Because of the nonspecificity of the symptoms, CO poisoning is often underdiagnosed or misdiagnosed as another illness.^1,4 In 1 study, approximately 30% of the CO poisoning cases were diagnosed incorrectly, and 43% of those were misdiagnosed as food poisoning.^8,9 Up to half of CO poisoning patients receiving treatment in ED may not receive an accurate diagnosis.^{8,9,37,40–42} Also, a substantial proportion of CO exposure cases result in on-site treatment and do not require stay at a health care facility.⁴³ These cases are unlikely to be captured. Furthermore, not all accounts of disaster-related CO poisoning cases are published, and therefore we excluded an unknown number of cases from this review.

Surveillance of CO cases is often limited to the geographic area that is most affected by disaster and excludes the population in other less affected areas. For example, although there were large scale power-outages over substantial periods of time in many states after Hurricane Katrina, only 5 gulf coast states (Alabama, Mississippi, Louisiana, Texas, Florida) with heavy Katrina damage had accounts of CO poisoning cases.^20,21,31,36 Similarly, during a 2009 ice storm in Kentucky, the CO poisoning surveillance focused on areas that were most greatly affected.³⁹

In the aftermath of a disaster, it is possible that some of the institutions that provide data (e.g. health care facilities, hyperbaric chambers) become nonoperational or operate under limited capacity. Therefore, surveillance and reporting of CO poisoning cases may be restricted. In this review, the number of facilities providing data on disaster-related CO poisoning varied (range = 1–109), but the median number was only 2 with 70% of the articles including cases from 4 or fewer facilities. Reporting by a handful of facilities may fail to include some of the CO poisoning cases.

Disaster-related CO poisoning surveillance is often conducted as a part of emergency response or preparedness. CO poisoning cases are, therefore, not reported until the apparent “emergency” or “crisis period” is over. Among the 25 articles that included information on case ascertainment period, the median was 10 days (range = 2–93 days) with 73% being 2 weeks or less. However, power outages and engagement in high-risk behaviors (e.g., improper placement of generators, fuel powered clean-up equipment use) might have continued, albeit at a lower frequency, past the initial postdisaster period.

No standard case definition was followed in identifying CO poisoning cases. Different facilities (e.g., hospitals, poison centers, HBOT facilities) used varying case definitions based on International Classification of Disease codes, symptoms and a positive history of CO exposure, or COHb levels, and sometimes based on a combination of 2 or more of the above. Some articles did not provide a case definition. This might have led to misclassification, underreporting, and even overreporting of some cases. The Council of State and Territorial Epidemiologists defines confirmed cases of CO poisoning as cases with COHb levels of 12% or more.⁴⁴ Although the levels of reported COHb levels varied (range = 0.2%–49.8%) in this review, the mean COHb level was 17.8% and the median was 14.1%. The median indicates that it is likely that the majority of cases included in this review were true cases of CO poisoning.

Some of the articles focused on specific populations,³⁷ specific types of events,¹⁷or exposures.²² This may result in underreporting of cases.

The selection of geographic area and population under surveillance, data collection period, or inclusion of health care or other facilities for disaster-related CO poisoning surveillance depends on, among other factors, the type and extent of disaster, the availability of resources and infrastructure, and the evidence-driven best judgment of public health personnel. However, standardization of case definition, data collection methods and tools, and the surveillance system will help to better understand the exact burden of disaster-related CO poisoning, identify high-risk populations, and develop appropriate prevention messages. Given the impediments and disruptions in the wake of a natural disaster, it is likely that the development of a well–laid-out disaster response plan, involvement of and partnership with stakeholders (e.g., disaster management personnel, media), identification of data sources, establishment of effective communication channels, and data sharing agreements (e.g., with hospitals, emergency medical services, medical examiner’s offices) before disaster onset will result in a more coordinated and effective public health effort.

Female individuals and children (< 18 years) constituted the majority of nonfatal cases whereas 79% of individuals who died were male. These findings are consistent with previous reports.^1–3 Higher mortality among men has been assumedly a result of engaging in high-risk behaviors such as using fuel-burning tools or appliances.^2,3 Biologically, children are more susceptible to CO poisoning because of higher basal metabolic rates and tissue oxygen demands. They manifest symptoms early but are also likely to recover sooner as a result of higher ventilation rates.^45,46 The higher nonfatal poisoning observed among women is likely because they, similar to children, manifest symptoms at lower levels of CO exposure because of a lower red blood cell count.⁴⁷ These factors might lead to earlier exposure recognition and, therefore, a lower exposure and shorter recovery time among women and chidren.⁴⁸

Twelve articles provided data on the race/ethnicity of the CO poisoning cases. According to the 2000 US Census, Whites constituted 81%, Blacks 12%, Asians 4.2%, and Hispanics 3.5% of the US population.⁴⁹ Racial/ethnic minorities represent a disproportionate fraction of both fatal and nonfatal disaster-related CO exposures. For example, after a 2002 ice storm in North Carolina, Hispanics accounted for 23% of all injuries and 65% of all CO exposures but represented only 5% of the population.¹⁴ Also, of the 7 studies that assessed the English language proficiency of the CO poisoning cases, 5 found that in more than 20% of cases (range = 23%–56%), the individual was not fluent in English or English was not the primary language spoken in the household. Excessive CO-related morbidity or mortality among minority, immigrant, or non–English-speaking populations has been reported by some studies.^{14–16,19,27,37,39} High-risk behaviors for CO poisoning, such as use of charcoal grills, have also been associated with racial/ethnic or cultural origin.^10,15,50,51 These findings have significant public health implications in terms of targeted intervention and dissemination of multilingual prevention messages to prevent disaster-related CO poisoning among racial/ethnic and linguistic minorities. It is of note that the reporting was not uniform or complete in most studies; therefore, careful and continuous review of the evidence is needed to determine disparities in disaster-related CO poisoning.

