A population‐based cohort of adult patients with diabetes mellitus in a Western District of Austria: The Diabetes Landeck cohort

Abstract Introduction Diabetes mellitus (DM) has become an important and exacerbating health epidemic, with severe consequences for both patients and health systems. The National Diabetes Strategy of Austria addresses the lack of high‐quality data on DM in Austria and the need for developing a national data network. The aims of our study are to establish a cohort including all adult diabetes patients in a district in western Austria, describe the demographic and clinical characteristics of this cohort, and provide an estimation of diabetes prevalence. Methods We recruited a population‐based cohort of adult patients with a diagnosis of DM in cooperation with a network of all caregivers. Data collection was based on a case report form, including patient characteristics, clinical parameters and long‐term complications. Results In total, 1845 patients with DM were recruited and analysed. We observed an overall prevalence of 5.3% [95% CI: 5.0%–5.5%]. For the subsequent main analysis, we included 1755 patients with DM after excluding 90 patients with gestational DM. There were significant differences between genders in the distribution of specific clinical parameters, patient characteristics, and the long‐term complications diabetic foot, amputation and cardiovascular disease. Conclusion To the best of our knowledge, we established the first diabetes cohort study in Austria. Prevalence and the proportion of specific long‐term complications were lower when compared to the international context. We assume that the Diabetes Landeck Cohort has reached a high degree of completeness; however, we were not able to identify independent data sources for a valid check of completeness.


| INTRODUC TI ON
Diabetes mellitus (DM) has become a considerable and exacerbating health epidemic that has severe consequences for both the patients and health systems. 1 According to the tenth edition of the International Diabetes Federation (IDF) Atlas, approximately 537 million people aged 20 to 79 years suffered from DM worldwide in 2021, representing a prevalence of approximately 10.5%. 2 The global prevalence is expected to increase in the future, with 783 million people predicted to suffer from DM by 2045, corresponding to a prevalence of around 10.9%. 2 In Europe, one in eleven adults has DM, which means that 61 million people suffer from DM, and the prevalence is estimated to increase from 9.2% in 2021 to 10.4% in 2045. Furthermore, the economic burden associated with DM is substantial: The IDF reports that the share of the global health expenditure for DM ranges from 3.4% in the Middle East and North African region to 19.6% in Europe and to 43% in North America and the Caribbean. 2 Patients with DM suffer from a lifelong burden caused by the treatment and disease complications. One of the main challenges is an increased risk of developing long-term complications such as neuropathy, nephropathy and cardiovascular disease. 3,4 In addition, patients with DM have an increased risk of early mortality. Worldwide, about 6.7 million people aged between 20 and 79 years are estimated to die from DM or its complications in 2021. 2 For Austria, the number of patients with DM aged 20-79 years in 2021 was estimated at about 450,000, corresponding to a prevalence of 6.6%. In addition, the number of undiagnosed DM cases was estimated at approximately 150,000. 2 Due to the increasing number of patients with DM and the long-term complications, considerable DM-associated healthcare expenditures are also expected in Austria. The annual costs of DM and its comorbidities in Austria amount to an estimated 3 billion euros. 5 Therefore, the main targets of the diabetes strategy of the Austrian Federal Ministry of Labor, Social Affairs, Health and Consumer Protection from 2017 are to reduce the incidence of DM and to prevent long-term complications. 6 This Austrian strategy document states that there is a lack and necessity of valid data on DM in Austria. Moreover, a report of the Austrian Court of Audition on DM prevention and DM care from 2019 points to the lack of high-quality data. 7 To provide valid data on DM in Austria, a project called the Diabetes Landeck Cohort was initiated, with the aim to set up a population-based cohort of patients with DM in a district in western Austria. This cohort should serve as a basis to collect and analyse valid and comprehensive data on DM.
The aims of this project and study are to establish a cohort including all adult DM patients in a district in western Austria, describe the demographic and clinical characteristics of these patients, and provide an estimation of diabetes prevalence, both at an overall level and stratified by gender and age groups.

