## Abstract

While largely preventable, fire and hot water-related injuries are common in the United States. Measures recommended to reduce these injuries are smoke alarms (SAs) and lowered hot water temperatures. This study aims to: (i) describe the prevalence of working SAs and safe water temperatures among low-income, urban communities and (ii) explore the relationship between these behaviors and individuals’ knowledge and beliefs about them. In this cross-sectional study, the Health Belief Model was used as a guide for understanding the safety behaviors. A total of 603 households had their SAs and hot tap water temperatures tested and were surveyed about their knowledge and beliefs related to these safety behaviors. We found that 40% of households had working SAs on every level and 57% had safe hot water temperatures. Perceived severity and self-efficacy were significantly associated with SA coverage, whereas perceived susceptibility and beliefs about benefits were significantly associated with safe hot water temperatures. This study demonstrates the need to increase the number of homes with working SAs and safe hot water temperatures. Messages focused on a safe home environment could communicate the ease and harm reduction features of SAs and benefits and risk reduction features of safe hot water temperatures.

## Introduction

While largely preventable, fire-related injuries including those from smoke, fumes and flames, and burns from hot water are common in the United States as well as globally. In 2009, there were 3195 fire and burn deaths in the United States (1.00/100 000) of which 2430 occurred in the home environment [1]. In addition to fatalities, 381 012 non-fatal fire and burn injuries were seen in emergency departments and hospitals that same year [2]. The fire and burn death and non-fatal injury rates are higher among children younger than 5 years of age and adults older than 60 years [3]. Lower socioeconomic status and urban environments are also well-documented risk factors for fire and burns [4–7].

In the United States, fires in the home lead to over three-quarters of the fire-related fatalities and injuries [8]. Unattended cooking, cigarettes or lighters, electric blankets and appliances or exposed wiring are the most frequent causes of house fires [3, 9]. Tap water has been found to cause almost 25% of scalds among children and is known to cause more serious burns than other hot liquids [10–13]. The most widely recommended measures to reduce these fire and burn injuries are smoke alarms (SAs) and lowered hot water temperatures. Approximately 4 of 10 residential fire deaths occur in homes without SAs, and working SAs reduce the risk of death in a house fire by at least 50% [9, 14]. Setting water heaters to 120°Fahrenheit (F) significantly reduces the risk of a serious burn because water temperatures at 140°F can cause a scald burn in 3 s, whereas at 120°F it takes approximately 10 min [15].

The extent to which families, especially those living in low-income urban neighborhoods, are aware of these risk factors and are protecting themselves by having working SAs and safe hot water temperatures is not well documented. National survey data find high self-reported rates of having at least one working SA (>95%) [3, 16]. However, other work suggests high rates of over-reporting in such types of surveys [17], and our home observation data from a low-income urban sample in the United States found that 82% of homes had at least one working SA. Having a working SA on every level is the recommended standard, and the 2007 American Housing Survey showed that reported rates for this level of protection are less common in some demographic groups, including renters, African Americans and homeowners >65 years [18]. On the issue of hot water safety, a national telephone survey in Canada found that only 14% of respondents reported testing the temperature of their hot water [19]. In a low-income urban area in the United States, we observed that 47% of homes had safe hot water temperatures [20].

Little is known about whether and to what extent knowledge and beliefs about fire and burn prevention are associated with these safety practices and may explain differences in rates across subpopulations. Hapgood et al. [21] examined sociodemographic correlates of a wide variety of safety practices, including working SAs and safe hot water temperatures in a large sample of parents with infants in Great Britain. They concluded that most of the variation in safety practices could not be explained by sociodemographic factors and urged further study of other potential determinants such as perception of risk, locus of control and social support. We could find only two recent studies, both qualitative, that addressed psychosocial factors related to fire and burn prevention [22, 23]. These studies suggest that low perceived risk, high perceived barriers and competing priorities may explain why individuals choose not to have working SAs or to test or turn down their hot water temperatures.

In the current study, the Health Belief Model (HBM) was used as a guide to explore beliefs associated with two fire and burn prevention safety behaviors, use of SAs and safe water temperatures. According to the HBM, preventive behaviors are influenced by six factors: (i) perceived susceptibility (e.g. likelihood of experiencing a home fire or scald burn); (ii) perceived severity (e.g. beliefs about the seriousness of a home fire or scald burn); (iii) perceived benefits of performing the preventive behavior (e.g. beliefs that taking action will reduce the susceptibility to or severity of a home fire or scald burn); (iv) perceived barriers to performing the preventive behavior (e.g. beliefs about the costs of taking action); (v) cues to action (e.g. factors that invoke action) and (vi) self-efficacy (e.g. an individual’s belief in his/her ability to perform the preventive behavior) [24–27]. The HBM was chosen because it has demonstrated utility in understanding other risk behaviors [28].

The current study sought to contribute to the literature on observed rates of working SAs and safe water temperatures among low-income, urban communities and to examine the sociodemographic risk factors for inadequate protection. Second, we sought to describe residents’ knowledge and beliefs about these home injury risks. In using the HBM to inform our measurement tool, we were able to explore the questions of why, what and how individuals can be influenced to adopt these safety behaviors and thereby also add to the small body of literature applying health behavior change theories to injury prevention behaviors [29]. Such information can contribute to effective messaging to promote participation in SA canvassing programs as well as encourage individuals to adopt the recommended safety behaviors.

## Methods

The data presented here are taken from a baseline survey of households in East Baltimore, MD, as part of a community intervention trial that evaluated a SA-canvassing program conducted by the Baltimore City Fire Department, as described in more detail in Gielen et al. [30]. Baseline data include interviews and home observations with 603 households selected from 12 census tracts. Data were collected between July and December 2009.

A random selection of 3503 of the 9826 households in the study area were initially contacted via mail with a project letter and then visited by interviewers Monday through Saturday during daylight hours. Selected households were visited up to five times before they were deemed unresponsive. There were a total of three random selection waves, each of which occurred approximately 2 months apart, after most households were resolved (surveyed, refused to be surveyed or deemed unresponsive) from the prior selection. Of the 3503 addresses, 193 who refused via telephone in response to the project letter were excluded and an additional 39 who participated in a pilot test of the survey application were also excluded. Data collectors attempted to visit the remaining 3271 addresses in person and subsequently excluded another 2144 addresses for various reasons. Of the remaining 1127 eligible households, 515 refused participation (45.7%). Baseline surveys were conducted with 612 participants (54.3% of eligible households), though 9 had incomplete data and were removed resulting in a final sample of 603 completed surveys (Fig. 1). Interviews were conducted inside the home with an English-speaking adult resident who gave informed consent to participate. Following the survey, data collectors tested the temperature of the hot water and all SAs following the procedures below. The Johns Hopkins Institutional Review Board approved the study.

