My name is Meryl Nass and I practice internal medicine in Freeport, Maine. I appreciate your invitation to be here today, to provide the perspective of a doctor in the trenches.
I am on the AHRP Board, and am known for my work with anthrax, having uncovered its use as a biological weapon in Rhodesia, and for my critiques of the military's anthrax vaccine program. I have given testimony to three Congressonal committees on anthrax vaccine, bioterrorism, and Executive order 13139, which gives the President the right to waive informed consent for military troops who are given experimental drugs. I also investigate Gulf War Syndrome and its relationship to various environmental exposures, including unlicensed drugs and vaccines that were used during Operations Desert Shield and Desert Storm, without informed consent, and without creating a medical record of their use[1].
Anthrax vaccine happens to be part of a much larger military medical program, funded first in 1997, termed the joint vaccine acquisition program, or JVAP, whose mission is the development of dozens of new vaccines for biowarfare threats. The vaccines will be administered to those in the military.
In 1997 I reviewed the JVAP's draft environmental impact statement, and thereby learned that the use of unlicensed vaccines on soldiers without informed consent was premeditated, having been mentioned in the draft EIS before Congress had even funded the program. My criticism of this (illegal) plan--to ignore the requirement for informed consent when administering experimental products--can be found in the final environmental impact statement's appendix.
The concept of forcing experimental products on humans now appears to be encroaching on the civilian sector of society. The model emergency health powers act gives either the governor, or the Health and Human Services Secretary, in the case of a national public health emergency, the right to enforce the administration of (possibly unlicensed and insufficiently tested) drugs and vaccines on all Americans, who may have no right of refusal.
I have brought up this matter because the waiver of informed consent at some institutions in the ARDS Network may be another example of what still remains an unlawful act in the United States: experimenting on humans without their informed consent.
I say unlawful because to the best of my understanding, the ARDS Network trial does not meet the criteria specified by OPRR on October 31, 1996 for emergency waiver of informed consent:
(http://ohrp.osophs.dhhs.gov/humansubjects/guidance/hsdc97-01.htm)
The related issue is whether the informed consent process and documents that were given to ARDSnet patients and surrogates was adequate. Most physicians are aware that the informed consents we use with patients are not always well-understood by patients, and may not always make clear that there are alternatives to the treatment or trial to which the patient is consenting.
One example stared me in the face as I composed this. An 8 page advertising section in last week's Newsweek magazine was designed to attract subjects for cancer clinical trials, in general. Nowhere in the 8 pages is there any discussion of Phase 1 or Phase 2 clinical trials. The following statement is included: "The majority of all clinical trial participants are enrolled in Phase 3 trials, where patients receive either the best existing treatment currently known, or the new therapy." The document later points out that phase 3 trials rarely use placebos. Had this advertisement mentioned specific clinical trials, it would have been considered a part of the informed consent process, and FDA could have sanctioned the advertiser for misleading claims.
Let's be realistic: do those in the cancer research establishment really want a cancer patient to understand all the implications of enrolling in a Phase 1 trial?
A recent Lancet article showed that even when 'top of the line' informed consent was given to cancer patients, with discussions lasting an average of one hour, the majority of the patients still did not understand that they might not receive the medicine proven to be best for their condition. This is probably not surprising, since more than 50% of the physicians providing the discussions did not understand that cancer clinical trials were designed to improve treatment for future patients, not necessarily for those enrolled in the current trial.[2] Clearly, the large issue in many clinical trials of what constitutes satisfactory and sufficient informed consent, and what the patient or their surrogate really understands, needs to be revisited.
In the case of ARDSnet, were patients and surrogates told that the low volume treatment might result in increased shortness of breath, and discomfort, and that the higher volume treatment might result in pulmonary pressures higher than usually considered optimal?
If, during the trial, other adverse events were discovered to occur at a higher than expected rate in one or more arms of the trial, was this then included in the consent process for subjects entering later?
After issues of informed consent, the second concern I have is the issue of whether enrolling in a clinical trial may degrade, rather than enhance, the care that the subject receives, and thus worsen the outcome. If the trial is not properly designed and honestly analyzed, the results could potentially worsen the care of future patients, by wrongly altering the standard of care.
This is not a theoretical but instead a very real problem, particularly since most trials with negative results never see the light of day. And because many different entities have vested interests in changes to the standard of care, practicing physicians are inundated with guidelines professing to inform us of improvements. Unfortunately, the introduction of so-called "evidence-based medicine" has not solved the problem of how to identify which guidelines are spurious. Those of us practicing clinical medicine in this country have become extremely circumspect about what the medical literature now tells us to do, as it has become polluted by special interests over the years. This does not bode well for the value placed on medical research in future.
