Skip to main content
Key Specialties
Cardiology Diabetology Endocrinology Gastroenterology Geriatrics and Gerontology Gynecology and Obstetrics Hematology Infectious Disease Internal Medicine Nephrology Neurology Oncology Pediatrics Respiratory Medicine Rheumatology Surgery
Congress Coverage
2024 ACC Congress 2023 ESMO Congress 2023 EASD Congress 2023 ERS Congress 2023 ESC Congress
Clinical Collections
Advances in Alzheimer’s

At a glance: The STEP trials

A round-up of the STEP phase 3 clinical trials evaluating semaglutide for weight loss in people with overweight or obesity.
ADVANCED SEARCH
Log in
Top
The European Journal of Health Economics
Published in: The European Journal of Health Economics 2/2012

Open Access 01-04-2012 | Original Paper

Defining care products to finance health care in the Netherlands

Authors: Machiel Westerdijk, Joost Zuurbier, Martijn Ludwig, Sarah Prins

Published in: The European Journal of Health Economics | Issue 2/2012

Login to get access

Abstract

A case-mix project started in the Netherlands with the primary goal to define a complete set of health care products for hospitals. The definition of the product structure was completed 4 years later. The results are currently being used for billing purposes. This paper focuses on the methodology and techniques that were developed and applied in order to define the casemix product structure. The central research question was how to develop a manageable product structure, i.e., a limited set of hospital products, with acceptable cost homogeneity. For this purpose, a data warehouse with approximately 1.5 million patient records from 27 hospitals was build up over a period of 3 years. The data associated with each patient consist of a large number of a priori independent parameters describing the resource utilization in different stages of the treatment process, e.g., activities in the operating theatre, the lab and the radiology department. Because of the complexity of the database, it was necessary to apply advanced data analysis techniques. The full analyses process that starts from the database and ends up with a product definition consists of four basic analyses steps. Each of these steps has revealed interesting insights. This paper describes each step in some detail and presents the major results of each step. The result consists of 687 product groups for 24 medical specialties used for billing purposes.
Footnotes
1
In Dutch `Diagnose Behandeling Combinatie’.
 
2
See for example, the yearly PCSE (patient classification systems europe) conferences and their proceedings.
 
3
College Tarieven Gezondheidszorg: The Dutch Healthcare Tariffs Organization.
 
4
The DBC data obtained from the hospitals went through data pre-processing activities and error correcting filters (consistency checks, check against registration guidelines, outlier filtering). This process is too complex to meaningfully describe for the purposes of this paper.
 
5
Note that we use a unique cost price per activity code, namely the median cost price over the hospitals (see Sect. 2.3). Differences in costs between two treatments are therefore caused by differences in which activities are performed, the number of activities and their median cost price.
 
6
More precisely, the Jaccard similarity between two vectors \( \vec{x} \) and \( \vec{z} \) is defined as \( J\left( {\vec{z},\vec{x}} \right) = \vec{z} \cdot \vec{x}/\left( {\vec{z} \cdot \vec{x} + \left\| {\vec{z} - \vec{x}} \right\|^{2} } \right) \). For DBCs a vector would represent its activity sequence, e.g. if there where N distinct activities \( \vec{x} = \left( {1,1,0,0, \ldots ,0,1} \right) \) would represent a DBC for which activity 1, 2 and N where performed. Instead of a binary (yes/no) representation, we can also represent the number of times the same activity was performed or the costs involved with the activity.
 
7
The (decision tree) algorithm that is used here is closely related to well-known algorithms in statistics such as CART [18] and C4.5 [20].
 
8
More precisely, the criterion used here is `minimum weighted average CV’, where the weight is given by the number of patients in each branch.
 
9
We do this by using a `pruning algorithm’, where we first fully develop the tree with all its detail and then prune those branches which do not result in sufficient decrease of the average CV value. To be more precise, the homogeneity of the episodes in a node is compared to the average homogeneity of all end nodes (leafs) of the whole subtree under the node. If the sub-tree does not significantly increase homogeneity, the whole subtree is pruned. Here, we also take into account the statistical uncertainty of the computed homogeneity in each node.
 
10
Cost homogeneity is defined using the coefficient of variance (CV), see for example Fischer [19]. The CV of the costs of a group of patients is defined as the standard deviation of the costs of in the group divided by the average costs. The lower the CV, the higher the cost homogeneity of the group. This measure of homogeneity has been widely used in evaluating casemix systems (e.g. Fischer [19]). In most research, a CV smaller than 1 is being accepted as reasonable. Because this criterion is based on DRG systems, which only involve clinical costs whereas this research uses clinical as well as outpatient costs, we will not compare our results with this standard. Here, the goal is simply to end up with an average CV as small as possible, preferably smaller than 1.
 
11
Note that the example treatment codes presented at the start of the paper in Table 1 are those after the refinement step.
 
12
We used a variant of a hierarchical agglomerative clustering algorithm, see for example Duda and Hart [17].
 
13
More precisely, the similarity between DBC codes is again computed using the Jaccard measure. Now, the Jaccard measure is used to compute the similarities between DBC codes based on their distribution over clinical pathways. If we wish to compute the similarity between two DBC codes by comparing their distributions z and x over clinical pathways using the Jaccard measure we get \( J_{W} \left( {\vec{z},\vec{x}} \right) = \vec{z}^{T} W\vec{x}/\left( {\vec{z}^{T} W\vec{x} + \left( {\vec{z} - \vec{x}} \right)^{T} W\left( {\vec{z} - \vec{x}} \right)} \right) \)where the components of the distribution vectors \( \vec{z} \) and \( \vec{x} \), for example\( z_{i} \), represent the frequency (a fraction between 0 and 1) with which the DBC code is observed in clinical pathway i. The matrix W is used to account for similarities in the clinical pathways themselves\( W_{ij} = \vec{v}_{i} \cdot \vec{v}_{j} \)where \( \vec{v}_{i} \) is the vector representing the average CTG activity profile of a clinical pathway, i.e., each component of \( \vec{v}_{i} \) corresponds to a CTG activity and its value is equal to the average (normalized) costs of that activity in pathway v.
 
