Study Population
This cross-sectional study was performed in the city of Hangzhou, the capital of Zhejiang Province in the people’s Republic of China from June 2015 to December 2016. The detailed description of this study has been reported previously[14]. A total of 2437 eligible participants were invited to attend the health checks at the Medical Center for Physical Examination, Zhejiang Hospital, where the participants were face-to-face interviewed by well-trained dietitians using written questionnaires. We excluded 453 participants with self-reported having diabetes,which may change their lifestyles and dietary habits. We additionally excluded 15 participants with a family history of diabetes.Further exclusion included 36 participants who provided incomplete anthropometric information, 138 participants who provided the missing information on their dietary intake, 14 participants who reported implausible energy intakes (<600 or >4000kcal/d), 20 participants who didn’t provide blood sample. After exclusions, the final analysis was conducted on data from 1761 participants (914 male and 847 female). All participants provided written informed consent when recruited for the study.
Dietary assessment
Dietary information was collected by well-trained dietitians, using a 138-item, validated semi- quantitative food frequency questionnaire(SQFFQ), which was designed to measure the dietary intake in middle-aged Chinese adults.The validity of reliability of SQFFQ has been previously validated[26]. Participants were requested to recall the frequency of 138 food items in the preceding year and the estimated portion size, using local weight units(1 Liang=50g) or natural units(cups).For each food item, the consumption frequency was reported using nine categories in this questionnaire, ranging from “never”, “< 1 time/month”, “1-3 times/month”, “1-2 times/week”, “3-4 times/week”, “5-6 times/week”, “1 time/day”, “2 times/day” to “3 times/day”. During face-to-face interviewing, the portion size of each food item was estimated using food models and standard serving sizes(e.g., one standardized portion of cooked rice is one small bowl, weighing approximately 100g).Then, the frequency of intake and portion size were used to calculate the amount of each food item consumed on average, using the Chinese Food Composition Table as the database. Finally, these data were converted to grams or millilitres per day and used in the following analysis.
Identification of dietary patterns
First, we collapsed 138 food items from the SQFFQ into 30 predefined food groups (Table S1) based on similarities in ingredients, nutrient profile, and culinary usage in a middle-aged Chinese population [14]. Before performing the factor analysis, the Kaiser-Meyer-Olkin Measure of Sample Adequacy and the Bartlett Test of Sphericity were used to assess data adequacy. If appropriate, factor analysis (principal component method) was used to derivate the major dietary patterns. During data analyses, the factors were rotated by orthogonal transformation(varimax rotation) to achieve simpler structure with greater interpretability; The eigenvalue and scree plot were applied to decide which factors to be remained[27]. After evaluating the eigenvalues, the scree plot test, and interpretability, factors with an eigenvalues ≥2.0 were retained. In the present study, the predefined food groups with a factor loading ≥|0.4| were considered as major contributors to a give dietary pattern. The labeling of dietary patterns was based on the interpretation of foods with high factor loadings for each dietary pattern[27]. Finally, dietary pattern scores were categorized according to quartiles with Q1 corresponding to the lowest quartile of dietary pattern score.
In our previous study[14], three dietary patterns have been extracted, naming the traditional southern Chinese pattern, which was characterized by high intakes of refined grains, vegetables, fruits, pickled vegetables, fish and shrimp, bacon and salted fish, salted and preserved eggs, milk, soya bean and its products, miscellaneous bean, fats, drinks; the Western pattern was characterized by high intakes of red meats, poultry and organs, processed and cooked meat, eggs, seafood, cheese, fast foods, snacks, chocolates, alcoholic beverages, coffee; the grains-vegetables pattern was characterized by high intakes of whole grains, tubers, vegetables, mushrooms,vegetable oil, nuts, honey, tea.
Assessment of blood pressure
Blood pressure was measured twice by a trained nurse after a 5-10 minutes rest in the sitting position, using a standard mercury sphygmomanometer. The mean of two measurements was considered as the participant’s final blood pressure.
Assessment of biomarker
All blood samples were drawn between 7:00 and 9:00 in the morning into evacuated tubes from each participant after 12h of fasting overnight, and stored temporarily at –20°C until subsequent analyses. Then, to measure 2-h plasma glucose (2-h PG), participants underwent an oral glucose tolerance test using 75 g of glucose. All blood samples were analyzed in the Medical Center for Physical Examination, Zhejiang Hospital for fasting plasma glucose(FPG), 2h PG, Glycosylated Hb (HbA1c), total cholesterol(TC), triglycerides(TG), high-density lipoprotein-cholesterol(HDL-C), low- density lipoprotein-cholesterol(LDL-C), serum uric acid(SUA), alanine aminotransferase(ALT) and asparagine aminotransferase(AST) using the Hitachi 7180 automatic biochemical analyzer (Hitachi, Tokyo, Japan).
Assessment of anthropometric measurements
Weight in light clothes and without shoes was measured with an accuracy of 100g with a digital scale, and height was measured using a stadiometer with an accuracy of 0.1cm. Body mass index(BMI) was calculated using weight(kg) divided by the square of height(m2). Waist circumference(WC) was measured at the midpoint between the lower rib edge and the upper iliac crest by means of a metric measure with an accuracy of 1 mm[28]. All anthropometric measurements were conducted by well- trained nurse according to standard procedures.
Assessment of other variables
Physical activity was assessed using the International Physical Activity Questionnaire (IPAQ), and expressed as metabolic equivalent hours per week(MET-h/week)[29]. Then, based on MET-h/week, three categories of physical activity were assigned, including light, moderate and heavy[30].Additional information including smoking habits (never, current, and former smokers), education level(primary school or below, middle and high school, junior college or above) was recorded with a structured questionnaire. Moreover, total energy intake was estimated through this semi-quantitative FFQ, and results were expressed in kilocalorie per day (kcal/day) and categorized according to quartile.
Definition of prediabetes
Prediabetes was defined as those without a previous diabetes diagnosis and the satisfaction of at least one of three conditions:(1)FPG test with values between 5.6 and 6.9mmol/L;(2)HbA1c of 5.7- 6.4%;(3)the 2-h oral glucose tolerance test(OGTT) with values between 7.8 and 11.0 mmol/L [31].
Statistical Analyses
Data were analyzed across the quartiles of each dietary pattern score, and results for categorical variables were expressed as number (percentages), and continuous variables were expressed as mean ±standard deviation(SD). First, Kolmogorov-Smirnov test was used to test the normal distribution of the variables. The continuous variables with normal and non-normal distribution were compared using the Independent-Samples t and Mann-Whitney tests, respectively. In addition, we used the chi-squared test to assess the difference in categorical variables. Multivariate logistic regression analysis was performed to evaluate the relationship between dietary patterns and prediabetes, adjusting for potential confounders. Model 1 was adjusted for age (years) and sex(male/female); Model 2 was further adjusted for educational level(<high school, high school, >high school), smoking status (never, current, former), physical activity (MET-h per week), hypertension(yes/no) and BMI(continuous); Model 3 was additionally adjusted for total energy intake(kcal/d). All statistical analyses were conducted with the use of the IBM Statistical Package SPSS version 23.0(SPSS Inc, Chicago, IL, USA), and a 2-tailed P<0.05 was considered statistically significant.