The Jepson Lecture 1991. Clinical databases and surgical research.

Modern science is preoccupied with basic mechanisms at the level of the gene, molecule, atom and fundamental particle. This preoccupation has affected attitudes to medical science. Paradoxically, politicians and Departments of Health are largely concerned with clinical epiphenomena--outcomes, interventions, effectiveness, efficiency--that are often disregarded by medical scientists and viewed without favour by granting bodies. In the surgical sciences, there is renewed interest in clinical research. The small computer and the readily available database with its compatible statistical package have added a new dimension to clinical research. Database design and analysis are as much a part of the surgical investigator's skills as are laboratory techniques. There are simple rules that govern database design. The database should be simple and flexible, but provide an adequate patient profile. Entry should be standardized by precise inclusion criteria and precise definitions. Data input should be numerical whenever possible. The database program should convert simply to a comprehensive statistical program. The clinical database is useful for both observational and investigational studies. Case series, case control studies and cohort studies can all be developed from well maintained databases. In the Department of Surgery at Westmead Hospital, databases have been maintained for 10 years. The hepatobiliary and pancreatic group of databases have led to 34 publications. The liver tumour database has produced 12 studies and nearly 20 papers. The process of development of one study is outlined in detail. The chance observation of an excess incidence of gallstones in patients having regular ultrasound examinations after major abdominal surgery has been confirmed in a case control study.(ABSTRACT TRUNCATED AT 250 WORDS)