The geriatric nutritional risk index mediated the relationship between serum uric acid and hypertension: a mediation analysis

This study used the data of wave 2009 of the China Health and Nutrition Survey (CHNS). As a national large- scale cohort study, the CHNS focused on the changes of health and nutrition, as well as the effects of the social and economic transformation on health and nutritional statuses in China. It was described in detail in the published literature elsewhere [15].

In the CHNS, a multistage and random cluster process was used to sample the study population from nine provinces, which were representative in geography, economic development, and health indicators. All participants were interviewed in household to collect information of lifestyle, health status, marriage and birth history, and detailed physical examinations including weight, height, arm and head circumference, and blood pressure. In the CHNS 2009, the biomarker data of 26 fasting blood measures were released including major cardiovascular biomarkers and important nutrition biomarkers, which has been published for an overview of these biomarkers [16].

The inclusion criteria were as follows: who were aged ≥ 60 years in the time of survey; and who have complete analyzed data, including height, weight, blood pressure, and uric acid. To explore the mediation effect of malnutrition not attributed to diseases on the association of SUA level with hypertension, participants with severe wasting diseases, such as hepatic disease and severe renal insufficiency, were excluded. Overall, 2371 participants were included in the final analysis. This study was approved by the Institutional Review Board of the National Institute for Nutrition and Food Safety, China Center for Disease Control and Prevention, and University of North Carolina at Chapel Hill. Written informed consent was obtained from all subjects.

Physical index including height and weight were measured by the trained health staff or the members of the CHNS following standardized approach. All index were measured for three times. And the averages were used to analysis. Height and weight were employed to calculate body mass index (BMI).

A valid questionnaire was used to collect the data of health behaviors and medical history. The smoking status was checked via a question as follows: Have you ever smoked cigarettes (including hand-rolled or device-rolled)? If respondents answered “never smoked”, he or she was identified as no smoking. If respondents answered “yes”, he or she was identified as smoking. And the status of alcohol consumption was interviewed as follows: Did you drink beer or any other alcoholic beverage last year? If respondents answered “no”, he or she was identified as no alcohol consumption. If respondents answered “yes”, he or she was identified as alcohol consumption. All participants were interviewed for their education levels as follows: What is the highest level of education you have attained? According to the responses, education levels were divided into three levels as follows: primary school or none, middle school, and college or above. Physical activities in this study included martial arts, gymnastics, track and field, soccer and basketball, badminton and volleyball, and others (ping pong, Tai Chi, etc.). Each participant was asked for each activity: Do you participate in this activity? If they answered “no” for all these items, physical activity was identified as “no”. If they answered “yes” for at least one of these items, physical activity was identified as “yes”. The history of diabetes was identified via a question: Has a doctor ever told you that you suffer from diabetes? If the answer was “no”, the history of diabetes was identified as “no”. If the answer was “yes”, the history of diabetes was identified as “yes”. Ethnicity was divided into Han and others according to the participants’ identification cards.

All participants were supposed to provide the fasting blood samples in wave 2009. The automated biochemistry analyzer (Hitachi 7600, Randox, UK and Kyowa, Japan) was used to assay SUA and serum albumin (ALB) in Beijing central laboratory.

The GNRI was calculated by ALB and weight as follows: GNRI= [ 1.489× albumin (g/L)] + [ 41.7× ( actual weight/ ideal weight)] [17]. Ideal body weight was calculated by the Lorentz formula [17]. Sex-specific Lorentz formula was: ideal weight of males = height − 100 - [(height − 150)/4] and ideal weight of females = height − 100 - [(height − 150)/2.5]. If actual weight / ideal weight was more than one, actual weight / ideal weight was set to 1.

After resting in a sitting position for 5 min and selecting suitable cuff size, blood pressure was measured for three times using standardized mercury sphygmomanometers. Systolic blood pressure (SBP) corresponds with the first Korotkoff sound, and diastolic blood pressure (DBP) corresponds with the fifth Korotkoff sound. And the averages were used to define hypertension as follows: SBP/ DBP ≥ 140/ 90 mmHg or taking antihypertensive medications [18].

Means ± standard deviations were used to describe continuous variables. And frequencies (percentages) were used to describe categorical variables. The between-group comparisons were conducted by t-test and chi-square test for continuous variables and categorical variables, respectively. Logistic regression was employed to examine the associations of the GNRI and SUA level with hypertension and obtain odd ratios (ORs) and 95 % confidential intervals (CIs). In the adjusted models, age, sex, current smoking, current drinking, education degree, physical activity, ethnicity, and history of diabetes were adjusted. In the mediate analysis, the GNRI, SUA, and hypertension were taken as the mediate variable, independent variable, and response variable, respectively. The Directed Acyclic Graph of relations among these variables is depicted in Fig. 1.

All mediation analyses were conducted using the SAS macro procedure package “mediation analysis” [19]. And the controlled direct effect (CDE), natural direct effect (NDE), natural indirect effect (NIE), and proportion mediated were provided. Based on the counterfactual framework, the CDE is the effect of exposure on outcome as mediator was controlled. The NDE, namely pure direct effect, is the effect of exposure on outcome with mediator was kept at the level it would naturally have taken. The NIE is the effect of mediator on outcome as exposure was controlled. When outcome is binary, the total effect is the overall effect of exposure on outcome regardless of mediator, and equals to odds ratio of the NDE multiplying by odds ratio of NIE. Similarly, the proportion mediated equals to odds ratio of the NIE divided by odds ratio of total effect. SAS 9.4 (SAS Institute Inc., Cary, NC, USA.) was employed to conduct all analyses. And a two-tailed P ≤ 0.05 was considered as statistical significance.