Abstract

Background The relationship between homicide and suicide has been studied extensively, but with conflicting results. The primary objective of this study was to examine the correlation between homicide and suicide rates in a large cross-sectional sample of UN member states.

Methods The study used age-standardized data on homicide and suicide for 65 international locales compiled by the World Health Organization. Weighted correlation coefficients between homicide and suicide rates were computed by sex, income level, and geographic region.

Results The overall correlation between homicide and suicide rates was weak and statistically insignificant (ρ = −0.08, P = 0.5178). However, when analysed by geographic region the data revealed two distinct patterns: homicide and suicide rates were positively correlated in European countries (ρ = 0.89, P < 0.0001), but negatively correlated in the Asia Pacific Region (ρ = −0.97, P < 0.0001), and the Americas (ρ = −0.62, P < 0.005).

Conclusions The strength and direction of the relationship between homicide and suicide vary significantly with geographic region. The divergent geographic patterns in the relationship between homicide and suicide might be due to regional differences in social and cultural variables.

The relationship between homicide and suicide as processes of killing has been studied extensively. Correlative research on homicide and suicide rates dates back to at least the early 1700s. Guerry1 considered the act of killing oneself and the crime of killing another as comparable phenomena. This conclusion was reached after he noticed higher rates of suicide in the north of France; while conversely, rates of homicide were higher in the south of France. Several other scholars24 extended geography as a determinant for understanding the relationship between homicide and suicide to other European countries.

In the 19th century scholars approached the topic through the lens of culture and society. Tissot5 perceived suicide as a direct response to larger societal issues of the time and provided little differentiation between the acts of homicide and suicide. According to Tissot5 homicide and suicide share a common aetiology (abnormal behaviour), result (death), and method (by one's own hands or by another's). Towards the end of the 19th century sociologists and social theorists furthered understanding on societal forces, precipitated in part by the work of Durkheim.3 While Durkheim's work focused primarily on suicide, his ideas and methods also had an affect on studies of homicide. Durkheim considered suicide to be a social fact and thus describable in terms of social factors. Durkheim thought suicide would be more common where social integration was weak. Understanding of suicide as a proportionality of one's social integration was further championed and expanded to include the topic of homicide by Henry and Short.6

Deviating from Durkheim's arguments, Henry and Short thought homicide and suicide causally related. Combining Durkheim's theories3 on suicide and social integration and Freud's theories7 on intrapsychic aggression, Henry and Short regarded homicide and suicide as forms of aggression predicated on measures of social integration. According to Henry and Short,6 aggression is a measure of the level of societal restraint over an individual's actions. Less restraint, e.g. more freedom, results in increased suicide and decreased homicide and, more control—forced conformation—would lead to high homicide and low suicide.

In conjunction with statistical data on the subjects, the idea that homicide and suicide be explained analogously in terms of aggression furthers the argument that the two are mutually associated. A similar position is expressed by Hentig:8 ‘murder and suicide are complimentary phenomena: the total amount of available destructiveness is discharged in two psychologically similar, socially distinct Gestalten’. The understanding that both homicide and suicide are antithetical to cultural norms and that each involves some element of self-destruction (with regard to both the killer and the individual killed)—there is some evidence that many homicides are victim-precipitated and represent suicides—has provided the stimulus to show that homicide and suicide are not unrelated and that one should consider them in aggregate, as a summation of deaths.9,10

Such a theory implies a positively correlated relationship, as evidenced by Holinger,9,11 who notes that rates of violent death tend to fluctuate in parallel; as over time, patterns of homicide and suicide rates in the US tend to be similar.12,13 While Holinger focused on rates of violent death in the US, others1416 have broadened the scope to include other countries; though findings have been contradictory.

In The Currents of Lethal Violence, Unnithan et al.17 revitalized the ideas of Henry and Short by arguing in favour of the stream analogy or integrated model: that homicide and suicide are alternate and causal forms of violent death depends on the forces of production (cultural and societal factors that indicate the amount of lethal violence) and direction (the cultural and societal factors driving the form of violence). Both Wu,18 in looking at regional rates of homicide and suicide in the US, and Batton19 in studying historical rates of death in the US, provide moderate support for the stream analogy theory, and the notion that homicide and suicide may share common social causes.

What is most apparent from texts on the subject is the diversity of opinions on the two forms of violent death, and though the seemingly disparate views may appear antithetical, they each point to a component of the relationship or possible connection between homicide and suicide. The purpose of this study is to approach theory by correlating contemporary rates of mortality from homicide and suicide from 65 countries and areas. The relationship between the two forms of violent death is further examined by geographic region, per capita income level, and the sex of those killed.

