Kent Fire and Rescue Service
Cognitive, affective and behavioural outcomes of preschool fire safety interventions: A narrative review

After conducting a systematic literature search in Google Scholar, 4,490 hits were returned. Google Scholar presents a maximum of 1,000 results, which are ordered according to relevance. Those 1,000 hits were screened, resulting in the identification of 16 papers that satisfied the inclusion/exclusion criteria after title and abstract and full-text screening. One additional paper (Vanselow and Hanley, 2014) was identified through reference lists. In total, 17 papers detailing 18 relevant studies were found to meet the inclusion/exclusion criteria (see Appendix B). In all 18 studies, preschoolers were the recipients of the interventions. Furthermore, all the studies consisted of primary research. The paper that consisted of two studies used a different research design for each (Jones and Kazdin, 1980). Seven evaluations employed a randomised controlled study design (quality grade range: A - B). One was an uncontrolled before-and-after study (quality grade: D). Three were controlled studies that lacked a pre-test (quality grade range: C - D). Five of the evaluations utilised a single-subject design (quality grade of each: E). Two evaluations employed an uncontrolled, post-assessment approach; both relied on qualitative assessment techniques, and one also used a cross-sectional survey (quality grade of each: E). Of the 17 papers, 12 originated from the USA, two from Canada, one from both the USA and Canada, one from New Zealand, and one from Australia. Accordingly, findings pertaining to the effectiveness of interventions in these papers may not generalise to the UK context, which differs or may differ in domains that could affect the feasibility of intervention delivery and/or moderate impacts, such as the school curriculum and the operations and priorities of fire services in each nation. A range of fire safety outcomes were assessed across studies, including: knowledge; attitudes; behavioural intentions; behavioural skills; and naturally occurring behaviours, such as fire safety promotion and incident prevention behaviours. None of the studies assessed preschoolers’ responses to real fire emergencies. Nine studies evaluated individual interventions, and nine studies evaluated group interventions. In this review, the results are structured first according to study design, then according to intervention type.

 

Randomised controlled studies

Individual interventions

Three randomised controlled trial (RCT) studies evaluated interventions that adopted an individual format (Coles et al., 2007; Morrongiello et al., 2012; Reim, 2011). In a study by Coles et al. (2007), 32 children with foetal alcohol syndrome or partial foetal alcohol syndrome, aged between 4 and 10 years, were recruited from a foetal alcohol and drug exposure clinic. They were randomly assigned to an intervention condition (n = 16) or an active control condition (n = 16). All children independently played a computer game in a single laboratory session for 30 minutes or less. In the intervention condition, the computer game pertained to fire safety, whilst in the control condition it pertained to street safety. The fire safety game presented a three-dimensional world characterised by a house with a meeting place in the backyard. The player was required to respond to a fire in the house. An animated dog modelled correct safety responses. Support from the animated dog was withdrawn with increased proficiency. The game provided feedback and reinforcement in relation to the player’s actions. Children’s fire safety knowledge was assessed via child interviews before, immediately after and one week after the intervention. In addition, fire safety behavioural skills amongst children in the intervention condition were assessed through observation in a real-world simulation immediately after and one week after the intervention. It was found that knowledge of fire safety steps significantly increased in the intervention condition relative to controls, and that knowledge gains were retained one week after the intervention. In addition, the children in the intervention condition performed well in enacting appropriate fire safety behaviours in the real world immediately after the session and at the one-week follow-up. Age was not a significant covariate in the analyses.

In addition, in a study by Morrongiello et al. (2012), a community sample of 76 children aged 3 – 6 years were recruited on a convenience basis. They were randomly allocated to an intervention condition (n = 38) or an active control condition (n = 38). In the intervention condition, children played a fire safety computer game at home. Although parental support was allowed (though not required), the game was designed so that children could play it independently; in particular, there were no reading demands. The game was two-dimensional and tasked children with managing different home fire scenarios by making choices. An in-game character modelled correct practices. A narrator taught children about fire safety and provided reinforcement and feedback in relation to the children’s choices. The children played the game for approximately 45 minutes over a period of three weeks on average. The control children played a computer game about dog safety. Children’s fire safety knowledge and behavioural intentions were measured through child interviews and observation before and after the intervention, with a period of three weeks between assessment time points. Children’s fire safety knowledge and behavioural intentions significantly improved relative to controls. A marginally significant positive association was found in the intervention group between play time and the extent of improvement.