| Settings of the study and data collection
The recruitment of the cohort titled 'Diabetes Landeck Cohort'  Direct documentation by the treating physician was also technically possible but conducted only in individual cases due to the heavy workload of the physicians caused by the COVID-19 pandemic. All cohort data are stored in the pseudonymized web-based database software ASKIMED. 8 ASKIMED provides the possibility to store visits for each patient at different care units and allows the documentation by different persons and the mapping of the necessary access rights. A pseudonym was generated based on each patient's social security number (with a SHA-2 procedure), which makes it impossible, due to the current computing performance, to identify the specific patient based on the pseudonym. For research purposes, data were transferred to the statistical software Stata (Version 17). 9 In order to address aspects of data privacy, all patients had to sign a written consent form. For data protection reasons, the data of the very few patients who did not agree to sign the consent have not been included into the database for analysis. In addition, every care unit signed a contract stating the rights and obligations be-

| Patient characteristics
We collected the following patient characteristics: diagnosis, age, migration background, diagnosis site, year of diagnosis, diabetes duration, family history of coronary heart disease and family history of DM, participation in the Austrian disease management program, smoking status, participation in a diabetes education program and sufficient knowledge on diabetes according to the physician.
All diagnoses were clinically confirmed according to the criteria of the Austrian Diabetes Society (ADS) based on a fasting glucose or oral glucose tolerance test or haemoglobin A1c (HbA1c). 10 The study protocol did not include patients with prediabetes in the study population. Gestational diabetes (GDM) was documented separately.
Diagnosis of GDM was based on the HAPO criteria. 11 Age at the last visit was used for the analysis and was cut into categories, namely 20-49, 50-64, 65-74 and ≥75 years, taking into account the definition of geriatrics of the elderly/old person. The migration background was classified based on Schenk's approach and adapted to the Austrian situation. 12 During documentation, participants were asked whether their DM diagnosis had been made in the hospital or at a physician's office (diagnosis site). For most patients, the year of diagnosis was recalled from patients' memory, and we decided to collect the exact year only for patients diagnosed during the past 10 years. The diabetes duration was defined as the difference between the year of DM diagnosis and the year of last contact with a care unit and was analysed in the following groups: 0-4 years, 5-9 years, and 10 years or longer. We documented family history of coronary heart disease and family history of DM. Family included children, parents and/or siblings. Participation in the Austrian Disease Management Program during the study period 13 was also documented. Additionally, the smoking status at the time of diagnosis was recorded retrospectively in the categories 'active smoker' or 'ex-smoker' and 'never smoker'.
We assessed whether each patient had participated at least once in a diabetes education program and whether the patient had sufficient knowledge on DM according to the physician.
Each patient's life status was verified by a study nurse who inspected hospital/private care records and local newspapers in case there were no up-to-date visit records. Pseudonymization prevented linking records with official mortality data. It was therefore impossible to systematically check the patients' life status.

| Clinical parameters
Data collection included the following clinical parameters: body mass index (BMI), HbA1c, low-density lipoprotein (LDL), systolic and diastolic blood pressure, microalbumin, physical activity, eye and foot inspection, hypoglycaemia (requiring external help for recovery), diabetes-specific therapy and lipid therapy.
Mean values were computed for BMI, HbA1c, LDL, and systolic and diastolic blood pressure, which were collected at each visit. BMI was calculated according to the formula BMI = weight/height 2 (kg/ m 2 ), and the classification was based on WHO recommendations. In addition, obesity was defined as a BMI ≥30 kg/m 2 . 14 The mean values of HbA1c were categorized into four groups (<6.5%, 6.5%-7.49%, 7.5%-8.99% and >9%) based on the ADS guidelines. 10 Increased blood pressure was defined as systolic pressure ≥140 mmHg or diastolic pressure ≥90 mmHg according to WHO guidelines. 15 It should be noted that increased blood pressure is based on blood pressure measurements only. We did not collect information on the diagnosis of hypertension and/or medication.
LDL is the primary therapeutic target for lipid control in patients with DM. The LDL classification was based on current recommendations of the ESC/EAS. 16 We did not collect information on the diagnosis of hyperlipidaemia, but we surveyed the proportion of patients with well-controlled LDL levels. We recorded whether microalbumin was determined at least at one visit. Patients were asked if they were physically active (defined as at least moderate activity for less than two and a half hours per week). We also documented if an ophthalmologist had performed at least one eye inspection during visits. The inspections followed the recommendations of the ADS. 10 Furthermore, we recorded foot inspections during the study period.

