The association of sleep problem, dietary habits and physical activity with weight status of adolescents in Nepal

A cross-sectional study was conducted between July and November 2019, among adolescents aged 11–19 years in Kathmandu a Metropolitan City and capital of Nepal. Kathmandu Metropolitan City is a residence to 20% of Nepal’s urban population from a diverse ethnic and socio-cultural background and has a density of 13,225 population per square kilometers [38]. The city was selected purposively as the study site due to its cosmopolitan nature, which can better represent the Nepalese urban adolescents’ population.

The sample size was calculated using the single population proportion formula: N = Z²pq/d², considering 95% confidence interval (CI), 50% assumed proportion (p) [39], and 5% margin of error (d). Assuming a 10% non-response rate and 1.5 design effect, the sample size was estimated to be 635.

A multistage cluster random sampling method was used to select the study sample. In the first stage, all 32 local administrative units (locally known as a “Ward”) of Kathmandu Metropolitan City were considered as clusters for this study. There were a total of 91 public and 638 private schools within the Kathmandu Metropolitan City, with approximately a total of 165,000 adolescents; more than 90% of adolescents pursuing their education in private schools within the Kathmandu Metropolitan City [40]. Considering the cluster size, the number of students per class, geographical coverage, and time frame of the study, two clusters out of 32 clusters were randomly selected for the study sites. In the second stage, the total number of schools within the selected clusters were categorized into two separate categories of public and private schools. Then, from the two randomly selected clusters, an equal number of schools i.e. four private schools and four public schools from each category were again randomly selected for this study. Finally, within each randomly selected eight schools (four private schools and four public schools), two-classes per year group (from grade 8–10) were randomly selected. Subsequently, all adolescents’ (10–19 years) students from the selected classes were invited to fill up the questionnaires. Adolescents with serious health problems, including physical or psychological difficulties, or those who were absent on the days of data collection were excluded from the study.

Self-administrated questionnaires were distributed to a total of 635 eligible adolescents. The instructions about questionnaires were explained to the adolescents jointly by core study team members and health teachers of the respective schools before the adolescents filled up the questionnaires. Also, we collected the student’s information on the socio-demographic part from their parents by sending a questionnaire along with the consent form. The socio-demographic questionnaires completed by parents were collected at the respective schools of their children within the 4 days of the questionnaire distribution. Anthropometric measurements of children were taken at their respective schools. All tools were originally developed in English and then translated into Nepali language and back-translated to the English language to ensure that the originality of the questionnaire remains unchanged. Also, the tools were pretested among 50 adolescents from a school that was not included in the analytical samples. The data collectors were undergraduate nurses (BSc Nursing), and they were provided a four-day training. The training included an explanation of the objective of the study, data collection procedure, sampling method, ethical principles and guidelines, and data entry techniques. Approximately 20 min was required for adolescents to fill out the questionnaires.

The collected data were entered into EpiData software 3.1v and transferred into StataMP version 14.1 (StataCorp LP, College Station, Texas) for statistical analyses. WHO Anthro Plus Software V.1.0.4 was used for BMI scores. The descriptive results are presented in the form of mean, standard deviation, frequency, and percentage. Both histogram and Kolmogorov-Smirnov normality test (p < 0.05) were applied to check the normal distribution of the data (z-score outliers). Pearson’s chi-square (χ2) and One-way analysis of variance (ANOVA) tests were applied to observe differences in socio-demographic characteristics, sleep measures, physical activity, and sedentary behaviors of participants by BMI categories. Multinomial logistic regression analyses were applied to observe the effects of the predictors on BMI categories considering “normal BMI” as a reference category. Multicollinearity amongst the predictor variables was checked using the Variance Inflation Factor (VIF). The mean value of VIF < 10 was the cut-off point [52]. The statistical significance was considered at p-value < 0.05 and 95% confidence intervals (CIs).

Four models were created to adjust the association between different predictors such as socio-demographic variables of parents and adolescents, sleep characteristics, dietary habits, and physical activity with the outcome of interest as BMI among adolescents using multinomial logistic regression analyses. Backward stepwise processes were used to identify parsimonious models to explain the predictors of adolescent’s weight status (all p < .05). Model-I was adjusted with the main predictor variables such as adolescents’ sleep characteristics and dietary habits. Model-II was adjusted for adolescents’ sleep characteristics, dietary habits, and socio-demographic characteristics of parents and adolescents. Model-III was adjusted with physical activity and socio-demographic characteristics of parents and adolescents. Model-IV adjusted for significant variables (p < 0.05) from Model- I, Model-II, and Model-III resulting in a parsimonious model in the multivariate analysis.

