Connection between percentage body fat and the risk of prediabetes in Chinese adults: a 5-year study with longitudinal cohorts

Study design

Research used a retrospective cohort study format, in which information from computer databases in China and the Nagala database were collected that were created by Chinese researchers (Chen et al.)¹⁹. The primary independent variable was the PBF at the beginning of the study. The future result was the development of prediabetes.

Data source

The Datadryad database (provided all the initial data. The data record was published from a publicly accessible study in 2018, with the title “Association of Body Mass Index and Age with diabetes used in Chinese adults: A population-based cohort study” that can be found in the interested parties: https: /doi.org0v.¹⁹. The data was extracted from a computer -aided database, which was set up by Rich Healthcare Group in China, which contained all medical records for participants who received a health check from 2010 to 2016. Rich Healthcare Group Review Board approved the original study. The terms of use of Dryad enable additional researchers to carry out secondary analyzes for the data and at the same time respect the rights of the authors.

Study population

Figure 1Displays flow diagram of study participants. The original group of Chinese participants consisted of 685,277 people, with 473,744 being removed from the study, which led to 211,833 people for further analysis. The study closed 26,247 participants with FPG levels ≥ 5.6 mmol/l in accordance with the standard 2021 American diabetes association^19.20. Ultimately, 185,586 participants were included in this study. The investigations followed the principles described in the explanation of Helsinki, and all procedures were carried out in accordance with the applicable standards and rules described in the declaration. Since it was a secondary analysis that was carried out retrospectively, no institutional ethical review or a declaration of consent was required for the study.

Data acquisition

Information for this research was collected by adults in China, including basic details such as age and gender, body mass index (BMI), blood pressure measurements (systolic and diastolic), soberglukosespiegel, triglycerides, overall cholesterol, old mirror, serum creatinine, buns and length of the follow-up. Trained researchers also collected initial information on alcohol recording, smoking habits (1 for current, 2 for the past, 3 for never and 4 for uncertainty) and family diabetes using standard surveys. When calculating the BMI, the weight in kilograms was divided by the square of the height in meters. The blood pressure was measured using a standard mercury -blood pressure meter. Venous blood samples were taken after at least 10 hours of fasting at every appointment. Covariates were selected based on clinical experience and published literature. The covariates contained the following variables: Continuous variables such as age, BMI, SBP, DBP, FBG, TG, TC, Alt, Bun, SCR; and categorical variables including gender, drinking status, smoking status and family history of diabetes.

definition

Prediabetes was characterized by increased sober glucose level in the range of 5.6 to 6.9 mmol/l in accordance with the definition (FPG 5.6–6.9 mmol/l)^21.22. The complicated procedure for determining the PBF was explained as PBF = 1.2 × BMI + 0.23 × age – 10.8 × gender – 5.4, with the gender being replaced by 1 for men²³.

Results measures

The appearance of prediabetes was our result variable. Prediabetes was determined by the FPG level in the follow-up examination and the absence of new cases of new cases of diabetes reported during the follow-up period. Prediabetes was displayed by an FPG mirror in the range of 5.6 to 6.9 mmol/l.

Missing data processing

Missing data is inevitable in observation studies. In this study, the missing data percentages were as follows: SBP at 0.00% (16 cases each), DBP at 0.00% (17 cases), TC at 2.28% (4.238 case), TG at 2.30% (4,269 case), SCR at 5.324 case), BNU, BNU, 1.07% (18.680). Several imputation was used by carrying out a linear regression using factors such as DBP, SBP, Alt, HDL-C, LDL-C, TG, SCR, Bun and TC. This process assumed that the missing data was missing by chance (Mar)²⁴.

Statistical analysis

Based on their PBF, the participants were layered in quarters. Continuous variables were summarized as mean values ​​and standard deviations for normally distributed data and as a median with interquartile areas for distorted distributions. Categorical variables were reported as frequencies and percentages. We used χ² tests to analyze categorical variables, and either a one-way anova (for normally distributed data) or the Kruskal Wallis-H test (for distorted data) to compare the differences between PBF groups. The probabilities of survival and the time of the event were determined using the Kaplan-Meier technology with comparisons with diabetes-free survival between PBF groups using the log-rank test.

In order to examine the relationship between PBF and the risk of prediabetes, we carried out Univariate and Multivariate Cox -proportional -Hazards regression analyzes. The analysis models including an unadjusted model, a minimally adjusted model (model I, which adjusted for systolic blood pressure and diastolic Blood Pressure), and a fully adjusted model (model II, which Adjusted for Systolic Blood pressure, diastolic Blood pressure, alanine Aminotransferase, Total Cholesterol, Triglycerides, Blood Urea Nitrogen, Creatinine, Smoking and Drinking Status, Family History of Diabetes, and Fasting Plasma Glukose at the start of the course). The Hazard Ratio (HR) and its corresponding 95% confidence interval (CI) were calculated. In order to evaluate the robustness of our results, we have carried out several sensitivity analyzes.

During the follow -up period, the insertion of diabetes can cover up the condition of prediabetes or influence the likelihood of its occurrence. Therefore, we carried out a competing risks of multivariate COX -proportional -hazard regression analysis in order to validate the connection between PBF and prediabetes. This method deals with the risk of diabetes as a competing event for prediabetes. Initially, participants from the age of 60 were excluded. Additional sensitivity analyzes have been carried out by being excluded or the same as 140 mmHg or diastolic blood pressure larger or 90 mmHg.

A COX proportional hazard regression model with cubic splines and smoothing curve was used to examine the non -linear connection between the PBF and the risk of prediabetes. In addition, a segmented COX regression model was used for proportional dangers to further clarify the non -linear relationship between PBF and the likelihood of prediabetes. The log-Likelihood ratio test was carried out to identify the model that best explains the connection between PBF and the risk of prediabetes.

Various subgroup analyzes were carried out using a layered COX proportional hazard regression model, which was layered according to gender, age, BMI, SBP and DBP. Age, BMI, DBP and SBP were categorized based on clinical threshold values: age categories were

The statistical software package (R Foundation) and the Empower statistics (X & y Solutions, Inc., Boston, MA) were used for all analyzes. The statistical significance was determined by a two-tailed P value below 0.05.