© 2003 American Thoracic Society
The Role of a Journal in a Scientific ControversyFive groups of investigators have undertaken randomized trials to determine whether or not low tidal volume decreases mortality in patients with the acute respiratory distress syndrome (1–5). Two groups reported that low tidal volumes had a beneficial This editorial has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org effect (1, 2) and three groups reported no benefit (3–5). In December 2002, the Journal published a Critical Care Perspective by Eichacker and colleagues (6) focusing on these five trials. The article has proven controversial. It has led to several exchanges in the Correspondence columns of the Journal (7–20), five articles in the New England Journal of Medicine (21–25), an article in Science (26), and several articles in the lay press. The incident raises a broader question: what is the impact of publishing a controversial article in a medical journal? A controversy arises from a disagreement between two schools of thought. There are also two viewpoints about the impact of a controversy. One sees controversy as a sign of disunity, highlighting the fragile nature of science, which otherwise is regarded as the most unified body of knowledge we have. The second view sees controversy as a sign of vitality, emphasizing the importance of a subject. And while the two sides in a controversy disagree on the surface, they have a common goal: the search for truth. Before considering the role of controversy in science, it is useful to consider the purpose of a medical journal. A journal serves as a means of holding together a community that shares common interests (27). That community consists of authors, readers (who are also authors and researchers), clinicians (who may or may not be authors or researchers), and a sponsoring society. A journal provides this community with a forum for bringing new ideas into the open for discussion. Even before the invention of printing, the advantage of getting a message in writing was understood: Moses did not rely on the oral tradition when bringing the Ten Commandments down from Mount Sinai. Although a journal is a place to register new research, it must not be perceived as an archival tome on a dusty library shelf (28). Instead, a journal should be relevant to day-to-day clinical life, providing a home of opinion and a stimulus for new research. Everyone who influences the practice of medicine is included in this community, and so a subspecialty journal plays a vital role in the advance of a medical discipline. A journal also plays a vital role in protecting patient safety by alerting the community of new research or of new interpretations of already published research. The search for truth is the ultimate goal of science. But when we read new research findings, we do not immediately say that they are true or false. Instead, we use words like interesting, suggestive, plausible, and persuasive (29). Even were we to judge an author's arguments as compelling or obvious, only time and the future decides whether or not they are true. The expectation for originality means that research is conducted at the margins of technical feasibility, where speculation and confusion are expected and contradiction and controversy are the rule. Journals publish papers worthy of examination, not revealed truth. Arnold Relman, editor of the New England Journal of Medicine from 1977 to 1991, pointed out: "We make no claims to omniscience, nor can we guarantee that everything published will stand the test of time. Indeed, we are sure it will not, for it is in the nature of medical progress that ideas are being continuously reshaped" (30). It is a mistake to regard journals as repositories of absolute truth. But if journals stimulate curiosity and foster open debate, then journals can help in the search for truth. The major cause of the acute respiratory distress syndrome is sepsis, and a most sought-after goal is therapy directed at mediators of sepsis (31, 32). The 1991 report by Ziegler and coworkers (33) of a randomized trial of an antibody (HA-1A) against endotoxin was initially hailed as a major breakthrough. The antibody had no effect on mortality in the overall group, but it decreased mortality in the subgroup with gram-negative bacteremia. The authors concluded "HA-1A antibody is safe and effective for the treatment of patients with sepsis and gram-negative bacteremia". Now, if mortality is significantly decreased in one subgroup but overall mortality is not changed, logic dictates that some patients must have had an increased mortality. But from the way the conclusion is worded, the reviewers of the manuscript must not have raised this concern, or, if they did, the editor did not consider it a serious reservation. A concern about subgroup analysis was raised in subsequent letters to the New England Journal of Medicine (34). Ziegler and coworkers (35) responded that they "conducted multivariate analyses to adjust the treatment effect of HA-1A for possible confounding factors ... After adjustment for all of these variables in a series of models, the treatment effect of HA-1A always remained significant." But the controversy did not go away. In a second randomized trial, overall mortality was again virtually identical for the antibody and placebo (36). At the first interim analysis, mortality in patients without gram-negative bacteremia was 41% for those who received antibody and 37% for those who received placebo. The increased mortality with the antibody exceeded the preset threshold and the trial was stopped. Here is an example of how a published debate led to the saving of patient lives and prevented the licensing of a therapy that was originally claimed as a major advance. Trauma is another cause of the acute respiratory distress syndrome, and traumatic brain injury causes neurogenic pulmonary edema (37, 38). In 1997, Marion and coworkers (39) published a randomized trial of hypothermia in 82 patients with traumatic brain injury. They concluded that hypothermia significantly improved outcome. Four years later, Clifton