Each year, an unknown number of young children die in the United States of heat stroke from being enclosed in parked motor vehicles. Physicians in the United States, Great Britain, and New Zealand warned about the risks of leaving infants and toddlers in hot cars in 1976 and in the early 1980s.^1–5 Medline contains only a few examples of more recent literature on this topic or on the vulnerability of young children to heat.^6–9 A 2002 report of injuries to unattended children in or around vehicles listed heat related causes as the most common type of fatality.¹⁰ A National Highway Traffic Safety Administration (NHTSA) study of death certificates of certain non-traffic and non-crash events found 24 heat related deaths in 1998, 23 to children under age 5.¹¹

There are no unique codes to identify motor vehicle heat related deaths in the International Classification of Diseases (ICD) or in any United States national or state data source. The WONDER and WISQARS (Web-based Injury Statistics Query and Reporting System) databases of the Centers for Disease Control and Prevention can be searched by relevant ICD-9 E codes (E900, excessive heat and its subcodes; E904.0, abandonment, and E904.3, exposure) but do not specify whether the decedents were in a vehicle. Fatality Analysis Reporting System (FARS) data from NHTSA are limited to deaths that occur as a result of a traffic crash.¹² We sought to describe the circumstances of heat related fatalities to children in the passenger compartments of parked vehicles by using news accounts as our primary source of information.

METHOD

Previous investigators have described the value of using news sources for injury surveillance.^13–16 Ideally, news accounts supplement other data, but with no national datasets for this cause of death, news became our primary source. A professional librarian searched for incidents occurring between 1 January 1995 and 31 December 2002, using newspaper indexes, news websites, and internet search engines.¹⁷ Keywords related to children, heat, motor vehicles, abandonment, enclosure, death, and injuries were used. Cases were defined as children aged ⩽5 years who died from heat related causes while confined in the passenger compartment of a parked motor vehicle in the United States.

A modified “snowball technique” tried to identify additional cases and to gather more detail. News accounts often referred to prior incidents, which we then sought. On several occasions, reporters contacted us and we identified new incidents or more details as a result. Queries seeking cases were sent to email discussion lists focusing on injury prevention, child death review, and child abuse.¹⁸ All these sources were supplemental; standard news accounts were required for each case to meet our criteria.

For each identified case, we created a unique record in a Filemaker Pro database. Multiple local news sources were searched for each case to verify details, reconcile inconsistencies reported in earlier accounts and track developments. Table 1 shows the study variables grouped by victim, vehicle, and other factors.

Cases were grouped and analyzed by incident: one or more children dying in a vehicle as a result of heat. Incidents were assigned mutually exclusive categories relating to the circumstances in which the victims were in the vehicles: playing or left unattended. We also looked at whether the adults knew when they exited the vehicle that they were leaving a child behind.

RESULTS

As of 31 December 2002, the database contained information on 233 heat related deaths to children in parked motor vehicles. Although this study focuses on the United States, deaths were also identified in Australia (6), Japan (4), England (2), Israel (2), Italy (1), and Malaysia (1).

A total of 171 deaths comprising 159 incidents met the case criteria (box 1).

In 10 incidents, two children died in each vehicle, and in one incident, three died. Forty six children (27%) died while playing and 125 children (73%) died when left unattended.

The children’s ages ranged from 10 days to 4 years; almost two thirds (64%) were male. The mode for girls was <6 months old, while for boys it was 2–2.5 years (table 2).

We grouped the circumstances by children who gained access while playing (27%) and by those left by adults (73%), which further divided into forgotten, left intentionally, and unclear intentions (fig 1).

High temperatures ranged from 63 to 115° F. Three quarters of incidents (124 of 159) were during the summer months of June, July, and August. There were no patterns to the number of deaths each year, ranging from seven in 1996 to 30 in 1999. Incidents occurred in 41 of 50 states (fig 2).

Most reports estimated how much time elapsed from when the child was last seen or placed in the car to when the child was found. The estimates ranged from less than 15 minutes to 10 hours; one child’s body was not found for several days. Eighty two percent (140/171) of children were in the vehicle for ⩾1 hour, and 29% (50/171) were in the vehicle for ⩾5 hours. A larger percentage of girls (39%) than boys (24%) were enclosed for ⩾5 hours.