Fig. 1.

Flow chart of sample selection.

Fig. 1.

Flow chart of sample selection.

### Measures

#### Sociodemographic characteristics

Respondents self-reported their sex, education, income, race/ethnicity, homeowner status (renter versus owner) and age, as well as the age of all residents in the home.

#### Fire and scald burn knowledge and HBM beliefs

Twelve questions were developed by the authors and used to examine participants’ knowledge about fires and scald burns. We developed 34 items to measure the HBM constructs. All of the knowledge and belief items were created for the purpose of this study and pretested using cognitive interviewing, which is a method that entails administering the items developed and collecting verbal information from the participants about their responses [31]. This information was then used to improve wording and comprehension.

#### SA status

All SAs in the home were tested by pressing the ‘test’ button. A summary variable was created to indicate whether there was a working SA on every level of the home (including basements) and a dichotomous variable (yes/no) was used in the analysis.

#### Hot water temperature

A standard candy thermometer was used to test the temperature of the hot tap water in the kitchen. Candy thermometers provide a measure of temperatures between 75°F and 400°F. After running the kitchen faucet at the hottest setting for 1 min, water was collected in a cup and tested with the thermometer. A dichotomous variable (yes/no) was used in the analysis to indicate whether the water temperature tested at or below the recommended level of 120°F.

### Data analysis

To address the study aims, we first examine the bivariate relationships between sociodemographic variables and the outcome variables using chi-square statistics. We then describe participants’ knowledge and beliefs, followed by an examination of their bivariate relationships with the outcome variables using chi-square and t-test statistics. Finally, unadjusted and main effects multiple logistic regression models are used to identify significant correlates of having working SAs on every level of the home and safe hot water temperatures. Interactions were included during model building; because they were not significant, the results are not presented. Inverse probability weights were used to examine the potential biases due to the study area having a higher frequency of African American and low-income respondents than all of Baltimore. Weights were calculated based on race and income distributions obtained from the 2000 Census data. Weighted and unweighted multiple logistic regression models were constructed. We present unweighted data here because the results changed only minimally when using the weighted data. The analyses were conducted using Intercooled STATA 11.2 [32].