The ARDSnet group sought to, and may have, changed current clinical practice.
According to Weinert[3] and Ricard[4], doctors have not, in fact, adopted the use of lower tidal volumes for ARDS patients since the trial's conclusion. Less than 1% of Weinert et al's patients received tidal volumes of 6.2 ml/kg or less. According to an editorial by Pinsky, after ARDSnet we remain uncertain of precisely what the best ways are of ventilating ARDS patients[5].
But clinical practice could have changed in response to this trial, and had it done so, following the ARDSnet guidance to use very low tidal volumes, would our patients be better or worse off? As one example, would we be seeing increased use of sedatives and muscle relaxants in ARDS patients, with resultant increases in post-ventilator myopathies? (ref: Neil McIntyre presentation: Mechanical ventilation strategies for lung protection, May 1999) After a significant change to clinical practice, it would require some time to "shake out" whether mortality rates had improved, and if they did so, whether it was due to the revision in ventilator management or other factors.
Because there have been a series of trials of tidal volumes in ARDS and acute lung injury patients, with diametrically opposed results, what can be reliably said about ventilator management for this condition?
Dr. Jack Noble, my colleague, comes to his analysis from the social sciences. He was surprised at the numerous factors that had not been controlled for or accounted for in the ARDSnet statistical analysis. I explained to him that the practice of ignoring variables other than the one that is being investigated is the rule, rather than the exception, in human clinical trials. Then I began to wonder if this is part of the reason why the conclusions of so many clinical trials conflict with one another?
Maybe the traditional large clinical trial, which generally focuses on and controls only a single variable, is not the best way to improve clinical practice. Perhaps we should reconsider our statistical methods, and adjust our techniques, in order to definitively resolve whether other variables may be confounding our results.
Several other ideas come to mind. Maybe we have lost sight of what can be gained from experiments in small numbers of people who are investigated very thoroughly.
Perhaps clinical trials need to enroll smaller numbers of subjects that more closely match each other. Perhaps we could derive additional benefit from studying one person at a time, but exhaustively, as was done in an ARDS case report by Kallet et al in 1999.[6] This patient happened to be enrolled in the ARDSnet trial, and was switched from a tidal volume of 11.4 to 7.4 ml/kg. He then rapidly developed pulmonary edema. When he was returned to his previous tidal volume, he rapidly improved. A single patient, meticulously monitored, demonstrated that in his particular case, "exacerbation of acute pulmonary edema coincided with the institution of a lung-protective strategy." For this type patient, a low volume strategy was clearly the wrong approach.
This well-documented case reminds us as clinicians that each patient is unique, that they often fail to emulate our models and may behave in entirely unexpected ways, and that we must be sensitive to and respond to individual patient reactions. There is no room in high quality medical practice for "cookie cutter" responses to patients and their illnesses. There is also the need to be able to adjust therapy on a moment to moment basis, especially in the critically ill.
But a clinical trial, in general, demands cookie cutter treatment by the clinician. Usually, given the enhanced attention the patient in a trial receives, the care is better and so are the outcomes.
But it may be that in certain cases, despite the increased attention, requiring clinicians to follow a trial protocol results in worse outcomes. Could this have happened in the ARDSnet trial?
Here I would like to identify several issues.
The first is the need for trial designers to properly distinguish what is the gold standard of treatment, against which their experimental intervention will be compared.
The second is the need to design the trial so that the experiment can be performed in a way that does not detract from clinical care.
The third issue is whether the bedside clinicians were sufficiently independent, so that they could use clinical judgment to vary from the protocol in the case of individual patients, when appropriate.
The fourth issue is whether those charged with ongoing data analysis performed properly, and identified concerns with the comparative mortality rates in the groups receiving low tidal volume, high tidal volume and the group of patients who declined trial participation, or identified any other concerns?
The fifth issue is whether the trial's principal investigators were apprised of such issues early on, while the trial was underway, and dealt with them appropriately, in terms of maximizing patient safety.
1. The first point is whether the current standard of care, when the trial was designed, was actually 12 ml/kg tidal volume for ventilated ARDS patients? This does not seem to have been the case; rather, the mean tidal volume used was closer to 9 ml/kg in 1993.[7] However, a comparison of the ideal body weights (and methods used to derive them) and the actual body weights measured in ARDSnet and similar trials would help to resolve this.
