To test the efficacy of installing safety devices in the homes of young children on total injury rates and on injuries deemed a priori modifiable by the installation of these devices.
A nested, prospective, randomized controlled trial.
Indoor environment of housing units.
Mothers and their children from birth to 3 years old participating in the Home Observation and Measures of the Environment study. Among 8878 prenatal patients, 1263 (14.2%) were eligible, 413 (32.7%) agreed to participate, and 355 were randomly assigned to the intervention (n = 181) or control (n = 174) groups.
Installation of multiple passive measures (eg, stair gates, cabinet locks, and smoke detectors) to reduce exposure to injury hazards. Injury hazards were assessed at home visits by teams of trained research assistants using a validated survey.
Main Outcome Measure
Modifiable and medically attended injury (ie, telephone calls, office visits, and emergency visits for injury).
The mean age of children at intervention was 6.3 months. Injury hazards were reduced in the intervention homes but not in the control homes at 1 and 2 years (P < .004). There was no difference in the rate for all medically attended injuries in intervention children compared with controls: 14.3 injuries (95% confidence interval [CI], 9.7-21.1 injuries) vs 20.8 injuries (95% CI, 14.4-29.9 injuries) per 100 child-years (P = .17); but there was a significant reduction in the rate of modifiable medically attended injuries in intervention children compared with controls: 2.3 injuries (95% CI, 1.0-5.5 injuries) vs 7.7 injuries (95% CI, 4.2-14.2 injuries) per 100 child-years (P = .03).
An intervention to reduce exposure to hazards in homes led to a 70% reduction in the rate of modifiable medically attended injury.
clinicaltrials.gov Identifier: NCT00129324
Despite a 25% reduction in housing-related injuries during the past 2 decades, the home environment remains the leading location of injury for young children, accounting for more than 13 million outpatient visits, more than 4 million emergency visits, 74 000 hospitalizations, and 2800 deaths each year in the United States.1 Risk factors that have been identified for housing-related injuries include physical hazards, poor quality housing, externalizing child behaviors in boys, low socioeconomic status, and persistent maternal depressive symptoms.2- 6
Multiple controlled trials to reduce housing-related injuries have been undertaken. They mostly have relied on anticipatory guidance and only occasionally have provided safety devices.7- 10 The largest trial to date, which was conducted in the United Kingdom and randomized more than 3400 families with children younger than 5 years, found a significantly higher rate of primary care visits for injury in intervention group children. None of the reported trials have undertaken comprehensive installation of safety products in the housing units for all participants. Therefore, it remains unclear whether such safety devices, if successfully installed and functioning, lead to a reduction in exposures to hazards and subsequent medically attended injuries in children.
The aim of this trial was to test the efficacy of installing home safety devices on a reduction in hazards and housing-related injuries in children. Specifically, we hypothesized that the installation of multiple passive safety measures in the homes of children randomized to the intervention group would reduce exposure to injury hazards and medically attended injury across unintentional residential mechanisms by 30% compared with children in control homes, including hazards and mechanisms whose exposure was modified by safety devices. The study was approved by the institutional review board at the Cincinnati Children's Hospital Medical Center.
We screened expectant mothers who attended 7 participating obstetrical practices for eligibility in the Home Observation and Measures of the Environment (HOME) Study, a prospective, randomized controlled trial of home injury and lead hazard control nested within a birth cohort examining the developmental effects of exposure to prevalent environmental neurotoxicants. Mothers had to be at least 18 years of age and no more than 19 weeks' gestation, live in a home built before 1978, have no plans to relocate in the next 12 months, not live in public housing or a shelter, speak English, and present to one of the participating prenatal clinics. Methods for enrollment and home surveys have been described previously.11 A biostatistician generated a random list of assignment codes in permuted blocks of 10 stratified by urban, suburban, and rural areas surrounding the Cincinnati, Ohio, regional area. The randomly generated assignments were placed in sealed, radiopaque envelopes and kept by a data manager. Randomization occurred after the baseline home visit and landlord consent for rental units. Families enrolled in the study whose landlord refused the intervention were not randomized and not included in this intention-to-treat analysis (n = 53; Figure 1). The biostatistician who performed all intention-to-treat analyses was masked to group status until these analyses were undertaken.