Methods

Age-adjusted homicide and suicide mortality data from 65 separate countries and regions of the world, collected between the years 1990 and 1999 and compiled by the World Health Organization (WHO) in the World Report on Violence and Health, were used to assess the relationship between homicide and suicide rates. Age-adjusted rates allow for comparison among countries with different age compositions and were calculated by relating age-specific rates to the World Standard Population.20 Definitions of homicide and suicide are based on those provided by the WHO. The WHO classifies death according to the International Classification of Diseases. Though initial classification of death was made by the individual member country, the WHO has paid particular attention to misreporting. The codes for homicide and suicide are, ICD-9 E960-E969 and ICD-9 E950-E959, respectively.20

Both the number and rate (per 100 000 population) of death due to suicide and homicide were compiled according to country and further categorized by sex (male and female), geographic region (Americas, Europe, and Asia Pacific) and income level (high and low–middle) (Table 1). Data from Kuwait and Israel (Middle East) and Mauritius (Africa) were excluded from this study because the small sample sizes precluded any meaningful regional analysis. Also excluded from the analysis were data for six Central Asian nations (Azerbaijan, Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan, and Uzbekistan). These Central Asian nations were grouped in the European region by WHO, but geographically are not part of Europe. A country's income level is based on estimates (1996) of gross national product (GNP) per capita (now referred to as gross national income), as initially compiled by the World Bank, and reported in the WHO report cited above. The GNP per capita of high-income countries is defined as US $9636 or more, while low and middle income countries are defined by a GNP per capita of less than US $9636.

Table 1

Age-adjusted rates of homicide and suicide per 100000 population in 65 international locales, as defined by region, sex, and income level

