Exercise and eating habits among urban adolescents: a cross-sectional study in Kolkata, India

Background Unhealthy eating and lack of exercise during adolescence culminated into earlier onset and increasing burden of atherosclerotic cardiovascular diseases (CVDs) worldwide. Among urban Indian adolescents, prevalence of these risk factors of CVD seemed to be high, but data regarding their pattern and predictors was limited. To address this dearth of information, a survey was conducted among urban adolescent school-students in Kolkata, a highly populated metro city in eastern India. Methods During January–June, 2014, 1755 students of 9th-grade were recruited through cluster (schools) random sampling. Informed consents from parents and assents from adolescents were collected. Information on socio-demographics, CVD-related knowledge and perception along with eating and exercise patterns were collected with an internally validated structured questionnaire. Descriptive and regression analyses were performed in SAS-9.3.2. Results Among 1652 participants (response rate = 94.1%), about 44% had poor overall knowledge about CVD, 24% perceived themselves as overweight and 60% considered their general health as good. Only 18% perceived their future CVD-risk and 29% were engaged in regular moderate-to-vigorous exercise. While 55% skipped meals regularly, 90% frequently consumed street-foods and 54% demonstrated overall poor eating habits. Males were more likely to engage in moderate-to-vigorous exercise [adjusted odds ratio (AOR) = 3.40(95% confidence interval = 2.55–4.54)] while students of higher SES were less likely [AOR = 0.59(0.37–0.94)]. Males and those having good CVD-related knowledge were more likely to exercise at least 1 h/day [AOR = 7.77(4.61–13.07) and 2.90(1.46–5.78) respectively]. Those who perceived their future CVD-risk, skipped meals more [2.04(1.28–3.25)] while Males skipped them less [AOR = 0.62(0.42–0.93)]. Subjects from middle class ate street-foods less frequently [AOR = 0.45(0.24–0.85)]. Relatively older students and those belonging to higher SES were less likely to demonstrate good eating habits [AOR = 0.70(0.56–0.89) and 0.23(0.11–0.47) respectively]. A large knowledge-practice gap was evident as students with good CVD-related knowledge were less likely to have good eating habits [AOR = 0.55(0.32–0.94)]. Conclusions CVD-related knowledge as well as eating and exercise habits were quite poor among adolescent school-students of Kolkata. Additionally, there was a large knowledge-practice gap. Multi-component educational interventions targeting behavioral betterment seemed necessary for these adolescents to improve their CVD-related knowledge, along with appropriate translation of knowledge into exercise and eating practices to minimize future risk of CVDs.


