Spatio-Temporal Trends in the Incidence of Type 2 Diabetes in Germany

Analysis of the Claims Data of 63 Million Persons With Statutory Health Insurance From 2014 to 2019

Background: There are no data on recent trends in the incidence rate of type 2 diabetes (T2D) in Germany. The aim of this study was to determine the sex-, age-, and region-specific trends in the T2D incidence rate between 2014 and 2019.

Methods: Based on nationwide data from statutorily insured persons in Germany, negative binomial regression models were used to analyze age- and sex-specific trends in the T2D incidence rate. Age- and sex-adjusted trends were calculated for 401 administrative districts using a Bayesian spatio-temporal regression model.

Results: During the period concerned, approximately 450 000 new cases of T2D were observed each year among some 63 million persons. Taking all age groups together, the incidence rate decreased in both women and men, from 6.9 (95% confidence interval [6.7; 7.0]) and 8.4 [8.2; 8.6] respectively per 1000 persons in 2014 to 6.1 [5.9; 6.3] and 7.7 [7.5; 8.0] per 1000 persons in 2019. This corresponds to an annual reduction of 2.4% [1.5; 3.2] for women and 1.7% [0.8; 2.5] for men. The incidence rate increased in the age group 20–39 years. The age- and sex-adjusted incidence rate decreased in almost all districts, although regional differences persisted.

Conclusion: The T2D incidence rate should be closely monitored to see whether the decreasing trend continues. One must not forget that the prevalence can rise despite decreasing incidence. For this reason, the findings do not necessarily mean a decrease in the disease burden of T2D and the associated demand on healthcare resources.

Over the past decades, the prevalence of type 2 diabetes (T2D) has been increasing continuously both in Germany and internationally (1, 2, 3). Modelling studies indicate that the number of people with T2D will continue to increase significantly in the future (3, 4, 5). The current and future prevalence of T2D is strongly influenced by the incidence of T2D, i.e. the number of new cases relative to the population without T2D. A meta-analysis showed that the incidence of T2D has been plateauing or declining in some high-income countries since the mid-2000s (6). The increase in prevalence over the same period of time is only an apparent contradiction, as improvements in medical care for people with T2D have resulted in increased life expectancy, which in turn has led to an increase in prevalence. As the result of the increased life expectancy, people live longer after their diagnosis of T2D; consequently, more people in the population have T2D at a given time (4, 7).

Studies on temporal trends in the incidence rate of T2D in Germany are rare. The most recent study from Germany included in the meta-analysis mentioned above found evidence of an increase in incidence during the period 2008–2010 (6, 8). In contrast, the German Healthcare Atlas of the Central Research Institute for Ambulatory Health Care (Zi, Zentralinstitut für die kassenärztliche Versorgung) reported a nationwide decrease in the incidence of T2D for the period 2012–2014. Both studies investigated a comparatively short and more distant period of time. Accordingly, Heidemann et al. (9) conclude in their review that it is difficult to assess temporal trends in the incidence rate of T2D in Germany, particularly with respect to age- and sex-specific trends. Apart from differences related to age and sex, previous studies also described considerable regional differences in T2D incidence (2). However, there is hardly any data on how these regional differences are evolving over time. In addition, previous studies were based on a comparatively rough regional breakdown at the level of the German Federal States or the regions of the 17 Associations of Statutory Health Insurance Accredited Physicians (KVs, kassenärztliche Vereinigungen).

Thus, the aim of this study was to determine the sex- and age-specific trends in the T2D incidence rate in Germany between 2014 and 2019. In addition, we wanted to describe recent trends with regard to regional differences in T2D incidence between the 401 administrative districts.

