Excerpt of the presentation


A Framework for Medical Costanalysis in XML and Cocoon

Christian Elsner 1, Christian Mazzi 2, Gerhard Hindricks 1
1 Heart Center Leipzig, 2 US Click4Care, 1, Germany

Cost-Effectiveness Analysis becomes an important tool for Providers and Payers in establishing new therapeutic options for patients. Despite a huge amount of published Information on Evidence based Data on Cost-Effectiveness is available over Medical Databases (e.g. MedLine) there is a huge time gap in "time to provider" and "time to payer" of this information. Objective of our work was to design a concept and data structure for easy and semi-automated set up of Cost-Effectiveness Markovmodels for rhythmologic topics and evaluate those with a portion of available literature. The Tools for Concept and Data structure were built using a XML-Designtool from Altova Inc. and a Visual Basic Program. A first alpha-prototype of a markov node representation layer webfrontend with condensed reference information was built with a Tomcat / Cocoon 2.0 Framework. For the Evaluation of the Model we used Review / Original Literature from Carol et al. on Cost-Effectiveness of Ablation for Supraventricular Tachykardia and Literature from Eckman et al. and Weerasooriya et al. on Cost-Effectiveness of PAF Treatments. As a result of our Work we developed two XML-Schemes, one to represent EBM / Cost-Effectiveness relevant information (Demographic, Clinical, Costs, Utilities and Disutilities) from Publications ( = MedEc-XML-REF) and a second to map and weight this information to defined markov nodes (MedEc-XML-MAP). With the System 60 References could be mapped into the new Data structure. With the prototype in Tomcat / Cocoon 2.0 we could rebuild the models from the literature graphically. The System could then give a web-enabled fast overview of the underlying referenced information and Clinical Evidence of the Data per node (for the Evidence Rating we used a modified model from U.S. Preventive Services Task Force). At least new data could be added fast through the integration of XML-structured data from a number of (fictive) new publications for evaluation purposes and validation of the system. The developed data scheme seems to be feasible to extract relevant information from the publications. Mapping the Information from the references still needs sort of interpretation which needs to be commented in both the schemes. A further granulation of the XML schemes may therefore be necessary. The process of mapping reference information to the markov nodes is highly enhanced due to semi-automation by the use of common classification codes for procedures and outcomes. The principle of the graphical representation may be useful for consecutive literature search / investigations being performed for further evidence of Cost-Effectiveness at certain points.