“Gray Murder”: Characteristics of Elderly Compared With Nonelderly Homicide Victims in New York City

Abstract

Objectives. We compared characteristics of homicides among New York City residents aged 18 years and older from 1990 to 1998 to determine differences in demographics, cause and place of death, and presence of illicit drugs and alcohol in the deceased’s system.

Methods. All medical examiner–certified homicides among New York City residents aged 18 years and older from 1990 to 1998 were studied (n = 11 850). Nonelderly (aged 18 to 64 years) and elderly (aged 65 years and older) victims were compared by gender, race/ethnicity, cause of death, place of death, and presence of illicit drugs or alcohol. Population-based homicide rates stratified by age, gender, and race were also calculated.

Results. Nonelderly homicide victims were significantly more likely to be male, non-White, to have been shot in the city streets, and to have evidence of illicit drug or alcohol use. Elderly victims were more likely to be female, White, to have been killed by nonfirearm injuries, and to have been killed in their own homes. The gender and race differences between age groups remained but were attenuated when population-based rates were compared.

Conclusions. The characteristics of homicide in nonelderly adults do not apply to elderly adults in New York City. Demographic factors and vulnerabilities of the elderly may underlie these differences, pointing to the need for oversight of isolated or homebound elderly persons and for protective interventions.

Previous research has emphasized the predominance of young men, particularly young African American and Hispanic men, as victims of homicide in the United States.1 There have also been suggestions that the availability of firearms and the high rates of cocaine or opiate use by young African American and Hispanic men in urban environments may contribute to these statistics.2^–4

By contrast, relatively little attention has been given to victims of homicide aged 65 years and older.5^,6 Geriatric homicides may be overlooked because of investigators’ presumption that advanced age and chronic illness provide sufficient explanations for most deaths.7 However, a better understanding of what has been termed “gray murder”7^,8 might contribute to the recognition of geriatric deaths as homicides. Clarification of the characteristics of elderly homicide victims could also inform the development of preventive, age-sensitive interventions by physicians, social workers, and adult protective service or law enforcement agencies.

We analyzed data from the Office of the Chief Medical Examiner of New York City to describe a group of elderly homicide victims (aged 65 years and older) and to compare them with nonelderly adult victims (aged 18 to 64 years). We hypothesized that the pattern of characteristics associated with nonelderly adult homicide victims in urban settings—overwhelmingly male, African American or Hispanic, and with evidence of illicit drug or alcohol use on autopsy1^–4—would not apply to elderly homicide victims in New York City. We also predicted that there would be differences in place and cause of death between elderly and nonelderly homicide victims.

METHODS

All deaths from 1990 through 1998 among New York City residents aged 18 years or older that occurred within city limits and were certified as homicides by the Office of the Chief Medical Examiner were studied. In each case, the medical examiner’s final determination of homicide was based on a review of evidence gathered by the investigator on the site, as well as on autopsy data when available and information from other corroborative sources, such as police reports. International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM ) codes E960 through E969 were originally used to designate cases of homicide.9 Some cases coded as ICD-9-CM E980 through E989, indicating that the fatal injury may have been accidentally or unintentionally inflicted, were also included but not individually identified in the data set. Cases of homicide–suicide—i.e., in which the perpetrator killed himself after killing the victim—were also included but not specified. Other suicides were not included.

Exclusions comprised individuals who were killed in New York City but were not city residents, as well as homicide victims who were homeless or whose New York City residence could not be verified. Selection of the 9-year study period was based on availability of data for analysis. Demographic characteristics of the victims, cause of death, place of death, and results of toxicological examinations were collected from the medical examiner’s files. These data were double entered and checked by trained research assistants. Analyses were conducted with SAS version 8.2 (SAS Institute Inc, Cary, NC).

Classification of Subjects

Demographics.

Each homicide case was classified into 1 of 2 age categories: non-elderly (aged 18–64 years) and elderly (aged 65 years or older). Classification of the race/ ethnicity of the homicide victims was made according to information from multiple sources, including, when available, the family members or others who identified the body; thus, determination of race/ethnicity did not rely solely upon the sometimes-altered appearance of the body. Four racial/ethnic groups were created on the basis of the mutually exclusive categorization scheme used in the 2000 US Census10 and in some iterations of the 1990 Census11: non-Hispanic White, non-Hispanic African American, Hispanic of any race, and other. The other group was comprised principally of non-Hispanic Asians, but also included small numbers of non-Hispanic American Indians or Alaska Natives, and non-Hispanic Native Hawaiians or Pacific Islanders.

