{"paper_id":"e485a053-693f-4a1f-a136-42929fa2e29c","body_text":"Cardiometabolic diseases are a series of metabolic dysfunctions, including cardiovascular diseases and diabetes mellitus. Among adults worldwide, they constitute one of the primary causes of morbidity and mortality ( 1 ,  2 ). Effective identification and active control of cardiometabolic diseases in their early stages represent a crucial determinant of disease development and prognosis. The development of cardiometabolic diseases is most closely related to obesity ( 3 ,  4 ). In the past, we usually used body mass index (BMI) to define obesity ( 5 ). However, this approach cannot determine the proportion of fat in the body composition and thus has many limitations ( 6 ). In recent years, a growing number of people with normal-weight obesity (NWO) has prompted a surge in research interest in the health implications of visceral obesity ( 7 ).\nVisceral obesity is directly related to cardiometabolic diseases ( 8 ), and many new measures have emerged for assessing visceral obesity. Wakabayashi et al. ( 9 ) were the first to propose CMI, which is derived from a combination of lipids and obesity parameters and can be calculated to detect cardiometabolic risk in humans simply and effectively, providing a comprehensive assessment of cardiometabolic risk ( 10 – 12 ).\nIn recent years, relevant studies have found that CMI may be a marker for predicting renal function ( 13 ), depression ( 14 ), impaired fasting glucose, insulin resistance, type 2 diabetes mellitus ( 15 ), non-alcoholic fatty liver disease and fibrosis ( 16 ), biological aging ( 17 ), endometriosis ( 18 ,  19 ), and erectile dysfunction ( 20 ) in older adults. Regarding the effect of CMI on mortality, recent studies have shown that elevated CMI is positively associated with all-cause mortality in middle-aged and older adults and is partially mediated by inflammation ( 21 ,  22 ). Moreover, in the general population, CMI has an L-shaped nonlinear association with all-cause mortality ( 23 ).\nBased on the above research, we have three research gaps. Firstly, we focused on adults over 18, the leading group who develop cardiometabolic diseases. Secondly, because CMI is associated with metabolic diseases, we focused on the correlation between CMI and all-cause, cardiovascular, and diabetes mortality. To our knowledge, this is the first study to include the correlation between CMI and diabetes mortality. Finally, there are differences in whole-body lipid metabolism between men and women, and trends in body fat accumulation in women are associated with an increased risk of metabolic-related diseases ( 24 ), so we further explored sex differences between CMI and different mortality correlations.\n\nA cross-sectional survey of U.S. citizens’ health and nutritional status, NHANES, was conducted in 1999 and contains a wide range of demographic, economic, nutritional, and health data ( 25 ). NHANES employs a sophisticated, four-stage sampling design, utilizing a stratified, multistage probability sampling approach to ensure the sample is representative of the entire nation ( 26 ). Fifteen different sites are selected each year from a sampling frame of all U.S. counties, from which it chooses a nationally representative sample of comprising approximately 5,000 individuals ( 27 ). Written informed consent and ethical review by the National Centre for Health Statistics (NCHS) Ethics Review Committee were obtained from all participants in the NHANES study.\nIn this study, we selected 101,316 participants who participated in the NHANES survey from 1999 to 2018 and linked them to NDI data to obtain information on participant follow-up. After excluding those who failed to meet the inclusion criteria, the study finally included 37539 participants. The flowchart in  \n Figure 1 \n  illustrates the process for screening the inclusion population.\nFlowchart for inclusion in the study population.\nCMI, as an exposure variable in this study, was derived by calculation from blood indicators and body data measured by professionals in a mobile examination van. The formula is as follows ( 9 ): CMI=TG(mg/dl)/HDL(mg/dl)×Waist(cm)/Height(cm).\nThe outcome variables in this research included all-cause mortality and two specific mortality rates. The specific mortality rates included cardiovascular mortality and diabetes mortality. We obtained mortality-related data from the NDI Death Certificate Record ( www.cdc.gov/nchs/data-linkage/mortality-public.htm ), which is currently updated until December 31, 2019 ( 28 ). Combining NHANES and NDI-related data provides comprehensive information on participant deaths, including time of death and cause of death, among other comprehensive information. Based on the fact that the study period is 1999-2018, we used the person-months from the date of interview to the date of death or the end of the period of death (PERMTH_INT) to calculate the follow-up time in this study. People can obtain specific information from the official website mentioned above. Specifically, the codes I00-09, I11, I13, I20-51, and I60-I69 identify cardiovascular mortality, while the codes E10-E14 identify diabetes mortality in the International Classification of Diseases (ICD)-10.