The Home Observation and Measures of the Environment (HOME) injury study participant enrollment flow diagram. OB indicates obstetrics.
Before randomization, all households underwent a baseline home survey of 5 predefined high-exposure and high-risk areas for injury hazards (kitchen, main activity room, stairways, child's bedroom, and child’s bathroom) by trained research staff. The number and type of injury-related hazards were quantified using a validated instrument.11 The instrument was developed on the basis of an analysis of the leading mechanisms of injury resulting in an emergency visit for US children (falls, cut/pierce, struck/strike, poison, and burn) by means of the National Hospital Ambulatory Medical Care Survey and a review of instruments used in similar studies.1,8,12 For the purposes of this study, we defined injury hazards to include tap water temperature exceeding 49°C; absent or nonfunctioning smoke alarms or carbon monoxide detectors; accessible and unlocked cabinets and drawers; unstable furniture or television stands; poorly maintained or ungated and accessible stairways; unsecured area carpets or rugs; accessible stove tops and ovens; easily accessible medications, cleaners, detergents, poisons, or needles or other sharp instruments; accessible windows (inside ledge <99 cm from floor or bed and outside ledge >1.2 m above ground); uncovered electrical sockets; lack of poison control or clinic telephone numbers; and unsafely stored firearms (no trigger lock or lock box for storage and/or ammunition not kept separate from firearm).
The density (number of hazards per 9.3 m2) of injury hazards was determined by dividing the number of hazards by the area of the room. Stairways were not included in these calculations. The 5-area survey, which was validated for its replicability, reproducibility, and representativeness of the entire living space,11 was repeated when the children were 12 and 24 months of age.
Interventions were directed primarily at areas less than 99 cm in height (the 75th percentile in height for a 3-year-old boy). Research staff, who had undergone extensive training, identified hazards at an intervention planning visit and discussed the study interventions and safety products with the families. Families had the option of rejecting any interventions recommended. Next, the research assistants installed all consumer product safety devices agreed upon by the families. If participants moved before their child turned 30 months old, their new homes were assessed for injury hazards and safety devices were installed. Participants in both arms of the trial were given the American Academy of Pediatrics' “The Injury Prevention Program” information sheets on developmentally appropriate injury risks and control measures.
Exposure to the indoor environment and injury outcomes were assessed by parental report using telephone and annual home visit interviews. Telephone interviews were conducted quarterly in the first year of the study and twice a year thereafter. The interview included initial questions about the child's time of exposure to the indoor environment, followed by questions on maternal supervisory behavior and any injuries the child experienced in the prior 3 months, and required approximately 15 minutes to complete. It was not possible to mask participants to treatment status, but research staff who conducted the interviews were trained separately and operated separately from intervention technicians to minimize the chances of interviewers becoming unmasked to treatment status. Participants were instructed not to reveal their group status to research interviewers whether they knew or suspected it. Investigators and analysts were masked to group status until data analyses were conducted.
The primary outcome for the HOME Study was medically attended injury events, especially those modifiable by the study safety devices. Parents were given a calendar with a box of small adhesive bandages attached to record events as they occurred at home and asked to report injury events (of known onset) that resulted in pain, anxiety, or a visible tissue mark lasting more than an hour. A medically attended injury was defined as an injury that prompted the parents to call or visit a physician's office, urgent care, or emergency department. We also classified injury events as modifiable before the start of the trial according to whether exposure to the hazard and mechanism causing the resulting injury could have been modified by one of the installed interventions (eg, wall-mounted stair gate). Modifiable mechanisms were stairway falls, struck/strike, cut/pierce, suffocation/submersion, poisoning, and burns. Reports of emergency visits for residential injury were confirmed through matching with an area-wide surveillance system (the Hamilton County Injury Surveillance System) for injury visits to emergency departments.
