Child Fatality Review: An Interdisciplinary Guide and Photographic Reference

Section I

INTRODUCTION

CASE EXAMPLE

A 20-month-old girl was brought to the ER by her parents. The girl was pale, limp, and breathing only occasionally. Her pupils were dilated and poorly reactive, and she gazed to the left. Neurological examination yielded a Glascow coma score of 5. Emergency medical staff intubated the child, established an intravenous line, and gave her pressor medications. Her blood glucose was 300 mg/dL, electrolyte levels were normal, her hemoglobin level was 8.8 g/dL, and the white blood count was 20 000 mm³. Blood cultures were drawn and antibiotic treatment initiated. The girl was taken for a computed tomography scan, which revealed a left subdural hemorrhage and cerebral edema.

In the pediatric intensive care unit, the girl’s condition deteriorated. An ophthalmology examination showed numerous bilateral multilayer retinal hemorrhages. The treating physician ordered a skeletal survey, but the order was deferred when it became apparent that the child would die. Brain death criteria were met and life support was withdrawn 40 hours after her admission.

The autopsy revealed extensive cerebral edema with herniation, bilateral retinal hemorrhages, a moderately sized left subdural hemorrhage, and scattered microscopic subarachnoid hemorrhages. No signs of impact to the head were observed, and the skeletal survey did not show any abnormalities. The cause of death was listed as shaken baby syndrome (SBS) and the manner as homicide.

Both parents were available to provide a history of the girl’s condition. The mother stated that she left home at about 8:15 AM to go to work. She fed her daughter and left her in the care of the father. The father reported that during the morning, the child seemed fine; she played and ate lunch. After lunch, she took a nap, from which she awoke at 2:30 PM, and then played some more. At about 4:00 PM, when the father was watching television, the child suddenly arched back from a sitting position. She was limp but may have made several jerking motions with her arms. The father called the mother at work, and she came home. Once the mother saw the child, the parents took their daughter to the ER.

The hospital child protection team reviewed the case. Based on this review, the father was arrested for felony murder and, at the time of the child death team review, was awaiting trial.

The parents had no prior police or child protective services reports, and this girl was their only child.

CHILD DEATH REVIEW TEAM ANALYSIS

—The team concurred with the diagnosis of SBS, the hospital child protection team’s analysis of the timing, and hence also the identification of the likely perpetrator.

—The professionals involved appeared to perform appropriately.

—The public health member of the team presented data revealing:

—The statewide hospital coding database recorded about 30 cases of SBS per year. About half were coded as 995.55 (International Classification of Diseases, 9th revision); the others were some combination of intracranial hemorrhage and retinal hemorrhage. Additional cases in which a child died before a diagnosis could be made in the hospital would have been missed with this analysis.

—The hospital costs for these cases averaged slightly over $32 000 per patient per year.

—This cost did not include physician fees, which would perhaps double the figure, nor clinic visits and tests for the survivors.

—Less than 10% of the lifetime costs of a brain-damaged individual are medical. Education, police, legal, and other factors are responsible for the other costs.

—A crude estimate is that the average SBS case costs more than $1 million over a lifetime.

—The team noted that funding for child abuse prevention programs related to physical abuse in some way totaled to approximately $1.2 million in the state.

—The conclusion was that this death was preventable.

SUBSEQUENT RECOMMENDATIONS FOR PREVENTION

—Support prevention programs on coping with crying. If even 1 of the 30 estimated annual cases of SBS in the state were prevented, the expenditure of $1 million on prevention efforts would equal the cost that the victim would have incurred to the state over his or her lifetime.

—Refer the mother to a parent support group for SBS.

Chapter 1

OVERVIEW

Randell Alexander, MD, PhD, FAAP

Death is a universal experience of all living creatures. At one time, many children died from diseases and conditions that are now treatable. The expectation for most 21st-century parents is that their children will outlive them. Unfortunately, this is not always true.

The creation of child death review teams (CDRTs) is foremost a formal attempt to answer the question, Why? The question comes from the perceived unnaturalness of a child dying, intellectual curiosity about how this could be so, and grief-inspired anguish. Additional questions include What could have prevented this? and What could we have done better? It is in the structure of CDRTs, the issues they face, and their commitment to children that prevention recommendations are made.

RATIONALE FOR CHILD DEATH REVIEW

The inspiration for child death review arose primarily from the concern about child abuse. While questions periodically arise about possible clusters of birth defects and deaths, environmental toxicology has not been the stimulus for the formation of child fatality review (and may indeed be underappreciated if or when it does exist). Motor vehicle–related fatalities are major causes of childhood death, yet this did not serve as the inspiration for the spread of CDRTs. Therefore, it was not the fact of child fatalities alone that caused such teams to arise, but rather concerns about the occurrence of undetected homicide caused by child abuse.

Child abuse first received attention in 1962 with the publication of The Battered Child Syndrome in the Journal of the American Medical Association.¹ This article described physical abuse, particularly cases with multiple and recurrent injuries. As a result of this publication, within the next 5 years all 50 states enacted child abuse reporting laws. Indeed, the Journal of the American Medical Association has cited this article as one of its landmark articles, acknowledging its pioneering importance in shaping the field of child abuse.²

These early descriptions of child abuse focused on physical abuse; sexual abuse and neglect were better appreciated later. Although much physical abuse is overt (eg, skin injuries) and potentially easily recognizable, there are some forms of physical abuse that may be very difficult to detect. For example, the findings of shaken baby syndrome were first linked with shaking as a mechanism of action in the early 1970s.³-⁵ In these early descriptions, cases were identified as shaken baby syndrome only if there were no signs of impact to the head. A child would be seen who had no apparent external signs of trauma but considerable brain injury. Shaken baby syndrome injuries (brain injury and swelling, intracranial bleeding, and retinal hemorrhages, in most cases) are internal and, thus, they can be missed in cases of sudden death without an autopsy. Later, coexisting impact trauma to the head⁶,⁷ and elsewhere on the body⁸ was better characterized. Nevertheless, many cases of shaken baby syndrome have no external signs of physical abuse.