\nThe covariates in the study consisted primarily of demographic characteristics of gender (male, female), age (years), race (Mexican American, non-Hispanic white, non-Hispanic black, other Hispanic, and other race), marriage (married, never married, separated, divorced, widowed, and living with partner), household income (Poverty Income Ratio, PIR), and educational qualification (less than 9th grade, 9-11th grade, high school graduate, some college, and college graduate or above). Other covariates included body mass index (BMI, kg/m 2 ), alcohol intake (never, light/moderate, heavy, former), and smoking status (never, current, former). Laboratory tests included triglycerides (TG, mg/dl), total cholesterol (TC, mg/dl), high-density lipoprotein (HDL, mg/dl), and glomerular filtration rate (eGFR, mL/min/1.73 m 2 ). A questionnaire determined coronary heart disease (CHD) diagnosis based on self-reported responses. A diagnosis of diabetes mellitus(DM) is the fulfillment of any of the following conditions: (1) doctor told you have diabetes, (2) glycohemoglobin HbA1c (%) >= 6.5, (3) fasting glucose (mmol/l) >= 7.0, (4) random blood glucose (mmol/l) >= 11.1, (5) two-hour OGTT blood glucose (mmol/l) >= 11.1, (6) Use of diabetes medication or insulin. The diagnosis of hypertension is based on the following criteria: (1) self-reported hypertension, (2) mean systolic blood pressure (SBP) ≥140 mmHg and/or mean diastolic blood pressure (DBP) ≥90 mmHg, and (3) use of antihypertensive medications. The diagnosis of hyperlipidemia is fulfilling the following conditions: (1) diagnosis of hypertriglyceridemia (TG>=150mg/dL), (2) diagnosis of hypercholesterolemia: (i) TC>=200mg/dL, (ii) low-density lipoprotein cholesterol (LDL-C) >=130mg/dL (iii) HDL<40mg/dL in men and <50mg/dL in women, and (3) current taking lipid-lowering drugs. The official NHANES website provides specific technical information regarding covariate determination.\nEvery analysis in this research utilized the R language (version 4.3.0) and achieved statistical significance with a two-tailed p-value below 0.05. We divided participants into four groups according to their CMI quartiles (Q1-Q4), and the lowest quartile was considered the baseline. We divided the population based on gender and CMI quartiles to investigate the relationship between CMI and all-cause and specific mortality. The weights of the analyses were adjusted to prevent oversampling. We used weighted means (95% confidence intervals [CIs]) and weighted percentages (95% confidence intervals) to describe continuous and categorical variables, respectively ( 29 ). To assess gender differences in the relationship between CMI and outcome variables, we devised three weighted COX regression models, with model 1 unadjusted for confounders, model 2 adjusted for age and ethnicity, and model 3 adjusted for age, race, marital status, smoking status, alcohol use, eGFR, and TC.\nIn addition, the RCS was employed with three nodes (5th, 50th, and 95th percentiles) to assess dose-response patterns between CMI and all-cause and specific mortality. We generated Kaplan-Meier survival curves for various mortality rates based on CMI quartiles and survival times.\n\nWe included 37,539 participants over ten consecutive survey cycles from 1999 to 2018.  \n Tables 1 \n ,  \n 2 \n  demonstrate the baseline characteristics of participants by CMI quartile (Q1-Q4) by gender, with a mean age of 45.954 for male participants and 47.811 for female participants. In Q1-Q4, there were notable disparities (p < 0.05) between males and females in age, gender, race, marriage, education, poverty level, BMI, smoking, alcohol consumption, CHD, DM, hypertension, hyperlipidemia, height, waist circumference, total cholesterol, triglycerides, HDL, and glomerular filtration rate.  \n Supplementary Table S1 \n  demonstrates the baseline characteristics of participants by survival status and shows that those who died exhibited higher levels of CMI compared with those who survived.\nBaseline characteristics of male participants.\nContinuous variables are expressed as weighted means ( ± SE) and categorical variables as unweighted frequencies (weighted percentages).\nBaseline characteristics of female participants.