125: Carbon Monoxide

Carbon monoxide (CO) is formed during the incomplete combustion of virtually any carbon-containing compound. Because it is an odorless, colorless, and tasteless gas, it is remarkably difficult to detect in the environment even when present at high ambient concentrations and is a leading cause of poisoning morbidity and mortality in the United States. Based on US national death certificate data, there were 439 annual deaths from unintentional non-fire exposure to CO from 1999 to 2004.^25,123 The groups with the highest risk were male gender and elderly age, possibly because of occupational exposure and inability to discern CO symptoms, respectively. CO related mortality remained essentially unchanged in 2002 despite increased CO detector use.^22,24,26 More than half of these cases (64%) occurred in homes with faulty furnaces, usually in the fall or winter months. Many clusters are associated with power failures during catastrophic weather, such as ice storms, blizzards, and hurricanes.^22,23 With improved data collection, the CDC WONDER database reported 1944 deaths in the United States in 2010 due to the toxic effects of CO.²⁷

Just as important as mortality, are the greater number of survivors from CO poisoning. Despite increased awareness for CO poisoning, in 2004 to 2006, there were still an average of 20,636 nonfatal, unintentional, non–fire-related CO exposures treated annually in the United States.²⁶ More than 40% of cases occurred in the winter, with almost 75% occurring in residences. However, exclusion of intentional and fire related cases severely underestimates the extent of the problem. Based on firsthand hospital data, a minimum of 50,000 CO cases present to US Emergency Departments (EDs) each year, up to half resulting in hospitalization.^80,91 More recent data using probable and suspected cases suggests that there were over 230,000 ED visits in 2007 alone that were unintentional and related to non-fire CO poisoning.⁹²

The bigger problem with CO poisoning may be the associated morbidity that survivors risk even after acute treatment. The most serious complication is persistent or delayed neurologic or neurocognitive sequelae, which occurs in up to 50% of patients with symptomatic acute poisonings.^68,144,196 There is still no highly reliable method of predicting who will have a poor outcome, requiring the threshold for HBO therapy for CO poisoning and follow-up be particularly low.

Potential sources of CO abound in our society, often resulting in unintentional poisoning²⁶ (Table 125–1). Although CO is found naturally in the body as a byproduct of hemoglobin degradation by heme oxygenase found in the liver and spleen,⁴⁰ it is readily available for inhalation from the incomplete combustion of virtually any carbonaceous fuel. Alternatively, absorption—dermal, ingestion, or inhalation—of methylene chloride may result in CO toxicity after hepatic metabolism¹²⁸ (Chap. 108). Despite catalytic converters and other emission controls, more than 50% of unintentional CO deaths are still caused by motor vehicle exhaust.^39,123 Occupants of motor vehicles are not the only victims of exhaust gases; CO poisoning is also reported in occupants of the beds of pickup trucks and on boats.^21,79 Workers can become symptomatic from use of propane powered equipment indoors such as ice skating rink resurfacers²⁰ and forklifts.⁵⁵ For optimal performance, propane-powered forklifts are typically adjusted to produce no less than 10,000 ppm of CO in exhaust, but in fact average more than 30,000 ppm.⁵⁶ In an enclosed warehouse with poor ventilation, even with proper emission control, CO levels could exceed safe recommendations within an hour.

In the past 10 years, non-vehicular sources of CO have increasingly accounted for the majority of unintentional poisonings.¹²³ Predominantly, these have involved the burning of charcoal, wood, or natural gas for heating and cooking.⁷⁶ Natural gas (methane or propane) burning furnaces for heating are often implicated, especially when the flue is blocked.^83,84 Gas kitchen stoves are also an important source of CO in indigent populations with marginal heating systems.⁸⁴ In fact, the use of gas stoves for supplemental heat is predictive of CO poisoning in patients who present to the ED with headache and dizziness. During ice storms, blizzards, hurricanes, and other natural disasters, the indoor use of gasoline powered generators¹⁸⁹ and charcoal burning grills, the latter particularly in immigrant populations, has resulted in epidemic CO poisoning outbreaks.^22,23,202

Fires are another important source of CO exposure, contributing substantially to the approximate 5613 smoke inhalation deaths each year.³⁹ CO is considered to be the most common hazard to smoke inhalation victims.^61,155 These cases are further compounded by the high incidence of hydrogen cyanide poisoning (Chap. 126).⁷¹

CO is readily absorbed after inhalation. The Coburn-Forster-Kane (CFK) model allows the prediction of COHb levels based on exposure history.⁴¹ This model has been simplified to allow estimation of the equilibrium based on the ambient concentration of CO in ppm: COHb (%) = 100/[1 + (643/ppm CO)].¹⁸¹ This assumes that the individual weighs 70 kg and is not anemic. With exponential uptake, it may take more than 4 hours for equilibrium to be attained. Therefore, within minutes of high CO exposures, the arterial COHb level may actually overshoot predicted estimates before equilibration.^8,13 Endogenous production of CO is not factored in because its contribution to COHb is only 2%.

After it has been absorbed, CO is carried in the blood, primarily bound to hemoglobin. The Haldane ratio states that hemoglobin has approximately a 200 to 250 times greater affinity for CO than for oxygen. Therefore, CO is primarily confined to the blood compartment, but eventually up to 15% of total CO body content is taken up by tissue, primarily bound to myoglobin.⁴¹ Therefore, the dissolved CO level in the plasma may better reflect the ultimate potential for poisoning because it is available for diffusion into all tissue compartments, including the muscle and brain.¹⁰⁷

Elimination of CO, like absorption, from the blood can be modeled mathematically using the CFK model. The equation predicts a half-life of 252 minutes. In actual volunteer studies, means of 249 and 320 minutes breathing room air are reported.¹⁴¹ With 100% oxygen, these half-lives can be reduced significantly to means ranging 47 to 80 minutes in studies of volunteers who attain COHb levels of 10% to 12%.^141,163 Patients poisoned with CO showed actual mean half-lives ranging from 74 to 131 minutes when treated with 100% oxygen.^196,197

Methylene chloride, a paint stripper, is another source of CO. It isreadily absorbed through the skin or by ingestion or inhalation and is metabolized in the liver to CO.¹⁵⁸ Reaching peak COHb levels may take 8 hours or longer and may range from 10% to 50%.^106,147 Because of ongoing production of CO, the apparent COHb half-life is prolonged to13 hours in these patients.¹⁴⁵ COHb levels after methylene chloride exposure appear to be proportional to the concentration and duration of exposure.¹⁴⁵

The most obvious deleterious effect of CO is binding to hemoglobin, rendering it incapable of delivering oxygen to the cells. Therefore, despite adequate partial pressures of oxygen in blood (PO₂), there is decreased arterial oxygen content. Further insult occurs because CO causes a leftward shift of the oxyhemoglobin dissociation curve, thus decreasing the offloading of oxygen from hemoglobin to tissue¹⁵⁰ (Fig. 22–2). This may result in part from a decrease in erythrocyte 2,3-bisphosphoglycerate (2,3-BPG) concentration. The net effect of all these processes is the decreased ability of oxygen to be delivered to tissue.