## Results

### Sample

The majority of respondents were African American (61%), female (70%), between the ages of 25 and 54 years (66%) and had a high school education or less (51%). Almost three-quarters of the respondents (74%) reported a per capita income of $25 000 or less and a little more than half (52%) rented their home. Forty percent of respondents lived with children under the age of 18 years. (Table I) Table I. Sociodemographic characteristics and bivariate relationship with smoke alarm and hot water temperature outcomes among a sample of residents in 12 census tracts in East Baltimore, 2009 Total sample Working smoke alarm on every levela Safe water temperature (<120°F)a Yes No χ2 P Yes No χ2 P Sex Male 180 80 (44) 100 (56) 2.1 0.15 109 (61) 70 (39) 1.8 0.18 Female 423 161 (38) 261 (62) 230 (55) 189 (45) Age (years) 18–24 89 28 (31) 61 (69) 16.3 0.003* 49 (56) 39 (44) 6.6 0.16 25–34 167 83 (50) 84 (50) 90 (55) 75 (45) 35–44 118 53 (45) 65 (55) 59 (50) 58 (50) 45–54 108 31 (29) 77 (71) 61 (57) 46 (43) ≥55 121 46 (38) 74 (62) 80 (66) 41 (34) Educationb Less than high school diploma/GED 79 31 (39) 48 (61) 14.6 0.002* 44 (56) 35 (44) 12.3 0.007* High school diploma/GED 230 72 (31) 157 (69) 115 (50) 114 (50) Some college 111 48 (43) 63 (57) 58 (53) 51 (47) Completed college 181 90 (50) 91 (50) 120 (67) 59 (33) Per capita incomeb$5000 or less 129 44 (34) 85 (66) 7.9 0.05* 73 (57) 55 (43) 2.1 0.54
$5001–10 000 103 42 (41) 61 (59) 52 (51) 50 (49)$10 001–25 000 124 49 (40) 74 (60)   73 (59) 50 (41)
$25 000 or more 125 64 (51) 61 (49) 74 (60) 50 (40) Race/ethnicityb Black or African American 359 127 (35) 232 (65) 8.3 0.004* 185 (52) 171 (48) 11.5 0.001* White or Caucasian and otherc 227 107 (47) 119 (53) 149 (66) 76 (34) Homeowner statusb Rent 297 128 (43) 169 (57) 2.6 0.11 148 (50) 146 (50) 8.2 0.004* Own or pay mortgage 280 102 (37) 177 (63) 173 (62) 105 (38) Children in home (<18 years) Yes 240 95 (40) 145 (60) 0.03 0.85 118 (50) 120 (50) 8.14 0.004* No 363 146 (40) 216 (60) 221 (61) 139 (39) Total sample 241 (40) 361 (60) 339 (57) 259 (43) Total sample Working smoke alarm on every levela Safe water temperature (<120°F)a Yes No χ2 P Yes No χ2 P Sex Male 180 80 (44) 100 (56) 2.1 0.15 109 (61) 70 (39) 1.8 0.18 Female 423 161 (38) 261 (62) 230 (55) 189 (45) Age (years) 18–24 89 28 (31) 61 (69) 16.3 0.003* 49 (56) 39 (44) 6.6 0.16 25–34 167 83 (50) 84 (50) 90 (55) 75 (45) 35–44 118 53 (45) 65 (55) 59 (50) 58 (50) 45–54 108 31 (29) 77 (71) 61 (57) 46 (43) ≥55 121 46 (38) 74 (62) 80 (66) 41 (34) Educationb Less than high school diploma/GED 79 31 (39) 48 (61) 14.6 0.002* 44 (56) 35 (44) 12.3 0.007* High school diploma/GED 230 72 (31) 157 (69) 115 (50) 114 (50) Some college 111 48 (43) 63 (57) 58 (53) 51 (47) Completed college 181 90 (50) 91 (50) 120 (67) 59 (33) Per capita incomeb$5000 or less 129 44 (34) 85 (66) 7.9 0.05* 73 (57) 55 (43) 2.1 0.54
$5001–10 000 103 42 (41) 61 (59) 52 (51) 50 (49)$10 001–25 000 124 49 (40) 74 (60)   73 (59) 50 (41)
$25 000 or more 125 64 (51) 61 (49) 74 (60) 50 (40) Race/ethnicityb Black or African American 359 127 (35) 232 (65) 8.3 0.004* 185 (52) 171 (48) 11.5 0.001* White or Caucasian and otherc 227 107 (47) 119 (53) 149 (66) 76 (34) Homeowner statusb Rent 297 128 (43) 169 (57) 2.6 0.11 148 (50) 146 (50) 8.2 0.004* Own or pay mortgage 280 102 (37) 177 (63) 173 (62) 105 (38) Children in home (<18 years) Yes 240 95 (40) 145 (60) 0.03 0.85 118 (50) 120 (50) 8.14 0.004* No 363 146 (40) 216 (60) 221 (61) 139 (39) Total sample 241 (40) 361 (60) 339 (57) 259 (43) aFrom a total of 603 homes completing the survey, 602 and 598 participated in the observational component of the survey evaluating presence of SAs on every level and safe water temperature settings, respectively. bOf the 60 homes completing the survey, 2 did not provide information on the highest level of education completed, 122 did not provide information on the family’s total annual income, 17 did not provide information on racial background and 26 did not provide information on whether they rented, owned or had a mortgage on their home. cWhite or Caucasian and other include 190 White or Caucasian respondents, 5 American Indian or Alaska Native respondents, 7 Asian or Asian American respondents, 4 Hawaiian or Pacific Islander respondents and 21 respondents identifying as more than one race. *P-value is significant (P ≤ 0.05). ### Outcomes and sociodemographic correlates Forty percent of the households had working SAs on every level, 14% of households had a working SA on the basement level/first floor only and 21% had a working SA on the second floor only. In the bivariate analysis, this outcome, working SAs on every level, was associated with respondent age, education, income and race/ethnicity, but not with homeowner status (Table I). Homes with and without children were equally likely to have working SAs on every level. Fifty-seven percent of the households had safe hot water temperatures. Water temperatures ranged from as low as 60°F to as high as 170°F. In the bivariate analysis, this outcome was associated with respondent education, race/ethnicity, homeowner status and having children in the home (Table I). Homes with children were less likely to have safe hot water temperatures in comparison to homes without children. ### Beliefs and knowledge For the outcome of having working SAs on every level, the mean scores on the belief scales ranged from 2.58 for perceived susceptibility to 4.38 for perceived benefits of working SAs (on a 5-point scale) and self-efficacy was scored 3.60 on average (on a 4-point scale). The average percent correct score on the fire safety knowledge measure was 57% (Table II). Table II. Health belief model constructs and knowledge about smoke alarms among a sample of residents in 12 census tracts in East Baltimore, 2009 (N = 603) Scale Items Mean Min Max Barriersa • Smoke alarms are a hassle when they go off by accident, like when we are cooking. (R) • We'd rather put up with the noise of a smoke alarm than take out the battery when it goes off by accident. • With everything going on in my life, it's hard to remember to change the smoke alarm battery. (R) • It's possible someone in this house might take a battery out of a smoke alarm to use for something else. (R) • Smoke alarms are more trouble than they are worth. (R) 3.55 1.6 Susceptibilityb • How likely is it that there will be a house fire in your neighborhood in the next few years? • How likely is it that there will be a fire in your home in the next few years? 2.58 Severitya • Compared to other household dangers, house fires are a serious problem. • Compared to other problems in Baltimore City, house fires are a serious problem. • Someone living in this house could die in a house fire. • The people living in this house could lose everything in a house fire. 4.03 1.75 Benefitsa • If there was a fire in this home, we'd be okay even if we didn't have a smoke alarm. (R) • Smoke alarms save lives. • Having a working smoke alarm is the best way to keep my family safe in a house fire. 4.38 2.67 Self-efficacyc • How confident are you that you can keep your smoke alarms working? • How confident are you that you can make sure no one in this house takes the batteries out of a smoke alarm to use for something else? • When your smoke alarm goes off by accident, how confident are you that you can stop the beeping without taking it apart or taking it down? 3.60 1.33 Knowledged • Some smoke alarm batteries last for 10 years, True. • Most fatal fires happen in homes between midnight and 6 AM. • In fires, smoke can be just as dangerous as the flames, True. • While you are sleeping, the most reliable way to know that a fire has started in your house is hearing the smoke alarm. • You should press the hush button or fan away the smoke if your smoke alarm goes off by accident (like when someone is cooking). • A good fire escape plan for your home should include making sure you have two exits from every room. • The most common cause of house fires is cooking. • For Baltimore City renters, landlords are legally responsible for putting smoke alarms in a home. • For Baltimore City renters, tenants are legally responsible for making sure smoke alarms are working. 