In Reim’s (2011) storybook study, 52 dyads of mothers and their children, aged 3 – 5 years, were recruited through schools, day-care centres and a university database. They were randomly allocated to an intervention condition (n = 27) or an active control condition (n = 25). In the intervention condition, children were exposed to a storybook which explored hazards, risks (including burns) and safe practices at home using concise text and photographs of preschool children. Mothers were asked to read the storybook with their children in the family home at least 5 times per week over 2 weeks. The storybook required children to make decisions as to whether particular behaviours in the home were safe or unsafe and to affix stickers to the storybook reflecting these decisions. In response, mothers were asked to discuss the risks with their children and encourage them to suggest safe practices. In the control condition, children were exposed to a storybook about healthy eating. Before and after the intervention, with a period of two weeks between assessment time points: children’s knowledge of home hazards was assessed through child interviews; and children’s behaviours in the form of interactions with hazards were assessed through observation. It was found that children’s knowledge significantly improved in the intervention condition relative to controls, and that more reading time was significantly associated with greater knowledge improvements in the intervention group. Additionally, risk behaviours increased marginally significantly in the control group, but there was no significant change in the intervention group.

 

Group interventions

Four RCT studies evaluated group format interventions (Jones and Kazdin, 1980; McConnell, Leeming and Dwyer, 1996; Mori and Peterson, 1986; Morrongiello et al., 2016). In Jones and Kazdin’s (1980) study, 60 children aged 3 – 6 years, who were enrolled in preschool programmes which were selected based on interest, received a teacher-delivered intervention at school which aimed to teach them how to contact the emergency services. Two sets of three classrooms were randomly assigned to one of three conditions: behavioural training; teacher-devised training; and a no-training condition in which children did not receive the intervention. In the behavioural training condition, teachers used simulated telephone apparatus and training instructions to provide children with training in emergency dialling characterised by modelling, practice, feedback, prompts, rehearsal and reinforcement. Training was conducted in a group format, but the children practised individually. In the teacher-devised training condition, teachers were provided with equipment and asked to teach dialling skills but were not given training instructions. Behavioural skills amongst the children were measured through child interviews which were conducted before and after the intervention, with a period of approximately four weeks between assessment time points. An analysis of covariance revealed that children’s emergency dialling skills improved significantly more in the behavioural training condition compared with the teacher-devised training and no-training conditions, and that there was no significance difference between the teacher-devised training and no-training conditions in improvement. However, it was also found through repeated measures t-tests that emergency dialling skills significantly improved in the behavioural training group and teacher-devised training group, but not in the no-training group.

In another study, 12 childcare centres were selected to participate in the research based on factors including representation of population diversity and willingness to engage. They were matched as closely as possible on children’s age, location and centre classification, were randomly assigned to an immediate treatment condition or a delayed treatment, control condition (McConnell, Leeming and Dwyer, 1996). In total, 443 children aged 3 – 5 years participated in the immediate treatment (n = 222) or delayed treatment (n = 221) condition. The intervention was a programme delivered by teachers to children in childcare centres. The programme focused on fire prevention, how to respond to burns and fire and the role of a firefighter. It involved education which focused on the cognitive aspects of fire safety, and training characterised by instruction, modelling, roleplay, and rehearsal in simulated emergencies. Teachers received necessary resources and received training via a seminar. In addition, the programme involved the participation of firefighters and promoted the involvement of parents in home-based fire safety activities such as establishing an escape plan. The programme lasted 30 hours overall, and was delivered in daily sessions lasting 20 minutes over a period of 18 weeks. Children’s fire safety knowledge was assessed through child interviews which were carried out before and after the intervention. It was found that  knowledge in the intervention condition significantly improved compared with controls.

Moreover, in Morrongiello et al.’s (2016) study, three schools, which were the first three schools to volunteer to participate based on a general information letter to schools, were recruited. They were randomly assigned to an intervention condition (six classrooms) or a control condition (three classrooms). In total 135 children aged 4 – 6 years participated in the intervention (n = 93) or control (n = 42) condition. The intervention was a classroom-based programme which researchers delivered to the children. The programme covered a range of home safety topics including burns and general home safety. The programme focused on the causal relationships between unsafe behaviours and injuries, grounded on specific experiences (an inductive reasoning approach). In addition, it presented a range of situational examples of hazards associated with particular injury types, encouraged children to apply their knowledge, and used cartoon imagery in the conveyance of information. It also sought to promote perceptions of vulnerability, harm severity and self-efficacy. The programme involved activities including storybook reading, singing, crafts and games. Furthermore, take-home activities were provided and parental involvement in these was encouraged. The programme comprised six sessions, each lasting 40 minutes. Fire safety knowledge amongst the children was measured through child interviews before and after the intervention. It was found that the intervention children’s knowledge of home safety hazards significantly improved relative to the controls.