   Homicide
 
     Suicide
 
     
   Total
 
 Male
 
 Females
 
 Total
 
 Male
 
 Female
 
 
Country
 
Year
 
Income level
 
No. of deaths
 
Death rate
 
No. of deaths
 
Death rate
 
No. of deaths
 
Death rate
 
No. of deaths
 
Death rate
 
No of deaths
 
Death rate
 
No. of deaths
 
Death rate
 
Americas               
    Argentina 1996 Low–medium 1611 4.7 1347 8.1 264 1.5 2245 6.5 1709 10.6 536 3.0 
    Brazil 1995 Low–medium 37076 23.0 33751 42.5 3325 4.1 6584 4.7 5174 7.6 1410 1.9 
    Canada 1997 High 431 1.4 285 1.9 146 1.0 3681 11.3 2914 18.1 767 4.6 
    Chile 1994 Low–medium 410 3.0 356 5.4 54 0.8 801 6.1 704 11.1 97 1.4 
    Colombia 1995 Low–medium 23443 61.6 21705 116.8 1738 9.0 1172 3.4 905 5.5 267 1.4 
    Costa Rica 1995 Low–medium 179 5.4 153 9.3 26 1.4 211 6.6 174 10.9 37 2.3 
    Cuba 1997 Low–medium 747 6.2 584 9.6 163 2.7 2029 17.1 1401 23.8 628 10.6 
    Ecuador 1996 Low–medium 1632 15.3 1501 28.2 131 2.5 593 5.5 396 7.8 197 3.2 
    El Salvador 1993 Low–medium 2480 55.6 2290 108.4 190 8.4 429 8.5 276 12.1 153 5.3 
    Guyana 1994–1996 Low–medium 42 6.6 35 11.8 2.0 84 12.5 64 20.5 20 5.6 
    Mexico 1997 Low–medium 13542 15.9 12170 29.6 1372 3.1 3369 3.9 2828 6.9 541 1.1 
    Nicaragua 1996 Low–medium 285 8.4 246 15.1 39 2.2 230 5.9 147 8.4 83 3.5 
    Panama (excluding Canal Zone) 1997 Low–medium 293 10.9 265 19.8 28 2.0 145 5.8 124 9.8 21 1.7 
    Paraguay 1994 Low–medium 459 12.6 420 23.4 39 2.2 109 3.2 82 5.1 27 1.5 
    Puerto Rico 1998 High 804 20.6 731 38.1 73 3.7 321 8.1 290 15.6 31 1.5 
    Trinidad and Tobago 1994 Low–medium 146 12.1 108 17.1 38 6.6 148 12.6 118 19.5 30 5.0 
    United States of America 1998 High 17893 6.9 13652 10.7 4241 3.1 30575 10.4 24538 17.3 6037 4.0 
    Uruguay 1990 Low–medium 136 4.4 105 7.1 31 1.9 318 9.6 251 16.4 67 3.7 
    Venezuela 1994 Low–medium 3353 16.0 3120 29.7 233 2.3 1089 6.1 890 10.3 199 2.1 
Europe               
    Albania 1998 Low–medium 660 21.0 573 38.7 87 5.3 165 5.3 104 7.1 61 3.6 
    Armenia 1999 Low–medium 98 2.6 77 4.2 21 1.0 67 1.7 49 2.6 18 0.9 
    Austria 1999 High 68 0.8 32 0.8 36 0.8 1555 15.5 1126 24.2 429 7.6 
    Belarus 1999 Low–medium 1123 10.5 784 15.6 339 5.8 3408 30.9 2877 56.9 531 8.5 
    Belgium 1995 High 169 1.6 100 1.9 69 1.2 2155 17.9 1550 27.1 605 9.4 
    Bulgaria 1999 Low–medium 238 2.6 174 4.0 64 1.3 1307 12.3 965 19.6 342 5.8 
    Croatia 1999 Low–medium 128 2.6 95 4.0 33 1.2 989 18.5 716 30.2 273 8.6 
    Czech Republic 1999 Low–medium 151 1.4 97 1.8 54 1.0 1610 13 1285 22.3 325 4.7 
    Denmark 1996 High 59 1.1 36 1.4 23 0.8 892 13.6 631 20.2 261 7.5 
    Estonia 1999 Low–medium 227 14.8 168 23.1 59 7.4 469 28.1 376 50.9 93 8.9 
    Finland 1998 High 125 2.2 90 3.3 35 1.2 1228 21.1 962 34.0 266 8.8 
