Violent
Death
in Children
in a Metropolitan
County:
Changing
Patterns
of Homicide,
1958 to 1982
Jerome A. Paulson, MD, and Norman B. Rushforth, PhD
From the Departments of Pediatrics, Biology, and Epidemiology and Biostatistics, Case Western Reserve University, Cleveland
ABSTRACT. Death rates from homicide in children younger than 15 years of age in the United States have increased during the last 30 years. Previous studies have suggested a typology consisting of fatal child abuse in young children and community violence in older children. We reviewed the data from the Cuyahoga County, Ohio, coroner’s office pertaining to homicides in children less than 15 years of age between 1958 and 1982 and obtained similar findings. The homicide rates for city children increased from 1.1 to 6.7/100,000 in the first 20 years and
then stabilized. Nonwhite boys had the highest death
rates except in one period. Assailants were usually ado-lescent and young adult men of the same race; however, 43% of children less than 5 years of age were killed by women. The older the child, the more likely the homicide was to have been committed by a nonrelative, outside of the home, and with a firearm. Overall, firearms are the leading cause of homicide (36.2%). The temporal char-acteristics of child homicides are also described. Pediat-rics 1986;78:1013-1020; homicide, child abuse, fatal inju-ries, community violence, firearms.
Whereas total US death mates of children younger than 15 years of age decreased systematically during the past 30 years, those from homicide have
dna-matically increased.1 Homicide has been one of the
leading causes of death in childhood2 and US rates are currently among the highest in the world.3
Several early studies 46 of pediatric homicide focused on the killing of children by their parents. These reports proposed typologic classification
Received for publication Aug 30, 1985; accepted Feb 19, 1986. Presented, in part, at the Ambulatory Pediatric Association Meeting, Washington, DC, May 9, 1985.
Dr Paulson’s current address is Joseph P. Kennedy, Jr, Foun-dation, Washington, DC 20005.
Reprint requests to (J.A.P.)
3800
Porter St, NW, 304,Washing-ton, DC 20016.
PEDIATRICS (ISSN 0031 4005). Copyright © 1986 by the American Academy of Pediatrics.
schemes of child homicide primarily based on the
state of mind of the perpetrators.
More recent analyses using data from the US Department of Health and Human Services, and the Federal Bureau of Investigations-Uniform
Crime Reporting Systems, provide yp78 of
child homicide perpetrated by parents and
nonpar-ents. Jason et al,8 in reviewing data for victims
younger than 18 years of age, identify two broad categories of child homicide. The first, defined as “fatal child abuse” is intrafamilial, occurs most
often in children 3 years of age and younger, and is
associated with bodily force and poorly defined
precipitating events. The second type is extmafa-milial, occurs most often in children older than 12
years of age, and frequently involves firearms or
knives as lethal weapons. It occurs often during
arguments or criminal acts by the assailant. Jason
et a! define this type as “fatal parental-society neglect,” because it represents the death of children unsupervised in an adult environment. They con-cluded that characteristics of child homicide begin
to resemble those of adult homicide at an early age.
Homicides that occur in children between 3 and 12 years of age appear to be a mixture of the two types, an increasing proportion of the adult type occurring with increasing victim age.
Christoffel,9 using US data on the perpetrators and weapon for differing ages of victims, proposes three subtypes for homicides in children younger than 18 years of age. This typology focuses on the environmental risk to child victims and is based on their changing developmental vulnerabilities with age. She characterizes the three types as (1) infan-ticide (parental impatience with infantile behavior
erupts into violence, resulting in CNS injury not
seen in older victims), (2) fatal child abuse and
neglect after infancy (toddlers and preschool
inappro-pmiate demands face fatal punishment), and (3) homicide in the community (homicide involving vulnerability in later childhood, resulting from a discrepancy between increased mobility out of the home and judgment on the part of the school-aged child). Based on this typology, Christoffel proposes a public health approach in reducing fatal violence against children.
