International Journal of Clinical Monitoring and Computing 5: 207-215 (1988) ©Kluwer Academic Publishers, Dordrecht- Printed in the Netherlands Experimental identifcation of technical and database factors that can afect the success of clinical computer systems Daniel J. Essin Department of Pediatrics, Universit of Souther Califoria School of Medicine, Los Angeles, Califoria, USA Accepted 12 January 1988 Key word: computer software, database management systems, transaction processing, software tools, clinical information systems Summary The entry of clinical data into computer systems is an extremely demanding form of transaction processing. High speed is important, especially if the collection involves real-time data. Clinicians must feel that they intuitively understand a system and that it is responsive. Medical data must be easily accommodated without sacrificing accuracy or completeness. Most systems cannot do this. Clinical systems that involve on-line storage of data fom patients should employ data-base technology. Systems that lack any of the following capabilities will not succeed: manual data entry, a data dictionary, a file system, utility functions, ad hoc query, and a statistical report generator. Tese general capabilities must satisfy a number of specific functional requirements if the entire system is to be a success. A group of such requirements have been experimentally validated. These will be discussed and a more comprehensive list presented. Introduction Computerization in a critical care or general hospi tal setting involves more than the purchase of pre fabricated commercial systems. Some institutions have chosen to specialize in technologically inten sive areas such as cardiac surgery. Others are re search oriented and collect large amounts of data, both textual and real-time physiological, that ex ceed the minimum required to provide treatment. Many government and academic institutions are subject to logistic and regulatory constraints that are uncommon in the private sector. These special ized circumstances necessitate the development of customized systems either as extensions of a cen tralized hospital information system or independ ently. Although the need to computerize in order to perform a particular function may be evident, the need alone does not guarantee success. The num ber of medical systems that have failed in the past attests to the need for a quantitative understanding and definition of the components of successful sys tems. The characteristics that defne general suc cess are listed in Table 1. The purpose of the present study is to experi mentally identify those specific features of comput er database management systems that have a mate rial impact on general success and functionality. Critical care and laboratory applications that re quire custom real-time programming are especially dependent on factors that promote general success since most of the available resources are invariably directed at achieving the specific success of the project.