There are nearly 2 million fire emergencies per year in the United States, resulting in thousands of deaths, tens of thousands of injuries, and billions of dollars lost (United States Fire Association 2004). Furthermore, statistics show that one- and two-family dwellings accounted for 73% of fires, 78% of deaths, 67% of injuries, and 76 % of dollar loss in 2001 (USFA 2004). Considering that more and more persons with disabilities are living in community residences, it is important to teach the fire safety skills that will keep them safe in a fire.

Researchers have successfully taught fire safety information to children and individuals with mental retardation (Matson 1980; McConnell et al. 1996; Mori and Peterson 1986). According to McConnell et al. (1996) these educational programs may increase the probability of a child living through a home-fire emergency. However, exhibiting knowledge of fire-safety skills does not guarantee the individual will take action to escape a dangerous fire situation.

This problem has been demonstrated with other safety skills showing that increases in knowledge do not necessarily correspond to increases in skills (Gatheridge et al. 2004; Himle et al. 2004; Lumley et al. 1998; Miltenberger et al. 1999; Olsen-Woods et al. 1998). Researchers may do a disservice to children or individuals with mental retardation by providing educational programs rather than teaching them the actual skills needed to stay safe (e.g. Himle and Miltenberger 2004; Roberts and Miltenberger 1999). What is needed are specific fire-safety training programs designed to teach and evaluate actual fire-safety skills.

Several studies have taught actual fire safety skills to children (Jones et al. 1981a, b, 1989) and individuals with mental retardation (Bannerman et al. 1991; Jones and Thorton 1987; Katz and Singh 1986; Rae and Roll 1985; Shields et al. 1999). However, the one common limitation in many of these studies is the lack of a naturalistic assessment. In some studies, training focused on teaching fire safety skills and then assessed knowledge of what to do in case of a fire, but did not assess the actual skills (Katz and Singh 1986; Matson 1980; McConnell et al. 1996; Mori and Peterson 1986). In other studies, researchers trained fire safety skills and then assessed the skills while a researcher prompted the fire safety behaviors (Jones et al. 1981a, b). Some researchers tried to make assessments as natural as possible by having unannounced fire drills but failed to structure the assessments in a way that ensured the participants were blind to the fact that they were being assessed (Bannerman et al. 1991; Jones et al. 1989; Jones and Thorton 1987).

In future research on teaching fire safety skills to vulnerable populations, assessments need to be structured in the most naturalistic way possible. That is, the assessments must present a seemingly real fire emergency (an activated smoke detector or fire alarm) without the participants’ knowledge that an assessment is being conducted (e.g. Gatheridge et al. 2004; Himle and Miltenberger 2004; Roberts and Miltenberger 1999). The purpose of this study was to evaluate training procedures for teaching individuals with severe and profound mental retardation to exit their residence upon hearing a fire alarm when they were unaware that an assessment was taking place.

Method

Participants and Setting

Seven participants with severe and profound mental retardation were selected from two local group homes. The participants’ legal guardians signed consent forms and the University Institutional Review Board and the participating agencies approved this study. Training and assessments took place in the participants’ bedroom or in non-occupied areas of the group home (i.e., living room or kitchen without staff or other residents present).

Target Behaviors

The target behavior was to immediately engage in exiting behaviors until safely out of the building upon hearing a smoke detector. Data were collected on (a) the latency to respond to the activation of a smoke detector by initiating exiting behaviors and (b) the duration of time to safely exit the building. The exiting behavior was defined as rising from a chair or bed or terminating an activity and walking toward and out of an exit. The goal was for participants to initiate exiting behavior within 10 s of hearing an alarm and to leave the building within 30 s of initiating exiting behavior.

Data Collection

In situ assessments were conducted during baseline and training phases. During an in situ assessment an experimenter entered the group home without the knowledge of the participant and hid in close proximity to the participant to observe his or her response to the activation of a smoke detector. Once the experimenter was hidden from the participant, the experimenter activated the smoke detector and a stopwatch. The experimenter recorded the latency and duration of exiting behavior.

Interobserver Agreement

A second researcher observed the participants’ behavior from a hidden location and recorded the latency and duration of exiting behavior. Percentage of agreement for latency and duration was calculated by dividing the smaller number of seconds by the larger number and multiplying by 100. For participants in Group Home 1, agreement was calculated for 35% of sessions. For participants in Group Home 2 agreement was calculated for 90% of sessions. For Group 1 average agreement was 97% for latency and 99% for duration. For Group 2 mean agreement was 99% for latency and 100% for duration.

Experimental Design and Procedures

A multiple baseline across subjects design was used to evaluate the effectiveness of behavioral skills training and in situ training. Following baseline three BST sessions were conducted followed by an assessment. If the participant did not engage in the safety skills during the assessment, in situ training was initiated.