Studies of violent death have been ongoing in
Cuyahoga County, Ohio (metropolitan Cleveland),
since the late 19505.10-13 In this urban county, the age-adjusted homicide mate for all victims has in-creased from a baseline level of 5.3/100,000 (1958 to 1962) to a peak of 19.6/100,000 county popula-tion (1973 to 1977) and then decreased to a level of 17.6 in the final 5-year period (1978 to 1982) under investigation. The most dramatic increase in hom-icide mates has been for homicide victims slain by firearms. The proportion of firearm homicide vic-tims increased from 54% to 74.2% during the 25-year time span of the study. Most victims were killed by assailants of the same race, with the highest mates among nonwhite men in the city.
15 has reported on cases in this popu-lation pertaining to child-child homicide and adult-child homicide.
The purpose of the present study was to system-atically analyze the data from the Cuyahoga County coroner’s office and to described trends in homicide rates in the county during the period 1958 to 1982 in children less than 15 years of age. Because the
typologies of child homicide previously proposed
are highly dependent on the age of the victim, we postulated that various characteristics of the hom-icides would be different for differing age groups of the victims.
MATERIALS AND METHODS
Cuyahoga County (“county”) consists of the city of Cleveland (“city”) and an adjacent aggregate of 38 cities, 18 villages, and four townships collectively referred to as “suburbs.” All known or suspected violent deaths in the county are investigated by the coroner, regardless of the fashion in which the violence arose (section 313.12, Ohio Revised Code). Moreover, only the coroner can sign a valid death certificate when injury, however sustained, is en-timely responsible for or is a contributory factor in causing death. Since 1936, Samuel R. Gerber, MD,
JD, has been county coroner. Under his direction, the professional staff has provided a set of uniform records based on consistent criteria for verdicts. Data from police reports, which include eyewitness accounts, and hospital records supplement the an-atomic, toxicologic, and other objective findings of the coroner’s staff. Such data are present for
vim-tually all cases.
Our analysis was based on statistical abstracts of the coroner’s files, broken down into five, 5-year time frames: 1958 to 1962, 1963 to 1967, 1968 to 1972, 1973 to 1977, and 1978 to 1982. Annual hom-icide mates were calculated using data from coroner’s
records for the number of deaths and from the US
Census Bureau publications for population figures.
The latter included a special census for the city of Cleveland carried out in 1965. Average annual hom-icide mates were calculated for the 5-year periods centered around the decennial census years, ie, 1958
to 1962 (1960), 1968 to 1972 (1970), and 1978 to
1982 (1980), using actual census figures as the
denominators. For the 5-year periods 1963 to 1967 (1965) and 1973 to 1977 (1975), the city, suburban, and county populations were estimated by linear interpolations of the decennial census figures, ex-dept for the period 1963 to 1967 (1965) for the city, where population data from the special census were used.
Theme were three justifiable homicides in the age group less than 15 years included in the data. Traffic manslaughters were not included. Fatalities were tabulated according to location of the homi-cide incident (city v suburbs) rather than the place of death or residence of victim. Race is designated
as white or nonwhite; the nonwhite county
popu-lation was 98.7%, 97.3%, and 92.6% black in 1960,
1970, and 1980, respectively.
Homicide rates were expressed as numbers of victims per 100,000 population in each of the 5-year periods, for age groups 0 to 4, 5 to 9, and 10 to 14
years. The number of homicides occurring in the
first week and first year of life were not sufficient for statistical analysis. Rates for the total age group,
0 to 14 years (Fig 1, left) are expressed as
age-adjusted mates for the periods, using the direct methods16 and US total population for 1970 as the standard population.
