© 2004 American Thoracic Society
Ethics and Standard of Care in Clinical TrialsTo the Editor:In their recent perspective, Miller and Silverman used the term "standard of care" to refer to physicians' usual care practices (1). (Other authors used the term "best current care" to describe usual care [2]). The words "standard" and "best" imply that usual care is well characterized and clearly recognizable. However, usual care frequently encompasses broad ranges of physician behavior (2) (as with tidal volumes in acute respiratory distress syndrome [ARDS], which necessitated the trials of volume and pressure limited ventilation). Moreover, "standard" and "best" imply "good," but as Miller, Silverman and others (3) report, usual care frequently strays from well established good care. Miller and Silverman (1) state "this standard of care had not been validated." Therefore, use of "standard of care" seems internally inconsistent. Moreover, one cannot know if usual care is similar or different from good care when the latter is undefined. This was the case with tidal volumes before the ARDS Network trial. The term "usual care" (4) is preferable because it is descriptive without implying positive attributes that may not exist. Miller and Silverman (1) state "it is hard to dispute that [the ARDS Network trial of lower tidal volume ventilation (5)] would have had greater potential clinical value if it had included a control group representing the standard of care." This seems logical on the surface, but it is easy to dispute. We excluded a usual care group because of the following reasons: (1) usual care for tidal volumes and inspiratory pressures in ARDS encompassed broad ranges because of differences of opinion and style (2); (2) reasons for physicians' usual care decisions are frequently unknown; and (3) usual care practices shift over time (6) and frequently for reasons that are unrelated to new evidence. Therefore, it would have been impossible to describe usual care in our trial. This would be a cardinal flaw in study design because the study could not be replicated. Moreover, true differences in outcomes between a protocol group and a usual care group could be obscured by the similarity of outcomes in the protocol group and the subset of usual care patients whose care resembled the protocol. Results of a negative trial that included a usual care group would be impossible to interpret and would violate Miller and Silverman's first ethical principle: generate scientifically valid data. If the trial produced uninterpretable results, potential risks to subjects would not be justified (7). A trial could demonstrate significant differences in outcomes between a protocol group and a usual care group that overlaps with the protocol group. However, the number of patients needed for adequate trial power is inversely proportional to the square of the anticipated difference in outcomes between study groups. Therefore, more patients would be required if there were substantial overlap between usual care and protocol groups. Under most assumptions, such a trial would have more deaths in the less favorable study group than a trial in which the two protocol groups represented distinct differences in practices within usual care (8). Hence, an adequately powered trial with a usual care study group would probably violate Miller and Silverman's second ethical principle: protect subjects from harm.
a Johns Hopkins University Baltimore, Maryland FOOTNOTES Conflict of Interest Statement: R.G.B., G.B., and A.M. do not have a financial relationship with a commercial entity that has an interest in the subject of this letter. REFERENCES
From the Authors: We described "the standard of care" as having a descriptive and a normative sense (1). Accordingly, there is nothing inconsistent in characterizing a given standard of care (descriptive sense) as invalid. Brower and colleagues take issue with our claim that the clinical value of the Acute Respiratory Distress Syndrome ARDS Network trial would have been enhanced by including a standard care group. We agree that the use of a control group receiving standard (or usual) ventilatory care poses potential methodologic difficulties, which we discussed in a previous article (2). However, the absence of such a control group precludes valid inferences about whether either the lower or the higher tidal volume strategy is better than standard care practices, thus diminishing the clinical value of the two-arm ARDS Network trial. Finally, Brower and colleagues refer to an abstract (3) that examines "extra deaths" calculations to refute the primary ethical reason for including a standard control group—to be able to detect whether patients receiving either the lower or higher tidal volume strategy (or both) have a higher mortality rate than those receiving standard care. Although the abstract does not exhaustively analyze the range of potential trial designs or the different conceptions of "extra deaths," it does point out the ethical relevance of considering standard care practices in the design of critical care trials. It was this ethical relevance that motivated us to write our perspective.
a National Institutes of Health Bethesda, Maryland FOOTNOTES Conflict of Interest Statement: F.G.M. and H.J.S. do not have a financial relationship with a commercial entity that has an interest in the subject of this letter. The opinions expressed are those of the authors and do not necessarily reflect the position or policy of the National Institutes of Health, the Public Health Service, or the Department of Health and Human Services. REFERENCES
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