Cause and place of death.

Four categories were used to classify the cause of death in each case: firearms, cutting, beating, and strangling. A fifth category was reserved for all other causes; these included fires or burns, pushing- or malevolently induced falls, heart attacks or strokes precipitated by criminal activity, train- or subway-related homicides, and unspecified causes.

Six categories were used to classify place of death: victim’s home, other residence, work-place, automobile, street, and outdoor public place other than the street (e.g., city park, open lot, bridge, or public transit). A seventh category for place of death was reserved for all other locations, including indoor public places (e.g., hotel lobby, bar or restaurant, retail store, jail) and places that were either unknown or not specified.

Toxicological tests.

Postmortem toxicological testing was performed at the Office of the Chief Medical Examiner for 3 illicit drug types (cocaine, opioids, or cannabis) and also for alcohol (ethanol). Homicide victims were considered to have undergone toxicological testing if the presence of any of the 3 illicit drug types or ethanol was assayed. Cases were included for data analysis of toxicological test results for illicit drugs only if estimated intervals of 3 days or less between the fatal injury and death could be documented. Injury–death intervals of 2 hours or less were required for ethanol assays. Victims were classified as illicit drug–positive if evidence of 1 or more of the 3 drug types was confirmed and as alcohol-positive if detectible levels of ethanol were confirmed.

Data Analytic Procedures

Nonelderly and elderly homicide victims were compared according to gender, race/ethnicity, cause of death, and place of death using 2-tailed χ² tests. In addition, population-based annual homicide rates were calculated for age by gender and age by race/ethnicity stratifications for New York City residents aged 18 years and older killed within city limits from 1990 through 1998. The stratum-specific denominators for homicide rates were estimated with population means for the 9-year study period.12 These means were derived from US Census counts of 2000 and 199010^,11 obtained from the Population Division of the New York City Department of Planning (2000 Census Summary Files 1 and 2; 1990 Census Summary Tape Files 1, 1A, and 2). All homicide rates were calculated per 100 000 population and are presented with 95% confidence intervals.13

Nonelderly and elderly homicide victims were also compared on rates of positive toxicological tests for illicit substances (cocaine, opioids, or cannabis) and alcohol (ethanol) using the χ² test. To assess for ascertainment bias among homicide victims who underwent toxicological testing, we compared the proportions of individuals who were tested for illicit drugs or alcohol across age, gender, and race/ ethnicity. Each statistical test used a 2-tailed α level of 0.05. We recognized that, given the number of homicides involved in these analyses, statistical power would be substantial. Therefore, the magnitude of differences is also described in the “Results” section.

RESULTS

For the 9-year period of 1990 through 1998, there were 11850 deaths of residents aged 18 years or older in New York City that were certified as homicides by the Office of the Chief Medical Examiner. For unknown reasons, the records of 1 homicide omitted designation of gender and records of 14 homicides omitted designation of race/ethnicity. Of the 11850 homicides, 9806 underwent postmortem toxicological testing for illicit drugs or alcohol; 2044 were not tested.

Demographic Variables

Men predominated among all adult homicide victims, but the proportion of men among nonelderly victims (86.3%) was significantly greater than that among elderly victims (57.3%; Table 1 ▶). As is seen in the population-based homicide rates (Table 2 ▶), this finding is accounted for in part, but not entirely, by the overrepresentation among women in the New York City population aged 65 years and older.

1.

Demographic Characteristics of and Circumstances of Death Surrounding Adult Homicide Victims: New York City, 1990–1998

