Study design and participants
The current study was conducted within the Tehran Lipid and Glucose Study (TLGS) framework, a population-based cohort study to determine the risk factors for non-communicable diseases (NCDs) among an urban population of Tehran. The study’s main objectives included: 1) Determining the prevalence of NCDs and their associated risk factors through a cross-sectional phase (1999–2001) followed by five follow-up examinations at 3-year intervals (2002–2004, 2005–2007, 2008–2010, 2011–2013, 2014–2016); 2) Investigating the effects of a healthy lifestyle intervention on NCD risk factors and outcomes.
The multistage cluster random sampling method was used to select the target sample. Of 20 health care centers located in district 13 of Tehran, three centers with complete information about the families covered were selected. In the first phase of the TLGS, participants were randomly selected from the population covered by the mentioned health care centers. As a field trial, in the second phase of the TLGS, participants under the coverage of one of the health care centers were considered an intervention group. Of the total 7151 families (intervention area: 2237, control areas: 2056 and 2858), 4751 families (intervention area: 1816, control areas: 1261 and 1674) were randomly selected to participate in the study. Thus, in total, 15,005 individuals (aged ≥ three years old) from randomly selected families participated in the baseline measurement of the TLGS. Of those, 5630 participants received the lifestyle intervention. The assignment of participants to the intervention and control groups was based on their area of residency. Far from the other two, one of the health care centers was selected as the intervention center, and the individuals residing in the area under its coverage received the intervention. Those living in areas covered by two other health care centers were selected as the control group and received routine and nationally approved health care [21, 22].
In the current analysis, data on 2374 adolescents (57.2% girls), aged 12–18 years, who participated in the baseline assessment of the TLGS were considered and monitored for over a median follow-up of 17.0 years. Of those recruited for the study, 1123 participants were excluded for the following reasons: lack of residential address at the time of data gathering (294, 12.4%), those who were displaced between study groups (753, 31.7%), and lost to follow up in all examinations (76, 3%). Final analysis has been conducted on 1251 eligible adolescents assigned to two groups control (932, 74.5%) and intervention (319, 25.5%). Further details of the current sampling design are illustrated in Fig. 1.
The outcome variable
In the current study, energy expenditure, defined as the calories an individual will burn during LTPA, was considered a continuous dependent variable from adolescence to adulthood. The LTPA data was collected using the validated Iranian version of the Modifiable Activity Questionnaire (MAQ) in all follow-up examinations [23]. In this regard, the adolescents’ version of the questionnaire [23] was used for participants aged 12–18 years, and an adult version [24] was used for those aged > 18 years old during follow-up. Both versions include 15 Iranian popular and common activities specified for the mentioned age groups during leisure time and time spent in each activity. To calculate the energy expenditure in LTPA, we multiplied the total weekly exercise minutes dedicated to each activity (Time spent in LTPA) by its typical intensity expressed in metabolic equivalents (METs) and individual’s weight to obtain METs.Kg- min/wk. [25]. Time spent in LTPA was estimated by the number of months a year and time per week that every activity was performed, considering possible seasonal variation.
Metabolic equivalent task (MET) is a physiological measure expressing the energy cost of physical activity. It is defined as the ratio of metabolic rate (or energy consumption) during a specific physical activity to a reference metabolic rate, set by convention to 3.5 ml O2 kg-1•min-1. In many studies, MET-value is defined as the energy expenditure of a particular physical activity divided by per kg body weight and hour and calculated by 1 MET = 1 kcal / (Kg*h). In other words, regardless of duration, MET-values are assigned to a particular activity to describe its intensity [26].
Moreover, baseline physical activity in the TLGS was assessed using the Lipid Research Clinics (LRC) questionnaire, which evaluated this behavior based on three sub-scales: regular, strenuous, and self-rated physical activity. Results have been classified as high (at least three times a week), moderate (less than three times a week), and low (none in the past week) [27].
Study measurements
Participants’ BMI [weight (kg)/height (m2)] was calculated based on weight and height values in each follow-up examination. Weight was measured, with subjects minimally clothed without shoes, using digital scales (Seca 707: range 0–150 kg) and recorded to the nearest 1 kg. Using a tape meter, height was measured in a standing position without shoes, while shoulders were in normal alignment. To determine body weight status in children aged ≤19 years, WHO percentiles for body mass index (BMI)-for-age and cutoffs for bodyweight status were used [28], and BMI between 25 to 30 kg/m2 and BMI ≥ 30 kg/m2 were defined as overweight and obesity cutoff points in adults.
