To Register      SMDM Homepage

Sunday, 17 October 2004

This presentation is part of: Poster Session - Public Health; Methodological Advances

USING COMBINED PATIENT CHARACTERISTICS TO PREDICT DIRECT AND INDIRECT COSTS IN PARKINSON’S DISEASE

Uwe Siebert, MD, MPH, MSc1, Bernhard Bornschein, MD, MPH2, Annika Spottke, MD3, Karin Berger4, Wolfgang H. Oertel, MD5, and Richard C. Dodel, MD3. (1) Massachusetts General Hospital, Harvard Medical School, Institute for Technology Assessment and Department of Radiology, Boston, MA, (2) Ludwig Maximilians University Munich, Germany, Program on HTA and Decision Sciences, Institute of Medical Informatics, Biostatistics, and Epidemiology, Munich, Germany, (3) University of Bonn, Department of Neurology, Bonn, Germany, (4) MERG Medical Economics Research Group, Munich, Germany, (5) University of Marburg, Department of Neurology, Marburg, Germany

Purpose: To identify independent predictors of disease-specific drug costs, direct non-drug costs, and indirect costs in Parkinson’s disease.

Methods: Data from an ongoing prospective cost study of the German Competence Network for Parkinson Syndromes (n=152) were analyzed using multivariate regression. Potential predictors were sociodemographic factors, clinical variables from the Unified Parkinson’s Disease Rating Scale (UPDRS) including disease stage (Hoehn & Yahr classification) and quality-of-life parameters (EuroQoL [EQ-5D], Parkinson’s Disease Questionnaire 39 [PDQ-39]). Data for disease-specific drug costs and direct non-drug cost were log-transformed. Indirect costs were calculated using the human capital approach. Modeling of indirect costs proceeded without transformation in two steps: first, the probability of presence of indirect costs was predicted by logistic regression, second, estimation of costs was performed by linear regression in those with non-zero indirect costs.

Results: Predictors for disease-specific drug cost were age (p<=0.001), sex (p=0.001), UPDRS (p<0.0001), and quality of life (EQ-5D, p=0.02). The model for the prediction of other direct costs included disease stage (Hoehn&Yahr scale, p=0.05 and p<0.001) and PDQ-39 (p=0.03). The probability for the presence of indirect costs was dependent on age (p<0.001), UPDRS (p=0.03), PDQ-39 (p=0.04), presence of depression (p=0.02), and falls (p=0.006). The magnitude of indirect costs was a function of clinical state (p=0.003) and falls (p=0.007). Variance explained by the models (adjusted R-square) ranged from 24% to 28%.

Conclusions: We identified UPDRS and quality-of-life as most important predictors of costs in Parkinson’s disease. Drug costs also depended on age and sex. However, these factors explained only about a fourth of the total variance in costs.


See more of Poster Session - Public Health; Methodological Advances
See more of The 26th Annual Meeting of the Society for Medical Decision Making (October 17-20, 2004)