Clustering of noncommunicable disease risk factors among adolescents attending higher secondary schools in Kasaragod District, Kerala, India [version 2; peer review: 1 approved with reservations]

Background Limited evidence exists on the presence of collective non-communicable disease (NCD) risk factors among adolescents in Kerala, India. We aimed to assess the prevalence and factors associated with multiple NCD risk factors and the clustering of these risk factors among adolescents in Kasaragod District, Kerala. Methods We selected 470 adolescents (mean age 16.6 years, male 53.8%) through multi-stage cluster sampling from higher secondary schools of Kasaragod district. Self-administered questionnaires were used, and anthropometric measurements were taken using standard techniques and protocols. Tobacco use, alcohol consumption, low fruits and vegetable consumption, inadequate physical activity, extra salt intake, overweight, consumption of soft drinks and packed foods were the eight NCD risk factors included.The factors associated with one, two and three or more NCD risk factors were analysed using multinomial logistic regression and the standard errors were adjusted for the four clusters. Results Risk factor clusters with two risk factors (dyads) and three risk factors (triads) were observed in 163 (34.7%) and 102 (21.7%) of the sample, respectively. Adolescents residing in urban areas (odds ratio (OR) = 3.55; 95% confidence interval (CI) = 1.45-8.73), whose father’s education level was lower (OR = 3.54; 95% CI = 1.24-10.10), whose mother’s education was lower (OR= 4.13; 95% CI = 1.27-13.51), who had restrictions on physical activity (OR = 5.41; 95% CI = 1.20-24.30) and who did not have a kitchen garden (an area where fruits and vegetables are grown for domestic use) (OR=4.51;95% CI = 1.44-14.12) were more likely to have three or more NCD risk factors compared to their counterparts. Conclusions Clustering of NCD risk factors was prevalent in more than half of the adolescents. Efforts are warranted to reduce multiple risk factors, focussing on children of low educated parents and urban residents.


Introduction
Non-communicable diseases (NCDs) such as cardiovascular diseases, cancer, diabetes, chronic respiratory diseases are the leading cause of death globally.According to the World Health Organization (WHO), NCDs contribute to 71% (41 million) of all deaths globally and 60% (5.87 million) of all deaths nationally 1 .About 4% of all NCD deaths in the year 2016 occurred in people below 30 years 2 .Exposure to NCD risk factors starts from the womb itself and aggravates in the later years of life.Prenatal exposure to tobacco and alcohol, maternal diabetes, overnutrition in utero, intrauterine growth retardation, premature birth, nutritional deficiency, and intergenerational factors have long-term impacts on health, including increased risk of adult cardiovascular disease, diabetes, etc 3 .NCD prevention is of utmost importance as it affects individuals in the productive life years and is a challenge to sustainable human development.Although lifestyle modification is an effective strategy to prevent and reduce the burden of NCDs 4 , their scope in prevention is limited in the adult stage.Compelling evidence suggests that the onset of NCDs starts in the early years of development and any intervention to combat the disease should be targeted in the early years of life 5 .One of the key actions to curtail the incidence of NCDs is to prevent and control NCD risk behaviours among the adolescent population.The adolescence stage shapes the nutritional, physical activity, and other lifestyle behaviours that are mostly taken forward to adulthood 6 .As per the Global Adult Tobacco Survey (GATS)-2, adolescents in the age group 15-19 years contribute to over 30% of tobacco users among youth in the age 15-24 years 7 .Within India, Kerala is the most advanced state in epidemiological transition (Lancet nations within a nation) and has a high prevalence of behavioural and metabolic risk factors of NCD 8 .According to the latest NCD risk factor survey in the state using the World Health Organization (WHO) stepwise approach to surveillance (STEPs) approach, the state had a diabetes prevalence of 19.2% and a hypertension prevalence of 30% in the adult population of the state 9 .Abdominal obesity was 60% (men 39%, women 73%) 9 .Studies conducted among adolescents reported the presence of soaring prevalence of certain NCDs risk factors globally and in India.
The need to target adolescents to prevent the development of NCD risk factors is globally recognized."Catch them young and keep them healthy" has become a catchphrase in global NCD prevention 10 .Though it is known that the collective impact of multiple risk factors could increase NCD risk, limited studies captured the presence of collective NCD risk factors and clustering of these risk factors among adolescents in Kerala.Given the context, this study was conducted with an objective to assess the prevalence of multiple NCD risk factors and clustering of these risk factors among adolescents in the Kasaragod District of Kerala and to find out the factors associated with the multiple risk factors among adolescents.