\nContinuous variables are expressed as weighted means (± SE) and categorical variables as unweighted frequencies (weighted percentages).\nThroughout the monitoring phase, participants experienced 5600 deaths due to any cause, comprising 1765 from cardiovascular causes and 200 from diabetes. As shown in  \n Figure 2 \n , we performed Kaplan-Meier analysis. The results showed that CMI differed significantly in predicting all-cause, cardiovascular, and diabetes mortality between Q1-Q4 groups in the total population, females and males. All three mortality rates showed an increasing trend with increasing follow-up time, and we found no gender differences. Nevertheless, the differences were more pronounced in the female than in the male group.\nKaplan-Meier curves between CMI and all-cause and specific mortality rates. The X-axis indicates follow-up time (months); the Y-axis indicates survival probability.  (A)  All-cause mortality for the whole population  (B)  All-cause mortality for females  (C)  All-cause mortality for males  (D)  Cardiovascular mortality for the whole population  (E)  Cardiovascular mortality for females  (F)  Cardiovascular mortality for males  (G)  Diabetes mortality for the whole population  (H)  Diabetes mortality for females  (I)  Diabetes mortality for males.\nTables 3 \n – \n 5 \n  show the results of the COX regression analyses between CMI and outcomes, respectively. When CMI was a continuous variable, CMI in model 3 was significantly correlated with increased mortality rates due to all causes, cardiovascular and diabetes in the whole population (all-cause: HR: 1.03, 95%CI: 1.02-1.05, P<0.0001; cardiovascular: HR: 1.04, 95%CI: 1.02-1.05, P <0.0001; diabetes: HR: 1.11, 95%CI: 1.09-1.14, P<0.0001) and participants of both genders (all-cause: male: HR: 1.03, 95%CI: 1.01-1.04, P<0.001; female: HR: 1.05, 95%CI: 1.02-1.07, P<0.001. cardiovascular: male: HR: 1.03, 95%CI: 1.01-1.05, P=0.01; female: HR: 1.05, 95%CI: 1.03-1.08, P<0.0001. diabetes: male: HR: 1.09, 95%CI: 1.06-1.12, P<0.0001; female: HR: 1.15, 95%CI: 1.09-1.22, P<0.0001). As a categorical variable based on quartiles, CMI demonstrated a notable rise in mortality rates for all causes, cardiovascular disease, and diabetes in Q4 compared with Q1 in the whole population (all-cause: P for trend=0.005; cardiovascular: P for trend<0.001; diabetes: P for trend<0.001) and female participants (all-cause: P for trend=0.03; cardiovascular: P for trend= 0.005; diabetes: P for trend=0.002). In contrast, male participants had no significant difference (all-cause: P for trend=0.06; cardiovascular: P for trend=0.06; diabetes: P for trend=0.09).\nModels of COX regression for CMI and all-cause mortality.\nModel 1: non-adjusted.\nModel 2: adjusted for age, gender, and race.\nModel 3: Whole population adjusted for age, gender, race, marital, smoking, alcohol, eGFR, and TC. Male and female people adjusted for age, race, marital, smoking, alcohol, eGFR, and TC.\neGFR, estimated glomerular filtration rate; TC, total cholesterol.\nHR, hazard ratios; CI, confidence interval.\nModels of COX regression for CMI and cardiovascular mortality.\nModel 1: non-adjusted.\nModel 2: adjusted for age, gender, and race.\nModel 3: Whole population adjusted for age, gender, race, marital, smoking, alcohol, eGFR, and TC. Male and female people adjusted for age, race, marital, smoking, alcohol, eGFR, and TC.\neGFR, estimated glomerular filtration rate; TC, total cholesterol.\nHR, hazard ratios; CI, confidence interval.\nModels of COX regression for CMI and diabetes mortality.\nModel 1: non-adjusted.\nModel 2: adjusted for age, gender, and race.\nModel 3: Whole population adjusted for age, gender, race, marital, smoking, alcohol, eGFR, and TC. Male and female people adjusted for age, race, marital, smoking, alcohol, eGFR, and TC.\neGFR, estimated glomerular filtration rate; TC, total cholesterol.\nHR, hazard ratios; CI, confidence interval.\nWe assessed the RCS using model 3 to investigate further the correlation between CMI and mortality rates ( \n Figure 3 \n ), which showed that in the total population, CMI and all-cause mortality were nonlinearly related (P for nonlinear = 0.0276), as well as diabetes mortality (P for nonlinearity = 0.001). Cardiovascular mortality was linearly related (P for nonlinear = 0.1272). All three mortality rates increased with increasing CMI. Across sexes, females showed a linear response relationship in all-cause mortality (P for nonlinear = 0.4645) and cardiovascular mortality (P for nonlinear = 0.7686) and a nonlinear response relationship in diabetes mortality (P for nonlinear < 0.001); males showed a nonlinear response relationship in all-cause mortality (P for nonlinear = 0.0324) and diabetes mortality (P for nonlinear = 0.0168), but there was no statistically significant relationship with cardiovascular mortality (P for overall = 0.1367).