    France 1998 High 436 0.7 269 0.9 167 0.5 10534 14.8 7771 23.2 2763 7.4 
    Germany 1999 High 720 0.9 418 1.0 302 0.7 11 160 10.6 8082 16.7 3078 5.1 
    Greece 1998 High 144 1.2 109 1.9 35 0.5 403 3.1 315 4.9 88 1.4 
    Hungary 1999 Low–medium 291 2.6 170 3.2 121 2.0 3328 26.9 2550 45.7 778 10.7 
    Ireland 1997 High 30 0.8 21 1.2 0.5 466 12.5 376 20.4 90 4.7 
    Italy 1997 High 720 1.1 561 1.8 159 0.5 4694 6.2 3547 9.9 1147 2.9 
    Latvia 1999 Low–medium 308 11.6 213 17.3 95 6.6 764 27.0 593 47.1 171 10.0 
    Lithuania 1999 Low–medium 297 7.5 209 11.3 88 1552 38.4 1287 69.2 265 11.2 
    Macedonia 1997 Low–medium 47 2.2 38 3.7 0.9 155 7.4 115 11.3 40 3.8 
    Moldova 1999 Low–medium 410 11.2 307 18.0 103 5.2 579 15.5 482 28.1 97 4.7 
    The Netherlands 1999 High 203 1.3 144 1.7 59 0.8 1517 8.3 1015 11.3 502 5.4 
    Norway 1997 High 41 0.9 28 1.3 13 0.5 533 10.9 387 16.1 146 5.9 
    Poland 1995 Low–medium 1088 2.7 785 4.0 303 1.4 5499 13.4 4562 23.2 937 4.2 
    Portugal 1999 High 118 1.1 82 1.6 36 0.7 545 4.0 407 6.7 138 1.9 
    Romania 1999 Low–medium 803 3.3 572 4.8 231 1.8 2736 10.8 2235 18.5 501 3.6 
    Russian Federation 1998 Low–medium 33553 21.6 25130 34.0 8423 9.8 51770 32.1 42785 57.9 8985 9.4 
    Slovakia 1999 Low–medium 132 2.3 89 3.2 43 1.4 692 11.5 590 20.8 102 3.2 
    Slovenia 1999 Low–medium 30 1.3 20 1.8 10 0.8 590 24.6 453 40.3 137 10.6 
    Spain 1998 High 355 0.8 255 1.2 100 0.4 3261 6.5 2499 10.5 762 2.8 
    Sweden 1996 High 110 1.2 74 1.5 36 0.8 1253 11.8 872 16.9 381 6.9 
    Switzerland 1996 High 77 1.1 48 1.4 29 0.8 1431 16.7 1010 25.1 421 9.1 
    Ukraine 1999 Low–medium 6260 11.7 4421 17.8 1839 6.1 14452 25.2 11806 46 2646 7.6 
    United Kingdom 1999 High 440 0.8 335 1.2 105 0.4 4448 6.8 3443 10.8 1005 2.9 
    England and Wales 1999 High 295 0.6 214 0.9 81 0.3 3690 6.3 2840 9.9 850 2.7 
    Northern Ireland 1999 High 24 1.4 20 2.4 0.5 121 7.3 103 12.5 18 2.2 
    Scotland 1999 High 121 2.4 101 4.1 20 0.7 637 11.7 500 18.8 137 4.7 
Asia Pacific               
    Australia 1998 High 295 1.6 201 2.1 94 2633 13.3 2108 21.4 525 5.2 
    China (Hong Kong SAR) 1996 Low–medium 63 1.0 39 1.3 24 0.8 788 11.2 501 14.6 287 7.9 
    China (selected areas) 1999 Low–medium 2405 1.8 1655 2.5 750 1.2 16836 13.7 8048 13.5 8788 14.0 
    Japan 1997 High 719 0.6 435 0.7 284 0.4 23502 14.5 15906 20.7 7596 8.5 
    New Zealand 1998 High 57 1.5 32 1.7 25 1.3 574 15.0 442 23.6 132 6.8 
    Philippines 1993 Low–medium 7726 14.2 7181 26.4 545 2.1 851 1.5 509 1.9 342 1.2 
    Republic of Korea 1997 Low–medium 987 2.0 602 2.4 385 1.6 6024 12.8 4162 18.8 1862 7.6 
    Singapore 1998 High 45 1.3 29 1.7 16 1.0 371 11.7 221 14.1 150 9.4 
    Thailand 1994 Low–medium 4161 7.5 3481 12.6 680 2.5 2333 4.1 1631 5.9 702 2.4 
   Homicide
 