Methods

Data source

Based on nationwide claims data of all German national health insurances (GKV, gesetzliche Krankenkassen) for the period 2014–2019, we estimated the nationwide and disease-specific incidence rates of T2D. The underlying data set comprises all statutorily insured persons with at least one outpatient contact with a physician per year. Approximately 85% of the German population have statutory health insurance. In line with the study by Goffrier et al. (2), we excluded persons already diagnosed with diabetes within the previous three years. The purpose of this diagnosis-free period is to ensure that people who already have diabetes and are therefore no longer at risk of newly developing the disease are excluded from the analysis. Choosing a diagnosis-free period that is too short may result in overestimation of the incidence rate (2). Furthermore, only insured with valid residence information (district), age < 111 years and male or female sex were included in the analysis. The diverse sex was excluded due to the low number of cases.

In order to ensure anonymity of the data, the Central Research Institute for Ambulatory Health Care (Zi) provided the data aggregated by age and sex, with a minimum of 30 cases in each age and sex stratum. At the federal level, the data made available were split into 20 5-year age groups (<5 years to ≥ 95 years), stratified by sex, and for each stratum the number of incident cases and the number of persons at risk was reported. Stratification of the data by age was not possible due to the granular regional stratification into 401 administrative districts; hence, only the total number of new cases of T2D was available at the district level. However, the number of persons at risk for newly developed T2D in 5-year age groups, stratified by sex, could be made available, allowing indirect standardization by age and sex.

Identification of incident cases of type 2 diabetes

New cases of T2D were identified based on the ICD-10 classification (10^th revision of the International Statistical Classification of Diseases and Related Health Problems) codes used for billing purposes, with only confirmed diagnoses considered. A minimum of one additional T2D coding in one of the three subsequent quarters was required after the first coding of T2D to reduce the likelihood of false positive diagnoses (m2Q criterion). In line with Goffrier et al. (2), the following ICD-10 codes were used to define T2D: E11 (T2D), E12 (malnutrition-related diabetes mellitus), E13 (other specified diabetes mellitus), and E14 (unspecified diabetes mellitus). Since at least two diagnoses in two different quarters were required to identify an incident case, various combinations of these ICD codes were possible. Detailed information about which specific combinations were defined as T2D can be found in Goffrier et al. (2). The diagnoses E12–E14 were included because the number of T2D diagnoses is presumably underestimated if only E11 diagnoses are considered (2). The data only comprise outpatient diagnoses and not inpatient diagnoses.

Statistical analysis

Age- and sex-specific incidence rates for the period 2014–2019 were estimated using negative binomial regression models. Temporal changes in incidence rates were reported as annual percentage changes (APCs).

On the district level, the standardized incidence ratio (SIR) was calculated to explore regional differences. SIR is defined as the ratio of the observed number of incident cases and the expected number of incident cases in a district. The latter number was calculated using the age- and sex-specific incidence rate of the available 2014 study population. SIR can be interpreted as a factor by which the incidence rate of a district differs from the nationwide incidence rate in 2014, regardless of differences in age and sex distribution. The standardized incidence ratio was calculated using a Bayesian spatio-temporal regression model. This model used information from neighboring districts to estimate the SIR of a district. This approach is based on the assumption that the T2D incidence rates of neighboring districts show greater similarity compared to the T2D incidence rates of non-neighboring districts. A detailed description of the statistical analysis can be found under eMethods and in eTable 1, eFigure 1 and eFigure 2.

eFigure 1

Age-specific incidence rate of type 2 diabetes from 2014 to 2019

eFigure 2

Age-specific incidence rate of type 2 diabetes from 2014 to 2019

eTable 1

Comparison of model fit of different splines

Results

Results at federal level

A total of approximately 450 000 new cases of T2D per year were observed among some 63 million persons at risk (eFigure 3).