CO toxicity cannot be attributed solely to COHb-mediated hypoxia. Neither clinical effects nor the phenomena of delayed neurologic deficits can be completely predicted by the extent of binding between hemoglobin and CO.^75,177 Furthermore, such a model fails to explain why even minimal levels of COHb (4%–5%) may result in cognitive impairment. An early study showed that dogs breathing 13% CO died within 1 hour and had COHb levels of 54% to 90%. However, exchange transfusion of this same blood into healthy dogs to reach similar COHb levels caused no untoward effects.⁶² Hemorrhaging the dogs to comparable degrees of anemia also produced no adverse effects. The appropriate conclusion was that inherent to CO toxicity is its delivery to target organs such as the brain and heart and that although COHb is easily measured; it rarely has a significant contribution to clinical toxicity. For CO to reach tissue, it had to be dissolved in the plasma rather than bound to hemoglobin.^66,67

CO interferes with cellular respiration by binding to mitochondrial cytochrome oxidase. Initial studies show that this binding is especially exaggerated under conditions of hypoxia and hypotension. In vitro rat models demonstrate that this oxidative stress causes mitochondrial damage with protein oxidation and lipid peroxidation, particularly in the hippocampus and corpus striatum.¹⁶⁴ In vivo models reveal that CO poisoning causes cell loss in the frontal cortex, which is associated with decrements in learning and memory.¹⁴² Although no comparable brain studies exist in humans, the peripheral lymphocytes and monocytes of CO-poisoned patients show cytochrome oxidase inhibition accompanied by increased lipid peroxidation.^59,118 In a small clinical series of CO-poisoned patients, normalization of this cytochrome activity lagged behind and seemed to agree better with symptom severity than COHb levels.¹¹⁹

Inactivation of cytochrome oxidase may be only an initial part of the cascade of inflammatory events that results in ischemic reperfusion injury to the brain after CO poisoning (Fig. A38–1). During recovery from the initial poisoning, white blood cells are attracted to and adhere to the damaged brain microvasculature.^169,170,171 This attraction may be partly attributable to endothelial changes from initial cytochrome oxidase dysfunction, mediated primarily through the free radical nitric oxide (NO).^169,175 CO displaces NO from platelets that in turn form peroxynitrites, which are even stronger inactivators of cytochrome oxidase.¹⁷⁵ Multiple animal studies demonstrate that NO is ultimately responsible for much of the endothelial damage from CO and that NO synthase inhibitors can prevent toxicity.^174,175 The NO formation promotes platelet–neutrophil aggregates that in turn lead to neutrophil adhesion to the brain microvasculature.¹⁷⁰ Myelin peroxidase activation in the area may further promote neutrophil adhesion with degranulation and release of proteases that convert xanthine dehydrogenase to xanthine oxidase, an enzyme that promotes formation of oxygen free radicals.¹⁶⁸ The end result of this process is delayed lipid peroxidation of the neurons, and the extent of destruction may be correlated with decrements in learning in rodents.¹⁶⁸ Rats depleted of xanthine oxidase through a tungsten modified diet show no changes in myelin basic protein and cognitive function after CO poisoning.⁸²

Simultaneously, with all this perivascular oxidative stress in the brain, there is activation of excitatory amino acids, which ultimately may be responsible for the subsequent neuronal cell loss.¹⁷⁵ In fact, in rat brains, glutamate concentrations increase after CO poisoning. Glutamate is an excitatory amino acid that can bind at N-methyl-d-aspartate (NMDA) receptors and cause intracellular calcium release, resulting in delayed neuronal cell death (Chap. 14). Blockade of NMDA receptors may prevent the neuronal death and learning deficits that accompany serious CO poisoning in mice.⁹³ Increases in the glutamate concentrations in rat brain in the first hour after severe CO poisoning are followed by a later increase in hydroxyl radicals.¹⁴² Ultimately, at 1 to 3 weeks, the animals show histologic evidence of both neuronal necrosis and apoptosis in the frontal cortex, globus pallidus, and cerebellum that are accompanied by deficits in learning and memory.

CO neuronal cell death may be caused by apoptosis, which has been confirmed in various models. In bovine pulmonary artery cells, CO exposure is accompanied by activation of caspase-1, a protease implicated in delayed cell death.¹⁷³ Confirmatory evidence was provided in the same study because both caspase-1 and NO synthase inhibitors blocked apoptosis. The end result of all these cellular processes is brain injury, particularly in the basal ganglia and hippocampus.¹⁹² In some studies this is accompanied by learning impairment.¹⁴² Thus, animal models correlate well with what ultimately occurs in victims of serious CO poisoning, namely, persistent or delayed deficits in learning and memory associated with structural changes in the brain.

Myoglobin, another heme protein, binds CO with an affinity about 60 times greater than it binds oxygen.⁴² About 10% to 15% of the total body store of CO is extravascular, primarily binding to myoglobin.^13,41 A dog model demonstrates that this binding is enhanced under hypoxic conditions.⁴² This binding may partially explain the myocardial impairment that occurs in both animal studies and low-level exposures in patients with ischemic heart disease. The combination of COHb formation, which decreases oxygen-carrying capacity, and reduced myoglobin in the heart, which decreases oxygen extraction, may explain the preterminal dysrhythmias that occur in animals.

Several studies suggest that CO effects on the cardiovascular system are necessary for ischemic reperfusion injury of the brain. Hypotension is an essential component and results from a combination of myocardial depression and vasodilation. CO, perhaps because of its similarity to NO, activates guanylate cyclase, which in turn relaxes vascular smooth muscle. Also, CO may further displace NO from platelets, resulting in additional vasodilation.¹⁷⁴ These factors contribute to the hypotension that occurs in animal experiments with exposure to high concentrations of CO.⁶² Such an episode of hypotension may present clinically as syncope, and this finding portends a worse clinical outcome.³⁴ In rhesus monkeys, cerebral white matter lesions correlate better with decreases in blood pressure than with COHb level.⁶⁵ Lipid peroxidation of the brain in rats develops an hour after a CO exposure that has produced syncope and hypotension.¹⁶⁷ This delay is comparable to the time that is necessary to produce mitochondrial destruction from oxidative stress in rats exposed to CO. In a feline model, central nervous system (CNS) damage from CO can be reproduced only when hypoxia is accompanied by an interval of ischemia, confirming the ischemic-reperfusion model.¹³²

Endogenous CO behaves like NO, binding to guanylate cyclase and thereby increasing cGMP concentrations.⁹⁴ Although low endogenous concentrations are physiologic, excessive concentrations of CO from exogenous sources may be problematic because CO persists much longer than NO. CO appears to be a neuronal messenger by virtue of the fact that as a gas, it can diffuse and signal adjacent cells.¹¹²

Acute Exposure

The earliest symptoms associated with CO poisoning are often nonspecific and readily confused with other illnesses, typically a viral syndrome¹⁹³ (Table 125–2). The initial symptom reported by volunteers within 4 hours of exposure to 200 ppm CO (producing COHb levels of 15%–20%) is headache; shorter exposures at 500 ppm also produce nausea.¹⁶⁰ The incidence of CO poisoning in symptomatic patients presenting to EDs in the winter with an influenzalike illness ranges from 3% to 24% in some series.^51,84 The typical presenting complaints include headache, dizziness, and nausea, and the most frequent exposures occur during the winter, explaining why influenza is the most common misdiagnosis.⁵¹ The most common symptom, headache, is usually described as dull, frontal, and continuous. CO poisoning is also frequently misdiagnosed as food poisoning, gastroenteritis, and even colic in infants. Similar to adults, children tend to develop nonspecific symptoms, complicating diagnosis.