I think it is fair to say that the Washington Manual reflects the standard of care in most teaching hospitals. The 1995 edition had this to say about ventilator settings:
"Initial tidal volumes can usually be set at 10-12 ml/kg. Patients with decreased lung compliance (e.g., ARDS) may need smaller lung volumes (6-8 ml/kg) to minimize peak airway pressures and their associated adverse effects." (28th Edition, page 201)
Prior to randomization, the subjects in this trial were receiving an average tidal volume of 10.3 ml/kg. Claiming that 12 ml/kg was the gold standard simply set up a straw man, against which the low tidal volume was competing. This created a trial design that was incapable of answering the question the trial was supposed to resolve, which was: what is the optimal tidal volume to minimize mortality in ARDS and acute lung injury?
A mistake was made - but why wasn't it discovered? A lot of people were involved in shaping and assessing this trial design, in a lot of centers. Each participating center also had a reviewing IRB. NHLBI funded the trial for $37 million dollars; it must have gone through extensive review there. How was it that such a basic error was allowed to stand? This is a fundamental question.
2. Point two asks whether the trial detracted from clinical care. Because of the higher than usual tidal volume in the control group, the answer is probably yes for the higher volume group, for the following reasons:
I believe the issue of plateau pressures that the investigators selected as thresholds for the high and low volume groups needs to be revisited. The assumption appears to have been that pressures up to 50 cm water were within the current standard of care, and were safe. (See "plateau pressure goals" on Page 2 of the ARDSnet protocol at: http://hedwig.mgh.harvard.edu/ardsnet/justvent911/node18.html )
The discussion of barotrauma in the May 2000 NEJM ARDSnet paper defines barotrauma as the presence of one or more of the following: pneumothorax, pneumomediastinum, subcutaneous emphysema and pneumatocoele. These events were not statistically more common in either the high or low volume group, which seems to imply that neither group's plateau pressure was more likely to predispose to barotrauma.
But the question of what degree of pressure causes damage on the cellular level, and increases mortality, is what is most important. The defining events for barotrauma that I just mentioned may be very poor surrogates for damage on the cellular level.
One reason to say this is that the current state of the art suggests that plateau pressures in ARDS patients should be maintained below 30 to 35 cm water. How did I conclude this?
I did a Pub Med search using the words "adult respiratory distress and plateau pressure". Every abstract that gave a target plateau pressure published in the last 18 months suggested keeping the pressure below 30-35 cm. [4] [8] [9] On the other hand, I found a 1989 review article on therapy of ARDS that used 50 cm water as the threshold for barotrauma.[10] Boussarsar et al performed a literature review and meta-analysis of ARDS studies for the determinants of barotrauma, which included 14 papers with a total of 2270 patients, and found that plateau pressures above 35 cm were highly correlated with barotrauma.[11] However, they found in their own prospective study of 116 patients, like the ARDSnet group, that pneumothorax was not a function of tidal volume.
The Esteban study[9] of not only ARDS patients, but 5000 patients on mechanical ventilation for a variety of indications, found that for the entire group plateau pressures above 35 were strongly related to increased mortality.
Analysis of the ARDSnet data[12] showed that only the level of PEEP, and not plateau pressure, was a determinant of gross barotrauma.
What does all this data mean for those receiving the higher tidal volumes? My interpretation is that plateau pressures in the range of 35 to 50 cm, associated with higher tidal volumes, may or may not increase the level of gross defining episodes of barotrauma, but there is sufficient evidence to conclude that despite this uncertainty, such levels may still increase mortality, and are considered by most experts undesirable. I therefore believe the higher tidal volume group received less than optimal care.
Did the lower tidal volume group receive less than optimal care? Probably, on the basis of their mortality rate being higher than the group of patients who declined to be part of the trial, but we cannot say so with absolute certainty. Nor can we say whether a meaningful number of patients, like the one described by Kallett et al who developed flash pulmonary edema, had significant worsening on the low volume protocol. I have not seen enough data to know.
3. Point three asks whether the treating physicians were able to go outside the parameters of the experiment in individual cases, according to their clinical judgment.
What we know is that it is unlikely that much deviation from the protocol took place, at least in the first three days, since the mean tidal volumes during the first three days were 6.2+/- 0.8 ml/kg and 11.8 +/- 0.8 in the low and high tidal volume groups respectively. We further note that low volume patients had to have "severe dyspnea"[13] in order to increase tidal volumes to 7 to 8 ml/kg of predicted body weight, according to the 2000 ARDSnet article mentioned previously.
Were the treating physicians able to judge whether the prescribed treatment was working well, having been told to tolerate elevations of pCO2, and patient discomfort, and to treat these manifestations with sedation? Given that treating physicians expected dyspnea, patient agitation and discomfort, it might have been extremely difficult to make a clinical decision as to the advisability of the prescribed ventilator treatment.