In another teacher-delivered intervention, 30 children aged 3 – 5 years were recruited from a university lab school. They were exposed to a school-based programme that taught safety response rules in a number of domains including fire (Mori and Peterson, 1986). The rules were displayed on cards, and the trainer played games with the children in small groups using the cards to impart the rules. The trainer provided reinforcement including praise and rewards such as stickers. The cards were progressively disused, and correspondingly children were asked to behaviourally demonstrate and rehearse, and verbally justify, the appropriate responses to different scenarios without visual cues. The children were randomly allocated to a condition in which the trainer had been provided with a training manual as well as professional supervision, or a training manual only condition. Children’s safety behavioural intentions and behavioural skills were assessed through observation before and two weeks after the intervention. It was found that the children’s behavioural intentions and behavioural skills significantly improved, and there was no significant difference in improvement between trainer preparation conditions.

 

Uncontrolled before-and-after study

One study, evaluating a group intervention, adopted an uncontrolled, pre-post assessment design (Jones, 2006). The intervention was a sociocultural, classroom-based programme which was delivered by teachers in two classrooms to 14 preschool children aged 3 – 6 years with a range of disabilities, including neurodevelopmental disorders and sensory impairments. The aim of the programme was to teach the children to stay away from hot things that can hurt and recognise the firefighter as a helper. Classes were supplied with intervention resources including toys, books and posters related to fire safety and instructional materials. The programme schedule was two and a half hours per day from Monday to Friday for two weeks. In the programme, children visited appropriate educational websites around fire safety and engaged in activities such as games and dramatic play, and a firefighter visited the school. Staff supported the children’s learning experience throughout, using a scaffolding approach (characterised by the measured withdrawal of support as the child’s competence increases) and promoting reflective dialogue (conversation amongst the children about acquired knowledge). In one of the two participating classrooms, where comparatively the teacher was observed to display a more flexible teaching style and the children’s access to the intervention resources appeared to be greater, the children engaged in significantly more reflective dialogue (assessed via observation). Children’s fire safety knowledge was measured through child interviews conducted one week before, and towards the end of, the intervention. It was found that the children’s knowledge of the relevant fire safety concepts significantly increased.

 

Controlled studies without a pre-test

Individual intervention

One paper presented an evaluation of an individual intervention which was controlled but lacked pre-assessment (Borzekowski et al., 2014). In this study, 320 dyads of parents and their 4 – 5 year old children were recruited via a recruitment company and reflected the ethnic and educational composition of the city population. The participants were shown, in a one-off laboratory session, two cartoon videos that displayed a burn injury and a house fire and the outcomes of behavioural responses to these incidents. The participants were randomly assigned to experimental conditions. Depending on condition assignment, participants saw either positive outcomes of correct behavioural responses (gain-framing) or negative outcomes of incorrect behavioural responses (loss-framing). Furthermore, after the video viewing, dyads interacted freely (no-mediation), parents were told to talk with their child about what they had seen (unscripted mediation), or parents were given a talking points sheet and asked to use this to discuss injury prevention points encapsulated in the videos with their child (scripted mediation). Children’s fire safety knowledge and attitudes were assessed via child interviews immediately after the intervention. Knowledge comprised recall, and understanding of the safety messages, of the videos. Attitudes related to self-efficacy regarding positive safety behaviours and perceptions of the safety behaviours of other children (social norm perceptions). It was found that only approximately half (55%) of the children were able to recall the videos without prompting immediately after exposure, and less than half understood the safety messages of the videos. Nonetheless, some (though not all) analyses found that gain-framing was significantly associated with greater safety understanding and self-efficacy. Furthermore, in certain contexts, video recall and understanding, self-efficacy, and social norm perceptions were significantly superior in the parental mediation conditions compared with the no-mediation condition. The findings of salutary outcomes connected to parental mediation were most robust where the mediation was scripted.