| Long-term complications
We documented the following long-term complications: neuropathy, nephropathy, retinopathy, cardiovascular and cerebrovascular disease, diabetic foot ulcers and amputations based on the recommendations of the ADS. 10 Neuropathy is defined as nerve injuries due to DM and is confirmed with a positive monofilament test. Nephropathy requires positive albumin results at two subsequent visits. Retinopathy is diagnosed according to the guidelines provided by the Austrian Ophthalmologist Society. 17 Cardiovascular late complication was defined as myocardial infarction, bypass or percutaneous coronary intervention. Cerebrovascular late complication was defined as minor and major strokes, including transient ischaemic attacks.
We used the strict definition of the long-term complication diabetic foot according to the guidelines of the ADS 10 and therefore only collected information on the presence of diabetic foot ulcers.
For amputations, we documented any non-traumatic amputation due to diabetic foot ulcers. For all late complications, the year of the first occurrence was recorded if the diagnosis was made within the past 10 years. In all other cases, we only documented the diagnosis of the respective late complication without the year of occurrence. For the analysis we counted every long-term complication, not only long-term complications diagnosed in the study period.

| Statistical analysis
Patient characteristics are described as counts and percentages for categorical data. We present all results stratified by gender and age groups (Tables S1-S2 only). Fisher's exact test or the chi-squared test were used to test differences across gender and age groups.
Statistical significance was established as p < .05. Cases with GDM that did not result in a life-long type of DM (called 'GDM only') were described in a separate part because they differ in many clinical aspects. The main analysis of demographic and clinical parameters does not include cases with 'GDM only'. In case of missing data, we report data and percentages for non-missing values in a first step and present the number of cases with missing values in a second step; this procedure was adopted for every variable reported in the result tables.
For the computation of prevalence figures, we included all patients with DM detected in our study in the region of Landeck.
Population data for the region of Landeck per age group were obtained from Statistics Austria. 18 According to this population data, confidence intervals (95% CI) are provided for the overall cohort and also stratified by gender, subregions and age groups. We applied the concept of period prevalence, that is, we included patients with at least one visit at a care unit stay during the study period as a prevalent case.
All statistical analyses were performed using Stata Version 17. 9  Table 1. We only describe results with significant differences between female and male patients. Age distributions differed between genders. For example, in the age group ≥75, the proportion of women with DM was higher than for men. Among patients with DM, women had a longer diabetes duration (55.9% ≥10 years vs.

| Study population/diabetes Landeck cohort
49.2% in men) and more often reported a family history of diabetes (38% in women vs. 31.7% in men). In men, we observed a substantially higher percentage of active smokers (14.1% in men vs. 7.2% in women) and ex-smokers compared to women (53.4% in men vs. 19.4% in women).
Gender differences were found for specific clinical parameters, such as BMI, HbA1c, LDL, determination of microalbumin, DM therapy and physical activity. Table 2 shows the comparison of these characteristics between genders. We observed more overweight subjects (i.e. BMI = 25.0-29.99) in men than women. The proportion of obesity (i.e. BMI≥30) was similar for both genders. Women had a significantly greater proportion of well-controlled HbA1c levels (HbA1c < 6.5) than men. The distribution of LDL showed a shift to higher values in men compared to women. Microalbumin was identified more frequently in men than in women, and more men were physically active (28.5% vs. 21.9%). We also observed differences in DM therapy. More women than men only adapted their lifestyle (22.3% vs. 16.2%). In men, metformin was more frequently used than in women, for example, more men (22.9%) received oral antidiabetic drugs therapy AND insulin than women (18.5%). Further details on clinical parameters are shown in Table 2.

