The Association of Mean Air Temperature With the Incidence of Type 2 Diabetes: An Ecological Study

An ecological study

Several hypotheses exist on a possible association between air temperature (AT) and the risk of type 2 diabetes (T2D), e.g., due to the influence of AT on health-related behaviors (1). One study from the USA estimated that per 1° C increase in annual average AT, the age-adjusted incidence of T2D increases by an absolute difference of 0.31 (95% confidence interval [0.19; 0.43]) (2). To date, only one study from Germany has investigated this issue (3). Hence, this study set out to estimate the sex-specific association between AT and T2D incidence at the level of the 401 rural and urban districts in Germany.

Methods

To investigate the long-term association between annual mean AT and the incidence of T2D, we conducted a longitudinal ecological study using district-level statutory health insurance (SHI) outpatient claims data between 2014 and 2019. In each study year, the data included all persons with SHI (approximately 85% of the German population) with at least one outpatient physician contact during that year. To determine the population at risk of T2D, persons in the database were excluded if they had already been diagnosed with T2D within the previous 3 years. The data were aggregated by age and sex with a minimum of 30 cases in each age and sex stratum. At the level of the 401 districts, this enabled stratification by sex but not by age. However, the number of individuals at risk for incident T2D could be stratified by 5-year age group and sex, which enabled us to control for differences in the age distribution among districts by means of indirect age standardization.

The annual number of incident cases of T2D in the 401 districts between 2014 and 2019 was determined on the basis of ICD-10 codes following the procedure described by Goffrier et al. (4). The annual average AT at district level was based on publicly available data from the German National Meteorological Service (Deutscher Wetterdienst, DWD, https://www.dwd.de/DE/klimaumwelt/cdc/cdc_node.html). The main determinant of a person’s experienced AT is their place of residence. Hence, variables that are associated with T2D incidence and that influence AT at the place of residence, and/or the place of residence itself, may confound the association between AT and the incidence of T2D. To adjust for this possible bias, we thus considered the following district-level variables as potential confounders: the German Index of Socioeconomic Deprivation (GISD) score (5), the extent of soil sealing, and the population density (both available in German from the Leibniz Institute of Ecological Urban and Regional Development at https://www.ioer-monitor.de/).

We estimated the association between AT and T2D incidence using a mixed negative binomial regression model with the observed number of incident T2D cases at district level as the dependent variable and a random effect for the districts. To adjust for age, the logarithm of the expected number of cases was included as an offset variable in the negative binomial regression, which is equivalent to indirect standardization and enables estimation of the standardized incidence ratio (SIR). Annual average AT, GISD score, soil sealing, population density, and calendar year were considered as independent continuous variables. These continuous variables were modeled as natural cubic splines if this improved the model fit according to the Bayesian information criterion (this was the case only for calendar year). We also included an interaction term between AT and sex to allow for sex-specific associations.

Results

Between 2014 and 2019, the annual number of incident T2D cases ranged between 443 384 and 479 993 among the approximately 63 million persons at risk each year. The number of incident cases in the 401 districts ranged between 158 and 22 334 (Table). The Figure shows the incidence rate ratios per 1° C increase in AT for the total population as well as for men and women separately. Overall, a 1° C increase in AT was associated with a 0.6% [–0.7%; 1.9%] increase in the incidence rate. The association was greater in men than in women. Considering the approximately 6°C range of observed annual average AT in Germany, this corresponds to an increase in incidence of 3.6% comparing the highest with the lowest observed AT. Given the mean incidence of 7.91 per 1000 in the 401 districts in the year 2014, a relative increase in incidence of 1% corresponds to an absolute increase of 0.079 per 1,000.

Figure

Incidence rate ratios (IRR) with 95% confidence intervals (CI) for the association between annual average air temperature in °C and the incidence of type 2 diabetes in Germany between 2014 and 2019.

Table

Description of the 401 rural and urban districts in Germany

Discussion

In this ecological study we found a small association between T2D incidence and annual AT: a 0.6% rise per 1°C increase. The single previous German study found comparable but somewhat smaller associations (3). In contrast, a study from the USA found a larger association, possibly explained by different population characteristics (2). In summary, our study provides no compelling evidence that AT is an important driver of T2D incidence at the population level.

Thaddäus Tönnies, Ramona Hering, Mandy Schulz, Oliver Kuss

Institute for Biometrics and Epidemiology, German Diabetes Center (DDZ), Leibniz Center for Diabetes Research at the Heinrich Heine University, Düsseldorf, Germany (Tönnies, Kuss); [e-mail]

Chair for Medical Biometry and Epidemiology, Witten/Herdecke University, Faculty of Health/School of Medicine, Witten, Germany (Tönnies)

Central Research Institute for Ambulatory Healthcare in Germany (Zi), Department of Healthcare Analyses, Berlin, Germany (Hering, Schulz)

Heinrich-Heine-University, Medical Faculty and University Hospital, Centre for Health and Society, Düsseldorf, Germany (Kuss)

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

Submitted on 13 March 2024, revised version accepted on 19 June 2024

Cite this as
Tönnies T, Hering R, Schulz M, Kuss O:
The association of mean air temperature with the incidence of
Type 2 diabetes—an ecological study. Dtsch Arztebl Int 2024; 121: 781–2. DOI: 10.3238/arztebl.m2024.0134

References

1. Edmondson D, Conroy D, Romero-Canyas R, Tanenbaum M, Czajkowski S: Climate change, behavior change and health: a multidisciplinary, translational and multilevel perspective. Transl Behav Med 2022; 12: 503–15.

2. Blauw LL, Aziz NA, Tannemaat MR, et al.: Diabetes incidence and glucose intolerance prevalence increase with higher outdoor temperature. BMJ Open Diabetes Res Care 2017; 5: e000317.

3. Badpa M, Schneider A, Schwettmann L, Thorand B, Wolf K, Peters A: Air pollution, traffic noise, greenness, and temperature and the risk of incident type 2 diabetes: results from the KORA cohort study. Environ Epidemiol 2024; 8: e302.

4. Goffrier B, Schulz M, Bätzing-Feigenbaum J: Administrative Prävalenzen und Inzidenzen des Diabetes mellitus von 2009 bis 2015. Versorgungsatlas-Bericht 2017; 17/03: 10.20364/VA-17.03.

5. Michalski N, Reis M, Tetzlaff F, et al.: German Index of Socioeconomic Deprivation (GISD): revision, update and applications. J Health Monit 2022; 7: 2–23.