0.57 Scale Items Mean Min Max Barriersa • Smoke alarms are a hassle when they go off by accident, like when we are cooking. (R) • We'd rather put up with the noise of a smoke alarm than take out the battery when it goes off by accident. • With everything going on in my life, it's hard to remember to change the smoke alarm battery. (R) • It's possible someone in this house might take a battery out of a smoke alarm to use for something else. (R) • Smoke alarms are more trouble than they are worth. (R) 3.55 1.6 Susceptibilityb • How likely is it that there will be a house fire in your neighborhood in the next few years? • How likely is it that there will be a fire in your home in the next few years? 2.58 Severitya • Compared to other household dangers, house fires are a serious problem. • Compared to other problems in Baltimore City, house fires are a serious problem. • Someone living in this house could die in a house fire. • The people living in this house could lose everything in a house fire. 4.03 1.75 Benefitsa • If there was a fire in this home, we'd be okay even if we didn't have a smoke alarm. (R) • Smoke alarms save lives. • Having a working smoke alarm is the best way to keep my family safe in a house fire. 4.38 2.67 Self-efficacyc • How confident are you that you can keep your smoke alarms working? • How confident are you that you can make sure no one in this house takes the batteries out of a smoke alarm to use for something else? • When your smoke alarm goes off by accident, how confident are you that you can stop the beeping without taking it apart or taking it down? 3.60 1.33 Knowledged • Some smoke alarm batteries last for 10 years, True. • Most fatal fires happen in homes between midnight and 6 AM. • In fires, smoke can be just as dangerous as the flames, True. • While you are sleeping, the most reliable way to know that a fire has started in your house is hearing the smoke alarm. • You should press the hush button or fan away the smoke if your smoke alarm goes off by accident (like when someone is cooking). • A good fire escape plan for your home should include making sure you have two exits from every room. • The most common cause of house fires is cooking. • For Baltimore City renters, landlords are legally responsible for putting smoke alarms in a home. • For Baltimore City renters, tenants are legally responsible for making sure smoke alarms are working. 0.57 aScores range from 1 to 5 (strongly agree to strongly disagree). bScores range from 1 to 5 (very likely to very unlikely). cScores range from 1 to 4 (very confident to not at all confident). dScores range from 0 to 1 (correct or incorrect). (R), item is reverse coded. For safe hot water temperature, the mean scores on the belief scales ranged from 3.08 for perceived susceptibility to 3.78 for perceived benefits, and self-efficacy was scored 3.19 on average. The average percent correct score on the knowledge measure was 34% (Table III). Table III. Health belief model constructs and knowledge about safe hot water temperatures among a sample of residents in 12 census tracts in East Baltimore, 2009 (N = 603) Scale Items Mean Min Max Barriersa • I have no idea how to change the temperature setting on my hot water heater. (R) • With everything going on in my life, I have more important things to worry about than testing the temperature of my hot water. (R) • The people living in this house will complain if the hot water temperature is set at the recommended temperature. (R) • Lowering the temperature of my hot water is more trouble than it's worth. (R) 3.47 Susceptibilityb • How likely is that someone in your home could get a scald burn by accident from the temperature of your hot water? • How likely is it that testing the temperature of your hot water could prevent a scald burn to someone living in this house? 3.08 Severitya • Scald burns from hot water can be as bad as burns from a fire. • Compared to other household dangers, scald burns from hot water are a serious problem. • Compared to other problems in Baltimore City, scald burns from hot water are a serious problem. • Someone living in this house could die from an accidental scald burn. 3.35 1.25 Benefitsa • The people living in this house will be safe from scald burns even if the hot water temperature is too high. (R) • Lowering the temperature on my water heater is the easiest way to prevent scald burns. • Lowering the temperature on my water heater can save me money. 3.78 Self-efficacyc • How confident are you that you can test the temperature of the hot water in your house? • How confident are you that you can change the hot water temperature setting on your water heater? • How confident are you that you can convince the people living in this house to keep the water temperature at the recommended level? 3.19 Knowledged • Babies’ skin burns faster than adults’, True. • The recommended temperature setting for hot water heaters is 120°F or warm setting. • The recommended water temperature for bathing is 90° to 100°F. 0.34 Scale Items Mean Min Max Barriersa • I have no idea how to change the temperature setting on my hot water heater. (R) • With everything going on in my life, I have more important things to worry about than testing the temperature of my hot water. (R) • The people living in this house will complain if the hot water temperature is set at the recommended temperature. (R) • Lowering the temperature of my hot water is more trouble than it's worth. (R) 3.47 Susceptibilityb • How likely is that someone in your home could get a scald burn by accident from the temperature of your hot water? • How likely is it that testing the temperature of your hot water could prevent a scald burn to someone living in this house? 3.08 Severitya • Scald burns from hot water can be as bad as burns from a fire. • Compared to other household dangers, scald burns from hot water are a serious problem. • Compared to other problems in Baltimore City, scald burns from hot water are a serious problem. • Someone living in this house could die from an accidental scald burn. 3.35 1.25 Benefitsa • The people living in this house will be safe from scald burns even if the hot water temperature is too high. (R) • Lowering the temperature on my water heater is the easiest way to prevent scald burns. • Lowering the temperature on my water heater can save me money. 3.78 Self-efficacyc • How confident are you that you can test the temperature of the hot water in your house? • How confident are you that you can change the hot water temperature setting on your water heater? • How confident are you that you can convince the people living in this house to keep the water temperature at the recommended level? 