eFigure 3

Composition of the study population

Figure 1a and eFigure 4 show the results of the negative binomial regression analysis at the federal level. A strong nonlinear relationship between incidence rate and age is noted, with higher age-specific rates among men compared to women. In addition, it is evident (Figure 1b) that the incidence rate in the older age groups decreased between 2014 and 2019. In the <60 age groups, some of the 2019 incidence rates are higher than the 2014 rates. The same can be seen in the linear trends for various age groups. In all age groups combined, the incidence rate decreased in both women and men, from 6.9 (confidence interval [6.7; 7.0]) and 8.4 [8.2; 8.6], respectively, per 1000 persons in 2014 to 6.1 [5.9; 6.3] and 7.7 [7.5; 8.0], respectively, per 1000 persons in 2019. This corresponds to an annual decrease by –2.4% [–3.2; –1.5] and –1.7% [–2.5; –0.8], respectively. Among older age groups, the decrease in incidence rate was more pronounced. In the age group from 20 to 39 years, by contrast, an annual increase in incidence rate by 2.9% [1.8; 4.0] and 2.4% [1.4; 3.3], respectively, was noted. This age group comprises approximately 7% of all incident cases during the period concerned.

Figure 1

Temporal trends in the age- and sex-specific specific incidence rate of type 2 diabetes in Germany from 2014 to 2019

eFigure 4

Temporal trends of age- and sex-specific type-2 diabetes incidence rate in Germany from 2014 to 2019

Results at district level

Figure 2 and eFigure 5 show the results for SIR, based on the spatio-temporal regression model. It is evident that, irrespective of the age and sex distribution in the districts, the incidence rate tends to be above the 2014 national average in the new federal states and Saarland and below the 2014 national average in the northwest and south of Germany. Like the nationwide incidence rate, the district-level SIR declines over the observation period. Thus, in 2019 considerably more districts were below the 2014 nationwide incidence rate compared to the situation in 2014 (eFigure 6). Accordingly, almost all districts show a decrease in SIR, with a median annual decrease of –2.2% (Figure 3). In 14 districts, however, an increasing trend is noted. In the Main-Kinzig, Dessau-Roßlau and Gotha districts, the SIR increased by more than 1% per year.

Figure 2

Spatio-temporal trends in the age- and sex-specific specific incidence ratio of type 2 diabetes

Figure 3

Change in age- and sex-adjusted standardized incidence ratio

eFigure 5

Spatio-temporal trends in the age- and sex-specific standardized incidence ratio (SIR) in Germany from 2014 to 2019

eFigure 6

Change in age- and sex-adjusted standardized incidence ratio (SIR) of type 2 diabetes in 401 administrative districts from 2014 to 2019

Even though the SIR decreased in almost all districts, the differences between the districts essentially remained. In 2014, the districts with the lowest SIR values were approximately 20–30% below the national average of 2014, while districts with the highest SIR values were 40–50% above it (eFigure 6). For 2019, the picture was similar; however, SIR distribution shifted toward lower values. A tabulation of the results on the district level is provided in eTable 2.

Discussion

Based on data of some 63 million statutorily insured persons annually, temporal and regional trends in T2D incidence rate in Germany were analyzed for the period 2014–2019. Overall, decreasing incidence rates were found nationwide; however, an increase was noted in the age group 20–39 years. With few exceptions, the age- and sex-adjusted incidence rates also declined on a regional level in almost all 401 administrative districts, although the considerable differences between the districts largely remained.

These results are in line with a recently published meta-analysis that found declining or stable incidence rates in some high-income countries since the 2010s (6). The previous study based on claims data of persons with statutory health insurance for the period 2012–2014 also showed a nationwide decrease in T2D incidence rate (2). This positive development of the T2D incidence rate may be attributable to the success of preventive measures (6). However, no clear trend was found for the current development of major T2D risk factors. For example, decreasing trends in smoking prevalence were observed between 1995 and 2018 (10). However, the prevalence of overweight remained almost stable between 2012 and 2020 and the prevalence of obesity even increased during the same period (11).

The significant regional differences in T2D incidence are likely caused by socioeconomic differences (12), but environmental factors, such as regional variations in exposure to air pollution or access to green spaces, may also play a role (13).

