing NAFLD with ultrasonography has limitations; while the sensitivity ranges between 75% and 91%, it is limited by its low sensitivity in subjects with less than 30% histologic steatosis. Due to the current study design, we could not assess the severity of NAFLD using semi-quantitative ultrasonographic indices, which correlate with metabolic and histological findings.[46] In this study, intra-observer and inter-observer agreement between the participating radiologists was 91.3% and 88.7% respectively. Fourth, there are potential biases arising from how data are collected in NHANES. Participants are told they will be asked questions about what they eat; this lends to the possibility of self-reporting bias on reported intake of foods as in other large-scale epidemiologic studies.[47] NHANES also uses a 24-hour dietary recall for dietary evaluation to estimate current dietary intake which has its own limitations. Also, the HEI used in this study was the earliest version and may not reflect current dietary guideline. Lastly, we were unable to obtain information specific to liver-related mortality as these data are restricted by NHANES and are not available. Access to these restricted data would provide a greater wealth of information in the context of diet quality, individual diet components and liver-related mortality based on NAFLD status.

In conclusion, high diet quality was inversely associated with NAFLD and was positively associated with a lower risk for cancer-related and all-cause mortality in those without NAFLD. Primary prevention against NAFLD with high diet quality in those without the condition may be the ideal way to reduce all-cause mortality in subjects without NAFLD and curb the rising incidence of NAFLD in the United States.