Likewise, abusive abdominal trauma often leaves no external signs. When hit, the skin of the abdomen is not trapped against adjacent underlying bone, as is the case with skin over the tibia, forehead, cheekbones, or other sites commonly seen with accidental bruising. Punching, stomping, or impacts with blunt objects (eg, knee, baseball bat) are not high velocity mechanisms like slaps, which can leave bruising on areas of skin without underlying bone. Unless these mechanisms of injury drive with such force and depth that they trap the skin against the spine and cause bruising, the skin of the abdomen is usually clear, even with considerable internal injury.

Mandatory autopsy laws and much of the motivation of CDRTs have stemmed from the fact that a child with no signs of external trauma (eg, shaken baby syndrome, abdominal trauma) might not be correctly diagnosed as being a victim of child abuse. The victim might then be buried with an incorrect diagnosis such as sudden infant death syndrome (SIDS). The consequences of this mistake could be considerable by allowing a perpetrator of fatal child abuse to remain in a position to hurt or kill other children. Underascertainment of specific causes of child fatality cases results in incorrect death certificates and a distorted picture of how children die.⁹,¹⁰

Although fear of missing a child abuse death may have been the initial rationale for the establishment of most CDRTs, the key charge to all such teams is prevention. For each death that is reviewed, many teams explicitly consider the degree to which the child’s death was preventable. Consider Case Study 1-1.

Case Study 1-1

A 3-year-old child was unrestrained in a car when a railroad crossing gate began to lower. Late for an appointment, the mother attempted to go around the gate but was hit by a train traveling at 48 km per hour. The car was dragged for approximately ½ km before the train was able to stop. The mother and the 3-year-old child were both killed at the scene. Upon review of the case, information was provided by the police that the car was severely crushed. Child protective services (CPS) did not get involved because there were no other children. The CDRT noted that the child was unrestrained in violation of state law and medical standards for proper restraint systems. Although proper child restraint often saves a child’s life, in this instance the crash was so severe that even if restrained, the child would have died. Thus this case was classified as unpreventable with regard to the seat restraint issue but preventable with regard to the mother’s actions.

Definitions of preventability vary, but many are similar to this: A preventable death is one in which an individual or a community could reasonably have done something that would have changed the circumstances that led to the death.¹¹

Three options of relative preventability are preventable, possibly preventable, or unpreventable. Teams may use other terminology but commonly face the issue that preventability is not all or none. The compilation of recommendations across cases comprises the reports made by state and local teams in an effort to reduce child deaths, and the degree of preventability helps with the prioritization of such recommendations. In Case Study 1-1, preventability is high with regard to the mother’s attempt to go around a railroad gate. This does not obviate the need for proper child restraint in other cases in which such restraint might make a difference in survival.

CHILD DEATH REVIEW TEAMS

Over time, the mission of CDRTs has expanded in many jurisdictions to include non—abuse-related deaths and older children rather than simply looking at child abuse—related deaths in young children. Yet a wide range still exists between teams that look only at deaths determined by CPS to be due to child abuse (eg, Florida) and those that look at deaths of all children (eg, Oklahoma). Some teams have begun to look at serious child abuse that is not fatal. Multidisciplinary fatality review teams for intimate partner violence (less precisely known as domestic violence) have arisen in some states and have been modeled after CDRTs.

There is also a spectrum of what child death review means. At a state level, the review process often includes cases that are 6 to 12 months old. Any decisions about whether a crime was committed are almost always made by local authorities before the time of review. The purpose of state CDRTs is prevention and is focused on public health. At a local level, child death review may occur at scheduled intervals or near the time of a child’s death. At times, such teams might be considered child investigation teams, and the composition of the team may be more slanted toward the prosecutor, medical examiner, CPS, and law enforcement.¹²

Internationally, CDRTs may function at a local, state/province (eg, Australia, Canada), or national level (eg, New Zealand).¹³

TERMINOLOGY

Different traditions among different professionals have resulted in a variety of somewhat conflicting or imperfect terminology. It is unlikely that there will be general agreement in the near future about any of them, but it is important that the reasons for each be understood. Working on CDRTs means becoming exposed to such terms and ensuring that professionals are communicating clearly with each other despite the seemingly different languages they use. Particularly important is the need to understand that the same word can mean different things to different people and that death review is impaired if such differences are ignored.

A number of specialized terms are used in child death review (eg, homicide, which refers to death at the hands of another, not necessarily a crime). However, certain terms can cause more argument.

CHILD ABUSE

When used generically this means all forms of child abuse. The 3 major types of child abuse include neglect, physical abuse, and sexual abuse. Thus it is redundant to say child abuse and neglect, though physical abuse and neglect makes sense. The next most common distinction is psychological maltreatment or emotional abuse. If not specified, these are included under neglect.

CHILD ABUSE VERSUS CHILD MALTREATMENT

Child maltreatment refers to a much larger set of conditions than child abuse. It could mean conditions such as poverty, poor intellectual stimulation for children, and a variety of behaviors not quite constituting child abuse. Child abuse is a subset of child maltreatment. Most child abuse cases do not result in court action, and when court action is pursued it primarily takes place in juvenile court. Thus, child abuse often does not equate a criminal action. When a child dies at the hands of a caregiver, it might be an accident, child maltreatment, child abuse, or murder. Some professionals find terms such as child abuse too strong and prefer even milder terminology (such as child maltreatment). From a child fatality viewpoint, this does not apply.

ACCIDENTS

Accidents are leading causes of death during childhood. Professionals who specialize in injury prevention and the Centers for Disease Control and Prevention do not like the term accident, arguing that this implies a certain randomness. However, many accidents are predictable. For example, many states have published lists of which street intersections have the highest rates of motor vehicle crashes. Thus, the risk of motor vehicle crashes is not random by location. (Note that such a viewpoint prefers motor vehicle crashes [MVCs] to the term motor vehicle accidents [MVAs]). Demographic variables help to predict the relative likelihood that a given individual will be involved in an accident; thus accidents can be at least partly preventable.

—Strengths. By focusing on the preventability of accidents, this approach helps to decrease morbidity and mortality. It also emphasizes a certain statistical orientation to the study of injury.

—Weaknesses. It is unlikely that the public will change their use of the term accident. Most who eschew it substitute an even worse term: unintentional.