3.19 Knowledged • Babies’ skin burns faster than adults’, True. • The recommended temperature setting for hot water heaters is 120°F or warm setting. • The recommended water temperature for bathing is 90° to 100°F. 0.34 aScores range from 1 to 5 (strongly disagree to strongly agree). bScores range from 1 to 5 (very unlikely to very likely). cScores range from 1 to 4 (not at all confident to very confident). dScores range from 0 to 1 (incorrect or correct). (R), item is reverse coded. Beliefs about barriers, severity and self-efficacy and knowledge were significantly associated with having a working SA in the bivariate analyses. Participants with a working SA had on average higher scores on the barriers, self-efficacy and knowledge scales compared with those without a working SA. Related, those without a working SA had on average higher scores on the severity scale compared with participants with a working SA. For safe hot water temperatures, all the belief items were significantly associated with the outcome, but knowledge was not. Participants who had safe hot water temperatures had on average higher scores on the barriers, benefits and self-efficacy scales compared with those who did not have safe hot water temperatures, while participants who did not have safe hot water temperatures had on average higher scores on the susceptibility and severity scales compared with those who had safe hot water temperatures (Table IV). Table IV. Bivariate relationship between health belief model constructs and knowledge and smoke alarm and hot water temperature outcomes among a sample of residents in 12 census tracts in East Baltimore, 2009 Working SA on every level if home Safe hot water temperature (<120°F) HBM constructs Yes (n = 241), mean (SD) No (n = 361), mean (SD) t P Yes (n = 339), mean (SD) No (n = 259), mean (SD) t P Barriersa 3.63 (0.55) 3.48 (0.59) −3.1 0.002* 3.54 (0.68) 3.36 (0.61) −3.3 0.001* Susceptibilityb 2.56 (0.82) 2.60 (0.83) 0.56 0.58 2.84 (0.96) 3.38 (1.1) 6.5 0.000* Severitya 3.93 (0.66) 4.09 (0.63) 3.1 0.002* 3.29 (0.73) 3.44 (0.65) 2.5 0.01* Benefitsa 4.41 (0.48) 4.35 (0.50) −1.3 0.2 3.83 (0.57) 3.70 (0.61) −2.5 0.01* Self-efficacyc 3.69 (0.42) 3.53 (0.55) −3.8 0.0001* 3.28 (0.81) 3.08 (0.79) −3.0 0.003* Knowledged 0.59 (0.13) 0.56 (0.15) −3.0 0.003* 0.33 (0.19) 0.34 (0.19) 0.62 0.54 Working SA on every level if home Safe hot water temperature (<120°F) HBM constructs Yes (n = 241), mean (SD) No (n = 361), mean (SD) t P Yes (n = 339), mean (SD) No (n = 259), mean (SD) t P Barriersa 3.63 (0.55) 3.48 (0.59) −3.1 0.002* 3.54 (0.68) 3.36 (0.61) −3.3 0.001* Susceptibilityb 2.56 (0.82) 2.60 (0.83) 0.56 0.58 2.84 (0.96) 3.38 (1.1) 6.5 0.000* Severitya 3.93 (0.66) 4.09 (0.63) 3.1 0.002* 3.29 (0.73) 3.44 (0.65) 2.5 0.01* Benefitsa 4.41 (0.48) 4.35 (0.50) −1.3 0.2 3.83 (0.57) 3.70 (0.61) −2.5 0.01* Self-efficacyc 3.69 (0.42) 3.53 (0.55) −3.8 0.0001* 3.28 (0.81) 3.08 (0.79) −3.0 0.003* Knowledged 0.59 (0.13) 0.56 (0.15) −3.0 0.003* 0.33 (0.19) 0.34 (0.19) 0.62 0.54 aScores range from 1 to 5 (strongly disagree to strongly agree). bScores range from 1 to 5 (very unlikely to very likely). cScores range from 1 to 4 (not at all confident to very confident). dScores range from 0 to 1 (incorrect or correct). *P-value is significant (P ≤ 0.05). ### Multiple logistic regression models of behavioral outcomes For having working SAs on every level, significant correlates included reporting low perceived severity and high perceived self-efficacy. For every unit increase in our measures, the odds decreased 0.66 for every unit increase in the measure of perceived severity, while the odds of having working SAs increased 2.29 times for self-efficacy (Table V). None of the sociodemographic variables retained their significance in the multivariate model. Table V. Multiple logistic regression analysis of association between sociodemographic variables, health belief model constructs and knowledge, and SA and hot water temperature outcomes among a sample of residents in 12 census tracts in East Baltimore, 2009 Working smoke alarm on every level Safe hot water temperature Unadjusted Modela Main effects model (n = 472) Unadjusted modelb Main effects model (n = 554) OR 95%CI OR 95%CI OR 95%CI OR 95%CI Sex Male Female 0.77 0.54, 1.10 0.78 0.55, 1.12 Age (years) 18–24 25–34 2.15* 1.25, 3.70 1.61 0.81, 3.21 0.96 0.57, 1.61 35–44 1.78** 1, 3.16 1.37 0.66, 2.81 0.81 0.46, 1.41 45–54 0.88 0.48, 1.62 0.66 0.30, 1.43 1.06 0.60, 1.86 ≥55 1.35 0.76, 2.42 0.82 0.38, 1.74 1.55 0.88, 2.73 Education Less than high school diploma/GED 1.41 0.83, 2.39 1.41 0.72, 2.74 1.25 0.75, 2.08 1.29 0.73, 2.30 High school diploma/GED Some college 1.66** 1.04, 2.65 1.43 0.79, 2.63 1.13 0.71, 1.78 0.82 0.49, 1.38 Completed college 2.16* 1.44, 3.23 1.27 0.70, 2.31 2.02* 1.34, 3.02 1.51 0.91, 2.52 Per capita income$5000 or less
$5001–10 000 1.33 0.78, 2.27 1.42 0.78, 2.59 0.78 0.46, 1.32$10 001–25 000 1.28 0.77, 2.14 0.94 0.49, 1.79 1.1 0.67, 1.82
$25 000 or more 2.03* 1.22, 3.36 1.38 0.69, 2.77 1.12 0.68, 1.84 Race/ethnicity Black or African American 0.61* 0.43, 0.85 0.73 0.45, 1.19 0.55* 0.39, 0.78 0.89 0.52, 1.15 Other Homeowner status Rent Own or pay mortgage 0.76 0.54, 1.06 1.63* 1.16, 2.27 1.18 0.95, 2 Children in home (<18 years) Yes 0.97 0.69, 1.35 0.62* 0.44, 0.86 0.73 0.52, 1.15 No HBM mean scale scores Barriers 1.58* 1.18, 2.11 1.02 0.70, 1.50 1.52* 1.18, 1.96 1.33 0.96, 1.84 Susceptibility 0.94 0.77, 1.15 0.59* 0.50, 0.70 0.63* 0.52, 0.76 Severity 0.67* 0.52, 0.87 0.66** 0.48, 0.91 0.74** 0.58, 0.94 0.98 0.73, 1.31 Benefits 1.24 0.89, 1.73 1.43** 1.08, 1.88 1.39** 1.02, 1.92 Self-efficacy 1.97* 1.38, 2.81 2.29* 1.42, 3.70 1.36* 1.11, 1.66 1.23 0.95, 1.59 Knowledge % items correct 1.02* 1.01, 1.03 1.01 1, 1.02 0.99, 1.01 Working smoke alarm on every level Safe hot water temperature Unadjusted Modela Main effects model (n = 472) Unadjusted modelb Main effects model (n = 554) OR 95%CI OR 95%CI OR 95%CI OR 95%CI Sex Male Female 0.77 0.54, 1.10 0.78 0.55, 1.12 Age (years) 18–24 25–34 2.15* 1.25, 3.70 1.61 0.81, 3.21 0.96 0.57, 1.61 35–44 1.78** 1, 3.16 1.37 0.66, 2.81 0.81 0.46, 1.41 45–54 0.88 0.48, 1.62 0.66 0.30, 1.43 1.06 0.60, 1.86 ≥55 1.35 0.76, 2.42 0.82 0.38, 1.74 1.55 0.88, 2.73 Education Less than high school diploma/GED 1.41 0.83, 2.39 1.41 0.72, 2.74 1.25 0.75, 2.08 1.29 0.73, 2.30 High school diploma/GED Some college 1.66** 1.04, 2.65 1.43 0.79, 2.63 1.13 0.71, 1.78 0.82 0.49, 1.38 Completed college 2.16* 1.44, 3.23 1.27 0.70, 2.31 2.02* 1.34, 3.02 1.51 0.91, 2.52 Per capita income$5000 or less
$5001–10 000 1.33 0.78, 2.27 1.42 0.78, 2.59 0.78 0.46, 1.32$10 001–25 000 1.28 0.77, 2.14 0.94 0.49, 1.79 1.1 0.67, 1.82
\$25 000 or more 2.03* 1.22, 3.36 1.38 0.69, 2.77 1.12 0.68, 1.84
Race/ethnicity
Black or African American 0.61* 0.43, 0.85 0.73 0.45, 1.19 0.55* 0.39, 0.78 0.89 0.52, 1.15
Other
Homeowner status
Rent
Own or pay mortgage 0.76 0.54, 1.06   1.63* 1.16, 2.27 1.18 0.95, 2
Children in home (<18 years)
Yes 0.97 0.69, 1.35   0.62* 0.44, 0.86 0.73 0.52, 1.15
No
HBM mean scale scores
Barriers 1.58* 1.18, 2.11 1.02 0.70, 1.50 1.52* 1.18, 1.96 1.33 0.96, 1.84
Susceptibility 0.94 0.77, 1.15   0.59* 0.50, 0.70 0.63* 0.52, 0.76
Severity 0.67* 0.52, 0.87 0.66** 0.48, 0.91 0.74** 0.58, 0.94 0.98 0.73, 1.31
Benefits 1.24 0.89, 1.73   1.43** 1.08, 1.88 1.39** 1.02, 1.92
Self-efficacy 1.97* 1.38, 2.81 2.29* 1.42, 3.70 1.36* 1.11, 1.66 1.23 0.95, 1.59
Knowledge
% items correct 1.02* 1.01, 1.03 1.01 1, 1.02 0.99, 1.01

