OUP user menu

Burden of overweight in Germany: prevalence differences between former East and West German children

Angela D. Liese, Thomas Hirsch, Erika von Mutius, Stephan K. Weiland
DOI: http://dx.doi.org/10.1093/eurpub/ckl052 526-531 First published online: 3 May 2006


Background: Given the increasing prevalence of childhood overweight, we aimed to quantify the population burden and evaluate potential regional differences in anthropometric characteristics and prevalence of overweight in fourth graders in two German cities. Methods: Data were analysed from a cross-sectional school-based study conducted in 1995–96 in Dresden (former East Germany) and Munich (former West Germany) as part of the International Study of Asthma and Allergies in Childhood. Height and weight of the children were measured, and the parents completed a questionnaire. 2474 children age 9–10 years provided anthropometric data. Overweight was defined based on the age-specific and gender-specific international cut-off values for body mass index. Results: Dresden children were on average 1.2 kg lighter and >1 cm taller than their Munich peers. The prevalence of overweight in Dresden was 15.2% in girls and 14.2% in boys compared with 24 and 22.2%, respectively, in Munich. Differences were observed between Dresden and Munich with respect to the proportion of children of non-German nationality, household smoking, breastfeeding practices, and individual dietary behaviours. Even in combination these factors were not able to explain entirely the between-city overweight differences. Conclusions: A substantial proportion of pre-adolescent children in Germany is now considered overweight with marked regional differences in prevalence. Comparison of population-level factors indicate that other unmeasured determinants of overweight may be responsible for the marked differences in the prevalence of overweight.

  • children
  • diet
  • overweight
  • regional differences

Overweight and obesity are increasing worldwide at an epidemic rate.1 Between 1975 and 1995 the prevalence of obesity more than doubled in former East German girls and increased by more than 60% in East German boys, most of the increase occurring in the latter 10 years.2 This trend continued in the 1990s,3 a time of marked changes in lifestyle and food availability in former East Germany because of the German reunification.

In contrast to other populations, very little data is available on the burden of overweight and obesity in German children, especially in the period of pre-adolescence and adolescence. Two recent publications documenting prevalence and time trends focused exclusively on former East German populations,2,3 however, no comparable data on children in former West Germany are currently available.

The purpose of this study was first to quantify the prevalence of overweight and obesity and document anthropometric characteristics in fourth grade children living in two German cities in 1995–96, Dresden in former East Germany and Munich in former West Germany, based on data collected under a standardized protocol. Subsequently, we explored the contribution of a variety of demographic and behavioural correlates of overweight to the between-city differences observed.


Two cross-sectional surveys were conducted in Germany between September 1995 and December 1996, one in the city of Dresden, state of Saxony in the South of former East Germany, and one in Munich, state of Bavaria in the South of former West Germany, as part of the International Study of Asthma and Allergies in Childhood (ISAAC). The methods of ISAAC have been described in detail elsewhere.4 ISAAC was initiated to gain insights into the aetiology of asthma and allergic disorders through standardized comparisons of diverse populations worldwide,5 thus, our analyses focusing on overweight are ancillary. Community-based random samples of children attending fourth grade were studied using the schools as sampling units. In the first study phase, parental questionnaires were distributed through the schools, self-administered by the parents and returned to study personnel via the children. Topics included respiratory and atopic disorders in the child, current dietary habits, infant nutrition, birth weight, pre-term birth, twin status, and characteristics related to socio-economic status (SES). The parental questionnaire included a brief, non-quantitative food frequency section, which queried the frequency of intake of 16 food groups or foods in the categories ‘never/less than once a week/1–2 times a week/3–6 times a week/once daily or more frequent’. The foods and food groups were cereal, grains or müsli, fresh grain kernels, whole grain bread, other breads, sausages or cold cuts, meat, fish; and the following foods or food groups as consumed in winter including spinach, carrots, cruciferous vegetables, fresh fruit, apple juice, orange juice, multi-vitamin juice, other fruit juices, multi-vitamin supplements.

Subsequently, a team led by a trained paediatrician following a standardized protocol during school hours conducted anthropometric examinations, including measurement of height and weight, and spirometry. Children were asked to remove shoes and heavy overcoats. Height was measured to the nearest 0.5 cm and weight to the nearest 0.1 kg on a stadiometer and balance beam scale. Parents had previously given informed consent separately for these physical examinations.

