Author: Becker, A; Dark, T; Mason, T; Goodwin, B
Date published: May 1, 2012
Journal code: PENV
Several reasons warrant revisiting carbon monoxide (CO) poisoning surveillance during the Florida 2005 hurricane season. Emergency responders provided on-scene detailed and reliable incident reports and field observations (e.g., generator placement, hospital transport) that included information not collected by medical facilities or accurately through questionnaires. Monitoring the emergency response stakeholders' interagency data sharing systems (e.g., State Warning Point [SWP], Florida Poison Control Information Network [FPCIN], and emergency management) provided the most complete and accurate information. This information was compared with additional multiple sources (National Response Center [NRC], coroner reports, medical records, emergency response incident reports, fire departments, hazmat) to collate the most accurate information and precise number of CO poisonings. In addition, it is important to expound on why African-Americans and Latinos had a higherthan-expected incidence of CO poisoning.
To summarize the 2005 hurricane season the following descriptions were provided by the Federal Emergency Management Agency (FEMA) for Hurricanes Dennis, Katrina, and Wilma in Florida (FEMA, 2006).
* On Sunday, July 10, Hurricane Dennis made landfall as a category three storm near Navarre Beach, Florida. Dennis left 680,000 people without power.
* On Thursday, August 25, Hurricane Katrina came ashore south of Fort Lauderdale, causing power outages to over 740,000 homes, and four days later came ashore in the Florida Panhandle, resulting in more than 100,000 homes and businesses losing power.
* On Monday, October 24, Hurricane Wilma, a category three hurricane, came ashore at Cape Romano located in Collier County and crossed the state, hitting Miami, West Palm Beach, and Fort Lauderdale. It resulted in widespread power outages affecting 3.5 million customers in 42 counties.
CO is a colorless, odorless, and nonirritating gas produced during incomplete combustion. CO is an asphyxiant and disrupts oxygen transport and delivery by interfering with oxygen binding to hemoglobin (Leikauf & Prows, 2001). CO poisoning is often associated with accidental exposure to exhaust from generators used during power outages related to natural disasters (Hampton & Zmaeff, 2005).
Low-level CO exposure to environmental levels (100 parts per million [ppm]) may cause nonspecific symptoms such as headache, nausea, and lightheadedness. Higher CO exposure levels (150-200 ppm) may result in central nervous system symptoms including disabled coordination, syncope, coma, convulsions, pulmonary edema, and death (Centers for Disease Control and Prevention [CDC], 2006a). The Consumer Product Safety Commission (CPSC) has collected death certificates for all CO fatalities since 1999 and found that fatalities have carboxyhemoglobin (COHb) levels from 40% to 60%, indicating that CO exposure levels in the air likely exceeded 1,000 ppm (CPSC, 2008). COHb levels do not correlate with the severity of the signs and symptoms (Benignus, Muller, & Malott, 1990). The blood sample often is not collected at the scene but collected in the emergency room after treatment with oxygen and after an extended period of time due to transport. The Council of State and Territorial Epidemiologists has recommended the use of a pulse oximetry to establish a standard COHb measurement that can be determined in a more timely manner.
Factors that influence COHb formation and elimination include CO concentration in air, duration of exposure, and physical activity (CPSC, 2008). Fatalities may also occur at lower COHb levels in at-risk groups such as fetuses, infants, and the elderly; and in at-risk people with conditions such as chronic heart disease, anemia, or respiratory illness (CDC, 2008; CPSC, 2008).
Low-level CO concentrations in the air may exceed the U.S. Environmental Protection Agency (U.S. EPA) National Ambient Air Quality Standard for CO in outdoor air of 35 ppm for one hour and 9 ppm for an eight-hour period (U.S. EPA, 2009). In addition, generators near an open window may contribute to indoor air levels of CO that may exceed the National Institute of Safety and Health exposure limit of 35 ppm for a 15-minute period and the Occupational Safety and Health Administration (OSHA)-recommended permissible exposure level of 50 ppm over an eight-hour period (CDC, 2005a). CO levels of 23.7 ppm may result in lower maximum aerobic power in people with a previous history of CO poisoning (Horvath, Raven, Dahm, & Gray, 1975).