     Suicide
 
     
   Total
 
 Male
 
 Females
 
 Total
 
 Male
 
 Female
 
 
Country
 
Year
 
Income level
 
No. of deaths
 
Death rate
 
No. of deaths
 
Death rate
 
No. of deaths
 
Death rate
 
No. of deaths
 
Death rate
 
No of deaths
 
Death rate
 
No. of deaths
 
Death rate
 
Americas               
    Argentina 1996 Low–medium 1611 4.7 1347 8.1 264 1.5 2245 6.5 1709 10.6 536 3.0 
    Brazil 1995 Low–medium 37076 23.0 33751 42.5 3325 4.1 6584 4.7 5174 7.6 1410 1.9 
    Canada 1997 High 431 1.4 285 1.9 146 1.0 3681 11.3 2914 18.1 767 4.6 
    Chile 1994 Low–medium 410 3.0 356 5.4 54 0.8 801 6.1 704 11.1 97 1.4 
    Colombia 1995 Low–medium 23443 61.6 21705 116.8 1738 9.0 1172 3.4 905 5.5 267 1.4 
    Costa Rica 1995 Low–medium 179 5.4 153 9.3 26 1.4 211 6.6 174 10.9 37 2.3 
    Cuba 1997 Low–medium 747 6.2 584 9.6 163 2.7 2029 17.1 1401 23.8 628 10.6 
    Ecuador 1996 Low–medium 1632 15.3 1501 28.2 131 2.5 593 5.5 396 7.8 197 3.2 
    El Salvador 1993 Low–medium 2480 55.6 2290 108.4 190 8.4 429 8.5 276 12.1 153 5.3 
    Guyana 1994–1996 Low–medium 42 6.6 35 11.8 2.0 84 12.5 64 20.5 20 5.6 
    Mexico 1997 Low–medium 13542 15.9 12170 29.6 1372 3.1 3369 3.9 2828 6.9 541 1.1 
    Nicaragua 1996 Low–medium 285 8.4 246 15.1 39 2.2 230 5.9 147 8.4 83 3.5 
    Panama (excluding Canal Zone) 1997 Low–medium 293 10.9 265 19.8 28 2.0 145 5.8 124 9.8 21 1.7 
    Paraguay 1994 Low–medium 459 12.6 420 23.4 39 2.2 109 3.2 82 5.1 27 1.5 
    Puerto Rico 1998 High 804 20.6 731 38.1 73 3.7 321 8.1 290 15.6 31 1.5 
    Trinidad and Tobago 1994 Low–medium 146 12.1 108 17.1 38 6.6 148 12.6 118 19.5 30 5.0 
    United States of America 1998 High 17893 6.9 13652 10.7 4241 3.1 30575 10.4 24538 17.3 6037 4.0 
    Uruguay 1990 Low–medium 136 4.4 105 7.1 31 1.9 318 9.6 251 16.4 67 3.7 
    Venezuela 1994 Low–medium 3353 16.0 3120 29.7 233 2.3 1089 6.1 890 10.3 199 2.1 
Europe               
    Albania 1998 Low–medium 660 21.0 573 38.7 87 5.3 165 5.3 104 7.1 61 3.6 
    Armenia 1999 Low–medium 98 2.6 77 4.2 21 1.0 67 1.7 49 2.6 18 0.9 
    Austria 1999 High 68 0.8 32 0.8 36 0.8 1555 15.5 1126 24.2 429 7.6 
    Belarus 1999 Low–medium 1123 10.5 784 15.6 339 5.8 3408 30.9 2877 56.9 531 8.5 
    Belgium 1995 High 169 1.6 100 1.9 69 1.2 2155 17.9 1550 27.1 605 9.4 
    Bulgaria 1999 Low–medium 238 2.6 174 4.0 64 1.3 1307 12.3 965 19.6 342 5.8 
    Croatia 1999 Low–medium 128 2.6 95 4.0 33 1.2 989 18.5 716 30.2 273 8.6 
    Czech Republic 1999 Low–medium 151 1.4 97 1.8 54 1.0 1610 13 1285 22.3 325 4.7 
    Denmark 1996 High 59 1.1 36 1.4 23 0.8 892 13.6 631 20.2 261 7.5 
    Estonia 1999 Low–medium 227 14.8 168 23.1 59 7.4 469 28.1 376 50.9 93 8.9 
    Finland 1998 High 125 2.2 90 3.3 35 1.2 1228 21.1 962 34.0 266 8.8 
    France 1998 High 436 0.7 269 0.9 167 0.5 10534 14.8 7771 23.2 2763 7.4 
    Germany 1999 High 720 0.9 418 1.0 302 0.7 11 160 10.6 8082 16.7 3078 5.1 
    Greece 1998 High 144 1.2 109 1.9 35 0.5 403 3.1 315 4.9 88 1.4 
    Hungary 1999 Low–medium 291 2.6 170 3.2 121 2.0 3328 26.9 2550 45.7 778 10.7 
    Ireland 1997 High 30 0.8 21 1.2 0.5 466 12.5 376 20.4 90 4.7 
    Italy 1997 High 720 1.1 561 1.8 159 0.5 4694 6.2 3547 9.9 1147 2.9 
    Latvia 1999 Low–medium 308 11.6 213 17.3 95 6.6 764 27.0 593 47.1 171 10.0 
    Lithuania 1999 Low–medium 297 7.5 209 11.3 88 1552 38.4 1287 69.2 265 11.2 
    Macedonia 1997 Low–medium 47 2.2 38 3.7 0.9 155 7.4 115 11.3 40 3.8 
    Moldova 1999 Low–medium 410 11.2 307 18.0 103 5.2 579 15.5 482 28.1 97 4.7 
    The Netherlands 1999 High 203 1.3 144 1.7 59 0.8 1517 8.3 1015 11.3 502 5.4 
    Norway 1997 High 41 0.9 28 1.3 13 0.5 533 10.9 387 16.1 146 5.9 
    Poland 1995 Low–medium 1088 2.7 785 4.0 303 1.4 5499 13.4 4562 23.2 937 4.2 
    Portugal 1999 High 118 1.1 82 1.6 36 0.7 545 4.0 407 6.7 138 1.9 
    Romania 1999 Low–medium 803 3.3 572 4.8 231 1.8 2736 10.8 2235 18.5 501 3.6 
    Russian Federation 1998 Low–medium 33553 21.6 25130 34.0 8423 9.8 51770 32.1 42785 57.9 8985 9.4 
    Slovakia 1999 Low–medium 132 2.3 89 3.2 43 1.4 692 11.5 590 20.8 102 3.2 
    Slovenia 1999 Low–medium 30 1.3 20 1.8 10 0.8 590 24.6 453 40.3 137 10.6 
    Spain 1998 High 355 0.8 255 1.2 100 0.4 3261 6.5 2499 10.5 762 2.8 
    Sweden 1996 High 110 1.2 74 1.5 36 0.8 1253 11.8 872 16.9 381 6.9 
    Switzerland 1996 High 77 1.1 48 1.4 29 0.8 1431 16.7 1010 25.1 421 9.1 
    Ukraine 1999 Low–medium 6260 11.7 4421 17.8 1839 6.1 14452 25.2 11806 46 2646 7.6 
    United Kingdom 1999 High 440 0.8 335 1.2 105 0.4 4448 6.8 3443 10.8 1005 2.9 
    England and Wales 1999 High 295 0.6 214 0.9 81 0.3 3690 6.3 2840 9.9 850 2.7 
    Northern Ireland 1999 High 24 1.4 20 2.4 0.5 121 7.3 103 12.5 18 2.2 
    Scotland 1999 High 121 2.4 101 4.1 20 0.7 637 11.7 500 18.8 137 4.7 
Asia Pacific               
    Australia 1998 High 295 1.6 201 2.1 94 2633 13.3 2108 21.4 525 5.2 
    China (Hong Kong SAR) 1996 Low–medium 63 1.0 39 1.3 24 0.8 788 11.2 501 14.6 287 7.9 
    China (selected areas) 1999 Low–medium 2405 1.8 1655 2.5 750 1.2 16836 13.7 8048 13.5 8788 14.0 
    Japan 1997 High 719 0.6 435 0.7 284 0.4 23502 14.5 15906 20.7 7596 8.5 
    New Zealand 1998 High 57 1.5 32 1.7 25 1.3 574 15.0 442 23.6 132 6.8 
    Philippines 1993 Low–medium 7726 14.2 7181 26.4 545 2.1 851 1.5 509 1.9 342 1.2 
    Republic of Korea 1997 Low–medium 987 2.0 602 2.4 385 1.6 6024 12.8 4162 18.8 1862 7.6 
    Singapore 1998 High 45 1.3 29 1.7 16 1.0 371 11.7 221 14.1 150 9.4 
    Thailand 1994 Low–medium 4161 7.5 3481 12.6 680 2.5 2333 4.1 1631 5.9 702 2.4 

