Understanding the linkage between different measure of Obesity/overweight and Pre-diabetes/diabetes Prevalence among Adolescents (10-19 years) in India: A Gendered perspective

The International Diabetes Federation (IDF) estimated that 1.1 million children and adolescents aged 14– 19 years are living with diabetes. Diabetes is a chronic, progressive disease characterized by elevated levels of blood glucose. It is also recognized as a complex disease that affects people of different ages due to different causes. The present study aims to estimate the prevalence of pre-diabetes/diabetes at the national level. Additionally, the respective study determines the factors associated with pre-diabetes/diabetes conditions among adolescents at the national level. The data for this study was carried out from Comprehensive National Nutrition Survey (CNNS), the rst-ever nationally representative nutrition survey of children and adolescents in India. The study used a sample size of 17,865 male and 17,965 female adolescents for the analysis. Descriptive statistics, bivariate analysis, and logistic regression analysis were done to carve out the results.

across the low and middle-income countries [2]. There is a global effort to curb down the rise of diabetes and obesity by 2025; however, a steady rise in the prevalence of diabetes is a matter of concern [3].
India alone contributes a larger proportion to premature deaths from non-communicable diseases (NCDs), including diabetes across all socio-economic groups [4]. Achieving the global agenda of Sustainable Development Goals (SDGs) by 2030, by reducing one-third of premature mortality from NCDs-including diabetes, a universal healthcare access approach is therefore required in most developing countries, including India (3)(4). India is currently going through the nutritional transition and it is evident with a rise in the prevalence of overweight and obesity [6]. Also, estimation shows that nearly 166 million adults are overweight/obese in 2016 in India [3]. In the case of adults who are affected by diabetes, the numbers are estimated to be nearly 73 million, the second-largest in the world [1].
Diabetes is a chronic, progressive disease characterized by elevated levels of blood glucose levels. It is also recognized as a complex disease that affects people of different agesdue to different causes. There are three main types of diabetes, type-1 diabetes (T1D), type-2 diabetes (T2D), and gestational diabetes mellitus (GDM) [3]. The type-1 diabetes incidence rate is increasing among the younger population and the contributing factors have remained unclear. Though there are factors that affect the heterogeneity of the respective disease such as genetic factors, behavioural risk factors, food and dietary habits, physical inactivity, lifestyle changes.
The spread of pre-diabetes/diabetes among the population is witnessed in light of socio-economic groups, rural-urban, sex, and family history of diabetes [15]- [17] Lack of physical activities and having a sedentary lifestyle among adolescents made responsible for cardiovascular disease and obesity-related risk factors which in turn may cause menace of pre-diabetes/diabetes among them [11], [18], [19].
Moreover, Triglycerides, Cholesterol, High-density lipoprotein (HDL), and Low-density lipoprotein (LDL) levels were found to be signi cant factors for the prevalence of pre-diabetes/diabetes among adolescents [20], [21]. The present study utilized the different forms of obesity which may affect the chaos of prediabetes/diabetes differently. Previous literature measured the ve types of obesity in their study. For instance, obesity measured through body mass index [21], waist circumference [21], waist to hip ratio [22], Subscapular skinfold thickness (SSFT) [23] and Triceps skinfold thickness (TSFT) [23].
There exists the dearth of literature focusing on the association of pre-diabetes/diabetes with obesity levels among adolescents at the national level in India. Therefore, the present study aims to estimate the prevalence of pre-diabetes/diabetes at the national level. Additionally, the respective study determines the factors for pre-diabetes/diabetes conditions among adolescents at the national level. The study hypothesized that there was no association between pre-diabetes/diabetes conditions with obesity levels and physical activity among adolescents in India.

Methods
The data for this study was carried out from Comprehensive National Nutrition Survey (CNNS), the rstever nationally representative nutrition survey of children and adolescents in India. The survey collects the data on the status of pre-school (0-4 years), school-age children (5-9 years), and adolescents (10-19 years) through interviews, a comprehensive set of anthropometric measures and biochemical indicators. The CNNS covered all 30 states of India using a multi-stage survey design for the selection of households and individuals aged 0-19 years. The data has been collected with the written consent of the participants. The survey was implemented under the guidance of the MoHFW, UNICEF, a Technical Advisory Group (TAG), and the Center for Disease Control (CDC). Data can be obtained on request from Population Council (India o ce) website. The detailed methodology, sampling design, and data collection procedure were published in the survey report [24]. The CNNS collected data from a total of 38,060 children aged 0-4 years, 38,355 children aged 5-9 years, and 35,830 adolescents age 10-19 years. The study used a sample size of 17,865 male and 17965 female adolescents for the analysis.

