Indian Journal of Public Health

ORIGINAL ARTICLE: DR. P.C. SEN BEST PAPER AWARD
Year
: 2019  |  Volume : 63  |  Issue : 3  |  Page : 171--177

Effect of dietary habit and physical activity on overnutrition of schoolgoing adolescents: A longitudinal assessment in a rural block of West Bengal


Arista Lahiri1, Arup Chakraborty2, Urmila Dasgupta3, Amal Kumar Sinha Roy3, Krishnadas Bhattacharyya4,  
1 Junior Resident, Department of Community Medicine, Medical College and Hospital, Kolkata, West Bengal, India
2 Assistant Professor, Department of Community Medicine, Medical College and Hospital, Kolkata, West Bengal, India
3 Associate Professor, Department of Community Medicine, Medical College and Hospital, Kolkata, West Bengal, India
4 Professor and Head, Department of Community Medicine, Medical College and Hospital, Kolkata, West Bengal, India

Correspondence Address:
Arup Chakraborty
240, Golpukur Road, Baruipur, 24 Parganas (South), Kolkata - 700 144, West Bengal
India

Abstract

Background: Overweight among adolescents has high prevalence on the eastern part of India, especially West Bengal. Objectives: The current study was conducted to estimate and compare the effects of different dietary habits and habits related to physical activity in the development of overweight and obesity among rural schoolgoing adolescents. Methods: A prospective repeated measures study was conducted on 645 schoolgoing adolescents from selected rural government-aided schools from June 2017 to December 2017. Dietary habits of the students and physical activity-related factors were the major predictors. Adjusting for the age and sex of the participants, effect of these factors on the development of overweight and obesity was analyzed by generalized estimating equations for 2 repeated measures, taken 6 months apart. Results: Most of the respondents were aged ≥16 years (56.90%), female (52.87%), Hindu (76.74%), from a nuclear family (76.74%), and studying in the secondary level (57.68%). There was a stark rise in proportion of overweight from 0.93% to 7.44%. The prevalence of unhealthy dietary habits was 68.99% at the baseline, and 66.82% on follow-up. The overall prevalence of inadequate physical activity increased to 48.68% from 47.91%. Female gender and older age group were at higher risk of being overweight or obesity. Overall fast food showed highest risk (3.04, 95% confidence interval [CI]: 1.86–4.95), while among the boys, it was with less vegetable consumption (4.64, 95% CI: 1.84–11.69). Conclusions: Strong evidence was generated of dietary practices being more rigidly related to overweight among the adolescents. Healthy dietary practices coupled with physical activity should be promoted to mitigate the risk of obesity.



How to cite this article:
Lahiri A, Chakraborty A, Dasgupta U, Roy AK, Bhattacharyya K. Effect of dietary habit and physical activity on overnutrition of schoolgoing adolescents: A longitudinal assessment in a rural block of West Bengal.Indian J Public Health 2019;63:171-177


How to cite this URL:
Lahiri A, Chakraborty A, Dasgupta U, Roy AK, Bhattacharyya K. Effect of dietary habit and physical activity on overnutrition of schoolgoing adolescents: A longitudinal assessment in a rural block of West Bengal. Indian J Public Health [serial online] 2019 [cited 2020 Oct 22 ];63:171-177
Available from: https://www.ijph.in/text.asp?2019/63/3/171/267210


Full Text



 Introduction



The prevalence of overweight and obesity is on a rising trend globally with a three times prevalence in 2016 compared to 1975.[1],[2] Regarded as a global epidemic, it is among the defined cluster of noncommunicable diseases (NCDs) called “New World Syndrome,” creating an enormous socioeconomic and public health burden, especially in middle- and low-income countries.[1],[3] The World Health Organization (WHO) estimates around 340 million children and adolescent to be overweight in 2016 with half of the under-five years burden of overweight and obesity being attributed to Asia.[2]