The ethical approval for this study was obtained from the ethical review board of the Nepal Health Research Council (Ref. no: 2777/2019). Permission was also taken from the participating schools. Written consent was obtained from the legal guardians or parents of eligible children aged 10–19 years. A separate questionnaire with an informed consent form and request letter from respective schools were sent to parents through their children. Written consent was also taken from the adolescents who participated in the study. Adolescents were also informed that they were allowed to leave any questions unanswered in the self-administered questionnaire. Participation was voluntary, and participants’ identity was kept confidential.

The mean (± SD) age of adolescents was 14.4 (± 1.1) years. More than three fourth (77.9%) of the adolescents were aged between 14 and 16 years, attended private schools (75.1%), and followed the Hindu religion (78.6%). More than half were male (54.1%), had a nuclear family type (55.3%). Slightly more adolescents were from grade nine. More parents were from the Brahmin/Chhetri ethnic group (40.6%), and more mothers attended below high school education (58.2%), while approximately half (48.4%) of fathers had attended university-level education. The majority of mothers were unemployed (63.1%), and three fourth of fathers were employed (74.9%), and more (41.4%) of parents had a family monthly income between 15,000–30,000 Nepalese Rupees (130–260 USD) per month (Table 1).

Underweight was most prevalent among 14–16 years (87.7%), males (54.1), private school attendant (54.3%), from Dalit and minority ethnicity (43.8%), and joint family (50.8%), whose mothers had less than high school education (50.8%) and unemployment (63.1%), unemployed fathers (63.1%), and having family monthly income between 15,000–30,000 NRs (33.3%) (Table 1).

Likewise, overweight/obesity was most prevalent among the age group 14–16 years (61.1%), males (51%), belonged to private school (75.1%), belonged to Brahmin/Chhetri ethnicity (44.9%), in a nuclear family (53%), amongst those whose mothers had below high school education (63.1%), in fathers university-level education (46.9%), in unemployment mothers (67.1%), in employed fathers (82.5%), and had family monthly income between 15,000–30,000 NRs (38.9%) (Table 1).

The mean (± SD) sleep duration for weekdays and weekends was 8.9 (± 1.1) and 8 (± 1.1) hours, respectively. About one in four adolescents had sleep problems with a mean (± SD) CASC score of 24.7 (± 4.5), and one in ten had OSA symptoms (Table 2).

Underweight was most prevalent among adolescents without sleep problems (85.9%) and history of OSA symptoms (92.9%). Conversely, overweight/obesity was most prevalent among adolescents who had sleep problems (56.3%) but no history of OSA symptoms (79.8%) (Table 2). The mean score ± SD of bedtime problem (6.6 ± 3.5), breathing and unstable sleep (6.1 ± 3.1), parasomnia and sleep movement (3.2 ± 2.3), and day time problem (3.2 ± 2.1) among overweight/obese adolescents (Table 2).

Only 38.9% of adolescents felt hungry due to not enough food in the home, half (51%) of adolescents never consumed fruits, one fourth (74.3%) of adolescents consumed vegetables more than two times a day, more (53.1%) of adolescents drank soft drinks, and one third (65.2%) of adolescents had consumed high-fat foods less than once a day in past 30 days. About nine in every ten children had consumed junk food, nearly half (46.1%) of adolescents have consumed junk food 1–3 times per week, more than half (53.4%) of adolescents took fast food 2–3 days per week, and more (55.8%) adolescents had consumed more than five food groups (Table 2).

Underweight was most prevalent among those adolescents who rarely/sometimes went hungry in school (50.8%), consumed fruits ≥1 times/day (61.4%), consumed vegetables ≥2 times/day (71.9%), consumed soft drinks < 1 time/day (75.4%), consumed high-fat food ≤1 time/day (66.6%), consumed fast food 2–3 days per week (45.6%), had consumed junk food (77.1%), consumed junk food 1–3 times per week (61.4%), and consumed less than 5 food groups (68.4%). Similarly, overweight/obesity was most prevalent among those adolescents who never went hungry in school (61.1%), consumed fruits ≥1 times/day (57.7%), consumed vegetables ≥2 times/day (73.1%), consumed soft drinks ≥1 time/day (83.22%), consumed high-fat food ≥2 time/day (50.3%), consumed fast food 2–3/days per week (47.6%), consumed junk food (89.2%), consumed junk food ≥4 times per week (49.6%), and consumed ≥5 food groups (53%) (Table 2).