Baseline

During baseline, situ assessments were conducted in the manner described previously. Latency to respond and duration to exit were recorded in seconds. Smoke detectors remained activated until the participant arrived safely outside or until 3 min had elapsed and the participant had not initiated exiting behaviors. At either point, the experimenter came out of hiding, stopped the smoke alarms, and brought the participant back to where he or she had started. If the participant initiated exiting behaviors prior to the 3 min limit, he or she was allowed an additional minute to exit the building before the smoke detector was turned off. No feedback was given to the participants following an assessment during baseline.

Behavioral Skills Training (BST)

Training sessions began with a short description of what to do if the participant heard a smoke detector. The experimenter then activated the smoke alarm while saying “(participant’s name), FIRE, GET OUT!” The first trial consisted of the experimenter modeling the correct behaviors. Each subsequent trial involved the experimenter using the least intrusive prompt necessary to get the participant to exit the building. The prompt order was verbal (i.e., “lets get out”, “come on”), gestural (i.e., waving to the participant to follow, pointing to an exit), simple physical (i.e., slight tug on sleeve or arm), and complete physical guidance (i.e., taking the participant by the hand or arm and guiding him or her outside). Training sessions lasted 20 min or until ten trials were completed. Prompts were faded over trials so that the least intrusive prompt was used to get the participant to initiate exiting behaviors and safely make his or her way outside. Once outside, the participant was praised and provided with highly preferred tangibles and edibles as determined by a multiple stimulus without replacement preference assessment (DeLeon and Iwata 1996).

In Situ Training (IST)

An in situ assessment as described earlier was conducted following the third BST session. In situ training was conducted if the participant failed to engage in the safety skills during the assessment. In situ training occurred when a researcher turned an assessment session into a training session. In situ training consisted of an experimenter jumping into the assessment situation while saying “(participant’s name), FIRE, GET OUT!” The experimenter then used the least intrusive prompt necessary to get the participant safely outside. Each in situ training session consisted of four trials. In situ training sessions were continued until the participant initiated exiting within 10 s and left the building with 30 s for three consecutive assessments or until 9 in situ training sessions were conducted with no behavior change.

Results

Figure 1 represents latency data for Group 1 (Lori, Deanna, and Rick). During baseline, Lori did not initiate exiting behaviors when a smoke detector was activated. Following BST and IST (indicated by arrows on the graph), Lori still did not initiate exiting behaviors. Deanna initiated exiting behaviors during three of the ten baseline assessments but not within 10 s of the activation of the smoke detector. Following training, Deanna’s latency to initiate exiting behavior decreased. Deanna met the training criteria (initiated exiting behavior with 10 s) on three consecutive assessments. Rick initiated exiting behaviors in all 11 of the baseline assessments. Furthermore, Rick initiated within 10 s from the activation of a smoke detector in three of the 11 baseline assessments. Post training assessments show that Rick decreased the latency from baseline and initiated exiting behaviors within 10 s following the activation of a smoke detector in four of eight assessments.

Fig. 1
figure 1

Latency to respond with exiting behaviors during in situ assessments for participants in Group home 1. Each arrow represents an in situ training session

Full size image

Figure 2 shows duration of exiting behaviors exhibited by participants in Group 1. Lori did not exit the building within 180 s during any baseline or post-training assessments. Deanna did not exit the building within 180 s during any baseline assessment. However, she did exit the building on six of seven posttraining assessments. In her final three assessments, Deanna exited the building upon the activation of a smoke detector in under 40 s. Rick exited the building before 180 s had elapsed only once during baseline. Following training, it was decided that should Rick initiate exiting behaviors, he would be allowed more time to exit the building, due to his limited mobility. Although Rick did not exit the building in the first posttraining assessment, he exited the building in every other assessment. In situ training was instituted following the fifth assessment in an attempt to get Rick to exit faster. Following in situ training, Rick exited the building in two of three assessments.

Fig. 2
figure 2

Duration of time to exit building for participants in Group home 1. Each arrow represents an in situ training session

Full size image

Figure 3 shows data on the level of prompting used during BST and IST sessions for the participants in Group 1. Data were collected on a 5-point scale to indicate the level of prompting needed for the participant to exit the building (0 = independent behavior, 1 = verbal prompt, 2 = gestural prompt, 3 = minimal physical prompt, 4 = full physical prompt). At the start of training, Lori required physical prompts. By the end of the IST phase, Lori started to exit the building when she saw a trainer enter the room but before the verbal prompt was delivered (indicated in the figure as a score of 0.5). Deanna required verbal prompts at the beginning of training but exited the building independently to the sound of the smoke detector by the end of training. At the start of training, Rick needed a verbal prompt. By the end of training, Rick independently exited the building.

Fig. 3
figure 3

Level of prompting required to exit building during training sessions for participants in Group home 1 (0 = independent behavior, 1 = verbal prompt, 2 = gestural prompt, 3 = minimal physical prompt, 4 = full physical prompt)

Full size image

Figure 4 shows latency data for Group 2. During baseline, Scott initiated exiting behaviors in two of three assessments but not within 10 s following the activation a smoke detector. None of the other participants initiated exiting within 180 s. Following three BST sessions, it was decided that the participants from Group 2 needed further training, so instead of a post-training assessment, an IST session was immediately implemented. Each arrow indicates three separate times in which IST was implemented. Nine IST sessions were conducted with each participant but none of the participants initiated exiting behavior when the smoke detector alarm sounded.