The data from National Family Health Survey-4 (NFHS-4) show an increase of the burden of overweight and obesity from 12.6% in NFHS-3 to 20.6% among women and 9.3%–18.9% among men. The prevalence in rural India was estimated to be 14.3% and 15.0% among men and women, respectively.[4] While the increasing trend was observed in the perspective of West Bengal as well, the proportions of adult (15–49 years) men and women being overweight or obese was observed to be 11.2% and 15.0%.[5] The largest study conducted regarding overweight, obesity, and risk factors of NCDs in India – the Global School-Based Student Health Survey (GSHS) in 2007 estimated around 10% overall burden among adolescents (13–15 years) where the prevalence of overweight and obesity was observed higher in boys compared to girls.[6] The trend of this morbidity was observed to be higher in the northern and also in the eastern parts of India among the schoolgoing adolescent age group.[7],[8] Several cross-sectional studies have reported an increasing prevalence of overweight and obesity in rural West Bengal, especially among children and adolescents.[7],[9],[10],[11]

While overweight and obesity contribute to a greater mortality burden due to NCDs, there are several factors associated, most common being unhealthy dietary habits and inadequate physical activity.[2],[3],[12] Several studies have shown, while family history, genetic factors, medical conditions also contribute in development of this condition, the mentioned behavioral and life-style factors are given more importance because of their strong association and the modifiable property.[7],[9],[10],[13],[14],[15],[16],[17] However, regarding the risk factors, there is lack of studies with serial measurements to infer on the causal effect of these factors in the development of overweight and obesity in the adolescence.[7] There are several follow-up studies in this regard, but most of them reported trends related to prevalence of obesity.[18],[19],[20]

With increasing urbanization, the problem of overweight, especially in the rural areas and among adolescents presents a steep challenge in healthcare. With the adoption of “life-cycle approach,”[21],[22],[23] the behavioral aspect of adolescent health has gained importance. Dietary factors and physical activity being the cardinal modifiable factors, measurement of their effects on development of overweight and obesity has become even more necessary. The current study was conducted to estimate and compare the effects of different dietary habits and habits related to physical activity in development of overweight and obesity among schoolgoing adolescents in rural parts of West Bengal.

 Materials and Methods



Study design and participants

A prospective follow-up study was conducted among adolescents (14–19 years) studying at secondary and higher-secondary level (Classes IX to XII) in four selected Bengali-medium coeducation government-aided schools of Barasat II block of West Bengal. The participants, who were permanent residents of the rural areas, were surveyed at the beginning, and a follow-up assessment was done after 6 months. Thus, two repeated measures were taken on each individual participant. The students who were absent during the days of initial data collection were excluded from the study. Those with nonresponse or partial response to questions related to dietary habit or physical activity were excluded during analysis.

Sampling and recruitment

The sample size for the study was calculated in STATA 14.2 software (Stata Corporation, College Station, Texas, USA) with the help of “lbpower” module.[24] For calculation of the sample size, the difference in percentage prevalence of overweight and obesity among males and females were considered. Now, for 6 months (i.e., half-year) period and two repeated measures, we take the detectable percentage point difference as 0.15 with repeated measure correlation at 0.5, at significance level of 5% and a power of 90% we get a minimum required sample size of 281 males and equal number of females.[10],[24] Accounting for 10% attrition, the revised sample size was 625 overall. Now, the block was divided into four geographical quadrants, and based on available data on population and school enrollment (available from block office), participants were selected by probability proportional to size method from four schools selected at random (one from each quadrant). Based on the enrollment and attendance, one section from each class (of IX-XII) was selected for the study. A total of 645 participants (304 male and 341 female respondents) were ultimately included in the analysis after getting response from 705 participants on initial data collection.