aNs for these analyses range from 480 to 602 due to missing data. bNs for these analyses range from 477 to 598 due to missing data. P-values: *≤0.01, **≤0.05. OR, odds ratio.

For having safe hot water temperatures, significant correlates were reporting lower perceived susceptibility and higher perceived benefits. The odds of safe hot water temperatures were decreased 0.63 for every unit increase in the measure of perceived susceptibility and the odds were increased 1.39 times for every unit increase of perceived benefits (Table V). None of the sociodemographic variables retained their significance in the multivariate model.

## Discussion

Among this large sample of urban households, we found unacceptably low rates of protection from fires and scald burns. Only 40% of households had working SAs on every level and 57% had safe hot water temperatures. Our results provide some insight into who is most at risk for inadequate protection and what might help address their needs.

### Smoke alarms

Our study indicates that the prevalence of homes with working SAs on every level may in fact be far below what is generally found in national surveys, but consistent with findings from earlier observational studies [33, 34]. For instance, in 1999, the most recent Behavioral Risk Factor Surveillance System report to assess SA prevalence found that 96% of participants self-reported having a SA in their homes [35]. This compares to the State of Home Safety in America, a national telephone survey conducted in 2002, which found that 90% of US households reported having at least one working SA, while 80% reported having one on each level of the home [36]. In a review of literature on SA prevalence rates in high-risk households, Liu et al. [37] found that out of 18 articles on the topic, 17 showed a prevalence below 95% and 7 showed a prevalence below 80%. The large gap between the rates of reported SA coverage and the observed rates that we and others have found [30, 31] should raise concerns for both researchers and practitioners interested in promoting SA use. In the absence of better ways of asking individuals about their SA behaviors, observing and testing SAs is warranted.

Although we found significant sociodemographic correlates of having working SAs on every level in the bivariate analyses, no sociodemographic variables retained significance in the main effects multiple logistic regression model. Our results are similar to those obtained by Hapgood et al. [21], leading us to also conclude that most of the variation in safety practices, particularly SA use, cannot be explained by sociodemographic factors. Of the HBM constructs we examined, lower perceived severity and higher self-efficacy increased the likelihood that working SAs were observed. It is interesting that these results differ from Roberts et al. [22] and Durand et al. [23] who suggested that low perceived risk and high perceived barriers may explain why individuals choose not to have working SAs. Differences in methods and samples may explain these inconsistencies and underscore the need for more research to understand how psychosocial factors such as health beliefs influence safety behaviors, which historically has been a neglected area of study [29, 38].

Knowledge levels were low on average, with a 57% correct response rate for the nine items we measured. These items tapped into content related to causes and consequences of house fires, how SAs should be used and who is responsible for their maintenance, as well as what is required for escape planning. Those with higher knowledge scores were only slightly more likely to have working SAs on every level (odds ratio = 1.01; 95% confidence interval: 1, 1.02), which suggests the need for increased public education to raise awareness of home fire safety. Despite years of public information concerning SAs (e.g. ‘change your clocks, change your batteries’), our results indicate that at least in this urban and generally low-income sample, efforts are needed to increase basic understanding about fire safety for households engaging in the safety behavior as well as those who are not.

On average, respondents scored high on confidence in their ability to keep their alarms in good working order and to stop nuisance alarms (a commonly heard complaint about alarms), which suggested that high scores on this scale were associated with having confidence in one’s ability to maintain the SA. These results suggest that messages demonstrating the ease of maintaining SAs could increase perceived efficacy and potentially increase their use. Interestingly, average scores on beliefs about benefits (e.g. SAs save lives) were unrelated to the outcomes but were quite high (4.38 out of 5) suggesting that the lifesaving ability of SAs is widely understood. However, respondents’ high perceived severity of house fires (4.03 out of 5 on average) was significantly inversely related to the outcomes. That is, those with low perceived severity were more likely to have working SAs, which makes sense given the cross-sectional nature of the data. For instance, those who were protected by having SAs were less likely to agree that people living in their homes would be harmed in the event of a fire and that house fires are a serious problem where they live. These findings suggest that rather than messages focused on ‘smoke alarms save lives’, messages framed around how SAs can reduce the seriousness of house fires for the individual residents as well as the community may be more effective.