More adolescents (63.1%) were not physically active, and half of the adolescents (51.8%) didn’t walk or ride a bicycle in the past 7 days. The majority of the adolescents (56.3%) didn’t get any physical education/training during the past school year, and more adolescents (42.5%) spent 3–4 h per day in sedentary behaviors (Table 2).

Underweight was most prevalent among those adolescents who weren’t physically active (84.2%), didn’t walk or ride a bicycle (63.1%) in the past 7 days, who didn’t get any physical education during this school year (73.6%), and who had spent 3–4 h/day in sedentary behavior (40.5%). Likewise, overweight/obesity was most prevalent among those adolescents who weren’t physically active (66.4%), didn’t walk or ride a bicycle as their primary means of transport to and from school or around the home (41.6%) in the past 7 days, who didn’t get any physical education during this school year (68.4%), and who had spent 3–4 h/day in sedentary behavior (34.9%) (Table 2).

In this study, Table 3 shows relative risk ratios of adjusted multinomial logistic regression models for underweight and overweight/obese relative to normal-weight adolescents.

After mutually adjusted all the confounding variables including socio-demographic, physical activity, dietary habits, and sleep characteristic, several factors showed a significantly higher risk of being overweight/obese: adolescents who had sleep problem compared to no sleep problem (RRR = 13.37, 95% CI: 7.14–25.05), adolescents who had OSA symptoms compared to no OSA symptoms (RRR = 3.21, 95% CI:1.31–7.86), who consumed soft drink ≥1 time/day compared to < 1 time/day in past 1 months (RRR = 5.44, 95% CI: 2.93–10.10), consumed high-fat food ≥2 times/day compared to ≤1 time/day (RRR = 2.17, 95% CI: 1.18–3.99), and adolescents who had consumed junk food compared to never consumed junk food (RRR = 5.71, 95% CI:2.55–12.82). Parents who belonged to Terai/Madhesi castes had more than two times (RRR = 2.81, 95% CI: 1.19–6.64) increased risk of overweight/obese children compared to Brahmin/Chhetri castes. Adolescents whose fathers were employed also had more than two times (RRR = 2.04, 95% CI: 1.04–3.98) significantly higher risk of being overweight/obese. Adolescents who used bicycle or walked 1–3 days/week as their primary means of transport to and from school or around the home (RRR = 4.33, 95% CI: 2.01–9.28), and ≥ 4 days/week (RRR = 1.86, 95% CI: 1.01–3.42) as their primary means of transport to and from school or around the home were significantly higher risk of being overweight/obesity compared to those who didn’t. Adolescents who had spent 5–6 h/day in sedentary behavior were 3.21 times (RRR = 3.21, 95% CI: 1.14–9.09) increased risk of being overweight/obese than those who had spent < 1 h/day in sedentary behavior. In contrast, adolescents aged 14–16 years were 71% lower (RRR = 0.29, 95% CI: 0.16–0.52) and adolescents who had consumed less than five food groups had 45% lower (RRR = 0.55, 95% CI: 0.31–0.97) risk of being overweight/obesity compared to 12–14 years age groups and who had consumed more than five food groups respectively.

The relative risk of underweight was found increased in adolescents who rarely/sometimes went hungry (RRR = 2.56, 95% CI: 1.27–5.16) compared to never went hungry, and Dalit and minority castes (RRR = 5.61, 95% CI: 2.12–14.82) compared to Brahmin/Chhetri castes. In contrast, those adolescents who consumed soft drink had 73% (RRR = 0.27, 95% CI: 0.12–0.61) than those who consumed < 1 time/day, physically active had 66% (RRR = 0.34, 95% CI: 0.14–0.80) compared to physically inactive, received 1–3 days physical education/training during this school year had 75% (RRR = 0.25, 95% CI: 0.08–0.80) compared to no physical education/training, and those who spent 3–4 h/ day in sedentary behavior had 43% (RRR = 0.57, 95% CI: 0.11–0.80) compared to those who spent < 1 h/day in sedentary behavior respectively were at lower risk of being underweight (Table 3).

