To the Editor.
—Groundbreaking though their work is, Dr Bates and colleagues1 and Dr Leape and colleagues2 may have collectively oversimplified the management tasks for reducing adverse drug events (ADEs). By launching into computerized systems changes without a deeper understanding of medication delivery process stability, they also encourage management decision making that appears contrary to knowledge-based approaches to such matters.Deming's3 knowledge-based management theory is grounded in the predictability of organizational systems. Prediction requires stable processes. Neither Bates et al nor Leape et al demonstrated that what was observed was in aUpper Control Limit Average per Week 1989 Weekly Control Chart 1990 Weekly Control ChartActual weekly process control charts of a corporation's Occupational Safety and Health Administration—reportable injuries for 1989 and 1990. The left side of the Figure shows an unstable system in which nonrandom, assignable causes of variation produced a spike of injuries after