(Continued from previous page)
Conclusions: CVD-related knowledge as well as eating and exercise habits were quite poor among adolescent schoolstudents of Kolkata. Additionally, there was a large knowledge-practice gap. Multi-component educational interventions targeting behavioral betterment seemed necessary for these adolescents to improve their CVD-related knowledge, along with appropriate translation of knowledge into exercise and eating practices to minimize future risk of CVDs.
Keywords: Eating habits, Exercise, Adolescent, Urban, Kolkata Background Exponential increase in the global burden of noncommunicable diseases (NCDs) marked the beginning of the twenty-first century. Population aging, improved survival opportunities and rapid rise in obesogenic environment were the potential contributors. Cardiovascular diseases (CVDs), globally the most common cause of NCD-related deaths [1], attributed for about 17.5 million deaths in 2012 [2]. About 75% of these CVD deaths were from low-andmiddle-income countries [2].
Compared to Western populations, in South Asian countries, higher prevalence and a decade earlier onset of CVDs were experienced owing to unique genetic predisposition and earlier exposure to risk factors [3,4]. Ischemic heart disease (IHD) continued to be the most common cause of mortality among working adults (15-69 years) in Asian countries [5]. Among all major ethnic groups in South Asia, Indians were found to be at highest risk for CVDs, especially premature coronary heart diseases [4,6,7]. Interplay of unhealthy diet (added high sugar and refined grains), vitamin-D deficiency, tobacco use and physical inactivity contributed to this elevated risk for CVDs among Indians, especially in urban areas. In addition, the rising epidemic of type-2 diabetes mellitus further increased the vulnerability of Asian Indians to IHD [8,9].
Thus a clear understanding of the potential predictors of CVDs appeared to be crucial for appropriate designing and timely (before the initiation of atherosclerosis) implementation of preventive interventions to control the rising tide of CVDs [10,11]. Clustering of cardiovascular risk factors and initiation (appearance of fatty streaks) of atherosclerotic CVDs start in the second decade of life and get influenced by genetic and environmental exposures (serum lipid concentrations/smoking/obesity/hyperglycemia) during lifetime [12][13][14]. Furthermore, sustained high blood pressure was observed to accelerate atherosclerosis in the third decade of life [12,[15][16][17]. Rapid urbanization probably exposed individuals to these life-threatening yet modifiable/reversible risk factors quite early in life [11,18]. For example, worldwide, among school-aged children, 10% were overweight [19] and >3% children and adolescents were hypertensive [20]. In addition, inadequate knowledge regarding CVDs coupled with low-risk perceptions among adolescents further heightened their susceptibility for CVDs [21].
Alike Western countries, South Asian children were also at high risk of developing CVDs in their future life, mostly because of deleterious life-styles and behaviors [22]. Among urban Indians, exposure to multiple risk factors of CVD was evidenced during adolescence and dramatically increased by 30-39 years of age [23]. Among Indian school children, high prevalence of overweight (14.4%), obesity (2.8%), sustained high blood pressure [24,25], coupled with maternal and fetal under-nutrition were suspected to increase the future risk of CVDs [13].
Quality data regarding exercise and eating habits along with related knowledge, perceptions and consequent practices related to the risk of future CVDs in this target population were limited in India. The aim of this study was thus to assess the CVD-related knowledge, health perceptions (especially about future CVD-risk), eating habits, exercise patterns and their interplay among urban adolescent students in Kolkata, a metropolitan city in eastern India.

Methods
Design A cross-sectional study was conducted among adolescent school students of Kolkata between January and December 2014. Students of 9th grade (aged 14-16 years) were selected as the study population group, as a proxy for the adolescent school-children of metropolitan Kolkata.

Sample size and sampling strategy
Cluster random sampling strategy was used for the study using schools as the clusters. The rate of homogeneity (roh) between clusters was assumed to be equal to the intra-cluster correlations owing to the single-stage sampling method [26,27]. Considering the possibility that characteristics relevant to this study (socio-demographics, CVD related knowledge and perception, exercise patterns and eating habits) were likely to be more correlated among the students within the same schools, we empirically assumed roh = 0.2, a relatively higher value, as per standard recommendations [26,28].
Based on the information collected from all the governing bodies of the operational educational boards in Kolkata, the average number of students studying in the 9th grade per school was found to be approximately 75. Hence assuming the average cluster size (b) to be 75, using the formula: D = 1 + (b-1)roh, the design effect, D was calculated to be = 13.8 [26][27][28].
Appropriate population parameter was not available from the study area. Hence following the most conservative approach to ensure the recruitment of adequate subjects for determining the estimates with 95% confidence interval having a standard error of 0.05 (s), we empirically used 0.5 as the expected proportion (p).
Using this design effect and other assumptions as mentioned before, according to the appropriate formula [required sample size, n = p(1-p)D/ s 2 ], 1580 students were required to be recruited from 21 schools [no. of clusters (schools), c = p(1-p)D/s 2 b = 21.067]. Assuming 10% non-response we planned to invite 1755 students and their guardians to participate in our study.

Selection of schools
Initially, an exhaustive list of 426 secondary-level schools in Kolkata metropolitan area was prepared. The schools were stratified according to the socio-economic strata (higher/middle/low) they generally catered as well as the type of students enrolled (co-educational/boys only/girls only). Ensuring the recruitment of at least two schools per strata, using stratified random sampling based on school types and socio-economic status (SES) with probability proportion to size, 21 schools were selected from the list. With administrative support from the Department of School Education, Government of West Bengal, all these schools were invited. Time, date and venue of data collection were finalized ensuring maximum attendance, after completion of necessary formalities in 19 schools who agreed to participate. All students of the selected schools and their parents (preferably mothers) were invited through written letters to participate in the study.

