Carbon monoxide (CO) poisoning is a leading cause of poison-related death in the U.S. and is responsible for 450 deaths and 20,000 nonfatal injuries every year (Centers for Disease Control and Prevention [CDC], 2012). The U.S. does not have a comprehensive national system of CO surveillance (Graber, Macdonald, Kass, Smith, & Anderson, 2007), however, so these numbers likely are a vast underestimate of the CO-related deaths and injuries. The incidence of CO poisoning might also be underrepresented nationally due to misdiagnosis resulting from the nonspecific nature of its symptoms (Iqbal, Law, Clower, Yip, & Elixhauser, 2012; Raub, Mathieu-Nolf, Hampson, & Thom, 2000). Between 2000 and 2009, more than 68,000 CO exposures were reported to poison centers (Annest et al., 2008). In 2007, unintentional, nonfire-related CO poisoning accounted for more than 2,000 hospitalizations with the cumulative total for hospitalizations in 2007 costing over $26 million (Iqbal et al., 2012).

Poisonings caused by CO occur when CO—an odorless, colorless, and tasteless gas—escapes from fuel-burning appliances and becomes trapped in enclosed spaces. The installation of a CO detector is the most effective step for protecting household occupants. Detectors are effective in alerting occupants to the presence of CO and reducing the number of individuals who experience poisoning symptoms. Nationally, less than one half of households own a CO detector (Runyan et al., 2005), yet most are unsure where to place CO detectors or how many they should install. In a recent Baltimore study, 26% of 603 surveyed households were observed to have a functioning CO detector and less than 20% of surveyed households correctly identified the best place to install a CO detector (McDonald et al., 2013). Common misuses (which lead to false alerts, decrease the effectiveness of the devices, or render the devices inoperable) are incorrect placement and failure to replace batteries every 6 months. Thus, there is a critical need for interventions to increase correct use of residential CO detectors.

Numerous methods and interventions have been developed and tested to distribute and increase the adoption and use of safety products. Evidence from previous meta-analyses showed that interventions to promote use of smoke alarms are effective at increasing smoke alarm ownership (DiGuiseppi & Higgins, 2000, 2001) and the prevalence of functioning alarms (DiGuiseppi & Higgins, 2001; Kendrick et al., 2007). Cooper and coauthors (2012) in a network meta-analysis showed that “more intensive” interventions (e.g., education with low-cost or free equipment, installation of equipment, and home inspection), compared with “less intensive” interventions had a higher probability of increasing possession of functioning smoke alarms (Cooper et al., 2012). A study by Harvey and coauthors (2004) determined that direct installation of smoke alarms by program staff resulted in functioning smoke alarms in 90% of households that received direct installation intervention compared with 65% in a voucher intervention group. To our knowledge, no similar interventions (or interventions that combined the aforementioned components) have examined the effectiveness of CO detector interventions or various distribution methods to increase CO detector ownership, functionality, and placement.

The purpose of this study was to describe changes in CO safety knowledge and observed CO detector use (ownership, functionality, and placement) following distribution of the identical CO intervention, that is, an educational tool, Fast Facts About Carbon Monoxide, along with a CO plug-in detector with battery backup in an emergency department (ED) setting (Columbus, Ohio) and in an urban community setting (Baltimore, Maryland).

The specific aims of the current study were to describe the 1) sociodemographic characteristics of each sample, 2) changes in CO safety knowledge 6-months postintervention, and 3) changes in observed CO detector use (ownership, functionality, and placement) 6-months postintervention.

Participants were part of larger studies: a randomized controlled trial based in Columbus, Ohio, and a community intervention trial based in Baltimore, Maryland. Participants in each group received an educational tool, Fast Facts About Carbon Monoxide, a new CO detector, and completed a 6-month follow-up home visit.

Fast Facts About Carbon Monoxide was developed as part of the Columbus, Ohio-based randomized controlled trial, which aimed to increase the use of correctly installed and maintained CO detectors in a population of parents recruited in a pediatric ED. The tool guides the recipients through a presentation in which CO is defined; the dangers, symptoms, and causes of CO poisoning are described; and the instructions on CO detector installation and maintenance are explained. The tool was written at a seventh-grade reading level so as to suit the needs of a low literacy population. Images and messages were chosen to be appropriate for the target audiences (Figure 1). The last page of the educational tool contained a removable magnet that included emergency and nonemergency phone numbers relevant for the city in which the educational tool was distributed.

A total of 125 participants in the Ohio sample and a total of 176 participants in the Maryland sample received the intervention and were included in our analysis. There were no differences on any single knowledge item or total knowledge score between those lost to follow up (that is, participants who did not complete the 6-month home visit in either study) and those who completed the 6-month follow-up home visit for the Ohio or Maryland sample.

