Specific Aim 1 is to test the hypothesis that there are specific characteristics of a voice smoke alarm (i.e., use of child's first name, behavior commands in the message content, use of mother's voice, and stimulus frequency) that will awaken children 5-12 years old in stage 4 sleep. The successful children's alarm will be tested among adult subjects to evaluate effectiveness across the age spectrum.

Specific Aim 2 is to test the hypothesis that there are specific characteristics of a voice smoke alarm (i.e., use of mother's voice and behavior commands in the message content) that will result in successful completion of simulated escape behaviors by children 5-12 years old after awakening from stage 4 sleep. The successful children's alarm will be tested among adult subjects to evaluate effectiveness across the age spectrum.

Being asleep at the time of a residential fire is an important risk factor for fire-related death. Children 5-12 years of age are unlikely to be awakened by a conventional residential tone smoke alarm in the event of a fire. However, findings from our preliminary research strongly suggest that an effective and practical alarm for this age group is achievable. Building on our previous work, the objective of this study is to determine key smoke alarm characteristics that will awaken children and prompt their escape. We will achieve our study objective through two specific aims.

Specific Aim 1 is to test the hypothesis that there are specific characteristics of a voice smoke alarm (i.e., use of child's first name, behavior commands in the message content, use of mother's voice, and stimulus frequency) that will awaken children 5-12 years old in stage 4 sleep. The successful children's alarm will be tested among adult subjects to evaluate effectiveness across the age spectrum.

Specific Aim 2 is to test the hypothesis that there are specific characteristics of a voice smoke alarm (i.e., use of mother's voice and behavior commands in the message content) that will result in successful completion of simulated escape behaviors by children 5-12 years old after awakening from stage 4 sleep. The successful children's alarm will be tested among adult subjects to evaluate effectiveness across the age spectrum.

Using a randomized, non-blinded, repeated measures, clinical intervention design, our two working hypotheses as stated in Specific Aims 1 and 2 will be tested in five linked studies.

Study 1. Identification of Specific Maternal Voice Smoke Alarm Characteristics Associated With Awakening and Escaping.

Using a randomized, non-blinded, repeated measures, clinical intervention design, Study 1 will identify the critical elements (i.e., use of child's first name and/or behavior commands in message content) in the maternal voice signal that are significantly associated with EEG-defined awakening (and completion of simulated escape behaviors by children after awakening from S4). A conventional residential tone smoke alarm meeting current NFPA 72 National Fire Alarm Code will be used as a reference stimulus to allow comparison of responses to the voice alarm stimuli with responses to a conventional residential tone alarm stimulus.

Alarm stimuli are being tested during S4 in our study, because this is the deepest stage of sleep and the most refractory to arousal. If alarm stimuli are successful in awakening children from S4 and prompting them to escape, then these stimuli will be expected to be at least this successful during other stages of sleep. We want to develop an alarm that provides an effective stimulus for sleeping children at highest risk of non-response.

During Study 1, subjects will be randomly assigned to a sequence of four alarm stimuli using a Latin Square design, which will minimize the possibility of a sequence effect. Block randomization by age range (5-6, 7-8, 9-10, 11-12 years) and gender will be used to ensure that these two variables are equally represented in each of the sequences. A different alarm stimulus will be given during each S4 period of the first and second sleep cycles on two separate non-consecutive nights. Consecutive nights of testing will not be done to avoid possible confounding effects of sleep deprivation. This study's repeated measures design avoids potential confounding effects due to variation of AATs among individuals (inter-subject variability) and takes advantage of the stability of AATs for an individual among sleep cycles during a night and from night to night (intra-subject variability) (Bonnet, et al., 1978; Zepelin, et al., 1984; Bruck, 2001).

Study 2. Comparison of Mother's Versus Stranger's Voice Smoke Alarms and Alarm Frequency.

Study 2 will take the voice alarm script that was the most successful in Study 1 in awakening and prompting children to perform the simulated escape behaviors, and will compare mother's voice to a female stranger's voice using this script. This will determine whether mother's voice is a critical factor for success of the voice smoke alarm. In addition, a Temporal-Three (T-3) pattern smoke alarm with dominant tones in lower frequency ranges similar to the human voice range will be included as a stimulus in Study 2 to evaluate the influence of alarm signal frequency on EEG-defined awakening (as well as completion of simulated escape behaviors by children after awakening from S4). As in Study 1, a conventional residential tone smoke alarm will be used as a reference stimulus in Study 2. This conventional residential tone alarm has a higher frequency signal than the other T-3 tone alarm.

If more than one alarm script ties for the most effective in Study 1, then the script that is the least personalized will be used in Study 2. This decision rule is based on injury prevention theory (Baker, 1981) that dictates that a voice alarm that requires the least effort by the parent (to personalize the script with the child's name in this case) will be more likely to be implemented. Indeed, the voice alarm that would be most likely to be used by parents, and the most likely to be used correctly, is one that requires no personalization at all. For example, if a voice alarm with a pre-recorded, effective, fire safety-appropriate, urgent evacuation message (equivalent to "Wake up! Get out of bed! Leave the room!" by the female stranger's voice in this proposed study) could be purchased, and installed by parents directly from its package, it would be the most likely to result in the needed protection. If our study demonstrates that personalization of an alarm message (by use of the child's first name and mother's voice) is not associated with improved awakening and escaping, then this could eliminate the need for a recording mechanism in a voice smoke alarm, such as currently found in the KidSmart alarm. Decreasing the technical complexity of the alarm will bring down the price, which could increase its use in lower income homes at higher risk of residential fire.

