gms | German Medical Science

10. Jahrestagung der GAA Gesellschaft für Arzneimittelforschung und Arzneimittelepidemiologie

16. bis 17.10.2003, Bonn

Is pharmacological data useful for the morbidity-adjustment in the context of integrated and sector-overlapping remuneration models?

Nutzung von Arzneimitteldaten zur Patienten-Risikoklassifizierung

Meeting Abstract

Suche in Medline nach

  • corresponding author Thomas Staffeldt - BKK Bundesverband, Kronprinzenstraße 6, 45128 Essen, Tel.: +49/201/179-1228, Fax: +49/201/179-1016
  • M. Nolting - BKK Bundesverband, Kronprinzenstraße 6, 45128 Essen

Gesellschaft für Arzneimittelanwendungsforschung u. Arzneimittelepidemiologie (GAA) e.V.. 10. Jahrestagung der Gesellschaft für Arzneimittelforschung und Arzneimittelepidemiologie (GAA) e.V.. Bonn, 16.-17.10.2003. Düsseldorf, Köln: German Medical Science; 2003. Doc03gaa05

Die elektronische Version dieses Artikels ist vollständig und ist verfügbar unter:

Veröffentlicht: 16. Oktober 2003

© 2003 Staffeldt et al.
Dieser Artikel ist ein Open Access-Artikel und steht unter den Creative Commons Lizenzbedingungen ( Er darf vervielfältigt, verbreitet und öffentlich zugänglich gemacht werden, vorausgesetzt dass Autor und Quelle genannt werden.



Background and Aim

Issues of morbidity and risk adjustment are increasingly important in the health system's discussion in Germany, regarding both financing (risk structure compensation) and remuneration. In particular, models of capitation-based managed care plans will require different reimbursement models to allow more rational distribution of resources. The aim of the study is to investigate the possible improvement of current remuneration models based on demographic data by further information derived from drug prescription data.

Material and Method

Data of hospital treatment, sickness benefits and drug prescriptions of approximately 400,000 German sickness fund members for 2 successive years has been used for the empirical analyses. The members were classified under 40 chronic disease groups according to their prescription patterns. With the help of regression models it was investigated if the inclusion of morbidity groups allows a more precise estimation of future (total) expenses of insured persons and of defined groups of insured persons than with the help of demographic-based models.


The predictive power of diagnostic- and demographic-based prediction models has been clearly increased by the morbidity groups. For the demographic models alone the R² results were only 4-6%. Additional information of pharmacological morbidity groups increased the R² values to 12-13%. Further inclusion of hospital information raised the R² values to 14-15%.

The advantage of the pharmacological morbidity groups is even more prominent with strongly selected collectives. The pharmacological based model applied to a selected group of diabetics showed an accurate prediction, measured by Predictive Ratio, whereas the demographic based model has been useless.

Analyses in detail on the 40 morbidity groups show that the validity of the separate groups is distinct. Especially strong predictive power can be seen by the groups of diabetics, coronary heart disease, cardiac failure, asthma, renal failure, alcoholism, epilepsy, as well as AIDS (HIV).


Pharmacological data has to be considered as a suitable morbidity indicator. Furthermore, it is comparatively quickly accessible and in contrast to billing-optimized diagnostic information from doctors, less manipulable. Especially the latter favours the use of pharmacological data in the context of sector-integrated capitation.

The depiction of chronic diseases by the means of pharmacological data is limited though. The morbidity groups have to be adjusted continuously to the developments on the pharmacological market.