Observance of increased risk of CO poisoning among minority populations following a weather disaster is not uncommon. In Washington State, a study showed that CO poisoning occurred in the Hispanic populations at four times the risk of the Caucasian population, and the African- American population had a risk of CO poisoning three times greater than the Caucasian population (Ralston & Hampton, 2000).
Many positive changes have occurred related to the CO surveillance conducted in our study as well as in previous studies in the Gulf Coast. Much still needs to be investigated, however, including identifying why minorities experienced a higher rate of exposure than expected, the development of a local concerted outreach community network, and improved first-responder investigations of CO incidents with patient monitoring and environmental monitoring.
The most robust data sets used for CO surveillance were the SWP and the FPCIN. Disposition of patient transport and subsequent request of medical records, emergency response incident reports, and contact with the first responder were possible in most cases.
The SWP receives reports of emergency response of naturally occurring and humanmade emergencies (e.g., chemical agents, hazardous materials) generally through the emergency management in the county. The SWP provides notification to incidents in real time and agencies have the opportunity to add to the information. Additional information could then be requested from the emergency response incident reports.
The FPCIN provides real-time case reports online through a password access to programs involved in department of health surveillance. A chemical query code developed through the American Association of Poison Control Centers and the FPCIN in Jacksonville was used to monitor selected chemical exposures, harmful effects, and particular contaminants.
CDC Classification for CO Poisoning
The cases were classified according to CDC's case definition for CO poisoning (CDC, 2005b). Cases were classified as probable or confirmed and suspected cases were excluded from the data analysis. The CO poisonings were tracked until the electricity was operational and generator use subsided.
A probable case was clinically compatible with a high index of suspicion (credible threat or patient history regarding location and time) for CO exposure or had an epidemiologie link to a laboratory-confirmed case (CDC, 2005b).
A confirmed case was clinically compatible and had laboratory evidence of CO poisoning from biologic samples confirming exposure (CDC, 2005b) or had a predominant amount of clinical and nonspecific laboratory evidence that CO was present.
A suspected case was defined as a potential that the exposed person was being evaluated by health care workers or public health officials for poisoning by a particular chemical agent, but no credible threat existed (CDC, 2005b). Once the cases were deemed to be suspect, noncases, or informational, they were excluded from subsequent analysis.
We excluded children from the confirmed category with a COHb level below 3% based on normal levels for COHb which may range from 0% to 3% (Pagana & Pagana, 2006). We also excluded adults from the confirmed category with COHb below 10%. The precedence for this is based on the surveillance completed in Alabama and Texas, which included CO cases for nonsmokers with CO levels greater than 2.5% and smokers greater than 9% (CDC, 2006b; Radford & Drizd, 1982).
The racial/ethnic makeup of Florida counties was calculated using the 2000 U.S. census (U.S. Census Bureau, 2000), table P8: Hispanic and Latino by Race. In addition, race/ethnicity data were available for 30% of the cases since FPCIN did not routinely collect this information.
A total of 126 nonfatal CO poisoning victims were reported after the 2005 Hurricanes Dennis, Katrina, and Wilma (Table 1). Of these exposures, 52% were confirmed and 48% were probable using the CDC classification criteria. Seventy-seven percent were associated with generator use. Thirty percent of these cases placed their generators in the home, and 43% placed their generator outside the home. The location of the generator was not reported for 27% of the cases.
Nine CO poisoning deaths were reported in Florida from Escambia (n = 2), Broward (n = 3), Collier (n = 1), Miami-Dade (n = 2), and Palm Beach (n = 1) counties. Two of these deaths were associated with Hurricane Dennis, four deaths were associated with Hurricane Wilma, and three deaths were associated with Hurricane Katrina.
Figure 1 shows the location by county of nonfatal CO poisonings of the three hurricanes of 2005. Most cases occurred in the southeastern part of the state in Palm Beach, Miami-Dade, and Broward counties. There were no nonfatal cases reported from Hurricane Dennis.
The sources of CO poisoning cases reported from Hurricane Wilma were FPCIN (n = 46), first responders (n = 14), and medical records (n = 22). The sources of CO poisoning cases reported from Hurricane Katrina (n = 38) were from FPCIN. Coroner reports were the source of the two poisonings from Hurricane Dennis.
