Study Subjects and Design
This was a cross-sectional study, conducted in the health examination center of the Affiliated Nanping First Hospital, Fujian Medical University from April 2015 through August 2017. As shown in Figure 1, we included individuals aged ³45 years old and who permanently resided in Nanping. Further exclusion criteria in our study were as follows: (A) Individuals whose daily consumptions of alcohol >40g (men) or >20g (women). (B) Individuals who had any other liver disease history, such as drug-induced liver disease, viral hepatitis, Wilson’s disease, autoimmune hepatitis and total parenteral nutrition. (C) Individuals who were taking hypolipidemic or diet pills. (D) Individuals who did not answer more than 25 food-related items on the questionnaire. (E) Individuals who did not provide information on smoking and tea consumption. All subjects provided their informed consent before participating in this study.
The current study was carried out in compliance with the Declaration of Helsinki, and the Ethics Committee of Fujian Medical University approved the study protocol (ethics number 2014096).
Data Collection
NAFLD ascertainment
NAFLD was diagnosed by abdominal ultrasonography using established criteria [26]. An abdominal ultrasonography examination was done by experienced radiologists who were unaware of the laboratory and clinical data.
Meat intakes assessment
Dietary information on typical food consumption of participants was collected using a semi-quantitative food frequency questionnaire consisting of 110 food-related items, which was developed and validated in a sample from southern China [27]. And the information was obtained from participants interviewed face to face by trained investigators. For each food item, participants used the following response options to indicate how often they ate the selected food on average: 3 times/day, twice/day, once/day, 5–6 times/week, 3–4 times/week, 1–2 times/week, 1–3 times/month, <once/month and rarely. Red meat consisted of pork, beef and lamb. Poultry was composed of chicken and duck. Processed meat contained fried and smoked meat. Fish included: fish, shellfish and crab. The nutritional components of each food item were taken from the China Food Composition [28].
Covariate assessment
The following variables were self-reported: age, sex, marital status, income, educational level, smoking status, tea intake status, physical activity, medication use and medical conditions. All subjects underwent physical examinations (height, weight, waistline, hipline and blood pressure) and blood tests (fasting plasma-glucose, low-density lipoprotein cholesterol, total cholesterol, TG, HDL-C, AST, ALT and GGT) performed by trained physicians. BMI was calculated as weight/ (height) 2. Participants with a systolic blood pressure ³140 mmHg or diastolic blood pressure ³90 mmHg were defined as having hypertension. And subjects who had one or more of the following abnormalities were diagnosed as dyslipidemia: total cholesterol ³6.2 mmol/L, TG >2.25 mmol/L, low-density lipoprotein cholesterol >4.13 mmol/L or HDL-C <1.03 mmol/L [29]. Diabetes was diagnosed as follows: fasting plasma glucose of 7.0 mmol/L or greater or 2-hour postprandial glucose greater than or equal to 11.1 mmol/L.
Statistical Analysis
Participants were classified into 4 groups based on the quartile of total meat consumption. The baseline characteristics of subjects were analyzed using the Nonparametric Kruskal-Wallis test for non-normal continuous variables and Chi-Square test for nominal variables. Continuous variables were expressed as median (interquartile range, IQR). Propensity scores were used to explain potential confounders by observed characteristics at baseline [30]. Age, gender and BMI were used to calculate the propensity score. Inverse probability of treatment weighting analysis was performed to evaluate associations of red meat, processed meat, poultry and fish intakes with NAFLD, adjusting for smoking status, tea intake status, weekly hours of physical activity, and presence of hypertension, dyslipidemia and diabetes. The lowest quartile (Q1) of each type of meat intake was served as the reference group and P for trend was calculated by setting the meat intake quartiles as a continuous variable. To evaluate dose-response relationships between continuous exposure variables (red meat, processed meat, poultry and fish intakes) and NAFLD, a logistic model with restricted cubic spline using five knots at 0.05, 0.275, 0.5, 0.725 and 0.95 was built, adjusting for age, sex, BMI, smoking status, tea intake status, weekly hours of physical activity, and presence of hypertension, dyslipidemia and diabetes.
We also performed subgroup analysis to examine relationships of red meat, processed meat, poultry, and fish with NAFLD by the following subgroups: age (<60 years or ³60 years), gender (men or women), BMI (<24 kg/m2 or ³24 kg/m2), smoking status (never, former or current), tea consumption status (yes or no), hypertension (yes or no), dyslipidemia (yes or no), diabetes (yes or no), and weekly hours of physical activity (<9 hours/week or ³9 hours/week). P value for interaction was calculated. Two sensitivity analyses were conducted: (1) logistic regression analysis without IPTW; and (2) propensity score-matching logistic regressions. Furthermore, to investigate the associations of meat subtypes intakes with the concentrations of serum GGT, ALT, AST, fasting plasma glucose, total cholesterol, TG, low-density lipoprotein cholesterol and HDL-C, spearman’s rank correlation was performed.
For statistical analyses, SPSS, version 19.0.0.1(IBM SPSS, 2010, Chicago, IL, USA) and R, version 4.0.0 (R Project for Statistical Computing) were performed. All P values were two-tailed and results were considered to be statistically significant at P values<0.05.