To the Editor.
—I can strongly agree with Dr Berwick1 that the "disciplined practice learning" that he espouses as "realtime science" is important and leads to clinically useful improvements in patient care, while still being concerned that he confuses the process that produces improvement with the knowledge gained from scientific investigation. As he states, "No one—not even the surgeons themselves—knows exactly what factors drove their mortality rates down." It is most admirable that the process in which these surgeons engaged has produced a significant lowering in mortality rate. We should all be encouraged to follow and, yes, report on the effects of similar processes. However, such processes usually involve many changes in procedure and practice, and in the absence of knowledge as to which specific changes lead to improvement, much superstitious behavior can be engendered. Where these behaviors do not create increased cost or risk, they are inconsequential. However,