The association of the age, period, and birth cohort with 15-year changes in body mass index and waist circumference in adults: Tehran lipid and glucose study (TLGS)

Obesity is highly prevalent in developed and developing countries [1‐3]. In the recent decades, Iran, as a developing country, has been experiencing an alarming rise in the prevalence of obesity; over an 8-year period, the overall prevalence of obesity in the country increased from 13.6% in 1999 to 22.3% in 2007 (OR = 1.08 per year) [4]. Increasing trends in general and abdominal obesity have been reported in both genders in Iran’s capital, Tehran. This trend has contributed to an overall 36 and 34% increased risk of general and abdominal obesity from 1999 to 2011, respectively [5].

Various patterns in trends of obesity can be explainable by many sociodemographic, socioeconomic, and lifestyle factors. To explore these patterns, it is necessary to assess overtime changes in measures of general obesity (e.g., body mass index [BMI]) and abdominal obesity (e.g., waist circumference [WC]) in individuals (i.e., the age effect); and based on population-wide behavioral changes and other age-independent exposures (i.e., the period effect); as well as based on intra-individual differences in the people born in different times (i.e., the birth-cohort effect) [6]. Rising trends in BMI and WC with aging, as observed in cross-sectional studies, could be the result of the age and birth-cohort effects. On the other hand, increases in BMI and WC with aging, as reported in longitudinal studies, might be due to period effects. The simultaneous assessment of the age, period, and birth-cohort effects on BMI and WC require longitudinal analyses with repeated measures in the same individuals [7‐9].

Over the last few decades, Iran has experienced an epidemiological transition from a traditional to a western lifestyle [10]. There are no recent studies evaluating age and/or cohort effects on BMI and WC based on repeated measurements in Iran. Here, we investigated the association of age, period, and birth-cohort with BMI and WC changes in the framework of a prospective cohort study within a 15-year follow up period.

The mean age of study participants at baseline was 41.4 ± 13.0 years, and 58.7% of the participants were women. After 15 years of follow-up, mean BMI in men increased from 26.0 ± 3.9 to 27.5 ± 4.3 kg/m² (P _trend < 0.001), and this trend was accompanied by an increase in mean WC from 88.8 ± 10.9 to 97.8 ± 10.4 cm (P _trend < 0.001). Women also showed elevations in mean BMI (from 27.5 ± 4.8 to 29.9 ± 5.4 kg/m²) and mean WC (from 87.3 ± 12.4 to 95.8 ± 12.1 cm) in this period (P _trend < 0.001). More detailed characteristics of the study population across different phases of the study are available in Table 1. In comparison with those who completed the follow-up and entered the study (the respondent group, n = 4895), non-respondents (n = 5133) were older and had lower BMIs (supplementary Table 1).

Cohort-specific mean BMI and WC values in the study population are presented in Table 2. Rising trends in mean BMI and WC in each age cohort were observed for both genders (except for the BMI of ≥70 years old in both genders and WC of ≥70 years old women). Additionally, sex-specific mixed effects were estimated on BMI and WC changes in the studied population, adjusted for age, age squared, time, time*age, smoking, physical activity, education, and marital status (Tables 3 and 4). In Figs. 1 and 2, adjusted estimated means of BMI and WC in six age cohorts are shown according to the mean age of the group in each phase of the study, in men and women.

In this 15-year population-based longitudinal study, gender-specific changes of BMI and WC in the adult population of Tehran were evaluated. The rises in BMI and WC were strongly associated with aging throughout the follow up period. Increases in BMI and WC were more prominent in younger birth cohorts in men, and older birth cohorts in women.

A large body of evidence suggests that weight, BMI, and WC increase throughout most of adult life [14‐17], which is in line with our findings regarding the positive association of aging with obesity measures, observed in both genders. A prospective longitudinal study in Norway recently showed that mean weight in middle-aged men and women (aged 26–54 years) increased significantly during a 10-year follow-up while it reduced in elderly men and women (aged 60–69 years); however, mean WC increased throughout the follow up in all age groups [18]. The recent findings were completely in accordance with those of the present study. We observed an overall positive association of aging with BMI in both genders which in combination with the observed inverse association of BMI with age squared suggested an inverted U-shaped relationship between age and weight gain. This implies that the rapid weight gain seen in young and middle-aged men and women will most likely be reversed in the elderly. The weight reduction with aging reported in the elderly could be a result of decreased appetite and calorie intake, as well as substantial loss of muscle mass due to the aging process [19]. Apart from age-related muscle loss, disease-related muscle loss could also explain this phenomenon, as many chronic diseases (with particularly higher prevalence in older individuals) could also accelerate decrease of muscle mass, such as diabetes, hypogonadism in men, growth hormone deficiency, hyperthyroidism, and hypercortisolism [20]. However, in older men and women, despite a drop in BMI, WC increased over years, which could be a result of body composition changes. Age has been shown to affect body composition with older people having higher proportions of the adipose tissue to the lean mass [21]. Furthermore, age affects the distribution of the adipose tissue, with more intra-abdominal and visceral adipose tissues vs. peripheral subcutaneous fat mass being observed in the elderly [22]. Besides, spinal degenerative changes and a consequent reduction in height during the aging process could explain the increase of WC despite decreased or sustained weight in this age group [23].