Despite the positive trends in T2D incidence rate, further close monitoring of the current development is required, especially in the light of increasing incidence rates among younger age groups. Findings from Denmark show that these dynamics can reverse quickly. While the T2D incidence rates in this country declined between 2011 and 2014, increasing T2D incidence rates were observed again between 2014 und 2016 (7). Another point to note is that despite declining incidence rates, the number of affected individuals may increase in the future, since, besides the incidence rate, changes in the life expectancy of people with or without T2D also have an impact on the future T2D prevalence (4). In addition, it is still unknown how the COVID-19 pandemic and its consequences will affect the incidence of T2D in the years to come. Existing studies suggest that COVID-19 may increase the risk to develop T2D (14). It is also possible that pandemic restrictions may have adversely affected health-related behaviors, such as physical activity or tobacco use (15). Data of the German Study on Tobacco Use (DEBRA) showed an increasing trend in smoking prevalence since end of 2020 (16). Thus, it is conceivable that T2D incidence will increase in the years after 2020. Future studies, examining this aspect in greater detail, can use the findings of our study for comparison.

In the light of the high current and future T2D-related disease burden in Germany (17, 18), our results should not be a reason to abandon efforts to effectively prevent T2D. For example, the German Diabetes Association recommends preventive measures at population level, such as the introduction of a value-added tax tiered based on nutritional profile as well as mandatory labeling with the Nutri-Score for all food products (19). Besides prevention, attention should be paid to providing adequate outpatient and inpatient care for people with diabetes, especially given the high number of people with diabetes receiving inpatient treatment (20). People with T2D seen in primary care may benefit from participating in disease management programs, as these can be associated with improved quality of care (21). As the population ages, the number of people with T2D in older age groups is likely to increase significantly in the coming decades (4). The diabetes care structures should be adapted and expanded accordingly.

Strengths and limitations

One of the strengths of our study is that it is based on a large data set which allowed precise estimates even on a small-scale level, in conjunction with the spatio-temporal regression model. In comparison to similar studies analyzing data from Germany, our study covers a longer observation period of six years. However, when interpreting our results, limitations should be taken into account. The underlying data only allow conclusions to be drawn about diagnosed T2D. As a result, the incidence rate is underestimated, since people with undiagnosed T2D (22) are not taken into account. This may also have an impact on temporal trends, for example, if the proportion of unknown T2D in all T2D cases changes over the period considered. Furthermore, the underlying data set is influenced by the coding behavior of the billing doctors. Thus, the possibility cannot be excluded that some temporal trends in T2D incidence rates are due to adjustments in documentation made e.g. for administrative reasons. Recently, it has been shown that in an older study based on statutory health insurance claims data some 200 000 false-positive diagnoses of type 2 diabetes were present (23). When compared to persons with private health insurance, persons with statutory health insurance tend to be in poorer health (24). Consequently, the T2D incidence may be overestimated, if our results are applied to the general population. However, since about 90% of the German population have statutory health insurance, this potential overestimation is likely to be minor. Despite the fact that our study’s observation period is rather long in comparison to previous studies, it is still too short to allow for estimation of long-term trends. For example, in the United States and Denmark data sources are available that permit tracking of T2D incidence rates over several decades (7, 25). Observing such long periods of time is necessary to be able to place short-term trends in the context of long-term developments.

In summary, our results suggest that between 2014 and 2019 the T2D incidence rate decreased in Germany by about 2% annually. Age-specific differences were found, with a stronger decrease in incidence rates in the age groups 60 years and older and increasing incidence rates in the age group from 20 to 39 years. Considerable regional differences in the age- and sex-adjusted incidence rate persisted over the entire period considered. The development should be further monitored to determine whether these short-term trends will continue in the longer term and also beyond the COVID-19 pandemic. Further research should be conducted to identify causes and measures to overcome ongoing regional differences in T2D incidence rate.

