Study design and participants
As part of the data collection for the National Survey for Nutrition and Adult Chronic Disease in Inner Mongolia, which was conducted across eight monitoring sites in Inner Mongolia, multi-stage stratified cluster sampling was used to ensure a representative cross-section of the participants. The study design has been described previously [21]. The sample size was calculated using a 19.15% prevalence of hypertension as reported in the fifth health service survey in Inner Mongolia, error of 3%, a design effect of three and a non-response rate of 10%. In total, 820 individuals over 55 years old participated in the study. This survey was approved by the Ethical Committee of the National Institute for Nutrition and Food Safety of the Chinese Center for Disease Control and Prevention. Participation in this survey entailed no treatments or interventions that could impact the health of participants. All participants provided written informed consent before the start of the investigation.
Data collection
Weighing method and 24-hour recall surveys
A weighing method and 24-hour recall surveys administered over 3 consecutive days were used to collect dietary data. Information about consumption of condiments, such as cooking oil, salt and monosodium glutamate, over 3 consecutive days was collected with a weighing method, using standardized weighing tools to weigh the amount of food to quantify consumption amount. The investigation lasted 4 days for each subject. On the first day, trained staff members visited the participant’s home and used a food scale to weigh and record all condiments, including the type of container used. On the second day, all condiments purchased and discarded were also weighed and recorded, and the investigator repeated this procedure on each of the remaining days of the investigation. At the end of the survey, the remaining condiments were weighed, and the amount of condiments consumed by participants over 3 consecutive days was estimated. In the 24-hour recalls, participants were asked to recall and describe all food and alcohol they had consumed over the same 3 consecutive days (2 weekdays and 1 weekend day), except for condiments.
Questionnaire
Trained health facility personnel interviewed each participant face-to-face using a uniform questionnaire after obtaining informed consent in person. This questionnaire, which was developed for the National Survey for Nutrition and Adult Chronic Disease, included items on demographic characteristics (e.g., gender, age, education level), health status (e.g., hypertension, diabetes) and health-related behaviours (e.g., smoking, physical activity). Physical activity was also assessed by the questionnaire, which covered three activity categories with 26 items: 20 items on physical activity, 4 items on rest and 2 items on sleep. These items asked the participants in what kind of activities they engaged, the frequency of activities per week and the total time spent on these activities per day. Physical activities were scored using the weighting procedure recommended by the International Physical Activity Guidelines for Americans [22], which calculates the total exercise metabolic equivalents (MET) of the three activity categories.
Participants’ height and weight were directly measured by trained and monitored workers, and blood sample was also collected.
Measures
Sodium intake
The dietary sodium intake measure included condiments assessed by the weighing method and all food in the 24-hour recalls. The sodium intake from each type of food was calculated using the China Food Ingredients Table (version II) [23]. Following the Chinese Nutrition Society, sodium intake was then categorized into two levels: sodium intake ≤ 2200 mg was defined as moderate, and sodium intake > 2200 mg was defined as excessive [23].
Alcohol consumption
A 24-hour recall survey was used to estimate each individual’s alcohol consumption on 3 consecutive days. Beverage type (liquor with high alcohol content, liquor with low alcohol content, beer, yellow rice wine, rice wine or wine) and amount consumed were measured on 3 consecutive days. With one standard drinking unit equal to 10 g of alcohol, alcohol consumption was calculated according to the Manual of Chinese Chronic Disease and Nutrition Surveillance Survey [24]. Each participant’s average alcohol consumption was divided into three levels, following the 2016 Dietary Guidelines for Chinese Residents: never (0 g/day), moderate (men: ≤ 25 g/day, women: ≤ 15 g/day) and excessive (men: > 25 g/day, women: > 15 g/day) [24].
Definition of hypertension
The main outcome was hypertension, as indicated by meeting one of the following conditions. The first condition was self-reported hypertension—having a diagnosis of hypertension and currently receiving hypertension treatment [14]. The second condition was field-measured hypertension, assessed as the average of three blood pressure measurements carried out by trained investigators using an electronic blood pressure monitor (Model HBP1300, Omron, Japan) with a precision of 1 mmHg. A standardized protocol for blood pressure measurement was used, following the recommendations issued by the Chinese Working Group on Blood Pressure Measurement [25]. Measurements were taken when the participant was seated, after a rest period of at least 5 minutes. Blood pressure was measured three times, at 1-minute intervals. Hypertension was defined as average systolic blood pressure ≥ 140 mmHg and/or average diastolic blood pressure ≥ 90 mmHg.
Definition of other variables
Ethnicity was categorized as Han, Mongolian or other minority. Marital status was categorized as single, married or other. Education level was categorized as primary school, junior high school, or high school and above. Smoking status was categorized as non-smoker (never having smoked), former smoker (previously smoked but quit) or current smoker (currently smoking). Total exercise MET was divided into tertiles, with physical activity categorized as low (total MET < 2988), medium (2988 ≤ total MET < 8400) or high (total MET ≥ 8400). Cut-offs for body mass index (BMI) were based on adjustments for the Chinese population issued by a working group on obesity in China [26]. BMI was categorized into three groups: < 23.9 kg/m2, 24.0–27.9 kg/m2 and ≥ 28.0 kg/m2.
Statistical analyses
The prevalence of hypertension was age standardized to the 2010 national demographic criteria in China using the direct method [27]. Chi-square tests were used to compare participants with and without hypertension in terms of demographic characteristics, sodium intake and drinking status. Additionally, jointly considering sodium intake status and drinking level, the participants were categorized into six subgroups: moderate sodium intake/no drinking, moderate sodium intake/moderate drinking, moderate sodium intake/excessive drinking, excessive sodium intake/no drinking, excessive sodium intake/moderate drinking and excessive sodium intake/excessive drinking.
Initially, we analysed the independent associations of sodium intake and drinking with hypertension by estimating odds ratios (OR) and 95% confidence intervals (CIs) using multivariable logistic models. A multiplicative interaction term between sodium intake and drinking was also included in the logistic regression models to test whether this interactive effect on hypertension was independent of sodium intake, drinking and other confounding factors. Then, a logistic regression model was used to compute the ORs for hypertension across the six subgroups by adjusting for important confounding factors to explore the main interaction effects of sodium intake and drinking on hypertension. We established three multivariate models: Model 1 was an unadjusted model. Model 2 adjusted for demographic variables including gender, ethnicity, educational level, marital status, BMI and family history of hypertension. Model 3 further adjusted for smoking, physical activity, diabetes and dyslipidaemia. Finally, the single effect of sodium intake and drinking on hypertension was compared.
The ‘Forward: LR’ method was used to select variables in the logistic regression. Statistical significance was set at a < 0.05. All statistical analyses were performed with SPSS, Version 19.0 (IBM Corp, Armonk, NY, USA).