At issue is the fact that the bedside clinicians in this trial were by and large physicians in training, working in a very hierarchical system in which they were subservient to the principal investigators at their institutions. Having experienced this conflict myself, as a resident, treating patients on experimental protocols when I sometimes did not feel the best interests of the patient were being served by the protocol, I can tell you now, twenty years later, that I still did not dare to vary from it, or request of the attending that the patient be treated off protocol. I'd be surprised if other physicians in the room haven't had the same experience.
This is a broader issue, and not specific to the ARDSnet trial, yet commonplace in clinical trials. However, I do not recall ever seeing it discussed, and it probably deserves to be studied and discussed.
4, 5. Regarding the fourth and fifth points, which have to do with the process of ongoing data monitoring, and whether signals of problems were identified, as they should have been, and whether they were properly handled by the core group of investigators, I have no information on which to comment. I leave these to the OHRP to resolve.
I leave the group with these questions on my mind:
1. Why did some centers feel informed consent was not needed from every patient or surrogate? How can we assure this problem is not a feature of future trials?
2. Was the informed consent process otherwise accurate and complete? If not, why not? How can the informed consent process be strengthened in future?
3. When did it become clear that trial subjects' mortality was greater than those who declined to be part of the trial, and how was this handled?
4. Are we ready as a society to give up our right of informed consent to experimental treatment - and if not, now that it is imperiled, how can we safeguard it?
5. In a trial like this, involving our top physicians and institutions, generous funding, a large staff, and oversight on many levels, how was the trial design allowed to stand? Will those of us with a stake in the institutions of medical research and patient care grapple effectively with the many systemic implications this brings to the surface?
6. What can be done to improve the ethical practices, design and practical value of future clinical trials? Can smaller, cheaper trials provide some of the answers we desire?
7. Finally, should a trial using the ARDSnet ventilation protocols be permitted to resume? I think not, at least until all the raw data has been made available, so that independent analysis can determine a) whether patients were actually harmed by their participation, and b) the entire community of research stakeholders decides whether we have important knowledge that can be gained by continuing.
Thank you.
[1] Rettig RA. Military Use of Drugs Not Yet Approved by the FDA for CW/BW Defense. Lessons from the Gulf War. MR 1018/9. OSD. 1999 (RAND).
[2] Joffe S, Cook F, Cleary PD. Quality of Informed Consent in Cancer Clinical Trials: A Cross-Sectional Survey. Lancet 2001; 358:1772-7.
[3] Weinert CR, Gross CR, Marinelli WA. Impact of randomized trial results on acute lung injury ventilator therapy in teaching hospitals. Am J Respir Crit Care Med 2003:167:936-941.
[4] Ricard JD. Are we really reducing tidal volume - and should we? Am J Respir Crit Care Med 2003;167:1297-8.
[5] Pinsky MR. Toward a better ventilation strategy for patients with acute lung injury. Critical Care 2000; 4:205-6.
[6] Kallett RH, Alonso JA, Luce JM. Exacerbation of acute pulmonary edema during assisted mechanical ventilation using a low tidal volume, lung-protective ventilator strategy. Chest 1999;116:1826-32.
[7] Jardin F, Fellahi JL, Beauchet A. Improved prognosis of ARDS 15 years on. Intensive Care Med 1999; 25:936-41.
[8] Page B, Vieillard-Baron A, Beauchet A. Low stretch ventilation strategy in acute respiratory distress syndrome; eight years of clinical experience in a single center. Crit Care Med 2003: 31: 765-9.
[9] Esteban A, Anzueto A, Frutos F, Alia I, Brochard L, Stewart TE, Benito S, Epstein SK, Apezteguia C, Nightingale P, Arroliga AC, Tobin MJ; Mechanical Ventilation International Study Group. Characteristics and outcomes in adult patients receiving mechanical ventilation: a 28-day international study. JAMA. 2002 Jan 16;287(3):345-55.
[10] Bresler MJ and Sternbach GL. The Adult Respiratory Distress Syndrome. Emerg Clin N America 1989;7:419-430.
[11] Boussarsar M, Thierry G, Jaber S, Roudot-Thoraval F, Lemaire F, Brochard L. Relationship between ventilatory settings and barotrauma in the acute respiratory distress syndrome. Intensive Care Med. 2002 Apr;28(4):406-13. Epub 2002 Jan 15.
[12] Eisner MD, Thompson BT, Schoenfeld D, Anzueto A, Matthay MA; Acute Respiratory Distress Syndrome Network. Airway pressures and early barotrauma in patients with acute lung injury and acute respiratory distress syndrome. Am J Respir Crit Care Med. 2002 Apr 1;165(7):978-82.
[13] ARDS network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the adult respiratory distress syndrome. NEJM 2000:342:1301-8.