INTENTIONAL AND UNINTENTIONAL INJURIES

These terms are used primarily by those who study injury prevention that does not include abuse. However, there is an overlap, in that neglectful situations and those who neglect disproportionately are overrepresented in even nonabusive injuries. To prosecutors and law enforcement on CDRTs, intention is a legal distinction and requires some degree of trying to understand motive.

—Strengths. This is an attempt to avoid the use of accident.

—Weaknesses. Neglect is primarily unintentional, yet neglect is not included as unintentional by those who use such terminology. Intent is difficult to study (eg, what was the person thinking?) and is inferential in most instances. Intentional injury does not equate with abusive injury. This is an attempt to provide a euphemism for abuse versus accident, and it creates even more problems.

ABUSIVE AND NONABUSIVE INJURIES

Use of this terminology is more prevalent by child abuse pediatricians. It avoids the use of accident but does not necessarily imply knowledge of a caregiver’s intent in situations in which the caregiver was a participant.

—Strengths. This avoids the presumption of motive and correctly distinguishes between child abuse and nonabuse (accidental) injuries.

—Weaknesses. The term accident will still be used by the public and most professionals. As with the other terms, it requires a judgment about whether something was abusive.

INFLICTED TRAUMA OR INJURY

This term is meant to be a substitute for abusive injury in an attempt to avoid the term abuse. It is meant to be less pejorative though still conveying the notion of an abusive injury. The National Institutes of Health (NIH) has helped to promote this usage.¹⁴ A subset of this approach is the use of the term inflicted neurotrauma as a substitute for shaken baby syndrome though, taken literally, inflicted neurotrauma could refer to a child injured in a driveway when a motor vehicle backs up (not the use apparently intended).

—Strengths. It is responsive to dissatisfaction by some professionals with the term abuse.

—Weaknesses. Theoretically, every trauma is inflicted. When a child falls and hits his or her head on a swing set, the swing set inflicted the injury. Hence it does not mean anything by itself, only if the custom becomes to substitute the term inflicted for abusive. This would mean that the term inflicted would take on a specialized professional meaning at odds with public meaning. To prosecution and law enforcement, such terminology can come across as a whitewash of the violence seen in abusive child deaths.

SIDS VERSUS SUDS VERSUS OTHER

The term sudden infant death syndrome (SIDS) arose with the observations of a pattern of infant death that had no apparent explanation. Although the cause(s) of SIDS are not yet clear, the major success of placing infants on their backs while they are sleeping has focused increased attention on what constitutes this diagnosis. Sudden unexpected death syndrome (SUDS) (or similar terminology) is sometimes used when a child was lying on his or her stomach or in other ambiguous situations. The notion of SUDS is to make a distinction that a risk factor was present that might partially explain the unexpected death in a way that is not classic for SIDS.

Some pathologists have described the cause of some deaths as sudden unexpected death while cosleeping when the child died in the parent’s bed but without evidence of overlaying. Although technically accurate, this diagnosis can distress the parents who were involved, making them feel guilty that they were responsible for the death of the child. The counterargument would be that cosleeping is a risk that the parent assumed, and the diagnosis merely reflects that. For the CDRT, a decision needs to be made about how to code this, which is an opportunity for considerable debate.

Another coding issue is the rare instance in which there appears to be a second SIDS case within a family. Although the rarity of such cases often leads to suspicions of hidden medical conditions or child abuse, even in situations in which it appears to be a genuine second SIDS death, how should it be coded? One of the key definitions of SIDS is that there be a clean review of the clinical circumstances. Does the presence of a prior infant death mean that the clinical circumstances are not clean and the second case should not be classified as SIDS? By using such an operational definition, there are no second SIDS deaths, they all would be undetermined. Not all would agree with this logic.

DEATH VERSUS FATALITY

Death and fatality are often used interchangeably and are used synonymously in this book. Interestingly, the common usage sometimes makes subtle distinctions. Fatality is more likely to be used if the death was sudden or from fate. Thus fatalities might describe deaths from natural disasters or fires but would be much less likely to be used when describing cancer deaths or HIV deaths. Whether the terms child death review or child fatality review are used, the meaning is clear.

CAUSE AND MANNER OF DEATH

The cause of death is the reason the child died. This can be straightforward (eg, gunshot) or complex (eg, pneumonia secondary to HIV acquired via sexual abuse). Sometimes it is difficult to ascertain the precise cause of death. For example, some children who are multiply abused may have many injuries, none of which would have been fatal by itself, yet the combination led to fatal shock. Thus the cause of death in this instance would be multiple injuries. Sometimes the cause of death is not precisely stated. For example, children do not die of a subdural hemorrhage, but rather of the brain swelling associated with the forces causing both the brain and subdural trauma.

The manner of death is the classification of how the child died: accidental, homicide, natural, suicide, or unknown. A given cause may represent different manners depending on the circumstances of the case (eg, head trauma as a cause of death may represent accidental, homicide, or suicide in a particular case).

It used to be said that the pathologist had the final word about the cause of death. Increasingly, this might be said for CDRTs. However, the manner of death can pose challenges for team members.

One example of differences in how teams may view a manner of death is SIDS. SIDS is one of the key diagnoses of interest to CDRTs. Numerous CDRTs highlight recommendations about SIDS ranging from smoking cessation during and after pregnancy to educating health care personnel and the public about putting a child on its back to sleep. Assuming the team reaches the correct diagnosis of SIDS as the cause of death, what is the manner of death?

Table 1-1 shows how SIDS is categorized in various states. The states in Table 1-1 were chosen randomly, and the classification of the manner of death that was assigned to the diagnosis of SIDS was extracted from the states’ latest reports. Most states classified SIDS as a natural death. The rationale for this determination is that SIDS is a discrete syndrome with strong diagnostic criteria. Although the cause or causes of SIDS may not be precisely known, lack of knowledge about the cause of a condition has been common throughout the history of medicine and still persists today for many diagnoses (eg, Alzheimer disease, many cancers, autism). Even if SIDS turns out to be from multiple causes, the presentation is stereotyped. Very rarely does an infant who was found to have died as a result of SIDS turn out to have been physically abused. Although cosleeping is a risk factor present in over 50% of all SIDS cases in many state reports, SIDS occurs in other settings as well. Whatever the exact cause of SIDS, it does not appear to be an accident, homicide, or suicide. The argument is that it is also not undetermined in that it is a definite entity. Furthermore, when it is suggested that SIDS is an unknown condition, the family is never truly absolved of a tragic occurrence for which it is not at fault.