Fig. 4
figure 4

Latency to respond with exiting behaviors during in situ assessments for participants in Group home 2. Each arrow represents 3 in situ training sessions

Full size image

Figure 5 shows duration data for Group 2. During baseline assessments, none of the participants exited the building following the activation of a smoke detector. Following BST, three additional IST sessions were implemented on three separate occasions and none of the participants exited the building upon the activation of a smoke detector.

Fig. 5
figure 5

Duration of time to exit the building during in situ assessments for participants in Group home 2. Each arrow represents 3 in situ training sessions

Full size image

Figure 6 shows prompting data collected during BST and IST sessions for Group 2. At the start of training, Scott, Jack, and Melissa needed physical prompts to exit the building to the sound of a smoke detector. As BST and IST continued, the level of prompting needed for Scott, Jack, and Melissa to exit the building did not decrease. For Megan, physical prompting was needed at the beginning of training. By the end of training only a verbal prompt was needed for her to exit the building to the sound of a smoke detector.

Fig. 6
figure 6

Level of prompting required to exit building during training sessions for participants in Group home 2 (0 = independent behavior, 1 = verbal prompt, 2 = gestural prompt, 3 = minimal physical prompt, 4 = full physical prompt)

Full size image

Discussion

The results of this study showed that one of the seven participants (Deanna) benefitted from training by learning to exit the group home quickly without prompts in response to the sound of a smoke detector. Rick was the only other participant who may have benefited from training—he initiated exiting more quickly following training and exited the building more often following training. However, the decision in baseline to terminate recording exiting behavior 3 min after the smoke detector was activated limits the conclusions that can be drawn about the effectiveness of training. Although Rick was ambulatory, he had substantial problems walking unassisted and thus walked very slowly. As a result, during the training phase, it was decided that if Rick initiated exiting behaviors and was still making an attempt to leave the building, the researchers would allow him more time without prompts to see if he did in fact independently exit the building. Because he was not given this same opportunity in baseline, it cannot be determined whether the intervention was effective for Rick.

All of the remaining participants did not meet the training criteria, thus the training appeared to be ineffective in teaching these individuals to exit their residence upon the activation of a smoke detector. However, the prompting data suggest the training did hold value for some of the remaining individuals in that following training less prompting from staff was required to get these individuals to exit the group home in response to the fire alarm.

Prompting data for Lori and Megan show that, at the start of training, they needed physical prompts to exit the building, but by the end of training they needed only verbal prompts. Furthermore, by the time the study ended, Lori exited the building to the sight of the trainer when the smoke detector was activated and the trainer was entering the room to deliver prompts. This is a valuable finding that has implications for staff behavior in the event of a real fire emergency. For example, if none of the individuals in Group home 2 had received training, in the case of a real fire emergency one could reasonably expect that their staff would have to physically guide each individual out of the home. Mere seconds in a real fire emergency can mean life or death to those inside. Based on the prompting data following training, one could conclude that in the event of a real fire emergency, staff may be able to verbally prompt Megan to get out while other residents might need full physical assistance. Having staff prepared with that knowledge could save time and potentially save lives.

Although it is not known why some participants benefitted more than others from training, one possible factor is level of mental retardation. Participants from Group home 1 were diagnosed with severe mental retardation, whereas participants from Group home 2 were all diagnosed with profound mental retardation. The relatively better training effects achieved with participants from Group home 1 may have been due in part to their higher level of functioning.

One limitation of this study is the lack of follow-up data. In future research long term assessments are needed to determine the longevity of the treatment effects. Another limitation is the lack of assessment of stimuli associated with real fire emergencies. Smoke detectors similar to those found in both group homes were used, but the researchers did not use heat or smoke to better simulate a real fire emergency. Future research might benefit from the use of such stimuli in training and assessment. The absence of nighttime assessments in this study is another limitation. Further research conducted in this area should assess fire safety skills during both daytime and nighttime assessments.

Another limitation of this study was the decision to stop data collection at 180 s with Rick during baseline. Although the treatment effects cannot be compared to baseline for this reason, it is highly unlikely that during baseline assessments Rick would have exited the building had he been given additional time. During the baseline assessments, Rick often walked pass the nearest available exit and wandered into different rooms; thus, Rick would not have met the criteria for additional time used during post-training (moving toward an exit).

This study extends the previous literature by demonstrating that behavioral skills training and in situ training were effective in teaching some individuals with developmental disabilities to engage in fire safety behaviors as assessed in naturalistic conditions. Furthermore, the results show that some of the individuals who did not learn to engage in the safety skills did respond to a less intrusive level of prompting and thus might be more likely to be safe in the event of a fire. Another beneficial outcome of this research is the knowledge that some individuals may not be able to learn these important safety skills in a reasonable period of time and, thus, staff would need to be prepared to use full physical guidance in the event of a fire.