Study population
All 9th grade students present on the day of data collection in the selected schools, were recruited for the study if accompanied by one of their guardians (preferably mothers), agreed to participate by providing written voluntary assents and their accompanying guardians or their legally authorized representative provided written informed consents. Students with any medical or psychiatric illnesses preventing normal communications were excluded from the study.

Data collection and measures
Information was collected through a self-administered, structured questionnaire, which was pre-tested for internal consistency and internally validated [29] in a sample of 160 students (approximately 10% of the total sample size) of same grade, recruited from two randomly selected schools of the study area. Collected socio-demographic information included: age, gender and SES of the students (based on family income).
CVD-related knowledge was assessed in the following five domains: CVDs in general (heard of heart attacks/ causes/symptoms), high blood pressure (cause/symptoms), risk factors for CVDs (smoking/obesity/low cholesterol/ raised sugar/stress/high fiber diet/positive family history/ physical activity), prevention of CVDs and healthy nutritional habits. Based on standard textbooks and guidelines, correct and incorrect responses to individual knowledgeassessment questions were scored as 1 and 0 respectively. To estimate domain-specific and overall knowledge related to CVDs, the corresponding and overall scores were respectively summed up, log-transformed and categorized based on tertile distributions (lower = poor/middle = average/upper = good level of knowledge).
Self-perceptions regarding body-weight (normal/underweight/overweight), future risk for CVDs (yes/no) and overall health (poor/average/good) were also assessed.
To understand the exercise habit, type of activity and corresponding duration were enquired. Based on guideline of World Health Organization (WHO) for adolescent health [30], moderate to vigorous activity for at least 60 min/day on average was considered as adequate exercise.
On the other hand, to elicit the eating habit and related behavior, frequency of major meals/day (1-2/3 />3 times), frequency of snacking (≤3/4/>4 times), history of skipping meals (never/sometimes/often) and history of eating outside home (never/sometimes/often) were recorded. Average frequency of food intake for five times (including meals and snacks) was regarded as approximately appropriate as per the standard recommendations [31,32]. Overall eating habit was determined by scoring the relatively poorer eating habits (captured based on inappropriate frequency of intake, skipping meals and eating fast foods) [31][32][33][34], in descending order and then after logtransformation of the scores, categorization as per the tertile distribution similarly as knowledge.

Statistical analysis
Descriptive analyses of the study variables were performed to determine the mean (for age) and proportions (for categorical variables) with corresponding 95% confidence intervals (CI). Bivariate and multiple (adjusting for age, gender, SES and family history) regression analyses were performed next to measure the associations [odds ratio (OR) for bivariate and adjusted odds ratios (AOR) for multiple logistic regressions] of sociodemographic characteristics/knowledge/perceptions with exercise habit and dietary practices as well as between knowledge and perceptions regarding CVDs. Multinomial logistic regression was used when outcome variables having more than two categories. SAS version 9.3.2 was used for all statistical analyses.

Results
Of 1755 invited students, 1652 did participate in our study (response rate of 94.1%). The mean age was 14.2 years. Proportion of females was slightly higher than males. Nearly two-third of the participants belonged to middle class families. Positive family history of CVDs was reported by about one-fifth of the students. While the majority had poor overall knowledge regarding CVDs and their preventions, a small proportion was found to have good knowledge. Domain-wise, the majority of the students had poor knowledge about CVDs, their risk factors, prevention, high blood pressure and related abnormalities as well as healthy nutritional habits. More than half of the students perceived their body weight as normal while less than a quarter perceived themselves as overweight. Less than a third of the interviewed subjects were engaged in regular moderate-to-vigorous exercise and adequate (at least 1 h) daily physically activity. Regarding dietary history, frequent snacking (four times or more/day) and eating street-foods (in the last week) were quite common (reported by more than half and one third respectively). On the other hand, consumption of more than three major meals/day and frequent skipping of meals were reported by more than 10%. Based on their dietary history, more than half of participants were found to have overall poor eating-habits (Table 1).
Better knowledge (compared to poor, average or good knowledge) regarding CVDs in general, its risk factors, prevention and healthy eating among these urban adolescent school-goers was associated with having the perception regarding future risk of CVDs [for good knowledge respective Adjusted Odds Ratio,

