CO: The COntroversy and COnfusion COntinues

abstract & commentary

Source: Scheinkstel CD, et al. Hyperbaric or normobaric oxygen for acute carbon monoxide poisoning: A randomized controlled clinical trial. Med J Aust 1999;170:203-210.

Using blinded cluster randomization, 191 carbon monoxide (CO) poisoned patients were assigned to hyperbaric oxygen (HBO) or normobaric oxygen (NBO) treatment schemes. All patients were included regardless of CO level (average CO level: HBO 20.5%, NBO 22%). Cluster randomization is used to assign all patients that present in a group to one of the randomized treatments. NBO patients received 100% oxygen at 1.0 atmosphere for 100 minutes. This was achieved using an occlusive non-rebreather face mask and a special oxygen reservoir. These treatments were administered in the hyperbaric chamber, which was flushed with air to simulate pressurization as a sham hyperbaric treatment. HBO patients received 60 minutes of 100% oxygen by occlusive mask at 2.8 atmospheres. Each treatment was administered daily for three days. In between, all patients received oxygen by non-occlusive non-rebreather face mask with high-flow oxygen at 14 L/min. Intubated patients received 100% oxygen. Neuropsychological testing was performed at day three; if abnormal, three more days of treatment were administered. The neuropsychological testing was performed at completion of treatment and at one month.

Randomization resulted in no significant differences between the groups. However, patients in the NBO group were treated one hour sooner than patients in the HBO group on average. Only 46% of patients were available for follow-up. Results showed that HBO patients required more additional treatments than NBO patients (28% vs 15%, P = 0.01). None of the 87 NBO-treated patients had delayed neurological sequelae (DNS); however, five of the 104 HBO patients had DNS. Because many practitioners believe HBO must be delivered within six hours, a post hoc analysis stratifying patients by time to treatment demonstrated equal outcomes for NBO or HBO. The data for that analysis are not reported. Scheinkstel and colleagues conclude that HBO cannot be recommended for treatment of CO poisoning.

Comment by Richard J. Hamilton, MD, FAAEM, ABMT

This report from Australia has been touted as the final word on the ongoing controversy of whether HBO improves outcomes in CO poisoning. On first analysis, Scheinkstel et al have used some clever mechanisms to overcome the typical problems associated with this sort of study — cluster randomization, sham chamber treatments, etc. There are a few small problems with this study: the inclusion of patients with levels lower than 25% without analysis of symptoms, the delays in treatment, etc. However, there is one glaring problem in that the NBO wing of treatment hardly received "normal" amounts of oxygen. These patients received continuous high-flow oxygen via a non-rebreather mask for 3-6 days, as well as 100% oxygen by "occlusive mask." The former is what most hospitals provide as "100% oxygen," and the latter is provided by a special aviator style or inflated-seal mask with rubber retainer straps. This is certainly not what most physicians would consider the treatment currently touted as the alternative to HBO, which is just high-flow oxygen by non-rebreather mask until the CO level drops to zero. The only legitimate interpretation of this study is that continuous high-flow oxygen with intermittent 100% NBO appears to be equivalent to continuous high-flow oxygen with intermittent 100% HBO.

My practice is to refer all patients with syncope, neurologic abnormalities, and cardiac abnormalities with elevated CO levels for HBO. In addition, patients with levels greater than 25%, patients who have been "soaked" in CO (long exposures but with lower levels), and pregnant patients also should be referred. My preference is to transfer them so that the HBO treatment can be initiated within six hours of their exposure. In general, they receive one 2.8 atmosphere dive with two 20-minute exposures to 100% oxygen by tight-fitting non-rebreather. They are subsequently discharged. The outcomes from my referral center are similar to those published here. I would remind those who believe that HBO treatment is of questionable value that the only demonstrably similar treatment protocol requires a stay in the hospital for 3-6 days on high-flow oxygen. In this day and age, that length of stay will be difficult to justify.

Methodological problems in the study by Scheinkstel et al on the treatment of carbon monoxide (CO) toxicity include all of the following except:

a. treatment delays.

b. the inclusion of patients with CO levels < 25% without symptom analysis.

c. lack of cluster randomization.

d. lack of generalizability of their normobaric oxygen delivery strategy to the general medical community.