|
Home
|
Issues
|
MJA shop
|
MJA Careers
|
Contact
|
Topics
|
Search
|
RSS |
Editorial
Hyperbaric oxygen for carbon monoxide poisoning
Are currently recommended regimens ineffective?
MJA 1999; 170: 197-199
|
In 1950 Pace
and colleagues published a report establishing that
hyperbaric oxygen (HBO) significantly accelerates the rate of
carbon monoxide (CO) elimination from haemoglobin.1 Further studies
have shown that, after CO exposure, HBO accelerates the dissociation
of CO from cytochrome a,a3 and induces a more rapid return
to normal of cytochrome redox state,2,3 reduces brain lipid
peroxidation,4 inhibits pathological
endothelial leukocyte adhesion5 and prevents intracranial
hypertension.6
It has therefore been no surprise that clinical observation and published evidence, including that of randomised trials, have supported the clinical efficacy of HBO for CO poisoning.7-12 When HBO is given to individuals with moderate or severe poisoning within approximately six hours of exposure, most published data strongly suggest that it not only induces a more rapid recovery, but also reduces the most-feared complication -- persistent or delayed neurological sequelae. Nevertheless, questions remain. As HBO therapy is not immediately available to all poisoned patients, who should be treated or transported to a hyperbaric facility? How many treatments should be given? There have been few randomised trials of HBO therapy, and no published randomised trial has addressed the question of whether it is effective in the treatment of severely poisoned patients. Therefore, a large trial from a reputable institution is welcome. In this issue of the Journal, Scheinkestel and colleagues13 have addressed the question of whether HBO should be used for CO poisoning. Their study design is among the most rigorous yet published in this area. Nonetheless, methodological questions remain, and some caution is required in accepting the authors' conclusions.
In the study, all 191 patients were treated for 100 minutes in a
hyperbaric chamber, but were randomised to receive either
normobaric oxygen (NBO) or HBO at 2.8 atmospheres absolute (ATA) for
60 of the 100 minutes. All patients received continuous high flow
oxygen between treatments, or 100% oxygen if they were intubated.
They received one treatment per day for three days, after which they
underwent neuropsychological assessment. If, after this, they had
clinical or neuropsychological abnormalities, they received an
additional three treatments. Most patients in this study were
poisoned during an attempted suicide, and 73% were categorised as
severe (defined by a carboxyhaemoglobin level >30%,
Mini-mental score The use of cluster randomisation for patients presenting simultaneously (to minimise the effect of the study on daily practice) engenders the risk of bias, as clusters of patients with similar baseline characteristics are simultaneously assigned to one or other of the groups. Generalised linear models were used to adjust for within-cluster correlation, but the authors have not provided enough information to allow an assessment as to whether the statistical analysis accounted for all of the resulting bias. A confirmatory analysis that would not have incurred this potential bias would have been a reanalysis of the results using only one patient from each cluster. This might have strengthened the report, as would a display of the numbers and sizes of clusters and a listing of the magnitude and consistency of the parameter estimates for the various models. The primary results of the study depend heavily on seven neuropsychological assessments that were performed after treatment and before discharge. Two or more "abnormal" test results represented a poor outcome, which then defined persistent neurological sequelae (PNS). Conclusions based on these tests must be tempered by the fact that, although baseline Mini-mental state tests were done, the study did not make baseline neuropsychological assessments, and hence could not quantify change for these patients. The only statistically significant difference between NBO and HBO patients at the end of treatment was in one of seven neuropsychological scales (Rey auditory verbal learning test), which favoured patients treated with NBO. Considering the multitude of statistical tests performed and the lack of a comprehensive baseline assessment, a single significant test may not be meaningful. In view of the weight of evidence in favour of the use of HBO for acute CO poisoning, how can the conclusions of Scheinkestel and colleagues that there is no difference in efficacy between HBO and NBO be explained? Firstly, it is possible that their method of NBO administration was more effective than in other studies. All their patients received three or six days of high-flow oxygen, and intubated patients might have been given 100% oxygen for six continuous days, a regimen that is considerably more intensive than common clinical practice. Secondly, their HBO regimen may have appeared less effective than in other studies, for several reasons. Depression can confound neuropsychological assessment, and the high proportion of depressed patients in the study may have minimised the apparent effect of HBO treatment and contributed to the high number of patients with a poor outcome. Unfortunately, neither the inspired oxygen concentration in either group nor the exact pressure-time profile of the hyperbaric treatments is provided. However, adding up to six HBO sessions to 100% oxygen for three or six days would be expected to produce a significant degree of pulmonary oxygen toxicity. Also, as Scheinkestel and colleagues point out, repetitive treatments at 2.8 ATA (a pressure higher than many clinicians use for CO poisoning) might have induced a neurotoxic effect14 that offset any potential benefit. Further, there are significant omissions from the article that preclude unfettered acceptance of the authors' conclusions. The surprising observation that there was no significant improvement in Mini-mental score in either group is weakened by the lack of information regarding administration of sedative drugs, especially to intubated patients, which might have confounded the testing. Importantly, other than mortality, no clinical outcomes or self-reported assessments of functional ability are reported. The overall relapse rate at follow-up, defined as new morbidity or deterioration in any neuropsychological test score, was higher in the group treated with HBO, but the relapse rate in the various subgroups of greatest interest (particularly those with short treatment delays) is not detailed. Moreover, the low follow-up rate (46%) makes it difficult to draw valid conclusions. The data regarding comparability of the two groups have two significant omissions -- the numbers of severely affected patients with long delays to treatment, and the number of mildly affected patients. Neither of these subgroups is likely to show a measurable response to treatment using the chosen endpoints. The possibility that both types were significantly represented is suggested by Mini-mental scores that appear disproportionately high for the degree of severity that is implied, as well as the high geometric mean of the delays to treatment (>6 hours). It therefore appears possible that a significant proportion of the patients in this study were treated at a time after CO exposure that HBO is likely to be ineffective.15-17 Including a large number of patients who are unlikely to respond (too mildly affected or treated too late) in a study will reduce the apparent effectiveness of the intervention, and might partly explain the surprisingly high proportion of patients with PNS (74% and 68% of patients, respectively, in the HBO and NBO groups). What might otherwise have been the most important conclusion of this investigation -- that even in the subgroup of severe poisonings treated within four hours, there was no difference between NBO and HBO -- would have been more convincing had the authors provided the observational and statistical details. What new information can be learned from the work of Scheinkestel and colleagues? Their results hint that, in the type of patients studied, prolonged administration of NBO may be more effective than the shorter regimens that are in general use. With respect to the primary question addressed by the investigators,13 we feel that there are still too many unresolved issues in their analysis to discard HBO as a treatment for acute CO poisoning. Richard E Moon Elizabeth DeLong
©MJA 1998 | |||
|
| |||
| |||
24, confusion, loss of consciousness, focal
neurological deficits, convulsions, pulmonary oedema,
electrocardiogram abnormalities or dysrhythmias, hypotension,
cardiac arrest or acidosis). Nevertheless, inclusion of mildly
poisoned patients and the high average Mini-mental scores suggest
that a significant number of patients had only mild central nervous
system impairment at the time of presentation.
2
integrins by hyperbaric oxygen in carbon monoxide-mediated brain
injury in rats. Toxicol Appl Pharmacol 1993; 123: 248-256.