### Hot water temperatures

Consistent with findings from previous studies, our results indicate that the prevalence of safe hot water temperatures in homes is low. In a study in East Baltimore conducted in the 1990s [20], we found that 47% of a clinic-based sample had safe hot water temperatures in their homes. Similarly concerning results were obtained two decades ago when Sharp and Carter [34] found that 73% of a predominantly low-income, African American population residing in an urban setting in the south had temperatures in excess of 130°F. More than two decades ago, water heater manufacturers voluntarily adopted a standard of pre-setting new water heaters at 120°F although many areas of the United States still do not regulate pre-setting of water heaters [39, 40], and as these data suggest many homes remain unprotected today. Other evidence suggests that thermostats on water heaters can be inaccurate, with the actual temperature of water stored in the heater up to 30°F above the thermostat setting; thermostatic mixing valves that regulate the temperature at the faucet are increasingly being preferred for safety [41, 42]. Similar to the findings regarding SAs, the significant bivariate relationships between sociodemographic characteristics and adequate protection, which included education, race/ethnicity and having children younger than 18 years living in the home, disappeared in the main effects multiple logistic regression model. When we looked at other potential determinants, we found that all five of the HBM variables were significant correlates in the bivariate analyses. However, in the main effects model, only the perceived susceptibility and perceived benefits variables retained significance.

Knowledge about safe hot water temperatures was woefully low (34% average percent correct). Likewise, we found that scores were generally lower for each of the beliefs scales when compared with those focused on SAs; for example, none of the four HBM mean scale scores (benefits, barriers, susceptibility, severity) was above 4 (out of 5) for the hot water items, whereas two of them were above 4 for SAs. These findings suggest that the issue of hot water scald burns is simply not as salient as SAs among the public. There certainly have not been the same long-standing public information campaigns to raise awareness and encourage residents to test water temperatures and turn down the temperature setting on their water heaters. Our findings show that much more can and should be done to promote public awareness of hot water scald burns and how to prevent them.

Messages should promote the perceived benefits of lowering water temperatures (scald burn prevention, saving money) and susceptibility to scald burns. We found perceived benefits to be significantly associated with having safe hot water temperatures. Perceived susceptibility to scald burns was significantly inversely related to the outcome, meaning that those who had safe hot water temperatures were less likely to believe that someone in their home could experience a scald burn and that testing the water temperature was necessary. Although this finding at first may seem counterintuitive in that high perceived susceptibility is thought to lead to behavior change to reduce risk, such cause and effect relationships cannot be determined with cross-sectional data and it is equally conceivable that because one has adopted the safer behavior, one’s perceived susceptibility is reduced.

### Summary

This study sheds new light on fire and scald burn risks in a large urban sample of predominantly low-income families and results should raise concerns about how inadequately protected most homes are. Findings about residents’ knowledge and beliefs can help shape future prevention efforts and persuasive messages.

The strength of our study design’s use of observed outcomes should be balanced against some study limitations. Despite our best efforts to obtain a population-based sample, the large number of randomly selected households that were unreachable and refused resulted in a selected sample. We used appropriate weighting procedures to address this limitation and found small and insignificant differences in the results. We also utilized multivariate models to control for sociodemographic variables in drawing inferences about the relative importance of knowledge and beliefs in distinguishing between residents with and without SAs and safe hot water temperatures. Finally, this study did not explore environmental factors that may serve as barriers to action, such as access to products and services needed to obtain and install SAs or to have help turning down the temperature settings on water heaters. However, Baltimore is advantaged in this regard in that the fire department will install SAs free of charge and most homes are equipped with gas water heaters, which residents can adjust easily if they have access to the water heater.

### Conclusion

This study demonstrates the urgent need to increase the number of homes with working SAs and safe hot water temperatures and provides suggestions for communication campaigns to address those needs. An estimated 60% of homes did not have a working SA on every level and 43% did not have safe hot water temperatures. Messages focused on how to provide a safe home environment could maximize their effectiveness by communicating the ease and harm reduction features of SAs and the benefits and risk reduction features of safe hot water temperatures.

## Acknowledgements

The authors wish to acknowledge and thank all the community partner organizations and individuals who contributed to the Johns Hopkins Home Safety Study. They include the Baltimore City Fire Department, Environmental Justice Partnership, Johns Hopkins CARES Mobile Safety Center, Johns Hopkins Urban Health Institute and Maryland Department of Health and Mental Hygiene.

## Funding

Centers for Disease Control and Prevention, National Center for Injury Prevention and Control (R18CE001339 to A.C.G.) and the National Institutes of Health, National Institute of Child Health and Human Development (RO1HD059216 to A.C.G.).

None declared.

## References

1
Centers for Disease Control and Prevention
Web-based Injury Statistics Query and Reporting System (WISQARS)—Fataly Injury Report [online]
,
2009

Available at: http://www.cdc.gov/injury/wisqars/fatal.html. Accessed: 23 March 2012
2
Centers for Disease Control and Prevention
Web-based Injury Statistics Query and Reporting System (WISQARS)—Non-Fatal Injury Report [online]
,
2009

Available at: ww.cdc.gov/ncipc/wisqars. Accessed: 1 April 2012
3
Runyan
CW
Johnson
RM
Yang
J
, et al.  .
Risk and protective factors for fires, burns, and carbon monoxide poisoning in US households
Am J Prev Med
,
2005
, vol.
28
(pg.
102
-
8
)
4
Flynn
J
Characteristics of Home Fire Victims
,
2010
Quincy, MA
National Fire Protection Association, Fire Analysis and Research Division
5
Bell
NJ
Schuurman
N
S
A small-area population analysis of socioeconomic status and incidence of severe burn/fire-related injury in British Columbia, Canada
Burns
,
2009
, vol.
35
(pg.
1133
-
41
)
6
Schachterle
SE
Bishai
D
Shields
W
, et al.  .
Proximity to vacant buildings is associated with increased fire risk in Baltimore, Maryland, homes
Inj Prev
,
2011
, vol.
18
(pg.
98
-
102
)
7
Diguiseppi
C
Edwards
P
Godward
C
, et al.  .
Urban residential fire and flame injuries: a population based study
Inj Prev
,
2000
, vol.
6
(pg.
250
-
4
)
8
National Fire Protection Association
An Overview of the US Fire Problem

n.d. Available at: http://www.nfpa.org/assets/files/PDF/FireOverview.pdf. Accessed: 26 March 2012
9
Karter
MJ
Fire Loss in the United States During 2009
,
2010
Quincy, MA
National Fire Protection Association, Fire Analysis and Research Division
10
McLoughlin
E
McGuire
A
The causes, cost, and prevention of childhood burn injuries
Am J Dis Child
,
1990
, vol.
144
(pg.
677
-
83
)
11
Feldman
KW
Schaller
RT
Feldman
JA
, et al.  .
Tap water scald burns in children
Inj Prev
,
1998
, vol.
4
(pg.
238
-
42
)
12
Baptiste
M
Feck
G
Preventing tap water burns
Am J Public Health
,
1980
, vol.
70
(pg.
727
-
9
)
13
Safe Kids USA
Burn and Scald Prevention Fact Sheet
,
2011