Note that 2 states (Ohio and Kansas) indicated that SIDS is a natural death but explicitly separates SIDS from other natural deaths in their reporting figures. The rationale may be that SIDS has a different age distribution at the time of incidence, is quite prevalent, and deserves special consideration in investigation, grief counseling, and prevention recommendations.

Iowa considers a SIDS death to be undetermined. The argument is that the cause of SIDS is by definition unknown. If its cause were known, then it could be classified into one of the other manners of death. Many of the risk factors parallel those of child abuse (eg, poverty, teenaged mothers) and undetermined deaths more than those of natural deaths. The role of positional asphyxia, environmental temperature, and possible rebreathing illustrates that some of what was called SIDS may have accidental causes. In this view, calling SIDS a natural death is inherently flawed, and it should be called undetermined.

The question of who is right will be decided over time. Yet the arguments illustrate the thoughtfulness of such teams in trying to understand this tragic manner of death.

Another problem posed in child death review is how to detect and assign relative weights to apparent multiple causes contributing to death. Although death certificates have lines to list secondary causes of death, humans are complex and sometimes die only when a series of unfortunate conditions align themselves to create a fatality. Consider how a CDRT would classify the situations in Table 1-2.

When annual reports are made that list the causes and manners of death, they represent the best estimates of the team, not certainty about every evaluation. Although this may pose problems with the prosecution or the follow-up of an individual case, the advantage of the CDRT is that it uses aggregate data to arrive at system conclusions. Thus, trends may be usefully tracked even when absolute numbers are not as certain.

INHERENT BENEFITS AND PROBLEMS WITH CHILD DEATH REVIEW

For many years, groups of adults undoubtedly gathered to discuss why a child died and what could have been done to prevent it. Within the last 30 years, more systematic, formal multidisciplinary teams have considered not just individual deaths but also deaths in the aggregate. In many spheres of medicine, multidisciplinary teams have become a paradigm of optimum care (eg, developmental disabilities, other specialty clinics). This approach has a number of benefits, including the following:

—The overlap of expertise may be synergetic in arriving at insights.

—The team process likely decreases errors in commission and omission.

—Learning is likely enhanced as a result of the team process.

—Emotional and professional support is provided for the team members.

Commonly, CDRTs issue periodic reports to inform and make prevention recommendations. The success of such recommendations depends on the extent to which decision makers are attentive and the extent to which the recommendations themselves represent successful strategies. However, there are limitations to an approach that reviews individual deaths and sums up their impact. For one, this process takes time.

One analogy could be brought to the team by an experimental psychologist. All animal senses are created to detect change. For example, a frog retina is built to detect dark, moderately fast-moving objects of a certain visual angle (corresponding to flies in their natural environment) but does not respond well or at all to slow moving objects.¹⁵ For the same reason, predators move slowly when they stalk prey so as not to trigger motion detectors in their victims. Contrast and motion are basic functions of sensory systems throughout the animal kingdom, and it is on this basis that brain structure is organized.

In contrast to anecdotal experiences (a sudden phenomenon), slow trends over time are a weakness in human detection. Thus an influenza epidemic, killing perhaps fewer than 100 children in the United States in a single winter, likely receives more media attention than child abuse, which kills about 2000 children in the United States over the course of an entire year. Steady state deaths become the norm and may elicit a tacit acceptance that nothing can be done to change them. For example, fatalistic statements are sometimes made that poverty, child abuse, and MVCs will always be present. The task for CDRTs is to overcome this cultural bias and, by use of statistics and process, generate attention to such issues and create new solutions to reduce such deaths.

Neglect is another example in which slow actions (or inactions) are more difficult to appreciate. Failure to thrive sometimes can be slow and difficult to recognize until it becomes pronounced.¹⁶ Children may be increasingly left unsupervised outdoors or near bodies of water as caregivers do not encounter any adverse events and thereby lessen their guard. Understanding these dynamics can help teams when formulating recommendations. Teams themselves can experience drift in their determinations as well. A team might slowly shift from viewing cosleeping as a risk factor to an accidental cause of death to one that is neglectful. A solution to individual and team drift is observation by outside viewers not habituated to the issue in question. Peer review is a quality assurance function that can help teams identify how they may have changed viewpoints on issues over time. With the possible exception of nested (hierarchical) child death review (eg, local team results critiqued by state teams), there is very little, if any, ongoing peer review of CDRTs.

Changing or different cultural standards can also pose dilemmas for teams. For example, 30 years ago, children involved in MVCs who were ejected and died were considered victims of unfortunate accidents. Today, many would consider failure to restrain a child to be neglect. Governmental policies or regional customs that enlist children as soldiers in warfare may be viewed as necessary or a form of criminal exploitation. Therefore, some changes in how deaths are viewed may justifiably be expected over time, and lack of agreement about how to view deaths sometimes occurs.

Small numbers also pose a problem for sensory detection. Formally known as signal to noise ratio, the difficulty is to distinguish between random events among the ongoing larger number of background events and those events that are significant. This theoretical construct applies when several children die of a certain cause or as a result of a given lapse in societal protection. When a local team reviews the situation, they may not see a pattern. Amassing larger numbers, a state team may detect enough cases to warrant attention. However, some relatively rare situations may require a larger dataset before a pattern can be seen. In the United States, this may mean that sharing a dataset across states or even the nation may be necessary before a small but significant pattern of deaths can be detected. Currently there is no national dataset in the United States and no international compilation of child death data.