We found nearly one in four adolescents (24.4%) had sleep problems and similar findings have been reported from several studies conducted in both developed and developing countries across the globe that showed sleep disorders as a common problem, particularly among the teenage population [53‐56]. In this study, adolescents’ sleep duration during weekdays and weekends, OSA symptoms, and sleep problems measured by CASC were significantly different among different groups for the BMI for the age of the adolescents. Also, adolescents who had sleep problems were found more likely to be overweight/obese. These findings are similar to the evidence from several previous studies conducted in different geographical regions of the world, where they found increased risks of overweight/obesity among young age groups of participants are those having sleep problems or short sleep duration [25, 57]. Several factors are responsible for the association of sleep problems and overweight/obesity, including both biological and habitual. The biological phenomenon can be explained with the evidence that sleep deprivation is associated with dysregulation of neuroendocrine control of appetites such as alterations in the hormones, including cortisol, leptin, and ghrelin [58, 59]. These hormonal changes are further responsible for alterations in demand for body calorie requirements. As a result, individuals adopt unhealthy food choices, consume additional foods, and decrease their physical activities.

Likewise, about one in every ten adolescents were found to have OSA symptoms, and those who had OSA symptoms were found to be at higher risks of overweight/obesity compared to those who did not have OSA symptoms. There is ample of evidence that suggests vice-versa associations of OSA and overweight/obesity among children and adolescents. The study conducted among Australian adolescents showed that with each standard deviation increase in BMI z-score, there is more than three times more risk of developing OSA, which supports the findings of our study [60]. Mitchell and Kelly [61] study claim that one in two obese children had OSA in the United States of America (USA). Severe upper ways obstruction in obese children can lead to sleep disorders further resulting in causing obesity and risk factors for cardiovascular disease, stroke, metabolic disease, and day-time sleepiness [62].

Due to “nutrition transition” in most of the low-income countries the food consumption patterns has shifted from locally available staple foods consumption towards the consumption of western diet pattern containing poor nutrients, foods high in fat, energy-dense foods, sugar-sweetened beverages, junk foods, fast foods and less consumption of vegetables and fruits [63]. In this study, adolescents who consumed soft drinks ≥ one time per day and high-fat food ≥ two times per day in the last 30 days and who consumed junk food were found to have higher risks of developing overweight/obesity. A recent review study shows that Asian pre-school children adhering to the westernized eating pattern, which is described as a high intake of sugary beverages, snacks, and red meat was found to have significantly higher odds of being obese [64]. Likewise, Brazilian children adhering to junk food diets [64], Chinese adolescents who followed the Western dietary patterns [64], Indian and Saudi adolescents who practiced adverse eating behaviors including consumption of high-fat foods and carbonate drinks were also found significantly at higher risk of being obese [65‐67]. However, the review by Togo et al. [68] presents mixed findings from 30 different studies where few studies showed negative or no association of high fatty, sweets, or energy-dense food consumption habits with overweight/obesity. The difference in the pattern of results might be due to the complexity of determining factors for overweight/obesity [31] and also, in fact, eating pattern alone is not only responsible for producing overweight/obesity.

Our study also revealed that adolescents who had consumed < five food groups within the last 24 h preceding to survey were found protective towards overweight/obesity as compared to those who consumed ≥ five food groups. These findings are supported by the results of the study conducted in the USA, where diversified food group consumption was positively associated with body fatness [69]. However, the systematic review by Otto et al. [70] shows that there is inconsistency in the results regarding the influence of dietary diversity on overweight/obesity [70].

On the other hand, only a few food-related factors were associated with an increased risk of underweight. The relative risk of underweight was found increased in adolescents who rarely/sometimes went hungry. Similar findings are reported by Johnson et al. [71], where they reported that going hungry most of the time was marginally associated with an increased risk of underweight [71]. Those adolescents who consumed soft drinks had a 73% lower risk of underweight than those who consumed <one time/day. This finding is in line with the previous study, where those adolescents who drank soft drinks had a 17% lower risk of underweight [63]. Food containing high sugar-sweetened beverages is linked to increased body weight [63]. In this study, adolescents who had consumed less than five food groups had a 77% lower risk of being underweight. Household food insecurity and lack of access to diversified food households were likely to suffer malnutrition because of very limited intake of nourishing food resulting in underweight adolescents. The decline in their fewer food groups was primarily due to lower consumption of meat and poultry, fish and shrimp, eggs, and dairy products [72].