Eighty-six CO nonfatal poisoning cases were reported to the chemical surveillance program from October 24 to November 8, 2005, as a result of effects from Hurricane Wilma (Table 1). Forty-five percent of these were confirmed. Seventy-nine percent of the cases were related to the use of a generator, with 16% of these having used a generator indoors and 47% having used the generator outdoors near the home. The location of the generator was unknown for the remainder of the incidents (37%).
Four confirmed deaths from Broward, Collier, Miami-Dade, and Palm Beach counties were documented after Hurricane Wilma.
Thirty-eight CO poisoning cases were reported to the chemical surveillance program after Hurricane Katrina (August 25-August 29, 2005) (Table 1). All reported CO cases (n = 38) were reported by FPCIN. Thirty-seven percent of these cases were confirmed and 63% were classified as probable. Seventy-one percent of the cases were related to the use of a generator. Of these, 59% resulted from generator use inside the home and 37% of the cases were related to generators used outside the home. The location of the generator was unknown for the remainder of the incidents (4%).
Eleven cases were attributed to having a grill or gas powered burner inside the home. The symptoms were a little more severe for these cases (five females; six males) and included altered consciousness, lethargy, and confusion with additional nonspecific symptoms associated with CO exposure such as headache, dizziness, and lightheadedness.
Three confirmed deaths were documented from coroner reports. Two of the deaths occurred in Broward County and one occurred in Miami-Dade County.
Two cases were documented from coroner reports. Both occurred in Escambia County.
Eleven CO cases occurred at work. Three emergency medical technicians were exposed and overcome with fumes when responding to a CO incident that involved a generator outside a window. Another eight cases occurred in various offices in which generators were running outdoors over an extended period and fumes drifted in open windows. No environmental measurements of CO concentration were measured for any of these cases and in some cases the workers were exposed to the fumes over an eight-hour period.
Fifty-three percent of all hurricane-related cases were female. Thirty percent of the victims were between O and 19 years, 43% were between 20 and 39 years, and 22% were between 40 and 59 years. The cases ranged in age from 1 month to 84 years old. The coroner's reports were the only source that had a significantly higher median age (55 years) than the overall median age (29 years) for all sources.
Limited race/ethnicity data were available for 2005. The first responder data, hospital da ta, and coroner reports included this information. The majority of the cases (Table 2) were among Hispanics (43.9%) followed by African-Americans (36.6%) and Caucasians (14.6%). Table 2 shows the expected and observed number of cases that occurred among each race/ethnicity group comparing expected number of cases by race/ ethnicity to the observed number of cases by race/ethnicity African-American and Hispanic/ Latino cases of CO poisonings were overrepresented (?2 = 22.45, ? = .0001). The 2005 race/ ethnicity data were available for 30% of the probable and confirmed cases.
During extended power outages associated with hurricanes, residents may decide to use alternative power sources. A strong relationship exists among the number of CO poisonings, number of homes experiencing power outages, length of power outage, and the number of homes utilizing generators.
It is very likely that the fumes reentering the home or business could cause CO poisoning over an extended time with similar or lower levels than allowed in the work environment. The National Institute of Standards and Technology (NIST) study in 2009 reported that studies of CO from generators outside the home should focus on generator exhaust direction, distance of generator placement from the house, size of open windows, and environmental factors (NIST, 2008, 2009). Generator use over an extended amount of time should be monitored in the work environment to prevent exceeding the OSHA ceiling of 50 ppm over an eight-hour period. Danger labels are now required on all portable generators manufactured or imported on or after May 14, 2007 (Consumer Product Safety Commission [CPSC] , 2007). Better outreach efforts need to be provided describing the dangers of generator fumes drifting into open windows.
A CO detector was available and worked properly in one incident in our study, which alerted the residents and enabled them to evacuate safely. It is estimated that half of all CO poisonings deaths could be prevented with the use of a CO alarm (Yoon, McDonald, & Parrish, 1998). A new requirement for CO detectors, Florida Statute 583.885, requires that a CO detector be installed in new construction in an attached garage and placed within 10 feet of bedrooms (Carbon Monoxide Alarm Required, 2009). CO detectors that meet the UL 2034 or CSA 6.19 safety requirements are recommended (CPSC, 2008).