Study variables and data collection

The data collection was conducted on June 2017 (baseline) and on December 2017 (follow-up). Data on each visit were collected in two parts. A predesigned pretested questionnaire was prepared reviewing the GSHS instrument[25] and the WHO-STEPS instrument[26] and comprised of a section on physical activity and another section on dietary habits. There was a question regarding family history as well. Sociodemographic information of the participants was collected during the initial visit. The questionnaire was translated to Bengali and was then backtranslated to English by two different experts. Validity of the questionnaire was checked by statistical tests, the discussion of which is beyond the scope of the current article (Cronbach's alpha[27] was estimated to be 0.82, ensuring good statistical reliability[28]). This instrument was administered among students of a selected section during a single period all at once. Each of them, already allotted an identification number against their roll numbers and name, were then subjected to measurements of height and weight by standardized stadiometer (IndoSurgicals® Height Measuring Scale) and bathroom type weighing machine (MCP® Analog Mechanical Weighing Scale), respectively, following standard protocol. Body mass index (BMI) was calculated as weight (in kilograms)/height (in meters).[2] The WHO adolescent BMI percentile chart was used to classify the nutritional status of the children.[29] A follow-up visit was done after 6 months to the selected schools, and the same process was repeated on those who were already allotted the identification number from the previous visit.

Study variables and statistical analysis

The data collected was compiled in Epi Info™ 7 (Epi Info™, Centers for Disease Control and Prevention, Atlanta, USA) and analyzed in STATA 14.2 software (Stata Corporation, College Station, Texas, USA). Nutritional status (presence of overweight and obesity) of the students was considered the time-dependent (repeated measure) outcome in the study. Gender of the participants was the only time-invariant predictor of nutritional status. Age of the students, dietary habits, and practices related to physical activity were considered time-dependent predictors directly influencing the occurrence of overweight or obesity. Family history of any NCDs or its risk factors (including obesity) was taken as a time-invariant predictor. Those who did not respond to the question were counted to have no family history for the sake of simplicity of analysis. Among dietary habits, episodes of going hungry (i.e., having no food), infrequently eating vegetables or fruits, frequent intake of soft drinks and fast foods were the variables used. Infrequent walking to commute to school, major duration of sitting activities, and infrequent other daily physical work were indicative of physical activity status of the participants. All these variables were measured in terms of a 5-point Likert-type scale. To arrive at a single variable describing the dietary factors, a principal component analysis was done on the related variables, and single factor with eigenvalue >1 was taken and then dichotomized into healthy or unhealthy habits. Physical activity variables were also reduced similarly to a single indicative factor.

The variables, both time-varying and time-invariant, were analyzed with the help of population average model or more commonly known as marginal population model – Generalized Estimating Equations (GEEs)[30],[31] to find the effect of the predictors with change in time. Robust standard errors were used to achieve an unbiased model fit, especially with respect to outliers. Effect size (epidemiological risk) was estimated in terms of the risk ratio obtained, and the 95% confidence intervals (95% CIs) were reported. P value was considered significant at <0.05. The interaction in-between was considered, and the model was compared to GEE model without the interaction terms. Although both the models showed overall fit (P χ2 < 0.001), model with interaction did not differ from the main model significantly. Next, another model was created with the composite dietary habit and physical activity variables to estimate how much an unhealthy habit overall contributes to the morbidity. In this model, interactions with age and gender w] ere included with the previously mentioned variables. In this case, however, the model with interactions differed significantly from that without interactions. A gender-dependent analysis on these factors was also performed. A statistical invariance of the baseline background information of the students excluded from the analysis was established to maintain sampling integrity and representativeness.

Ethical considerations

Ethical permission was taken from the Institutional Ethics Committee of Medical College and Hospital, Kolkata (Ref. No. MC/Kol/IEC/Non-Spon/360/11-2016, dated 19 November, 2016). Following the administrative approvals, data collection was conducted maintaining confidentiality and on obtaining written consent from the respondents and the heads of the institutions. Participation was ensured to be on a voluntary basis with assent from the students and written informed consent from the parents during parent–teachers' meetings.