Continued exposure to CO may lead to symptoms attributable to oxygen deficiency in the heart. Low-level exposures, leading to COHb levels of 2% to 4%, in volunteers with stable angina results in decreased exercise tolerance as well as signs and symptoms of myocardial ischemia.² At higher levels (COHb 6%), there is a greater frequency of premature ventricular contractions during exercise. Myocardial infarction and dysrhythmias are described in victims of CO poisoning, and acute mortality from CO is usually a result of ventricular dysrhythmias.^1,2 Prolonged exposure to CO or high COHb levels are associated with temporary myocardial stunning, lasting usually less than 24 hours and reflected by a decrease in left ventricular ejection fraction (LVEF).⁹⁵ This stunning is reflected by a decrease in LVEF and is correlated with increased β-type natriuretic peptide. Troponin may also be elevated in the absence of any coronary artery disease or even ECG or ECHO changes.^28,46 These patients have an increased propensity for cardiac mortality, with almost one-third dying within 8 years after serious CO poisoning.⁸⁵

The CNS is the organ system that is most sensitive to CO poisoning. Acutely, otherwise healthy patients may manifest headache, dizziness, and ataxia at COHb levels as low as 15% to 20%; with higher levels or longer exposures causing syncope, seizures, or coma.¹⁹³ Patients may present with focal neurologic symptoms suggestive of a cerebrovascular accident. The electroencephalogram (EEG) may show diffuse frontal slow-wave activity. Within a day of exposures that result in coma, computed tomography (CT) and magnetic resonance imaging (MRI) can show decreased density in the central white matter and globus pallidus (Fig. 125–1).¹⁰⁸ Autopsies show involvement of other areas, including the cerebral cortex, hippocampus, cerebellum, and substantia nigra.¹⁰⁴

Metabolic changes may reflect CO’s toxic effects better than any particular COHb level. Patients with mild CO poisoning may develop respiratory alkalosis in an attempt to compensate for the reduction in oxygen-carrying capacity and delivery. More substantial exposures result in metabolic acidosis with lactate production that accompanies tissue hypoxia.¹⁵⁷ Even in the absence of hypotension, lactate was an independent predictor of worse mental status and inpatient complications.¹²² The importance of metabolic acidosis was highlighted in a retrospective series of 48 CO-poisoned patients, in which hydrogen ion concentration was a better predictor of poor recovery during initial hospitalization than was COHb level.¹⁸⁶

Although the brain and heart are the most sensitive, other organs may also manifest the effects of CO poisoning. One-fifth to one-third of patients with severe CO poisoning—those who required endotracheal intubation—develop pulmonary edema.⁶⁹ This can be due to cardiac depression directly from CO and ARDS from associated smoke inhalation.^71,95,155 This does not appear to be a direct effect of CO on lung tissue because sheep with prolonged exposure to CO, resulting in COHb levels greater than 50%, showed no anatomic or physiologic changes in lung function.¹⁵⁴ Although myonecrosis and even compartment syndromes occur, patients rarely develop acute kidney injury (AKI). Retinal hemorrhages may develop with exposures greater than 12 hours.⁹⁶ Cherry-red skin coloration occurs only after excessive exposure (2%–3% of cases referred to one hyperbaric center) and may represent a combination of CO-induced vasodilation, concomitant tissue ischemia, and failure to extract oxygen from arterial blood.¹⁴⁸ Another classic but uncommon phenomenon is the development of cutaneous bullae after severe exposures. These bullae are thought to be caused by a combination of pressure necrosis and possibly direct CO effects in the epidermis.

The persistent or delayed effects of CO poisoning are varied and include dementia, amnestic syndromes, psychosis, parkinsonism, paralysis, chorea, cortical blindness, apraxia and agnosia, peripheral neuropathy, and incontinence.¹⁰⁵ If not diagnosed at initial poisoning, neurologic deterioration can be delayed and preceded by a lucid period of 2 to 40 days after the initial poisoning.³⁴ In patients admitted to an intensive care unit for severe CO poisoning and treated with 100% oxygen, 14% of survivors had permanent neurologic impairment.¹⁰⁰ In a Korean series of 2360 CO-poisoned patients, 3% continued to show memory failure or parkinsonian features one year after exposure.³⁴ Another series of 63 seriously poisoned patients showed memory impairment in 43% and deterioration of personality in 33% at 3-year follow-up.¹⁵⁷ Children also develop behavioral and educational difficulties after severe poisoning.¹⁰³ However, patients older than 30 years of age appear to be more susceptible to the development of delayed sequelae.^34,198 Most cases of delayed neurocognitive sequelae are associated with loss of consciousness in the acute phase of toxicity.³⁴

Neurocognitive sequelae probably involve lesions of the cerebral white matter.⁶³ Weeks after exposure, autopsies show necrosis of the white matter, globus pallidus, cerebellum, and hippocampus. MRI studies confirm the damage to the white matter and hippocampus.^58,108,193 Animal studies show that having a markedly elevated COHb level alone cannot cause similar white matter lesions but that there must also be an episode of hypotension.^65,132 The fact that the areas permanently damaged in serious CO poisoning cases are the areas with the poorest vascular supply in the brain is consistent with these findings.

Often, patients complain of persistent headaches and cognitive problems after long-term exposure to low concentrations of CO. Unfortunately, to date, there have been no controlled studies demonstrating that in the absence of a severe acute poisoning episode, this type of exposure results in any long-term sequelae. Warehouse workers who are chronically exposed to CO from propane combustion have intermittent problems with headache, nausea, and lightheadedness.⁵⁵ Fortunately, unless there has been an episode of severe poisoning with acute deterioration, most workers go on to have resolution of their symptoms.⁵⁶ One series of chronic CO poisoning demonstrates a high incidence of headache and memory complaints along with motor slowing and memory problems on neuropsychologic testing.¹²⁵ Although many of the objective deficits improved with elimination of the exposure and HBO treatment, many continued to have posttraumatic stress and conversion disorders. Although it is unclear that chronic exposure to low concentrations of CO can cause permanent damage, health care providers still should be vigilant for symptomatic individuals to prevent continued or catastrophic outcomes.

The most useful diagnostic test obtainable in a suspected CO poisoning is a COHb level. Normal levels of COHb range from 0% to 5%. Levels at the high end of this range occur in neonates and patients with hemolytic anemia because CO is a natural byproduct of the breakdown of protoporphyrin to bilirubin.³³ Of note, in blood samples from neonates, falsely high COHb levels up to 8% can occur due to interference of fetal hemoglobin with spectroscopy.¹⁹¹ COHb levels average 6% in one-pack-per-day smokers but may range as high as 10%.¹⁵⁹ Although high COHb levels confirm exposure to CO, particular levels are not necessarily predictive of symptoms or outcome.^75,140

The usual method for measuring COHb is with a cooximeter, a device that spectrophotometrically reads the percentage of total hemoglobin saturated with CO. Traditionally, arterial blood was used for this determination; however, venous blood levels are accurate as there is little CO extraction from hemoglobin across the capillary bed.¹⁸⁵ Refrigerated heparinized samples yield accurate COHb levels for months and at room temperature for 28 days, making retrospective clinical and postmortem evaluations reliable.^73,102

Bedside tests using ammonia or sodium hydroxide are unable to differentiate reliably various levels of COHb versus control subjects.¹³⁴ Because of the similarities in extinction coefficients, COHb is misinterpreted as oxyhemoglobin on most types of pulse oximetry (Chap. 29).⁷ Thus, the pulse oximetry reading is usually normal in the setting of even severe CO poisoning.⁷² Some newer pulse oximeters, called pulse cooximeters, have the ability to measure COHb noninvasively.⁵ A study in 10 healthy volunteers who inhaled CO at 500 ppm until they reached a peak COHb level of 15% found good agreement between pulse cooximetry and cooximetry.⁶ Early models of a commercial bedside pulse cooximeter showed very poor agreement, mischaracterizing half the patients with levels over 15% as lower.¹⁸⁵ Subsequently, another large cohort of ED patients, using a later model of the same device, found that it measured COHb well, with a bias of 3% and precision of 3.3%.¹⁴⁹ Because it tends to overestimate COHb, the authors recommend a normal upper limit of 6.6% as triggering intervention. Because the pulse cooximeter is noninvasive, it may be useful in screening ED patients for occult CO poisoning who present with nonspecific symptoms.³¹ In addition, it may be used in the field for screening fire victims, patients with potential CO exposure, and rescuers.