We estimated a sample size of 400 households to measure a 30% reduction in medically attended injury with an α of .05 and a statistical power of 80%.
This analysis was by intention to treat with a priori hypotheses of a reduction in injury hazards and in modifiable and medically attended home injuries. Data were entered into a relational database and downloaded for analysis into SAS software, version 9.1 (SAS Institute Inc, Cary, North Carolina). Univariate analysis of the demographic characteristics for mothers, their children, and homes was conducted. The mean number and density of hazards were compared by group assignment using t tests with adjustment for multiple comparisons. Numbers of all reported events (medically attended and emergency visits) for both modifiable and total injuries were based on maternal report at the telephone survey. To account for clustering of injuries within an individual child and correlation of measurements over time, generalized estimating equations were used to model injury outcomes by means of a Poisson link function. The multivariable models were developed using the injury outcome as the dependent variable and the group assignment and time × group assignment as independent variables.13 To develop rate estimates, the overall 3-month estimate from the generalized estimating equation model was converted to a rate per 100 child-years.
We screened expectant mothers presenting to obstetrical practices for eligibility between February 13, 2003, and January 12, 2006. Of the 8878 women screened, 7287 (82.1%) were ineligible and another 328 (3.7%) did not respond to the invitation letter and could not be contacted (Figure 1).
A total of 1263 women (14.2%) were eligible for participation. Of these women, 795 declined to participate and 55 dropped out during the run-in period for the trial before randomization. After the baseline home visit but before randomization, an additional 53 women became ineligible for randomization because the landlord refused to participate, and 5 dropped out. This left 355 expectant mothers to be randomized. Before delivery of their infants, 181 women were randomized to the intervention group and 174 to the control group. Each arm of the dual-arm, parallel group study acted as the other's control. The first birth in the cohort was on July 16, 2003, and the final birth was on July 18, 2006. Of the 181 women in the intervention group, 167 delivered live-born singletons and had at least 1 follow-up questionnaire completed during the 24-month follow-up. Of the 174 women in the control group, 159 delivered live-born singletons, completed at least 1 questionnaire, and were not lost to follow-up. Therefore, the intention-to-treat analyses included the 167 participants in the intervention group and the 159 participants in the control group. Of the 167 participants, 149 (89.2%) had an intervention completed during the 24-month period of study. Satisfaction reporting forms were provided to each participant. Of the 149 participants with completed interventions in the intervention arm, 104 (69.8%) had returned completed satisfaction surveys; 103 (99.0%) indicated good or high satisfaction with their experience in the trial.
The demographic characteristics of mothers and their children are shown by group assignment (Table 1). The mean age of the mothers at study entry was 30 years, and the mean age of the children at the time of the intervention was 6.3 months. The median income for the cohort was $70 000 annually ; approximately one-quarter of families in both groups earned less than $30 000.
At the baseline visit, there was no difference in the mean number and density of hazards in the 4-room area (kitchen, main activity room, child's bedroom, and bathroom) by group assignment (Table 2). Longitudinal analysis of the number (P < .001) and density (P = .01) of injury hazards in homes during the 24-month follow-up period showed a significant group × time interaction. The mean number and density of hazards in the intervention group decreased by 10% across the 4 rooms from baseline to the 12-month follow-up visit (P = .001). The number and density of hazards were decreased by 15% at the 12-month survey in the intervention homes compared with the control homes (P < .005). Although the mean number and density of hazards were lower in the intervention homes at 24 months, only the mean number of hazards remained statistically significantly reduced from baseline (P < .02) and compared with controls (P = .001). Tests were adjusted for multiple comparisons. Selected safety practices and installed products over time are displayed in Table 3.
There were 134 total modifiable injury events in 92 children during the first 24 months of the study (1.4 events per child). Modifiable injuries accounted for 21 of 74 (28.4%) medically attended events in children (Table 4). The rate of modifiable, medically attended injuries in the home was reduced by 70%, at 2.3 injuries (95% confidence interval [CI], 1.0-5.5 injuries) per 100 child-years for children in intervention households vs 7.7 injuries (95% CI, 4.2-14.2 injuries) per 100 child-years in controls (Table 3; P = .03). The cumulative rates of injury by group assignment are displayed in Figure 2.
