Overweight and Underweight Prevalence Trends in Children from Romania - Pooled Analysis of Cross-Sectional Studies between 2006 and 2015

In adults, overweight and obesity represent a major health burden worldwide due to associated comorbidities such as cardiovascular disease, type 2 diabetes, stroke, cancer and osteoarthritis as well as due to social and psychological problems. Almost one of two adult people in Europe is obese [1]. Excess body weight in children poses a high risk for adult obesity. Research by Guo et al. [2,3] indicated that being overweight at age 35 years may well be predicted from the age of 13 years. Therefore, the prevention of obesity onset should include preventative policies and measures against childhood obesity, especially when considering that weight loss has proven to be extremely difficult [4,5]. Moreover, childhood overweight may be associated with immediate negative health outcomes such as high cholesterol [6], high blood pressure [7,8], prediabetes [9], osteoarticular problems, sleep apnea, and psychological problems such as stigmatization and poor self-esteem [10].

Epidemiologic data provide the rationale for addressing the issue of childhood obesity prevention. In spite of increasing concerns related to childhood obesity, a systematic review comparing estimates of the prevalence of overweight and obesity in school-aged children in 34 nations found many countries lacking high-quality data [11]. There is no study to provide national representative data on obesity in children from Romania. However, several studies on childhood overweight and obesity analyzing representative cohorts at subnational levels have been published in recent years [12,13,14,15,16,17,18,19,20]. Additionally, very few international reviews or collaborative studies have included Romania in their analyses. A review by Cattaneo et al. [21] included data of pre-schoolers from Romania that had been collected in 2004, reporting a prevalence of overweight plus obesity at age 4 equal to 11.8%. Another study indicated that Romanian children 6-11 years old had the highest prevalence (31.4%) of overweight (including obesity) when compared to 7 other European countries [22]. This study used mother-reported heights and weights of children (n = 1,004) [22]. The World Health Organization (WHO) European Childhood Obesity Surveillance Initiative (COSI) collection round in 2013 reported an overweight prevalence of 26.8% (11.6% for obesity) in 8-year-old children from Romania, using the WHO reference [23], which was among the highest percentages reported in the COSI international samples [24]. Moreover, some underdeveloped or developing countries face a ‘double burden' of under- and overnutrition in children and adolescents [25,26].

Undernutrition has not been investigated in a representative cohort from Romania. This pooled analysis of cross-sectional data aims to present, for the first time, national prevalence, trends, and other factors associated with body weight status (overweight, obesity, and underweight) in Romanian children and adolescents for the last decade (2006-2015).

Data sources used, were representative for a population of children from one or multiple Romanian counties and included measurements of height and weight in a school or kindergarten setting. Hospital-based cohort studies were not included. Self-reported data on height and weight were not used because these are subject to significant biases that are influenced by geographic area, age, sex, and socioeconomic characteristics such as education and ethnicity [27,28].

Multiple methods for identifying and accessing data were used. MEDLINE and Google Scholar searches identified published studies that measured the variables of interest in Romanian public school and kindergarten children. The search included the combination of the following terms (in Romanian and English): obesity, overweight, body fat, body weight, or body mass index and children or school and Romania. Thus, eight data sources were identified. The authors of these studies were contacted and asked to provide the raw, anonymized individual record data according to a common protocol. Five out of the eight identified sources provided the requested data. Additionally, the authors were asked if they had knowledge of other unpublished data sources. With this approach, another five data sources with suitable study design were identified of which four also provided the data. Before calculating means and prevalence, participants younger than 6 years and older than 19 years were excluded. Participants with BMI considered implausible were excluded (below -6 SD or above +6 SD). Information was provided about the study population, period of measurement, sampling approach, and measurement. Data sources used are listed in table 1 (11 cohorts from 9 studies).

For the studies included in this article, random sampling methods for each county were applied to obtain selected schools and kindergartens where the measurements were performed. The sampled schools and kindergartens included in this study were assessed between 2006 and 2015. All included studies were cross-sectional. Two studies included data from the same counties at multiple time points (Timis and Bucharest), obtained several years apart. However, the included schools and kindergartens were different; therefore, the likelihood of measuring the same children at different time points was very small. All anthropometric data were collected by physicians, school nurses, or medical students. All examiners were trained in taking accurate measurements according to international measurement techniques. Each child underwent anthropometric assessments that were carried out during one-day sessions. Height and body mass was obtained using calibrated scales. The subjects wore normal, light, indoor clothing without shoes. BMI was calculated as weight (kg) divided by squared height (m²).