The basis of our present analyses are 1999 children in Dresden and 2019 children in Munich who were randomly selected from the ISAAC schools and invited to the lung function test and anthropometry. Participation rates were 62.8% in Dresden and 66.0% in Munich. After excluding 114 children outside the 9–10 year age range and one child with an implausible weight measurement, our analysis population of 9- to 10-year old children for anthropometric characteristics and prevalence estimates of overweight and obesity comprised 1190 children in Dresden and 1284 children in Munich (total n = 2474). A comparison of participants with non-participants revealed that participants were slightly more likely to be of German nationality but did not differ with respect to other demographic characteristics.

For the second set of analyses exploring population-level determinants of overweight, we excluded 538 children with incomplete information on food intake and 43 children with missing information on the relevant confounders (breast feeding, SES, smoking in household) resulting in a total of 1893 children, 913 in Dresden and 980 in Munich.

Body mass index (BMI) was calculated as weight/height2 (kg/m2). In order to facilitate both international comparisons and utilization by German national health care providers, we present prevalence of overweight and obesity based on two sets of reference values. First for international comparisons, prevalence estimates of overweight and obesity are based on the age-specific and gender-specific BMI cut-offs for children that have been developed by Cole et al.6,7 and members of the International Obesity Task Force (IOTF) to correspond to an adult BMI of 25 and 30 kg/m2. Second, overweight and obesity are defined as a BMI greater or equal to the age-specific and gender-specific 90th and 97th percentile, respectively, of the German BMI-for-age reference values.8 Thus, the categories for overweight and obese are not mutually exclusive, i.e. obese children are also overweight. The cut points used correspond to the percentiles at ages 9.5 and 10.5 years, respectively, since for epidemiological use of age groups of 1 year width we used the mid-year value, i.e. 9.5 for the 9.0–10.0 age group to get an unbiased estimate of prevalence. The German BMI-for-age reference values were developed based on data on height and weight from 17 studies of 34 422 youth (17 275 girls) in Germany.8 These BMI reference values were calculated according to the LMS-method developed by Cole.9 This method allows via Standard Deviation Scores (SDSLMS) the classification of individual values within the distribution of reference values.8 The results were basically identical to those obtained when applying the 90th and 97th percentile cut-offs generated from our own study population.

Variables ascertained from the parental questionnaire included information on demographic characteristics, early childhood characteristics such as low birth weight, pre-term birth, twin status, infant diet, and also some indicators of current nutritional behaviours. Breast-feeding was defined as ever having been breast-fed. The years of schooling and highest academic degree of the parents were used to derive a variable describing SES. The questionnaire also ascertained the nationality of the child because Germany has a substantial proportion of non-German residents, who have relocated from various countries and cultures, who have lived and worked in Germany for many years but are not German nationals. Under German immigration laws, German nationality can generally be equated to having parents of German ancestry, i.e. being of Caucasian ethnicity.

Estimates of means, standard deviations, and prevalence of the children's anthropometric, demographic and behavioural characteristics are presented by city. Differences between the cities were calculated and tested by two-sample t-test and Chi-square tests.

A subsequent set of analyses focused on explaining the prevalence differences of overweight between Dresden and Munich. Multivariate logistic modelling was used with the goal of identifying the most informative variables that explained between-city differences, i.e. reduced the beta estimate of the city variable in the model. All analyses were conducted using SAS version 8.02.


Table 1 describes anthropometric characteristics and prevalence of overweight and obesity among children aged 9–10 years in Dresden and Munich by gender. Both girls and boys in Dresden were lighter and taller than their Munich peers and consequently had a markedly lower mean BMI and BMI percentiles (P-values <0.005). The prevalence of overweight and obesity was lower in Dresden among both girls and boys (15.2 and 14.2%, international reference data) than in Munich children (24 and 22.2%). Of note, the prevalence of obesity in Dresden girls was particularly low at 1.4% (95% CI 0.5–2.4%) (table 2).