The strength of the relationship between rates of homicide and suicide was measured by the Pearson product-moment correlation coefficient (ρ) weighted by the population size for a given rate.21 The formula for the weighted Pearson product-moment correlation coefficient is as follows:  

\[{\tau}_{xy}\ =\ \frac{{\sum}w_{i}\left(x_{i}\ {-}\ {\bar{x}}_{w}\right)\ \left(y_{i}\ {-}\ {\bar{y}}_{w}\right)}{\sqrt{{\sum}w_{i}\left(x_{i}\ {-}\ {\bar{x}}_{w}\right)^{2}\ {\sum}w_{i}\left(y_{i}\ {-}\ {\bar{y}}_{w}\right)^{2}}}\]
where,  
\[{\bar{x}}_{w}\ =\ {\sum}w_{i}x_{i}/{\sum}w_{i}\]
 
\[{\bar{y}}_{w}\ =\ {\sum}w_{i}y_{i}/{\sum}w_{i}\]
xi, yi, and wi denote the homicide rate, suicide rate, and the population, respectively for each country. The weighted correlation coefficients of homicide and suicide were computed by stratification of sex, income, and geographic region.

Graphical depictions of rates of homicide and suicide for each of the categories listed above were prepared for visual inspection. Such graphs were used as a way of assessing general patterns in data, and to identify potential outliers.

Results

When all countries were assessed together there was a very weak negative correlation between homicide and suicide rates (ρ = −0.08, P = 0.5178). This weak and statistically insignificant correlation between homicide and suicide was virtually unchanged when stratified by sex or income level (Table 2). However, when the data were analysed by geographic region homicide and suicide rates correlated closely and positively in Europe (ρ = 0.89, P < 0.0001; Figure 1), but negatively in the Americas (ρ = −0.62, P = 0.0048; Figure 2) and the Asia Pacific (ρ = −0.97, P < 0.0001; Figure 3). The correlation between homicide and suicide rates in each of the three geographic regions was more pronounced among males than among females. The effects of per capita income level on the relationship between homicide and suicide depended on geographic region (Table 3).

Figure 1

Age-adjusted homicide and suicide rates in Europe

Figure 1

Age-adjusted homicide and suicide rates in Europe

Figure 2

Age-adjusted homicide and suicide rates in the Americas

Figure 2

Age-adjusted homicide and suicide rates in the Americas

Figure 3

Age-adjusted homicide and suicide rates in the Asia Pacific

Figure 3

Age-adjusted homicide and suicide rates in the Asia Pacific

Table 2

Weighted correlation coefficients (ρ) between homicide and suicide rates for different geographic regions, sexes, and income levels


 
Weighted ρ
 
P-value
 
Region   
    Americas −0.61785 0.0048 
    Europe 0.88889 <0.0001 
    Asia Pacific −0.96895 <0.0001 
Sex   
    Male 0.04767 0.7061 
    Female −0.22715 0.0688 
Per capita income   
    Low–medium −0.13132 0.4319 
    High −0.06465 0.7487 
Overall −0.08167 0.5178 

 
Weighted ρ
 
P-value
 
Region   
    Americas −0.61785 0.0048 
    Europe 0.88889 <0.0001 
    Asia Pacific −0.96895 <0.0001 
Sex   
    Male 0.04767 0.7061 
    Female −0.22715 0.0688 
Per capita income   
    Low–medium −0.13132 0.4319 
    High −0.06465 0.7487 
Overall −0.08167 0.5178 
Table 3

Weighted correlation coefficients (ρ) between homicide and suicide rates by geographic region, sex, and income level

 Male
 
 Female
 
 Both sexes
 
 
Region
 
Weighted ρ
 
P-value
 
Weighted ρ
 
P-value
 
Weighted ρ
 
P-value
 
Americas       
    Low–medium income level −0.43864 0.0892 −0.12656 0.6405 −0.35661 0.1752 
    High income level −0.98332 0.1164 −0.64675 0.5523 −0.99992 0.0078 
    Both income levels −0.66085 0.0021 −0.27616 0.2524 −0.61785 0.0048 
Europe       
    Low–medium income level 0.81832 <0.0001 0.78263 <0.0001 0.83466 <0.0001 
    High income level −0.04323 0.8605 0.66601 0.0019 0.20044 0.4106 
    Both income levels 0.89145 <0.0001 0.74089 <0.0001 0.88889 <0.0001 
Asia Pacific       
    Low–medium income level −0.98435 0.0156 −0.95340 0.0466 −0.97010 0.0299 
    High income level −0.48313 0.4097 −0.51485 0.3747 −0.90165 0.0365 
    Both income levels −0.83310 0.0053 −0.49567 0.1748 −0.96895 <0.0001 
 Male
 
 Female
 
 Both sexes
 
 
Region
 
Weighted ρ
 
P-value
 
Weighted ρ
 
P-value
 
Weighted ρ
 
P-value
 
Americas       
    Low–medium income level −0.43864 0.0892 −0.12656 0.6405 −0.35661 0.1752 
    High income level −0.98332 0.1164 −0.64675 0.5523 −0.99992 0.0078 
    Both income levels −0.66085 0.0021 −0.27616 0.2524 −0.61785 0.0048 
Europe       
    Low–medium income level 0.81832 <0.0001 0.78263 <0.0001 0.83466 <0.0001 
    High income level −0.04323 0.8605 0.66601 0.0019 0.20044 0.4106 
    Both income levels 0.89145 <0.0001 0.74089 <0.0001 0.88889 <0.0001 
Asia Pacific       
    Low–medium income level −0.98435 0.0156 −0.95340 0.0466 −0.97010 0.0299 
    High income level −0.48313 0.4097 −0.51485 0.3747 −0.90165 0.0365 
    Both income levels −0.83310 0.0053 −0.49567 0.1748 −0.96895 <0.0001 