Circumstances: children playing

Twenty seven percent of deaths (46/171) were children who gained access to unlocked vehicles, most (35) in the family’s own vehicle at home. Seven incidents (11 victims) occurred in other vehicles in the neighborhood, mostly abandoned cars. Children who died while playing in vehicles tended to be boys (78%), between 2 and 3 years old (48%), and often were with other children who did not enter the vehicle. More than one third of children who died while playing were in the vehicle for less than one hour; three were thought to be in the vehicle for less than 15 minutes. In five incidents, there were multiple deaths: four pairs of children (including two sets of twins), and one vehicle containing three children.

In one fourth (10/41) of the playing incidents, the adults were napping or showering when the children entered the vehicle. Three accounts reported that families told investigators the child had previously entered the vehicle, or had just recently learned to open the car door. In four incidents, the children were sought for hours before the vehicle was searched.

Circumstances: left unattended

Seventy three percent of deaths (125) occurred when adults left children unattended in vehicles. Unattended children tended to be younger than children who died while playing: more than half (64) were ⩽1 year. Three fifths (74) of unattended children were boys. We identified 137 adults responsible for leaving children (in six cases, two adults were involved): 83 women and 52 men. In two cases, the adults’ genders are unknown.

Just over half (68/125) of unattended deaths occurred when adults forgot or were unaware of the child. Thirteen of these involved a change in routine—a large family gathering (often around the Fourth of July), move, or other upheaval.

Twenty seven percent (34) were deliberately left in vehicles by adults who were aware that they were doing so. Some adults were reluctant to disturb a sleeping child and failed to realize how hot the vehicle would become. In other incidents, the car—and the child safety seat—were used to restrain the child so the adults could sleep, work, drink or use drugs, or gamble. Twelve children were left for ⩾5 hours, seven for ⩾8 hours.

In 12 incidents involving 15 children, the responsible adult had a documented history of problems with alcohol or other drugs or mental illness. Three children were left behind by adults reportedly too intoxicated to notice the child.

Almost half (56) of unattended deaths were associated with child care. Twenty four children (19%) were left by child care providers, drivers, or babysitters, most of whom were formal or licensed providers. Thirteen of these victims were left in child care vans or school buses. Another 32 children were left by adults who meant to transport them to child care but went to work instead. Usually, the children were not found until the parent returned to the vehicle at the end of the day, or the other parent went to retrieve the child from child care.

The vehicles

Sport utility vehicles, vans, and minivans accounted for 33% (57/171) of the vehicles. At least 16 vehicles had tinted windows which may have prevented the victims from being seen. In some instances, electric door locks and heavy sliding doors kept children from exiting, even when they had been able to enter the vehicles. Nine vehicles were inoperable or abandoned. Although news accounts did not consistently mention child restraints, 45% (77/171) of the victims were noted as being found in child safety seats.

DISCUSSION

Previous reports have identified heat risks to children in cars, but have primarily described single cases. Our analysis found that heat related deaths to young children in motor vehicles occur each year, primarily during warmer months, and throughout the country. Three quarters of deaths were due to adults leaving children unattended, intentionally and unintentionally. Lack of adult supervision was frequently an issue, including in the one quarter of fatalities occurring to children playing in unlocked vehicles.

Overall, almost twice as many boys died as girls. The Institute of Medicine stated that “injury death rates were higher for males than for females in each age group except for infancy when the rates were similar. In 1995 for children 1–9 years of age, injury death rates for males were about 1.5 times the rates for females, and the difference increases with age”.¹⁹ Our results show a greater gender disparity for this cause of death: young boys accounted for three quarters of playing cases and three fifths of unattended cases.

The frequency with which children play or are left in motor vehicles is unknown. However, a survey in the United States reported that 10% of parents of young children thought it “acceptable” to leave children unattended in vehicles, in contrast to a Japanese survey ranking this behavior second on a list of “crimes, unethical and immoral acts”.^20,21

Virtually all cases in our study were identified through searching multiple standard online news sources. The NHTSA special study found 23 deaths in 1998 by searching Nexis; using a broader search strategy, we found 25 in that year meeting our case criteria. No new cases were identified through email lists.