Our study revealed that very low percentages of adolescents were found to be involved in physical activity and more than half of the adolescents were found to have more than 3 h of sedentary lifestyle per day. These results were found congruent with the findings from similar studies conducted in the past [48, 73]. However, these results are explicitly indicating that some of these modifiable risk factors linked to obesity presented among Nepalese adolescents, if not changed at the early stages of life, may continue to adulthood and exacerbate the burden of chronic illness in the future. Also, our study presents those who spend 5–6 h of sedentary lifestyle per day were more than three times higher risk of overweight/obesity compared to those who spend less than one-hour sedentary lifestyle per day. On the other hand, those who attained physical education or training ≥4 days/week were less likely to be overweight/obese when compared to those who do not attain any physical education or training in the recent week in this school year. It is more likely that those who were involved in insufficient physical activity and spent less sedentary time had a positive influence on maintaining normal BMI [74]. It was also found that those who were involved in walking or riding a bicycle ≥4 days/week were slightly more at risk of overweight/obesity, which sounds quite unexpected. This may be due to the reason that adolescents who were overweight/obese might have recently engaged in walking or riding a bicycle as a form of physical activity to reduce their weight. Another reason may be due to certain limitations such as behavioral findings were assessed using a self-administered questionnaire. Thus, social desirability bias leading to over and under-reporting may occur, and tools used in physical activity were taken only for the past 1 week.

In the current study, adolescents belonging to Terai/Madhesi castes were found more likely to be overweight/obese. There is a paucity of evidence demonstrating the association between ethnicity and overweight/obesity among adolescents; as a result, we couldn’t justify it anymore in this study. These findings recommend prospective researchers to explore the causal relationship between ethnic variations and BMI within the Nepalese context. However, the possible reason could be that the elite group or richest people in Nepal comes from Madhesi, Newar, Thakali, and Thakuri ethnic groups, alongside Bhramin and Chhetri. Those richest people who belonged to Terai/Madhesi caste were migrated from Terai to Kathmandu and are privileged to study in high-quality English-medium schools. Moreover, the richest peoples have access to energy-dense, fast foods in their family, and luxurious lifestyle and able to high daily expenses resulting in overweight/obese children. In contrast, adolescents who belonged to Dalit and minority caste were found to be underweight. Dalit castes, one of the most marginalized groups in Nepal, are still facing several forms of discrimination in all sectors, including the workplace. As a result, Dalits today stand at the bottom of most indicators of socio-economic development [75]. Peoples from low-income families lack access to resources to purchase healthy and sufficient foods for their families, resulting in their children being underweight [9]. Despite the formulation of several plans, policies, and legislation to abolish caste-based discrimination in Nepal, the disadvantaged ethnic groups are facing different forms of discrimination [76]. To uplift, the socio-economic status of such disadvantaged ethnic groups warrant effective implementation of existing governmental plans, policies, and legislation. Prevention of modifiable risk factors such as poor dietary habits, physical inactivity, sedentary behavior, and sleep deficiency for both under and over-nutrition among adolescents can be achieved through the implementation of nutrition-specific and nutrition-sensitive interventions [77, 78]. Our findings suggest that the father’s employment status was found to have increased the risk of overweight/obese children. This finding is in line with the other study, where they found that father’s employment to be associated with overweight/ obese childhood [79]. The possible reason could be those adolescents whose fathers were in employment have higher socio-economic income sources and access to purchasing more calorie-dense and nutrient-poor foods. Those adolescents who were between aged 14–16 years were less likely to be overweight/obese as compared to those who were between aged 12–13 years. These findings are consistent with Smetanina et al. [80] results, where they reported that overweight/obesity is decreased with an increase in age. The plausible reason could be that adolescents are more likely to increase the physical exercise in school/home, and are self-aware about their health status with an increase in their age.

Our study has some limitations. First, due to the cross-sectional study design of the study, the directional cause and effect relationship between independent and dependent variables cannot be inferred. Generalizing the pattern of dietary habits, sleep characteristics, and physical activity throughout the entire calendar year would be misleading as the data were collected at a specific point of time. Second, the data collected were self-reported and it was retrospective, we could not completely ignore the recall bias. Third, calculating the actual amount of energy expenditure or sedentary behavior was not possible due to the use of self-reported information. Likewise, food consumption in terms of the nutrient amount was not calculated, which could result in overestimating or underestimating the actual extent of dietary habits influences on the BMI of adolescents. Fourth, since the participants were recruited from the schools located in urban settings, the results of this study cannot be generalized among adolescents living in rural areas of the country. Despite the mentioned limitations, our study has some strengths. This study is the first of its kind to explore the association of sleep characteristics with BMI of adolescents in the Nepalese context. Therefore, this study fills the gap of evidence regarding how sleep characteristics with dietary habits and physical activity influence the BMI status of Nepalese adolescents. Further, this study has attempted to bring the attention of researchers for the future in-depth study for the explicit understanding of the multifactor and the complexity of the overweight/obesity mechanism among Nepalese adolescents, which is one of the major current public health concerns in the country.