View this table:
Table 1

Anthropometric characteristics of 9- to 10-year old children in Dresden and Munich in 1995/1996 (n = 2474)

DresdenMean (Std)MunichMean (Std)Difference
Girls, 9–10 years(n = 585)(n = 670)
    Weight (kg)35.6 (7.2)36.5 (7.8)−0.80.0495
    Height (cm)143.2 (7.3)141.5 (6.8)+1.7<0.0001
    BMI (kg/m2)17.3 (2.6)18.1 (2.8)−0.8<0.0001
    BMI percentile48.157.4−9.3<0.0001
Boys, 9–10 years(n = 605)(n = 614)
    Weight (kg)35.9 (7.1)36.5 (7.7)−0.60.1210
    Height (cm)143.3 (6.7)142.0 (6.4)+1.30.0005
    BMI (kg/m2)17.4 (2.6)18.0 (2.9)−0.6<0.0001
    BMI percentile48.656.6−7.9<0.0001
View this table:
Table 2

Prevalence of overweight and obesity of 9- to 10-year old children in Dresden and Munich in 1995/1996 (n = 2474)

Prevalence(95% CI)Prevalence(95% CI)%P-value
Girls, 9–10 years
Boys, 9–10 years
  • a: International reference values6

  • b: German reference values9: Overweight ≥90th, obese ≥97th age-and-gender-specific BMI percentile

The prevalence of overweight for girls and boys combined is shown in figure 1. In Dresden, 14.7% of the children were overweight (or obese) compared with 23.1% of the children in Munich, applying the IOTF international cut-off values. The prevalence estimates using the German reference values are provided for comparisons with German national data. Children of German nationality in Munich had a lower prevalence of overweight than non-German children (data not shown). In Dresden the number of non-German children in the sample was too small for meaningful estimates.

Figure 1

East–West comparison of prevalence of overweight in 9- to 10-year old German children. ****P-value of difference in prevalence <0.0001

Children in Dresden and Munich differed with respect to a number of demographic and behavioural characteristics (table 3). Data are presented by city combined for boys and girls since gender differences were negligible. The proportion of children of German nationality (which is roughly equivalent to having parents of German nationality) was significantly higher in Dresden than in Munich (99.8% versus 86.2%). Parents of children in Dresden were on average 3 years younger. Smoking in the household occurred in a smaller proportion of families in Dresden than in Munich while there were no observable differences in the percentage of households with a high SES based on parental education. Children in Dresden were slightly more likely to have ever been breast-fed. Indicators of current nutritional behaviours revealed that on average Dresden children consumed whole grain bread, sausages or cold cuts, and fresh fruit more frequently in winter than children in Munich. Dresden children had a lower intake of multi-vitamin supplements.

View this table:
Table 3

Demographic and behavioural characteristics of 9- to 10-year old children in Dresden and Munich in 1995/1996 (n = 1893)

Dresden (n = 913) PercentMunich (n = 980) PercentDifferenceP-value of difference
German nationality99.886.2+13.6<0.0001
Age of mother at birth24.8 (4.3)a28.1 (5.3)a−3.3<0.0001
High socio-economic statusb53.150.9+2.20.3377
Current smoking in household30.643.5−12.9<0.0001
Mean (Std)% DailyMean (Std)% DailyDifference% Daily
Intake frequency per week
    Whole grain bread2.77 (2.60)20.22.05 (2.22)9.6+0.72 ****+10.6****
    Sausages or cold cut5.04 (2.13)45.23.62 (2.29)20.3+1.42****+24.9****
    Fresh fruit in winter6.00 (1.67)68.85.16 (2.12)48.7+0.84****+20.1****
    Multi-vitamin supplements0.60 (1.56)4.41.04 (1.91)5.7−0.44****−1.3
  • **** P < 0.001

  • a: Mean (Std)

  • b: High socio-economic status: one parent had at least the degree of Abitur or Fachabitur which is approximately equivalent to 13 years of schooling

We subsequently explored the extent to which correlates of overweight could explain the observed prevalence difference in overweight and obesity between Munich and Dresden (crude OR 1.59, 95% CI 1.21–2.07) (table 4). Age, breastfeeding, SES, and smoking individually and in combination reduced the prevalence odds ratio for the Munich versus Dresden difference, but did not eliminate it (see Model 2, table 4). In contrast, the additional inclusion of sausage and cold cut intake as an indicator of a dietary pattern increased the difference to an odds ratio of 1.55 (95% CI 1.15–2.09) (see Model 3, table 4) while none of the other dietary variables were influential in the model. Exploration of additional confounders such as birth weight, birth order, or maternal age had no effect.