DECISIONS SUBSEQUENT TO CHILD DEATH REVIEW

Child death review teams commonly make recommendations for legislative and regulatory changes. In addition, they may advocate for professional and public education that would directly or indirectly reduce the chances for child death. Challenges facing teams are being heard and assessing whether the suggested interventions actually work. A cost-benefit consideration is implicit in each recommendation, but rarely is a price attached to a death in such a way as to make the comparison overt. For example, consider a home visitation program that costs about $3000 each year per family in a jurisdiction with 100 000 births per year and covering the first 2 years of life. At a cost of $300 000 000 per year how many lives would need to be saved to reach a break-even point? Is an infant life worth $1 million, $10 million, $100 million? And what are the odds that the program can be shown to work, especially if it has a low rate of success?

Even when deaths are correctly classified and tabulated and effective recommendations are made, the motivation for change by individuals or organized entities remains. Based on the number of recommendations made by CDRTs over the years, it is legitimate to question whether society cares enough about children to commit the necessary resources to reduce child fatalities. Although team members are no doubt committed to the health and welfare of children, it is less clear that public policy strongly shares this concern. For example, there are approximately 42 000 traffic deaths (all ages) in the United States each year, a number not much lower than the approximately 52 000 American deaths in the Vietnam War. More than 60 000 children have died of child abuse in the United States since the Vietnam War ended. However, the outcry about the slaughter of people on the highways and children in their homes is relatively absent, and there is no national monument in Washington, DC, to memorialize these tragedies. Until prevailing attitudes about children and the inevitability of some of their deaths is changed, CDRTs will continue to be busy counting dead children.

CONCLUSION

A child’s death is an unexpected event. The original inspiration for child death review arose from concerns about child abuse, but the focus has broadened to include other child deaths as well. The functions of CDRTs vary from state to state and have evolved over time, but the common goal is to understand the reasons for child deaths and to learn how to prevent such deaths in the future.

REFERENCES

1.Kempe CH, Silverman FN, Steele BF, Droegemueller W, Silver HK. The battered-child syndrome. JAMA. 1962;181:17-24.

2.Landmark article July 7, 1962: the battered-child syndrome. By C. Henry Kempe, Frederic N. Silverman, Brandt F. Steele, William Droegemueller, and Henry K. Silver. JAMA. 1984;251:3288-3294.

3.Guthkelch AN. Infantile subdural haematoma and its relationship to whiplash injuries. Br Med J. 1971;2:430-431.

4.Caffey J. On the theory and practice of shaking infants. Its potential residual effects of permanent brain damage and mental retardation. Am J Dis Child. 1972;124:161-169.

5.Caffey J. The whiplash shaken infant syndrome: manual shaking by the extremities with whiplash-induced intracranial and intraocular bleedings, linked with residual permanent brain damage and mental retardation. Pediatrics. 1974;54:396-403.

6.Duhaime AC, Gennarelli RA, Thibault LE, Bruce DA, Margulies SS, Wiser R. The shaken baby syndrome. A clinical, pathological and biomechanical study. J Neurosurg. 1987;66:409-415.

7.Alexander R, Crabbe L, Sato Y, Smith W, Bennett T. Serial abuse in children who are shaken. Am J Dis Child. 1990;144:58-60.

8.Alexander R, Sato Y, Smith W, Bennett T. Incidence of impact trauma with cranial injuries ascribed to shaking. Am J Dis Child. 1990;144:724-726.

9.Crume TL, DiGuiseppi C, Byers T, Sirotnak AP, Garrett CJ. Underascertainment of child maltreatment fatalities by death certificates, 1990-1998. Pediatrics. 2002;110(pt 1):e18.

10.Overpeck MD, Brenner RA, Cosgrove C, Trumble AC, Kochanek K, MacDorman M. National underascertainment of sudden unexpected infant deaths associated with deaths of unknown cause. Pediatrics. 2002;109:274-283.

11.Iowa Child Death Review Team. Review of Child Deaths for Calendar Year 2003. Des Moines: Iowa Child Death Review Team; 2004. Available at: http://www.idph.state.ia.us/common/pdf/publications/cdrt_report_2004.pdf. Accessed February 8, 2006.

12.US Advisory Board on Child Abuse and Neglect. A Nation’s Shame: Fatal Child Abuse and Neglect in the United States. Washington, DC: US Dept of Health & Human Services; 1995.

13.Durfee M, Durfee DT, West MP. Child fatality review: an international movement. Child Abuse Negl. 2002;26:619-636.

14.Reece RM, Nicholson CE, eds. Inflicted Childhood Neurotrauma: Proceedings of a Conference Held in Bethesda, MD. Elk Grove Village, Ill: American Academy of Pediatrics; 2003.

15.Lettvin JY, Maturana HR, McCulloch WS, Pitts WH. What the frog’s eye tells the frog’s brain. Proc IRE Inst Radio Eng. 1959;47:1940-1951. Reprinted in McCulloch WS, ed. Embodiments of Mind. Cambridge, Mass: MIT Press; 1965.

16.Amundson J, Sherbondy A, Van Dyke DC, Alexander R. Early identification and treatment necessary to prevent malnutrition in children and adolescents with severe disabilities. J Am Diet Assoc. 1994;94:880-883.

Chapter 2

EPIDEMIOLOGY OF CHILD DEATH

Randell Alexander, MD, PhD, FAAP

Megan Meisner

Key functions of child death review teams are the collection, characterization, and public documentation of data about deaths. Thus, most child death review teams spend considerable time reviewing cases, determining how a child died and assigning the death to a category, assembling case demographics, examining patterns, and preparing reports that summarize these data. The result is a snapshot of how children have died for the period in question (usually over the course of a calendar year). If done well, this then serves as a powerful aid in developing prevention recommendations and prioritizing prevention efforts.

A basic operating assumption underlying this process is that how children died in one year is a good predictor of how they will die in the next unless some type of effective prevention or intervention is implemented (eg, a vaccine or superior medication). Although episodic disasters may be missed with this short-term focus of looking only at the last several years, most teams have found that one year’s performance is usually a good predictor of the next year’s.

The epidemiology of child fatality is, therefore, an important part of the child death review process and enables the monitoring of trends over time, which is essential when tracking the possible effectiveness of prevention programs.

PROBLEMS WITH COUNTING CHILD FATALITIES

No one knows how many children die each year or of what causes. Official databases (eg, vital statistics) are subject to numerous limitations, basically consisting of poor data leading to poor summary statistics. It is the job of the child death review team to attempt to come close to the ideal of describing who dies and why, something no person or organization has been able to accomplish to date.