 Results



Background information

The mean age of the participants was 15.9 (±1.54) years (range: 14–19 years). Most of the respondents (56.90%) were aged 16 years or below. Among the participants, majority were female (52.87%), Hindu (76.74%), and belonging from a nuclear family background (76.74%). Majority of the participants (57.68%) were studying in the secondary level (Classes IX and X). Rest of the adolescents were studying at higher secondary level. The mean per-capita monthly income was ₹ 3358.1 (±319.8). While 13.50% of the respondents belonged to Class I of BG Prasad socioeconomic status classification (December, 2016 modification),[32] 31.55% were from Class IV.

Among those who responded regarding their father's education, majority of the fathers were educated up to higher secondary level (27.41%). On the contrary, the majority of the mothers had received primary level of education (43.44%), while 19.75% were educated up to higher secondary level. While majority students reported that their fathers were involved in farming as occupation (48.12%), majority of their mothers were homemakers (69.84%).

Trend of obesity and overweight

Figure 1 depicts the trend of nutritional status of the participants over two observations 6 months apart. [Figure 1]a shows that the normal nutritional status was observed among 85.58% of the participants initially; but after 6 months, the proportion was 77.36%. There was a stark rise in proportion of overweight from 0.93% to 7.44% with obesity rising to 1.86% from previously 0.93%. Interestingly, the other spectrum of adolescent malnutrition – thinness also increased in prevalence from 12.56% to 13.33%, though not as sharp as that of overweight. [Figure 1]b and [Figure 1]c, when compared, shows that the increase in the prevalence of overweight was more in male adolescents. While girls show an improvement in thinness status over time, boys reported a higher burden of thinness as well. These observed differences were found to be statistically significant as well (P < 0.001).{Figure 1}

Trend of diet and physical activity related risk factors

[Table 1] compares the distribution of the risk factors related to dietary habits, physical activity, and family history of NCDs and risk factors between the two observation points. Among the dietary habits, the students reported that 12.40% went hungry most of the times, 47.60% were eating fruits less than once a day, and 23.88% were eating vegetables less than once a day. Regarding frequent intake of soft drinks and fast foods, the prevalence was 46.51% and 53.02%, respectively, at the time of initiation. However, on follow-up visit, all the prevalence were seen to be higher except frequent intake of junk foods, which fell down to 48.22%. Infrequent intake of vegetables was prevalent among 28.99% of the participants on follow-up. Now, this difference was statistically significant. While the prevalence of unhealthy dietary habits overall was noted to be 68.99% at the baseline, there was a marginal decrease to 66.82%. However, this difference also was not significant statistically. Risk factors pertaining to physical activity among adolescents, for example, usually not walking or cycling to school increased to 32.56% on follow-up compared to 30.80% at the baseline. However, there was a marginal decrease in prevalence of sitting activities from 14.11% at the baseline to 13.64%. Similarly, for inadequate other daily physical work, the prevalence decreased from 25.27% to 23.88%. However, the overall prevalence of inadequate physical activity increased to 48.68% from the baseline value of 47.91%. However, none of these differences were statistically significant. After completion of follow-up, 9.92% of the students were found to have reported the presence of family history.{Table 1}

It is important to note that while there is a statistical difference in the outcome variable (overweight and obesity) over time, such differences are mostly not there for the risk factors of interest. Although effect of the predictors on the outcome cannot be simply written off basing on only this disparity, it can be well understood that the cause of disparity is most likely hidden in the effect of the risk factors over time (i.e., the time interaction).