| Long-term complications
The most frequent long-term complication was cardiovascular disease (N = 332, 19.2%), followed by nephropathy (N = 313, 18.1%) and neuropathy (N = 223, 12.9%). We observed significant differences between women and men for diabetic foot, amputation and cardiovascular disease. More men than women suffered from these three long-term complications ( Figure 1). Further details for the total population and genders are presented in Table 3.
In our study, we observed 90 cases with GDM that did not develop T2DM. A mean value of HbA1c < 6 was documented for 94% of patients with GDM. Only three patients with GDM showed a HbA1c ≥ 6.5. Lifestyle adaptation was sufficient in 82.5% of patients with GDM, and insulin therapy was necessary for 13.4%.
Briefly, we observed differences in the following patient characteristics: diabetes duration, migration background and smoking status.
We also found a difference in the participation in education programmes and sufficient diabetes knowledge, but the percentage was still very high across all age groups. For the clinical parameters stratified by age group, we observed differences in the BMI, LDL, microalbumin, diabetes therapy, physical activity, foot inspection, and for most of the long-term complications. Further details on patient characteristics and clinical parameters are described in Tables S1-S2.

| DISCUSS ION
We established, to the best of our knowledge, the first populationbased cohort of patients with DM in Austria following the recommendations of the National Diabetes Strategy. Our cohort should support data-driven healthcare decision-making and should contribute improving outcomes of patients with DM.
In total, we analysed 1845 adult cases with DM (including 'GDM only') and observed an overall prevalence of 5.3%, with no statistically significant difference between genders but substantial differences between subregions. In addition, we identified differences between women and men in the following patient characteristics: Ministry of Health showed that mortality due to DM in western Austria is lower than in eastern Austria. 19 There is also an evident east-west variation in cardiovascular mortality among persons over 64. Overweight/obesity, which is a major risk factor for DM, 20 is an even more significant problem in the population over the age of 64 in the east of Austria than in the west. 19 However, these estimates are not standardized, for example, for age, sociodemographic characteristics or proportion living in urban areas.
Our prevalence estimation for the district of Landeck is in line with the reported prevalence of 4% in Switzerland, 21

