Tuesday, February 10, 2009

West's estimations of PaO2 on Everest Confirmed - but SaO2 remains an estimation

Recently, Grocott et al published results of an intriguing study in which they drew blood gas samples from climbers near the summit of everest and analyzed them at one of the high camps with a modified blood gas analyzer. (See: http://content.nejm.org/cgi/content/abstract/360/2/140 ) This is no small feat, and the perhaps shocking results confirm earlier estimations of low arterial oxygen tension derived from samples of exhaled gas. The PaO2 of these climbers is often under 30mmHg - a difficult to believe number for clinicians who are accustomed to a danger zone represented by much higher numbers in clinical practice.

As intriguing as the numbers may be, the authors have made a crucial assumption in the estimation of arterial oxygen saturation (SaO2) that leads us to be circumspect about the accuracy of this estimated value. A letter written by me and my colleagues emphasizing several caveats in these estimations was not accepted for publication by the NEJM so I will post it below.

In the article by Grocott et al, an important limitation of using calculated SaO2 values for the estimation of arterial oxygen content is neglected. The equation used for the calculation of SaO2 in the article does not take into account changes in hemoglobin affinity induced by increased 2,3-DPG levels which are known to occur during acclimatization (1;2). Errors resulting from these estimations will be magnified for values of PaO2 on the steep portion of the oxyhemoglobin dissociation curve. The PaO2 values of the subjects studied are on this portion of the curve. Can the authors comment on 2,3-DPG levels in these climbers and how any resulting changes in hemoglobin affinity may have affected calculated values? Were the climbers taking acetazolamide, which has variably been demonstrated to affect the oxygen affinity of hemoglobin (3;4)? Is there any evidence that acclimatization induces increased production of fetal hemoglobin as occurs in some other species (5)? Because of such caveats and possibly other unknown variables, co-oximetry remains the gold standard for determination of arterial oxygen saturation.

Reference List

(1) Wagner PD, Wagner HE, Groves BM, Cymerman A, Houston CS. Hemoglobin P(50) during a simulated ascent of Mt. Everest, Operation Everest II. High Alt Med Biol 2007; 8(1):32-42.
(2) Winslow RM, Samaja M, West JB. Red cell function at extreme altitude on Mount Everest. J Appl Physiol 1984; 56(1):109-116.
(3) Gai X, Taki K, Kato H, Nagaishi H. Regulation of hemoglobin affinity for oxygen by carbonic anhydrase. J Lab Clin Med 2003; 142(6):414-420.
(4) Milles JJ, Chesner IM, Oldfield S, Bradwell AR. Effect of acetazolamide on blood gases and 2,3 DPG during ascent and acclimatization to high altitude. Postgrad Med J 1987; 63(737):183-184.
(5) Reynafarje C, Faura J, Villavicencio D, Curaca A, Reynafarje B, Oyola L et al. Oxygen transport of hemoglobin in high-altitude animals (Camelidae). J Appl Physiol 1975; 38(5):806-810.

Scott K Aberegg, MD, MPH
Leroy Essig, MD
Andrew Twehues, MD

Monday, February 9, 2009

More Data on Dexmedetomidine - moving in the direction of a new standard

A follow-up study of dexmedetomidine (see previous blog: http://medicalevidence.blogspot.com/2007/12/dexmedetomidine-new-standard-in_16.html )
was published in last week's JAMA (http://jama.ama-assn.org/cgi/content/abstract/301/5/489 ) and hopefully serves as a prelude to future studies of this agent and indeed all studies in critical care. The recent study addresses one of my biggest concerns of the previous one, namely that routine interruptions of sedatives were not employed.

Ironically, it may be this difference between the studies that led to the failure to show a difference in the primary endpoint in the current study. The primary endpoint, namely the percentage of time within the target RASS, was presumably chosen not only on the basis of its pragmatic utility, but also because it was one of the most statistically significant differences found among secondary analyses in the previous study (percent of patients with a RASS [Richmond Agitation and Sedation Scale] score within one point of the physician goal; 67% versus 55%, p=0.008). It is possible, and I reason likely, that daily interruptions in the current study obliterated that difference which was found in the previous study.

But that failure does not undermine the usefulness of the current study which showed that sedation comparable to routinely used benzos can be achieved with dexmed, probably with less delirium, and perhaps with shorter time on the ventilator and fewer infections. What I would like to see now, and what is probably in the works, is a study of dexmed which shows shorter time on the ventilator and/or reductions in nosocomial infections as primary study endpoints.