Predictors of overweight and obesity

[Table 2] shows the GEE models done for different dietary risk factors and risk factors of physical inactivity. Female gender and older age group (i.e., >16 years of age) were at higher risk of developing overweight and obesity. In the cumulative (overall) model, intake of fast showed highest risk (3.04, 95% CI: 1.86–4.95) in favor of development of overweight. The risk estimate for fast food intake causing overweight was also the highest among the girls (4.49, 95% CI: 1.90–10.64). Less fruits and vegetables consumptions were statistically linked with overweight and obesity. Those who had a regular intake of soft drinks were at 2.31 (95% CI: 1.41–3.77) times risk of getting overweight. In the GEE model for boys, older age did not have any statistically significant effect. However, similar to the cumulative model fruit and vegetables ingestion, soft drinks, and fast food, intake did have a statistically significant relationship. Among the boys, the maximum risk was observed with eating less vegetable (4.64, 95% CI: 1.84–11.69). In case of the girls, risk of older girls becoming overweight was 4.99 (95% CI: 1.71–14.58). While none of the variables related to physical activity were significant statistically in any of the models, inadequate daily work was observed to have statistically significant risk of 2.48 (95% CI: 1.11–5.55). Eating less vegetable was not statistically significant among girls. However, intake of soft drinks had a risk (4.40, 95% CI: 1.84–10.50) comparable to that of fast food intake.{Table 2}

[Table 3] summarizes the effects of overall dietary habit and physical activity adjusting for gender age group of the respondents. In [Table 3], the two-way interaction terms are incorporated to identify the significant interactions between the modifiable and nonmodifiable risk factors depicted in [Table 3]. Overall unhealthy dietary habit was found to contribute a risk of 8.62 (95% CI: 5.38–13.80), which was higher compared to risk contributed by inadequate physical activity (6.58, 95% CI: 3.92–11.03). Both these risks increase in boys and girls separately. The unhealthy dietary practices lead to 14.80 times (95% CI: 8.57–25.55) cumulative risk among the girls, which is higher compared to that in boys. On the other hand, inadequate physical activity among the boys leads to a very high cumulative risk of developing obesity. Along with the main effects, the interactions also appeared significant statistically in the mentioned models. However, this statistical significance implies the interrelationship of the modifiable and nonmodifiable risk factors.{Table 3}

 Discussion



In consonance with the findings of the other researchers,[7],[9],[15],[16],[20],[33] prevalence of obesity and overweight was observed to be on a rising note in the current study as well with a sharp rise of around 7% among the participants. Since overweight and obesity are itself time-varying dynamic condition, the difference observed is actually a net increase (newly developing overweight – those returning below overweight BMI category); therefore, this can be regarded as a “net incidence,” which is in fact a proxy marker for incidence of the morbidity under discussion. In a comparative study in West Bengal, Ghosh[9] found the prevalence of overweight and obesity to be higher in Urban area. However, the proportions depicted in that study in the rural areas were very much comparable to the baseline findings of the current study. Craig et al.[13] found that in rural South Africa, the effect of female sex on development of obesity was very high. The findings were supported by the current study through the obtained risk estimates, which were in deed high. This was also the case in several of the India studies[7],[10],[20],[33] and also consistent with NFHS-4 report.[5] Several authors studying the risk factors of obesity have concluded with behavioral modification, diet, and physical activity being their part[8],[15],[16] but were not able delineate the effect size or the relative risk of these behavioral factors. In the current study, by virtue of its longitudinal design, the risk estimates for different modifiable factors adjusting for the nonmodifiable factors were obtained.

There is strong evidence of dietary practices being more rigidly related to overweight among the adolescent. Although inconsistent conceptually, the variables of physical activity did not show statistical significance individually in the primary model [Table 3]. However, the significance of cumulative effect was derived from the interaction model. It was observed that age itself was a more important predictor of overweight and obesity among females than males. Among the dietary factors studied, fast food intake followed by soft drinks intake were the most vital risk factors identified. These are even more important in case of females. The cause may be that the girls tend to eat and drink these items more compared to boys in the rural area. However, less vegetable intake was more detrimental in case of boys, as observed in the study. The likely explanation for this lies within the fact of less consumption of vegetables by boys as compared to girls, with boys receiving more calorie and protein intensive diet. Goyal et al.[33] showed that the lifestyle factors have an enormous effect on overweight and obesity, which are similar across different socioeconomic groups. However, the invariance of effect across different groups was not studied currently, rather the cumulative or crude measure of effect was noted.