	Aged 18–64 y, No. (%)	Aged ≥65 y, No. (%)	Total, No. (%)
Gendera
Men	9883 (86.3)	229 (57.3)	10 112 (85.3)
Women	1566 (13.7)	171 (42.8)	1737 (14.6)
Total	11 449 (100)	400 (100)	11 849 (100)
Race/ethnicityb
Non-Hispanic African American	5621 (49.2)	141 (35.3)	5762 (48.7)
Non-Hispanic White	1065 (9.3)	180 (45.0)	1245 (10.5)
Hispanic	4300 (37.6)	73 (18.3)	4373 (36.9)
Other	450 (3.9)	6 (1.5)	456 (3.9)
Total	11 436 (100)	400 (100)	11 836 (100)
Cause of deathc
Firearms	8573 (74.9)	92 (23.0)	8665 (73.1)
Cutting	1577 (13.8)	86 (21.5)	1663 (14.0)
Beating	583 (5.1)	102 (25.5)	685 (5.8)
Strangling	385 (3.4)	59 (14.8)	444 (3.7)
Other	332 (2.9)	61 (15.3)	393 (3.3)
Total	11 450 (100)	400 (100)	11 850 (100)
Place of deathd
Victim’s residence	2484 (21.7)	265 (66.3)	2749 (23.2)
Other residence	735 (6.4)	5 (1.3)	740 (6.2)
Workplace	565 (4.9)	11 (2.8)	576 (4.9)
Automobile	513 (4.5)	5 (1.3)	518 (4.4)
Street	5279 (46.1)	64 (16.0)	5343 (45.1)
Other outdoor public place	444 (3.9)	13 (3.3)	457 (3.9)
Indoor public place/unknown	1430 (12.5)	37 (9.3)	1467 (12.4)
Total	11 450 (100)	400 (100)	11 850 (100)

Note. Of 11 850 confirmed homicides, records of 1 homicide omitted designation of gender and records of 14 homicides omitted designation of race.

^aχ² = 261.108; df = 1; P < .001.

^bχ² = 528.402; df = 3; P < .001.

^cχ² = 748.675; df = 4; P < .001.

^dχ²=441.665; df = 6; P < .001.

2.

Annual Homicide Rates (With 95% Confidence Intervals [CIs]), by Demographic Characteristics: New York City, 1990–1998

	Aged 18–64 y Homicide Rate (95% CI)	Aged ≥65 y Homicide Rate (95% CI)	Total Homicide Rate (95% CI)
Gender
Mena	48.03 (47.09, 48.99)	7.18 (6.28, 8.17)	42.55 (41.72, 43.38)
Womenb	6.89 (6.55, 7.24)	3.27 (2.8, 3.8)	6.21 (5.92, 6.51)
Totalc	26.43 (25.95, 26.92)	4.75 (4.3, 5.24)	22.9 (22.49, 23.32)
Race/Ethnicity
Non-Hispanic African Americand	52.72 (51.35, 54.12)	9.09 (7.66, 10.73)	47.18 (45.97, 48.42)
Non-Hispanic Whitee	6.17 (5.81, 6.56)	3.38 (2.9, 3.91)	5.51 (5.21, 5.83)
Hispanicf	38.37 (37.23, 39.53)	6.55 (5.14, 8.24)	35.49 (34.45, 36.56)
Otherg	10.72 (9.75, 11.75)	1.41 (0.52, 3.06)	9.86 (8.97, 10.80)
Totalh	26.4 (25.92, 26.89)	4.75 (4.3, 5.24)	22.88 (22.47, 23.29)

Note. Homicide rates are per 100 000 population.

^aPopulation denominators = 2 286 226, 354 626, and 2 640 852, respectively.

^bPopulation denominators = 2 526 930, 581 253, and 3 108 183, respectively.

^cPopulation denominators = 4 813 156, 935 879, and 5 749 035, respectively.

^dPopulation denominators = 1 184 601, 172 273, and 1 356 874, respectively.

^ePopulation denominators = 1 916 685, 592 444, and 2 509 129, respectively.

^fPopulation denominators = 1 245 226, 123 772, and 1 368 998, respectively.

^gPopulation denominators = 466 645, 47 392, and 514 037, respectively.

^hPopulation denominators = 4 813 157, 935 881, and 5 749 038, respectively.

Differences between nonelderly and elderly homicide victims were found in the distribution of racial/ethnic groups. Non-Hispanic African Americans (49.2%) and Hispanics (37.6%) predominated among nonelderly victims, whereas non-Hispanic Whites (45.0%) and non-Hispanic African Americans (35.3%) predominated among elderly victims (Table 1 ▶). When population-based homicide rates were compared, non-Hispanic African Americans and Hispanics had higher rates than did non-Hispanic Whites in both nonelderly and elderly groups, but the ethnic disparities were greatly reduced among elderly victims (Table 2 ▶).