Level of education was categorized based on the participants’ total years of education; 0–5 years as “primary”, 6–12 years as “secondary”, and over 12 years as “higher” (including university courses). Employment status was categorized depending on whether the participant has a job or not.
Multi-setting intervention
Face-to-face counseling, written materials, public health education, and the promotion of health policies to facilitate healthy lifestyles among the intervention group were strategies used at the family, school, and community levels. The TLGS scientific committee prepared the intervention content in each context delivered under the supervision of the public health care center located in the intervention area.
Schools
Twelve schools in the intervention area farthest from the control area were selected as “Health Promoting Schools”. These schools implemented the school-based subprogram, which focused on improving children’s physical activity and healthy eating as well as reducing smoking. The scientific committee trained the principals and volunteer teachers regarding a healthy lifestyle, taking students’ age and needs into account, at the beginning of each year. All students in the first year of both guidance (12–15 years) and high schools (16–18 years) participated in nine 45-minute classes. Volunteer students formed a “school health society” under the supervision of their teachers and transferred health messages to their peers. In addition, family involvement was an essential part of the school-based program. Parents were educated to form a supportive and healthy environment at home. Recurrent parent-teacher meetings and annual seminars were held to maintain families’ engagement. Also, pamphlets/booklets with healthy lifestyle-related content were distributed between families. The number and duration of educational sessions and the number of participants (students and parents) were registered based on the study protocol. Almost 70% of the planned school-based interventions were successfully implemented.
Families
Parents were involved in forming a supportive and healthy home environment. The family involvement component aimed to introduce parents to the school-based lifestyle modification program and assist them in creating a supportive environment to improve healthy behaviors in adolescents, including LTPA. Through ordinary meetings of the parent-teachers in each school, interactive forums were conducted to inform parents regarding alarming rates of NCD risk factors in Iran, the necessity of lifestyle modification, and relevant practical recommendations.
In addition, family-based interventions were delivered by inviting families for group sessions (for > 2 hours, 10–12 individuals in each session) between baseline and the first follow-up examination. Families were also involved in the lifestyle modification program by “Health liaisons”, who were volunteers responsible for delivering the program under the supervision of the public health care center in the intervention area. All families in the intervention area received pamphlets/booklets containing information and benefits of healthy lifestyle behaviors (physical activity, food pyramid, smoking consequences, cessation tips, and coping with stress). They also received a seasonal newsletter named “Courier of Health,” containing affordable recommendations regarding healthy lifestyle behaviors. To sensitize and motivate families regarding lifestyle change, some results of the TLGS were reported in the mentioned health newsletter (mainly about the prevalence and the effect of risk factors in their area). Telephone surveys showed that 50% of households had received and paid attention to educational pamphlets and health newsletters.
Community
Community-based components of the TLGS intervention engaged various sectors, i.e., municipality, police, media, and community and religious leaders. Two to four times a year, on religious occasions such as Ramadan and special days such as World No Tobacco Day or World Diabetes Day, ceremonies were held at the local mosque or amphitheater. In these ceremonies, health messages were conveyed to the public through lectures on healthy living. In addition, city billboards in the intervention area were used for advertising a healthy lifestyle. More than 80% of the households participated in at least one public gathering between every two examinations.
Statistical analysis
Mean ± SD and frequency (percent) were reported as data descriptions for continuous and categorical variables, respectively. The Chi-square test and independent samples T-test were conducted for group comparisons. Generalized Estimating Equations (GEE) as a robust statistical method was used to evaluate educational intervention’s effect on leisure-time physical activity in adolescents during 15.9 years of follow-up. The GEE procedure considered the “identity” link function and “autoregressive” working correlation matrix. The main effects of follow-up times and intervention and their interaction terms on physical activity were examined. GEE models were adjusted for age, education, and occupation at each follow-up and BMI at baseline. In the present study, parental characteristics that significantly differed in the intervention and control groups were explicitly adjusted in each age- and sex-specific group. Hence, parental ages and paternal occupation for younger girls and maternal education for older boys were considered adjusting factors Regarding parental characteristics, those factors that were significantly different between the control and intervention groups were adjusted considering adolescents’ sex and age. Parental ages and paternal occupation for younger girls and maternal education for older boys were considered adjusting factors. Adolescents’ and parental characteristics were compared between responders and non-respondents participants. All analyzes were performed based on gender and age (early and late adolescents). IBM SPSS Statistics 23 was used for statistical analysis.