The authors have declared that no competing interests exist.
Conceived and designed the experiments: WM AM. Analyzed the data: WM. Wrote the paper: WM. Critical review: CSN RH LK TL YD CH AM. Interpretation of data: CSN RH LK TL YD CH AM. Contribution to discussion: CSN RH LK TL YD CH AM.
There is increasing evidence that prevention programmes for type 2 diabetes mellitus (T2DM) and obesity need to consider individual and regional risk factors. Our objective is to assess the independent association of area level deprivation with T2DM and obesity controlling for individual risk factors in a large study covering the whole of Germany.
We combined data from two consecutive waves of the national health interview survey ‘GEDA’ conducted by the Robert Koch Institute in 2009 and 2010. Data collection was based on computer-assisted telephone interviews. After exclusion of participants <30 years of age and those with missing responses, we included n = 33,690 participants in our analyses. The outcome variables were the 12-month prevalence of known T2DM and the prevalence of obesity (BMI ≥30 kg/m2). We also controlled for age, sex, BMI, smoking, sport, living with a partner and education. Area level deprivation of the districts was defined by the German Index of Multiple Deprivation. Logistic multilevel regression models were performed using the software SAS 9.2.
Of all men and women living in the most deprived areas, 8.6% had T2DM and 16.9% were obese (least deprived areas: 5.8% for T2DM and 13.7% for obesity). For women, higher area level deprivation and lower educational level were both independently associated with higher T2DM and obesity prevalence [highest area level deprivation: OR 1.28 (95% CI: 1.05–1.55) for T2DM and OR 1.28 (95% CI: 1.10–1.49) for obesity]. For men, a similar association was only found for obesity [OR 1.20 (95% CI: 1.02–1.41)], but not for T2DM.
Area level deprivation is an independent, important determinant of T2DM and obesity prevalence in Germany. Identifying and targeting specific area-based risk factors should be considered an essential public health issue relevant to increasing the effectiveness of diabetes and obesity prevention.
There is sound evidence that the prevalence of type 2 diabetes mellitus (T2DM) is strongly associated with obesity and that both T2DM and obesity are inversely associated with individual socio-economic status (SES)
In Germany, as in all industrialized countries, T2DM poses a major public health problem. According to the International Diabetes Federation (IDF), the prevalence of diagnosed diabetes in Germany is still lower than in the US, but higher than in a number of other European countries
Studies from other countries are often limited as well, because they did not adjust for individual SES
Against this background, the present study aimed to (1) determine whether area level deprivation is associated with the prevalence of T2DM and obesity in Germany at the national level, independently of individual risk factors and (2) further explore sex-specific differences.
The German Health Update (‘Gesundheit in Deutschland Aktuell’, GEDA) survey system consists of periodically repeated representative national health interview surveys. GEDA is an integral part of the continuous health monitoring conducted by the Robert Koch Institute. We used cross-sectional data from the 2009 and 2010 GEDA survey waves, which were conducted between July 2008 and June 2009 (GEDA 2009) and between September 2009 and July 2010 (GEDA 2010) using highly standardized computer-assisted telephone interview (CATI) techniques. The methods have been described in detail previously
We intended to target T2DM but the assessment of diabetes in the GEDA surveys allows no distinction between different types of diabetes. T2DM accounts for about 90–95% of all diabetes cases
As the participation in the National Health Telephone Interview Surveys is voluntary, not arising any costs for survey participants, and because the study has no medical relevance for individual survey participants (no medical research involving human subjects is being conducted) an ethics approval was not compulsory. In terms of data protection and informed consent the study was approved by The Federal Commissioner for Data Protection and Freedom of Information. Verbal informed consent was provided by all participants prior to the interview.
Information regarding individual data is based on self-report as collected by CATI. For the present analysis, we included information on sex and chronological age, history of known diabetes mellitus, BMI, smoking status, sport activity, living with a partner and educational level.