Statistical significance of changes in mates were
tested using x2 tests for Poisson variables.’7 The rates for the five periods were used in a hetemoge-neity test. Sets of homicide rates not compatible with Poisson variables with the same mean were subsequently tested for linear and nonlinear trends.
x2 goodness of fit tests and
x2
tests for contingency tables were used in testing for temporalrelation-ships in homicides, mode of lethal violence, and
assailant characteristics.17
RESULTS
Long-Term Trends in Homicide Rates
In the 25 years from 1958 to 1982, theme were
7 Victims <l5yrs 6 City 70 5 Victims l5yrs 60 4 50 City 3 County 40 2 30 1 20 1960 10
1970 1980 0
Suburbs
1960 YEAR
Fig 1. Age-adjusted homicide rates for city, suburbs, and county: Left, victims younger than 15 years of age; right, victims older than 15 years of age.
1970
5,153 (90%) occurred in the city. During this period, theme were 235 homicides of children between birth and 14 years of age. Of these, 189 (80.4%) occurred
in the city, 137 (58.2%) were nonwhite, 128 (58.7%)
were boys, approximately one half (117) were less than 4 years of age, 51 (21.7%) were 5 to 9 years, and 67 (28.5%) were 10 to 14 years.
The changing patterns of homicide mates for vic-tims less than 15 years of age were similar to those of victims 15 years and older (Fig 1). The homicide mates for children in the city increased nearly sev-enfold from the first 5-year interval to the fourth interval (P
<
.05) and then stabilized (Fig 1, left). The suburban mate remained stable throughout the quarter century. Overall, for the county, the rate increased from 0.9 to 3.2 (P<
.005, Fig 1, left).0 0 0 0 0 Co
0
> w I-. w 0 0 0 IHomicide trends for older victims mirrored these changes, although the rates were almost tenfold greater throughout the time span of the study (Fig 1, might).
The numbers of homicides and homicide mates for children less than 15 years of age by mace and sex for the city and county are given in Table 1. (County census is given in Table 2.) Rates for nonwhites were higher than for whites of the same sex, nonwhite boys having the highest mates except for the initial period. During the last 10 years, white male rates were higher than those of white girls in both the city and the county.
Age-specific rates for victims 0 to 4, 5 to 9, and 10 to 14 years of age are given in Fig 2. Throughout the entire time span of the study, homicide mates
TABLE 1. Homicides for
Cuyahoga County, Ohio: 1
Children Young 958 to 1982*
em Than 15 Years of Age by Race a nd Sex for the City of Cleveland and
Group and Location
of Homicide 1958-1962 No. Rate 1963-1967 No. Rate 1968-1972 No. Rate 1973-1977 No. Rate 1978-1982 No. Rate
City of Cleveland White boys Nonwhite boys White girls Nonwhite girls Total Age-adjusted Cuyahoga County White boys Nonwhite boys White girls Nonwhite girls Total . Age-adjusted 0 0.0 5 2.2 3 0.8 6 2.7 14 1.1 LO 5 0.5 5 2.2 7 0.7 7 3.0 24 1.0 0.9 6 1.7 14 5.4 6 1.8 8 3.0 34 2.8 2.5 11 1.1 14 5.0 8 0.9 8 2.8 41 1.7 1.6 2 0.7 17 7.0 8 2.9 13 5.2 40 3.8 3.7 6 0.6 18 6.4 14 1.6 14 4.9 52 2.2 2.2 12 5.4 29 13.8 7 3.3 9 4.3 57 6.7 6.7 18 2.5 32 12.2 8 1.1 10 3.8 68 3.5 3.5 12 7.8 14 7.9 6 4.2 12 6.9 44 6.8 6.6 13 2.4 16 6.5 8 1.6 13 5.4 50 3.2 3.2
0 0 0 0
0
Cl)
I-0
>
w
I-w
0
0
0
I
5.0
4.0
3.0
2.0
1.0
0
TABLE 2. Census for Cuyahoga County, Ohio, Population Younger Than 15 Years of Age
Yr City Suburbs
Boys Girls Boys Girls
White Nonwhite White Nonwhite White Nonwhite White Nonwhite
1960 1970 1980
82,881 58,062 30,586
44,960 48,588 35,259
78,869 55,654 28,918
45,177 49,487 35,008
119,599 125,774
77,663
740 7,473 13,750
115,258 120,035
74,056
756 7,576 13,413
1970
YEAR
Fig 2. Age-specific homicide rates for victims younger
than 15 years of age for Cuyahoga County, Ohio.
