gms | German Medical Science

26th Annual Meeting of the German Drug Utilisation Research Group (GAA)

Gesellschaft für Arzneimittelanwendungsforschung und Arzneimittelepidemiologie

21.11. - 22.11.2019, Bonn/Bad Godesberg

Multimorbidity and polypharmacy for diabetes patients using regional secondary data of the Statutory Health Insurance

Meeting Abstract

Gesellschaft für Arzneimittelanwendungsforschung und Arzneimittelepidemiologie e.V. (GAA). 26. Jahrestagung der Gesellschaft für Arzneimittelanwendungsforschung und Arzneimittelepidemiologie. Bonn/Bad Godesberg, 21.-22.11.2019. Düsseldorf: German Medical Science GMS Publishing House; 2019. Doc19gaa18

doi: 10.3205/19gaa18, urn:nbn:de:0183-19gaa187

Published: November 19, 2019

© 2019 Schuster et al.
This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 License. See license information at



Background: Diabetes mellitus is one of the most challenging public health problems throughout the world. The number of people with diabetes is increasing due to worldwide population growth and increasing prevalence of obesity and physical inactivity in more developed countries. A detailed analysis of the regional prevalence and interactions between diagnoses and drug treatment are important starting points for improvements in the healthcare system and of treatment strategies. The negotiations between regional Statutory Health Insurance Funds and the corresponding Associations of Statutory Health Insurance Physicians had to address treatment quality on the one hand and drug economic aspects on the other. New anti-diabetic drugs are also present in the evaluations of the Joint Federal Committee (GBA). Graph theoretic methods help to understand interactions and crosslinks.

Materials and methods: We use drug prescription data of the second quarter of 2018 of all statutory insured patients in Schleswig-Holstein as well as diagnostic data. This data pool contains about 1.7 million patients. Diabetes patients are determined by drug prescriptions with the codes A10A (insulins and analogues) and A10B (blood glucose lowering drugs, excl. insulins) drugs in the international ATC (Anatomic Therapeutic Chemical) system with German specifications provided by the German Institute of Medical Documentation and Information (DIMDI). There are around 208,000 patients with diabetes E10-E14 (Diabetes mellitus) or O24 (Diabetes mellitus in pregnancy) and 131,000 patients of 65 and older among them using the ICD-10 system (International Statistical Classification of Diseases and Related Health Problems 10th Revision).

Results: 42% of the patients are treated with cost up to 50 euros per quarter using glibenclamide (A10BB01), metformin (A10BA02), glimepiride (A10BB12), gliquidone (A10BB08) and gliclazide (A10BB09). Mean drug costs per quarter above 200 euros occur for empagliflozin (A10BK03), insulin aspart fast-acting (A10AB05), insulin lispro intermediate- or long-acting combined with fast-acting (A10AD04), insulin lispro fast acting (A10AB04), dulaglutide (A10BJ05), exenatide (A10BJ01) and liraglutide (A10BJ02). The inequality of cost can be described with a Lorentz curve, it has a Gini coefficient of 0.43. The mean values of multimorbidity and polypharmacy differ much between different antidiabetic drugs. Low scores of multimorbidity and polypharmacy go with A10BD15 (metformin and dapagliflozin) and A10AB05 (sulin aspart fast-acting), whereas the highest position in polypharmacy and multimorbidity is connected to A10AD01 (insulin human intermediate- or long-acting combined with fast-acting). Frequent comorbidities of diabetes patients are G63 (polyneuropathy in diseases classified elsewhere), E66 (obesity), E79 (disorders of purine and pyrimidine metabolism), I25 (chronic ischaemic heart disease) and E78 (disorders of lipoprotein metabolism and other lipidaemias). One can classify antidiabetic drugs with respect to diagnoses using a graph clustering method and an integer linear program (ILP) which is NP-hard (not polynomial time for solving ). One of five clusters: Patients of cluster 1 have the smallest mean value of multimorbidity, another cluster has an increased incidence of code F32 (depressive episode).

Conclusion: Community structures of graphs give new insights in therapeutic backgrounds and offer the possibility for improved health care decisions at the negotiation level. Further on it leads to improved medical decisions from a patient centered point of view. The drug economic MRG system (Mordbidity Related Groups) used in Schleswig-Holstein is helpful to identify comorbidities by looking at their aggregated costs at ATC four digit level: in descending order we get B01A (antitrombotic agents), V04C (other diagnostic agents, here tests for diabetes), C09D (angiotensin II receptor blockers, combinations), R03A (adrenergics, inhalants), N02A (opioids), C10B (lipid modifying agents, combinations), L04A (immunosuppressants), C01E (other cardic preparations), J05A (direct acting antivirals), L01X (other antineoplastic agents), R03B (other drugs for obstructive airway diseases, inhalants) and S01E (antiglaucoma preparations and miotics).


International Diabetes Federation. IDF Diabetes Atlas. 8th edition. 2017. Available from: External link
Shi Y, Hu FB. The global implications of diabetes and cancer. Lancet. 2014 Jun;383(9933):1947-8. DOI: 10.1016/S0140-6736(14)60886-2 External link
World Health Organization. Global Report on Diabetes. 2016. Available from: External link
Songer TJ, Ettaro L; Economics of Diabetes Project Panel. Studies on the Costs of Diabetes. Atlanta, GA: Centers for Disease Control and Prevention; 1998.
American Diabetes Association. Economic consequences of diabetes mellitus in the US in 1997. Diabetes Care. 1998;21:296-309.
Klein BE. Overview of epidemiologic studies of diabetic retinopathy. Ophthalmic Epidemiology. 2007;14 (4):179-83.
Agarwal D, Kempe D. Modularity-Maximizing Graph Communities via Mathematical Programming. EPJ. 2019. (manuscript).
Charikar M, Guruswami V, Wirth A. Clustering with qualitative information. Journal of Computer and System Sciences. 2005;71(3):360-83.
Schuster R, Emcke T. Graph-theoretic network analysis of interactions between patients, physicians and prescribed drugs. Archives of Data Science, Series A. (submitted).
Schuster R, Ostermann T, Emcke T. Modeling (Multi-)Morbidity and (Poly-)Pharmacy in Outpatient Treatment with Gamma Distributions. BIOSTEC. 2019;5:353-8.
Shinano Y, Fujie T, Kounoike Y. Effectiveness of Parallelizing the ILOG-CPLEX Mixed Integer Optimizer in the PUBB2 Framework. In: European Conference on Parallel Processing 2003. p. 451-60.
Agarwal G, Kempe D. Modularity-Maximizing Graph Communities via Mathematical Programming. The European Physical Journal B. 2008;66(3):409-18.
Newman M. Finding community structure in networks using the eigenvectors of matrices. Physical Review E. 2006 Sep 11;74. DOI: 10.1103/PhysRevE.74.036104 External link
Schuster R, Emcke T, von Arnstedt E, Heidbreder M. Morbidity Related Groups (MRG) for Epidemiological Analysis in Outpatient Treatment. Stud Health Technol Inform. 2016;228:783-7.
Schuster R, Ostermann T, Heidbreder M, Emcke T. Relations of Morbidity Related Groups (MRG), ICD-10 codes and age and gender structure in outpatient treatment. In: Proceedings of the 11th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2018) – Volume 5: HealthInf. 2018. p. 322-8.