The lower frequency T-3 alarm that will be used in this Study 2 is a Simplex 1996, 4100 Fire Alarm with a fundamental frequency at 500Hz with additional dominant frequencies at 1500Hz and 2500Hz. This is the same alarm previously used by Proulx and Laroche (2003) and Bruck, et al. (2004). Although use of this alarm does not evaluate the influence of alteration of the frequency of the voice alarm stimulus on awakening, it does allow for testing the effects of alarm frequency on awakening by comparing the success of the lower frequency T-3 alarm with the other alarms in Study 2. Since July 1996, the NFPA 72 National Fire Alarm Code has required that buildings equipped with a fire alarm system must sound the Temporal-Three (T-3) pattern as defined by ISO 8201 (International Organization for Standardization, 1987). The international standard addresses the signal's length, sound pressure level, and temporal aspects, but does not stipulate requirements for signal frequency. However, Bruck, et al. (2004) have suggested that alarm frequency may be the key factor in awakening sleeping children.

Using the same design as in Study 1, subjects in Study 2 will be block randomized by age range and gender to a sequence of four alarm stimuli using a Latin Square design. Using a repeated measures design, a different alarm stimulus will be given during each S4 period of the first and second sleep cycles on two separate non-consecutive nights.

Study 3. Active Comparator: Adults 20-49 Years of Age Using a randomized, non-blinded, repeated measures, clinical intervention design, our two working hypotheses as stated in Specific Aims 1 and 2 will be tested in a linked study, Study 3, with adult participants 20-49 years of age to ensure that the alarms tested will also work for adults in this age group. This arm will use the alarm signal identified in Study 2 that is significantly associated with Electroencephalography (EEG)-defined awakening and successful completion of simulated escape behaviors by children after awakening from slow wave sleep. A lower frequency tone smoke alarm will evaluate the influence of alarm signal frequency on awakening. A conventional residential tone smoke alarm will be used as a reference stimulus. Both a male and a female voice will be used as alarm stimuli. Note that these will be strangers' voices, and not a mother's voice.

Study 4. Active Comparator: Older Adults 60-84 Years of Age Study 4 of this project will take the voice alarm script in Study 2 and compare it with a low-frequency 520 Hz square wave tone smoke alarm in awakening older adults 60-84 years of age from slow wave sleep and prompting their performance of a simulated escape procedure. Note that this will necessarily be a female stranger's voice, and not a mother's voice, in this older age group. As in Studies 1 and 2, a conventional residential tone smoke alarm will be used as a reference stimulus in Study 4. In order to maintain the same experimental design across these studies, a fourth alarm type will be introduced. This fourth alarm will be a hybrid of the low-frequency 520 Hz square wave tone smoke alarm and the voice alarm, i.e., the stimulus will begin with the 520 Hz square wave tone in a T-3 pattern followed by the voice script, with this stimulus being repeated until the subject completes the escape procedure. The simulated escape procedure for older adults will be different from that of the child and adult (20-49 years of age) studies (Studies 1, 2, and 3). It will consist of the participant sitting up in his/her bed and pressing a button anchored to the wall beside the bed to stop the alarm. This will avoid any concerns of falls in this older age group upon awakening.

Study 5. Children 5-12 Years of Age (Testing Male Voice and Hybrid Tone/Voice Alarm) Using a randomized, non-blinded, repeated measures, clinical intervention design, our two working hypotheses as stated in Specific Aims 1 and 2 will be tested in Study 5 among children 5-12 years of age using the following 4 alarm stimuli: female stranger's voice, male stranger's voice, hybrid of the low-frequency 520 Hz square wave tone smoke alarm and the voice alarm (from Study 4), and conventional high frequency tone residential alarm. This study arm will allow comparison of a male versus female voice and also evaluate the hybrid low frequency tone/voice alarm among children 5-12 years of age. The same protocol will be used for children in this study arm as was used in Studies 1 and 2.

Inclusion Criteria:

• Child is at least 5 years of age and has not yet had his/her thirteenth birthday (for Studies # 1, 2, and 5).
• Subject (and child's caretaker for Studies #1, 2, and 5) speak English (2000 Census data indicate that less than 5% of the population older than five years of age in Franklin County, Ohio [where the study is located] speaks English less than "very well").
• Subject/family is able to be contacted by telephone (to obtain pre-study information and to remind the family about the study appointment and confirm that the child is in a normal state of health on day of study).
• For adult study arms: Study 3: Subjects are 20-49 years of age; Study 4: Subjects are 60-84 years of age.

Exclusion Criteria:

• Subject has a clinical diagnosis that may affect sleep, arousal or ability to perform the escape procedure.
• Subject has a hearing impairment.
• Subject is taking medication that may affect sleep, arousal, or ability to perform the escape procedure.
• Subject has an acute illness at the time of the sleep study.

The target population for Study 3 consists of adults 20-49 years of age, who have not yet had his/her fiftieth birthday. Adult participants must not have a clinical diagnosis that may affect sleep, arousal or ability to perform the simulated escape procedure. He/she must not have a hearing impairment; not be taking medication that may affect sleep, arousal, or ability to perform the escape procedure; and not have an acute illness at the time of the sleep study. The adult must speak English and be able to be contacted by telephone (to obtain pre-study information and to remind the participant about the study appointment and confirm that he/she is in a normal state of health on the day of the study).

The target population for Study 4 consists of older adults 60-84 years of age, who have not yet had his/her eighty-fifth birthday. The older adult must not have a clinical diagnosis that may affect sleep, arousal or ability to perform the escape; not have a hearing impairment; and not be taking medication that may affect sleep, arousal, or ability to perform the simulated escape procedure. The older adult must not have an acute illness at the time of the sleep study. The older adult must speak English and be able to be contacted by telephone (to obtain pre-study information and to remind the participant about the study appointment and confirm that he/she is in a normal state of health on the day of the study).

The target population for Study 5 is the same as that for Studies 1 and 2.