Over the 14-year study period, we observed 172,489 poison center exposure cases in 50 states, 105,833 ED visits in 16 states (representing 42.9% of the US population), 19,230 hospitalizations in 26 states (representing 60.1% of the US population), and 14,448 deaths in 50 states due to unintentional carbon monoxide poisoning. Per 1 million, these represent 39.5 exposure cases, 56.5 ED visits, 7.3 hospitalizations, and 3.3 deaths due to unintentional carbon monoxide poisoning during 2005 to 2018(Table 1). Non–fire-related carbon monoxide poisoning annual estimates are 5,593 ED visits, 825 hospital visits, and 422 deaths in the United States annually.

Carbon monoxide exposures were most frequently reported among children aged 0 to 9 years in poison center exposure data (17.6%). ED visits were high for young adults aged 20 to 29 years (17.1%) and 30 to 39 years(15.8%). Adults aged 50 to 59 years represented the highest proportions for hospitalizations (19.9%) and deaths (17.3%) (Table 1). Children aged 0 to 9 years had the highest reported rates in poison center exposure and ED visit data (54.1 and 70.5 cases per 1 million, respectively). Conversely, adults older than 80 years had the highest hospitalization rates and deaths (20.2 and 9.9 per 1 million, respectively) (Table 1). Although women comprised more than half of poison center exposure calls (51.8%) and ED visits (51.3%), men were more prevalent among reported hospitalizations (56.6%) and deaths (66.3%) due to unintentional carbon monoxide poisoning. The northeast US census region represented the highest rate of poison center exposure calls (57.2 per 1 million), and the Midwest had the highest mortality rate (5.0 per 1 million).

Non–fire-related cases comprised approximately 74.0% ED visit data, 60.1% of hospitalization, and 40.9% of deaths compared with other unintentional causes. Across all data sources, unintentional carbon monoxide poisonings peaked in the winter season (December to February).

In the poison center exposure case data, residential (N=138,493; 80.3%) and workplace exposure sites (N=16,369; 9.5%) were most frequently reported regardless of the management site (on-site versus health care facility) (Table 2). Of the cases with available information, almost half of the poison center exposure calls were treated at a health care facility (45.9%). The minor effect was the most often reported medical outcome treated at a health care facility. Major and moderate effects were reported by 19.6% of poison center exposure calls treated at a health care facility versus 1.5% of the poison center exposure calls were managed on-site (P<.001). During the 14-year study period, NPDS reported 551 deaths. Clinical symptoms were reported for 42.6% (N=73,558) of the total carbon monoxide exposure cases, and headache (24.8%), nausea (14.3%), dizziness/vertigo (12.4%), drowsiness/lethargy (5.9%), and vomiting (5.8%) were the most frequently reported symptoms related to carbon monoxide exposure. Other, less commonly reported symptoms included cough/choke (2.6%), confusion (2.1%), dyspnea (2.0%), syncope (1.9%), throat irritation (1.6%), chest pain (1.4%), tachycardia (1.0%), muscle weakness (0.9%), coma (0.8%), and ocular irritation/pain (0.8%) (Table 2).

The crude rate for unintentional carbon monoxide poisoning decreased over the study period for 3 data sources (ED visits, hospitalizations, deaths). Overall, the non–fire-related carbon monoxide poisoning crude rates of ED visits (average annual percentage change, 1.6%; 95% CI −1.43 to −1.77; P<.0001), hospitalizations (average percentage change, −0.23%; 95% CI 0.22 to −0.68; P=0.3118), and deaths (average annual percentage change, −1.4%, 95% CI −0.76 to −2.01; P<.0001) decreased from 2005 to 2018. However, the hospitalization and ED visit rates showed temporal increases from 2016 to 2018 (Figure).

The findings in this report are subject to limitations per each data source. First, calls to the poison center centers come from health care professionals or persons voluntarily reporting a poison exposure (a passive reporting system). There are no qualifications necessary to report, which can raise concerns about incorrect identification of exposure and toxicity syndromes. Second, cause classifications (fire-related, non–fire-related, unknown mechanism or intent) were only available from the hospitalization, ED visit, and death data. We classified poison center exposure cases as fire-related or unknown if fire-related on the basis of the case’s substance. Cases coded as carbon monoxide only (either generic code or not otherwise specified carbon monoxide product code) were classified as unknown if fire-related. Cases using the substance code smoke, toxic products of combustion, or fire hazards (toxic products of combustion) were grouped as fire-related. Substance codes are chosen by the poison specialist on the basis of the caller’s history and clinical presentation. Third, the death and poison center exposure data provided national estimates on carbon monoxide mortality and exposure. However, the analysis using ED visit and hospitalization data did not cover all 50 states. Therefore, the annual rates or cases could not represent the national data. Fourth, ICD-10-CM replaced ICD-9-CM to code medical terminology and disease classification in October 2015. This coding change affects information classifications for hospital discharge and ED records in all health care settings. Differences in the counts and rates in years before 2015 (ICD-9-CM) compared with 2015 (ICD-9-CM and ICD-10-CM) and subsequent years (ICD-10-CM) could be a result of this coding change in October 2015 and not an actual difference in the number of events. Fifth, although duplicate records for the same ED visit are excluded, the ED visits and hospitalizations are based on occurrences, not individuals, because a unique identifier is not always available. When multiple admissions for the same person during the year are not identified, the rate is calculated as the number of events per 1 million, which is an overestimate of the proportion. Finally, death investigation laws vary by location, potentially leading to variations in how medical examiners/coroners/physicians assign intentionality.