Capturing all child deaths is not as simple as may be imagined. Some reasons this may be difficult are listed in Table 2-1. in remote areas of the world, formal record keeping may be spotty, especially for indigenous peoples. In some parts of the world there is not an obvious central repository of child deaths. Local records (eg, church records, family plots) may be the only way to find information on the deaths of children. In addition, for a number of reasons, children may become separated from their families and have no one to confirm their identity. These reasons include natural causes (eg, parents dying of HIV, disasters) as well as human causes (eg, wars).

Table 2-2 illustrates some of the ways in which the cause of death can be difficult to ascertain or to compare across jurisdictions. Even if there were an accurate tally of all child deaths, their causes of death are not always recorded in the same way by different authorities. Superficially, this could be as simple as sudden infant death syndrome (SiDS) being labeled as cot death in England. These types of synonyms pose no problem for a human coder but cause problems with electronic databases. More of an obstacle is seen when such a death is coded by a medical examiner as sudden unexpected infant death while cosleeping. Is this the equivalent of SIDS, or is cosleeping (without obvious overlaying) another category or a subcategory? Shaken baby syndrome is coded as 995.55 in the International Classification of Diseases, listed as shaken infant syndrome.¹ If the hospital or the medical examiner refers to the death as craniocerebral trauma, inflicted neurotrauma, or blunt force head trauma, it would take a sophisticated system to infer that these are probably the same, since the coding may be different. Teams routinely wrestle with trying to impose order on the efforts of many different certifiers of death.

Another problem with assigning manner to a child’s death is deciding what factors were responsible, filtered through the cultural lens of the certifier. For example, if a child dies in a motor vehicle crash, is it strictly an accident? Might poor highway maintenance be listed as a factor if it applies? Do automobile manufacturers bear some responsibility for failure to build safer cars? Are politicians responsible for failing to enforce safety issues and simply allowing predictable and high rates of motor vehicle deaths? A given culture might have different views of these priorities and a different sense of preventability.

In many instances, autopsies of children are not obtained. With a sophisticated medical system, many causes of death can be determined without an autopsy, but not all. Obvious accidents, chronic diseases, and thorough medical workups can yield a high degree of accuracy in diagnosis. Yet the level of certainty of all cases may be lacking. With less sophisticated medical systems, the accuracy of diagnosis may be considerably less and the data about how children die may be compromised to a small extent (with the most sophisticated medical systems including strong autopsy laws and practices) or to a large extent (systems with rare autopsies and less sophisticated medical practices).

LIFE EXPECTANCY

Life expectancies have improved dramatically around the world within the last several hundred years. More reliable food sources, sanitation, and, more recently, the advent of antibiotics and effective medical care have increased the probability of a child surviving to adulthood. Survival to adulthood improves the overall average life expectancy of any given cohort followed from birth until death.

In the Middle Ages, the life expectancy at birth was 20 to 30 years.² Table 2-3 illustrates the large gains in life expectancy at birth following that time period. The gains in life expectancy accelerated mostly within the last 150 years in Western Europe and the United States. Gains in India and China have accelerated since 1950. Africa made lesser gains in the last 55 years, but has recently been contending with the AIDS epidemic, which may slow or reverse life expectancies. The world rise in life expectancy understandably parallels a major acceleration in world population.

During the Roman Empire the average newborn had a life expectancy of 22 years, and the average life expectancy in England during the Middle Ages was approximately 30 years. However, this does not mean that each person had a 20- to 30-year life span. An extreme example of statistical reasoning would be if half of the children died around birth and half lived to be 50 years old. The resultant bimodal distribution of mortality would yield the average of 25 years. For a gain in life expectancy to occur, 1 of 4 possibilities must occur: older people live longer, more children live to older ages, both, or the birth rate is significantly reduced so that there are fewer children (a falling birthrate means more older people proportionately). (Arguably, a major proportional increase in only the healthy cohort of the population could also inflate the results, but the mechanisms and numbers to accomplish this are too unlikely to be seriously considered and could in some scenarios be the equivalent of eugenics.) Figure 2-1 shows that in the United States, the life expectancy for those who have already reached 65 years of age has not dramatically increased in the last 100 years—perhaps 6 to 10 years at most. In contrast, the life expectancy for a person at birth has increased from around 50 years in 1901 to the mid to upper 70s today. Therefore, this increase is mostly the result of more people surviving to 65 years of age, primarily from a decrease in child fatality.

Not all countries have enjoyed the benefits of a significant increase in life expectancy. Table 2-4 shows countries in which an actual decline occurred in either female or male life expectancies between the period of 1955 to 1960 and the period of 1995 to 2000. Two patterns emerge: an increase in adult mortality or an increase in mortality for all ages. Although the mortality rate for children younger than 5 years declined in Russia, Ukraine, and Uganda, life expectancy at birth declined somewhat. In Russia and Ukraine, adults have presumably engaged in destructive activities such as alcoholism, increasing the death rate during adulthood itself. In Rwanda, Zambia, and Zimbabwe, the mortality rate for children younger than 5 years increased and life expectancy decreases were seen. War or high rates of an epidemic such as AIDS can account for countries in which such declines are seen, because they involve mortality that affects all ages.

Figure 2-1. Life expectancy by sex of people at birth and aged 65 years in the United States. Reprinted from National Center for Health Statistics.⁴

DEFINITIONS

An infant is defined as a person younger than 12 months. Sometimes an additional distinction is made for neonate as a person between birth and 30 days of age. A child is someone between 1 and 18 years of age. It is often useful to distinguish between toddlers, grade school—aged children, and teenagers, since dynamics of these age groups and causes of death vary. Some of these age differences are noted in subsequent chapters when causes of death are discussed. For the purposes of this chapter, the only subdivisions are between infant and child to avoid a virtual atlas of tables and statistics. Nevertheless, it is important to note that useful distinctions do apply between these different ages, distinctions that also may be confounded between different cultures in which the expectations and roles of children may vary greatly (eg, children as farmers, soldiers, other types of workers).