for victims younger than 5 years of age were highest, increasing systematically with successive periods
(P
<
.02). An approximately parallel increase in mates for victims 10 to 14 years of age occurred (P< .01), whereas mates for the age group 5 to 9 years peaked during the fourth quinquenium (Fig 2). Seg-megating the data by city v suburbs for the four mace-sex groups indicated that nonwhite city male victims younger than 5 years of age had the highest homicide mates of any group: 17.1 and 17.4 victims per 100,000 in the last two periods.
Characteristics of Homicides
Assailant Characteristics. During the last 10 years of the study when the age of the assailant was recorded, the majority of children (51.1%) were killed by assailants between 25 and 39 years of age. Only 7.6% of the victims were killed by other chil-dren (5 to 14 years of age); no deaths were caused by children less than 5 years of age (Table 3). The age group of the victim and the assailant were associated (P
<
.05). For victims 10 to 14 years of age, assailants were younger than those who per-petmated lethal violence on children 0 to 9 years of age.The race and sex of the assailant by mace, sex, and age group of the victim are given in Table 4. Child homicide was almost exclusively intramacial:
96% of victims were killed by members of the same mace. The majority of assailants were men, but a
significantly higher proportion of the children less
than 5 years of age were killed by women (43% v
15%) than victims in the two older age groups,
irrespective of race (P
<
.005).Although approximately one half of the homi-cides for all children were perpetrated by relatives,
the proportion of homicides by relatives decreased
systematically with increasing age of the victim (P
< .005): 71.7% for victim 0 to 4 years, 60.8% for
victim 5 to 9 years, and 13.4% for victim 10 to 14 years. For 1.7% of victims, the relationship between the victim and the assailant was unknown. A sim-ilam gradient was seen in the relative frequency of homicide occurring in the home of the victims (P
< .005): victims age 0 to 4 years (87.2%), 5 to 9
years (80.4%), and 10 to 14 years (56.7%). In 1.7%
ofcases, the location ofthe homicide was unknown.
Only five homicides were perpetrated by
assail-ants while committing another crime, four of which
occurred among victims was 10 to 14 years of age. Forty-seven homicides occurred during a quarrel. The proportion of such homicide increased with increasing age of the victim (P
<
.005): victims age0 to 4 years (11.1%), 5 to 9 years (25.5%), and 10
to 14 years (31.3%).
Methods of Lethal Violence. In the county, the
mode of death varied significantly with the age of the victim (P
<
.005, Table 5) and is known for allvictims in the study. Among children 0 to 4 years
of age, the leading cause of death was violence with a blunt instrument, including hands and feet (42.7%), followed by asphyxiation (17.1%) and
fire-arms (15.4%). With increasing age of the victim,
violence with a blunt instrument decreased in mel-ative frequency as a method of lethal violence,
whereas deaths from firearms markedly increased.
Ten- to 14-year-old victims were killed most
fre-quently by firearms (62.7%), and violence with a
blunt instrument accounted for only 4.5% of the homicides.
Firearms were used more frequently in killing
male than female victims, 41.3% v 28.9%, but this
was due exclusively to a preponderance of male
propor-TABLE 3. Age of Assailant for Homicide Victims Younger Than Cuyahoga County, Ohio, 1973 to 1982, by Age of Victim*
15 Years of Age in
Age of Assailant (yr) Age of Victim (yr) Total
0-4 5-9 10-14
5-14 15-24 25-39 :40
Subtotal More than one Unknown Total
1 17 19 0 37 3 8 48
(2.7) (45.9) (51.4) (0) (77.1)
(6.2) (16.7) (100)
1 (4.0) 3 (12.0) 19 (76.0) 2 (8.0) 25 (80.6) 1 (3.2) 5 (16.1)
31 (100)
5 13 9 3 30 1 8 39
(16.7) (43.3) (30.0) (10.0) (76.9) (2.6) (20.5) (100)
7 (7.6) 33 (35.9) 47 (51.1) 5 (5.5) 92 (78.0)
5 (4.2) 21 (17.8)
118 (100)
Results are numbers (percentages) of victims.