To enable comparison with other studies we used three different criteria panels to classify BMI in children.

The WHO 2007 BMI reference criteria for age and gender in children 5-19 years [29] classify children's BMI using standard deviations (SD) as follows: overweight: >1 SD (equivalent to BMI 25 kg/m² at 19 years); obese: >2 SD (equivalent to BMI 30 kg/m² at 19 years); severely obese: >3 SD; thin: <-2 SD; severely thin: <-3 SD.

Considering the International Obesity Task Force revised references (IOTF, Cole 2012) [30], the BMI cut-offs at 18 years were: thinness <17 kg/m² (includes thinness grades 2 and 3); overweight ≥25 kg/m², obese ≥30 kg/m²; morbidly obese ≥35 kg/m²[30]. For this study, thinness grade 1 (defined by the cut-off BMI < 18.5 kg/m² at age 18 years), was included in the ‘normal' category, in order to enable appropriate comparisons with the other references used.

The Centers for Disease Control and Prevention (CDC) reference (2000) for children and adolescents aged 2-20 years, were used to classify children's BMI as: ‘overweight' if BMI was between the 85th and 95th percentiles and ‘obese' if BMI was above the 95th percentile, while ‘underweight' was considered if BMI was <5th percentile for specific age and gender [31].

In this study, the term ‘underweight' encompassed thinness and severe thinness together, while ‘excess weight' summarized overweight, obese and severely obese children altogether.

The children were classified according to their age as being prepubertal (6-9 years old), in early puberty (10-12 years old), in middle puberty (13-14 years old), in late puberty (15-16 years old) and postpubertal (17-19 years old).

In order to assess a possible socioeconomic impact on children's BMI status, we used the Human Development Index (HDI) yearly values reported for Romania by Human Development Report Office for the United Nations Development Program [32]. HDI incorporates the level of socioeconomic development of a country by combining three indicators: income per capita, literacy rate, and life expectancy.

Two methods were used for obtaining informed consent from the participants in the selected studies. The first method used by all data sources, except for the one in Timis (2011) [15], consisted of each participating child being required to obtain written, informed consent from his or her legal guardian. The second method used a ‘passive consent' or ‘opt out' method, where parents were fully informed about all study procedures through a school information meeting [15]. Parents who refused the participation communicated this to the school principal. In all studies, prior to enrolment, the children also gave their assent to participate. All data sources came from studies that were approved by the Institutional Ethical Advisory Board of the respective research organization or University, in accordance with the Declaration of Helsinki.

Descriptive statistics are outlined for all variables assessed as mean and standard deviation (SD) for continuous variables and as percentages for categorical variables. The Student's t test was used for evaluating differences between groups of continuous variables. Pearson's chi-square test was used to evaluate differences between groups for categorized variables. Multinominal multivariate logistic regression was used to examine the combined effects of year of assessment, gender, environment, pubertal age, and county as predictors of underweight and excess weight. Statistical significance was reported at p < 0.05. Data were analyzed with SPSS 20 (SPSS Inc, Chicago, IL, USA). To analyze the BMI classification for children, we used an application tool provided by the WHO - a macro for SPSS that has been described elsewhere [33]. For classifying BMI according to the CDC criteria, we used the commercial software EZ BMI Calculator [34]. For classifying BMI using the IOTF revised cut-offs, a script was created (in Phyton programming language) using the cut-off values provided by Cole et al. [30]. This script is available at https://github.com/dragulescubogdan/COFFS.

Complete data were received from 25,768 participants in the selected studies, of which 394 individuals were younger than 6 years, 251 individuals were older than 19 years, and 95 children were considered to have improbable BMI (above +6 SD or below -6 SD), when using the WHO reference criteria. Thus, 25,060 children from 8 Romanian counties were included in the analysis, of which 48.7% were males and 51.3% were females. Most of the participating children were from the urban environment (80.4%), and only 19.6% were from rural areas. The numbers, means ± SD for age (years) and BMI SDS values using WHO reference are presented in table 2 for all participating children and separately for males and females.

Overall, in the last decade, in Romanian 6- to 19-year-old children , the prevalence of underweight status was 5% (WHO), 4.5% (IOTF), or 8.5% (CDC) (table 3), while the prevalence of overweight status (including obesity) was 28.3% (WHO), 23% (IOTF), or 23.2% (CDC). The prevalence of underweight, overweight and obese participants using age and sex stratification is also presented in table 3.