Outcome variable
The outcome variable was pre-diabetes/diabetes conditions among adolescents aged 10-19 years. The pre-diabetes/diabetes condition was measure through glycosylated haemoglobin (HbA1c) [25]. The cutoff points for pre-diabetes/diabetes among adolescents were as: pre-diabetes (> 5.6% and ≤ 6.4%) and diabetes (> 6.4%) [26]. We merged diabetes and pre-diabetes into one category due to the small sample in the diabetes condition. The outcome variable was divided into two categories as 0 "no prediabetes/diabetes" and 1 "pre-diabetes/diabetes".

Exposure variables
Body mass index was categorized as not overweight/obese (BMI ≤ + 1 SD) and overweight/obese (BMI > + 1 SD) group, abdominal obesity was categorized as no (waist circumference-for-age ≤ + 1 SD) and yes (waist circumference-for-age > + 1 SD), Triceps skinfold thickness (TSFT) was grouped as not overweight (TSFT-for-age ≤ + 1 SD) and overweight (TSFT-for-age > + 1 SD) and Subscapular skinfold thickness (SSFT) was divided into two categories: not overweight (SSFT-for-age ≤ + 1 SD) and overweight (SSFTfor-age > + 1 SD). Physical activity was categorized as a low, medium, and high using principal component analysis approach as15 questions were used to generate the respective variable. Serum triglycerides (assessed by spectrophotometry and enzymatic endpoint method) was categorized into low (< 130 mg/dl) and high (≥ 130 mg/dl), cholesterol (assessed by spectrophotometry using cholesterol oxidase esterase peroxidase) was recoded as low (< 200 mg/dl) and high (≥ 200 mg/dl), low-density lipoprotein (LDL) (assessed by spectrophotometry and direct measure cholesterol oxidase) was categorized as low (< 130 mg/dl) and high (≥ 130 mg/dl) and high-density lipoprotein (HDL) cholesterol (assessed by spectrophotometry and direct measure polyethylene glycol-modi ed cholesterol oxidase) was recoded as low (< 40 mg/dl) and high (≥ 40 mg/dl). Age was categorized into early adolescents (10-14 years) and late adolescents (15-19 years), education was divided into no schooling, dropouts and currently studying, Media exposure was categorized as no "no exposure to television & radio & newspaper & internet" and yes "exposure to television or radio or newspaper or internet), parent education was recoded as (both uneducated, anyone educated and both educated), parent diabetes status was categorized as (no or anyone had), the diet was categorized as (vegetarian and non-vegetarian), Junk food consumption was categorized as (no, 1-2 days and 3-7 days), substance use was categorized as (no and yes). Caste was categorized as (Scheduled tribe/Scheduled caste (SC/ST) and non-Scheduled tribe/Scheduled caste), religion was categorized as (Hindu, Muslim, Christian, and Others), wealth quintile was categorized as (poor, middle and rich), the residence was categorized as (urban and rural) and region were divided as north, central, east, north-east, west, and south.

Statistical analysis
Descriptive analysis is used to show the sample pro le of the respondents. Further, bivariate analysis was carried out to estimate the prevalence of pre-diabetes/diabetes by selected variables, and a proportion test was applied to test the signi cant difference between male and female adolescents for prediabetes/diabetes. Finally, multivariate logistic regression was performed to nd out the effects of different predictors on pre-diabetes/diabetes. The results are presented in the form of an odds ratio (OR) with a 95% con dence interval (CI). The model is usually put into a more compact form as follows: Where are the regression coe cient indicating the relative effect of a particular explanatory variable on the outcome, the study further examined the possible interaction between body mass index, abdominal obesity, triceps skinfold thickness (TSFT), subscapular skinfold thickness (SSFT), and gender in models 2, 3, 4, and 5 respectively.