14
Mallonee
S
Istre
GR
Rosenberg
M
, et al.  .
Surveillance and prevention of residential-fire injuries
New Engl J Med
,
1996
, vol.
335
(pg.
27
-
31
)
15
Moritz
AR
Henriques
F
Jr
Studies of thermal injury: II. The relative importance of time and surface temperature in the causation of cutaneous burns
Am J pathol
,
1947
, vol.
23
pg.
695

16
Ballesteros
MF
Kresnow
M
Prevalence of residential smoke alarms and fire escape plans in the US: results from the Second Injury Control and Risk Survey (ICARIS-2)
Public Health Rep
,
2007
, vol.
122
(pg.
224
-
31
)
17
Chen
L
Gielen
A
McDonald
E
Validity of self reported home safety practices
Inj Prev
,
2003
, vol.
9
(pg.
73
-
5
)
18
Ahrens
M
Smoke Alarms in US Home Fires
,
2009
Quincy, MA
National Fire Protection Association, Fire Analysis and Research Division
19
Macarthur
C
Evaluation of Safe Kids Week 2001: prevention of scald and burn injuries in young children
Inj Prev
,
2003
, vol.
9
(pg.
112
-
6
)
20
Gielen
AC
McDonald
EM
Wilson
MEH
, et al.  .
Effects of improved access to safety counseling, products, and home visits on parents’ safety practices: results of a randomized trial
,
2002
, vol.
156
(pg.
33
-
40
)
21
Hapgood
R
Kendrick
D
Marsh
P
How well do socio-demographic characteristics explain variation in childhood safety practices?
J Public Health
,
2000
, vol.
22
(pg.
307
-
11
)
22
Roberts
H
Curtis
K
Liabo
K
, et al.  .
Putting public health evidence into practice: increasing the prevalence of working smoke alarms in disadvantaged inner city housing
J Epidemiol Community Health
,
2004
, vol.
58
(pg.
280
-
5
)
23
Durand
MA
Green
J
Edwards
P
, et al.  .
Perceptions of tap water temperatures, scald risk and prevention among parents and older people in social housing: a qualitative study
Burns
,
2011
, vol.
38
(pg.
585
-
90
)
24
Hochbaum
G
Public Participation in Medical Screening Programs: A Sociopsychological Study
,
1958
Washington, DC
Government Printing Office

Contract No. 572
25
Rosenstock
I
The health belief model and preventive health behavior
Health Educ Monogr
,
1974
, vol.
50
(pg.
354
-
86
)
26
Rosenstock
IM
What research in motivation suggests for public health
Am J Public Health
,
1960
, vol.
50
(pg.
295
-
302
)
27
Rosenstock
IM
Strecher
VJ
Becker
MH
Social learning theory and the health belief model
Health Educ Behav
,
1988
, vol.
15
(pg.
175
-
83
)
28
Chamption
VL
Skinner
CS
Glanz
K
Rimer
BK
Viswanath
K
The health belief model
Health Behavior and Health Education Theory, Research, and Practice
,
2008
4 edn
San Francisco, CA
Jossey-Bass
29
Trifiletti
L
Gielen
A
Sleet
D
, et al.  .
Behavioral and social sciences theories and models: are they used in unintentional injury prevention research?
Health Educ Res
,
2005
, vol.
20
(pg.
298
-
307
)
30
Gielen
A
Shields
W
Frattaroli
S
, et al.  .
Enhancing fire department home visiting programs: results of a community intervention trial
J Burn Care Res
,
2012

Dec 12 [Epub ahead of print, doi:10.1097/BCR.0b013e3182685b3a]
31
Beatty
PC
Willis
GB
Research synthesis: the practice of cognitive interviewing
Public Opin Q
,
2007
, vol.
71
(pg.
287
-
311
)
32
StataCorp
Stata Statistical Software: Release 11
,
2009
College Station, TX
StataCorp LP
33
Jones
AR
Thompson
CJ
Davis
MK
Smoke alarm ownership and installation: a comparison of a rural and a suburban community in Georgia
J Community Health
,
2001
, vol.
26
(pg.
307
-
29
)
34
Sharp
GB
Carter
MA
Prevalence of smoke detectors and safe tap-water temperatures among welfare recipients in Memphis, Tennessee
J Community Health
,
1992
, vol.
17
(pg.
351
-
65
)
35
Centers for Disease Control and Prevention
Behavioral Risk Factor Surveillance System Survey Data
,
1999
Atlanta, GA
US Department of Health and Human Services
36
Runyan
CW
Casteel
C
The State of Home Safety in America: Facts about Unintentional Injuries in the Home
,
2004
2nd edn
Washington, DC
Home Safety Council
37
Liu
Y
Holland
AE
Mack
K
, et al.  .
Disparities in the prevalence of smoke alarms in US households: conclusion drawn from published case studies
J Safety Res
,
2011
, vol.
42
(pg.
409
-
13
)
38
Gielen
A
Sleet
D
DiClemente
R
Injury and Violence Prevention: Behavioral Sciences Theories, Methods, and Applications
,
2006
San Francisco, CA
Jossey-Bass
39
Erdmann
TC
Feldman
KW
Rivara
FP
, et al.  .
Tap water burn prevention: the effect of legislation
Pediatrics
,
1991
, vol.
88
(pg.
572
-
7
)
40
Staunton
CE
Frumkin
H
Dannenberg
AL
Doll
L
Sandra
S
Sleet
P
, et al.  .
Changing the built environment to prevent injury
Handbook of Injury and Violence Prevention
,
2007
New York, NY
Springer
(pg.
257
-
75
)
41
George
R
Plumbing to Prevent Domestic Hot Water Scalds
,
2011

Available at: http://www.hgexperts.com/article.asp?id=5135. Accessed: 4 March 2013
42
Shields
W
Perry
E
McDonald
E
, et al.  .
Too hot to handle: an examination of hot water temperature and water heaters in a sample of urban homes
American Burn Association Annual Meeting
,
2012
Seattle, WA