View this table:
Table 4

Explaining prevalence differences in overweight among 9- to 10-year old children in Dresden and Munich in 1995/1996 (n = 1893)

VariablesModel 1Model 2Model 3
OR95% CIP-valueOR95% CIP-valueOR95% CIP-value
Smoking in household0.840.72–0.970.01630.850.74–0.990.0315
Socio-economic status0.680.52–0.890.00590.700.53–0.920.011
Sausage/cold cut intake1.111.05–1.180.0008
  • a: Munich versus Dresden

  • b: German versus non-German

View this table:
Table 5

International comparison of the prevalence of overweight in children (IOTF definition)

CountryReferenceTimeAgeBoys (%)Girls (%)
United States151988–949–1125.226.4
  • a: Based on reported values of height and weight


Despite a growing awareness of the epidemic of overweight and obesity in youth among health care providers in Germany, there is a remarkable paucity of systematically collected epidemiological data quantifying the burden of this condition, especially in pre-adolescent and adolescent populations. Our data from fourth grade children residing in Munich, Germany, are unique in that there currently seem no comparable statistics from another West German population. Applying the international cut-off values,6 ∼23% of 9–10 year olds were overweight in 1995–96. In contrast, the prevalence in the formerly East German city of Dresden was markedly lower, 14.7% of children being overweight.

The remarkable regional difference in the prevalence of overweight raises the question whether this difference is in fact real. First, all data were collected under a standardized protocol for both sites with technicians having received centralized training. Given that school attendance is mandatory in Germany, the sampling frame of the study, all schools educating fourth graders with the exception of those serving severely handicapped children, is representative of the underlying population of fourth graders. Second, while the overall participation (questionnaire, anthropometry, lung function testing) of the children was satisfactory at >63%, it does raise the question of generalizibility. A response analysis revealed that participants were more likely to be of German nationality than the non-participants. The participants did not, however, differ significantly with respect to other demographic characteristics (age or gender) from the non-participants. In both cities, our anthropometric data on non-German children indicate that they had on average higher BMI values than their German classmates, a finding recently confirmed in another German population.10 Thus it is highly likely, that at least for Munich, a city with a somewhat larger population of non-German residents, the total population prevalence estimates for overweight and obesity actually represent underestimates. Owing to its history, Dresden has a very small proportion of non-German residents, thus estimates would not be affected as strongly by the lower response of non-German nationals. Furthermore, the anthropometric characteristics of Dresden children in this study are very similar to those reported from Jena, also formerly East Germany, in 1995.2 We conclude that our study most probably yields conservative estimates of the true burden of overweight in two populations of fourth graders and that the differences observed between the cities are most probably underestimates.

To interpret these findings of a substantial difference in the prevalence of overweight and obesity between two populations of children in Germany, it is useful to distinguish between two aims, first, an assessment of the actual population burden of disease, and second, an investigation of the determinants of population rates of a disease. The prevalence estimates of overweight and obesity based on the German reference values are clearly relevant for German public health officials, health care providers, and policy makers as they present an estimate of the burden of disease present in the population.11

International comparisons have been greatly facilitated since the publication of the IOTF BMI cut-off values for children.6 Based on an internationally compiled reference population, these cut-offs for overweight and obesity have been developed to correspond to an adult BMI of 25 and 30 kg/m2, respectively. A comprehensive international review of childhood obesity has been recently published.1 Compared with children of similar age studied in the 1990s1218 (table 5), the prevalence of overweight among Dresden children ranked very low in both girls (15.2%) and boys (14.2%) while Munich children ranked in the upper range of prevalence rates at 24 and 22%. Populations in the US and Canada are clearly suffering from a much more widespread distribution of this condition. In contrast to the British estimates, which were limited to white children living in England or Scotland, our estimates comprised the entire population of school children, regardless of their nationality or ethnic background. Thus while the absolute burden of overweight and obesity in Germany may not have reached the proportions of some other countries, trend analyses in Germany indicate similar increases over time.2 It is also interesting to note that recently, substantially higher prevalence estimates ranging from 19 to 20% were reported by Frye and Heinrich3 from another former East German area located in Saxony–Anhalt.