But to show endpoints such as these, we are going to need to carefully standardize our ascertainment of infections (difficult to say the least) and also to standardize our approach to discontinuation of mechanical ventilation. In regard to the latter, I propose that we challenge some of our current assumptions about liberation from mechanical ventilation - namely, that a patient must be fully awake and following commands prior to extubation. I think that a status quo bias is at work here. We have many a patient with delirium in the ICU who is not already intubated and we do not intubate them for delirium alone. Why, then, should we fail to extubate a patient in whom all indicators show reaolution of critical illness, but who remains delirious? Is it possible that this is the main player in the causal pathway between sedation and extubation and perhaps even nosocomial infections and mortality? (The protocols or lack thereof for assessing extubation readiness were not described in the current study, unless I missed them.) It would certainly be interesting and perhaps mandatory to know the extubation practices in the centers involved in this study, especially if we are going to take great stock in this secondary outcome of this study.

Another thing I am interested in knowing is what PATIENT experiences are like in each group - whether there is greater recall or other differences in psychological outcomes between patients who receive different sedatives during their ICU experience.

I hope this study and others like it serve as a wake-up call to the critical care research community which has heretofore been brainwashed into thinking that a therapy is only worthwhile if it improves mortality, a feat that is difficult to achieve not only because it is often unrealistic and because absurd power calculations and delta inflation run rampant in trial design, but because of limitations in funding and logistical difficulties. This group has shown us repeatedly that useful therapies in critical care need not be predicated upon a mortality reduction. It's past time to start buying some stock in shorter times on the blower and in the ICU.

Tuesday, February 3, 2009

Cost: The neglected adverse event / side effect in trials of for-profit pharmaceuticals and devices

Amid press releases and conference calls today pertaining to the release of data on two trials of the investigational drug pirfenidone, one analyst's comments struck me as subtly profound. She was saying that in spite of conflicting data on and uncertainty about the efficacy of the drug (in the Capacity 1 and Capacity 2 trials - percent change in FVC [forced vital CAPACITY] at 72 weeks was the primary endpoint of the study) IPF is a deadly and desperate disease for which no effective treatments exist (save for lung transplantations if you're willing to consider that an effective treatment) and therefore any treatment with any positive effect however small and however uncertain should be given ample consideration, especially given the relative absense of side effects of pirfenidone in the Capacity trials.

And I thought to myself - "absense of side effects?" Here we have a drug that, over the course of about 1.5 years reduces the decline in FVC by about 60ccs (maybe - it did so in Capacity 2 but not in Capacity 1) but does not prolong survival or dyspnea scores or any other outcome that a patient may notice. So, I'm picturing an IPF patient traipsing off to the drugstore to purchase pirfenidone, a branded drug, and I'm imagining that the cash outlay might be perceived by such a patient as an adverse event, a side effect of sorts of using this questionably effective drug to prevent an intangible decline in FVC. The analyst's argument distilled to: "why not, there's no drawback to using it and there are no alternatives", but this utterly neglected the financial hardships that many patients endure when taking expensive branded drugs and ignored alternative ways that patients with IPF may spend their income to benefit their health or general well-being.

This perspective is even more poignant when we consider the cases of "me-too" drugs that add marginally to the benefits or side effect profiles of existing drugs, and which are often approved on the basis of a trial comparing them to placebo rather than existing generic alternatives. One of the last posts on this blog detailed the case of Aliskiren, and I am reminded of the trial of Tiotropium published in the NEJM in October, among many other entire classes of drugs such as the proton pump inhibitors, antidepressants, antihistamines, inhaled corticosteroids, antihypertensives, ACE-inhibitors for congestive heart failure, and the list goes on.

Given todays economy, soaring healthcare costs, and increasing financial burdens and co-pays shouldered by patients especially those of limited economic means or those hit hardest by economic downturns, we can no longer afford (pun intended) to ignore the financial costs of "me too" medications as adverse events of the use of these drugs when cheaper alternatives exist.

In terms of trial design, we should demand that new agents be compared to existing alternatives when those exist, and we need to develop a system for evaluating the results of a trial that does not neglect the full range of adverse effects experienced by patients as a result of using expensive branded drugs. Marginally "better" is not better at all if it costs ridiculously more, and the uncertainty relating to the efficacy of a drug must be accounted for in terms of its value to patients, especially when costly.