\nRCS curves between CMI and all-cause and specific mortality rates. Red lines represent the estimated hazard ratio, and the shaded red area corresponds to the 95% confidence intervals.  (A)  All-cause mortality for the whole population  (B)  All-cause mortality for females  (C)  All-cause mortality for males  (D)  Cardiovascular mortality for the whole population  (E)  Cardiovascular mortality for females  (F)  Cardiovascular mortality for males  (G)  Diabetes mortality for the whole population  (H)  Diabetes mortality for females  (I)  Diabetes mortality for males.\n\nUsing NHANES data, we conducted a cohort study of 37,539 participants. The results showed significant gender differences between CMI and all-cause, cardiovascular, and diabetes mortality among the general adult population in the United States. In women, we found that CMI independently predicted mortality from all causes, cardiovascular diseases, and diabetes, but this association did not hold for men. Moreover, we further explored the trend changes by Kaplan-Meier and RCS curves. It is the first study to address gender differences in the correlation between CMI and mortality in the general adult population.\nResearchers initially used CMI, a relatively new measure of obesity and lipid levels, to screen for diabetes and obesity ( 9 ). Subsequently, many related studies have shown a correlation between CMI and an increased risk of developing diabetes, with Zha et al. ( 30 ) finding a nonlinear positive relationship between CMI and diabetes risk in Japanese adults and Qiu et al. ( 31 ) finding the same positive association in a study of middle-aged and older adults in China. In addition, studies have also found associations between CMI and a variety of metabolic disorders, such as CVD. For example, Cai et al. ( 32 ) found that CMI was positively associated with the risk of new-onset CVD in patients with hypertension and obstructive sleep apnea. The above studies suggest that CMI can not only be used to assess diabetes but is also significantly associated with metabolism-related diseases. It is consistent with this study’s finding that CMI is significantly and positively associated with all-cause, cardiovascular, and diabetes mortality in the U.S. adult population.\nOur study population was the general adult population, and we further analyzed that there was a significant correlation between CMI and all-cause mortality, cardiovascular mortality, and diabetes mortality in females. We tried to explain the reasons for the emergence of this gender difference.\nBased on previous reports, it appears that cardiovascular and diabetic disease risk differs by gender. In older American populations, white men have a higher risk of obesity, whereas white women have a higher risk of hypertension and diabetes ( 33 ). Similarly, in Asian populations, research in a Japanese population showed a stronger relationship between obesity and cardiometabolic risk factors in women ( 34 ) and a greater cardiometabolic risk of glucose and dyslipidemia in women than in men ( 35 ). Shi et al. ( 36 ) found that higher AIP levels in women significantly increase the risk of developing prediabetes and diabetes. In women, impaired glucose tolerance (IGT) is more prevalent than in men, and obesity-associated type 2 diabetes mellitus is associated with significant gender and regional differences ( 37 ). The above findings suggest that gender differences in cardiovascular-related and diabetes-related diseases are more common among populations in various regions. Few studies have now reported sex differences between CMI and cardiovascular and diabetes-related diseases. Elevated CMI is associated with eccentric and centripetal left ventricular hypertrophy in females compared with males ( 38 ). It suggests that CMI is likely to be a marker for explaining the adverse cardiovascular effects of central obesity ( 39 ,  40 ). Moreover, the correlation between CMI and diabetes mellitus is more pronounced among female patients ( 9 ,  15 ). The above result is in line with the findings in the current study.