Adherence to the Dietary Approaches to Stop Hypertension Diet and Non-alcoholic Fatty Liver Disease

Yuanyuan Sun; Shuohua Chen; Xinyu Zhao; Yanhong Wang; Yanqi Lan; Xiaozhong Jiang; Xiang Gao; Shouling Wu; Li Wang


Liver International. 2022;42(4):809-819. 

In This Article


In this cohort-based case-control study with 11 888 adults, we observed that greater adherence to DASH, including most of the DASH components except dairy, meats and beverages, was associated with a lower risk of NAFLD. The association was modified by sex, BMI, FBG, LDL-c, HDL-c, CRP and ALT.

Although studies have shown that an excess of energy alone could increase the risk of NAFLD, the relationship between diet and NAFLD development of is complex. It extends beyond total energy intake.[26,27] NAFLD is not only related to diet quantity but also related to diet quality. We used the energy-adjustment method by the residual model to control for confounding by total energy intake and remove extraneous variation due to total energy intake.[16] We observed that the participants with the highest quintile of DASH score were 20% less likely to have NAFLD relative to those with the lowest DASH score after adjusting demographic characteristics, lifestyle factors, BMI and metabolic-related factors. The conclusion was consistent after we calculated adherence using the dry weight of each DASH component. Many other dietary patterns, such as the Mediterranean diet and the ketogenic diet, have been proposed to treat NAFLD.[11] DASH and Mediterranean diet are both moderate macronutrient diets, while ketogenic diet is a typical very-low-carbohydrate diet.[28] Although the ketogenic diet has outstanding performance in weight reduction, it may lead to a low intake of fibres and water-soluble vitamins and LDL-c elevation.[29] The Mediterranean diet has long been considered the most recommended diet intervention for NAFLD. Mediterranean diet is a plant-based diet characterized by a high intake of grains, vegetables, fruits and beans, which shares many common parameters with the DASH diet.[30] Meanwhile, the Mediterranean diet is also characterized by the high intake of olive oil as the main source of fat. Considering the cooking habit and food cultural differences, the Mediterranean and ketogenic diets may be hard to generalize and apply in the Chinese population. Thus, the DASH diet may be considered as a substitution for NAFLD control and prevention.

Previous observational studies about the association between the DASH diet and NAFLD have shown inconsistent results. A cross-sectional study conducted in 3051 adults aged 40–75 years in Guangzhou, China, reported a reverse association between the DASH diet and prevalent NAFLD,[8] consistent with our results. Also, a case-control study with 102 newly diagnosed NAFLD and 204 controls found that participants in the top quartile of DASH diet score were 30% less likely to have NAFLD (OR = 0.70; 95% CI: 0.61–0.80). However, the association changed to non-significant (OR = 0.92; 95% CI: 0.73–1.12) after adjusting for dyslipidemia and BMI.[7] The protective association between adherence to DASH and NAFLD could be partially explained by the beneficial effect of the DASH diet on losing weight, which is supported by a meta-analysis of randomized controlled clinical trials published before December 2015[31] and a clinical trial published after December 2015.[6] All these studies have shown that the DASH diet has a beneficial effect on losing weight, especially in overweight and obese individuals.[6] This mechanism is also supported by the modifying effect of BMI on the associations between the DASH diet and NAFLD in our study. In the overweight/obesity participants, the reverse association between DASH and NAFLD was observed in the highest quintile group (OR = 0.77, 95% CI: 0.62–0.96).

In most scenes of subgroup analysis, we observed that the odds of the case from the crude data were similar or even higher in quintile 5 than those of quintile 1. However, after adjusting energy, the OR changed to <1 but was not statistically significant. Furthermore, ORs for quintile 5 vs quintile 1 were not only lower than 1 but also significant after we further adjusted BMI, physical activity, TG and CRP separately based on the energy-adjusted model. These findings prompted confounding from energy intake, BMI, physical activity, TG and CRP might be responsible for the association between DASH and NAFLD.

Subgroup analysis showed a different association between DASH and NAFLD according to sex. The protective association was observed in quintile 4–5 groups in women but not in men. These could be explained by the higher energy intake in quintile 4 and quintile 5 groups in men.