The strong geographic effects could not be explained by exceptional values in the datasets. Excluding Colombia and El Salvador increased the correlation coefficient between homicide and suicide rates in the Americas from −0.62 to −0.78 (P = 0.0002). And excluding the Philippines did not affect the strength of the relationship between homicide and suicide in the Asia Pacific (ρ = −0.98, P < 0.0001).

Discussion

Our analysis indicates, overall, the correlation between homicide and suicide rates across all nations is very weak and statistically insignificant. Two opposing patterns, a positive correlation in European countries and a negative correlation in the countries of the Asia Pacific and the Americas compound this weak correlation. Furthermore, the correlation between homicide and suicide rates in the Americas, Europe, and Asia Pacific is more pronounced among males than among females.

In looking at the degree of difference and diversity among the countries selected, the lack of a strong correlation between rates of homicide and suicide is not surprising. A similarly weak correlation between international rates of homicide and suicide was also reported by Unnithan et al. (ρ = −0.10)17 and Lester (ρ = 0.15).14 Although such findings do not provide evidence for a strong correlation between homicide and suicide rates, they do not necessarily negate the possibility that a significant relationship does exist in certain geographic regions and among population groups.

In the present study the lack of a correlation between suicide and homicide rates for all nations resulted from divergent regional patterns. Region, as a defining characteristic, has a profound impact on the correlation between rates of homicide and suicide.18,22 The relationship between rates of homicide and suicide in each of the three regions studied—Europe, the Americas, and the Asia Pacific—is distinct. A positive correlation in Europe, and conversely, negative correlations among rates in the Asia Pacific and the Americas may be indicative of larger differences of culture, values, and of attitudes towards life and death.23 While the correlation of homicide and suicide appears to be stronger in the Asia Pacific and Europe regions than in the Americas, the variation may be a result of a less-historically solidified culture of the Americas. Although the measuring and quantification of culture or cultural components based on large geographic regions such as Europe or the Americas is difficult, a general comparison, or acknowledgement of cultural and societal difference is worthwhile. Thus cultural differences as expressed through regional distinction may account for differing signs and strengths of the correlation. Bohannan2 stressed the cultural components of the relationship between homicide and suicide, and in turn alluded to the problems of using a single sociological model to define relational rates of violent death. The strong regional effects on correlations of homicide and suicide give credence to such critiques. Analysis of rates of homicide and suicide in other regions, such as Africa, Southeast Asia, and the Eastern Mediterranean would be useful, but was not possible in this study, as sufficient data were not available.

Per capita income level has been recognized as a significant determinant of homicide and suicide rates,14,17,18,24 but it does not seem to play an independent role in the relationship between homicide and suicide rates. Per capita income level, however, appears to have a significant interaction effect with geographic region on the relationship between homicide and suicide. Specifically, the correlation between homicide and suicide was stronger among low–medium income countries than among high income countries in Europe and Asia Pacific regions and weaker in the Americas.

In general, males die as a result of homicide and suicide at far greater rates than females. Verkko4 noted that where homicide rates were higher there were more male victims involved. Yet, sex appeared to somewhat modify regional patterns in the relationship between homicide and suicide. Specifically, the correlation appears to be more pronounced among males than among females in all three regions. Differences in the correlative strength may owe in part to the much greater death rates among males.

Comparisons of individual country rates to larger patterns in regional data signify potential outliers. El Salvador, Colombia, and the Philippines are each indicative of a pattern of a high rate of homicide, coupled with a low rate of suicide. When rates of violent death in El Salvador and Colombia are compared with data from other countries in the Americas the differences are quite significant. The exceptionally high rates of homicide seen in El Salvador and Colombia are conceived, by Desjarlais et al.25 as the ‘unhinging of social relations [where] a general atmosphere of tension and mistrust is the rule’. In the Philippines high homicide rates and low suicide rates differ widely with other countries in the Asia Pacific Region. The inverse correlation—of high rates of homicide and correspondingly low rates of suicide—is a well-documented trend in times of conflict or war, as noted by Henry and Short,6 Marshall,26 and Archer and Gartner.27

However, alternate effects of war and conflict on suicide have also been reported. Desjarlais et al. point to the profound effects of violence and social upheaval on rates of suicide in Sri Lanka. In 1991, Sri Lanka, also in the midst of a civil war, reported higher levels of suicide than any other country.25 Pinguet,28 who calls attention to the increased numbers of suicides in Japan during and just preceding Second World War, also shares such a perspective. Although Japan is widely referred to as a ‘suicide nation’, current rates of suicide in Japan are similar to European countries.28 The complex relationship between war and suicide may reflect in part the varying effects of conflict on peoples' attitudes towards death and dying in different societies and cultures.

