[Analysis of the cause of death of ICD patients during long-term follow-up]. / Analyse der Todesursachen von ICD-Patienten in der Langzeitbeobachtung.

BACKGROUND: Implantable cardioverter defibrillators represent an effective therapy to prevent sudden cardiac death in patients with ventricular arrhythmias. This implies a change in the spectrum of causes of death. There exist no large studies providing an exact differentiation of these causes of death. METHODS: During a mean observation period of 41+/-29 months (3 d-12 yrs), we analyzed the outcome of 429 ICD patients (CAD n=274, dilative CMP n=97, others n=23, no structural heart disease n=35), mean age 62+/-12 years (23-87 yrs). RESULTS: The overall survival rate was 79.7% whereby 3.5% of these patients underwent successful heart transplantation and 2.1% died suddenly. 60.9% of all deaths showed a non-sudden cardiac cause (heart failure 56.3%, myocardial infarction 4.6%). In 28.7% a non-cardiac fatal event was present (cancer 11.5%, septicemia 6.9%, stroke 4.6%). The deceased showed significantly more often structural heart disease (100 vs 92%, p<0.02) and a lower left ventricular ejection fraction (37+/-14 vs 44+/-18%, p<0.02). Significant higher survival rates were associated with medication with beta blockers (88 vs 75%, p<0.02) and the class III antiarrhythmic agents sotalol und amiodarone (85 vs 77%, p<0.03). CONCLUSIONS: Mortality due to sudden death was reduced to 2.1% in ICD patients associated with a shift of causes of death towards severe heart failure. Therefore, special attention should be paid to the signs of heart failure in these patients to enable early initiation of appropriate therapeutic strategies.

Complexity of biomedical data models in cardiology: the Intranet-based AF registry.

Atrial fibrillation (AF) is the most common cardiac arrhythmia and is associated with major complications. Ongoing research is focused on new pacing devices for alternative treatment of this disease. The objective of an AF registry is to store prospectively all relevant data covering clinical information, quality of life and device parameters and by this means provide a platform for long-term follow-up. For statistical analysis, categorical and numerical items are required, thus a high-granular data structure must be defined and implemented in the clinical setting. Facing the limits of formalization, we developed an XML-based documentation scheme consisting of 619 items in eight tables and implemented it with state-of-the-art Intranet technology. At present detailed information on 88 patients is recorded. The pacing device generates per patient and follow-up visit a file consisting of approximately 400-500 parameters provided on a floppy disk, which are transferred by means of a specific interface into the database. Success factors for integration of a complex research database into the routine workflow of a busy university hospital are interfaces between data sources to enable non-redundant data entry, intensive fine tuning by iterative software engineering and benefit for the clinical users in form of clinical reports and patient-specific summaries. Data quality must be assured by plausibility checks. To get an overview of this complex dataset we developed a dedicated visualization tool. Due to the high number of items a large patient collective must be recruited for statistical evaluation. Interinstitutional cooperation is required for a consensus on common minimal documentation schemes to enable pooling of data.

XML-based visual data mining in medicine.

Medical databases in general are characterized by a high degree of complexity in terms of quantity of items, number of parameter values and data types (free text, categorical, numerical and other). Substantial domain knowledge is required for adequate formalization of medical entities. In this context we developed medical database plot (mdplot), a data mining tool to visualize both structure and quality of data in medical databases to identify items suitable for evaluation. Data models are provided in XML format. Missing data is identified to enable targeted efforts to improve data quality prior to analysis. Database items are classified as 1:1- related to the patient (i.e. variables are collected once per patient) and 1:n related. mdplot provides a list of all classes contained in a database, the number of records each and a condensed bar chart for semi-quantitative description of completeness according to four types of items: categorical, numerical, text and other. All items in a category are grouped from left to right, the height of each bar represents the proportion of non-missing values with respect to the total number of records in the class; thus the amount of content in a specific class is visualized. By selection of a specific class, a detailed description of it is provided including mean completeness in each item category as well as number of values per item. The new methodology was applied to a cardiological research database consisting of 619 items on 88 patients.