The interactive effect of diabetes and central obesity on stroke: a prospective cohort study of inner Mongolians

All participants were at least 20 years of age and were recruited from 32 villages in two neighboring townships located in the counties of Kezuohou Banner and Naiman Banner in Inner Mongolia. Most of the residents of these townships are Mongolians who have lived there for many generations and maintain a traditional diet and lifestyle. At baseline, in 2003, this study included 2,589 individuals (1064 males and 1525 females) all without previous cardiovascular disease (CVD) including stroke. The subjects were selected from the 32 villages. The selection criteria were to meet all of the followings: (1) age: ≥ 20 years, (2) ethnicity: Mongolian. There were a total of 3,475 eligible residents in the study fields. The exclusion criteria were to meet one of the followings: (1) self-reported history of CVD, stroke, or tumors, (2) taking antihypertensive medications, (3) being pregnant. The 886 people who refused to participate or met the exclusion criteria were excluded. In 2012, 2,583 individuals (99.8%) were successfully contacted to provide comprehensive health information. All participants provided written informed consent. The present analysis is based on the baseline and follow-up examinations. The ethics committee at Soochow University in China approved this study.

Participants underwent a thorough physical examination at baseline and the last year of follow-up where anthropometric information, blood pressure measurements, and blood samples were obtained. Data on demographic information, lifestyle risk factors, family history of CVD, and personal medical history were gathered from standard questionnaires written in Chinese and administered by trained staff. Smoking and drinking were two lifestyle risk factors that were pertinent to this study. An individual was classified as current cigarette smoker if they smoked at least 1 cigarette per day for 1 year or more leading up to the start of the study. An individual was classified as current drinker if they consumed any type of alcoholic beverage at least once per week during the last three years.

A physical examination was then administered. After the participants had been resting for 5 minutes, they remained seated and three consecutive blood pressure measurements (3 minutes between each) were made with a standard mercury sphygmomanometer and an appropriately sized cuff [12]. The first and fifth Korotkoff sounds were recorded as systolic and diastolic blood pressure, respectively. The mean of the 3 physician-obtained easurements constituted the reported blood pressure used in further analyses. In our study, hypertension was defined as systolic blood pressure (SBP) of at least 140 mmHg and/or diastolic blood pressure (DBP) of at least 90 mmHg. Body weight and height were measured using standard methods, and body mass index (BMI) was calculated as weight in kilograms divided by the square of the height in meters (kg/m²). Waist circumference (WC) was measured 1 cm above the umbilicus. Women with a WC greater than 80 cm and men with a WC greater than 85 cm were classified as centrally obese [13].

Blood samples were obtained by venipuncture in the morning after a requested overnight fast (at least 8 hours). All plasma and serum samples were frozen at −80°C until laboratory testing. Fasting plasma glucose (FPG) was measured using a modified hexokinase enzymatic method [14]. Diabetes was defined as one of the following: (1) FPG ≥ 7 mmol/L (2) self-reported history of diabetes or (3) current use of either insulin or oral diabetes medication [15]. Serum total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), and triglycerides (TG) were assessed enzymatically using commercial reagents [14]. Low-density lipoprotein cholesterol (LDL-C) concentration was calculated by means of the Friedewald equation for participants who had less than 400 mg/dL TG [16].

Participant follow-up was executed between June 2003 and July 2012. In this study, stroke as defined by both ischemic and hemorrhagic stroke was the event of interest. Stroke was defined as a sudden critical onset of neurological symptoms lasting at least 24 hours [17]. Participants were also diagnosed by cranial computed tomography or magnetic resonance imaging (MRI). Trained staff interviewed either the participants or their relatives, if participants were dead or unable to communicate, every two years to find new stroke cases. When a new case was found during follow-up, the staff reviewed the hospital records and completed a standard event form. An end point review committee made the final decisions regarding a participants’ stroke diagnosis.

Baseline characteristics were analyzed for the total population. In the results continuous variables that were not normally distributed were expressed as median and interquartile range, normally distributed continuous variables were expressed as mean ± standard deviation, and binary variables were expressed as frequency and percent.

Participants were then divided into four categories: individuals having neither risk factor, only diabetes, only central obesity, or both risk factors. The distributions of other conventional risk factors were compared across the four subgroups mentioned. Normally distributed continuous variables were assessed by ANOVA, continuous variables with a skewed distribution were assessed by Wilcoxon rank-sum test, and categorical variables were assessed by a Chi-square test. Univariate and multivariate Cox proportional hazard models were utilized to calculate the hazard ratios (HRs) and 95% confidence intervals (95% CI) for each category compared to individuals with neither risk factor. A multiplicative interaction term of diabetes and central obesity was set in the models to test the presence of an effect. In the multivariate models, some important confounders at baseline such as age, gender, TC, TG, family history of CVD, smoking, drinking, and hypertension were all included as covariates. The biological additive interaction between diabetes and central obesity on stroke was evaluated by three indexes: relative excess risk because of interaction (RERI), attributable proportion because of interaction (AP), and synergy index (S) [18]. If there was no biological interaction the 95% CI of RERI and AP would include 0 and the 95% CI of S would contain 1. Finally, the yearly cumulative incidence of stroke among the four subgroups was estimated using Kaplan-Meier survival curves and compared using the log-rank test. A two-tailed P value < 0.05 was considered statistically significant. All analyses were completed using SAS 9.1 software.