Generator use was implicated in most of the fatal and nonfatal cases and was the primary source of CO exposure for all posthurricane studies (14 articles). Nearly 67% of the fatal cases resulted from using the generator indoors or in a basement. Indoor placement is particularly dangerous because CO content in the exhaust from a typical portable generator (5.5 kW) is equal to that of 6 idling automobiles.^12,29,52,53 In many cases, placing a generator outside in an attached garage, near open doors or windows, or near the air conditioning vent can still put residents at risk for CO poisoning. Accordingly, 10 of the 30 fatalities in this review were from generators placed outside. Although the recommended distance for the placement of a generator outside of a home is now recognized as a minimum of 20 feet,^54,55 the public should be encouraged to place generators as far from their homes as possible, but also at a safe distance from any nearby dwellings. Approximately 17% to 31% of disaster-affected populations have reported using generators for restoring power after disasters.^56,57 Communicating the risks of improper generator placement can impact a large segment of the population. Still, the need for electricity, fear of theft, lack of portability or of an adequately long electric extension cord, and the risk of operating a generator in wet conditions can all pose challenges to the safe operation of portable generators.²⁹

Indoor use of charcoal grills was also a major source of disaster-related CO exposures, particularly for winter storms. It has been suggested that persons of Asian, Middle Eastern, or African origins or persons from warmer climates where solid fuel burning indoors for cooking or heating purposes is common might be inclined to use charcoal briquettes or grills indoors.^{10,15,37,50,51,58} Differences in CO exposure source by disaster type can guide development of disaster-specific prevention messages that might also lead to differential regional strategies for prevention. The Northeast and the Midwest regions of the country are more likely to experience adverse CO-related health events from ice and winter storms or blizzards, whereas the South experiences more hurricane-related events. Although federal mandates require CO warning labels on both generators and charcoal bags,^10,59,60 multilingual education or warning materials distributed with sales of generators and charcoal bags might be helpful in targeting specific minority populations and reducing the risk of CO poisoning. Also, most CO poisoning cases occurred in residential settings, highlighting the importance of having battery powered or battery backup CO alarms in homes to prevent CO poisonings during power outages after disasters and in nonemergency situations.^19,61 Installing a battery powered or battery backup CO alarm at home is mandated by many state and local governments and is a primary recommendation from CDC for the prevention of CO poisoning.⁵⁵

Predisaster risk communication might result in better public health effectiveness in reducing disaster-related CO exposures, because most cases occur within days of event onset, and most natural disasters and subsequent high-risk behaviors (e.g., generator use during power outages) are quite predictable. Absent a generator engineering solution (e.g., auto shutoff, integrated CO alarm) and more effective policies requiring CO alarm installation, surveillance and risk communication are currently the 2 major components of disaster-related CO poisoning prevention. CDC's current national communication strategy for CO prevention is based on 3 components: (1) improving awareness of the dangers of CO poisoning, (2) advocating the use of CO alarms, and (3) promoting proper use and maintenance of fuel-burning appliances (including generators and furnaces). A key component of CDC's communications efforts has been targeting of racial/ethnic and linguistic minorities through the sharing of basic guidelines on UNFR carbon monoxide poisonings and other materials available in 17 languages.⁵⁵ These materials are made available to state and local public health partners for appropriate use in their communities on both an emergency and nonemergency situations. Recently, in cooperation with a range of public and private partners, CDC conducted outreach via Spanish-language radio stations, prepared video footage adaptable to multiple languages, and developed other materials designed for diverse audiences.⁵⁵ This communication model could be effective in predisaster risk communication. Predisaster risk communication should take into account the racial/ethnic and linguistic makeup of at-risk population and effectively utilize the local mass media (e.g., local non-English newspapers, radio, or TV stations) and community platforms (e.g., shopping centers, community centers).

Despite the magnitude of the health burden posed by CO poisoning among postdisaster populations, the public does not always perceive CO poisoning as a major health concern or is not aware of safe health practices.^56,62,63 Although evidence suggests that receiving health information on CO poisoning before, during, and after a disaster might lead to safer health practices, information is sometimes not widely received by affected populations.^23,26,57 Effective communication in pre- and postdisaster situations, overall, is critical to improving public understanding, awareness, and cooperation. These findings emphasize the continued need for communicating CO-related health information and surveillance of CO poisoning cases as a central component of public health emergency preparedness, response, and prevention efforts during natural disasters.