Results
Anthropometric and socio-demographic pro les of adolescents aged 10-19 years are presented in Table 1. Overall, around ve percent of male and female adolescents were overweight/obese and about two percent and one percent male and female adolescents have abdominal obesity. Nearly, ve percent male and two percent female adolescents had triceps skinfold thickness whereas seven percent male and four percent female adolescents had subscapular skinfold thickness. About 14 percent male and 18 percent of female adolescents had high serum triglycerides. Moreover, almost the same proportion of adolescents (3% vs. 5%) had high cholesterol and low-density lipoprotein respectively. The proportion of adolescents was higher in the 10-14 years age group irrespective of gender. The majority of adolescents are currently studying and, around 17% male and 21% of female adolescents were school dropouts. More than three-fourth of adolescents had some mass media exposure. One-fourth of the parents had no schooling and a very low proportion of parents (anyone) had diabetes (4%). About 81 percent male and 73% of female adolescents were non-vegetarian. Seven out of ten adolescents have consumed 1-2 days of junk food. The substance use was higher among male (8%) adolescents than female ones. One-third of adolescents belonged to SC/ST group and the majority of adolescents were Hindu. Two-fth of adolescents belonged to rich families and one-fourth of adolescents lived in urban areas.
Gender differentials for pre-diabetes/diabetes among adolescents presented in Fig. 1. Prediabetes/diabetes was more prevalent among 15 years of adolescent and it is lower in 10 years of the adolescent. Though, the prevalence of pre-diabetes/diabetes was higher among male adolescents compared to females irrespective of their age.
Table 2shows the gender differences in the prevalence of pre-diabetes/diabetes among adolescents aged 10-19 years across different socio-demographic characteristics in India. The prevalence of prediabetes/diabetes was higher among male adolescents across all the selected socio-demographic characteristics and differences were statistically signi cant. Substantial differences are observed among adolescents those who were having abdominal obesity (11.7%), overweight (SSFT) (9.8%), belonged to Northeast region (10.1%), consumed junk food for 3-7 days (8.5%), and belonged to the middle (6.4%) and rich families (5.6%).
Binary logistic regression estimated for pre-diabetes/diabetes among adolescents aged 10-19 years were presented in Table 3. Results from model-1 found that overweight/obese adolescents had 41% higher chances [Odd ratio (OR), 1.41; CI: 1.08-1.82] to get pre-diabetes/diabetes compared to not overweight/obese adolescents. Moreover, SSFT has a signi cant effect on pre-diabetes/diabetes among adolescents. Overweight adolescents (SSFT) were signi cantly more likely (OR, 1.49: CI: 1.15-1.94) to suffer from pre-diabetes/diabetes than those who were not overweight. There was a negative relationship between the physical activity of adolescents and their pre-diabetes/diabetes condition. Adolescents involved in high physical activity were less likely to suffer from pre-diabetes/diabetes compared to those who did low physical activities. Though the result was not signi cant for physical activities.
Model 2 represents the interaction between body mass index and gender for pre-diabetes/diabetes. Interestingly, a female adolescent with not overweight/obese category were signi cantly less likely to suffer (OR, 0.72; CI: 0.62-0.83) from pre-diabetes/diabetes compared to male adolescent belonged to not overweight/obese category. Interaction between gender and abdominal obesity for pre-diabetes/diabetes condition were presented in Model 3. Male adolescents having abdominal obesity were more likely (OR, 1.26; CI: 0.74-2.17) to suffer from pre-diabetes/diabetes than a male adolescent with no abdominal obesity. Moreover, a female adolescent with abdominal obesity were signi cantly less likely (OR, 0.73; CI: 0.63-0.84) to suffer from pre-diabetes/diabetes compared to a male adolescent with no abdominal obesity. Results from Model 4 revealed the interaction between gender and triceps skinfold thickness (TSFT) for pre-diabetes/diabetes among adolescents. The likelihood of pre-diabetes/diabetes among overweight (TSFT) male adolescent were lower (OR, 0.96; CI: 0.63-1.45) than a not overweight male adolescent. However, the chances of pre-diabetes/diabetes among females with not overweight category were signi cantly lower (OR, 0.72; CI: 0.62-0.83) compared to males with no overweight category. Model 5 shows the interaction between gender and subscapular skinfold thickness (SSFT) for prediabetes/diabetes among adolescents. SSFT has a signi cant effect on diabetes among both male and female adolescents. The probability of getting diabetes was signi cantly higher (OR, 2.23; CI: 1.53-3.23) among overweight (SSFT) male adolescents than not overweight males. Moreover, a female adolescent from the not overweight category was signi cantly less likely (OR, 0.72; CI: 0.62-0.84) to suffer from prediabetes/diabetes compared to males from not overweight category.