Although limited data were available on race/ethnicity for this study, the data show an overrepresentation of CO-poisoned AfricanAmericans and Hispanics/Latino compared to Caucasians. Several possible explanations for these findings may exist, including a higher use of generators among these populations, length of time needed to restore power, and the condition of the housing.
It appears that from 2004 to 2005, CO poisonings in the Caucasian population were reduced from 45.5% (CDC, 2005c) to 14.6% (Becker et al., 2006), respectively. This is possibly due to outreach that may have selectively reached the Caucasian population through the more traditional methods of distributing fact sheets and media through the Florida Department of Health and other agencies. More information needs to be collected concerning race/ethnicity and outreach efforts should be targeted to the African-American and Hispanic/Latino populations at the local level. In addition, African-Americans and Hispanics/ Latinos were overrepresented in hospital and coroner data, suggesting that these population groups were both more frequently and more severely poisoned than others.
The predictability of generator use and CO poisoning during severe weather seasons underscores the necessity of public health initiatives and provides the opportunity to obtain safety information from different sources such as product packaging and public service announcements via mass media. Since the quality of the sources from which consumers can obtain information may influence the information received, the manner of information distribution is of upmost importance. We must keep in mind the fact that financially disadvantaged minorities are more likely to reside in areas in which electricity is slow to be restored (e.g., outside city limits). Exacerbating these potential problems is the possibility that these populations are less likely to have access to information regarding the hazards related to portable generator use.
Collection of detailed information on those exposed to CO in order to determine unexplored and underexplored risk factors for CO poisoning is imperative. Employment, income, education, and neighborhood socioeconomic characteristics may be predictors of increased risk of CO poisoning among disadvantaged minority populations. Such information will allow for the exploration of issues expanding beyond the mere presence and legibility of safety (proper use) information on product packaging, but also the comprehension level and the successful/ unsuccessful delivery of the safety message. This research is imperative.
Limitations The findings of our study are subject to several limitations. First, the symptoms of CO poisoning may be nonspecific. Therefore possible underreporting due to underdiagnosis and misdiagnosis must be considered. No active CO poisoning surveillance system existed at the county level, so the suspect cases could not be evaluated or fully investigated due to manpower and time constrains during major and multiple disasters.
Conclusion The major accomplishments after the completion of our study that will likely reduce future CO exposure following hurricanes in Florida included the addition of CO poisoning as a reportable condition, using the more active classification system developed for county health departments to conduct surveillance at the local level, and requiring CO detectors and warning labels on generators. Additional work needed includes monitoring COHb in a standardized manner to assure accuracy; better prediction of severity of exposure; predictive prognosis; developing networks of reporting that include first responders, local county health department, and local emergency management; and the development of outreach materials and techniques for all race/ethnicity populations. Funding is needed in counties with high population centers most affected by hurricanes (e.g., Miami-Dade, Broward, Palm Beach) to further develop the local surveillance network to improve environmental monitoring, incident causation, and investigation. In addition, an outreach web to the community needs to be developed to effectively communicate CO risks to all populations. It is imperative that county health departments collaborate with local first responders, hazmat, and fire departments to provide outreach to all citizens of Florida to address the race/ethnicity disparity in the surveillance of CO poisonings.
Acknowledgements: This work was funded by a grant received from the Agency for Toxic Substances and Disease Registry (U61/ ATU474148-3). Data were previously reported in Becker et al., Florida Environmental Health Association journal, 195, 20-24,
Becker, A., Jones, J., Goodwin, B., Mason, T., Patel, P.S., & Blackmore, C. (2006). Florida hazardous substances emergency events surveillance activities in classifying the severity of carbon monoxide exposures of the 2005 hurricane season. Florida Environmental Health Association journal, 195, 20-24.
Benignus, YA., Muller, K.E., & Malott, C.M. (1990). Dose-effects function for carboxyhemoglobin and behavior. Neurotoxicology and Teratology, 12(2), 111-118.
Carbon Monoxide Alarm Required (Building and Construction Standards), Florida Statute 583.885 (2009).