* Age group of assailant was recorded only for the last 10 years of the study period.
TABLE 4. Race and Sex of Assailant for Homicide Victims Younger Than 15 Years of Age in Cuyahoga County, Ohio, 1958 to 1982, by Age, Race, and Sex of Victim*
Assailant Victim’s Age (yr)
Boys Girls
White Nonwhite White Nonwhite
0-4 5-9 10-14 0-4 5-9 10-14 0-4 5-9 10-14 0-4 5-9 10-14
Male
White 12 7 16 0 0 1 10 8 5 0 0 0
Nonwhite 1 0 2 20 12 21 0 0 1 11 6 9
Female
White 9 4 0 1 1 0 10 2 0 0 1 0
Nonwhite 0 0 0 12 3 1 0 0 0 9 3 0
* Results are numbers of victims. Additional data: three nonwhite male, one white female,
and three nonwhite female children were the victims of more than one assailant; the
assailant was unknown in cases involving two white and ten nonwhite boys and eight white and ten nonwhite girls.
TABLE 5. Mode of Homicide for Victims You County, Ohio, 1958 to 1982, by Age of Victim*
nger Than 15 Y ears of Ag e in Cuyahoga
Mode of Homicide
0-4
Age of Victim (yr) Total
5-9 10-14
Firearms 18 (15.4)
Violence with blunt instrument, 50 (42.7) including hands and feet
Cuts and stabs 5 (4.3)
Asphyxia 20 (17.1)
Other 24 (20.5)
25 (49.0) 7 (13.7)
4 (7.8) 11 (21.6)
4 (7.8)
42 (62.7) 3 (4.5)
9 (13.4) 8 (11.9) 5 (7.5)
85 (36.2) 60 (22.5)
18 (7.7) 39 (16.6) 33 (14.0)
* Results are numbers (percentage) of victims.
tions of male and female homicides due to firearms
were equivalent. The proportionate use of firearms
was similar for white and nonwhite victims (white
male, 43.4%; nonwhite male 40.0%; white female, 28.9%; nonwhite female, 28.8%). For victims 5 years of age and older, homicides from firearms increased from the baseline period, peaking in the fourth quinquenium (Fig 3, left). For victims 0 to 4 years of age, firearm homicide mates tended to remain low throughout the 25-year span of the study. In
con-trast, rates of nonfirearm homicides increased
sys-tematically for this age group and were consistently higher than those for victims 5 to 9 and 10 to 14 years of age (Fig 3, might).
Among children 0 to 14 years of age, handguns accounted for 55.4% of fatalities caused by firearms. For the entire age spectrum, handguns accounted for 49.4% of homicides due to firearms (Table 6).
Temporal Characteristics of Homicides. The tern-pomal characteristics of homicides (time of the day, day ofweek, and month ofyear) for victims younger than 15 years of age were compared with those of culpable homicide for victims 15 years of age and older. For victims 15 years of age and older, a pattern of increasing homicide from noon until
midnight emerged, with continued high frequency
for the next three-hour period (1:00 AM to 3:00 AM)
Non-firearms
5.0 - Firearms 5.0
4.0 4.0
3.0- 3.0
10-l4yrs
::
S ,‘\5-gyrs::
YEAR
1960 1970 1980
..
,
.R..
10-l4yrs
#{163}..‘?:.#{232}’
I1960 1970 1980
* Results are numbers (percentages) of homicides.
25
Cl)
w
20
0
I
15- U-0
I-z w
0
LU Q.
10
5-0
1-3 4-6 7-9
AM.
Fig 4. Time of day of homicide by age of victim. -,
0 to 4 years (n = 108); A-A, 5 to 9 years (n = 51);
.. . .-U, 10 to 14 years (n = 67); x- . _ .x, 15 years (n
= 4,539).