Circumstances of Death

Firearms were a significantly more frequent cause of death among nonelderly than elderly victims, accounting for 74.9% of homicides among those aged 18 to 64 years (Table 1 ▶). By contrast, firearms (23.0%), cutting (21.5%), and beating (25.5%) contributed in similar proportions among the elderly victims (Table 1 ▶). Nonelderly victims were significantly more likely than were elderly victims to be killed in the street (46.1% vs 16.0%, respectively), whereas elderly victims were more often murdered in their own residences than were nonelderly victims (66.3% vs. 21.7%, respectively; Table 1 ▶).

Among the homicide victims who underwent postmortem toxicological analysis, the nonelderly group had a significantly higher rate than the elderly group of detection of either illicit drugs (41.5% vs 8.6%; χ² = 121.834; df = 1; P < .001) or alcohol (33.6% vs 20.0%; χ² = 12.694; df = 1; P < .001). However, significantly higher proportions of nonelderly than elderly (83.2% vs 69.8%; χ² = 49.022; df = 1; P < .001), male than female (84.0% vs 75.5%; χ² = 74.268; df = 1; P < .001), and non-White victims (non-Hispanic African American, 82.5%; Hispanic, 84.8%; and other race, 86.2%, vs non-Hispanic White 76.1%; χ² = 55.192; df = 3; P < .001) were tested for either illicit drugs or alcohol than were victims from other demographic groups.

DISCUSSION

From these data emerges a picture of elderly homicide victims that differs from that of nonelderly victims. The nonelderly adult homicide victim in New York City was likely to be male, non-Hispanic African American or Hispanic, and to have died after being shot in the city streets.1^–4 However, this pattern could not be applied directly to elderly homicide victims. Elderly victims had lower population-based homicide rates overall and included an increased representation of women, non-Hispanic Whites, and individuals who died from nonfirearm injuries, such as cutting, beating, or strangling compared with nonelderly victims. The data also suggested that many elderly homicide victims were killed in their own homes or in locations other than on city streets. Finally, nonelderly homicide victims were more likely to have toxicologically detectible quantities of cocaine, opiates, cannabis, or alcohol than were elderly victims.

At least in part, the demographic composition of the elderly population of New York City during the period of the study may have influenced findings.14 For those aged 65 years or older, the 1990 US Census11 for New York City showed 1.6 times as many women as men and more than 3 times as many non-Hispanic Whites as non-Hispanic African Americans; the same comparisons yielded similar proportions in the 2000 US Census counts.10 However, annual age-, gender-, and race/ethnicity–specific homicide rates suggested that risk factors or vulnerabilities other than demographics alone underlay the increased presence of women and non-Hispanic Whites among elderly homicide victims in New York City. For example, although men, non-Hispanic African Americans, and Hispanics had higher homicide rates than did women and non-Hispanic Whites, respectively, in both elderly and nonelderly groups, the gender and racial/ethnic disparities were much less apparent in the elderly.

The more frequent postmortem evidence of illicit drug and alcohol use among nonelderly than among elderly homicide victims was unsurprising. However, that as many as 20% of the elderly victims had recently consumed alcohol, if not necessarily to excess, was a less expected finding. It is possible that continued use of alcohol among some elderly homicide victims contributed to their vulnerability to assault, because of ethanol-related slowing of reactions or alertness. However, for the interpretation of these data, it should be cautioned that the population of individuals at risk based on illicit drug use or alcohol consumption was unknown.

Comparisons With National and Other Regional Data

The characteristics of elderly homicide victims reported here are mostly in accord with extant reports. In the National Crime Victimization Survey,5 which recorded homicide data across the United States from 1992 to 1997, a period with which this New York City study overlapped, males and non-Hispanic African Americans also predominated in all age groups when annual homicide rates were used as the basis for comparison.

Moreover, in the National Crime Victimization Survey, as in the present New York City data, elderly homicide victims were more likely to have been killed in their own homes and were less likely than nonelderly victims to have been murdered with a firearm.5 In a Canadian study, older homicide victims were twice as likely as younger victims to be killed in their homes and were more likely than younger victims to have sustained physical injuries consistent with beating or strangling.6

Street Dispute Versus Home Vulnerability

We speculate that although younger adults may be stronger and more resilient in the event of injury than are the elderly, they are more likely to be in interpersonal disputes outside the home, e.g., at work or on the city streets, and they therefore have greater exposure to danger in these locations. By contrast, compared to nonelderly persons, the elderly, who are less mobile, are not as likely to be involved outside the home in conflicts over money, sex, or criminal activity, but they are more vulnerable to physical assault, which for them is more likely to occur at home and more likely to result in death.15^–17 Perpetrators may conclude that firearms are not needed to kill frail home-bound elderly victims, those who may be not be able to flee or to defend themselves. It is also possible that elderly women may be perceived by perpetrators as particularly vulnerable.