Study participants were asked whether they had ever been diagnosed with diabetes by a physician (lifetime diagnosis) and, if yes, whether they had also been suffering from diabetes in the past 12 months. We defined the dependent variable ‘12-month prevalence of diabetes’ as a positive answer to both questions, which we used in order to reduce misclassification bias due to under- or overreporting. T2DM was defined by restricting the dataset as described above. Obesity was defined as a BMI ≥30 kg/m2, based on self-reported weight and height
In order to control for potential confounding, the following covariates were included in our analysis: sex, age (three categories: 30–49, 50–64 and ≥65 years), BMI (three categories: <25 kg/m2, 25 to <30 kg/m2 and ≥30 kg/m2), sport activity (measured by hours of sport activity, three categories: no sport activity, up to 4 hours/week, more than 4 hours/week)
Individual SES was defined by educational level
Area level deprivation was assessed by the German Index of Multiple Deprivation (GIMD), a recently introduced area-based deprivation measure that has not yet been applied to the GEDA dataset. The GIMD has been established based on the method used in the UK
We carried out univariate and bivariate analyses calculating chi-square statistics and Cochran–Armitage tests for trend. Then we performed logistic multilevel regression models and fitted two-level binomial logit-link models (level 1: individuals; level 2: districts) with random intercept, calculating first crude odds ratios (ORs) with their 95% confidence intervals (95% CIs). We tested for associations between district deprivation and the prevalence of T2DM (12-month prevalence) and obesity in subsequent models, controlling for potential confounding or effect modification. Sex-specific results were obtained by stratified analysis. Finally, we report ORs with their 95% CIs and area level variances (VA) with their standard errors (SE). In order to quantify the relevance of area level variation, we also calculated the median odds ratios (MORs), which can be calculated as a simple function of the area level variance VA
All analyses were performed as complete case analysis using the software SAS 9.2 (SAS Institute Inc., Cary, NC, USA). The logistic multilevel models were estimated with the SAS procedure GLIMMIX, using a maximum likelihood estimation based on Laplace approximation.
Men | Women | Total | |
Participants (n) | 14,402 | 19,288 | 33,690 |
T2DM |
7.7 | 6.1 | 6.8 |
Obesityb (%) | 16.4 | 14.2 | 15.2 |
Independent variables (%) | |||
Age (years) | |||
30–49 | 46.8 | 47.7 | 47.3 |
50–64 | 29.6 | 30.2 | 29.9 |
≥65 | 23.6 | 22.2 | 22.8 |
BMI (kg/m2) | |||
<25 | 36.6 | 56.3 | 47.9 |
25–<30 | 47.0 | 29.5 | 37.0 |
≥30 | 16.4 | 14.2 | 15.2 |
Smoking status | |||
never smoker | 35.6 | 50.2 | 44.0 |
ex-smoker | 34.8 | 24.9 | 29.1 |
current smoker | 29.6 | 24.9 | 26.9 |
Sport activity | |||
>4 h/week | 24.4 | 19.2 | 21.4 |
up to 4 h/week | 41.2 | 49.5 | 46.0 |
no sport activity | 34.4 | 31.3 | 32.6 |
Partner | |||
living with a partner | 74.3 | 65.9 | 69.5 |
living without a partner | 25.7 | 34.1 | 30.5 |
Educational level | |||
high level | 46.9 | 38.4 | 42.0 |
medium level | 25.1 | 34.6 | 30.5 |
low level | 28.0 | 27.1 | 27.5 |
GIMD quintiles (Q) | |||
Q1 ( = least deprived) | 24.0 | 23.6 | 23.8 |
Q2 | 21.9 | 21.6 | 21.8 |
Q3 | 19.4 | 19.3 | 19.4 |
Q4 | 16.6 | 17.3 | 17.0 |
Q5 ( = most deprived) | 18.1 | 18.2 | 18.1 |
Crude 12-month prevalence of type 2 diabetes; b crude prevalence of obesity (BMI ≥30 kg/m2).