The Home Observation and Measures of the Environment (HOME) injury study intention-to-treat analysis of modifiable, medically attended injury rates per 100 child-years by group assignment. Error bars indicate mean (SD).
Rates for all medically attended injuries among children were decreased during the 24-month follow-up by 31%, at 14.3 injuries (95% CI, 9.7-21.1 injuries) per 100 child-years in the intervention group vs 20.8 injuries (95% CI, 14.4-29.9 injuries) per 100 child-years in the control group (P = .17; Table 4). There were 12 emergency visits for all home injury-related mechanisms in the intervention group compared with 10 for controls (P = .92); and the rates of emergency visits were almost identical in control and intervention groups at 5 per 100 child-years.
There were 2 deaths to child participants enrolled in the HOME Study and reported to the institutional review board at Cincinnati Children's Hospital Medical Center, both deemed unrelated to the participation in the trial. The first involved perinatal complications related to premature delivery. The second occurred under the supervision of the mother and involved a drowning of a 1-year-old sibling of a child born into the study. The case was referred to the Cincinnati Child Protective Services Division of the police department.
Children who lived in housing units that were randomly assigned to the intervention group had significantly fewer injury hazards and a 70% reduction in the rate of modifiable and medically attended injury from 7.7 to 2.3 injuries per 100 child-years in controls compared with intervention group children (Table 4). Of all medically attended injuries, 37.2% (16 of 43; Table 4) in the control group met our definition of “modifiable.” Thus, if confirmed, this trial suggests that large-scale implementation of this type of intervention could result in a 26% reduction (0.70 × 0.37) in all medically attended, housing-related injuries, estimated at 5 million medically attended visits for injury in US children younger than 5 years each year.1,14- 16 To our knowledge, this is the first trial of home safety that installed and maintained safety devices in all intervention homes.8- 10,17,18
A recent Cochrane Review summarized findings from previous trials of home safety, noting that “there was a lack of evidence that interventions reduced rates . . . [over] a range of injuries.”7 There are 2 conflicting trials that have reported housing-related injury outcomes in children. The first study, a multicenter trial by King et al,19 recruited and randomized 1172 participants, 951 of whom completed a follow-up home visit at 1 year, at 5 Canadian pediatric emergency departments and enrolled children younger than 8 years with specific injuries, including home-related events (eg, tap water scalds) in addition to bicycle crashes and other injuries outside the home. Intervention homes underwent a home survey for hazards, and the parents were educated about hazard exposure amelioration and were given coupons to purchase safety devices (including car seats and bicycle helmets). This trial found that the intervention was effective in reducing the overall self-reported occurrence of injury visits (inside and outside the home) at 4, 12, and 36 months of follow-up.20
In a larger controlled trial in the United Kingdom, Watson et al8 randomized more than 3400 families in 47 general practices in Nottingham to receive a standardized safety consultation and provision of free and fitted stair gates, fire guards, smoke alarms, cupboard locks, and window locks for low-income families and of reduced-cost equipment for families with relatively higher income. Control families received usual care. A total of 1163 families (68%) in the intervention arm received safety counseling, 619 (36%) had free equipment fitted, and 26 (1.5%) bought equipment at low cost. Paradoxically, the attendance rate for a medically attended injury visit was 37% higher for children in the intervention arm compared with those in the control arm (P = .003).
There are important differences in the study design and interventions of these trials that make it difficult to compare directly to this study. The multicenter study conducted by King et al19 included children up to 8 years of age and interventions outside the indoor environment (eg, bicycle helmets and automobile seat restraints for children) and did not install the safety products. In the UK study8 that showed an increase in injury rates in the intervention group, only approximately one-third of families randomized to the intervention arm actually had safety products installed, and differences in safety practices from baseline to 12- and 24-month follow-ups for all intervention and control families were small (<10% for most practices). Also, this study measured all medically attended injuries in children younger than 5 years (as opposed to hazards and mechanisms directly related to the products provided or “fitted”), possibly diluting the effect of the intervention on modifiable injuries among younger children. Finally, this UK population came from relatively lower socioeconomic environments compared with the more affluent, educated Cincinnati population.