Conflict of interest statement
The authors declare that no conflict of interest exists.

Manuscript received on 8 September 2022, revised version accepted on 14 December 2022.

Translated from the original German by Ralf Thoene, MD

Corresponding author

Dr. PH Thaddäus Tönnies
Deutsches Diabetes-Zentrum (DDZ)
Auf’m Hennekamp 65, 40225 Düsseldorf, Germany

Cite this as:
Tönnies T, Hoyer A, Brinks R, Kuss O, Hering R, Schulz M: Spatio-temporal trends in the incidence of type 2 diabetes in Germany—analysis of the claims data of 63 million persons with statutory health insurance from 2014 to 2019. Dtsch Arztebl Int 2023; 120: 173–9. DOI: 10.3238/arztebl.m2022.0405

►Supplementary material

eReferences, eMethods, eTables, eFigures, eBox:
www.aerzteblatt-international.de/m2022.0405

References

1. Heidemann C, Scheidt-Nave C, Beyer A-K, et al.: Gesundheitliche Lage von Erwachsenen in Deutschland—Ergebnisse zu ausgewählten Indikatoren der Studie GEDA 2019/2020-EHIS. J Health Monit 2021; 6: 3–27.

2. Goffrier B, Schulz M, Bätzing-Feigenbaum J: Administrative Prävalenzen und Inzidenzen des Diabetes mellitus von 2009 bis 2015 Zentralinstitut für die kassenärztliche Versorgung in Deutschland (Zi). Versorgungsatlas-Bericht Nr. 17/03. Berlin 2017.

3. International Diabetes Federation: IDF Diabetes Atlas. 10 th edition. https://diabetesatlas.org/atlas/tenth-edition/?dlmodal=active&dlsrc=https%3A%2F%2Fdiabetesatlas.org%2Fidfawp%2Fresource-files%2F2021%2F07%2FIDF_Atlas_10th_Edition_2021.pdf(last accessed on 08 July 2022).

4. Tönnies T, Röckl S, Hoyer A, et al.: Projected number of people with diagnosed type 2 diabetes in Germany in 2040. Diabet Med 2019; 36: 1217–25

5. Voeltz D, Tönnies T, Brinks R, Hoyer A: Future prevalence of type 2 diabetes—a comparative analysis of chronic disease projection methods. PLoS One 2022; 17: e0264739

6. Magliano DJ, Islam RM, Barr ELM, et al.: Trends in incidence of total or type 2 diabetes: systematic review. BMJ 2019; 366: l5003

7. Carstensen B, Rønn PF, Jørgensen ME: Prevalence, incidence and mortality of type 1 and type 2 diabetes in Denmark 1996–2016. BMJ Open Diabetes Res Care 2020; 8: e001071

8. Boehme MWJ, Buechele G, Frankenhauser-Mannuss J, et al.: Prevalence, incidence and concomitant co-morbidities of type 2 diabetes mellitus in South Western Germany—a retrospective cohort and case control study in claims data of a large statutory health insurance. BMC Public Health 2015; 15: 855

9. Heidemann C, Scheidt-Nave C: Prevalence, incidence and mortality of diabetes mellitus in adults in Germany—a review in the framework of the diabetes surveillance. J Health Monit 2017; 2: 98–121.

10. Seitz N-N, Lochbühler K, Atzendorf J, Rauschert C, Pfeiffer-Gerschel T, Kraus L: Trends in substance use and related disorders: analysis of the epidemiological survey of substance abuse 1995 to 2018. Dtsch Arztebl Int 2019; 116: 585–91

11. Schienkiewitz A, Kuhnert R, Blume M, Mensink GBM: Übergewicht und Adipositas bei Erwachsenen in Deutschland—Ergebnisse der Studie GEDA 2019/2020-EHIS. J Health Monit 2022; 7: 23–31.