News sources provide information that is not available from traditional public health datasets. We found it helpful to learn the location of the vehicle; circumstances in which the child was entrapped; length of time the child was left; child’s name; vehicle description; parents’ names, ages, and employment status; ages and presence of other children; and legal outcomes. News accounts could also be used to analyze non-fatal events, although these reports frequently lack identifiers, and health and legal outcomes are not reported if the case is referred to child protective services.

Limitations

This study is a descriptive analysis from which relative risk cannot be calculated because of insufficient information about risk exposure (the number of children left unattended in motor vehicles and the amount of time spent in the vehicle). Nor can we calculate incidence without population based data.

Despite our best efforts, we do not believe we captured all incidents occurring in the time period covered. We excluded 20 deaths from this study because we could not confirm they were heat related or learn enough details to ensure that they were not duplicates. Only deaths covered in the news, and particularly in news sources available online, were identified. Different accounts of the same death sometimes contained inconsistent information. To address this, we relied on the most recent news accounts as more accurate than reports issued early in the investigations.

We excluded deaths occurring before 1995 because older internet based news archives are more limited in their availability and scope, and are more expensive to access.

Lack of controlled vocabulary in news archives presents search difficulties. We employed multiple search terms to capture as many variants as possible (for example, child, toddler, tot, baby, infant, boy, girl). We also searched for “hypothermia” deaths using the same strategies as previously described but found only two (non-fatal) cases of child hypothermia in vehicles and a number of instances of reporters, investigators, and judges misnaming hyperthermia deaths.

Prevention

Injury prevention practitioners can work broadly with health professionals and child care providers to increase their knowledge and to set relevant policies to prevent heat related deaths, as well as urge motor vehicle and child safety seat manufacturers to build in warning systems and other design features.

Education

• Remind parents to keep cars locked when not in use, to avoid leaving children in cars, and to look in the car first if a toddler is missing. Emphasize that vehicles are inappropriate places for children to play or to nap: keep them locked and remove sleeping children.²² Parents and caregivers should supervise young children more closely during changes in routine and times of upheaval or family chaos.

• Educate parents about heat vulnerability of young children and about the “greenhouse effect” that can occur rapidly within vehicles. Provide parents with simple reminding devices that a child is in the back seat (a teddy bear in the front seat, or placing their briefcase in the rear seat with the child).

• Promote appropriate use of child safety seats. Child passenger safety advocates should counsel parents that seats are not a safe substitute for adult supervision.

Regulation

• Establish absentee child care policies requiring that absences be phoned in, or parents will be contacted. Parents can encourage their providers to make these procedures part of their safety routines even if no regulation exists.

• Stores and shopping centers could institute frequent patrols of their lots and report incidents to public safety officials.

• National and federal agencies should examine the risks to children confined in vehicles along with other non-traffic events. Motor vehicle trunk entrapment came to national attention in the United States through the efforts of survivor advocates, and fueled by three widely publicized incidents in which 11 children died in one summer.^23,24 NHTSA was mandated to form an expert panel, which recommended vehicle design changes. As a result, a federal motor vehicle safety standard was issued in 2000, requiring interior trunk releases in new vehicles.²⁵ This simple design change, along with widespread public education about the risks of trunk entrapment, may have already reduced this cause of a small number of deaths.

Engineering

• Auto manufacturers should aggressively research technology to reduce the ease with which children gain access to vehicles, or can be inadvertently left behind, especially in child safety seats. The National Aeronautics and Space Administration (NASA) has developed a device to alert drivers when a vehicle is locked from the outside and weight is sensed in the child safety seat.²⁶ (This initiative resulted from the heat related death of an infant, forgotten in a NASA employee’s vehicle.)

Legislation

• Enact policy change by passing relevant laws. Three states—California, Illinois, and Kentucky—have recently passed laws to hold adults responsible for leaving children unattended in motor vehicles.²⁷

SUMMARY

No national data sources track heat related incidents to children in motor vehicles, keeping this cause of death invisible. However, practitioners do not need to wait for national statistics in order to mount public awareness campaigns and other interventions to reduce these preventable deaths. Drivers should be frequently reminded to keep cars locked when not in use, and to avoid leaving children in vehicles for any length of time. Child care providers should implement more effective absentee policies. Auto manufacturers should provide warning systems as an important safety component of family vehicles. More localities should pass laws to make it illegal to leave children unattended in vehicles.