0 0 0 0
0
Cl)
I-0
>
w
I-w
0
0
0
I
Fig 3. Age-specific homicide rates by firearms and nonfirearms for victims younger than 15 years of age for Cuyahoga County, Ohio.
TABLE 6. Type of Firearm Used in hoga County, Ohio, 1953_1982*
Homicide,
Cuya-Types of Firearm Victim 0-14 Yr of Age
All Homi-cides
Handgun
Nonautomatic 38 (46.9)
Automatic 7 (8.4)
Subtotal 46 (55.4)
Other gun
Rifle 9 (10.8)
Shotgun 15 (18.7)
Miscellaneous 5 (6.0)
Unknown 8 (8.6)
Subtotal 37 (44.6)
1,422 (45.4) 141 (4.5) 1,563 (49.9)
93 (3.0) 272 (8.7) 93 (3.0) 1,113 (35.5)
1,571 (50.1)
age and older, homicides increased on Friday,
peaked on Saturday, and remained high on Sunday.
This pattern is significantly different from that
seen in victims younger than 15 years of age (P
<
.05). Although Saturday and Sunday had an above
average percentage of child homicides, Tuesday also had a relatively high frequency. In contrast,
Wednesday had relatively fewer child homicide
vic-tims. No distinct differences were noted among the
temporal relationships of the age groups 0 to 4, 5
to 9, and 10 to 14 years, but all three showed the
high Tuesday-low Wednesday pattern.
Although the relative homicide frequencies for
victims older than 15 years of age tended to increase
slightly throughout the year, in the case of the child
homicide victim, no systematic monthly patterns
were discernible. This was evident for the individual age groups and the combined group age 0 to 14
years.
DISCUSSION
We have described the pattern of child homicides
in a single metropolitan county during a time span
of 25 years. The homicide rate that we report, the
relative proportions of the various modes of death,
I I 1 1 I I I I the relationships in terms of race, sex, and age
10-12 1-3 4-6 7-9 10-12 between the victims and assailants are all
consist-P.M. ent with other reports in the litematune.8’10’21
A recent study of homicides in the city of Los
Angeles between 1970 and 1979 did have some
different findings than our study.22 They did not
see an increase in the age-adjusted homicide rates
different from those of younger victims (P
<
.005), for those less than 15 years of age, although thewhose pattern in time depended on the age group overall age-specific homicide rate of 3.9/100,000 for
of the victim. The differences in patterns among that age group is similar to ours. Those less than
the three age groups noted in Fig 4 presumably 15 years of age accounted for 5.4% of all homicide
relate to school attendance. cases in the city, compared to 3.7% in our study.
in Cuyahoga County. Because subunban mates
me-mained constant, this increase in county homicide
rates was due to the dramatic sixfold increase in
child homicide in the city of Cleveland. We consider that such increases reflect greater levels of lethal
violence directed toward children in this urban
community rather than enhanced reporting of
vio-lent child deaths as homicides. Although we do not
believe that theme were systematic errors in
classi-fication of incidents, we cannot eliminate the
pos-sibility of some misclassifications. During the study
period, the mates of classification of incidents as
fatal home accidents in children in the county
remained constant, and increases in the rates of
classification of incidents as violence of unknown
etiology closely paralleled those of child homicide
victims. Thus, theme was no evidence of decreases
in the. rates of other forms of violent death in
children that would have accompanied increased
homicide mates due to increased diagnosis of
homi-cides among violent death in this population. We
cannot eliminate the possibility of concomitant
in-creases in accidental deaths and classification of
incidents as homicide, but we believe this to be
unlikely.
Patterns of child homicide were highly dependent
on the age of the victim. Homicide rates were
consistently higher for children younger than 5
years of age and increased systematically with
suc-cessive 5-year periods. Rates for 5- to 9-year-old
children were lower and peaked during the fourth
quinquenium.This pattern was observed for victims
of firearms in the 5- to 9- and 10- to 14-year age
groups and was evident in the adult homicide rates.