Breath-sampling methods may be used for screening patients.^43,45,53 A cutoff of 53 ppm in patients breathing air and 43 ppm in those breathing oxygen has an approximate reliability of approximately 80% in predicting COHb levels above 10%.⁵⁷ Breath sampling for CO has been advocated as a quick screening device for detecting recent exposure to CO in ED patients.

Some clinical laboratories measure CO directly in blood samples rather than COHb. This technique involves assaying CO directly with infrared spectrophotometry after it is extracted from the blood sample with a manometer. Based on calculations rather than true experimental data, the assumption has been made that for a patient with a normal hemoglobin, a CO level of 1 mmol/L corresponds to an 11% COHb concentration. A simpler method to measure serum CO content is to add a known solution of hemoglobin followed by sodium dithionite to form COHb. The resulting COHb is measured spectrophotometrically with the assumption that 1 mole of hemoglobin binds 4 moles of CO. Interestingly, in one study, serum CO ranged from 0.14 to 0.6 mg/L but was the same in smokers (average, 4.6% COHb) and nonsmokers (average, 1% COHb).²⁰⁰ At this time, further research is required to determine the clinical importance of serum CO content.

Additional laboratory tests may be useful in severe poisoning cases. An arterial or venous blood gas analysis will confirm the presence of metabolic acidosis, and a measurement of serum lactate concentration can be performed simultaneously. Metabolic acidosis with elevated lactate concentration may serve as a more reliable index of severity than a measurement of the COHb concentration.¹⁵⁷ Unfortunately, arterial pH does not correlate well with either initial neurologic examination or the COHb level, making it a poor criterion for deciding the need for HBO treatment.¹²⁴ Specifity of lactate may also be low in smoke inhalation victims, where it may indicate concomitant cyanide poisoning (Chap. 126).

Cardiac monitoring and a 12-lead electrocardiography (ECG) are essential to identify ischemia or dysrhythmias in symptomatic patients with preexisting coronary artery disease or severe exposure. Mild elevations of creatine phosphokinase are common (ranging 20–1315 IU/L in one series of 65 cases), usually because of rhabdomyolysis rather than cardiac sources.¹⁵³ However, because CO may cause myocardial infarction in the presence of normal coronaries,⁹⁵ it is not surprising to see nonspecific increases in troponin concentrations, which may reflect diffuse cardiac myonecrosis rather than focal coronary artery disease.²⁹ Congestive heart failure or hypotension can be evaluated with a β-type natriuretic peptide or echocardiography (or both), looking for evidence of myocardial stunning.⁹⁵ Because of the potential for increased cardiovascular mortality,⁸⁵ patients with ECG changes or elevated cardiac enzymes may benefit from further cardiac testing, a stress test, or angiography.

The problem with using COHb levels to base treatment is that there is a wide variation in clinical manifestations across patients with identical COHb levels. Furthermore, particular COHb levels are not predictive of symptoms or final outcome.^114,131,157 In a large prospective study of CO poisoning, COHb levels did not correlate with loss of consciousness and were not predictive of delayed neurologic sequelae.¹⁴⁴ Admission COHb levels are inaccurate predictors of peak levels, and the use of nomograms to extrapolate to earlier levels has not been validated. Their credibility is also suspect because of the great variability in COHb half-lives and differences in treatment with oxygen.

Because of the inherent unreliability of COHb levels in predicting outcome, researchers are searching for other surrogate markers. Rats have early increases in glutathione released from erythrocytes, a potential marker for CO oxidative stress that could ultimately lead to brain injury.¹⁷⁶ Another promising marker is serum S100B, a structural protein in astroglia that is released from the brain after hypoxic stress. A series of 38 consecutive patients poisoned with CO showed that those who presented with normal neurologic findings and no loss of consciousness had normal S100B concentrations.¹⁵ Patients who presented with loss of consciousness and neurologic deficits all had elevated concentrations. CO-poisoned rats treated with HBO did not develop elevated S100B concentrations unlike those treated with ambient oxygen therapy.¹⁴ Although markers used for cerebral injury, specifically S100B and neuron-specific enolase, do rise after CO poisoning, they have not be shown to be predictive of final outcome.^17,146 Levels at the target organ, the brain, show that S100B in the cerebrospinal fluid predicted extreme neurological sequelae, that is, persistent vegetative state, only.⁹⁰ A recent pilot study concluded that almost 100 different plasma proteins—cytokines, chemokines, and other biomarkers—increase after CO poisoning in patients; their significance awaits more definitive studies.¹⁷²

The extent of neurologic insult from CO can be assessed with a variety of tests. The most basic is documentation of the normal neurologic examination with a quick mini mental status examination. A more sensitive indicator of the acute effects of CO on cortical function is a detailed neuropsychologic test battery developed specifically for CO patients.¹¹⁷ The advantages of such testing, which usually takes about 30 minutes, are that it can reliably distinguish 79% of the time between CO-poisoned patients and control subjects, and it shows improvement with appropriate HBO treatment.¹¹⁷ Unfortunately, such testing shows a sensitivity of only 77% and specificity of 80% for CO poisoning. There may be practice effects as well if repeated testing is performed. Another study suggested that the degree of impairment CO patients had on a test of short-term rote and context-aided verbal memory correlated well with the number of HBO treatments needed.¹¹⁵ The biggest problem with such neuropsychiatric testing is that it is unclear if deficits in the test during the acute CO poisoning phase are at all predictive of the development of neurologic sequelae and therefore the necessity of HBO treatment.

Acute changes on CT scans of the brain occur within 12 hours of CO exposure that resulted in loss of consciousness.^108,120 Symmetric low-density areas in the region of the globus pallidus, putamen, and caudate nuclei are frequently noted.⁸⁸ Changes in the globus pallidus and subcortical white matter early within the first day after poisoning are associated with poor outcomes¹³⁶ (Fig. 125–1). Alternatively, in one series of 18 patients, a negative CT within a week of admission was associated with favorable outcome.¹⁸² The use of contrast may enhance early isodense changes not visible on initial CT scan²⁰⁶ but is not routinely performed.

MRI appears to be superior in detecting basal ganglia lesions after CO poisoning.¹⁰⁸ One study found a much higher incidence of periventricular white matter changes on MRIs done within the first day after exposure. However, such changes had no correlation with COHb level or cognitive sequelae.¹³⁶ These periventricular changes are more common and probably more sensitive than globus pallidus lesions. Globus pallidus lesions were present on MRI in only one patient (1.4%) in a prospective study of CO-poisoned patients, half of whom had loss of consciousness.⁸⁸ Diffusion-weighted MRI may have more promise in detecting changes in subcortical white matter within hours of serious CO poisoning.¹⁶⁵ Regardless, neuroimaging usually does not influence patient management and can be reserved for patients who show poor response or have an equivocal diagnosis.