Study design
The study was conducted using multi-stage stratified cluster sampling method in February 2018.Amongst the two educational districts in Kasaragod district, one was selected randomly.This study targeted only higher secondary students i.e., from the age group of 15 -19 years (late adolescent stage) from both public and private schools.Keeping in view of the anticipated prevalence of 5.2% 11 , confidence interval of 95%, precision factor of 3%, a design effect of two and non-response rate of 10%, a total sample size of 470 was estimated.Those who were physically absent during the data collection were excluded from the study.
A self-administered structured questionnaire was used for data collection 12 .The questionnaire captured socio-demographic characteristics, known predictors of behavioural risk factors and NCDs such as family history of obesity, body image perception, physical activity behaviour and factor influencing it, exposure to tobacco, alcohol use among parents and dietary habits.In addition to this, the Indian Adolescent Health questionnaire (IAHQ), validated in India 13 and B.G Prasad scales 14 were used to measure the outcome variables.The modified questionnaire was piloted in a smaller sample of 15 students for the feasibility and comprehension of the questions among the students, who were excluded in the final survey.Minor rephrasing of the questions was made as per the feedback and the questionnaire was found to be comprehendible and feasible to be employed and was finalized thereafter.
All behavioural risk factors were based on the Indian Adolescent Health Questionnaire.Physical activity was defined as any activity that increases the heart rate and makes them get out of breath some of the time.If the participants were physically active for at least 30 minutes per day for five days in the previous week of the survey they were considered as physically active.Tobacco use was defined as the use of any form of tobacco products (smoking and smokeless) in the previous month.Current alcohol use was defined as use of any alcoholic products such as wine, vodka, beer, or whiskey, etc in the last six months except drinking few sips of wine for a religious purpose.Ever user was defined as one who used tobacco products/ consumed alcohol at least once in their life-time.Inadequate intake of fruits and vegetables was defined as consuming less than five servings of fruits and vegetables per day.Anthropometric measurements such as height in centimetres and weight in kilogram were taken at school on the day of the survey using standardized instruments like SECA wall-mounted measuring scale and SECA electronic weighing scale using standard protocol 15 .Body mass index (BMI) for age was computed using WHO growth reference for school aged children and adolescents 16 .Adolescents were categorized overweight if

Amendments from Version 1
In the revised version of the manuscript, the factors associated with one, two and three or more NCD risk factors were analysed using multinomial logistic regression and the standard errors were adjusted for the four clusters.The independent variables that were potential confounders and effect modifiers such as age, education, gender and behavioural risk factors were included in the model.
Any further responses from the reviewers can be found at the end of the article BMI for age was greater than one standard deviation from median 16,17 .

Ethical approval
Ethical clearance was obtained from the Institutional Human Ethics Committee (IHEC) of the Central University of Kerala (CUK/IHEC/2018/018, 19 February 2018).Authorization for data collection was taken from the higher secondary district coordinator and the respective school heads.Students who met the inclusion criteria, were briefed about the study and provided with participant information sheet and informed consent, a day prior to the data collection.Participants were recruited only after obtaining consent from parents and assent from the students.The survey and anthropometric measurements were taken on the subsequent day at the school.Anthropometric measurements were taken after ensuring sufficient privacy to the participants.Confidentiality of the individuals was ensured by masking the personal identifiers with a participant identification number.

Data analysis
The data were entered, cleaned and analysed using SPSS version 23.0.Frequencies, proportions and percentages were used to descriptively analyse the data.Prevalence of individual risk factors were analysed.Descriptive analysis of the clustering of NCD risk factors were done.The factors associated with one, two and three or more NCD risk factors were analysed using multinomial logistic regression and the standard errors were adjusted for the four clusters.The independent variables that were potential confounders and effect modifiers such as age, education, gender and the behavioural risk factors were included in the model.