Children included in this analysis were recruited from eight Romanian counties covering all four macro-regions in the country, including Bucharest (2.1 million inhabitants, the capital of Romania), and the other two counties with high population density and higher economic development as well as less populated and less developed counties. Prevalence of excess weight (WHO) in children aged 6 to 19 years is presented in figure 1 for each participating county.

No significant trend for overweight prevalence was observed over the years of the last decade (chi-square test p = 0.609 for all years). However, the prevalence of excess weight (WHO) peaked in 2009 (30.1%), 2012 (29.8%) and 2015 (33.2%) The lowest prevalence was observed in 2010 (24.5%) and 2013 (24.6%) (fig. 2). The dips of excess weight prevalence coincided with a decrease in HDI in the same year. However, the data are cross sectional and measured in different counties and in different years; thus the variation could be attributed to regional characteristics. Data sources collected in the same county several years apart were available for Bucharest city and Timis county, as previously mentioned. In Bucharest city, the prevalence of excess weight was 37.3% in 2006 and 31.9% in 2011 (p < 0.001), while in Timis County it was 31% in 2011 and 37.7% in 2014 (p = 0.027). We found a weak but statistically significant Spearman correlation between HDI and BMI SDs (r = 0.028, p < 0.001).

Conversely, prevalence of underweight was higher in 2006 (8.9%), and 2013 (13.3%); and lower in 2008 (2.7%) and 2014 (3.3%).

Prevalence of excess weight children (WHO) was higher in urban (29.5% = 18.3% overweight + 11.2% obese; n = 20,137) versus rural environment (22.9% = 14% overweight + 9.9% obese; n = 4,923) (p < 0.001). A significant difference was maintained in both males and females when stratifying by sex. 33.2% of males from urban environment had excess weight and 4.5% were underweight, compared to 23.5% and 8.6%, respectively, in the rural environment (p < 0.001). Of the females from urban environment, 26.1% had excess weight, while 4% were underweight, whereas in the rural environment the values amounted to 22.4% or 7.1% (p < 0.001).

The prevalence of underweight children (WHO thin + severely thin) remained relatively stable throughout all ages (average 5%), ranging from 3.1% in postpubertal youths to 5.7% in middle puberty (p < 0.001). In contrast, the percentage of children with excess weight (overweight + obese+ severely obese) changed dramatically over different ages. On average, 28.3% of children, aged 6-19 years, presented excess weight. Highest percentages of excess weight were observed in prepubertal children (34.2%) and early puberty (33.8%), while lowest percentages were observed in late puberty (18.1%) and postpubertal youths (13.6%) (p < 0.001) (see supplemental fig. S1; available at http://content.karger.com/ProdukteDB/produkte.asp?doi=444173). A significant difference between pubertal ages is maintained when controlling for gender (chi-square test p < 0.001), as shown below.

On average, 25.4% of females aged 6-19 years had excess weight, and 4.6% were underweight (WHO). Highest percentages of excess weight were observed in prepubertal females (33.5%) and early puberty (30.5%), while lowest percentages were observed in late puberty (12.4%) and postpubertal females (10.0%) (see supplemental fig. S2; available at http://content.karger.com/ProdukteDB/produkte.asp?doi=444173)

In males aged 6-19 years, 31.2% had excess weight, and 5.3% were underweight (WHO). Highest percentages of excess weight were observed in prepubertal males (34.8%), and in early puberty (37.1%), while the lowest prevalence was observed in postpubertal males (18.4%) (see supplemental fig. S3; available at http://content.karger.com/ProdukteDB/produkte.asp?doi=444173).

Bivariate associations between assessment year, birth year, pubertal age, gender, environment or county and underweight or excess weight (WHO) as main effects were tested, then significant associations were included in a multinominal multivariate logistic regression model (table 4). Birth year was not included in the model as it was not significantly associated with underweight or excess weight. The assessment year was negatively correlated with underweight (OR = 0.74, 95% CI = 0.68-0.79), and with overweight - including obesity (OR = 0.96; 95% CI = 0.93-0.99), when using year 2006 as reference. Male gender associated with both increased risk for being underweight (OR = 1.28, 95% CI = 1.13-1.44) and overweight (OR = 1.37, 95% CI = 1.29-1.45), when compared to females. Prepubertal age increased the risk for being underweight (OR = 1.50, 95% CI = 1.17-1.94), and even more for the risk of being overweight (OR = 3.86, 95% CI = 3.41-4.36), when compared to postpubertal age. The urban environment decreased the risk for being underweight (OR = 0.77, 95% CI = 0.62-0.94), while it increased the risk for being overweight (OR = 1.12, 95% CI = 1.01-1.26), when compared to rural environment. The region (county) had a significant but relatively minor effect on the risk for underweight and excess weight in several counties, as shown in table 4.

