Estimation of adverse events in medical care

Abstract

Author's reply Sir—Poses agrees with us that the issue of what is the most appropriate way to study inappropriate care in hospitals is important. However, he underestimates the usefulness of the prospective observational study and overestimates the feasibility (a requirement of usefulness) of the randomised controlled trial (RCT) for this purpose. We, of course, know that the RCT is the method by which a segment of medical treatment is guided. From RCT or other systematic studies the correct dose of a drug or the proper antidote to a drug given in error are determined. When an incorrect drug dose was pointed out as an error in the settings we studied in a teaching hospital, the actual journal articles that provided the basis for the judgment were sometimes cited. Other kinds of events, for example, leaving an anaesthetised patient unattended, or finding an IV-line draining on the floor may not even have been studied by RCT or any systematic method. We are not sure how anyone could carry out an RCT of these or many of the other general errors discussed in the settings we studied. It would be inconceivable to randomly subject patients to such inappropriate care just so we could study whether they were actually harmed as a result. Poses is concerned that we have overestimated the frequency of adverse events since we relied on health care providers' assessments. We were concerned that the frequently cited estimates of such events, carefully gleaned from reviews of patient records, greatly underestimates the frequency of such occurrences. Studies with various methodologies have produced and will continue to produce varied estimates. The pivotal question is, which estimates should be used for which purposes? Attempts to prevent errors or reduce their frequency whether through education or through policy changes should be based on a greater understanding of what actually occurs in hospital. In our study analysing health care providers' discussions, there were incentives to keep misidentification of an error to a minimum. The person accused of the error was sometimes present to explain his or her rationale. Since medical students and residents were present to learn, senior personnel would have been remiss if they had not corrected a person who misidentified an error. And, unlike in a survey or a medical records review, there were full discussions of the events that might be perceived to be errors. Reviews of patient records have been elegantly done, but few would assert that all adverse occurrences in hospitals are clearly recorded in them. Although we specifically stated in our report that the true number of adverse events can never be known, we offered an alternative strategy for identifying adverse events in hospitals to better understand their magnitude. Estimation of adverse events in medical careThe conclusions drawn by Andrews and collegues (Feb 1, p 309)1 seem to go beyond their data. They report a qualitative (ethnographic) observational study of surgical teams, including house staff, attending physicians, and nurses. Ethnographers recorded instances when a team member alluded to an inappropriate decision. Their definition of an inappropriate decision was one made “when, at the time, an appropriate alternative could have been chosen”. They made no distinction between decisions deemed unquestionably and possibly appropriate, inappropriate by a senior surgeon or an intern, or inappropriate by the entire team or one member. Full-Text PDF