It is well understood that the risk factors of the NCDs itself form a spectrum with intricate interrelation. The current study attempted to capture the effect of two selected factors contributing to obesity adjusting for some nonmodifiable factors. One of the drawbacks in the study however remains with the use of questionnaires, as conscious falsification by the respondents would always create a skewed result. Because sections of a class were selected, the study could have encountered selection bias at the initial phase. The current study despite being school-based, an individual-level longitudinal study was conducted thus overcoming the probable source of ecological fallacy. An interventional design with a higher power (larger sample size) will help to get a more precise result in future.

The recommendations that emerged from the study were that healthy dietary practices and physical activity should be promoted to mitigate the risk of obesity. Increasing awareness among the girls regarding the ill-effects of junk foods and soft drink is also needed. Since in the rural areas, still farming remains the dominant profession, adolescents should be encouraged to consume more fruits and vegetables; this, however, is more required for the male child. The teachers and the parents should be educated and motivated about behavioral change trainings of the adolescents. With the increase in overweight and obesity in such epidemic proportions, the regular school-based health checkups are a generic requirement.

Acknowledgment

The authors would like to acknowledge the participants in the study and the staffs and faculties of the institutions. The authors would also like to thank the Indian Public Health Association for the opportunity to present this study in Dr. P. C. Sen Memorial Award Session.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