There were several limitations of the current study. First, it is not possible to conduct a double-blind trial for this type of study. Nevertheless, although the participants were not masked to group assignment, research interviewers who conducted surveys assessing medically attended injuries were masked to group assignment. Furthermore, intervention technicians performing installation and maintenance visits were maintained as separate, functioning teams throughout the study. Second, we relied on maternal report of injuries. However, we verified parental report using a county-wide surveillance system for emergency visits. Third, although mothers participating in this study were representative in age and racial background of those who gave birth in the 5-county regions from which they were enrolled, they may not be representative of US and other households.
The installation of multifaceted home safety devices led to a significant reduction in injury hazards and a 70% reduction in medically attended and modifiable injury among children in the first 2 years after birth. Healthcare expenditures for injury in US children amount to more than $2.3 billion annually, and emergency visits for children due to injury cost on average approximately $800 per visit.16,21 Because US children younger than 5 years account for more than 1.7 million emergency visits and 5 million ambulatory visits annually for injury in the home environment,1,16,22 this intervention, if replicated in larger populations of mothers and their children, could reduce pain and suffering and save millions of dollars in health care costs.
Correspondence: Kieran J. Phelan, MD, MS, The James M. Anderson Center for Health Systems Excellence, S Bldg, Room 2.347, MLC 7014, Cincinnati Children's Hospital Medical Center, 3333 Burnet Ave, Cincinnati, OH 45229-3039 (email@example.com).
Accepted for Publication: October 11, 2010.
Author Contributions: Drs Phelan, Khoury, and Lanphear and Ms Liddy had full access to all the data in the study and take responsibility for the integrity of the data and accuracy of the data analysis. Study concept and design: Phelan and Lanphear. Acquisition of data: Phelan, Liddy, and Lanphear. Analysis and interpretation of data: Phelan, Khoury, Xu, Hornung, and Lanphear. Drafting of the manuscript: Phelan. Critical revision of the manuscript for important intellectual content: Phelan, Khoury, Xu, Liddy, Hornung, and Lanphear. Statistical analysis: Phelan, Khoury, Xu, and Hornung. Obtained funding: Phelan and Lanphear. Administrative, technical, and material support: Phelan, Liddy, and Lanphear. Study supervision: Phelan and Lanphear.
Financial Disclosure: None reported.
Funding/Support: This study was supported by a Career Development Award (1K23HD045770-01A2) from the National Institute of Child Health and Development and a New Investigator Award (R49CCR523141-01) from the National Center for Injury Prevention and Control at the Centers for Disease Control and Prevention (Dr Phelan); and by grant PO1-ES11261 from the US Environmental Protection Agency and grant R01ES014575 from the National Institutes for Environmental Health Sciences (Dr Lanphear).
Role of the Sponsors: Funding organizations reviewed the initial designs and protocols on a blinded, competitive peer-reviewed basis to determine funding on the basis of design and scientific merit but had no role in data collection, management, analysis, interpretation, or write-up of the results.
Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the Eunice Kennedy Shriver National Institute of Child Health & Human Development, the US Centers for Disease Control and Prevention, or the National Institutes of Health.
Previous Presentations: This study was presented at the Pediatric Academic Society Meeting; May 3, 2010; Vancouver, Canada; and at the World Safety Conference; September 21-24, 2010; London, England.
Phelan KJ, Khoury J, Xu Y, Liddy S, Hornung R, Lanphear BP. A Randomized Controlled Trial of Home Injury Hazard ReductionThe HOME Injury Study. Arch Pediatr Adolesc Med. 2011;165(4):339-345. doi:10.1001/archpediatrics.2011.29