High Prevalence of Metabolic Obesity in India: ICMR-INDIAB-23

Background & objectives While obesity usually produces cardio-metabolic dysfunction, some obese individuals are metabolically healthy, and conversely, some nonobese individuals have significant metabolic dysfunction. This study aims to assess the national prevalence of various obesity subtypes and their association with type 2 diabetes (T2D), coronary artery disease (CAD), and chronic kidney disease (CKD) in the Indian Council of Medical Research-India Diabetes (ICMR-INDIAB) study. Methods The ICMR-INDIAB study is a nationally representative cross-sectional survey of 1,13,043 individuals aged ≥20 yr from urban and rural areas across 31 Indian States and Union Territories. In every fifth individual (n=19,370), venous blood glucose and lipids were measured. A body mass index (BMI) ≥25 kg/m2 was defined as being obese, and metabolic obesity was diagnosed if two risk factors, out of the following: high waist circumference, high blood pressure, elevated blood glucose, raised serum triglycerides, or low HDL cholesterol, were present. Four subgroups were identified: Metabolically Healthy Non-Obese (MHNO), Metabolically Healthy Obese (MHO), Metabolically Obese Non-Obese (MONO), and Metabolically Obese Obese (MOO). Results The prevalence of various obesity subtypes was as follows: MONO: 43.3 per cent [95% confidence interval (CI): 42.6-44%], MOO: 28.3 per cent (27.7-28.9%), MHNO: 26.6 per cent (26-27.2%), and MHO: 1.8 per cent (1.6-2%). MONO was more prevalent in rural areas [Rural vs. Urban: MONO: 46 per cent (45-46.9%) vs. 39.6 per cent (37.8-41.3%), P<0.001]. MOO showed the highest risk for T2D and CAD, while MONO showed the highest risk of CKD, especially among females. Interpretation & conclusions Individuals with MONO have a distinct phenotype with adverse metabolic consequences, highlighting the need to shift from body weight-focused approaches to broader strategies to identify and tackle non-communicable diseases (NCDs) in India. Keywords: Asian Indians, coronary artery disease, chronic kidney disease, diabetes, dyslipidaemia, hypertension, obesity, South Asians Obesity, characterised by an elevated body mass index (BMI), has numerous adverse metabolic consequences on overall health. Worldwide, it is estimated that there are five million deaths every year from noncommunicable diseases (NCDs) attributable to high BMI, type 2 diabetes (T2D), cardiovascular disease (CVD), cancer, neurological disorders, chronic respiratory diseases, and chronic kidney disease (CKD)1. As per the World Health Organization (WHO)2 there are 2.5 billion overweight adults and 890 million with obesity worldwide (representing 43% and 16%, respectively, of the global adult population). Generalised obesity is defined based on BMI cut points, while abdominal obesity is defined based on waist circumference (WC) or various waist-related indices like waist to hip ratio (WHR) or waist to height ratio (WHtR). Compared to other ethnicities, Asian Indians have a distinct susceptibility to develop T2D and other obesity-related metabolic disorders at a lower BMI. The ‘Asian Indian Phenotype’, marked by high levels of abdominal fat, insulin resistance, and dyslipidaemia with low HDL cholesterol and high serum triglycerides even with normal BMI, is believed to be a primary factor underlying this heightened risk3-5. The substantial burden of obesity in India was confirmed by a recent publication from the Indian Council of Medical Research (ICMR)-India Diabetes (INDIAB) study6, which documented that there are an alarming 254 million and 351 million adults with generalised and abdominal obesity, respectively, in India6. Metabolic dysfunction is often linked with obesity, but some individuals with obesity have no cardiometabolic risk factors. A recent Lancet Commission on Clinical Obesity has distinguished between individuals with excess body fat who have evidence of obesity-associated illness (termed ‘clinical obesity’) and those who do not (termed ‘preclinical obesity’)7. Conversely, some normal-weight or lean individuals may have significant cardiometabolic risks. In the 1980s, Ruderman8,9 introduced the concept of Metabolically Unhealthy Normal Weight (MUHNW) or Metabolically Obese Normal Weight (MONW) to describe individuals who are not obese based on BMI but show traits like hyperinsulinemia, insulin resistance, high triglycerides, and increased risk of coronary artery disease (CAD) and T2D. Another group of individuals who do not exhibit high-risk metabolic profiles yet meet conventional BMI criteria for obesity are classified as ‘metabolically