[Medical data in pathology--evaluation of a large collection. (530,000 diagnoses coded in SNOMED II)]

Abstract

The paper is describing the design and the performance of the computerized system, from its introduction in 1982 until the present day. The first device, using the MUMPS language on a mini-computer, followed by a VAX computer with terminals have been replaced in 1996 by an application program, using ORACLE, based on the client-server concept. The content and the particularities of the different data groups are discussed, concerning the functional components of the data bank: 'PATIENTS', 'SPECIMEN', 'SENDERS', 'REPORT' and 'DIAGNOSES'. By means of examples, we demonstrate the chronological evolution of the registration of persons, the distribution of the diagnoses according to the organ systems, the possibilities to combine various lesions and an algorithm to assure that a given lesion is registered only once per patient. In first place, the efficiency and the reliability of manual coding by a pathologist using the Systematized Nomenclature of Medicine (SNOMED; 2nd edition [1979/1982]) is discussed. The data bank currently contains 530,000 diagnoses, distributed among on SNOMED's five main modules, obtained from 1500 autopsies, 140,000 surgical and 180,000 cytological specimens. Concluding the article, an analysis is made of desirable developments in the future with the aim of a better integration of the acquired information in routine work and an enhanced use of the medical content for epidemiological research or statistical analysis.