 

Group interventions

Two controlled studies that lacked pre-testing evaluated group interventions (Cole, Crandall and Kourofsky, 2004; Folkman and Taylor, 1972). The programme evaluated by Cole, Crandall and Kourofsky (2004) sought to impart behavioural skills to promote safety in the event of a fire (“go to the firefighter”; “crawl low under smoke”; and “stop, drop and roll”) and prevent fire play (“go tell a grown up”). The participants were 109 children aged 3 – 5 years from 13 preschool classrooms. The classrooms were randomly drawn from six early childhood programmes. Where participating teachers had two classes, one was randomly selected to receive the programme. Eight classes received the programme and five participated as controls. The programme was delivered by teachers to the children using a programme kit. One lesson was taught per week over the course of a month. The programme involved the use of video and multiple activities, including behavioural rehearsal of safety skills. Fire safety knowledge, behavioural intentions and behavioural skills were assessed amongst the children by interviewing and observing them two weeks after the intervention. Post-intervention, the children in the intervention condition had significantly superior fire safety knowledge, behavioural intentions and behavioural skills compared with controls.

In a study by Folkman and Taylor (1972), 88 classes, which comprised 1,500 children who were enrolled in the Riverside County Headstart Project and tended to be aged 4 – 5 years (though some were as young as 3 years and others were over 6 years), were matched as closely as possible socio-demographically and randomly allocated to an intervention condition or control condition. In the intervention condition, teachers delivered a fire safety programme, using the Headstart Fire Prevention kit, to the children. The programme was delivered over the course of the school year, and aimed to satiate natural curiosity around fire, improve understanding of causal links between the child’s actions, fire production and associated consequences, and promote positive attitudes regarding fire safety behaviours amongst the children. The key mode of delivery was aural-visual presentation, using coloured illustrations. Additionally, the programme involved the use of colouring sheets, posters and adhesive badges, and leaflets regarding the programme were sent to parents. Fire safety knowledge and attitudes amongst the children were measured through post-intervention child interviews towards the end of the school year. It was found that children in the intervention condition displayed significantly superior understanding of only some fire-related causal relationships relative to controls after the intervention. No significant differences in fire-related attitudes were found.

 

Single-subject design studies

Five studies evaluated individual interventions using a single-subject research design (Jones and Kazdin, 1980; Morgan, 2017; Padgett, Strickland and Coles, 2006; Rossi et al., 2017; Vanselow and Hanley, 2014). Jones and Kazdin (1980) reported on an intervention in which six preschoolers aged 3 – 4 years, who were enrolled in preschool programmes which were selected based on interest, were taught when to call the emergency services. They were shown thirty pictures representing emergency and non-emergency scenes (including fire-related, injury-related and neutral scenes) and were asked who, if anyone, should be called in relation to each. The researcher provided the children with feedback and reinforcement and modelled correct responses as appropriate. Children’s knowledge was assessed via child interviews over the course of the intervention. The training process was found to be associated with an enhanced understanding amongst the children as to when to make an emergency call. In another study, carried out by Morgan (2017), three children with Autistic Spectrum Disorder (ASD) who were enrolled at a particular preschool and were aged between 3 and 5 years were shown, twice per day for four weeks at school, a two-minute video on an iPad of a child modelling the correct safety behaviours in response to a fire alarm. Children were given non-verbal reinforcement, such as pats on the back, whilst they learnt the fire safety skills. The children’s behavioural skills in responding to a researcher-induced fire alarm sound were assessed via observation before, during and two weeks after the intervention, and the children’s behavioural skills in the context of a school-wide fire drill were assessed via observation two weeks after the intervention. It was found that the children’s behavioural skills improved during the intervention, these improvements were maintained two weeks later and the children were able to generalise the skills to a school-wide fire drill.

In addition, in one intervention, the participants were three children with ASD aged between 5 and 6 years who attended a behavioural programme. They received behavioural skills training (BST) at school, which involved exposure to a range of ostensibly dangerous physical stimuli (e.g., a lighter; Rossi et al., 2017). BST involves instruction, modelling, rehearsal and feedback. For each stimulus, the child was exposed to the stimulus and it was explained that it was not safe to handle. The child was instructed not touch it, to move away from it and to inform an adult of its presence, and was asked to verbally repeat these instructions to the researcher (instruction). Next, the trainer modelled the correct safety behaviour in relation to the stimulus (modelling). The child then practised the safety behaviour (rehearsal). Finally, the trainer provided reinforcement and feedback as appropriate (feedback). These steps were repeated until the child was able to demonstrate the desired safety behaviours. Training session length ranged from 6 to 11 minutes, and total training time was between 109 and 244 minutes. Children’s behavioural responses to ostensibly dangerous physical stimuli in the school and home contexts were assessed via observation before, immediately after, two weeks after and four weeks after the intervention. Improvements in the children’s behavioural responses were observed, such that all three children were able to respond safely to the stimuli at school and in the generalised context of the home after the intervention. These safety responses were maintained two to four weeks after training.