T2DM |
Obesityb | |||||
Men | Women | Total | Men | Women | Total | |
Participants (n) | 14,402 | 19,288 | 33,690 | 14,402 | 19,288 | 33,690 |
Independent variables (%) | ||||||
Age (years) | ||||||
30–49 | 1.8 | 1.7 | 1.8 | 13.5 | 10.5 | 11.8 |
50–64 | 9.6 | 6.7 | 7.9 | 20.8 | 17.4 | 18.8 |
≥65 | 17.2 | 14.6 | 15.8 | 16.8 | 18.0 | 17.5 |
P<0.0001 | P<0.0001 | P<0.0001 | P<0.0001 | P<0.0001 | P<0.0001 | |
BMI (kg/m2) | ||||||
<25 | 3.3 | 2.3 | 2.7 | – | – | – |
25–<30 | 7.3 | 6.9 | 7.2 | – | – | – |
≥30 | 18.7 | 19.1 | 18.9 | – | – | – |
P<0.0001 | P<0.0001 | P<0.0001 | ||||
Smoking status | ||||||
never smoker | 6.2 | 6.9 | 6.6 | 13.8 | 14.9 | 14.5 |
ex-smoker | 11.7 | 6.4 | 9.1 | 21.0 | 16.0 | 18.6 |
current smoker | 4.9 | 4.1 | 4.5 | 14.1 | 11.2 | 12.6 |
P<0.0001 | P<0.0001 | P<0.0001 | P<0.0001 | P<0.0001 | P<0.0001 | |
Sport activity | ||||||
>4 h/week | 6.1 | 4.2 | 5.1 | 11.2 | 9.3 | 10.2 |
up to 4 h/week | 5.6 | 4.0 | 4.7 | 14.1 | 12.1 | 12.9 |
no sport activity | 11.5 | 10.4 | 10.9 | 22.8 | 20.7 | 21.7 |
P<0.0001 | P<0.0001 | P<0.0001 | P<0.0001 | P<0.0001 | P<0.0001 | |
Partner | ||||||
living with a partner | 7.5 | 4.6 | 5.9 | 16.4 | 13.2 | 14.7 |
living without a partner | 8.4 | 9.0 | 8.8 | 16.4 | 16.3 | 16.3 |
P = 0.109 | P<0.0001 | P<0.0001 | P = 0.969 | P<0.0001 | P<0.0001 | |
Educational level | ||||||
high level | 5.6 | 2.9 | 4.2 | 11.8 | 8.6 | 10.1 |
medium level | 6.7 | 4.4 | 5.2 | 17.6 | 13.4 | 14.9 |
low level | 12.3 | 12.6 | 12.5 | 23.1 | 23.4 | 23.3 |
P<0.0001 | P<0.0001 | P<0.0001 | P<0.0001 | P<0.0001 | P<0.0001 | |
GIMD quintiles (Q) | ||||||
Q1 ( = least deprived) | 6.9 | 4.9 | 5.8 | 14.9 | 12.7 | 13.7 |
Q2 | 6.9 | 5.4 | 6.0 | 15.5 | 13.5 | 14.4 |
Q3 | 8.2 | 5.8 | 6.8 | 17.9 | 14.1 | 15.7 |
Q4 | 8.1 | 6.5 | 7.2 | 16.8 | 15.2 | 15.9 |
Q5 ( = most deprived) | 9.1 | 8.3 | 8.6 | 17.7 | 16.4 | 16.9 |
P = 0.008 | P<0.0001 | P<0.0001 | P = 0.004 | P<0.0001 | P<0.0001 |
Crude 12-month prevalence of type 2 diabetes; b crude prevalence of obesity (BMI ≥30 kg/m2).
P values: Chi-square test.