The most promising area of neuroimaging after CO poisoning is in assessing regional cerebral perfusion.^37,135 Single-photon emission CT (SPECT) gauges regional blood flow noninvasively using an iodine or technetium tracer. In one series of 13 patients with delayed neurologic sequelae, all cases showed patchy hypoperfusion throughout the cerebral cortex within 11 days of poisoning.³⁵ These changes in perfusion may occur as early as one day after poisoning and primarily involve watershed regions such as the temporoparietooccipital area.⁴⁹ Perfusion defects on SPECT scanning appear to be associated with neuropsychological impairment months after serious CO poisoning.⁵⁸ Unfortunately, because of the scant availability of the procedure and the lack of comprehensive studies, SPECT scanning is not the definitive tool at this time for determining prognosis or need for HBO. In addition, when imaging patients 1 to 2 years post poisoning, it appears that T2 weighted imaging on MRI is more sensitive than SPECT scanning.³²

Positron emission tomography (PET) can also be used to evaluate regional blood flow as well as oxygen metabolism in the brain after CO exposure. In one series of severely CO-poisoned patients, PET examination after HBO treatment showed increased oxygen extraction and decreased blood flow in the frontal and temporal cortices.⁴⁷ Of note, patients with permanent deficits persisted in showing these abnormalities on PET scanning. One delayed PET study demonstrated that increases in dopamine D₂ receptor binding in the caudate and putamen after CO poisoning were improved with bromocriptine, at which time neuropsychiatric symptoms resolved.²⁰⁵ Although PET scanning cannot be used to predict outcome, abnormalities that persist on the scan may be indicative of patients with permanent neurologic sequelae.

To complement perfusion studies, EEG mapping has also been performed on CO-poisoned patients. Although initial studies demonstrate that many patients have regional EEG abnormalities after poisoning, it is unknown if these are predictive of persistent or delayed neurologic problems. EEG mapping may be discrepant relative to SPECT scanning because EEG preferentially demonstrates subcortical lesions.

The mainstay of treatment is initial attention to the airway. One hundred percent oxygen should be provided as soon as possible by either non-rebreather face mask or endotracheal tube. Although concerns have been raised regarding toxicity from excess oxygen, patients poisoned with CO can still have cellular hypoxia in spite of normal oxygen saturation.^12,72 It is important to remember that a non-rebreathing mask only delivers 70% to 90% oxygen; a positive pressure mask or an endotracheal tube is necessary to achieve higher oxygen concentrations. The immediate effect of oxygen is to enhance the dissociation of COHb.¹⁵⁰ In volunteers, the half-life of COHb is reduced from a mean of 5 hours (range, 2–7 hours) when breathing room air (21% oxygen) to approximately one hour (range, 36–137 minutes) when breathing 100% oxygen at normal atmospheric pressure.¹⁴¹ Actual poisonings show a range in half-lives of 36 to 137 minutes (mean, 85 minutes) when breathing 100% oxygen; the longer elimination half-lives appear to be most often associated with long, low-level exposures.^126,197 With oxygenation and intensive care treatment, hospital mortality rates for serious exposures range from 1% to 30%. The duration of treatment is unclear, with a valid end point being the resolution of symptoms, usually accompanied by a COHb below 5%.

Cardiac monitoring and intravenous (IV) access are necessary in any patient with systemic toxicity from CO poisoning. Hypotension can initially be treated with IV fluids; inotropes may also be necessary to treat myocardial depression. An evaluation for cardiac ischemia, including ECG and cardiac markers, should be considered in symptomatic patients at risk. Standard advanced cardiac life support protocols can be followed for the treatment of patients with life-threatening dysrhythmias. Patients with a depressed mental status should have a rapid blood glucose checked. Animal studies of CO poisoning suggest that hypoglycemia can be deleterious.¹³⁸ Correction of any acidemia with bicarbonate is controversial and may result in further cellular injury secondary to a left shift of the oxyhemoglobin dissociation curve, so it is not recommended unless the academia is profound and persistent.

HBO therapy appears to be the treatment of choice for patients with significant CO exposures.¹⁷⁸ But the most obvious effect, may not be the most important. One hundred percent oxygen at ambient pressure reduces the half-life of COHb from about 320 minutes to 85 minutes; at 2.5 atmospheres absolute (ATA), it is reduced to 20 minutes.^141,197 Actual CO-poisoned victims treated with HBO have half-lives ranging from 4 to 86 minutes.¹²⁶ HBO also increases the amount of dissolved oxygen by about 10 times, which is sufficient alone to supply metabolic needs in the absence of hemoglobin (Chap. 29).¹⁰ This is rarely an important clinical issue because most patients have already been stabilized and have appreciably decreased COHb with ambient oxygen alone before arrangements can be made for an HBO treatment.

Therefore, HBO is more than just a modality to clear COHb more quickly than ambient oxygen (Antidotes in Depth: A38). More importantly, in rats after loss of consciousness from CO exposure, hyperbaric, but not normobaric, oxygen therapy prevents brain lipid peroxidation.¹⁶⁶ HBO appears to prevent ischemic reperfusion injury by a variety of mechanisms. First, in animal models, HBO accelerates regeneration of inactivated cytochrome oxidase, which may be the initiating site for CO neuronal damage.¹² Unlike 100% oxygen at room pressure, in clinical trials, HBO was much more effective at restoring mitochondrial function within peripheral white blood cells in CO poisoned patients.⁵⁹ Second, HBO also prevents β-integrin mediated neutrophil adhesion to brain microvascular endothelium, a process essential for amplification of CNS damage from CO.¹⁷⁸ This may explain why HBO, but not 100% oxygen at atmospheric pressure, prevented delayed deficits in a learning and memory maze model.¹⁷⁰

Clinical studies of the effectiveness of HBO in preventing neurologic damage from CO are not as convincing as basic science studies would suggest. In uncontrolled human clinical series, the incidence of persistent neuropsychiatric symptoms, including memory impairment, ranged from 12% to 43% in patients treated with 100% oxygen and was as low as 0% to 4% in patients treated with HBO.^{69,114,127,131}

More recently, several controlled clinical trials have evaluated the efficacy of HBO in CO poisoning (Table 125–3). The first randomized study of CO poisoning included more than 300 patients and failed to show a benefit from HBO in patients who had no initial loss of consciousness.¹⁴⁴ Unfortunately, seriously ill patients were not randomized to surface pressure oxygen; they received either one or three treatments of HBO. Flaws in the study included significant delays to treatment and the use of suboptimal pressure of 2.0 ATA. A smaller (n = 60) controlled study avoided some of these flaws and showed that HBO decreased delayed neurologic sequelae at 3 to 4 weeks from 23% to 0% in CO-poisoned patients who presented without loss of consciousness.¹⁸⁰ However, patients with syncope, a marker of serious poisoning, were excluded. A very small study (n = 26) of patients presenting with Glasgow Coma Scores (GCS) above 12 after CO poisoning included almost half with loss of consciousness.⁵² Randomization to HBO versus 100% normobaric oxygen resulted in decreased EEG abnormalities and less reduction in blood flow reactivity to acetazolamide at 3 weeks. Unfortunately, all of these studies failed to definitively study all CO-poisoned patients, including those with syncope or coma.

The first randomized trial to directly address the issue of HBO efficacy in seriously CO-poisoned patients evaluated 191 CO-poisoned patients referred for HBO treatment.¹⁵² Patients were randomized to a minimum of three daily treatments of HBO (2.8 ATA for 60 minutes) or 100% oxygen at 1.0 ATA for 3 days. Although the HBO group had a higher incidence of persistent neurologic sequelae at one month, there was no significant difference between the two groups; more than two-thirds of each group had persistent problems. This study, although the largest controlled, randomized study to date, suffered from several flaws. Fewer than half of the patients had follow-up at one month. Disproportionate numbers of suicide cases (about two-thirds) and drug toxicity (44%), with accompanying neuropsychologic defects, could have confounded any beneficial effect from HBO. Finally, HBO treatment was delayed for 6 hours, making it much less likely to be effective.^69,144

A more recent randomized, double-blind, placebo-controlled study identified a beneficial effect of HBO in CO-poisoned patients.¹⁹⁶ Most of these patients were ill, with a mean initial COHb level of 25% and a 50% incidence of loss of consciousness. Patients were all treated within a 24-hour window after exposure, but the success of the study might be partially attributable to the rapid mean time to treatment of less than 2 hours. Patients received HBO three times at intervals of 6 to 12 hours, each at 2.0 ATA, except for the first hour of the first treatment, which was at 3.0 ATA. Control patients received sham treatments in the HBO chamber with 100% oxygen at 1.0 ATA. At 6 weeks, the HBO group had a 24% incidence of cognitive sequelae versus 46% in the control group. Based on these data, the number of patients needed to treat to prevent one case of cognitive impairment is only five. Critics of this study point out that the neuropsychiatric tests were not significantly different between the groups except for digit spam and trail making, and there was no difference in activities of daily living. However, untreated patients had increased self-reported memory problems at 6 weeks (28% vs. 51%), and the beneficial effect on cognitive sequelae lasted well into 12 months.