Furthermore, in Padgett, Strickland and Coles’ (2006) study, five children with foetal alcohol syndrome or partial foetal alcohol syndrome, aged 5 – 7 years, were recruited from a foetal alcohol clinic. They independently played a three-dimensional computer game featuring a house. At one stage, the house was on fire. The player was required to respond to the situation. Children received support with navigating the world from an animated character and/or arrows in the game. The level of support given was reduced with increasing competence. Furthermore, the children received reinforcement and feedback within the game in respect to their actions. Children’s fire safety knowledge was assessed via child interviews before, immediately after and one week after the intervention. In addition, children’s fire safety behavioural skills were measured through observation in the context of a real-world simulation immediately after, and one week after, the intervention. It was found that, whereas none of the children demonstrated fire safety knowledge at baseline, four children demonstrated optimal fire safety knowledge and optimal behavioural skills immediately after the intervention. At the one week follow-up, whilst only three children demonstrated optimal fire safety knowledge, all five children showed optimal behavioural skills.

In a further single-subject study, Vanselow and Hanley (2014) reported on an intervention in a school setting in which 16 children aged between 5 and 6 years, who had been recruited on a convenience basis, played a computer game for approximately 20 minutes. The game provided virtual BST aimed at promoting safety behaviours around different hazards, including lighters. The game included cartoon illustrations, and a character who provided instructions, reinforcement and feedback. Children were tasked with putting safety responses in sequential order, categorising objects as safe or unsafe and responding to dangers. Textual stimuli were used but these were accompanied by pictures and vocal instruction. If the child did not achieve competence in the desired safety behaviours after exposure to the computer game only, real-world BST was provided using ostensibly dangerous physical stimuli, including lighters (in-situ training). Children’s behavioural responses to ostensibly dangerous physical stimuli were assessed in the school setting via observation before the intervention, immediately after BST and immediately after in-situ training (if this was required). It was found that children’s safety responses improved in all participants over the course of the intervention. Whilst some participants demonstrated competence in safety responses after playing the computer game only, in other participants competence was only observed after in-situ training.

 

Uncontrolled post-assessment studies

Two studies conducted post-assessments of group interventions which employed qualitative techniques and did not use a control group (Dickinson et al., 2014; Lidstone, 2006). The intervention evaluated by Dickinson et al. (2014) was a programme delivered within early childhood centres by teachers to preschool children aged 2 – 5 years. It aimed to teach children about the nature of fire, the role of a firefighter, fire prevention and how to respond to a fire in the home. The programme involved use of visual (e.g., a storybook) and audio resources, a visit from the fire service, experiential activities (e.g., singing, exploration of a fire engine), and provision of take-home materials for parents which promoted home-based activities to reinforce learning. Children’s fire safety knowledge, attitudes, and behavioural skills were assessed through post-intervention qualitative interviews with teachers and the fire service representative who had visited the centres. In addition, fire safety-related behaviours amongst the children were assessed through the qualitative interviews and by a post-intervention survey completed by 23 parents. It was found through the qualitative interviews that the children had gained fire safety knowledge and behavioural skills, were applying the learning they had acquired in their play and viewed the firefighter as helpful and safe after the visit. In addition, the parent survey revealed that most parents agreed that their children had talked about fire issues spontaneously and had encouraged relatives to address fire safety issues.

Finally, one study evaluated an educative school assembly intervention (Lidstone, 2006). A single performance, using puppetry, was delivered by external producers in a school assembly format to children aged 4 – 8 years within multiple schools. The performance lasted 45 minutes and sought to impart fire safety messages including how to reduce fire risk and how to respond to a fire. The performance was followed by an audience question-and-answer segment in which the children were asked recap questions relating to the key messages. Children’s fire safety knowledge was explored qualitatively through classroom discussions which were conducted immediately after, and four to six weeks after, the intervention. It was found that, whilst the children tended to be able to recite the key safety slogans (e.g., “stop, drop and roll”) with high accuracy immediately after the performance and a month later, knowledge of the meanings of and rationales for these messages, and the circumstances in which they should be applied, was relatively low.