Carbon monoxide (CO) is an odorless, tasteless, colorless gas that is formed from the incomplete combustion of most carbon containing compounds. CO is extremely difficult to detect in the environment without the use of a specific detector, secondary to being odorless and tasteless. The US Centers for Disease Control and Prevention (CDC) estimates 50,000 annual emergency department visits are due to CO poisoning.  Approximately 400 people die per year of nonintentional, non-fire–related CO poisoning.
The most common sources of CO include exhaust systems, furnaces, water heaters, kerosene space heaters, boats with engines, cigarette smoke, gasoline or diesel powered generators, gas ranges, and fireplaces. CO is also found in mining explosions, methylene bromide, methylene chloride, and anesthetic absorbents. CO poisonings are more common in the winter months and following natural disasters.
CO is readily absorbed following inhalation. CO binds to hemoglobin with an affinity of 200 times greater than that of oxygen. This leads to the inability of hemoglobin to deliver oxygen to the cells. CO also causes a shift in the oxyhemoglobin dissociation curve to the left, decreasing the offloading of oxygen from hemoglobin at the cellular level. 
The toxicity of CO cannot be completely explained by these hypoxic-ischemic effects. The potential long-term neurologic deficits cannot be predicted based purely on the CO level, as cognitive deficits are seen at low levels of CO as well as elevated ones. CO also causes cellular toxicity secondary to cytochrome malfunction. Cytochrome oxidase binds to CO leading to prolonged mitochondrial dysfunction.
These effects on cytochrome oxidase are thought to only be the start of the inflammatory cascade that leads to the ischemic reperfusion injury seen in the brain, particularly in the hippocampus and corpus striatum. Brain lipid peroxidation from CO is thought to be the effect of the two above theories of hypoxia-ischemia and cytochrome dysfunction. Cytochrome oxidase malfunctioning can lead to brain lipid peroxidation along with the fact that lipid peroxidation is more dependent on perfusion and hypotension versus actual CO levels.  CO also causes activation of excitatory amino acids, which ispostulated tocause subsequent neuronal cell loss.
Early CO poisoning presents with various nonspecific symptoms, which are often confused for a viral syndrome. Symptoms of CO poisoning include headaches, nausea, vomiting, dizziness, and chest pain. As the level increases dysrthymias, myocardial ischemia, syncope, vision changes, ataxia, and changes in mental status can be seen.
A syndrome, often referred to as delayed neurologic sequelae, consists of persistent or delayed cognitive effects of CO poisoning, including attention deficits, ataxia, apraxia, chorea, peripheral neuropathies, psychosis, agnosia, incontinence, and cortical blindness. Delayed effects are usually seen within 30 days after CO exposure and are most commonly associated with a loss of consciousness at the time of the initial exposure. These delayed sequelae are likely a result of white matter lesions. 
Obtaining a carboxyhemoglobin (COHb) level is the most useful test to determine CO poisoning. An arterial blood gas sample is classically indicated; however, a venous blood sample has been shown to be as accurate in measuring a CO level. Normal levels of COHb in nonsmokers range from 0-5%. Smokers can have COHb levels ranging from 2-10%. People who are chronically exposed to CO at their place of employment, such as heavy machinery operators, and taxi drivers may have elevated levels at baseline (up to 12%).  Estimated ranges on COHb levels and the expected symptoms are shown below; however, levels are not always predictive of outcomes: 
COHb level 20-30% - Headaches, nausea, vomiting, confusion
COHb level 30-40% - Dizziness, vision problems, tachycardia, tachypnea, hypotension
COHb level 50-60% - Loss of consciousness
COHb level >60% - Seizures, coma, death
COHb levels are usually measured with co-oximetry. A co-oximeter is a device that is able to read the percentage of hemoglobin that is bound/saturated with CO. This device functions using spectrophotometrics. Spectroscopy uses various wavelengths of light to determine the quantitative measurement of a substance. Various wavelengths of light are exposed to specimens and depending on the amount of absorption by the specimen a concentration of a particular substance is calculated.
Pulse oximeters use a similar technique to measure the percent of hemoglobin saturated with oxygen noninvasively. Most pulse oximetry machines confuse CO for oxygen and give falsely elevated oxygen saturation levels. This is because most pulse oximetry machines only measure two wavelengths of light. Therefore, do not depend on the O2 saturation to evaluate the oxygenation status of the patient. 
Newer forms of pulse CO-oximeters have been produced that are able to measure a COHb level without a blood draw. The evidence for the accuracy of these hand-held CO oximetry devices has shown variable results; however, they show some promise as screening tools. Piatkowski et al found that CO oximetry was accurate as long as the sensor was positioned correctly and the fingertips were clean.  Nilson et al found that CO oximetry may prove to be a helpful screening aid for CO poisoning.  Ruppel et al found the CO oximetry had a wide variability range and may be helpful in screening.  Feiner et al used the bedside CO oximetry meter for methemoglobin levels and stated it became inaccurate as oxygen saturation dropped below 95%. 
As mentioned above, CO is odorless and tasteless, making it difficult to detect in the air. An increase in awareness of CO poisoning has led to an increase in the presence of CO detectors in personal homes. However, the National Fire Protection Association states that although more than 90% of homes have a smoke detector, only 15% have a CO detector.  These detectors function similar to a smoke detector and sound an alarm when a dangerous level of CO is detected in the home. When a CO detector alarms, the home must be evacuated and medical attention must be sought. Emergency services should be dispatched to evaluate the environment using more sophisticated detectors.
Myelin basic protein (MBP) in cerebrospinal fluid (CSF) is an investigational marker for CO poisoning. Toshimitsu et al found in a small case study that although an invasive technique of CSF sampling was needed, MBP concentrations could predict the development of delayed sequelae.  Photon-emission tomography (PET) scanning has also been suggested in studies to help differentiate white matter lesions with more accuracy. 
Treatment & Management
The most important initial treatment of CO poisoning is 100% oxygen. The half-life of CO in breathing room air is approximately 5 hours. This is reduced to 1 hour if breathing 100% oxygen at normal atmospheric pressure. It is further reduced to 20 minutes if breathing 100% oxygen in a hyperbaric oxygen chamber (HBO) at 2 atm.  HBO is the treatment of choice for significant CO poisonings. Not only does HBO clear COHb more quickly than 100% oxygen but it is proposed to prevent or improve delayed neurologic sequelae. Nultiple academic societies have indications for HBO, none of which have been prospectively validated. Most common indications for HBO treatment in CO poisoning include syncope, coma, seizures, confusion/altered mental status, age older than 36 years, fetal distress in pregnancy, prolonged CO exposure for greater than 24 hours, or a COHb level of more than 25%. 
Multiple studies have evaluated the use of HBO for delayed neurological sequelae in the CO poisoned patient. Few have been randomized, controlled studies. The set-up and definition used in the studies have varied, making them difficult to compare. The studies have had large bias and flaws, making adequate conclusions difficult. Also, the results of the studies have provided conflicting evidence on the efficacy of HBO therapy.
Annane et al found no benefit of HBO versus normobaric oxygen;  however, severely poisoned patients were not randomized. Severely poisoned patients automatically received HBO, so this does not disprove a positive effect in severely poisoned patients, only in patients with initial low CO levels. Thom et al found benefits from the use of HBO;  however, many questions were raised about the statistical analysis in this research. Scheinkestel et al found no benefit from HBO treatments; however, less than half of their enrollees followed-up.  Weaver et al found benefit for HBO on delayed neurologic sequelae; although it had its limitations, this study was the best set-up among current publications. [15, 16] However, such varying results indicate that more research is needed to determine the role for HBO in CO poisoning.