This report provides a summary of unintentional carbon monoxide exposure/poisonings from poison center exposure case data, ED visits, hospital discharges, and death data for 2005 to 2018. Our study focused only on unintentional carbon monoxide poisoning cases, including fire-related and non–fire-related causes.

Our demographic trends for unintentional carbon monoxide poisoning were consistent with those of some previous studies.^3,6,9,20 Children and younger adults represented the most frequently reported exposures in poison center exposure case and ED visit data; however, carbon monoxide exposure-related deaths occurred more often among men and older adults (aged ≥80 years) and in the Midwestern region.^1,6,9,20,21 Exposures most often occurred during the winter season across all study years.

Children and the elderly may be more vulnerable to carbon monoxide poisoning because of their increased metabolic demand, inability to vocalize symptoms, and/or mobility restriction.¹¹ Mobility may be especially important considering seeking medical care and procurement of rapid diagnosis and treatment increase the likelihood of desirable outcomes.^22,23 Older adults may have higher rates of underlying comorbidities or conditions at the time of carbon monoxide exposure, potentially leading to further complications.^22,23 Awareness, recognition and management of the signs and symptoms of carbon monoxide poisoning requires educational programs be provided for emergency physicians, clinical toxicologists, and neuroradiologists to speed up diagnosis and treatment to reduce unintentional carbon monoxide exposure-related morbidity and mortality rates among vulnerable populations.^7,24,25

We observed significantly decreased crude rates in non–fire-related carbon monoxide poisonings from ED visits and mortality over the study period (ie, 2005 to 2018). Using the population-based discharge data from the tracking program, our study shows a slight decrease in the ED visits (56.5 per 1 million) and hospitalization (7.3 per 1 million) rates compared with the previous study.⁹ The previous study reported 71 ED visits and 8 hospitalizations per 1 million during 2000 to 2009; however, their estimates were based on the nationwide ED and inpatient sample data in the Healthcare Cost and Utilization Project.⁹ We also observed a slightly increasing trend in ED visits and hospitalization rates from 2016 to 2018. This temporal trend shift may be related to the transition in reporting ICD-9-CM to ICD-10-CM codes in October 2015 and may make temporal comparison challenging in trends of hospital discharge data moving forward.^6,16 A previous study reported that non–fire-related carbon monoxide poisoning mortality did not substantially decrease from 2000 to 2014.¹ Using the same data source, non–fire-related carbon monoxide poisoning mortality showed a significant decrease during our study period. The death certificate data provide a more comprehensive estimate across the states where ICD-10 codes were assigned consistently during the full study period.

The CSTE position statement defines and classifies cases as confirmed, probable, and suspect by data source and medical outcome.¹⁶ After exclusions, all probable poison center cases with moderate, major, and death medical outcomes and suspect calls (minor medical outcome) were included in this analysis. Only confirmed or probable hospital discharge cases and confirmed death cases were included in this analysis. Among 172,489 poison center exposure calls, we observed 17,589 probable (10.2%) and 43,435 suspect (25.2%) carbon monoxide exposures.

Our surveillance report provides a summary of the morbidities and mortality of unintentional carbon monoxide poisonings. In 2019, as a part of the NNDSS Modernization Initiative, CDC developed the carbon monoxide poisoning message mapping guide to collect standardized data elements across multiple data sources.¹⁷ This single platform containing pertinent data sources will save valuable time and resources, improve data transmission, and reduce the burden of case reporting and duplicates. The standalone surveillance system is important to summarize and update the surveillance report for carbon monoxide exposures and poisoning timely and will incorporate different data sources and make it possible to link cases using personal identifiers or exposure information for the event. The new NNDSS system will timely collect case-based carbon monoxide poisoning data across data sources; therefore, it will drive accurate public health policy and actions ultimately aimed at reducing mortality and morbidity caused by carbon monoxide poisoning.^26,27

In conclusion, this surveillance report provides the baseline data for the new CDC case surveillance, a national NNDSS system for carbon monoxide poisoning. The more standardized and continued public health surveillance of carbon monoxide poisoning is necessary to monitor the public health burden and assess the effectiveness of targeted prssevention strategies. A comprehensive approach and public health responses from federal, state, and local public health agencies will play a role in enhancing public health surveillance and highlight the importance of the case-based carbon monoxide poisonings surveillance.