Centers for Disease Control and Prevention. (2005a). National Institute for Occupational Safety and Health pocket guide to chemical hazards. Retrieved from http://www.cdc.gov/nioslVnpg/npgd0105.html
Centers for Disease Control and Prevention. (2005b). Case definition for chemical poisoning. Morbidity and Mortality Weekly Report, 54(RROl), 1-24.
Centers for Disease Control and Prevention. (2006a). Carbon monoxide: A model environmental public health indicator (National Workgroup on Carbon Monoxide Surveillance). Retrieved from http://www.myfloridaeh.com/programs/Environmental_Public_Health_Tracking/ PDFs/CO_A_Model_Public_Health_Indicator.pdf
Centers for Disease Control and Prevention. (2006b). Carbon monoxide poisonings after two major hurricanes - Alabama and Texas, AugustOctober 2005. Morbidity and Mortality Weekly Repon, 55(09), 236-239.
Centers for Disease Control and Prevention. (2008). Emergency preparedness and response, clinical guidance for carbon monoxide (CO) poisoning after a disaster. Retrieved from http://emergency.cdc. gov/disasters/co_guidance.asp
Consumer Product Safety Commission. (2007). Portable generator hazards. Retrieved from httpy/www.cpsc.gov/cpscpub/pubs/portgenhtml
Consumer Product Safety Commission. (2008). Carbon monoxide questions and answers. Retrieved from http://www.cpsc.gov/cpscpub/pubs/466.html
Federal Emergency Management Agency. (2006) . News releases . Retrieved fromhttpy/wwwfema.gov/news/newsarchive.fema?year=2006&month=8
Hampton, N.B., & Zmaeff, J. L. (2005). Carbon monoxide poisoning from portable electric generators. American journal ofPreventative Medicine, 28(1), 123-125.
Horvath, S.M., Raven, PB., Dahm, T.E., & Gray, DJ. (1975). Maximal aerobic capacity at different levels of carboxyhemoglobin. journal of Applied Physiology, 38(2), 300-303.
Leikauf, G., & Prows, D. (2001). Inorganic compounds of carbon, nitrogen and oxygen. In E. Bingham, B. Cohrssen, & C. Powell (Eds.), Patty's toxicology. New York: John Wiley and Sons.
National Institute of Standards and Technology. (2008). NIST to study hazards of portable gasoline-powered generators. Retrieved from http://wwwnist.gov/public_affairs/techbeat/tb2008_0305.htm
National Institute of Standards and Technology. (2009). Modeling the effects of outdoor gasoline-powered generator use on indoor carbon monoxide exposures (NIST Technical Note 1637). Retrieved from http://www.fire.nist.gov/bfrlpubs/build09/PDF/b09009.pdf
Pagana, K., & Pagana, T. (2006). Carboxyhemoglobin test reference. In K. Pagana & T. Pagana (Eds.), Manual of diagnostic and laboratory tests (pp. 159-161). St. Louis: Mosby
Radford, E. P, & Drizd T.A. (1982). Blood carbon monoxide levels on persons 3-74 years of age: United States, 1976-80. Advance Data, 76, 1-24. Retrieved from http://www.cdc.gov/nchs/data/ad/ ad076acc.pdf
Ralston, J. D., & Hampton, N.B. (2000). Incidence of severe unintentional carbon monoxide poisoning differs across racial/ethnical categories. Public Health Reports, 115(1), 46-51.
U.S. Census Bureau. (2000). County -specific data (table P8). Retrieved from http://www.census.gov/census2000/us.html
U.S. Environmental Protection Agency. (2009). Carbon monoxide (CO). Retrieved from http://www.epa.gov/iaq/co.html
Yoon, S., MacDonald, S., & Parrish, G. (1998). Death from unintentional carbon monoxide poisoning and potential for prevention with carbon monoxide detectors. Journal of the American Medical Association, 279(9), 685-687.
A. Becker, MPH, PhD,
Florida A&M University
T. Dark, PhD,
Florida A&M University
T. Mason, PhD,
University of South Florida
B. Goodwin, PhD,
National Center for Advancing
Corresponding Author: Tyra Dark, Assistant Professor, College of Pharmacy and Pharmaceutical Sciences, Institute of Public Health, Tallahassee, FL 32307. E-mail: tyra.darkŪ famu.edu or email@example.com.