Models for both men and women combined (
Men and women | ||||
Variables | Crude | Model 1 | Model 2 | Model 3 |
OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | |
Sex ( |
||||
men | 1.30 (1.20–1.42) | 1.29 (1.18–1.40) | 1.32 (1.20–1.45) | 1.19 (1.08–1.31) |
Age ( |
||||
50–64 | 4.80 (4.18–5.52) | 4.78 (4.16–5.49) | 4.06 (3.52–4.67) | 3.67 (3.18–4.23) |
≥65 | 10.48 (9.17–11.98) | 10.31 (9.02–11.79) | 7.16 (6.21–8.25) | 7.30 (6.32–8.43) |
BMI ( |
||||
25–<30 | 2.81 (2.50–3.16) | – | – | 2.09 (1.85–2.36) |
≥30 | 8.54 (7.58–9.61) | – | – | 5.96 (5.26–6.76) |
Smoking status ( |
||||
ex-smoker | 1.42 (1.29–1.56) | – | 1.31 (1.18–1.45) | 1.24 (1.12–1.38) |
current smoker | 0.66 (0.59–0.75) | – | 0.81 (0.72–0.93) | 0.96 (0.84–1.09) |
Sport activity ( |
||||
up to 4 h/week | 0.90 (0.79–1.03) | – | 1.04 (0.91–1.18) | 0.93 (0.81–1.06) |
no sport activity | 2.26 (2.00–2.55) | – | 1.92 (1.70–2.18) | 1.52 (1.34–1.73) |
Partner ( |
||||
living without a partner | 1.53 (1.40–1.67) | – | 1.29 (1.17–1.41) | 1.30 (1.18–1.43) |
Educational level ( |
||||
medium level | 1.25 (1.10–1.40) | – | 1.25 (1.11–1.42) | 1.14 (1.00–1.29) |
low level | 3.29 (2.96–3.64) | – | 1.89 (1.70–2.11) | 1.49 (1.33–1.67) |
GIMD quintiles ( |
||||
Q2 | 1.05 (0.91–1.21) | 1.05 (0.91–1.21) | 1.00 (0.87–1.15) | 0.99 (0.86–1.15) |
Q3 | 1.18 (1.02–1.37) | 1.19 (1.03–1.37) | 1.11 (0.97–1.28) | 1.08 (0.94–1.25) |
Q4 | 1.27 (1.10–1.48) | 1.21 (1.04–1.40) | 1.14 (0.99–1.32) | 1.08 (0.93–1.25) |
Q5 ( = most deprived) | 1.59 (1.37–1.84) | 1.37 (1.19–1.58) | 1.26 (1.10–1.45) | 1.18 (1.03–1.35) |
Variances | ||||
VA (SE) | – | 0.011 (0.011) | 0 | 0 |
MOR | – | 1.11 | 1.00 | 1.00 |
*: Reference group; OR (odds ratios); 95% CI (95% confidence intervals); bold type = significant.
VA (area level variance); SE (standard error); MOR (median odds ratio).
Men and women | |||
Variables | Crude | Model 1 | Model 2 |
OR (95% CI) | OR (95% CI) | OR (95% CI) | |
Sex ( |
|||
men | 1.18 (1.12–1.26) | 1.18 (1.11–1.26) | 1.22 (1.14–1.30) |
Age ( |
|||
50–64 | 1.74 (1.62–1.86) | 1.73 (1.61–1.86) | 1.44 (1.34–1.55) |
≥65 | 1.60 (1.48–1.72) | 1.58 (1.46–1.70) | 1.03 (0.95–1.12) |
Smoking status ( |
|||
ex-smoker | 1.35 (1.26–1.45) | – | 1.27 (1.18–1.36) |
current smoker | 0.85 (0.78–0.92) | – | 0.71 (0.65–0.77) |
Sport activity ( |
|||
up to 4 h/week | 1.30 (1.18–1.42) | – | 1.37 (1.25–1.50) |
no sport activity | 2.42 (2.21–2.64) | – | 2.24 (2.05–2.46) |
Partner ( |
|||
living without a partner | 1.16 (1.08–1.23) | – | 1.15 (1.07–1.23) |
Educational level ( |
|||
medium level | 1.54 (1.42–1.66) | – | 1.53 (1.41–1.66) |
low level | 2.67 (2.48–2.88) | – | 2.33 (2.16–2.53) |
GIMD quintiles ( |
|||
Q2 | 1.07 (0.95–1.20) | 1.06 (0.94–1.19) | 1.03 (0.92–1.15) |
Q3 | 1.19 (1.06–1.34) | 1.19 (1.06–1.34) | 1.14 (1.02–1.27) |
Q4 | 1.21 (1.07–1.37) | 1.20 (1.06–1.35) | 1.17 (1.04–1.32) |
Q5 ( = most deprived) | 1.38 (1.23–1.56) | 1.33 (1.18–1.50) | 1.27 (1.12–1.42) |
Variances | |||
VA (SE) | – | 0.043 (0.010) | 0.031 (0.009) |
MOR | – | 1.22 | 1.18 |
*: Reference group; OR (odds ratios); 95% CI (95% confidence intervals); bold type = significant.
