Marilyn Sue Bogner, Ph.D.
Institute for the Study of Medical Error
The computer chip and resultant automation as well as technology in general have revolutionized equipment across domains. Early in this technological revolution, there were expressions of concern about the unquestioning acceptance of automation -- acceptance that contributes to problems. Among the early cautions are lack of user oriented system linking ( Chapanis, Garner, & Morgan, 1949) and lack of operator knowledge about the state of the system when human intervention is necessary ( Bainbridge, 1983). These concerns and others are increasingly expressed for aviation as well as process control and nuclear power ( Parasuraman & Riley, 1997). However, automation in medicine, a domain in which technology is pervasive, has received a paucity of attention ( Bogner, 1994). It is particularly important to address medical automation concerns because the current fiscal constraints of managed care incorporate the implicit if not explicit assumption that technology can compensate for fiscally induced changes in health care delivery.
In health care facilities, the workload of the clinician is increasing and the work environment is complicated by greater cognitive demands and accelerated work tempo afforded by automation ( Cook, 1994). Such system complexity contributes to an increased number of problems while making it more difficult to identify and correct the source of a problem. These conditions give rise to the inappropriate attribution of a problem to human error. The complications, however, reflect well documented problems with technology such as lack of feedback ( Norman, 1992); a need for simplicity, transparency, comprehensibility, and predictability in system behavior ( Billings, 1997); and information overload ( Stokes & Wickens, 1988).
The environments in which complications are most evident, those in which fast and accurate decisions and treatment are most critical and stress and fatigue the highest, are the emergency room ( Xiao et al., 1996), the operating room ( Finley & Cohen, 1991), and intensive care unit ( Donchin et al., 1995). Health care delivery is further impacted by the presumed cost-saving personnel changes mandated by managed care such as replacing experienced registered nurses with lower-salaried, less experienced licenced practical nurses. This is occurring as demand is increasing for clinical experience and skill to safely use technologically sophisticated medical devices. Other manifestations of fiscal constraints are discharging a patient in a sub-acute state with attendant medical devices to home care provided by inexperienced, stressed friends or relatives as lay care givers, and involving the patient in self-care using sophisticated and difficult to use devices ( Obradovich & Woods, 1996). Concurrent with increasing the role of lay care givers in providing care using medical technology is the reduction in assistance to those care givers by restricting reimbursement for home visits by nurses. These scenarios that mismatch inexperienced users with sophisticated devices establish the conditions for incidents with medical technology that have adverse outcomes. Such outcomes often require further medical treatment with additional trauma and expense, and affect the quality of life of the patient, the lay care giver, and the clinician through the specter of litigation.
For medical automation to be a help rather than a hazard, its design must be user-driven ( Bogner, 1996), complex displays must be configured with regard for the users' needs ( Chapanis, 1996), and situations comparable to computer controlled flight in aviation that might invite large blunders while eliminating small manual system errors ( Wiener, 1993) must be avoided. In short, to be a help rather than a hazard, medical automation and technology must be socialized from being dictators that demand conformance to their terms forcing users to persevere in tailoring their behavior to accommodate those terms however inappropriate they may be in the operational context, to being colleagues with the users on the users'terms.