Literature
1.
Fetter, R.B., Shin, Y., Freeman, J.L., Averill, R.F., Thompson, J.D.: Case mix definition by diagnosis-related groups. Med Care 18(Supplement) (1980), February
2.
Welvaarts, R., Ludwig, M., Hofdijk. J.: Issues in comparing the DBC and IR DRG patient classification systems. In: 19th Annual PCS/E Conference, Washington, DC, 8–11 Oct 2003
3.
Schreyögg, J., Stargadt, T., Tiemann, O., Busse, R.: Methods to determine reimbursement rates for diagnosis related groups (DRG); a comparison of nine European countries. Health Care Manag. Sci. 9, 215–223 (2006)PubMedCrossRef
4.
Steinbusch, P.J.M., Oostenbrink, J., Zuurbier, J., Schaepkens, F.J.M.: The risk of upcoding in casemix systems: a comparative study. Health Policy 81, 289–299 (2007)PubMedCrossRef
5.
Zuurbier, J.J.: Episode based hospital reimbursement based on cost accounting: mitigating financial risk. In: Proceedings of the 33th Annual Congress European Accounting Association, Istanbul, 2010
6.
Coombs, R.: Accounting for the control of doctors: management information systems in hospitals. Acc. Organ. Soc. 12, 389–404 (1987)CrossRef
7.
Jackson, T.: Using computerized patient-level costing data for setting DRG weights: the Victorian (Australian) cost weight studies. Health Policy 56, 149–163 (2001)PubMedCrossRef
8.
Beaver, C., Zhao, Y., McDermid, S., Hindle, D.: Casemix-based funding of northern territory public hospitals: adjusting for severity and socio-economic variations. Health Econ. 7, 53–61 (1998)PubMedCrossRef
9.
Drain, M., Godkin, L.: A portfolio approach to strategic hospital analysis: exposition and explanation. Health Care Manage. Rev. 21(4), 68–74 (1996)PubMed
10.
Lynk, W.J.: One DRG, one price? The effect of patient condition on price variation within DRGs and across hospitals. Int. J. Health Care Finance Econ. 1, 111–137 (2001)PubMedCrossRef
11.
Eldenburg, L., Kallapur, S.: Changes in hospital service mix and cost allocations in response to changes in medicare reimbursement schemes. J. Acc. Econ. 23, 31–51 (1997)CrossRef
12.
McCue, M.J., Clement, J.P., Luke, R.D.: Strategic hospital alliances: do the type and market structure of strategic hospital alliances matter? Med. Care 37(10), 1013–1022 (1999)PubMedCrossRef
13.
Eldenburg, L., Kallapur, S.: The effects of changes in cost allocations on the assessment of cost containment regulation in hospitals. J. Account. Public Policy 19, 97–112 (2000)CrossRef
14.
Reed, S.D., Friedman, J.Y., Gnanasakthy, A., Schulman, K.A.: Comparison of hospital costing methods in an economic evaluation of a multinational clinical trial. Int. J. Technol. Assess. Health Care 2, 396–406 (2003)CrossRef
15.
Younis, M.Z., Rivers, P.A., Fottler, M.D.: The impact of HMO and hospital competition on hospital costs. J. Health Care Finance 4, 60–74 (2005)
16.
Ross, T.K.: Analyzing health care operations using ABC. J. Health Care Finance 3, 1–20 (2004)
17.
Duda, R.O., Hart, P.E.: Pattern Classification and Scene Analysis. Wiley, NY (1973)
18.
Breiman, L., Friedman, J.H., Olsen, R.A., Stone, C.J. (eds.): Classification and Regression Trees. Wadsworth International Group (1984)
19.
Fischer, W.: Are DRG’s homogeneous with regard to resource consumption? In: 16th Annual PCS/E Conference, Washington, DC, Oct 2000
20.
Quinlan, R.: C4.5: Programs for Machine Learning. Morgan Kaufman Publishers Inc (1993)
Metadata
Title
Defining care products to finance health care in the Netherlands
Authors
Machiel Westerdijk
Joost Zuurbier
Martijn Ludwig
Sarah Prins
Publication date
01-04-2012
Publisher
Springer-Verlag
Published in
The European Journal of Health Economics / Issue 2/2012
Print ISSN: 1618-7598
Electronic ISSN: 1618-7601
DOI
https://doi.org/10.1007/s10198-011-0302-6

Other articles of this Issue 2/2012

The European Journal of Health Economics 2/2012 Go to the issue

Original Paper

Decision-making in general practice: the effect of financial incentives on the use of laboratory analyses

Original Paper

Exploring the disparities of regional health care expenditures in Switzerland: some empirical evidence

Original Paper

Malpractice liability, technology choice and negative defensive medicine

Editorial

Separating the wheat from the chaff

Original Paper

Treating pneumonia in critical care in the United Kingdom following failure of initial antibiotic: a cost-utility analysis comparing meropenem with piperacillin/tazobactam

Original Paper

Towards a social discount rate for the economic evaluation of health technologies in Germany: an exploratory analysis