Discussion
The overall prevalence of pre-diabetes/diabetes was 12.3% and 8.4% among male and female adolescents in India respectively. Moreover, it was found that body mass index (BMI) and subscapular skinfold thickness (SSFT) was signi cantly associated with pre-diabetes/diabetes among adolescents in India. Additionally, it was revealed that female adolescents were less likely to suffer from prediabetes/diabetes than male adolescents in every category of obesity measure.
Earlier studies found that no accessibility of proper care such as early diagnosis, diabetes care management, and blood glucose testing available in primary healthcare settings in the early phases of life for children and adolescents [15], [27]. At this moment, it needs to be diagnosed (adolescent's diabetes or pre-diabetes) that can have a lesser risk in the longer term than the person who lives with undiagnosed and untreated diabetes.
The current study reveals the fact that there was a positive association between BMI, SSFT, and low physical activity status with pre-diabetes/diabetes conditions among adolescents in India. The results were consistent with previous ndings that increased BMI was positively associated with prediabetes/diabetes conditions among adolescents [19]. Moreover, SSFT was also a signi cant risk factor of pre-diabetes/diabetes conditions among adolescents as argued by previous literature [28]- [30]. Apart from obesity, genetic and metabolic factors there are other signi cant risk factors but in the present study, those were not taken into account due to data limitations [2], [27], [31]. Additionally, low physical activity status was found to be a substantial risk factor for pre-diabetes/diabetes conditions among adolescents as found in previous studies [32]. It was too argued that even a moderate level of physical activity could help in the prevention of pre-diabetes conditions by reducing obesity levels among adolescents [32].
Dietary practices with unhealthy food such as high intake of junk foods and sugar-sweetened beverages are associated with a high prevalence of obesity/overweight that may lead to pre-diabetes/diabetes conditions among adolescents [33], [34]. This nding is consistent with ndings of the present that adolescents with junk food consumption frequency of 3-7 days a week had a high prevalence of diabetes/diabetes condition. Additionally, the menace of substance use among adolescents was a staunch factor for pre-diabetes/diabetes conditions. The nding was paralleled with the previous study that abnormal alcohol and drug use among adolescents increases the risk of pre-diabetes/diabetes conditions [35].
Although it was earlier argued that diabetes is an a uent disease and it occurs among the population from higher socio-economic strata [15], however, the present study revealed that the disease pattern is now shifting towards lower a uent population, for instance, per-diabetes/diabetes condition was more among SC/ST, poor and adolescents from rural areas especially in case of females. Further research is needed to look into the issue.
The results showed that pre-diabetes/diabetes vary across gender for instance gender differential was over 4% i.e., males are more prone to pre-diabetes/diabetes than females. Moreover, for most of the background factors males were at high risk than females. Even previous literature argued that females do have a lower probability to suffer from pre-diabetes/diabetes [36]. The plausible reasons can be derived through descriptive statistics in the present study where it was clearly visible that obesity levels and substance use were low among females and which may further protect them from the burden of prediabetes and diabetes condition in comparison to their male counterpart. Additionally, interaction models revealed that females were less likely to be pre-diabetic/diabetic in comparison to males for the obesity measure. The ndings are found to be consistent with the previous one as it was argued that female adolescents are less likely to be overweight or obese [37].

Limitation
The study has potential limitations. This study used cross-sectional data that does not allow to establish a causal relationship between variables. Despite the limitation, the study also has its own strength. This study used large scale survey data of adolescents which is the rst-ever survey on nutrition and gives reliable estimates on diabetes at the national level.

Conclusion
Our study found that pre-diabetes/diabetes was highly prevalent among adolescents especially among males in India. Interaction models in the present study clearly reveal the fact that females are less likely to suffer from pre-diabetes/diabetes than males. BMI and SSFT are the two most important predictors of pre-diabetes/diabetes among adolescents in India. The high prevalence of pre-diabetes/diabetes among female adolescents from lower socio-economic strata poses a high concern in the Indian scenario. The emerging type 1 diabetes (T1D) among children and adolescents cannot be prevented with existing knowledge [3]. However, type 2 diabetes can be prevented with effective approaches available, and premature deaths can be curtailed off [3]. To avoid diabetes and related risk factors, regular exercise, maintaining a healthy diet, avoiding substance use, and controlling blood pressure and lipids as recommended by WHO guidelines [3]. The present study also suggests that to prevent prediabetes/diabetes conditions among adolescents certain health interventions are required. Utilization of knowledge, awareness, and programme (KAP) re-intervention to control obese/overweight among adolescents on the one hand and the regular health check-ups among adolescents on the other hand will surely help to curtail pre-diabetes/diabetes conditions among adolescents in India. Adolescent subpopulations are out of the coverage of the special programme that addresses the crucial public health The study is based on secondary data which is publically available on request from cnns.pc@gmail.com. Therefore ethical approval and consent to participate is not required.

Consent for publication
Not applicable

Availability of data and materials
We have provided details of the data in the methodology section. The CNNS data can be obtained on request from the Population Council, Delhi (India). The report and the survey tools are also available on the website:https://www.popcouncil.org/uploads/pdfs/2019RH_CNNSreport.pdf and cnns.pc@gmail.com respectively.