The Centers for Disease Control and Prevention (CDC) estimate that every year, approximately 15,000 emergency department (ED) visits and 475 deaths result from unintentional non-fire-related carbon monoxide (CO) poisoning (Centers for Disease Control and Prevention [CDC], 2005a, 2007). CO is undetectable to the senses and causes more fatal poisonings than any other toxicant (Varon, Marik, Fromm, & Gueler, 1999). CO poisoning is preventable, however, with proper understanding of CO risks and education about CO poisoning prevention.
Previous studies have documented CO poisoning during power outages, often following severe weather events when nonelectrical devices are used to generate electricity or for heating and cooking (CDC, 2005b, 2006; Daley, Smith, Paz-Argandona, Malilay, & McGeehin, 2000; Houck & Hampson, 1997; Van Sickle et al., 2007). These devices include portable generators, gas kitchen ranges, barbeque grills, and unvented heaters that use carbon-based fuels. Improper use of these devices can lead to CO poisoning.
On October 12 and 13, 2006, an early season snowstorm produced up to 24 inches of snow around Buffalo, New York (National Weather Service, 2006), causing widespread long-term power outages due to electrical wire damage from fallen tree limbs. Over 370,000 customers of the area's two primary electric utilities were without power on October 13. Approximately 100,000 people were still without power on October 18. For some customers, power was not restored for two weeks (New York State Department of Public Service, 2007).
Following the storm, ED visits for CO-related illness increased. Follow-up telephone interviews were conducted with people diagnosed with CO poisoning. The purpose of this investigation was to identify exposure sources, assess local access to CO warning information, and determine the presence of CO alarms. Staff from CDC, Consumer Product Safety Commission (CPSC), and the Maine and Connecticut Health Departments provided advice. The results of the investigation, discussed here, are intended to help public health professionals develop messages to prevent power outage-related CO poisoning.
One month after the storm, a letter was sent to administrators at 20 hospitals in the four counties in which 97% of the power outages occurred. The letter described the project and requested assistance in accessing medical records of patients who were evaluated in the ED for CO exposure between October 13, 2006, and October 28, 2006. Hospital medical records staff identified individuals with probable CO-related illness from medical records containing a International Classification of Diseases 9th Revision, Clinical Modification (ICD-9-CM) (U.S. Department of Health and Human Services, 2004) principle discharge diagnosis code of 986 (toxic effect of CO) or external cause of in jury codes E868.3, E868.8, E868.9, E981.1, or E982.1. Demographic data were recorded from the medical record.
Telephone and address information was obtained from medical records to contact individuals for interview. In households where only a child (younger than 16 years old) was diagnosed with CO poisoning, a parent or guardian was interviewed as a proxy. Individuals in households where a CO fatality occurred were not interviewed. Telephone interviews were conducted by Erie and Niagara County Health Department staff and New York State Department of Health (NYSDOH) staff between December 2006 and mid-March 2007.
Individuals who provided verbal consent, the "respondents," were asked to confirm the power outage and estimate its duration for their household. Respondents were then asked about usage patterns for various exposure sources, access to and clarity of CO warnings (i.e., through public service announcements or on device warning labels), and presence and use of CO alarms or smoke alarms.
All data analyses were performed using SAS software, version 9.1 (SAS Institute Inc., Cary, NC). Analyses were performed either by individual or by household. The ?2 test was applied to test for significant differences (p < .05) in demographic information collected from medical records between interviewed and noninterviewed groups. Univariate and bivariate analyses were conducted to examine the distribution of categorical variables.
During the power outage, 264 individuals--187 adults and 77 children--representing 155 households were diagnosed with CO poisoning. Telephone interviews were …