A more recent study of 179 patients with transient loss of consciousness after CO poisoning showed no benefit from a single HBO treatment.⁴ Neurological recovery at one month was approximately 60% regardless of HBO or normobaric oxygen. However, the study was done only at 2.0 ATA, which is below the customary initial dose for CO in positive studies. In addition, over 20% of patients were lost to follow-up.

Based on the strong animal and basic science experience, the positive human studies mentioned above, and few adverse effects, it is not surprising that the Underwater and Hyperbaric Medical Society (UHMS) recommends HBO for all CO patients with signs of serious toxicity.¹⁹⁵ With the low risk of this procedure,^151,156 almost 1500 patients are treated with HBO for CO poisoning in the United States each year.⁷² Therefore, HBO has become the standard of care for serious CO poisoning, even though there is substantial disagreement in the interpretation of the existing evidence.^83,109,201

Indications for Hyperbaric Oxygen Therapy.

Although specific indications for HBO after acute CO poisoning are listed (Table 125–4), they have not been prospectively evaluated. The patients most likely to benefit are those most at risk for persistent or delayed neurologic sequelae, such as those presenting in coma or with a history of syncope.¹⁹⁸ These may be clinical markers for the episode of hypotension that are necessary for causing neuronal damage from CO-induced ischemic–reperfusion injury in animal models.^65,133 However, syncope is neither particularly sensitive nor specific marker for cognitive sequelae. Patients with long exposures, or “soaking” periods, typically longer than 6 hours, are also at greater risk for neurologic sequelae.¹¹ The presence of a significant metabolic acidosis may be a surrogate marker.^157,186 Patients who present with decreased level of consciousness, a GCS <9 in one series, had an odds ratio of 7.0 for development of neurological sequelae.¹⁴⁰ Some authors advocate ongoing myocardial ischemia as an indication for HBO; however, in our experience, these patients usually already meet neurologic criteria for treatment, such as loss of consciousness or ongoing mental status changes. Isolated cardiac ischemia, more importantly, deserves immediate proven myocardial salvaging therapy rather than delayed treatment with an unproven therapy such as HBO.

Some authors advocate treating all patients with COHb levels of 40% or greater with HBO. Many HBO centers arbitrarily use a more conservative level of 25% as an indication for HBO.⁷⁸ More important than actual level are patient history and examination. Further analysis of data from the most recent controlled trial demonstrates that in patients not treated with HBO, there were no reliable factors (COHb level, loss of consciousness, or base excess) for predicting who progressed to cognitive sequelae.¹⁹⁸ This recent multivariate analysis showed that of all factors (loss of consciousness, age, exposure time, and COHb levels) only age of 36 years or older and CO exposure duration of 24 hours or longer predicted cognitive dysfunction at 6 week follow-up. More problematic is the incidence of cognitive sequelae in patients without those risk factors: 32% in those younger than age 36 years and 36% in those with less than 24 hours of exposure. In conclusion, it appears that there are no reliable predictors for screening out patients who will do well and not develop cognitive sequelae, and therefore can avoid HBO treatment with mild CO poisoning.

Multiple studies have looked for serum markers after acute CO poisoning that would predict neurological sequelae. COHb, although confirming exposure, does not correlate with future outcome, let alone acute symptoms.^75,198 Multiple serum markers, including cytokines, do increase after CO poisoning, but their predictive accuracy is unclear.¹⁷² Although impaired mitochondrial cytochrome function and elevated lipid peroxidation are seen in peripheral lymphocytes and monocytes after clinical CO poisoning, they only confirm CO exposure and are too nonspecific to use as predictors of neurological sequelae.⁵⁹

Therefore, at this time, it is prudent to refer for HBO treatment those patients with the most serious neurologic symptoms, regardless of their COHb level. Such symptoms include coma, seizures, focal neurologic deficits, altered mental status (GCS <15), and although controversial, loss of consciousness. Patients who have had cardiac arrest from CO poisoning and had the return of spontaneous circulation may be poor candidates for HBO therapy because all these cases have been fatal.⁸¹ In fact, such deaths from CO poisoning are no contraindication to organ donation.

Excluding patients from HBO with milder symptoms after CO poisoning may be problematic because even they are susceptible to neurocognitive sequelae. One series of 55 patients with mild poisoning as defined by absence of loss of consciousness and maximum measured COHb level less than 15% found that even one-third of these individuals had neurocognitive sequelae up to 12 months after exposure.³⁰ This was no different than that occurring in the severely poisoned group, although the milder group had a much longer duration of exposure as well as a greater delay to COHb level drawn. Brain imaging studies have confirmed that mild exposures, marked by no LOC and COHb levels lower than 15%, may result in visible changes.^58,143 Taken to its logical but impractical conclusion, because even apparently mild cases of CO poisoning may have poor neurocognitive outcomes, HBO treatment of every CO-exposed patient, regardless of severity, could be justified.

It is still unclear if mild neurologic symptoms, such as confusion, headache, dizziness, visual blurring, or abnormal mental status testing on initial presentation after CO poisoning, are prognostic for cognitive sequelae, which would necessitate HBO treatment. These symptoms simply represent CO poisoning, which, at COHb levels approaching 10% in volunteers, may cause temporary impairment of learning and memory.³ In one prospective clinical trial of CO poisoning, the incidence of cerebellar dysfunction portended a higher incidence of cognitive sequelae (odds ratio, 5.7 {95% confidence interval, 1.7–19.3}).¹⁹⁶ Therefore, difficulty with finger-to-nose, heel-to-shin, rapid alternating hand movements, or even ataxia, should be considered indications for HBO. Patients with other mild neurologic findings, such as headache, warrant at least several hours of oxygen by non-rebreather facemask until symptoms resolve. If symptoms do not resolve, HBO may be considered; however, any delay in HBO may decrease its efficacy.

A more promising method to discern patients who may respond to HBO may be the genotype, apolipoprotein E, specifically the isoform ε4.⁸⁹ This particular polymorphism allele is present in up to one-quarter of the population, and it is associated with worse neurologic outcome from trauma and stroke. In the presence of CO poisoning, it is associated with lack of response to HBO for preventing neurocognitive sequelae. Further studies may support not treating patients with this particular allele, focusing on those with the potential for response to HBO therapy.

Because of the confusion in determining which CO poisoned patients really need HBO treatment, several professional societies have developed evidence-based guidelines. As alluded to previously, the American College of Emergency Physicians has noted that no clinical variable can be used to predict patients at risk of cognitive sequelae and therefore most likely to benefit from HBO.²⁰¹ Similarly, the Cochrane Collaboration review on the use of HBO in CO poisoning concluded that because of so much conflicting data, there is insufficient evidence to support HBO for CO poisoning at this time.¹⁶ The collective odds ratio for protective effect at 4 to 6 weeks with HBO versus normobaric oxygen was 0.78 [95% CI 0.54–1.12] based on a collective experience of 1361 patients in six studies. The most recent guideline from the UHMS states that CO-poisoned patients should be referred for HBO if they have serious poisoning, such as unconsciousness, whether it is transient or persistent; age 36 years or older; or CO exposure duration of 24 hours or longer, even if intermittent.¹⁹⁵ This is all consistent with the prior studies discussed earlier. The UHMS guidelines also state that many physicians treat when neuropsychologic testing is abnormal or COHb levels are greater than 25% to 30%.