In this study, a significant negative association of birth cohort with BMI and WC was observed in adult men, indicating notable general and central weight gain in the youngest compared with older birth cohorts. This finding was consistent with the observations of previous studies [7, 9, 24‐30]. Iran, as a country located in the Middle East and North Africa region, is subjected to a dramatic epidemiological transition from a traditional to modern nutrition and lifestyle [10]. This modernization and associated lifestyle changes by encouraging sedentary habits and excess calorie intake have increased the risk of obesity and overweight in the new generations of Iranians, and as a consequence, the prevalence of obesity among children and adolescents has increased [31‐33]. Overweight or obese children and adolescents are likely to remain so in adulthood, and this may explain why people who were born in the most recent birth cohorts were more susceptible to BMI and WC elevations than older individuals born in the years of economic and social instability. Moreover, in older men, this negative association of birth cohort could be partially explained by reduced calorie intake and aging-associated loss of appetite [19]. In addition, exposure to stress might lead to unhealthy behaviors which could contribute to weight gain as well [34]. Another explanation for the observed association of weight gain with younger birth cohorts could be more job commitments of younger generations and their less time for physical exercises and self-care, delivering them more prone to obesity. In contrast, the individuals born in earlier birth cohorts spent their youth in more active lifestyles and were less exposed to cumulative obesogenic health behaviors [35, 36]. Another explanation for the negative impact of birth cohort on men could be related to the survival bias which might be a consequence of premature mortality among older obese individuals [37]. In contrast to our findings, a study on Chinese adults found that being overweight was more common among middle-aged men born between 1950 and 1975 and revealed that the prevalence of overweight among Chinese adults decreased rapidly in the most recent birth cohorts (i.e., after 1975) [38]. The observed difference between the two studies may be due to economic and social differences between the two populations. China has become one of the world’s fastest growing economies since the 1980s and with increased access to health information and the implantation of new health policies, has somehow managed to overcome the epidemiological transition [39].

In women, however, we observed a positive association of birth cohort with BMI. Regarding the difference between men and women in this area, both biological and social factors may be accountable for justifying the reported disparity between the two genders. The younger women born in recent birth cohorts are more likely to have higher education, have fewer pregnancies, and adhere to feminine beauty standards paying a premium for slimness as a desirable body form [40]. Moreover, menopausal changes affect women in their later years, making elderly women more prone to obesity [41, 42]. Another factor contributing to the observed gender disparity may be a higher health awareness level among young women rather than men [43].

We are aware that our research may have some limitations. First, since our subjects were only urban residents in Tehran, our findings cannot be extrapolated to the entire country’s population. Second, in order to reach solid conclusions about the association of the “time period” with BMI and WC, more measurements and longer follow-up periods are needed. Third, the body composition of the participants was not recorded in our study; therefore, we could not speculate to what degree the reported weight fluctuations might correspond to body fat percentage changes. Moreover, we did not consider the role of the participants’ dietary intake and nutritional habits on their BMI and WC. However, the current study has several strengths worth mentioning. Anthropometric measurements were performed by trained staff instead of relying on self-reports, increasing the accuracy of measurements. Moreover, our research included participants from a wide adult age spectrum and birth cohorts, who were tracked for 15 years and underwent three measurement intervals.

In conclusion, this study demonstrated rising age-dependent trends in the BMI and WC of Tehranian adult men and women. However, there were substantial gender-specific associations with birth cohorts, indicating that the men born in recent birth cohorts and the women born in earlier birth cohorts had the most alarming trends in BMI and WC. Given Iran’s epidemiological transition state, it seems that the nation has a heterogeneous tendency for weight gain, and our findings suggest that effective population-based obesity preventive measures should be targeted towards gender, age, and birth cohort, as the most potent determinants of obesity.