In contrast, deaths from firearms in 0- to
4-year-old victims remained at low levels throughout this
study.
Inspection of the age-specific homicide mates for
children in the three age groups (0 to 4, 5 to 9, 10
to 14 years) for the 5-year periods of the study
indicated increases in the mates of homicide for
successive cohorts. Such relationships have been
observed for mates of adolescent and young adult
suicides in this population.23 The analysis of hom-icide mates into age, period, and cohort components is relatively complex and will be the subject of
subsequent study. However, the increasing rates of
homicide for child victims suggest the possibility of
changes in parenting behavior coincident with
other observed increased mates of violent death in
the 225
Relatively fewer child homicides occurred in the
home, and fewer were perpetrated by relatives as
the age of the victim increased. Those committed
with other felonies occurred almost exclusively with
10- to 14-year-old victims, and in older victims a
greaten proportion of the homicides occurred during
or following a quarrel. A significantly higher
pro-portion of victims in the youngest age group was killed by women, whereas in the older age group
the perpetrators tended to be younger men or boys.
Most of the victims 0 to 4 years of age were killed
by blunt instruments, but the frequency of this
mode decreased systematically with the age of the victim. Conversely, with increasing age, more
vic-tims were killed by firearms or by means of cuts
and stabs. Surprisingly, as many as 15% of the
younger victims were killed by firearms, with
sim-ilar mates in male and female victims, irrespective of race. In the oldest age groups, many more boys
were shot to death than girls. Handguns accounted
for more than 55% of all childhood homicides.
Victims 10 to 14 years of age had a closer
tem-poral pattern to that seen in adults in terms of the hour of the day for the homicide incident. Patterns
for the other two age groups of child homicide were
distinct, with higher frequencies of homicide during different periods of the day.
The observed difference in (1) homicide mate
trends, (2) age and sex of the assailant and
mela-tionship to the victim, (3) location of the homicide and method of lethal violence, and (4) time of day
of the homicide event, for the three age groups of
child victims were somewhat consistent with the
typologies of child homicide advocated by both
Jason7 and Chmistoffel.9 For the youngest age group
(0 to 4 years), the characteristics of the homicide
suggest fatal child abuse. Such homicides tend to
be intrafamilial, occurring primarily in the home,
and are associated with bodily force administered
to an infant or small child. Homicides for victims
10 to 14 years of age have many of the
character-istics of adult homicides: extrafamilial, perpetrated
outside the home, primarily by firearms with a
preponderance of male assailants and victims. Such homicides bear the properties of “fatal
parental-society neglect” proposed by Jason7 and “homicide
in the community” proposed by Chmistoffel.9 Many
of these cases “may represent children drifting into an adult life style and environment for which they
are not developmentally or emotionally prepared
and in which they do not receive adequate parental
societal supervision.”8
The homicide committed against victims 5 to 9
years of age showed characteristics somewhat
in-temmediate between those of the younger and older
age groups. This suggests that homicides in this age
group may be a mixture of the other two types,
consistent with the findings of Jason.7 As with the
data for lethal violence against children in the
United States studied during a more restricted time
began to resemble those of adult homicides at an early age of the victim.
CONCLUSIONS AND RECOMMENDATIONS
1. Children 4 years of age and younger are most likely to be killed by a parent, at home, by a blunt instrument. This is most consistent with a pattern of fatal child abuse. We are currently reviewing the individual coroner’s reports, as opposed to the sta-tistical data summaries used for this paper, to de-temmine how often there was evidence of prior abuse. Are the children who die known to the medical and/or social systems in the community? Had any attempt at intervention been made? Were the episodes of violence that lead to death a first-time event for the child and their families? The answers to these questions have implications for structuring surveillance and intervention systems. Clearly, because large numbers of young children are dying at the hands of their parents, such sys-tems are inadequate.
2. Children of all ages die as a result of injuries sustained from firearms. More than 15% of those younger than 4 years of age, 49% of those 5 to 9 years, and nearly 63% of those 10 to 14 years die of firearm injuries. Slightly more than 55% of these firearm fatalities are due to handguns.