Some authors recommend selective use of HBO because of cost and difficulties in transport if the primary facility lacks a chamber. However, complications that may make such transfers and treatment unsafe are rare.¹⁵⁶ At the present time, we routinely recommend HBO for selected patients poisoned by CO based on the indications in Table 125-4. Fortunately, even without HBO, anywhere from one-third to three-quarters of cases with persistent cognitive sequelae resolve over the subsequent year.^34,196

Delayed Administration of Hyperbaric Oxygen.

The optimal timing and number of HBO treatments for CO poisoning is unclear. Patients treated later than 6 hours after exposure tend to have worse outcomes in terms of delayed sequelae (30% vs. 19%) and mortality (30% vs. 14%).⁶⁹ This may explain the failure of one of the first randomized trials on HBO in CO, which had a mean time to treatment of over 6 hours after poisoning.¹⁵² Meanwhile, HBO treatments delivered within 6 hours after poisoning in patients with loss of consciousness after CO seem to almost completely prevent neurologic sequelae.²⁰⁷ However, patients may benefit if they are treated even later. In the most recent randomized clinical trial showing beneficial effects of HBO, although all patients were treated within 24 hours of exposure, 38% of patients were treated later than 6 hours after exposure. Therefore, it is not unreasonable to consider HBO, contingent on transport limitations, within 24 hours of presentation for symptomatic acute poisoning.

One case series suggests beneficial effects for HBO used up to 21 days after exposure, even after patients have developed neuropsychologic sequelae.¹²⁷ The problem with studies showing HBO benefits days after an acute poisoning or after chronic poisoning is that these cases are all anecdotal and lack control subjects. In fact, delayed neurocognitive sequelae frequently resolve within 2 months in patients with mild CO poisoning,¹⁸⁰ and in those with serious CO poisoning who survive to HBO treatment, one-third resolve within one year.¹⁹⁶ It is possible that these delayed or chronic cases may simply represent the placebo effect of HBO.

Repeat Treatment with Hyperbaric Oxygen.

A randomized clinical trial demonstrated that three HBO treatments within the first 24 hours improved cognitive outcome.¹⁹⁶ Unfortunately, there was no group treated with only one or two HBO sessions in that study. Regardless, multiple treatments are advocated for patients who have persistent symptoms, particularly coma, and do not clear after their first HBO session. In a pilot study, one hyperbaric oxygen treatment was enough to promote almost total mitochondrial cytochrome activity, as measured in peripheral lymphocytes, after CO poisoning.⁵⁹ In a nonrandomized retrospective study, CO-poisoned patients who received a second HBO treatment had a reduction in delayed neurologic sequelae from 55% to 18% compared with control subjects who had only one treatment.⁶⁸

It is not clear that more HBO treatments are better. A large recent study showed that in seriously poisoned CO patients, two hyperbaric treatments resulted in a worse outcome than one treatment, with complete recovery 47% versus 68% respectively.⁴ There were serious flaws in that study, including lack of formal neuropsychological testing and the use of only 2.0 ATA of pressure, well below the 3.0 ATA used initially in favorable studies. With the lack of prospective studies comparing single versus multiple courses of HBO therapy, multiple treatments cannot be recommended routine at this time. The most recent clinical guidelines from the UHMS state that the optimal number of HBO treatments for CO poisoning is unknown and that one should consider reserving multiple treatments for patients who fail to fully recover after one treatment.¹⁹⁵

The management of CO exposure in the pregnant patient is difficult because of the potential adverse effects of both CO and HBO. A literature review of all CO exposures during pregnancy revealed a high incidence of fetal CNS damage and stillbirth after severe maternal poisonings.¹⁸⁸ A series of three severely symptomatic patients who did not receive HBO had adverse fetal outcomes: two stillbirths and one case of cerebral palsy.⁹⁹ There have even been cases of limb malformations, cranial deformities, and a variety of mental disabilities in children poisoned in utero.^18,110,111 A recent epidemiological study in Guatemala showed that CO exposure from wood smoke during the third trimester was inversely associated with neuropsychologi­cal performance at ages 6 to 7 years when corrected for socioeconomic confounders.⁵⁰

Traditionally, it was thought that fetal hemoglobin had a high affinity for CO. Pregnant ewe studies show a delayed but substantive increase in COHb levels in fetuses, exceeding the level and duration of those in the mothers.¹¹¹ Thus, it appeared that fetuses are a sink for CO and could be poisoned at levels lower than mothers. However, such data may not apply to humans because in vitro work shows that as opposed to sheep, human fetal hemoglobin actually has less affinity for CO than maternal hemoglobin, at a ratio of 0.8. Under conditions of low oxygenation and high 2,3-BPG, as in serious CO poisoning, the affinity of human fetal hemoglobin starts to approach that of maternal.¹⁹⁹ The more important issue with maternal CO exposure is the precipitous decrease in fetal arterial oxygen content that occurs within minutes at CO concentrations of 3000 ppm.⁶⁴ Therefore, the ensuing hypoxia of the fetus, rather than increase in fetal COHb, may be of more concern.

Maternal COHb levels do not accurately reflect fetal hemoglobin or tissue levels.⁴⁴ In primate studies, a single CO exposure insufficient to cause clinical disease in the mother led to intrauterine hypoxia, fetal brain injury, and increased rate of fetal death.^64,65 In humans, there are a few cases of fetal demise with maternal levels of COHb less than 10%.¹⁸ However, in that series, some mothers were treated with oxygen before obtaining their COHb levels. Another issue with some of these data is that often the mother has been chronically “soaked” with CO, making levels difficult to interpret. Rodent studies show that chronic low level CO exposure in pregnant mothers may result in permanent cognitive deficits in the subsequent progeny.⁴⁸

Because maternal COHb does not necessarily predict fetal demise, clinicians must direct their attention to maternal symptoms of CO toxicity. Multiple case series demonstrate that pregnant women who present with normal mental status and no loss of consciousness after CO poisoning have excellent outcomes in terms of normal deliveries.^18,99 These infants have no subsequent delay in attaining their developmental milestones. Therefore, it appears that mothers who appear well after acute CO poisoning will have good outcomes with respect to their pregnancies.

The bigger dilemma for clinicians is the approach to treatment of seriously symptomatic CO-poisoned pregnant patients. All patients should receive 100% oxygen by facemask, at least until the mother is asymptomatic. However, CO absorption and elimination are slower in the fetal circulation than in the maternal circulation.¹¹¹ A mathematical model predicts that elimination of CO from fetuses takes 3.5 times longer than maternal CO elimination.⁸⁶ However, based on the fact that some of these data are based on sheep fetal hemoglobin kinetics, the optimum time for treatment of the mother cannot be recommended at this time.

For exposed pregnant women with a loss of consciousness or high COHb levels, HBO might be considered. Unfortunately, pregnant patients were excluded from all prospective trials documenting efficacy of HBO. However, treatment of pregnant patients with HBO is not without theoretical risk. Animal studies show conflicting results on the effects of HBO on fetal development. Some studies have shown that HBO causes developmental abnormalities in the central nervous, cardiovascular, and pulmonary systems of rodent fetuses. This is in marked contrast to the extensive Russian experience, in which hundreds of pregnant women were treated with HBO, apparently without significant perinatal complications and with improvement in fetal and maternal status for their underlying conditions of toxemia, anemia, and diabetes.¹²¹ Cases in the United States have been published in which HBO used for mild CO poisoning resulted in infants who were normal at birth. However, less than optimal outcomes have occurred in cases of sicker patients in which the mother has had loss of consciousness or presented comatose.⁵⁴ Thus, it appears that HBO should be safe and have the same efficacy for pregnant patients as in nonpregnant patients. However, its effect in preventing adverse fetal outcomes is unclear.