INFANT AND CHILD MORTALITY STATISTICS

INFANT MORTALITY: WORLD

Approximately 8 million infants die annually worldwide. Table 2-5 shows infant mortality rates for selected regions. Sweden, Iceland, and Hong Kong are among those with the lowest rates. The economic development of a country is a strong predictor of its infant mortality rate, though the infant mortality rate in the United States lags behind more than 20 other countries.

Table 2-6 lists the 10 countries with the highest infant mortality rates. Nearly all are in Africa. Reductions in world infant mortality might, therefore, best begin with analysis and intervention into the underlying reasons for such high rates in Africa.

The causes for infant deaths can be difficult to ascertain for reasons cited above. Table 2-7 illustrates the causes for infant deaths as reported by the World Health Organization (WHO). The numbers reflect the percentage of deaths in each country attributable to a particular disease entity. The considerable variability in percentages likely reflects more than true differences in the rates of causes. Differences in classifications of these and other conditions likely contribute to the different percentages as well. Nevertheless, it is apparent that infectious and parasitic diseases are more prevalent in South Africa than in well-developed countries such as Japan.

INFANT MORTALITY: UNITED STATES

Table 2-8 lists the 10 leading causes of death in the United States for infants and older children in 2002 as recorded by the Centers for Disease Control and Prevention (CDC). Perinatal problems (including at least the categories of congenital anomalies, short gestation, maternal pregnancy complications, placenta/cord problems, and intrauterine hypoxia) are the leading causes of death for infants, totaling at least 13 579 per year.⁸ Although concentrated in a short period of time after birth, perinatal problems are the leading cause of death as compared with any other childhood cause at any age. As compared with causes in developing countries, infectious diseases are not a major cause of death for infants in the United States.

Primarily in the first 6 months, SIDS is the leading cause of death not related to perinatal factors. Thereafter, child abuse is the leading cause of death until sometime later in the preschool years, when accidents become the leading cause. Accidents remain the leading cause of death into adulthood. (Note that the term unintentional injury used in Table 2-8 does not refer to all unintentional injuries [accidents and neglect] but only to accidental injuries. The use of the concepts of intentional and unintentional do not follow the legal meanings of these terms, the common meaning of the word intent, or child abuse concepts; rather, they emerge from the injury prevention community, terminology also promoted by the CDC.)

Note that all of these numbers are ultimately derived from death certificate data, which do not accurately reflect some causes of death. For example, there are at least 2000 child abuse deaths each year in the United States,⁹ and many of these children die in the first year of life. Table 2-8 indicates only about 2000 homicides in children older than 1 year, which would also include teenaged murders. Clearly the numbers for homicide on death certificate data systems are substantially wrong by not including as many child abuse deaths as exist.

Changes in the US infant mortality rate over the last half century are shown in Figure 2-2. Neonatal and postneonatal mortality have declined. However, the largest decline in rate has been with neonatal mortality. Most recently, data from 2002 showed a small increase in infant mortality to 7.0 deaths per 1000 live births from 6.8 in 2001, the first rise in US infant mortality since 1958.⁹ It may be that, along with a decline in late fetal mortality, some infants are born alive who would otherwise have been stillborn. Further understanding of these figures and whether they represent a trend awaits analysis of data from later years.

Infant mortality in the United States is associated with race and cultural background, in part likely explained by socioeconomic cofactors. Figure 2-3 illustrates that as a class, infant mortality is lowest for Asian or Pacific Islander populations, about equal for white and Hispanic populations, and highest for black, non-Hispanic populations.

Figure 2-2. Mortality rates for infants, neonates, and postneonates in the United States. Reprinted from National Center for Health Statistics.⁴

Figure 2-3. Infant mortality rates by mother’s race/ethnic origin. Reprinted from National Center for Health Statistics.⁴

Figure 2-4. Infant mortality rates by state per 1000 live births. Reprinted from Kids Count.¹⁰

However, within such groupings there is considerable variability. Black (African American) infants have more than double the mortality rate as compared with whites or Hispanics. American Indian or Alaskan Native infants have the second highest infant mortality rate. Although the rate is higher for Hawaiian infants versus whites, Japanese or Chinese ancestry yields the lowest infant mortality rates in the United States. Along with other health issues, the study of racial/ethnic disparities has received increasing attention from federal funding authorities.⁴

Figure 2-4 shows infant mortality rates for each state. Note that the highest infant mortality rates tend to be in the southeast and the lowest in the upper plains, Midwest, and New England. This is unlikely to be the result of significant differences in medical expertise between regions but rather correlated with poverty and health care accessibility.

CHILD MORTALITY: WORLD

Table 2-9 illustrates child mortality rates for children younger than 5 years for selected countries for the years 1960, 1990, and 2000. Thus in the year 2000, children younger than 5 years in Angola were nearly 74 times as likely to die as their counterparts in Iceland, Norway, Sweden, and Japan. Although infectious diseases (eg, diarrhea, malaria, HIV) are much less a problem in the countries with the lowest younger-than-5 fatality rates, poverty, war, and poor health care systems also bear much of the blame.

Several causes of death for several countries are listed in Table 2-10. The first row (all causes) lists the percentage of children between 1 and 14 years of age who die in each country (crude death rate). The other rows show proportions of deaths that are due to specific causes. In South Africa, infectious and parasitic diseases and HIV account for a much more substantial proportion of deaths than they do in Japan. In contrast, of the children who die in Japan, proportionately more die of heart disease. However, it is the reduction of other causes of death that makes a higher rate of heart disease, as shown by the fact that the overall death rate is lower. In developing countries, diarrhea, malaria, and measles—diseases that are infrequently seen or pose little problem in developed countries—account for significant numbers of deaths. The shift in patterns of causes of death parallels that seen in the general population.

Figure 2-5. Child mortality rates by state per 100 000 children. Reprinted from Kids Count.¹⁰

CHILD MORTALITY: UNITED STATES

Figure 2-5 shows state comparisons for child fatality rates. As compared with infant mortality (Figure 2-4), geographic distinctions are less clear-cut.