For the first time, to our knowledge, this study presented epidemiologic data on excess weight and underweight in the last decade (2006-2015) with national representation for Romanian children aged 6-19 years. A large number of children (25,060) from 8 Romanian counties were included in the analysis. Prevalence for excess weight and underweight was determined using the WHO, the IOTF and the CDC reference criteria sets to enable comparison with other studies. As such, these different criteria used to classify BMI generated different prevalence rates. IOTF criteria provided lower rates of overweight and obesity, when compared to WHO and CDC criteria, and these findings align with similar differences as already published [24,30,35,36].

Excess weight in 6- to 19-year-old Romanian children was found in 28.3% (WHO reference) [29]; 23% (IOTF cut-offs) [30] and 23.2% (CDC reference) [31]. A comparison between different references is subject of several other studies, but this is beyond the scope of this paper. The prevalence of excess weight in Romanian children aged 6-9 years was on average 35% (WHO reference) in this study. The value is similar to those from countries such as Slovenia, Portugal or Spain, as assessed by the COSI study (WHO reference) [24]. The World Health Organization (COSI) included 2,800 children from Romania that were measured in 2013. The reported prevalence of overweight was 26.8% (11.6% for obesity) in 8-year-old children from Romania using the WHO reference [23]. In our study, we found a prevalence of 34.7% for overweight (16.2% for obesity) in 8-year-olds (n = 2,688). The difference could be explained by the fact that our sample included more children form the urban environment, while the COSI report indicated rates of 31.4% for overweight children in urban environment, as compared to 21.6% in rural. Additionally our data represents an average across the last decade. The prevalence for excess weight reported by the ENERGY project in Romanian children at the age 10-12 years was 34% and thus similar to that in Hungary, Slovenia, and Spain (WHO reference) [37]. Although the prevalence of overweight children was slightly lower in Romania as compared to other countries such as Italy or Greece [24,37], these findings are worrisome and provide the rationale for devising and implementing further prevention strategies.

In the last 20 years, Romania went through important political and economic transition stages and sociodemographic changes that might have influenced nutritional patterns [38]. However, a significant trend over the years in the last decade (2006-2015) was not observed in our study. Two peaks were noticed for 2009 and 2015, while two dips were found for 2010 and 2013 that coincided with a decrease in HDI. Only a very weak positive correlation between BMI SDS and HDI was found. A detailed analysis in the two counties (Bucharest and Timis) where measurements were performed several years apart, showed a peak in 2006-2009 and a dip in 2014-2015. The relation between trends of overweight prevalence and HDI as a marker of socioeconomic status must be interpreted with caution, as individual data related to socioeconomic status of the family were not available for this study.

In Romania, boys were 1.4 times more likely to have excess weight when compared to girls. The COSI study 2013 collection round in Romania also indicated that 8-year-old boys had a higher prevalence of overweight and obesity (29.8%) when compared to girls (23.7%) [23]. A higher prevalence of overweight in boys was also found in several other European countries such as Slovenia, Spain, Italy, Greece, or Hungary [37]. The same study indicated that potential risk behaviors for excess weight are generally more likely in children of parents with fewer years of formal education. Thus, interventions aiming to prevent overweight and obesity in school-aged children should pay specific attention to boys and children of low-educated parents [37].

Excess weight was more prevalent in the urban than in the rural environment (OR = 1.12; 95% CI =1.01-1.26). Similar findings were reported by the COSI study in Romania [23] and by studies in East Slovakia and Poland [39,40], but not by others in Greece [41] or the USA [42]. This variation may be attributed to differences related to education, physical activity, sedentary behavior, economical family status, or even self-perception [41,42].

The different regions in Romania included in the pooled analysis had a minor impact on excess weight prevalence. This finding may reflect the different economic status throughout the counties; however, more studies are needed to identify the causes of this outcome.