healthy obese (MHO)’10,11. As cardiometabolic risk varies in each of these subtypes, it is important to assess their prevalence to plan preventive or treatment strategies. This is particularly relevant in the Asian Indian context, where individuals tend to develop obesity-related comorbidities even in the non-obese ranges of BMI, leading to a delay of medical interventions when screening is based solely on BMI. There is no national data on the various obesity subtypes in India. We used data from the large, nationally representative, ICMR-INDIAB study to report on the prevalence of various obesity subtypes and evaluate the risk of each subtype for T2D, CAD, and CKD among adults in India. Materials & Methods This cross-sectional survey was undertaken by the department of Epidemiology and Research Operations and Diabetes Complications, Madras Diabetes Research Foundation, Chennai, Tamil Nadu, India. The study was approved by the Institutional Ethics Committee of the coordinating centres and individual States. Written informed consent was obtained from all study participants. The study was registered with the Clinical Trials Registry of India (CTRI/2019/03/018095). Sampling and study population Adults aged ≥20 yr were recruited from the ICMR-INDIAB study12-19, a cross-sectional, population-based survey in India. The study methodology, including sampling strategies, sample size, and phases, have been described previously12 and provided in the supplementary material. Of the total 1,19,022 individuals from 31 States studied, 1,13,043 individuals participated in the study, yielding a response rate of 95 per cent. Supplementary material IJMR-161-5-461-SM.pdf (273.2KB, pdf) Assessments Data on medical history, family history of diabetes, physical activity, and socioeconomic status were collected using a standardised and structured questionnaire in all participants. Self-reported data included alcohol and smoking (current or in the prior six months). Physical activity was measured using a validated Physical Activity Questionnaire18. Individuals were classified into two categories based on their physical activity level (PAL), which was determined by dividing their total energy expenditure for 24 h by their basal metabolic rate: 1.4-1.69 for inactive individuals and 1.7-2.4 for active individuals. Dietary information was obtained using the MDRF-Food Frequency Questionnaire (M-FFQ)20, a validated, interviewer-administered tool. This meal-based questionnaire lists 222 common food items from urban and rural areas. Daily intake of calories, macronutrients, dietary fibre, and fatty acids was calculated using ‘EpiNu’ Software (Version 2.0). The nutrient densities expressed as the percentage of energy derived from carbohydrates, proteins, and fats were estimated and used in the analysis. Standardised methods were used to determine blood pressure (BP) and anthropometric measurements, including weight, height, and waist circumference21. Height (in centimetres) was measured using a stadiometer (SECA Model 214, Seca Gmbh Co, Hamburg, Germany). Body weight (in kilograms) was measured using an electronic weighing scale (SECA Model 807, Seca Gmbh Co, Hamburg, Germany) placed on a flat horizontal surface. BMI was calculated using the formula: weight (kg)/height (m)2. Waist circumference (centimetres) was measured using a non-stretchable measuring tape at the smallest horizontal girth between the costal margins and the iliac crest at the end of expiration. Using an electronic sphygmomanometer (Omron HEM-7101; Omron Corporation, Tokyo, Japan), BP was recorded to the nearest 1 mm Hg. The final reading was recorded as the average of two measurements taken five minutes apart. Inter-observer and intra-observer coefficients of variation between the field technicians were documented and were less than 5 per cent. Equipment with the same specifications was used during the investigations as a quality control measure. A One Touch Ultra glucose meter (LifeScan Johnson & Johnson, Milpitas, California) was used to assess each individual’s capillary blood glucose (CBG) after confirming an overnight fast of 8-12 h. Participants were administered 82.5 g of glucose (75 g of anhydrous glucose) for an oral glucose tolerance test, and the 2-h post-load CBG was measured. For individuals with self-reported diabetes, only fasting CBG was measured. In every fifth participant and individuals with diabetes, a venous sample was taken for the measurement of lipids, glycated haemoglobin (HbA1c), and creatinine. The VariantTM II Turbo machine (Bio-Rad, Hercules, CA) was utilised for high-pressure liquid chromatography