We believe that the problem of child homicides and, indeed, that of all homicides is an extremely complex one, involving numerous social, economic, and other influences. Nevertheless, it seems ob-vious that firearms are in the same environment as children and that the removal of those firearms would lead to a decline in the number of childhood homicides.
REFERENCES
1. Characteristics ofAmerican Children and Youth: 1980,
Cur-rent Population Reports. US Bureau of the Census publica-tion No. 114. Government Printing Office, 1982, p 23 2. Better Health for Our Children: A National Strategy, vol III:
A Statistical Profile. US Department of Health and Human Services publication No. (PHS) 79-55071. Government Printing Office, 1981, tables 20 & 21
3. World Health Statistics Annual Yearbook. Berne, Switzer-land, World Health Organization, 1981
4. Resnick PJ: Child murder by parents: A psychiatric review of felicide. Am J Psychiatry 1969;126:325-334
5. Scott, PD: Parents who kill their children. Med Sci Law 1973;13:120-126
6. D’Orban PT: Women who kill their children. Br J Psychia-try 1979;134:560-571
7. Jason J: Child homicide spectrum. Am J Dis Child
1983;137:578-581
8. Jason J, Gilliland JC, Tyler CW Jr: Homicide as a cause of pediatric mortality in the United States. Pediatrics
1983;72:191-197
9. Christoffel KK: Homicide in childhood: A public health problem in need of attention. Am J Public Health
1984;74:68-70
10. Hirsch CS, Rushforth NB, Ford AB, et al: Homicide and suicide in a metropolitan county: I. Long term trends. JAMA
1973;223:900-905
11. Rushforth NB, Hirsch CS, Ford AB, et a!: Accidental firearm fatalities in a metropolitan county (1958-1973). Am J
Epi-demiol 1975;100:499-505
12. Rushforth NB, Ford AB, Hirsch CS, et al: Violent death in a metropolitan county: Changing patterns in homi-cide(1958-1974). N Engi J Med 1977;297:531-538
13. Ford AB, Rushforth NB, Rushforth N, et al: Violent death in a metropolitan county: II. Changing patterns in suicides (1958-1974). Am J Public Health 1979;69:459-464
14. Adelson L: The battering child. JAMA 1972;22:159-161
15. Adelson L: Slaughter of the innocents: A study of 46 homi-cides in which the victims were children. N EngI J Med
1961;264:1345-1349
16. Colton T: Statistics in Medicine. Little, Brown & Co, Boston, 1974, p 47
17. Snedecor GW, Cochran WG: Statistical Methods, ed 7. Ames, Iowa State University Press, 1980, pp 197-201 18. Klein D, Reizen MS, Van Amburg GH, et al: Some social
characteristics of young gunshot fatalities. Accident Anal Prey 1977;9:177-182
19. Jason J, Carpenter MM, Tyler CW: Underrecording of infant homicides in the United States. Am J Public Health 1983;73:195-197
20. Kaplun D, Reich R: The murdered child and his killers. Am
J Psychiatry 1976;133:809-813
21. Copeland AR: Homicide in childhood: The Metro-Dade County experience from 1956 to 1982. Am J Forensic Med Pathol 1985;6:21-24
22. University of California at Los Angeles, Centers for Disease Control: the Epidemiology of Homicide in the City of Los Angeles, 1970-79. US Department of Health and Human Services, Public Health Service. Atlanta, Centers for Disease Control, August 1985
23. Rushforth NB, Ford AB, Sudak HS, et al: Increasing suicide rates in adolescents and young adults in an urban commu-nity (1958-1982): Tests of hypotheses from national data, in Sudak HS, Ford AB, Rushforth NB (eds): Suicide in the
Young, Boston, John Wright, 1984, pp 45-68
24. Holinger PC: Violent deaths among the young: Recent trends in homicide and accidents. Am J Psychiatry
1979;136:1144-1147