Mortality rates for 10- to 24-year-old people have declined over the last 2 decades as seen in Figure 2-6. In 2001, children between 10 and 14 years of age had a nearly 5 times lower death rate than did 20 to 24 year olds. Figure 2-7 breaks this down by gender and age. As compared with their 10- to 14-year-old counterparts, 20- to 24-year-old women have about 3 times the death rate and men have about 6 times the death rate. When 20- to 24-year-old women and men are compared, the young women have about one third the death rate of young men. This gender disparity tends to be because of greater lifestyle differences between genders (eg, homicide, risk taking, motor vehicle crashes). Gender differences are similar across ethnic/racial groups as seen in Figure 2-8. Black, American Indian, and Hispanic male adolescents (10 to 24 years of age) have the highest mortality rates. Asian/Pacific Islanders have the lowest mortality rates for both sexes.

Figure 2-6. Mortality rates by age. Reprinted with permission from the National Adolescent Health Information Center (NAHIC).¹¹

Figure 2-7. Mortality rates by age and gender for 2001. Reprinted with permission from the NAHIC.¹¹

As shown in Table 2-8, accidents are the leading cause of death for children between the ages of 1 and 18 years in the United States. As discussed previously, child abuse is undercounted, and thus the homicide rate is actually somewhat higher than that listed. Cancer is the third leading cause of death. Note that suicide is more skewed to the older end of this age range. This is also seen in Figure 2-9, which examines the causes of death for 10 to 24 year olds. Combining all causes of accidents (motor vehicle and other unintentional injuries), they account for 44% of all deaths in this age range.¹¹ Figure 2-10 illustrates that motor vehicle crashes are the leading cause of death for these age ranges, followed by homicide, then suicide. A strong increase in each category is seen with increasing age. Note that the homicide rate for 15 to 24 year olds is 14.8 times that of 10 to 14 year olds. Racial/ethnic disparities in mortality rates for individuals between 10 and 24 years old are shown in Figure 2-11. The largest disparity is seen with homicide rates, which are highest in the black popu-lation. Black males in particular are at higher risk for homicide than for motor vehicle crashes. American Indian/Alaskan Native (non-Hispanic) adolescents have the highest motor vehicle crash rates. In Alaska, this is related to widespread use of all-terrain vehicles, an inherently unsafe vehicle.¹² Other causes of deaths include cancer (21.7%), heart disease (11.6%), congenital anomalies (6.2%), and other accidents (12.6%).

Figure 2-8. Mortality rates by race/ethnicity and gender for 2001. Reprinted with permission from the NAHIC.¹¹

Figure 2-9. Leading causes of death for adolescents in 2001. Reprinted with permission from the NAHIC.¹¹

Figure 2-10. Mortality rates by cause and age for 2001. Reprinted with permission from the NAHIC.¹¹

Figure 2-11. Mortality rates by cause and race/ethnicity for 2001. Reprinted with permission from the NAHIC.¹¹

CONCLUSION

All of the data presented in this chapter are subject to certain limitations. There is no single, completely reliable clearinghouse for child death information within the united States or between countries. (See Table 2-11 for a list of organizations that can be used as resources.) For example, different data sources can yield somewhat conflicting answers about a cause of death. Inconsistent labeling practices and failure to consistently identify all the factors contributing to a death add to uncertainties about overall conclusions. Child death review teams can be of major assistance in resolving many of these issues, but they still are in their relative infancy in this regard.

It is clear that at the beginning of the 21st century, wide variations in infant and child mortality exist in different parts of the world. Resources no doubt exist to reduce such disparities, but the political imperative to do so has not yet been achieved. The first day of life carries high risk in that prematurity, congenital anomalies, and other perinatal factors interplay to determine whether a newborn is viable. Surviving the risk of SIDS, then child abuse, and later accidents, a child faces an increasing risk of death in adolescence from a variety of lifestyle choices. In other parts of the world, famine, AIDS, war, and poverty exact a substantial toll. By more accurately identifying these numbers and causes, interventions can be more carefully planned and the effects of such interventions measured.

REFERENCES

1.World Health Organization. International Classification of Diseases (ICD-9-CM). 9th rev, clin mod, 6th ed. Washington, DC: National Center for Health Statistics; 2001.

2.Walton GM. A Brief History of Human Progress. Rev ed. Davis, Calif: Foundation for Teaching Economics; 2005. Available at: http://www.fte.org/capitalism/downloads/ CAPWalton EssayRevised5-12-05.doc. Accessed October 23, 2006.

3.United Nations Population Division. World Population Prospects: The 2004 Revision. Population Database. New York, NY: United Nations; 2005. Available at: http://esa.un.org/unpp/index.asp. Accessed October 24, 2006.

4.National Center for Health Statistics. Health, United States, 2005, With Charts on Trends in the Health of Americans. Hyattsville, Md: National Center for Health Statistics; 2005.

5.United Nations Children’s Fund. Trends in mortality and life expectancy. Progress Since the World Summit for Children: A Statistical Review. New York, NY: United Nations Children’s Fund; 2001. Available at: http://www.childinfo.org/cmr/revis/db1.htm. Accessed October 3, 2006.

6.Population Reference Bureau. 2001 World Population Data Sheet. Washington, DC: Population Reference Bureau; 2001.

7.Central Intelligence Agency. Rank order: infant mortality rate. CIA: The World Factbook Web site. Available at: https://www.cia.gov/cia/publications/factbook/rankorder/2091rank.html. Accessed October 3, 2006.

8.Kochanek KD, Martin JA. Supplemental analyses of recent trends in infant mortality. National Center for Health Statistics Web site. Available at: http://www.cdc.gov/nchs/products/pubs/pubd/hestats/infantmort/infantmort.htm. Accessed June 29, 2006.

9.US Advisory Board on Child Abuse and Neglect. A Nation’s Shame: Fatal Child Abuse and Neglect in the United States. Washington, DC: US Dept of Health & Human Services; 1995.

10.State-level data online. Kids Count Web site. Available at: http://www.aecf.org/kidscount/sld/compare.jsp. Accessed July 27, 2006.

11.National Adolescent Health Information Center. 2004 Fact Sheet on Mortality: Adolescents & Young Adult. San Francisco: National Adolescent Health Information Center, University of California; 2004. Available at: http://nahic.ucsf.edu//downloads/Mortality.pdf. Accessed July 27, 2006.

12.American Academy of Pediatrics.