VA (area level variance); SE (standard error); MOR (median odds ratio).
Separate analyses revealed important differences between men and women for the prevalence of T2DM (
Men | ||||
Variables | Crude | Model 1 | Model 2 | Model 3 |
OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | |
Age ( |
||||
50–64 | 5.77 (4.69–7.10) | 5.77 (4.69–7.09) | 5.02 (4.07–6.19) | 4.65 (3.77–5.75) |
≥65 | 11.39 (9.32–13.92) | 11.34 (9.28–13.86) | 8.87 (7.19–10.95) | 9.42 (7.61–11.66) |
BMI ( |
||||
25–<30 | 2.29 (1.92–2.74) | – | – | 1.90 (1.59–2.28) |
≥30 | 6.65 (5.54–7.98) | – | – | 5.19 (4.28–6.30) |
Smoking status ( |
||||
ex-smoker | 1.99 (1.73–2.29) | – | 1.48 (1.28–1.72) | 1.35 (1.16–1.57) |
current smoker | 0.78 (0.65–0.93) | – | 0.89 (0.74–1.08) | 0.98 (0.81–1.19) |
Sport activity ( |
||||
up to 4 h/week | 0.93 (0.78–1.10) | – | 1.06 (0.89–1.27) | 0.98 (0.81–1.17) |
no sport activity | 2.00 (1.70–2.36) | – | 1.71 (1.44–2.03) | 1.42 (1.19–1.69) |
Partner ( |
||||
living without a partner | 1.12 (0.98–1.28) | – | 1.24 (1.08–1.44) | 1.28 (1.11–1.49) |
Educational level ( |
||||
medium level | 1.22 (1.03–1.44) | – | 1.33 (1.12–1.58) | 1.23 (1.03–1.46) |
low level | 2.40 (2.08–2.76) | – | 1.53 (1.32–1.77) | 1.31 (1.12–1.52) |
GIMD quintiles ( |
||||
Q2 | 1.00 (0.82–1.20) | 0.98 (0.80–1.19) | 0.93 (0.77–1.14) | 0.92 (0.76–1.13) |
Q3 | 1.20 (0.99–1.45) | 1.21 (1.00–1.47) | 1.13 (0.93–1.38) | 1.07 (0.88–1.31) |
Q4 | 1.18 (0.97–1.43) | 1.16 (0.95–1.42) | 1.10 (0.90–1.36) | 1.04 (0.85–1.29) |
Q5 ( = most deprived) | 1.34 (1.11–1.61) | 1.19 (0.98–1.44) | 1.12 (0.92–1.36) | 1.07 (0.88–1.30) |
Variances | ||||
VA (SE) | – | 0 | 0 | 0 |
MOR | – | 1.00 | 1.00 | 1.00 |
*: Reference group; OR (odds ratios); 95% CI (95% confidence intervals); bold type = significant.
VA (area level variance); SE (standard error); MOR (median odds ratio).
