Carbon monoxide poisoning surveillance in the Veterans Health Administration, 2010–2017

Background Exposure to carbon monoxide (CO), the odorless, colorless gas resulting from incomplete combustion of hydrocarbons, is preventable. Despite the significant risk of morbidity and mortality associated with CO poisoning, there currently exists no active national CO surveillance system in the United States (U.S.). Our study aims to use electronic health record data to describe the epidemiology of CO poisoning in the Veterans Health Administration healthcare population. Methods We identified unique inpatient and outpatient encounters coded with International Classification of Diseases (ICD) codes for CO poisoning and analyzed relevant demographic, laboratory, treatment, and death data from January 2010 through December 2017 for Veterans across all 50 U.S. states and Puerto Rico. Statistical methods used were 95% CI calculations and the two-tailed z test for proportions. Results We identified 5491 unique patients with CO poisoning, of which 1755 (32%) were confirmed/probable and 3736 (68%) were suspected. Unintentional poisoning was most common (72.9%) overall. Age less than 65 years, residence in Midwest U.S. Census region versus South or West, and winter seasonal trend were characteristics associated with confirmed/probable CO poisoning. Twenty-six deaths (1.5%) occurred within 30 days of confirmed/probable CO poisoning and were primarily caused by cardiovascular events (42%) or anoxic encephalopathy (15%). Conclusions Our findings support the use of ICD-coded data for targeted CO poisoning surveillance, however, improvements are needed in ICD coding to reduce the percentage of cases coded with unknown injury intent and/or CO poisoning source. Prevalence of CO poisoning among Veterans is consistent with other U.S. estimates. Since most cases are unintentional, opportunities exist for provider and patient education to reduce risk.