Carbon monoxide poisoning in the 21st century

The long-term consequences in survivors can range from severe brain damage (which is fortunately uncommon) to a much more common syndrome of less severe but persistent problems. Neurological sequelae are often divided into persistent neurologic sequelae (PNS) and delayed neurologic sequelae (DNS) [1, 21]. The incidence of neurological sequelae depends very much on the definition applied, the extent and timing of the assessment, and the population studied.

The concept of persistent neurological injury after a hypoxic brain injury is straightforward. DNS, in contrast, lacks a consistent definition, diagnostic criteria, or an established mechanism. However, the apparent development of the first neuropsychological symptoms or signs occurring days to weeks after CO poisoning clearly occurs. DNS varies very widely in studies from a few percent to two thirds of patients [22–25]. However, the ratio of DNS to PNS appears much lower when studies are prospective and careful monitoring is done from the beginning (that is, DNS may sometimes reflect delayed diagnosis rather than delayed appearance of symptoms).

Neurological sequelae are most commonly subjective and affect mood, short-term memory, attention, and concentration. The most common problems encountered are depressed mood (even in those accidentally exposed) and difficulty with higher intellectual functions (especially short-term memory and concentration). More severe problems include areas typically affected by ‘watershed’ infarcts (for example, basal ganglia and memory). In some cases, these are not noted initially but present later after initial recovery (typically within a week of the exposure). Neuro-psychological testing may be useful to provide objective measures of subtle deficits not found with routine bedside mental state examination and also to monitor the progress of these sequelae. Long-term follow-up is necessary in those at risk, as more subtle defects can develop or become apparent over a few weeks to months. However, the long-term prognosis is favorable in the majority of cases, and symptoms gradually resolve over the first few months [26], and the overwhelming majority of patients with CO poisoning return to full-time work [27].

Identification of patients who are at risk of neurological sequelae serves an important role in terms of counseling and indicating the extent of follow-up warranted. The best identified risk factors for long-term neurological effects are early and obvious neurological damage or a sustained loss of consciousness during the CO exposure. Most studies have found that significant neuropsychological sequelae are confined largely to those who have loss of consciousness at some stage [27–29]. However, if less stringent criteria are used for neurological sequelae (that is, slightly low test scores), other risk factors are thrown up by univariate analysis (for example, prolonged or repeated exposures and older age). These risk factors may also represent confounding, or reverse causality (for example, older age is linked to a higher risk of poor memory and executive function irrespective of CO poisoning; and impaired cognition prior to exposure is a risk for prolonged or repeated CO exposures) [28]. The recent promulgation of such criteria as age of more than 36 years (irrespective of the absence of other more established risk factors) to guide risk assessment [26, 28] or to alter treatment has little to recommend it; it greatly inflates the numbers perceived to be at risk and goes against the much stronger evidence of a relatively benign long-term outcome for CO poisonings without the established risk factors.

However, other objective ways to identify patients at risk of sequelae are also required, as the history of loss of consciousness may be complicated in some settings. Neuron-specific enolase (NSE) is a glycolytic enzyme that is localized primarily to the neuronal cytoplasm in the central nervous system. S100B is a calcium-binding protein localized to astroglial cells [30]. They are both released after hypoxic damage as a result of neuronal and astroglial cell death [31]. These markers show considerable promise as intermediate outcome measures for brain injury in both animals [32] and humans [19].

Studies to date in acute CO poisoning confirm this promise. In one recent Taiwan study, 10 out of 71 patients developed DNS. These patients not only had longer loss of consciousness but also had 15-fold higher S100B levels. Further statistical analysis demonstrated that this was an independent predictor of the development of DNS after acute CO poisoning: serum S100B of more than 0.165 μg/L predicted DNS with a sensitivity of 90% and a specificity of 87% (odds ratio 121, 95% confidence interval 4 to 3,467) [19]. The timing of S100B measurement is critical in interpretation, and high initial levels were associated with coma and cardiac injury but these dropped fourfold within 6 hours [33]. This and other studies have reported generally lower levels and only a minor association with CO poisoning with loss of consciousness but without such sequelae [34]. Further larger and long-term studies including more people with severe poisoning are required to clarify the optimal timing and threshold and the extent to which a normal S100B can be regarded as reassuring with respect to long-term prognosis. In general, results for NSE have shown a less obvious relationship to severity than S100B [33, 34], although one study found that it was better linked to level of consciousness and also had a longer apparent half-life [30]. No studies to date have examined the relationship between NSE elevation and long-term sequelae.