Basis of data control for anatomical pathology.

A principal function of the morbid anatomist is the generation of tissue diagnoses and the recording of them. This activity requires skill and produces large amounts of data. At the rate of 600 necropsies per year, and 20 diagnoses per necropsy, a single department may generate over 12,000 morphological terms in a year's time. These terms are of considerable importance; they express the frequency of various disease states, they reveal the relationships among different lesions, and, as the record continues, they disclose the changing incidence of human illnesses. It is, therefore, important to medicine and to the welfare of communities and perhaps nations, that these data be recorded and stored in such a way that retrievability is enhanced and analyses are possible. As the record enlarges, and the usefulness of it increases, the need for generating a retrievable account becomes imperative, and the responsibility for satisfying this need rests squarely on the shoulders of the morbid anatomists. While the anatomical record grows rapidly, the methods for retrieving those data have not been overlooked. There are simple methods and complex methods. There are code-by-hand methods (Systemized Nomenclature of Pathology, 1965), there are retrieve-by-machine methods (Systemized Nomenclature of Pathology, 1965; Carpenter, 1962), and there are methods that both code and retrieve by machine alone (Smith and Melton, 1964; Lamson and Dimsdale, 1966; Paplanus, Shepard, and Zvargulis, 1969). This welter of methods, however, does not entirely solve the problem before pathologists. Between the mountain of records to be retrieved on the one side, and the welter of methods for accomplishing the retrieval process on the other, lies the bottleneck of translating the English words of the former into the code signals of the latter. The success of any system will depend on the ease with which this step is accomplished. The bottleneck of coding is depicted in Figure 1. It is the coding procedure that requires special attention. Coding begins with a definition of what the retrieval process is expected to accomplish. The retrieval process should be expected to accomplish two main objectives: the first is the identification of any record that lists a particular diagnosis, and the second is the enumeration of the anatomical site involved and the pathological process independently of each other, and on hierarchical levels progressing from generality to specificity. Thus, enumerating the pathological involvement of anatomical systems regardless of the lesions, and on the other hand, enumerating pathological processes irrespective of their sites, may be easily undertaken. For example, the frequency of abnormalities of the central nervous system may be of interest regardless of what the abnormalities were. And in the same way, the incidence of pathological processes such as carcinoma in a necropsy population is a figure of interest irrespective of the sites at which they arose. The ability to scan each of these categories of information independently of the other is thus useful. The dual objectives of identifying specific diagnoses and of scanning processes and sites independently of each other, and on different hierarchical levels, thus provides for virtually unlimited control of the data. This is especially true when other aspects such as age, weight, race, sex, and organ weights are also included as independent variables.