12. Maier W, Holle R, Hunger M, et al.: The impact of regional deprivation and individual socio-economic status on the prevalence of type 2 diabetes in Germany. A pooled analysis of five population-based studies. Diabet Med 2013; 30: e78–86

13. Dendup T, Feng X, Clingan S, Astell-Burt T: Environmental risk factors for developing type 2 diabetes mellitus: a systematic review. Int J Environ Res Public Health 2018; 15: 78

14. Rathmann W, Kuss O, Kostev K: Incidence of newly diagnosed diabetes after Covid-19. Diabetologia 2022; 65: 949–54

15. Klosterhalfen S, Kotz D, Kastaun S: Did smoking, alcohol consumption, and physical activity change during the COVID-19 restrictions in Germany in spring 2020? Sucht 2022; 68: 129–38

16. Deutsche Befragung zum Rauchverhalten (DEBRA) Studie: Aktuelle nationale Statistiken. www.debra-study.info (last accessed on 29 November 2022).

17. Tönnies T, Hoyer A, Brinks R: Productivity-adjusted life years lost due to type 2 diabetes in Germany in 2020 and 2040. Diabetologia 2021; 64: 1288–97

18. Wengler A, Rommel A, Plaß D, et al.: Years of life lost to death—a comprehensive analysis of mortality in Germany conducted as part of the BURDEN 2020 project. Dtsch Arztebl Int 2021; 118: 137–44

19. Deutsche Diabetes Gesellschaft: Empfehlungen der DDG zur Nationalen Diabetesstrategie: Welche konkreten Maßnahmen müssen nun folgen? www.ddg.info/fileadmin/user_upload/Politische_Empfehlungen_DDG_2021.pdf (last accessed on 08 July 2022).

20. Auzanneau M, Fritsche A, Icks A, et al.: Diabetes in the hospital—a nationwide analysis of all hospitalized cases in Germany with and without diabetes, 2015–2017. Dtsch Arztebl Int 2021; 118: 407–12

21. Laxy M, Stark R, Meisinger C, et al.: The effectiveness of German disease management programs (DMPs) in patients with type 2 diabetes mellitus and coronary heart disease: results from an observational longitudinal study. Diabetol Metab Syndr 2015; 7: 77

22. Heidemann C, Du Y, Paprott R, Haftenberger M, Rathmann W, Scheidt-Nave C: Temporal changes in the prevalence of diagnosed diabetes, undiagnosed diabetes and prediabetes: findings from the German health interview and examination surveys in 1997–1999 and 2008–2011. Diabet Med 2016; 33: 1406–14

23. Brinks R, Tönnies T, Hoyer A: Importance of diagnostic accuracy in big data: false-positive diagnoses of type 2 diabetes in health insurance claims data of 70 million Germans. Front Epidemiol 2022; 2: 887335

24. Stauder J, Kossow T: Selektion oder bessere Leistungen—Warum sind Privatversicherte gesünder als gesetzlich Versicherte? Gesundheitswesen 2017; 79: 181–7

25. Geiss LS, Wang J, Cheng YJ, et al.: Prevalence and incidence trends for diagnosed diabetes among adults aged 20 to 79 years, United States, 1980–2012. JAMA 2014; 312: 1218–26

e1. Bernardinelli L, Clayton D, Pascutto C, Montomoli C, Ghislandi M, Songini M: Bayesian analysis of space—time variation in disease risk. Stat Med 1995; 14: 2433–43

e2. Besag J, York J, Mollié A: Bayesian image restoration, with two applications in spatial statistics. Ann Inst Stat Math 1991; 43: 1–20

e3. Rue H, Martino S, Chopin N: Approximate Bayesian inference for latent Gaussian models by using integrated nested Laplace approximations. J R Stat Soc Series B Stat Methodol 2009; 71: 319–92

e4. Blangiardo M, Cameletti M, Baio G, Rue H: Spatial and spatio-temporal models with R-INLA. Spat Spatiotemporal Epidemiol 2013; 4: 33–49