Results
A sample of 470 adolescents in the ages 16-19 years were analysed 12 .The sample comprised of 53.8% (n=253) males and 46.2% (n=217) females.The mean age (in completed years) of the participants was 16.6 years.The majority of the participants was from urban areas (66.4%).A detailed outline of the socio-demographic characteristics of the sample are given in Table 1.
Prevalence of NCD risk factors NCD risk factors were highly prevalent among the study sample.This study assessed the prevalence of eight major NCD risk factors.Among the NCD risk factors, consumption of packed food was most prevalent, with 67% adolescents reporting consumption of packaged food in the last one week.Prevalence of inadequate fruit and vegetable intake was 49%, followed by physical inactivity (41.9%).The least prevalent risk factor was tobacco use (smoke form or smokeless form) of 4.7% (n=22).The detailed outline of the eight risk factors such as fruit and vegetable intake, consumption of soft drinks, overweight, physical inactivity, alcohol consumption, tobacco use, extra salt intake and consumption of packed food is given in Table 2.

Clustering of NCD risk factors
Among 470 adolescents sampled, at least one of the eight NCD risk factors were observed in 94.1% (n= 442) of the sample.Interestingly, the NCD risk factors were found to be clustering with each other in most of the sample.Five of the eight NCD risk factors were reported by 5.1% (n=24) of the total sample.Risk factor clusters with two risk factors (dyads) and three risk factors (triads) were observed in 163 (34.7%) and 102 (21.7%) of the sample respectively.Overall, 39.8% of the total sample were found to have at least three NCD risk factors.More than five risk factors were not reported by anyone.Figure 1 represents the NCD risk factor profile of the sample.
Dyads and triads of NCD risk factors were highly prevalent among the adolescents accounting for a combined prevalence of 56.4% (n=265).We decomposed the dyads and triads to identify the NCD risk factor combinations.Among the dyads, the combination of "inadequate fruit and vegetable intake + consumption of packaged food" were most prevalent.Among the triads, the combination of "extra salt + consumption of packaged food + physical inactivity" were most prevalent.Figure 2 and Figure 3 report the decomposition of NCD risk factor cluster dyads and triads, respectively.

Factors associated with clustering of NCD risk factors
Unadjusted odds ratios were computed to identify the factors associated with clustering of NCD risk factors among the adolescents.Factors found to have significant unadjusted odds were included into multi-variate analysis.Multivariate analysis was conducted using multinomial logistic regression with NCD risk factors (i.e., one risk factor, two risk factors, three or more risk factors) as dependent variable.The multinomial logistic regression yielded a significant model with an acceptable model fit.Several predictive factors such as gender of the participant, place of living, educational status of mother, father's education, restrictions on physical activity, and having an income generating job were found to be significantly predicting the clustering of NCD risk factors.The resultant adjusted odds ratios with 95% confidence intervals obtained through multinomial logistic regression are outlined in Table 3.