Women | ||||
Variables | Crude | Model 1 | Model 2 | Model 3 |
OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | |
Age ( |
||||
50–64 | 4.09 (3.39–4.94) | 4.05 (3.36–4.89) | 3.27 (2.69–3.96) | 2.88 (2.37–3.50) |
≥65 | 9.76 (8.16–11.67) | 9.53 (7.96–11.40) | 5.50 (4.51–6.71) | 5.45 (4.45–6.67) |
BMI ( |
||||
25–<30 | 3.10 (2.64–3.65) | – | – | 2.19 (1.85–2.58) |
≥30 | 9.89 (8.45–11.58) | – | – | 6.44 (5.45–7.61) |
Smoking status ( |
||||
ex-smoker | 0.95 (0.82–1.09) | – | 1.18 (1.02–1.37) | 1.14 (0.98–1.33) |
current smoker | 0.59 (0.50–0.69) | – | 0.77 (0.65–0.92) | 0.95 (0.80–1.14) |
Sport activity ( |
||||
up to 4 h/week | 0.96 (0.79–1.16) | – | 1.04 (0.86–1.26) | 0.91 (0.75–1.11) |
no sport activity | 2.64 (2.20–3.16) | – | 2.17 (1.80–2.61) | 1.64 (1.36–1.99) |
Partner ( |
||||
living without a partner | 2.07 (1.84–2.34) | – | 1.33 (1.17–1.52) | 1.34 (1.17–1.53) |
Educational level ( |
||||
medium level | 1.51 (1.26–1.81) | – | 1.31 (1.09–1.57) | 1.14 (0.95–1.38) |
low level | 4.80 (4.09–5.62) | – | 2.44 (2.06–2.89) | 1.77 (1.49–2.10) |
GIMD quintiles ( |
||||
Q2 | 1.10 (0.90–1.35) | 1.12 (0.91–1.37) | 1.09 (0.89–1.33) | 1.08 (0.88–1.32) |
Q3 | 1.19 (0.97–1.45) | 1.17 (0.95–1.44) | 1.10 (0.90–1.35) | 1.11 (0.90–1.36) |
Q4 | 1.37 (1.12–1.68) | 1.26 (1.03–1.55) | 1.17 (0.96–1.43) | 1.12 (0.91–1.38) |
Q5 ( = most deprived) | 1.82 (1.49–2.21) | 1.54 (1.27–1.89) | 1.39 (1.14–1.69) | 1.28 (1.05–1.55) |
Variances | ||||
VA (SE) | – | 0.022 (0.021) | 0.004 (0.021) | 0 |
MOR | – | 1.15 | 1.06 | 1.00 |
*: Reference group; OR (odds ratios); 95% CI (95% confidence intervals); bold type = significant.
VA (area level variance); SE (standard error); MOR (median odds ratio).
In contrast, the stratified analyses for obesity (
Men | |||
Variables | Crude | Model 1 | Model 2 |
OR (95% CI) | OR (95% CI) | OR (95% CI) | |
Age ( |
|||
50–64 | 1.68 (1.52–1.87) | 1.68 (1.52–1.86) | 1.41 (1.27–1.57) |
≥65 | 1.31 (1.16–1.46) | 1.30 (1.16–1.46) | 0.93 (0.82–1.06) |
Smoking status ( |
|||
ex-smoker | 1.66 (1.50–1.84) | – | 1.49 (1.34–1.66) |
current smoker | 1.02 (0.91–1.15) | – | 0.84 (0.74–0.95) |
Sport activity ( |
|||
up to 4 h/week | 1.30 (1.14–1.48) | – | 1.33 (1.17–1.51) |
no sport activity | 2.33 (2.06–2.64) | – | 2.14 (1.88–2.43) |
Partner ( |
|||
living without a partner | 1.02 (0.92–1.13) | – | 1.06 (0.95–1.18) |
Educational level ( |
|||
medium level | 1.57 (1.40–1.76) | – | 1.51 (1.34–1.70) |
low level | 2.22 (2.00–2.46) | – | 1.90 (1.70–2.12) |
GIMD quintiles ( |
|||
Q2 | 1.05 (0.90–1.23) | 1.04 (0.89–1.22) | 0.99 (0.86–1.15) |
Q3 | 1.24 (1.06–1.46) | 1.25 (1.06–1.46) | 1.18 (1.01–1.37) |
Q4 | 1.15 (0.98–1.36) | 1.15 (0.97–1.36) | 1.12 (0.96–1.32) |
Q5 ( = most deprived) | 1.31 (1.11–1.55) | 1.28 (1.08–1.51) | 1.20 (1.02–1.41) |
Variances | |||
VA (SE) | – | 0.049 (0.017) | 0.025 (0.015) |
MOR | – | 1.24 | 1.16 |
*: Reference group; OR (odds ratios); 95% CI (95% confidence intervals); bold type = significant.