In our study, the protective effect of DASH on NAFLD was only observed in the subjects with CRP > 2.0 mg/L. Compared to the lowest quintile, the OR (95% CI) was 0.56 (0.40–0.78) for the highest quintile group. The beneficial effect of the DASH diet on the improvement of inflammation can explain this result. Our former study from the same cohort has reported a positive association between CRP level and development of NAFLD.[32] A meta-analysis from six randomized clinical trials has reported that the DASH diet significantly decreased the CRP level (mean difference: −1.01, 95% CI: −1.64 to −0.38) compared to usual diets, and the reduction of CRP levels was more significant in trials lasting for 8 weeks or longer.[33] Thus, the protective effect of the DASH diet on NAFLD may be related to the beneficial impact of DASH on the reduction of CRP.

Glucose and lipids metabolism disorder has been recognized as the main pathogenesis for the development of NAFLD.[34] Previous studies have reported that the DASH diet can help improve the metabolism of glucose and lipids.[35,36] However, the stratified analysis in our study showed that the reverse association between DASH and NAFLD only existed in participants with normal FBG, HDL-c and abnormal LDL-c levels. The non-significant results in abnormal FBG, HDL-c and normal LDL-c levels can be explained by the small sample size in these stratifications.

ALT elevation has been recognized as one of the characteristics in NAFLD patients, and a clinically significant index closely related to the risk of developing end-stage liver disease.[37] The clinical trial also found that DASH diet intervention for 8 weeks can significantly reduce ALT,[6,38] which may explain the protective effect of DASH on NAFLD in those with ALT < 40 U/L.

When individual DASH items were studied, we found that adherence to all components of the DASH diet except fat was associated with low NAFLD risk. The result was consistent with previous studies which explored the association between single food groups and NAFLD.[39–44] The association between individual DASH items and NAFLD implied that the beneficial effect of the DASH diet on NAFLD was driven by the high intake of vegetables, fruits, dairy, beans, grains and low intake of meats, sodium and beverages.

There existed several limitations in this study. First, the dietary exposure assessment and the diagnosis of NAFLD were conducted at the same time in our study. However, we have excluded those who have been diagnosed with NAFLD or had no ultrasound examination history before. All the participants in this study were incident cases, which would avoid the incident-prevalent bias. Second, the diet information was collected by self-reported FFQ. Although the FFQ has been validated and applied to the Chinese national nutrition survey in 2002,[12] the FFQ, for example, test-retest precision study, was not validated in the Kailuan cohort, leading to measurement error. However, we have excluded those with impossible daily energy intake, partially avoiding the information bias. Total energy intake may also be underestimated due to the small number of food items in the FFQ. Thus, we adjusted total energy intake as a covariate in all multiple regression models to reduce the influence of the systematic error. Thirdly, we did not evaluate the beneficial effect of the DASH diet on the prevention of NAFLD progression because we did not collect information on aspartate aminotransferase and cannot calculate fibrosis markers. Fourthly, although experienced radiologists blinded to clinical presentation and laboratory findings were invited to conduct abdominal ultrasonography, the diagnosis of NAFLD was based on abdominal ultrasound. So, NAFLD may be misclassified as non-NAFLD when the liver hepatic far content is <20%[1] and underestimate the existing association. Also, all the participants underwent abdominal ultrasonography once for each follow-up due to the large population size. However, we selected the subjects diagnosed with fatty liver in 2010 and found 79.5% of them being diagnosed as fatty liver in 2014. Considering the remission rate of NAFLD (7-year remission rate of 36.4%),[45] NAFLD diagnosis in our cohort is reliable. Fifthly, although follow-up was required every 2 years for the entire cohort, the follow-up intervals were not the same for each individual. Thus, interval-censoring existed in our study. However, we furtherly compared the DASH adherence between the participants with and without each follow-up interval of 2 years (data not shown). The adherence was comparable, indicating the interval-censoring would not influence our conclusion. Finally, we cannot exclude the possibility of residual confounding by uncollected risk factors, such as the detailed disease history, which may influence the health status of participants and lead to the change of dietary pattern. However, we have excluded those who had cardiovascular disease history of controlling the possible residual confounding.

In conclusion, we found that greater adherence to the DASH diet is associated with a lower risk of NAFLD. The inverse association was observed in women and participants with overweight/obesity, normal FBG, TG, HDL-C, LDL-c, ALT and elevated CRP, which suggested that a DASH diet should be particularly recommended for these people.