The majority of participants was female (Ohio: 90.4%; Maryland: 85.8%), 25–34 years of age (Ohio: 41.6%; Maryland: 31.8%), and employed either full or part time (Ohio: 50.4%; Maryland: 61.2%). Most participants had a per capita income of $5,000 or less (Ohio: 43.2%; Maryland: 34.2%) or $5,001–$10,000 (Ohio: 26.4%; Maryland: 35.5%). The Maryland sample had significantly more participants who reported their race as Black (p < .01). Educational attainment differed significantly between the two samples (p < .01); Ohio participants were more likely to have completed some college (Ohio: 49.6%; Maryland: 23.9%), while Maryland participants were more likely to report completing high school or less (Ohio: 33.6%; Maryland: 61.9%).

The amount of time living in current residence significantly differed between samples (p < .01); Ohio participants were more likely to report living in their current residence less than 1 year, while Maryland residents reported living in their current residence more than 2 years (Table 1).

CO poisoning is a leading cause of poison-related death in the U.S. (CDC, 2012) and a significant public health concern. A properly installed and functioning CO detector is an effective tool to protect household occupants from residential, nonfire-related CO poisoning. The purpose of this study was to describe changes in CO safety knowledge and observed CO detector use following distribution of the same CO intervention (educational tool Fast Facts About Carbon Monoxide plus a plug-in CO detector with battery backup) in an ED setting (Columbus, Ohio) and in an urban community setting (Baltimore, Maryland).

Overall, both groups significantly improved in knowledge scores and the majority of participating households was protected by a CO detector at follow up (>70% for Ohio and Maryland). The detectors were not consistently installed, however, in the correct recommended location, i.e., near sleeping areas in either sample. Differences in postintervention outcomes were detected between samples. The Ohio sample that had higher postintervention knowledge scores was more compliant on having a working CO detector than the Maryland sample. Other indicators of improved behavior were participants who lived at their current residence for 1–2 years, identified their race as White, and were older in age.

There are several differences in the target populations and delivery methods that may partially explain these differences; however, these differences were adjusted for in these analyses. First, there were key demographic differences between the two samples, namely, educational level (lower in Maryland sample) and minority composition (more Blacks in Maryland sample). Other significant differences were the age of participants and time living at current residence. Although the educational tool was written at a seventh-grade reading level and with a low literacy population in mind, perhaps the tool could be further refined in this manner (text shortened, lower reading grade level, etc.).

Second, the “intensiveness” of the intervention from a resource standpoint and from a content and information standpoint differed between the samples. The Maryland sample received the intervention as part of another study where smoke alarms and hot water temperature were also addressed. The Ohio sample received only information and intervention on the CO detector. The difference in the amount of information that participants had to process may have contributed to the Maryland sample’s difficulty in following through on the recommendations. The CO intervention might be better as a stand-alone intervention, rather than combined with other safety messages and recommendations.

Third, the setting in which the interventions were distributed varied. The Maryland sample received the intervention in their homes (Baltimore City Fire Department staff and data collector were present); the Ohio sample received the intervention in a pediatric ED (study recruiter delivered the intervention). As the Ohio participants “had time to wait” in the ED, they might have had more time to read the tool, absorb the information, and were then motivated to install the device when they returned home.

Despite these differences, it is promising that the less-resource intensive distribution method in Ohio (i.e., simply delivering the tool and device in a clinical setting) had higher knowledge gains and more uptake of CO detectors. A positive note about the home distribution is that you can conserve resources by restricting distribution to homes in need or address other safety issues within the home.

Messaging around the importance and timing of battery replacement need improvement. Our results suggest that the educational tool and messages on battery replacement were not effective in motivating participants to change the battery, even when a replacement battery was provided. Methods to increase battery replacement should be further investigated in future studies.

The two study samples (Maryland and Ohio) were derived from other larger studies and were not originally designed or selected to be comparable; it was timing and launching of both studies and convenience that drove the comparison. As such, these studies were not collectively powered for this comparison. Other limitations included minor variations in how the intervention was distributed and how follow-up home visits were conducted at each site, including: 1) how children were defined in each study (≤18 years in the Ohio sample and ≤17 years in the Maryland sample); 2) length of time between enrollment and 6-month follow up; 3) length of time to conduct follow-up home visit (average 30 minutes for Ohio and 60 minutes for Maryland); and 4) amount of information shared with participants. The groups received identical educational materials, CO detectors, and batteries. Both sites were assessed using the same survey items and observation criteria.

An intervention designed to improve CO safety knowledge and CO detector presence, functionality, placement, and battery replacement behaviors can be distributed successfully with positive results in a pediatric ED and/or door-to-door in an urban setting. The success of the intervention varied between settings and distribution methods, but both methods showed positive changes in knowledge and behavior. CO safety knowledge was better among the Ohio sample (more improvement in knowledge from enrollment to follow up) and CO detector use (installation, location, and functionality) was significantly better at follow up. All participants, regardless of setting or distribution method, would benefit from improved battery replacement messages or reminders. Future educational efforts around this topic should focus on the less well-known information about CO poisoning and prevention such as the causes of CO, symptoms of CO poisoning, and where CO detectors should be installed. Despite the differences in the improvement shown in knowledge and behaviors between the sites, both distribution methods (ED and community distribution) were promising for getting this life-saving technology into homes.