VA (area level variance); SE (standard error); MOR (median odds ratio).
Women | |||
Variables | Crude | Model 1 | Model 2 |
OR (95% CI) | OR (95% CI) | OR (95% CI) | |
50–64 | |||
≥65 | 1.05 (0.93–1.18) | ||
ex-smoker | – | ||
current smoker | – | ||
up to 4 h/week | – | ||
no sport activity | – | ||
living without a partner | – | ||
medium level | – | ||
low level | – | ||
Q2 | 1.08 (0.93–1.25) | 1.08 (0.93–1.25) | 1.06 (0.92–1.22) |
Q3 | 1.13 (0.98–1.32) | 1.13 (0.97–1.31) | 1.09 (0.94–1.27) |
Q4 | |||
Q5 ( = most deprived) | |||
VA (SE) | – | 0.045 (0.015) | 0.034 (0.015) |
MOR | – | 1.22 | 1.19 |
*: Reference group; OR (odds ratios); 95% CI (95% confidence intervals); bold type = significant.
VA (area level variance); SE (standard error); MOR (median odds ratio).
Differences between area variances (VA, MOR) were generally low in the T2DM models. In the obesity models, there was a larger variation between districts, but this was quite similar for men and women (
Our objective was to evaluate the relationship between area level deprivation and the prevalence of T2DM and obesity, looking also at the role of educational level and potential differences between men and women. Our findings suggest that living in very deprived districts and having a low educational level are both independently associated with a higher prevalence of T2DM and a higher prevalence of obesity. The increased prevalence of obesity in these highly deprived areas applies to both men and women, but the increased prevalence of T2DM in the most deprived districts is confined to women. Also, the increased prevalence of T2DM and obesity associated with low educational level is stronger for women than for men. Concerning models with the dependent variable T2DM, it is important to note that controlling for BMI (like other covariates, e.g. smoking) may lead to potential overadjustment, and BMI could act as an important intermediate factor between area level deprivation and T2DM
Our findings are in good agreement with results reported from other countries. In the Diabetes Study of Northern California (DISTANCE), Laraia et al.
In our analyses, controlling for individual educational level seems to have little influence on the effect of area level deprivation. This indicates that individual SES and area level deprivation may act through different pathways
However, the role of individual educational level should not be neglected. Having a low educational level may lead to low health literacy. This could result, for instance, in less benefit from diabetes disease management programmes
Our results show that the effects of individual SES and area level deprivation are more pronounced among women than among men. Kavanagh et al.
Fano et al.
Some limitations of our study have to be taken into account. The potential for non-response bias has to be considered. Low educational level and diabetes have both been associated with non-response
Education is well accepted as being a good indicator of individual SES. For example, a study conducted in nine European countries demonstrated that the relationship between overweight and low education is stronger and more consistent than for other SES variables such as household income
Also, our analyses are based on districts. These administrative units vary considerably in area and population size. Therefore, the classification of individuals by area level deprivation may be more sensitive in smaller districts than in larger ones. Moreover, when assessing the association between area deprivation and health, the influence of the modifiable areal unit problem (MAUP) depends on the size of spatial units: using smaller areas (e.g. municipalities instead of districts) may provide even more significant results
Finally, the cross-sectional design of the dataset does not allow any causal interpretation of our findings.
Some important strengths of our study should be pointed out. We used an extensive database, based on a large representative nationwide dataset including individual data from two consecutive national health interview surveys conducted across the whole of Germany. This is an excellent resource for studying regional differences in the prevalence of T2DM and obesity. Using an established area-based deprivation measure for Germany, we were able to quantify the effect of area deprivation on the prevalence of T2DM and obesity controlling for individual educational level. To our knowledge, this is the first study looking at the association of area level deprivation, T2DM and obesity covering the whole of Germany at a district level.
In Germany, higher area level deprivation is associated with a higher prevalence of type 2 diabetes mellitus and obesity at the national level, independent of individual educational level and established risk factors. In order to reduce health disparities, diabetes and obesity prevention strategies need to consider individual as well as area-based risk factors