and coworkers (40) reported a larger randomized trial in 392 patients. Hypothermia had no benefit. These are just a couple of studies that the reviewers and editors judged believable, and would have had enormous impact on the management of critically ill patients. But then more convincing data came along and overturned the initial beliefs. As clinicians, how should we select reports on which to base our practice? Proponents of evidence-based medicine say you can grade the reliability of research reports (41). Level 1 evidence is a randomized controlled trial or metaanalysis (see Figure E1 in the online supplement). At the bottom of the pyramid is level 5, consisting of a case series. The pyramidal system implies that there is a hierarchal approach to generating knowledge. Randomized trials, at the peak, have epistemologic superiority—the information is more reliable than other forms of research. The name, evidence-based medicine, is clever and seductive. It naturally implies that any critic practices self-indulgent medicine, based on anecdote, and never reads a journal. A pyramid needs a solid foundation. But the pyramid of evidence-based medicine is not solid—it's a house of cards. The assumptions behind evidence-based medicine fly in the face of research outside medicine and ignore the revolutionary changes in the philosophy of science of the twentieth century (42). In their blueprint paper, the developers of evidence-based medicine write: "A new paradigm for medical practice is emerging. Evidence-based medicine deemphasizes [reasoning based on] pathophysiologic rationale" (43). As such, we are talking about a practice of medicine divorced from the scientific principles that are its foundation. Poynard and coworkers (44) analyzed 474 original articles on cirrhosis and hepatitis published between 1945 and 1999, and found that only about half the conclusions were still considered true after 50 years (see Figure E2 in the online supplement). Truth survival of randomized studies was not different from nonrandomized studies, nor did a grading of the quality of the methodology influence continued validity of the conclusions. These findings of Poynard and coworkers represent a microcosm of the twentieth-century philosophy of science. At the turn of the twentieth century, the research of Isaac Newton (1642–1727) had been accepted as guaranteed truth for over 300 years. Indeed, Newton was credited as having discovered laws of nature. Then Albert Einstein (1879–1955) published his first theory of relativity in 1905, which aroused considerable controversy. In 1919, crucial experiments confirmed that Einstein was right and Newton was wrong. The realization that Newton was wrong shook physics to the core (45, 46). But, at least, mathematics was believed sound. In their magisterial Principia Mathematica, published between 1910 and 1913, Alfred North Whitehead (1861–1947) and Bertrand Russell (1872–1970) argued that the whole of mathematics could be derived from principles of logic. In 1931, Kurt Gödel (1906–1978) published his incompatibility theorem—pointing out that axioms are unprovable in a system on which they are based—and wrecked the entire edifice. The upheaval caused by Einstein and Gödel meant that even the most rigorous scientific methodology—methods far, far more rigorous than a clinical trial—can never guarantee knowledge (46). This revolution in the philosophy of science means that the five-level pyramid of evidence-based medicine has no solid foundation. "There is no absolute knowledge," commented Jacob Bronowski (47), "and those who claim it, whether they are scientists or dogmatists, open the door to tragedy." Three papers stimulated much of our understanding of ventilator management of patients with the acute respiratory distress syndrome (48). The 1967 report of Ashbaugh and coworkers (49, 50) is a case series, placing it at level 5 on the evidence-based medicine scale. The reports of Mead and coworkers (51) and of Webb and Tierney (52) are experimental research and do not show on the scale—they are located somewhere below level 5. Within 3 years of the Ashbaugh report, Mead and coworkers concluded "mechanical ventilators, by applying high transpulmonary pressure to the nonuniformly expanded lungs ... may cause the hemorrhage and "hyaline membranes" found in such patients' lungs at death" (51). What amazing prescience. Here we see that research far down on the evidence-based medicine scale can ultimately lead to major advances in how we take care of patients. One common form of ventilator assistance is continuous positive airway pressure (53). The landmark article on its use in sleep apnea was a report of five cases by Sullivan and coworkers (54). Myocardial ischemia is a major problem in critically ill patients. The first report in the English language of transluminal angioplasty by Gruentzig (55) was an account of five cases in a letter to the Lancet. These two reports highlight that there is no dependable grading system to tell us what type of new research findings will lead to better care of patients. When first published, the reports of Sullivan and Gruentzig were met with skepticism, and each aroused considerable controversy. Controversy has been a constant companion of science. When Copernicus (1473–1543) discovered that the earth rotated around the sun, he delayed publication for 30 years for fear of upsetting the establishment. He famously received the first copy of De Revolutionibus on his deathbed. The reticence of Copernicus was wise. A hundred years later, Galileo (1564–1642) again advanced the Copernican thesis in his Dialogue. The Inquisition forced him to recant, and he spent the rest of his life under house arrest. Pope John Paul II admitted some error in 1992. The list of researchers condemned and ridiculed for controversial new ideas is long, and includes Harvey, Darwin, and Semmelweis to name a few. It is a mistake to think of communication in science as the unidirectional transfer of information from one person to another. Instead, communication is achieved by dialog, and knowledge is gained through