At the VASCUNET Collaborators meeting in Helsinki, experts from across European discussed the proposal for a European wide collaboration in vascular surgery. Dr Peter Walton, Dendrite Ltd, UK, who in co-operation with Professor Bruce Keogh, and on behalf of the European Association of Cardio-Thoracic Surgeons (EACTS), has produced the Second European Adult Cardiac Surgical Database. This registry aims to produce measurable procedural (and procedural subset data) for cardiac surgeons across Europe.
Walton began the meeting by highlighting the lessons learned from cardiac surgeons. He said that current practice for data collection benefited from low cost-low risk, national registries. Walton argued that compared with collecting data from hospitals, national registries establish clear lines of authority, are nationally owned, result in clearer data which can then be reproduced on a usable model. He stated that the problem with hospital data collection encompasses issues such as quality, ownership, unclear authority, difficulties in establishing and maintaining effective communication between institutions.
Following Walton, Mr Chris Gibbons, Morriston Hospital Swansea, UK, discussed the difficulties in established a European wide collaboration. He said that several problems were apparent, not least because of different collection methods throughout Europe. However, Gibbons suggested that a feasibility study, examining a single operation over a finite length of time should be established. He proposed that aortic abdominal aneurysms (AAA) was perhaps the ‘best’ procedure to begin with. He proposed looking at AAA over a two year period, using datasets common to existing registries, under which mortality, morbidity and practices could be compared. Gibbons said the confidential data would be owned by the ESVS/VASCUNET and any publication would be by VASCUNET as a whole. As with cardiac surgeons, individual surgeons would not be named and only national figures would be produced. Financial backing would be required and various avenues were being explored.
Dr Juerg Schmidli, Head of Vascular Surgery, Department of Cardiovascular Surgery, University Hospital, Berne, Switzerland, then discussed the merits of from the Swissvasc Registry. This registry gathers as many vascular procedures as possible which are performed in Switzerland on a voluntary basis (except for university and other teaching hospitals). Schmidli said the Registry is a valuable tool of quality control, could provide a benchmark for surgeons and institutions alike and can educate vascular surgeons. He reminded the audience of the words of Michael Horrocks, President of the Vascular Society of Great Britain and Ireland: “I believe that voluntary submission of accurate and complete data by surgeons themselves is the best form of defence for the profession, and such data should be regarded as a shield rather than any kind of threat.”
Next, Dr Leif Panduro Jensen, Chief Vascular Surgeon, Department for Vascular Surgery, Gentofte University Hospital, Copenhagen, Denmark, discussed the use of data-mining, from the Danish Vascular Registry. Jensen said that the Registry reported results to medical professionals and allowed a comparison (benchmarking) between professionals and institutions. In addition, the Registry serves as a basis for audits, assessing the quality and the risks for the patient undergoing treatment. He said that the publication of crude rate and case-mix correction will not correct errors in the data, indeed, it might even increase the significance of bias. However, such rates are of interest when under discussion among peers (e.g. very high crude rate mortality may be caused by bad selection of patients). He said that the causes of bias include incomplete registration of cases and the fact that parameters are not defined equally in reporting centers, making it difficult to compare results.
Jensen explained that data-mining is a method where statistics is used to recognize patterns in selected variables on an output variable, the purpose of which is to simplify data before doing further analysis. He concluded by stating that case-mix correction is needed, but crude rates contain additional information for discussion. “Data-mining seems to be an easy and effective tool for making case-mix correction. Public presentation of our data is a demand from society, so development of an model for case-mix correction is of urgent importance,” Jensen concluded.