to estimate HbA1c. An autoanalyzer [model 2700/480; Beckman Coulter AU (Olympus, County Clare, Ireland)] was used to measure serum cholesterol (cholesterol esterase oxidase-peroxidase-amidopyrine method), serum triglycerides (glycerol phosphate oxidase-peroxidase-amidopyrine method), and high-density lipoprotein cholesterol (direct method; polyethylene glycol-pre-treated enzymes). Serum creatinine was measured using the Jaffe Kinetic method. For biochemical assays conducted at the central laboratory, the intra-assay and inter-assay coefficients of variation ranged between 3.1 per cent and 7.6 per cent. A resting 12-lead electrocardiogram (ECG) was recorded in every fifth individual, and those with diabetes, and were graded using Minnesota coding. For the present analysis, the sample size (n) was 19,370, as lipids, creatinine, and ECG (needed to categorise individuals for metabolic obesity and assess complications) were performed only on every fifth participant. Definitions Metabolic obesity was defined as having ≥2 components of metabolic syndrome: (i) waist circumference ≥90 cm in males and ≥80 cm, in females, (ii) fasting blood glucose (FBG) ≥100 mg/dl, (iii) BP ≥130/85 mmHg or on anti-hypertensive medications, (iv) serum triglyceride levels ≥150 mg/dl or (v) HDL cholesterol <40 mg/dl for males and <50 mg/dl for females. Four obesity subtypes were defined as below: Metabolically Healthy Non-Obese (MHNO): absence of metabolic obesity and BMI <25 kg/m2, MHO: absence of metabolic obesity and BMI ≥25 kg/m2, Metabolically Obese Non-Obese (MONO): presence of metabolic obesity and BMI <25 kg/m2, Metabolically Obese Obese (MOO): presence of metabolic obesity and BMI ≥25 kg/m2. Diabetes was defined as fasting CBG ≥126 mg/dl (7.0 mmol/l), or 2-h post-oral glucose load CBG ≥200 mg/dl (11.1 mmol/l), or a physician diagnosis of diabetes22. CAD was diagnosed based on a recorded history of myocardial infarction (MI) or drug treatment for CAD and/or Minnesota codes: Q wave changes (1-1-1 to 1-1-7), ST segment depression (4-1 to 4-2) T-wave abnormalities (5-1 to 5-3) on the resting ECG23. CKD was diagnosed if the estimated glomerular filtration rate (eGFR) was <60 ml/min/1.73m2 at the time of study. eGFR24 was derived using the following formula: GFR=141*min (Scr/𝜅,1)𝛼*max⁡(Scr/𝜅, 1)−1.2⁢0⁢9*0.⁢993Age*1⁢.0⁢18 (if female) [κ=0.7 if female, κ=0.9 if male; α=-0.329 if female, α=-0.411 if male; min=the minimum of Scr/κ or 1, max=the maximum of Scr/κ or 1; Scr=serum creatinine (mg/dl)] Statistical analysis The proc survey (frequency/mean) procedure was used to analyse the data collected from complex survey designs, ensuring that the statistical analyses and inferences drawn were accurate and representative of the target population. These procedures are used when dealing with survey data that involves stratification, clustering, and survey weights. Supplementary material (page 9) provides the sampling weights calculation to adjust for sampling at different levels within each State. Survey-adjusted linear regression was used to compute the mean, and the Wald χ2 test was applied to compare the proportions of variables between subtypes (MHNO vs. MHO, MONO, and MOO; urban vs. rural within each subtype of obesity). Univariate logistic regression was used to estimate risk for T2D, CAD, and CKD for individuals with different subtypes using MHNO as a reference group. The variables that were clinically relevant and/or significant in the univariate logistic regression, such as, sex, age, education, smoking, income, family history of diabetes, physical activity, and dietary nutrient densities (Carbohydrates %E, Fat %E and Protein %E) were adjusted in multiple logistic regression. A P <0·05 was considered statistically significant. To analyse data, we used Statistical Data Analysis Software (version 9.4; SAS Institute, Cary, NC, USA). Results Figure 1 presents the prevalence of different subtypes of obesity in the study population. The most prevalent obesity subtype was MONO, accounting for 43.3 per cent [95% confidence interval (CI): 42.6-44.%], followed by MOO (28.3%; 95% CI: 27.7-28.9%), MHNO (26.6%; 95% CI: 26.0-27.2%) and MHO (1.8%; 95% CI: 1.6-2.0%). The obese subtypes (MHO and MOO) were more common in the urban areas (urban vs. rural: MHO: 2%; 95% CI: 1.5-2.4% vs. 1.7%; 95% CI: 1.5-2%; P=0.304; MOO: 39%; 95% CI: 37.2-40.8% vs. 22.8%; 95% CI: 22-23.7%; P<0.001). MOO was more common among females (male vs. female: 23.2%; 95% CI: 22.2-24.3% vs. 32.9%; 95% CI: 31.7-34.2%; P<0.001), whereas MHO was more common in males (male vs. female: 2.1%; 95% CI: 1.8-2.4% vs. 1.5%; 95% CI: 1.3-1.8%; P=0.011).