References

1Ng M, Fleming T, Robinson M, Thomson B, Graetz N, Margono C, et al. Global, regional, and national prevalence of overweight and obesity in children and adults during 1980-2013: A systematic analysis for the global burden of disease study 2013. Lancet 2014;384:766-81.
2World Health Organization. Obesity and Overweight. Available from: http://www.who.int/news-room/fact-sheets/detail/obesity-and-over weight. [Last accessed on 2018 Sep 21].
3World Health Organization. Obesity: Preventing and Managing the Global Epidemic. Available from: http://www.who.int/entity/nutrition/publications/obesity/WHO_TRS_894/en/index.html. [Last accessed on 2018 Sep 20].
4India Fact Sheet. In: National Family Health Survey – 4. Available from: http://rchiips.org/NFHS/pdf/NFHS4/India.pdf. [Last accessed on 2018 Jul 24].
5State Fact Sheet West Bengal. In: National Family Health Survey – 4. Available from: http://rchiips.org/NFHS/pdf/NFHS4/WB_FactSheet.pdf. [Last accessed on 2018 Jul 24].
6India (CBSE) 2007 Fact Sheet. Available from: http://www.who.int/ncds/surveillance/gshs/2007_India_CBSE_fact_sheet.pdf. [Last accessed on 2018 Sep 29].
7Ranjani H, Mehreen TS, Pradeepa R, Anjana RM, Garg R, Anand K, et al. Epidemiology of childhood overweight & obesity in India: A systematic review. Indian J Med Res 2016;143:160-74.
8Panda SC. Overweight and obesity and lifestyle of urban adolescent school children of Eastern state of India. Int J Res Med Sci 2017;5:4770.
9Ghosh A. Rural-urban comparison in prevalence of overweight and obesity among children and adolescents of Asian Indian origin. Asia Pac J Public Health 2011;23:928-35.
10Mittal PC, Srivastava S. Diet, nutritional status and food related traditions of Oraon tribes of new mal (West Bengal), India. Rural Remote Health 2006;6:385.
11Bhowal K, Mukherjee I, Chaudhuri D. BENGAL, INDIA Sci 2015;7:4.
12Obesity – Symptoms and Causes. Mayo Clinic. Available from: http://www.mayoclinic.org/diseases-conditions/obesity/symptoms-causes/syc-20375742. [Last accessed on 2018 Sep 25].
13Craig E, Reilly JJ, Bland R. Risk factors for overweight and overfatness in rural South African children and adolescents. J Public Health (Oxf) 2016;38:24-33.
14Jane Scott NA. Physical Activity and Sedentary Behaviors as Risk Factors of Obesity among Rural Adolescents. J Child Adolesc Behav [Internet]. 2015;03(01). Available from: http://www.esciencecentral.org/journals/physical-activity-and-sedentary-behaviors-as-risk-factors-of-obesity-among-rural-adolescents-2375-4494.1000185.php?aid=39015. [cited 2018 Sep 29].
15Meharda B, Sharma SK, Singhal G, Kumar LD. Overweight and obesity: A rising problem in India. Int J Community Med Public Health 2017;4:4548.
16Urmila K, Divya K, Sudakaran, Nambiar M. Prevalence and risk factors of obesity among higher secondary school students in urban and rural schools of North Kerala. Int J Contemp Pediatr 2017;4:1851.
17Sharma DM, Sharma DA. Prevalence of obesity and overweight amongst adolescents in rural and urban areas of Rajasthan India. Int J Med Health Res 2017;3:01-7.
18Ramachandran A, Snehalatha C, Vinitha R, Thayyil M, Kumar CK, Sheeba L, et al. Prevalence of overweight in urban Indian adolescent school children. Diabetes Res Clin Pract 2002;57:185-90.
19Gupta R, Rastogi P, Arora S. Low Obesity and High Undernutrition Prevalence in Lower Socioeconomic Status School Girls: A Double Jeopardy. Human Ecology Special Issue No. 14: 65-70.
20Subramanyam V, Jayashree R, Rafi M. Prevalence of overweight and obesity in affluent adolescent girls in Chennai in 1981 and 1998. Indian Pediatr 2003;40:775-9.
21Nair MK, Mehta V. Life cycle approach to child development. Indian Pediatr 2009;46 Suppl:s7-s11.
22Rai RK, Kumar S, Sekher M, Pritchard B, Rammohan A. A life-cycle approach to food and nutrition security in India. Public Health Nutr 2015;18:944-9.
23MoHFW. Reproductive, Maternal, Newborn, Child and Adolescent Health Programme. Available from: https://mohfw.gov.in/sites/default/files/5665895455663325.pdf. [Last accessed on 2018 Sep 28].
24Sanchez ADQ. LBPOWER: Stata Module to Calculate Approximate Power (or sample size) for Longitudinal Studies with Binary Response and two Equally Sized Treatment Groups 2015. (Statistical Software Components). Available from: https://econpapers.repec.org/software/bocbocode/s457211.htm. [Last accessed on 2018 Sep 29].
25WHO. Global School-based Student Health Survey. Available from: http://www.who.int/chp/gshs/bangladesh/en/. [Last accessed on 2018 Jan 11].
26NCDs. The STEPS Instrument and Support Materials. Available from: http://www.who.int/ncds/surveillance/steps/instrument/en/. [Last accessed on 2018 Sep 26].
27Cronbach LJ. Coefficient alpha and the internal structure of tests. Psychometrika 1951;16:297.
28Tavakol M, Dennick R. Making sense of Cronbach's alpha. Int J Med Educ 2011;2:53-5.
29World Health Organization. Growth Reference – BMI-for-Age Charts and Tables. WHO,BMI-for-age (5-19 years). Available from: https://www.who.int/growthref/who2007_bmi_for_age/en/. [Last accessed on 2018 Sep 27].
30Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics 1986;42:121-30.
31Zeger SL, Liang KY. An overview of methods for the analysis of longitudinal data. Stat Med 1992;11:1825-39.
32Prasad Social Classification Update. Prasad Social Classification Scale Update. Available from: http://prasadscaleupdate.weebly.com/real.html. [Last accessed on 2018 Sep 27].
33Goyal RK, Shah VN, Saboo BD, Phatak SR, Shah NN, Gohel MC, et al. Prevalence of overweight and obesity in Indian adolescent school going children: Its relationship with socioeconomic status and associated lifestyle factors. J Assoc Physicians India 2010;58:151-8.