which borders
Landeck. Germany also shows a southwest-to-northeast gradient.
The regional standardized prevalence was highest in the east, with 12.0% (95% CI: 10.3%-13.7%), and lowest in the south, with 5.8% (95% CI: 4.9%-6.7%). 22  In the following, we compare the parameters obesity, glycaemic control, smoking status, foot inspections and participation in a diabetes education program with data from the Scottish diabetes registry, 23 the Swedish registry, 24 and with results from the DAWN2 study. 25 We also compare diabetes therapies with Austrian data 26 and with data from a diabetes surveillance system for Germany at the Robert Koch Institute. 22 In our study, 42.2% of women and 40.1% of men were obese. In Scotland, the proportion of obese patients was at 55%. 23 In Sweden, obesity was also more frequent, with a prevalence of 61% for women  Abbreviation: 95% CI, 95% confidence interval. and 54% for men, although data from Sweden are only available for patients in the age group 30-60. 24 Overall, the proportion of obese patients in our study is lower compared to global numbers and the Austrian average, 27 which fits the east-west gradient in Austria mentioned above.
Concerning glycaemic control, the proportion of patients with HbA1c < 7.5% was 55.4% in Scotland 23 and 72.3% in our study (75.2% in women and 69.9% in men). One reason could be that the population in this western part of Austria is less obese and more physically active. Even overweight patients perform substantial physical activity. In addition, diabetic care is rather well structured.
The participation in a diabetes education program was 78.7% in the DAWN2 study, 25 which is nearly identical to our results of 78.2%. The DAWN2 study reported 62.5% foot inspections (vs. 79.7% in the Diabetes Landeck Cohort); however, the DAWN2 data were based on self-assessments. In the Diabetes Landeck Cohort, the proportion of active smokers was 10.9% (two times as many men than women), which is lower compared to 14.3% of the study of Panisch et al. 28 and to 17.7% in Scotland. 23 Our findings on diabetes therapy data do not describe the patients' current therapy (at the last contact) but rather summarize all therapy modalities that were documented during the study period.
The proportion of patients with DM who did not need diabetesspecific therapy (lifestyle adaptation only) is relatively high (19%) but is similar to the Swedish registry. 24 The percentage of DM patients treated with OADs corresponds to 72% in the publication of Engler et al., 26 which describes patients in the diabetes register in Tyrol.
Among 45-to 79-year-old people with T2DM in Germany, 29.6% of women and 37.2% of men received metformin monotherapy in 2010. 22 This percentage is higher in our study (38.9% in women and 42.1% in men). Also, the percentage for OAD+ insulin is higher in our cohort, which included all patients with DM compared to the study across Germany that only assessed patients with T2DM. 22 Comparisons of diabetes-specific therapies with the literature should be carefully interpreted as many studies exclude elderly patients. In contrast, in our cohort, all patients with DM were registered, and the proportion of patients aged ≥75 was 36%.
When comparing our results with published data, it is essential to consider that the frequency of the respective long-term compli-

| CON CLUS IONS
As explicitly stated in Austria's National Strategy Report on diabetes, there is a lack of high-quality data on DM in the country. 6 There are, to the best of our knowledge, no systematic population-based data on patients with DM in Austria. The Diabetes Landeck Cohort closes this gap for one region in Austria and provides, for the first time in Austria, a nearly complete set of patients with DM living in a welldefined region. Some results, such as the diabetes prevalence or the frequency of some long-term complications, are lower compared to international data. We succeeded in establishing a population-based cohort and related database; however, we were not able to identify independent sources to verify our results. Therefore, for the future, we strongly suggest evaluating both completeness and comparability of data with well-accepted methods. In general, documentation by study nurses who should ideally be located in the care units is recommended to obtain valid data, because many important data are not stored in the practice systems or cannot be accessed in a systematic way. To access patients with specific diagnoses, support for the coding of diagnoses by physicians in private practices should be developed and applied in the practice systems. To make the best use of already existing data in the Austrian healthcare system, we recommend developing and/or optimizing systems to link different databases (e.g. civil registration and death data). The Diabetes Landeck Cohort should allow to evaluate and improve the quality of care of patients with DM in the future. In general, the cohort should be optimized and updated because high-quality data provide an essential basis to optimize the care of patients with DM. The data could also be used to supplement a biobank, for long-term monitoring of diabetes patients, for questions in health services research and healthcare economics, and for the investigation of new electronic communication methods between physicians and patients. Further research is needed, and in a subsequent step, we will extend this study by carefully taking the limitations into account.

ACK N OWLED G EM ENTS
We would like to thank all participating general practitioners, diabetes specialists in private practices and the Hospital Zams for their collaboration.

FU N D I N G I N FO R M ATI O N
We thank the Tyrolean Health Fund (Tiroler Gesundheitsfond, TGF) for funding this project.

CO N FLI C T O F I NTE R E S T
The authors have no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from Tiroler Gesundheitsfond. Restrictions apply to the availability of these data, which were used under licence for this study.

E TH I C S A PPROVA L
The present study was approved by the Ethics Committee of the