Discussion
Current distribution reveals that persons belonging to 10-24 years age-group (1.8 billion) constitute the largest component of the global population and 1.5 billion of them hail from resource-poor settings [25,35]. Adoption of healthy behaviors during adolescence is expected to serve as the foundation for ensuring longer life expectancy for this population group. Adolescence is considered as the period of vulnerability as well as the optimum opportunity to modify health-related risk behaviors [35]. On the contrary, young adults are always considered to be healthy and global health planners mostly neglect their health needs. Although some standardized health indicators are available for young in Western countries, such indicators are almost nonexistent in developing world. Moreover, adolescents' risk perceptions in relation to health-related behaviors appears to be crucial in determining long-term health consequences [36][37][38][39][40][41]. Thus, educating adolescents regarding negative impacts of risktaking and encouraging them to take responsibility of their own health seemed crucial in controlling adolescent health situation in countries like India.

Knowledge regarding CVDs and their prevention
In this relatively large cross-sectional study, involving adolescent school-students of a metro city (Kolkata) in India, overall knowledge regarding CVDs and their prevention was observed to be poor among participants. The level of knowledge varied considerably across domains. About 60% students had average/good knowledge regarding CVDs in general. Similar findings were also reported from other states in India [42,43] In our study about 23% of participants had good knowledge regarding risk factors for CVDs. Knowledge regarding lifestyle risk factors of NCDs was observed to be even lower among students in Kerala [44], as well as in Michigan [21] and Arizona [45] in US but relatively higher in Nepal [46]. A significant proportion of students in Pune reported obesity, physical inactivity and smoking as predictors of CVDs but could not identify other potential risk factors like serum cholesterol and hypertension [42]. Innovative and friendly educational intervention strategies should be developed for youth so that the gaps in knowledge about CVDs could be addressed successfully.

Perceptions
Understanding the risk-associated thoughts and response was found to be fundamental in shaping perceptions to promote healthy behaviors. Slovic suggested that unfamiliar risks (new/unknown) were mostly avoided by people whereas familiar risks if perceived as controllable or selfchosen culminated into reduction of importance [47]. In addition, body image perception of adolescents and resultant dissatisfaction might negatively affect adolescent health behaviors [48,49]. In our study, only 23.7% perceived themselves as overweight and they demonstrated relatively healthy eating habits. Care should also be taken to ensure that youth perception of overweight does not culminate into depression and harmful weight control practices. About 82% of the participating adolescents did not perceive themselves to be at risk for future CVDs and even those who perceived the risk showed poor dietary practices. Similar findings were also reported from Nepal and Michigan (US) [21,46]. One of the probable explanations might be that adolescents considered CVDs to be old age related problem and underestimated their future risks. We also observed that subjects with average/good knowledge regarding CVDs were more likely   (compared to those with poor knowledge) to perceive themselves at risk for future CVDs. Promotion of school-based cardiovascular health programs might be crucial in dispelling myths and misconceptions with eventual prevention of early onset atherosclerotic changes in arterial walls.

Positive family history
Consistent with previous studies [43], little above onefifth participants had positive family history of CVDs while in Kerala about 2.9% had such history [44]. Early screening of these at-risk children and appropriate interventions might be effective in prevention of early onset CVDs.

