Randomized, Controlled Trials in Critical Care: An Expert Interview With John J. Marini, MD

Interviewer: Antonios Liolios, MD

June 03, 2003


After having attended Dr. John J. Marini's debate entitled, "Randomized Trials Determine Best Practice in the ICU: Pro-Con Debate,"[1]at the 23rd International Symposium on Intensive Care and Emergency Medicine in Brussels, Belgium in March 2003, Dr. Antonios Liolios wrote an article as part of the conference coverage for Medscape Critical Care.[2] (See the conference coverage section of Medscape Critical Care.) Dr. Liolios had a few more questions to ask about the topic, so he spoke again to Dr. Marini.

Dr. Marini is Director of Academic Programs in Research and Education for the Department of Medicine of the Regions Hospital in St. Paul, Minnesota. Dr. Liolios is Attending Intensivist in the Intensive Care Unit at the University Hospital of Saint Luc in Brussels, Belgium and is a member of the Medscape Critical Care Editorial Board.

Antonios Liolios, MD

John J. Marini, MD

Dr. Liolios: Were there any particular events that made you critical of the utility of randomized, controlled trials (RCTs)?

Dr. Marini: Although there are many specific examples of RCT conclusions that conflict with my personal experience, my primary concern is that the results of those trials -- however internally valid they might be -- pertain to population averages and not directly to the individual patient under consideration. Many of the conditions we address in critical care are imprecisely defined (eg, acute respiratory distress syndrome, sepsis), and the complexity of their illnesses render such population-based recommendations questionably valid. Comorbidities, cointerventions, and variation of practice styles unaddressed by these trials concern me for using RCT-based evidence as the gold standard to guide practice.

Dr. Liolios: Some compare evidence-based medicine (EBM) to a modern religion. Do you believe that EBM limits independent thinking and personal initiative? Is this necessarily detrimental in everyday clinical practice?

Dr. Marini: Who can argue with using the "best evidence" to aid decision-making? But for many problems in critical care, RCTs provide only a fragment of the relevant database -- not a prescription for action. With so much variability, the best we can hope for is a guideline for the broad category of patients who are similar to the ones in the trial and managed in the same way. The best informed EBM practitioners and advocates understand this. Yet, in an effort to define "best practices" and standardize care, the results of RCTs are increasingly promoted as a "trump card" that overrides the other key elements of clinical judgment. In my view, this is a dangerous practice built on a flawed paradigm. Whatever their value, large clinical trials are seldom repeated and stand unopposed as the best evidence for years afterward, even as the practice environment changes. In some notable cases, treatments that are valuable for some patients -- steroids and prone positioning are two examples -- are left unconsidered.

Dr. Liolios: Do you use renal dose dopamine in your ICU practice? Why?

Dr. Marini: Maybe it isn't the best EBM approach, but I admit to trying it in oliguric patients who have failed to respond to other measures. For me, the practice of critical care often requires making informed and physiology-based management decisions that involve cautious tests and challenges to our assumptions. I know that dopamine may not routinely enhance the distribution of renal blood flow, and usually (on average) it won't help a bit. Yet, in certain instances it does help -- perhaps by increasing cardiac output, by the redistribution mechanism, or perhaps by another mechanism yet to be defined. If it fails, I stop it.

Dr. Liolios: What type of knowledge (scientific, empiric, etc.) do you consider evidence?

Dr. Marini: Carefully conducted studies of either type are part of the useful database. Evidence can also take the form of one's own well-documented clinical experience, even if no published information is available to confirm it. In the latter instance, there must be a plausible physiologic mechanism consistent with that experience that could potentially explain it.

Dr. Liolios: Are there any established guidelines for the management of certain diseases or conditions in your ICU? Who determines these guidelines and their validity?

Dr. Marini: Of course. Most guidelines involve the implementation of well-established principles and are intended to assure the avoidance of iatrogenic misadventures. We have an open ICU arrangement (as do the majority of hospitals in the United States). Guidelines are helpful for residents in training, nurses, generalists, and intensivists in order to standardize the consensus reached by physicians with advanced credentials, nursing leaders, and ancillary care providers.

Dr. Liolios: What do you believe about complementary and alternative medicine? Do you think they have a position in the ICU?

Dr. Marini: Interesting question. At the present time, there are so few scientific or physiologic rationales for the great majority of these options that we confine their use to those instances in which we think them innocuous and the family or patient emphatically believes that they should be tried. These cases arise very infrequently and are almost always based in ethnic practices.

Dr. Liolios: Do you think there should be a practical alternative to EBM? If yes, what would you suggest?

Dr. Marini: What do you mean by EBM? I think physicians were attempting to use the best evidence base well before the formalization of that term in the early 1990s and should always strive to do so. Randomized trials should be undertaken with great caution, however. To paraphrase an old adage -- "Don't just do something -- stand there!" Certainly RCTs have utility when the question is of pressing clinical relevance and can be studied effectively. They can: (1) characterize the collective result of applying a certain management strategy across a population of patients with varying practice styles; (2) confirm a consistent difference between therapeutic alternatives in well-defined circumstances; or (3) test in the clinical setting a promising approach with a solid preclinical research base. RCTs with exacting definitions and methods can demonstrate a "proof of principle." But as you already must have guessed, the vast majority of management questions can either not be meaningfully or even ethically tested by RCT. Finally, the soul of science is reproducibility -- and this form of expensive clinical science is seldom replicated.

Dr. Liolios: In your talk, you mentioned the example of use of steroids in ARDS; based on previous negative trials, physicians withheld a potentially helpful treatment for years until recently. Were you using steroids before the Meduri trial for late ARDS?[3] Why?

Dr. Marini: Not routinely then and not routinely now. Along with many practitioners, I used steroids back then in those unusual patients in whom immune-mediated inflammation was likely to figure prominently and in whom deterioration was evident. Now, I consider them when resolution progress is slow, using them (unless obviously contraindicated) for a 3- to 5-day trial period. They are stopped if no radiographic and gas exchange benefit is seen. Of course, many of our most seriously ill patients are now tested for relative adrenal insufficiency and may receive low-dose steroids for that indication.

Dr. Liolios: Are you concerned that by assuming such a position against EBM, you may encourage the re-emergence of various nonscientific approaches?

Dr. Marini: Again, I emphatically am NOT against developing a logical approach to clinical practice in which the best available scientific data are considered. But I do believe that the overemphasis of RCTs as the pre-emptive lynchpin is misdirected for many of the problems we face. The complexity of critical illness and our inability to precisely define who we are talking about severely limit RCT methodology in most instances. These theoretical limitations are quite apart from my ongoing concerns regarding the practical elements -- motivation, funding, selection, design, execution, and interpretation.

Dr. Liolios: What type of research do you consider meaningful and mainly engage yourself in?

Dr. Marini: The bulk of my research activity has been investigator-initiated experiments centered on helping to define physiologic mechanisms of disease or treatment. Because of escalating clinical and educational commitments, time spent on research is increasingly precious to me. Therefore, I try to focus on those topic areas that leverage my ability to contribute to the advancement of useful medical knowledge. That said, on rare occasion, I have participated in multicenter trials because of my desire to gain experience with an exciting innovation (eg, liquid ventilation). I tend to refrain from clinical trials activity when anonymous participation is unlikely to provide new insight, to be intellectually stimulating, or to promise great benefit (with minimal risk) to the specific patients I enter into the protocol. Unfortunately, the overwhelming majority of trials fall into the latter categories. Of course, if a clinical trial is superbly designed, well motivated, addresses a key area of my interest, and does not require a large personal time investment, I would be happy to help the general cause.