Methods

Each month, all team members received a list of all deaths involving youth 21 years and younger from the Philadelphia vital statistics office. To assure that all deaths were reviewed accurately, the lists were compared with other lists obtained from the state vital statistics, medical examiners' database, and various participating agencies. Although many non-Philadelphia County residents died in the city, only deaths involving county residents were reviewed. Members were responsible for searching their own agencies for all information on the decedents. Because the information discussed by the fatality review members was highly confidential, all members signed a confidentiality statement. To ensure further confidentiality, information was only shared verbally.

At the monthly meetings, agencies presented their information and a staff person entered the relevant information on each of the variables onto specific data forms. There were separate forms for homicides, suicides, and unintentional or natural deaths. Once all the available data were collected on each death, the committee determined whether the death was preventable. A preventable death was defined as one in which, with retrospective analysis, a reasonable intervention may have prevented the death. “Reasonableness” was defined to take into consideration the condition, circumstances, or resources available. At no point during these deliberations was any determination of preventability construed to equate predictability. If the death was judged to be preventable, the committee then discussed or developed interventions and policies to be implemented at the individual, agency and/or community level. The potential intervention strategies and the aggregated data were reviewed at quarterly policy meetings involving team members, as well as invited guests such as city council members, community activists, and agency leaders. Selected injury related death cases were analyzed by demographic characteristics and circumstantial variables.

Results

In 1995, 607 youth aged 21 years and younger died in Philadelphia. This number represented an 8% decrease from 1994 (662 deaths). Sixty per cent were due to natural deaths, 18.6% resulted from homicides, 16.3% resulted from unintentional injuries, 2.3% were due to suicides, and 1.2% were undetermined. This distribution was similar to that in 1994. The leading causes for unintentional injury deaths were residential fires/burns (29.3%), occupancy of motor vehicles (15.2%), drowning (10.1%), and pedestrian injuries (7.1%).

PREVENTABLE DEATHS

Of all deaths, 37% were judged to be preventable. While the majority of natural deaths were judged to be non-preventable, all suicide deaths and the majority of homicides and unintentional deaths were judged to be preventable. Because of incomplete information on a few of the cases, preventability was not determined for 7% of the cases (table 1).

HOMICIDES (N=112)

Among the leading causes of death, homicides ranked first in youth ages 13–21 years. Even among infants, unintentional injuries and homicides tied for the fifth position among the 10 leading causes of death. The majority of homicide victims were African American males (81.4%). Most homicides were perpetrated by firearms, largely automatic handguns (43.4%), while a few were perpetrated by fists, knives, and through hanging or suffocation. Most homicide victims had histories of substance abuse, prior arrests, problems at school such as truancy and poor grades, as well as prior weapon related injuries.

FIRE/BURN DEATHS (N=29)

All 29 deaths due to fire/burns were judged to be preventable with more than half involving situations where the fire was started by a child aged 3–6 years. In half of the cases, the source of the fire was matches while the child was playing, and in the remaining cases the source was a cigarette (n=4) or a lighter (n=3). Other sources of the fire included space heaters and faulty electrical wires (table 2).

Discussion

The concept of using the multidisciplinary death review team to analyze child deaths and from the data develop cost effective interventions is important. Regardless of the age group, injury was found to be a significant cause of youth death in Philadelphia. Child death review teams have contributed significantly to knowledge about child abuse and neglect. These traditional child death review processes have often been referred to as a “witch hunt” by some practitioners outside of the child abuse and neglect arena because of their focus on identifying the perpetrators. The PIYFRT process, on the other hand, has been expanded beyond this perspective and reflects a public health mission to prevent youth mortality, but most importantly to promote a healthy lifestyle. This process can be viewed as a “gold mine”. It is a gold mine in as much as by examining deaths, a cost effective alternative injury surveillance methodology can be created to better understand youth fatality in any community. In addition, this process alleviates or decreases “turfing” and increases collaboration among agencies, the stakeholders of children's health. Perhaps the greatest asset is that these data, in conjunction with data from other communities (currently child death review teams now exist in almost every state and many local communities), offer a wealth of information to examine the many factors that increase the likelihood of injury that may lead to death.

As a result of this process, the following products and policy changes were generated and are being implemented in Philadelphia:

• Improved coordination among PIYFRT members.

• More comprehensive data collection to reduce the number of missing data; a decrease in inadequate death certificates¹²; a link between birth and infant death certificates; use of height/weight charts in the medical examiner's office; and access to the 911 history for selected deaths.

• Prevention of interpersonal violence through development of a program for first time juvenile violators of the Uniform Firearm Act; increasing agency communication on juvenile bench warrants; recognizing the need for school nurses' training on domestic violence; recognizing the need to highlight to mothers the potential risk of assuming that all intimate partners are appropriate caregivers to their young children; and identifying the need for multidisciplinary, multiagency school intervention teams.

• Prevention of unintentional injuries through increasing the enforcement of the child safety seat law; debating traffic regulations of public transportation buses used as school buses during school hours; increasing school based pedestrian safety education; recognizing the need for “Fire Starter” programs for children ages 3–7; and recognizing the need for non-battery powered smoke detectors.

This process allows for innovative data collection. The diverse data from many agencies allow a broader perspective and multidisciplinary input, thus the steps to more complete research. To illustrate this point, a scenario is the 5 year old who dies after playing with matches. When all fire related deaths are aggregated, most reviews by researchers would identify the lack of a smoke detector as the “potential” risk factor and a working smoke detector use campaign as the primary prevention strategy. There is a need to better understand the barriers and enabling factors as they relate to behavior. Through the PIYFRT process, the “barrier” identified was inadequate supervision by a family member who left matches available to a 5 year old. Both reviews would have come to the same conclusion—that the death was preventable. But only the PIYFRT process would identify the “barrier,” the lack of supervision by the caretaker and the intervention strategy—that is, highlighting the need for programs for young children who start fires. In this era of scarcity, understanding barriers and enabling factors is essential in making decisions about efficient resource allocations.

Many times an agency appears with what they believe is the “only story” as opposed to understanding that they have only one piece of the puzzle. The process of sharing confidential information frequently involves exposing the limitations of agency systems and the uncomfortable realization that many of those limitations fail to protect children. For team members to become effective in this process, a critical component is the willingness to participate. Participation increases as members develop trust in the process and better understand that the limitations of failures become the basis for system change.

The PIYFRT review process involving community participation with agencies (public and private) allows many of the pieces of the puzzle to be brought together. The comprehensive data generated serves to complete the puzzle needed by decision makers to formulate solutions. The benefits of sharing information can be viewed through the development of prevention strategies, interventions, products, policies, etc.

The challenge is to expand child death reviews beyond the focus of child abuse and neglect to one of public health so as to identify preventable child deaths and achieve effective prevention. Presently, this process may serve as the one source of comprehensive data on the many factors contributing to youth deaths. It also offers a menu of what can be done, as well as providing the necessary information for decision makers competing for scarce public health resources. The PIYFRT serves as a model on the local level of a public health/community based approach which generates science based information and converts it into concrete action, a gold mine for injury prevention and control.