Between-population comparisons can also be useful for investigations into the determinants of population prevalence or incidence rates.19 These determinants can be conceptualized operating both at the individual and at the environmental level.20 When we explored the individual-level characteristics that might explain the difference in the prevalence of overweight observed between Dresden and Munich, our results indicate that factors related to SES such as parental school degree or smoking in the household and differences in infant feeding practices between the cities explained part but not all the prevalence difference, which suggests that a combination of other, unmeasured, individual-level factors and environmental determinants may be operating. For example, residual confounding by socio-economic factors remains a possibility because it is questionable whether parental education is an appropriate measure of SES in former East Germany. Another clear limitation of our study is the lack of information on the children's physical activity and parental overweight, both of which have been shown to be highly associated with overweight among children.21 While we have no data on physical activity habits in East versus West German children in the mid-1990s, we do know from the National Health Interview and Examination Survey16 conducted on East and West German adults in 1998 that younger adults (i.e. in the age range of the ISAAC children's parents, ∼30–45 years) in former East Germany had lower mean BMIs and a lower prevalence of obesity than their West German peers, similar to the direction of the differences observed among children in our study. Thus unmeasured factors such as parental history, lifestyle, and children's physical activity, or differences in sexual maturation may well explain a further part of the differences observed. Finally, our between-population exploration is limited by the fact that it is based on only two cities, one in each part of former East and West Germany.

At the same time, the importance of environmental determinants should not be overlooked.20,22 Egger and Swinburn22,23 have outlined an ecological framework for understanding obesity. They distinguish specific types of environments, including the physical, economic, and socio-cultural environment,22 all of which can be envisioned as being different between Dresden and Munich in our study. Children in Dresden grew up under very different environmental circumstances for the first 3–4 years of their life. In former East Germany, the availability of typical ‘Western’ foods favoured by children such as soft drinks, chocolate, snacks, and fast food was extremely low, and increased only gradually after reunification in 1990. Access to and attractiveness of sedentary activities such as watching television or playing computer games has increased greatly. It is interesting that 5 years after reunification, prevalence of overweight in Dresden children remained substantially lower than in Munich, despite the wide-ranging changes in lifestyle that occurred. Several studies of former Eastern European populations undergoing enormous political and economic changes in the 1990s have now shed light on the impact of these changes on overweight and obesity.2,3,2426 A study in East Germany has indicated that the prevalence of overweight has increased substantially between 1985 and 1995 compared with the previous decade,2 similar to findings in Czech children.26 In Russia, opposing trends seem to have occurred in children versus adults. While the prevalence of overweight in 6- to 18-year old Russian youth decreased markedly from 15.6% in 1992 to 9% in 1998,25 in adult women, the prevalence of overweight remained remarkably stable between 1992 and 2000, although an increase in obesity was observed.24 Future studies should aim to determine whether, and, if so, when during the lifecourse these profound changes in the living environment result in an increased rate of weight gain, and try to identify unique characteristics of these less ‘obesogenic’ environments.22

Key points

  • This study quantified the burden of overweight in fourth grade children in two German cities and examined regional differences.

  • A substantial proportion of children were overweight.

  • The prevalence of overweight in Dresden (14.7%) was lower than in Munich (23.1%).

  • Individual-level risk factors were not able to explain regional differences in overweight.

  • Public health policy should consider directing prevention efforts both at individual and environmental determinants of overweight.


We are indebted to Dr Katrin Kromeyer-Hauschild, Institute of Human Genetics and Anthropology, Friedrich-Schiller-University Jena, Germany, for generating the BMI percentiles. We thank the students, their parents, the teachers, and secretaries of the schools for their help and co-operation during data collection. The study was funded by the German Ministry of Education and Research.

Contributors list

A.D.L. developed the manuscript idea and statistical analyses and wrote the manuscript. T.H. made significant contributions to the concept for the analyses and revisions of the manuscript. E.von M. made significant contributions to the design of the study and revisions of the manuscript. S.K.W. provided detailed information about the study data and contributed significantly to the design and writing of the manuscript.


View Abstract