Physical activity
Less than 30% subjects reported to get engage in moderate-to-vigorous exercises regularly which was lower than that reported among school-aged children in Delhi [43] and in Gujarat [50] but higher than in Kerala [44]. Inadequate physical activity was also reported from Ontario, US [51] and in Uttarakhand, Maharashtra, Madhya Pradesh and Andhra Pradesh in India [52]. Consistent with previous research in Brazil [53] and Taiwan [35], male students were more likely to be physically active in our study. Previous studies suggested a probable decline in physical activity during transition from adolescence into adulthood, especially between 15 and 23 years [54,55]. One of the probable reasons might be that as adolescents get older, they lack self-motivation and become less compliant to healthy advices. On the other hand we observed that students with good/average knowledge regarding CVDs were somewhat physically active. Based on this observation, it seemed reasonable to hypothesize that adolescents who were well informed about negative health effects of physical inactivity were possibly more motivated to be physically active.

Eating habits
More than half of the participants of the current study demonstrated poor eating-behaviors. Older subjects, males and those who were economically better-placed, took more than appropriate number of meals (major meals & snacks). Similar observations were also reported among adolescents from Arab [56], Europe [57], China [58] and other states in India: Baroda [31], Mysore [30], Uttarakhand, Maharashtra, Kerala, Madhya Pradesh and Andhra Pradesh [52]. There were huge gaps between knowledge about CVDs and eating behavior among study participants. Subjects with good knowledge were likely to take higher number of meals and snacks per day. This observation might be explained by possible influence of frequency and pattern of eating in family. Friendly interactions between adolescents and their parents/teachers/nutritionists at regular interval might be effective in shaping their eating habits. Although higher physical inactivity has often been linked with increased risk of morbidity and mortality, an estimated 60% of population in the world do not perform recommended exercise regularly. As per WHO estimates, 80% of premature heart diseases and 80% of diabetes could be prevented by interplay of healthy diet, physical activity and tobacco avoidance [35]. Moreover, analysis of Northern Finland birth cohort data revealed that risk of obesity was significantly lower in adolescents who ate five meals/day [32] while intake of inappropriate number of meals/day was associated with future development of obesity [59]. Thus, early identification of these unattended risk factors through screening, timely interventions and raising awareness about healthy lifestyle during adolescence might be more effective in controlling CVD epidemics in India. Furthermore, lifestyle modifications at younger age might be more successful than  changing acquired harmful habits in adults. Thus, a concerted public health effort towards modifying food and physical activity environments in schools and in communities seemed to be the need of the hour.

Study limitations
There were some major limitations in the present study. Alike any other observational study associations observed here, should not be interpreted as causal owing to the potentials for residual confounding and other systematic errors. Because of the cross-sectional design, potentials for temporal ambiguity should also be kept in mind. Nonresponse might have affected the representativeness of the study population. Thus, efforts for extrapolation of results beyond the study sample to all adolescent urban students in the study area need some caution. But we still believe that the observations were quite generalizable, owing to the robust sampling strategy and good response rate. Selfreported history of dietary pattern and exercise habits were also likely to suffer from some social desirability bias and issues of recall, although we considered them to be non-differential. Moreover, to minimize the chances of information bias, we urged the subjects to recall for only a short period of 1 week. Despite these limitations, by virtue of large sample size, robust methodology and advanced statistical analyses, we believe that the results of this research will be useful in understanding the scenario pertaining to CVD related knowledge, perceptions regarding related risk, exercise habit, dietary practices and interplays thereof among adolescents of Kolkata, India.

Conclusions
Eating and exercise habits were found to be quite poor among large proportion of adolescent school-students of Kolkata. Lack of adequate knowledge and poor perception about CVD, related risk and importance of their prevention were potential contributors. Also, there existed a large gap between CVD-related knowledge and eating and exercise related practice. As such, our results suggest that there is a critical need for not only the improvement of knowledge and awareness regarding CVD and its related risks, but also actions that will help to bridge the knowledge-practice gap for eating and exercise habits. Given childhood origin of CVDs and possibility of reversing/slowing these atherosclerotic changes through early life style modifications, youth friendly, multicomponent cardiovascular health promotion programs are urgently required to raise awareness and appropriate translation of the awareness in to healthy practices in this target population.