The percentage of children with excess weight changed dramatically in children of different ages. The highest percentages (average 34%) of excess weight were observed in 6- to 12-year-olds (prepubertal and in early puberty), while the prevalence decreased to less than half in 15- to 19-year-olds (late puberty and postpubertal youths). The logistic regression indicated that prepubertal and early pubertal children were 3.9 times more likely to have excess weight as compared to postpubertal youths. The variation was maintained after stratifying by gender, albeit with lower values for females as compared to males. Other studies have also found higher prevalences in pre- and early adolescence [43,44]. Weight gain before puberty promotes earlier initiation of the maturation process mediated by luteinizing hormone concentrations in girls [45]. Therefore, there is a urgent need for early interventions in pre-adolescent years, in both genders, in Romanian children.

The prevalence of children aged 6-19 years that were underweight (below -2 BMI SDS using the WHO reference) was low in Romania, with an average of 5%. The logistic regression showed a slightly negative trend from 2006 to 2015. In detail, males had a higher prevalence of underweight as compared to females (5.3% vs. 4.6%). Both males and females from rural environment had higher prevalence of underweight when compared to urban participants. The relatively low levels of underweight children in Romania could be explained by the capitalist economy, which is promoting a consumption-driven behavior, inducing widespread overnutrition. In Romania, prepubertal and early pubertal children were 1.5 times more likely to be underweight, as compared to postpubertal participants, indicating that the peripubertal age needs to be under healthcare surveillance for both risks of under- and overnutrition.

This study has several limitations. The study included more children from urban compared to rural environment. The study did not assess, nor adjusted, for factors such as dietary factors, birth weight, breastfeeding, food security, maternal and childcare, health services, and other environmental factors that could influence growth. Puberty has major implication upon growth and development, but in our study, Tanner stage was not assessed due to methodological and ethical issues. However, some insights are provided by assessing the data set in accordance to age groups known to be associated to a certain pubertal stage. Socioeconomic information about children's families was not available.

Overall, according to the WHO reference, in Romania almost one in four children aged 6-19 years was overweight or obese between 2006 and 2015. This prevalence remained relatively stable over the years during the last decade. Prepubertal age, urban environment, and male gender were variables associated with increased risk for excess weight in Romanian children. The prevalence of underweight in Romania was low, affecting more females, preadolescent children, and those from rural environment. This study indicated that a significant part of children in Romania are either underweight or overweight. The prevalence of overweight and obese children is at levels comparable with those from other European countries, which includes Romania among the countries in which childhood obesity is at epidemic levels. Therefore, this study reinforces the need for a comprehensive approach in the fight against childhood obesity and for the implementation of public health policies for the prevention of the obesity epidemic.

We thank Bogdan Dragulescu from Politehnica University of Timișoara for rendering the vector map of Romania and designing a script (Phyton programming language) that uses the revised IOTF cut-offs to calculate prevalence of Thinness, Overweight and Obesity in Children. We thank Dr. Mihai Niculescu (The University of North Carolina at Chapel Hill) for critical appraisal of the manuscript and proofreading. We also acknowledge Silvia Mandasescu (Roman Municipal Hospital), Lecturer Cornel Vălean and Prof. Mircea Nanulescu (Clinical Emergency Hospital for Children Cluj) for their involvement in data collection.

Adela Chirita-Emandi was funded through a project for young researchers: Internal Competition University of Medicine and Pharmacy ‘Victor Babes' Timisoara, Program II-C4-TC-2016. Pascanu M. Ionela and Raluca Pop were supported by an internal research grant from University of Medicine and Pharmacy Targu-Mures, Romania (Nr. 17802/1/22.12.2015). Maria Puiu was funded under operational program: Development of Existing Infrastructure and Creation of New Infrastructure (Laboratories, Research Centers) POSCCE-A2-O2.2.1-2013-1, in the Center of Genomic Medicine of the University of Medicine and Pharmacy ‘Victor Babes' Timisoara. These funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. All other authors received no specific funding for this work.

Adela Chirita-Emandi performed part of initial data collection, designed the study plan, coordinated and supervised data collection from all data sources, performed the statistical analysis, drafted the initial manuscript, revised the manuscript, and approved the final manuscript as submitted.

Carmen Gabriela Barbu, Eliza Cinteza, Bianca Ioana Chesaru, Michaela Dobre, Simona Vasilica Fica, Mihai Gafencu, Gabriela Edita Ichim, Veronica Mocanu, Ionela M. Pascanu, Raluca Pop, Simona Alexandra Tatar, each contributed, in part, to initial data collection, reviewed and revised the manuscript, and approved the final manuscript as submitted.

Maria Puiu conceptualized the study, performed part of initial data collection, supervised data collection, critically reviewed the manuscript, and approved the final manuscript as submitted.

All authors have no conflict of interest to declare.