There currently is no scientific validation for an absolute level at which to provide HBO therapy for a pregnant patient with CO exposure. Arbitrarily, COHb levels greater than 20% are recommended as an indication in a pregnant patient regardless of symptoms. Pregnant patients should not be treated any differently if they meet criteria for HBO that have already been mentioned (Table 125–4). Additional criteria include any signs of fetal distress, such as abnormal fetal heart rate.

It has been suggested that children are more sensitive to the effects of CO because of their increased respiratory and metabolic rate.³³ Epidemiologic studies suggest that children can become symptomatic at COHb levels less that 10%, which is lower than commonly expected in adults.⁹⁸ The other problem is that these patients may have unusual presentations. Although most children manifest nausea, headache, or lethargy, an isolated seizure or vomiting may be the only manifestation of CO toxicity in an infant or child.

While obtaining COHb levels in infants, clinicians must be aware of two confounding factors. First, many cooximeters give falsely elevated COHb levels in proportion to the amount of fetal hemoglobin present.¹⁹⁰ Second, CO is produced during breakdown of protoporphyrin to bilirubin. Therefore, infants normally have higher levels of COHb, which are even higher in the presence of kernicterus. Some neonates not exposed to CO can have COHb cooximetry readings approaching 8%.¹⁹¹ Thus, before it is assumed that an elevated COHb level implies CO poisoning in an infant, the contribution of jaundice and fetal hemoglobin must be considered in the final analysis.

Although children may be more susceptible to acute toxicity with CO, their long-term outcomes appear to be more favorable than adults. In a series of 2360 serious CO cases, all incidences of delayed neurologic sequelae were in adults older than age 30 years.³⁴ Pediatric series of CO poisoning demonstrated an incidence of delayed neurologic sequelae of 10% to 20% of children after severe CO poisoning.³³ This low incidence, in patients treated only with 100% oxygen at 1.0 ATA, has been used as an argument to avoid HBO. However, there still is a real risk of such sequelae, and HBO has been used successfully to prevent it.²⁰⁴ Children do well with HBO treatment, with mortality mainly related to concomitant smoke inhalation if present.³⁶ If the use of surface-pressure oxygen is selected to treat a child, it is comforting to know that the COHb half-life is approximately 44 minutes, faster elimination than that seen in adults.⁹⁸ Often, children exposed to CO in similar circumstances, but appear well, are treated simultaneously with the sick parent, especially if a multiplace chamber is available.

A variety of neuroprotective agents have been tested in animal models. They are targeted primarily at preventing the delayed neurologic sequelae associated with serious CO poisoning. One of the simplest treatments tested is insulin. Hyperglycemia exacerbates neuronal injury from stroke as well as in arrest situations. In CO poisoning of rodents, it is associated with worse neurologic outcome.¹³⁸ However, insulin, independent of its glucose-lowering effect, may be the protective agent after ischemia. In rodent studies, improved neurologic outcome, as measured by locomotor activity, occurs after those with CO poisoning treated with insulin. In light of these findings, it is reasonable to aggressively treat documented hyperglycemia with insulin in patients with serious CO poisoning.

Many neuroprotective agents involve blockage of excitatory amino acids that are implicated in neuronal cell death after CO poisoning. Pretreatment of mice with dizocilpine (MK-801), which blocks the action of glutamate at N-methyl-d-aspartate receptors, ameliorates learning, memory, and hippocampal deficits with CO poisoning.⁹³ Ketamine, another glutamate antagonist, decreases the mortality rate of rats poisoned with CO after carotid ligation.¹³⁹ Treatment of mice with various glutamate antagonists prevents learning and memory deficits in a model of CO poisoning.⁶² Blockage earlier in the immunologic cascade, with a neuronal NO synthase inhibitor also prevented NMDA receptor activation, thus protecting mice from learning deficits after CO poisoning.¹⁷⁵ One exciting approach is the use of antioxidants, such as dimethyl sulfoxide and disulfiram, that prevent learning and memory deficits when given after CO poisoning in mice.⁶² Use of these or related therapeutics, although promising, awaits further animal testing because of potential adverse effects.

Other modalities have been tested in preventing neuronal damage from CO without much success. Hypothermia, rather than being beneficial, actually increases mortality in animals.¹⁶² Allopurinol which prevents formation of free radicals through xanthine oxidase inhibits lipid peroxidation in CO poisoning when given prior to exposure.¹⁶⁸ This strategy has not been promising because of the necessity for pretreatment.

Early diagnosis prevents much of the morbidity and mortality associated with CO poisoning, especially in unintentional exposures. The increased quality of home CO-detecting devices allows personal intervention in the prevention of exposure.¹⁰¹ If a patient presents complaining that his or her CO alarm sounded, it is important to realize that the threshold limit for the alarm is set roughly to approximate a COHb level of 10% at worst. Therefore, manufacturers must have their alarms activate within 189 minutes at 70 ppm CO, 50 minutes at 150 ppm, and 15 minutes at 400 ppm (Underwriters Laboratories, UL2034). Alarms are not to activate for prolonged exposures below 30 ppm to prevent epidemic alarms during winter thermal inversions in large cities.⁹ Government ordinances for obligatory CO alarms could potentially prevent many poisonings, particularly during winter storms.^22,70 Although most serious CO poisonings are associated with the absence of CO alarms, a recent series of such patients showed that even with alarms, patients become seriously poisoned.³⁸

Routine laboratory screening of ED patients during the winter is not very efficacious in diagnosing unsuspected CO poisoning; the yield is less than 1% when patients are tested in whom the diagnosis of CO exposure was already excluded by history. Instead, selecting patients with CO-related complaints, such as headache, dizziness, or nausea, increases the yield to 5% to 11%.^53,161 During the winter, risk factors such as gas heating or symptomatic cohabitants in patients with influenzalike symptoms such as headache, dizziness, or nausea, particularly in the absence of fever, is the most useful method for deciding when to obtain COHb levels for potential patient.

The issue of symptomatic cohabitants is especially important from a preventive standpoint. Alerting other cohabitants to this danger and effecting evacuation may prevent needless morbidity and mortality. This is especially critical for multifamily domicile, like hotels, that have resulted in dramatic collective exposures and even deaths.¹⁹⁴ Most communities have multiple resources for onsite evaluation. Usually the local fire department or utility company can either check home appliances or measure ambient CO concentrations with portable monitoring equipment. Current workplace standard for ambient CO exposures is 35 ppm averaged over 8 hours with a ceiling limit of 200 ppm (measured over a 15-minute period).¹²⁹ Just a 4-hour exposure to 100 ppm of CO may result in COHb level greater than 10% with symptoms.

• Unintentional exposures to CO are easily missed or misdiagnosed. Patients with a suspected influenzalike illness should be screened for potential home sources of CO, and symptomatic cohabitants should be alerted.

• CO exposure should also be considered in patients with unexplained coma, acidosis, or signs of cardiac ischemia, especially if attempted ­suicide is suspected.

• Fire victims, in addition to airway complications and potential cyanide toxicity, may succumb to CO toxicity.

• The mainstay of treatment in CO poisoning is good supportive care with early oxygenation to increase the elimination of COHb.

• Because of the overwhelming clinical successes with HBO and its limited risks, early use of this treatment modality in severe exposures is encouraged.

• Discussion with a regional poison center or hyperbaric facility will help in identifying patients who are most likely to benefit from such treatment.

References