argument. Scholars back to the time of Socrates (470–399 BC) have used dialog as a method for truth seeking. It was Georg Hegel (1770–1831) who identified the key steps whereby dialog, or the dialectic process, leads to a change in understanding (46). We begin with the thesis—the initial state of affairs. This provokes an opposing viewpoint, the antithesis. The two sides engage in debate. Then, by weighing arguments and applying rules of logic, the thesis and antithesis are united into a synthesis. The synthesis has the seeds of a new controversy, and the process begins all over again but from a higher level of sophistication. As such, Hegel's dialectic is often called the law of change because it presumes that reality is a dynamic rather than a static state. Few participants in a debate possess the equanimity of Dominic Corrigan (1802–1880): "Whether my observations and opinions be disproved or supported, I shall be equally satisfied. Truth is the prize aimed for; and, in the contest, there is at least the consolation that all the competitors may share equally the good attained" (56). A willingness to listen to an opposing viewpoint is especially important in clinical research, because the stakes—patient safety—are so high (57). Being able to see a debate in print allows for more dispassionate and critical scrutiny than is possible during an oral debate. Several topics are being vigorously debated in the Correspondence Columns of the Journal (58–85). The revolution in the philosophy of science caused by Einstein and Gödel led Karl Popper (1902–1994) to formulate a full-scale theory of knowledge in "The Logic of Scientific Discovery" (86). Popper pointed out that we can never guarantee that a new proof will remain true in the future. Instead, all we can do is show that a theory is in error. A good theory—or explanation—will be replaced by a better theory, which, in turn, will be replaced by an even superior theory. Knowledge is created by humans, and it will always be fallible precisely because it is created by humans. Even the best knowledge—created by a genius like Newton, and corroborated repeatedly over 300 years—cannot be certain. Popper extended his theory of research to sociology and politics in a second book, "The Open Society and Its Enemies" (87). He argued that an open society, which permits open discussion, criticism, and opposition, is the best social system. In scientific publications, authors feel free to express independent views. The alternative is silence: secrecy when self-imposed, censorship when imposed by others. The trust that readers have in a journal is an expectation that the editor is not beholden to any special interest group (88), feels free to publish a controversial article that might offend the proprietor (88), and favors integrity over concerns with profitability (88). Proprietors may think otherwise and fire the editor (89–91), but their journals lose years of hard-won credibility in the process (92, 93). How do we reach closure in a controversy? In a court of law, lawyers determine the facts, and the jury and judge make a ruling. In a scientific controversy, however, the validity of the facts—the data—is rarely disputed. Instead, what's at stake is the relative believability of contradictory interpretations of the data. Should the opponents in a controversy go to a hierarchical authority, such as the president of a National Academy of Science, who would make a ruling like the Pope, with the expectation that the parties obey the ruling? Religion emphasizes faith, revelation, and constancy. But science is devoted to doubt, discovery, and change. Sovereignty over science resides in the hands of numerous independent individuals rooted in a common tradition (94). A hierarchical authority, making pronouncements from a central office, would be as destructive to science as sovereignty shared among individuals is essential to its survival. It is the nature of science that no authority is conceivable that could completely overrule the general opinion formed by dialog among individual scientists (94). The motto selected by the oldest scientific society still in existence, the Royal Society (95), founded in 1660, was "Nullius in Verba," best translated as "Take nobody's word for it; see for yourself." Most serious scientists cringe at the very idea of some central authority telling them how to interpret data. When true scientists engage in debate, they do not appeal to some higher authority, seek testimonials from grandees of a field, or engage in ad hominem arguments—instead, they stick to the data and its interpretation (42, 45). Science has no official machinery for closing a controversy (42). Proponents of a minority view are not expected to recant in public, and they can still advance their views, as did Linus Pauling (1901–1994) on vitamin C (96). Discredited claims are not killed-off. They simply fade away. The final sanction of the scientific community is the sanction of the skeptic: the less believable side of an argument is ignored. In conclusion, controversy is not a sign of confusion or a breakdown in the scientific process, but a sign of vibrancy and a motor for advance. Controversy highlights the tentative nature of the conclusions in any paper, no matter how rigorous the design and convincing the data. In clinical research, concern for patient welfare must take precedence over all other concerns. Journals must never be seen as closing their channels to a dissenting view that is well thought out, because the appearance of suppressing sincere criticism is far more destructive to the credibility of science than is the criticism itself. Censorship and secrecy are the enemies of science. Controversy and debate are signs of health. And journals play a key role in helping the two sides in a controversy to cooperate in the search for truth. FOOTNOTES Conflict of Interest Statement: M.J.T. is editor of AJRCCM. He receives a fixed stipend from the American Thoracic Society. He receives royalties for two books published by McGraw-Hill, Inc. REFERENCES
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