\nGender differences in CMI reflecting adverse cardiometabolic outcomes may be related to the characteristics of body fat distribution in different gender populations. The specific mechanisms may be related to sex hormones, adipocyte characteristics, and the adipose microenvironment. Premenopausal women usually have more subcutaneous adipose tissue (SAT). In contrast, men have more trunk and visceral adipose tissue (VAT), and this characteristic shifts after menopause, with an increase in the distribution of trunk and VAT, the development of centripetal obesity, and an increase in waist circumference in women compared with that in men, which dramatically increases the risk of diabetes mellitus and cardiovascular disease in women ( 41 – 44 ). The CMI measures the lipid profile of the body, and the CMI is an indicator of cardiometabolic disease risk in women. While CMI is a comprehensive indicator for detecting dyslipidemia and abdominal obesity, it can reflect VAT distribution well. It appears to be more relevant to women in predicting adverse outcomes in cardiometabolic-related diseases. At the same time, we should also be concerned about the impact of obesity on adverse outcomes in other pre-existing conditions. The clinical phenotype of obesity is similar to that of hypertrophic cardiomyopathy (HCM) ( 45 ). Relevant studies have shown that among patients with HCM, women are older than men and have a higher risk of HCM-related complications and death ( 46 ,  47 ). Obesity is also a significant risk factor for increased rates of adverse events in patients with conditions such as stroke ( 48 ), thromboembolic disease ( 49 ), and atrial fibrillation ( 50 ).\nIn addition, an imbalance in female sex hormones significantly increases the risk of cardiometabolic issues. Sex hormones may influence the distribution of adipose molecules. Estrogen may affect fat distribution through its effect on adrenergic receptor distribution in long-lasting adipocytes. Testosterone may also play a specific role in adipose tissue distribution through its impact on lipoprotein lipase (LPL) activity and rate-limiting fatty acid uptake ( 51 ). Sex hormones also influence adipose cycling factors (e.g., leptin and lipocalin), which are directly related to the accumulation of SAT ( 52 ). The absence of positive results in males in this study may be related to the role of androgens. Androgens are involved in regulating lipid metabolism and may inhibit fat deposition ( 53 ). Therefore, high androgen levels in men may protect against lipid metabolism-related diseases.\nIntrinsic properties of adipocytes may also account for gender differences in fat distribution and function. Adipocytes from SAT and VAT have different physiological properties, with STA adipocytes undergoing hyperplasia and hypertrophy, whereas VAT adipocytes experience hypertrophy only ( 54 ). Researchers have linked this intracellular difference to distinct patterns of gene expression. In addition to the intrinsic properties of adipocytes, differences in the extrinsic microenvironment and extracellular matrix can also lead to gender differences in adipose distribution, which is related to the regulation of the activation of adipocyte precursors that drive adipocyte proliferation ( 55 ).\nIn summary, the results of our study indicate that CMI is significantly linked to an increased risk of all-cause mortality, cardiovascular mortality, and diabetes mortality among adult women. It suggests that we may be able to judge the prognostic risk by the change of CMI value in clinical practice, especially for patients with cardiac disease, diabetes, and other underlying diseases. It is also important to pay attention to the influence of gender differences on disease prognosis. We recommended that adult women with high CMI values regularly engage in moderate exercise and pay attention to the management of lipids and waist circumference in order to reduce the risk of adverse prognosis of cardiometabolic-related diseases ( 56 ). More generally, it is worthwhile for medical personnel to consider whether there is a need to formulate different treatment standards and protocols according to gender in clinical diagnosis and treatment. At the same time, it is also necessary to pay attention to the screening of the relevant population. Clinical staff can apply new electrocardiography, echocardiography, and cardiac CT imaging to better screen high-risk groups ( 57 – 59 ) and reduce or delay the occurrence of cardiometabolic-related adverse outcomes.\nHowever, this study has some limitations: (i) Since this study only examined the U.S. population, more studies on other races are required for generalizability. (ii) Human reporting factors may affect the accuracy of NHANES data on causes of death, population classification, and death certificates. (iii) We considered potential confounders, but other factors may remain unconsidered.\n\nIn our study, we found that CMI was significantly positively associated with all-cause mortality, cardiovascular mortality, and diabetes mortality in adult females in the United States but not in males. According to the findings, it is plausible that CMI can be employed as a predictive indicator for mortality, particularly in relation to all-cause mortality, cardiovascular-specific mortality, and diabetes mortality among adult females.","source_license":"CC-BY-4.0","license_restricted":false}