Discussion
The study was conducted to assess the prevalence of NCD risk factors among adolescents and identify the clustering of risk factors and their correlates.Our study found that NCD risk factors among the adolescents were majorly on unhealthy diet and physical inactivity.High prevalence of these risk factors among Indian adolescents were documented in earlier studies.A study comparing NCD risk factors among adolescents in five southeast Asian countries observed that over 85% of the adolescents in India had inadequate fruit and vegetable consumption (i.e., < 5 servings per day) 18 .While limited evidence exists concerning the consumption of packaged food among adolescents in Indian context, studies from other developing countries argue that affordability of packaged foods, peer influence, absence of healthy alternatives and perception of packaged foods as safer options make them popular food choice among adolescents 19 .
Interestingly, the study observed the clustering of the NCD risk factors among the adolescents.Among the risk factor dyads, "Inadequate fruit and vegetable intake along with consumption of packaged foods" was the most prominent followed only by the dyad of "inadequate fruit and vegetable intake plus inadequate physical activity".Among NCD risk factor triads "extra salt + consumption of packaged food + physical inactivity", "inadequate fruit and vegetable intake + consumption of packaged food + physical inactivity" and "alcohol+ extra salt + inadequate fruit and vegetable consumption" were among the major ones.
A recent study from north India reported that physical inactivity and inadequate fruit and vegetable intake make up the largest of the behavioural risk factor dyads among adolescents 20 .While in our study inadequate fruit and vegetable intake, physical inactivity and consumption of packaged food were found to be strong contributors to NCD risk factor clusters, earlier studies reported obesity and overweight as predominant risk factors in NCD clusters among adolescents in north India 21 .Physical inactivity along with consumption of packaged food during adolescent period is likely to contribute to overweight while these adolescents become adults.Several factors were found to be significantly predicting the NCD risk factor clustering among adolescents.Being a female, living in urban area, father having an education of up to higher secondary schooling, mother's education of graduation and above, and possessing an income generating job were found to be significantly predicting the clustering of NCD risk factor dyads among the adolescents.In the study it was found that females had higher odds (OR-5.51) of having at least two NCD risk factors (dyads) compared to males.This is in contrast to a recent study from north India which reported a higher prevalence of NCD risk factor dyads among males compared to females 20 .Similarly, adolescents from urban regions had higher odds of NCD risk factors and their clustering compared to rural counter parts.It could primarily be due to the reason that adolescents in urban region have better transport facilities, fewer possibilities to undertake physical activity, easy availability of unhealthy foods, and other risk factors.The higher prevalence of NCDs and NCD risk factors among urban adults is well known 22 .An interesting observation made with respect to NCD risk factor clustering was with regard to restriction in physical activity.Adolescents who reported restriction in physical activity had a high odds (OR = 5.40, 95% CI = 2.83-10.32) of developing NCD risk factor clusters with three or more NCD risk factors.Literature from other developing country settings reported that parenting practices influence development of NCD risk factors among adolescents 23 .While restrictions on physical activity prevent the development of certain peer influenced NCD risk factors such as tobacco and alcohol, it can substantially increase the chances for NCD risk factors of physical inactivity, overweight, consumption of packaged food etc. Tobacco use was the least prevalent risk factor among this population.Tobacco consumption in most Indian states reduced as per the Global Adult Tobacco Survey 2 and Kerala reported the highest reduction of tobacco use among the major Indian states 7 .Therefore, this finding is in line with the findings of the GATS survey.
One of the limitations of the study is that it surveyed adolescents attending an educational institution and hence may not be representative of the community Behavioural risk factors such as physical activity, diet, alcohol and tobacco use were self-reports may likely have reporting bias.
In conclusion, there was high prevalence of individual NCD risk factors and risk factor clusters among the adolescents in Kasaragod, Kerala.Most NCD risk factors were dietary in nature, specifically around consumption of packaged food or inadequate consumption of fruits and vegetables.Indian policy environment gives a lesser emphasis to encourage healthy eating among adolescents compared to its LMIC counter parts 24 .
There is a need to prioritize healthy eating by the governments, education department and schools.Moreover, targeted interventions should also focus on improving physical activity and preventing the initiation of alcohol and tobacco use.
The manuscript on Clustering of noncommunicable disease risk factors among adolescents attending higher secondary schools in Kasaragod District, Kerala, India, is of public health important and well written, in which the authors have justified the study by stated the problem and burden of Non-communicable diseases globally and in India.In addition, the authors stated that all health risk behaviour usually "started" during adolescence and limited information required the study to be carried out in Kerala.The authors clearly stated how the sampling was carried out and the valid tool was used for data collection.
However, there are some lacking aspects of the analysis the results, in which the authors only categorised the clustering based on the descriptive aspect, while there are statistic methods such as clustering analysis which can further enhance the quality of the analysis.In addition, the authors did not describe on the fit of the multinomial logistic model, in which several assumption need to be fulfilled to ensure the results of the study are valid.(such as linearity, no outlier etc).In addition, the possible two way interaction between the independent variables (IV) need to be tested to ensure no "Effect Modification" among IV.
The results are presented in systematic way and discussion is quite comprehensive (based on the results in the text).But this can be improved by expanding it to the findings from more developed and developing countries, (consistent and contrast and why) and not limited it on the plausible reasons for the findings in this study.Reviewer Expertise: adolescent health, smoking, non-communicable diseases I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

Is the work clearly
and accurately presented and does it cite the current literature?YesIs the study design appropriate and is the work technically sound?YesAre sufficient details of methods and analysis provided to allow replication by others?PartlyIf applicable, is the statistical analysis and its interpretation appropriate?PartlyAre all the source data underlying the results available to ensure full reproducibility?YesAre the conclusions drawn adequately supported by the results?YesCompeting Interests: No competing interests were disclosed.

Table 3 . Predictors of non-communicable disease (NCD) risk-factors among adolescents: Results of multinomial logistic regression
. (CI=confidence interval, Standard errors adjusted for the four clusters).