“Safeware”: Safety-Critical Computing and Health Care Information Technology

Information technology (IT) is highly promoted as a mechanism for advancing safety in health care. Ironically, little attention has been paid to the issues of safety in health care IT. Computer scientists have extensively studied the problem of assured performance in safety-critical computing systems. They have developed a conceptual approach and set of techniques for use in settings where incorrect or aberrant operation (or results from correct operation that are aberrant in context) might endanger users, the public, or the environment. However, these methods are not commonly used in health care IT, which generally has been developed without specific consideration of the special factors and unique requirements for safe operations. This article provides a brief introduction for health care professionals and informaticians to what has been called “safeware,” a comprehensive approach to hazard analysis, design, operation, and maintenance of both hardware and software systems. This approach considers the entire joint sociotechnical system (including its operators) over its entire lifecycle, from conception through operation and on to decommissioning. Adoption of safeware methods should enhance the trustworthiness of future health IT. Introduction Twenty-five years ago, Lissane Bainbridge coined the phrase “ironies of automation” to refer to the observation that introducing automation into a complex sociotechnical system to improve safety and performance often simultaneously introduced new problems into the system that degraded safety and performance. 2 Despite this experience, the belief that advanced information technology (IT) is a critical mechanism by which to improve the safety of health care is strongly held by academics, public officials, and vendor, business, and civic groups. 4, 5, 6, 7, 8, 9 The anticipated benefits of health care IT are presented in these discussions as a sort of manifest destiny—difficult, to be sure, but ultimately inevitable. While there have been many discussions about the challenges, costs, priorities, and other planning issues implementing IT, there has been virtually no discussion about how to make health IT itself safe for patients, practitioners, and health care organizations. The irony of seeking safety through systems that may not be safe to begin with seems to have been lost in the enthusiasm for remaking health care via IT. in Past experience with IT has not shown it to be an unequivocal success. 14 Hardware failures have propagated in unexpected ways to remote, ostensibly unrelated components on a common network ; system upgrades have lead to missing or false laboratory information; programming mistakes have similarly led to incorrect guidance in decision support; and