Thus, with aging, there may be a shift from a “street-dispute” model to a “home-vulnerability” model. The relatively lower rates of homicide among elderly non-Hispanic African Americans could reflect a reduction in time spent on the streets of the higher-crime areas in which many African Americans still live in New York City, a decline in the use of illicit drugs, or other factors.

Limitations

The principal limitation of this study was the unavailability of information on the perpetrators of the homicides. Victim characteristics are, in a sense, half of the story. Nationally, males committed the vast majority of the homicides of both elderly and younger victims in the years 1992 through 1997. The racial composition of perpetrators was more variable, with the percentage of non-Hispanic White perpetrators increasing with the age of the victims.5 In this study, we also lacked information on whether elderly homicide victims were killed by persons known or unknown to them. Because many of the geriatric homicides occurred in the victim’s own home, known perpetrators might be intimates such as relatives or others with access to the victim, such as healthcare attendants or domestic employees; unknown perpetrators would presumably be intruders in the course of a robbery or other crime.18

Another limitation is that these findings could, like homicide rates generally, be most relevant to a particular location and cohort.19 More recent data, when available, may reveal differences based upon declines in the New York City crime rate beginning during the period of the study, changes in the age distribution of illicit drug use, and demographic shifts such as an increase in the Hispanic population. Also, New York City had a higher rate of homicide for the elderly than pooled national data for the mid-1990s (4.75 per 100 000 population from 1990–1998 vs 3 per 100 000 population from 1992–1997,5 respectively). The reasons for New York City’s higher homicide rates among the elderly are unclear. However, despite overall declines in the late 1990s,20 the largest US cities still tended to have the highest homicide rates nationally for all ages.21 Although factors related to the urban setting or unique local conditions might have contributed to our findings, most of the characteristics and circumstances of death of elderly homicide victims reported here for New York City were in agreement with national trends for the period.5

The scope of New York City’s medical examiner files allowed for numbers of elderly homicide victims sufficient to make comparisons based on age but not large enough to examine secular trends. Elderly homicide is a relatively rare event, and case numbers may have been further reduced by a tendency not to perform autopsies on the elderly,7 women, or non-Hispanic Whites. Autopsy was in most cases used to corroborate the determination of homicide. Because toxicological analysis was done only as part of the autopsy procedure, the lower percentages of elderly, female, and White victims who underwent comprehensive toxicological analysis indirectly raises the possibility that ascertainment biases influenced the numbers of certified homicides in these groups.

Implications and Future Directions

Despite limitations, this study presents a complete composite description of elderly homicide victims in New York City for the period of the study. Further studies that cover wider time frames will help to expose trends as well as heterogeneity within the aggregate age groups used here.16

Additional studies are also needed to provide more detailed information on the risk factors for geriatric homicide.22 For example, many older persons live with a family care-giver or an unrelated home-health aide. Both situations involve the potential for elder abuse; the family caregiver may be an overstressed individual with economic gain to be had from killing the victim, and the home-health aide in New York City is typically a poorly paid immigrant with significant language and cultural differences from his or her employer. Because elderly homicide victims are frequently killed in their own homes, an understanding of the dynamics of relationships with various kinds of caregivers may offer clues to the risk factors for geriatric homicide.

These data on geriatric homicide were gathered to describe the characteristics and circumstances of the death of victims of the most extreme form of elder abuse and to distinguish them from younger adult victims. The findings underscore the vulnerability of some of society’s most fragile individuals, namely, elderly persons living in the community, whether alone or with a family or unrelated caregiver. Elderly women in particular appear to be at risk for victimization in their own homes. Improved accountability for their safety and well-being must involve a collaboration of adult protective, medical, social service, and law enforcement entities. Increased support for—and supervision of—caregivers, and greater attention to home security will also be required. In the broadest terms, it may be necessary to establish more consistent oversight of community-dwelling elderly persons than is presently available in most urban settings.