It is noteworthy that the correlation coefficient measures the linear relationship only. Non-linear relationships between homicide and suicide were not rigorously probed in this study. Scattered plots of homicide vs suicide rates did not exhibit any apparent non-linear relationships between the two. The current study was also limited to three control variables (sex, region, and income). Age and race, for example, have been correlated with rates of both homicide and suicide, though the impact of both age and race on the relationship between homicide and suicide rates is not conclusive.2931 Future studies including a larger number of social and cultural variables are needed to further understand regional patterns in the relationship between homicide and suicide rates.

Initial classification of reported death rates was made by the individual UN member country. Definitions, collection, reporting, and understanding of suicide and homicide vary from country to country. Suicide may be highly stigmatized, troubling to survivors, and even illegal in some countries. Official suicide statistics, for example, may underestimate the problem by 30–200%.32 Similarly, homicide rates may often be underreported as higher rates may be a symbol of instability. Thus, the validity and reliability of country-level rates of homicide and suicide rates are questionable.14,25,33 Both homicide and suicide as, ‘an aggregate of acts that had widely different meanings for the very varied people who took their lives’, is an important point, and should be taken into account when making comparisons between multiple countries.34 Field research can help to minimize the problems associated with official statistics.

Although the relationship between homicide and suicide has been studied for centuries, no one theory has emerged as the uniform framework for understanding homicide and suicide. Most contemporary understandings of violent death assume some kind of common link or connection between homicide and suicide, and support for such a conclusion is also presented here. The most developed theory explaining patterns of homicide and suicide rates is that of the integrated model or stream-analogy.6,17 Though the theoretical framework put forth by Unnithan et al. is a good starting point it does not provide an adequate basis for understanding differences in the relationship between homicide and suicide across geographic regions and between sexes. Furthermore, while Unnithan et al. stress the effects of inequality and economic development on rates of homicide and suicide, region, rather than income level, is the far greater determinant for understanding patterns of the relationship between rates of violent death in the present study. While rates of violent death may be higher in low–medium income countries, overall, the relationship between homicide and suicide rates among high and low–medium income does not differ significantly. Such a finding should not be seen as conflicting, but rather as an indication that several factors affect rates of homicide and suicide. Such intricacy, moreover, implies the need for a more culturally nuanced understanding of the motivations and intentions behind violent death. Again, individual country, or sub-population, analyses are needed, and would help to shape the understanding of the regional patterns seen in this study.

While suicide was originally defined in relation to homicide, most current theoretical understandings of homicide arise from theories on suicide. This continual convolution of definitions and theories makes it difficult to sort out the relationship between the two forms of death. To further understand patterns of death, homicide and suicide rates need to be addressed within a uniform framework of social, cultural, economical, political, and environmental determinants. It is important to construct the relationship between violent death and the various determinants as mutually causative given that homicide and suicide can have a profound impact on the stability of a culture and its ability to cultivate and maintain meaning and identity.

Disclaimer

The researchers' views were in no way influenced by the funding agencies; similarly, the views of the researchers do not necessarily reflect the ideas of the funding agencies.

KEY MESSAGES

  • Homicide and suicide are theorized as alternate and causal forms of violence.

  • Direction and strength of the relationship between homicide and suicide rates vary with geographic regions.

  • Homicide and suicide rates are positively correlated in European countries and inversely correlated in the Americas and Asia Pacific.

  • Cultural and socioeconomic differences may have contributed to the divergent geographic patterns.

We thank Yandong Qiang for her statistical assistance, and Drs Timothy Baker, Susan Baker, and Yandong Qiang, as well as Rachel Cohen for their constructive comments. This research was supported in part by Grants R01AA09963 and R01AG1364 from the National Institutes of Health, and Grant CCR302486 from the Centers for Disease Control and Prevention.

References

1
Guerry A. Essai sur la statistique morale de la France. Paris: Crochard,
1833
.
2
Bohannan C. African Homicide and Suicide. Princeton, NJ: Princeton University Press,
1966
.
3
Durkheim E. Suicide: A study in Sociology (trans. Spaulding J, Simpson G). Glencoe, IL: Free Press,
1951
.
4
Verkko V. Homicide and Suicide in Finland. Copenhagen: C.E.S. Gads Verlag,
1951
.
5
Tissot J. De la manie du suicide et de l'esprit de revolte, de leurs causes et de leurs remèdes. Paris: Ladrange,
1840
.
6
Henry A, Short J. Suicide and Homicide: Some Economic, Sociological and Psychological Aspects of Aggression. London: The Free Press of Glencoe, Collier-Macmillan Ltd,
1954
.
7
Freud S. The Ego and the Id (trans. Riviere J). London: The Hogarth Press,
1947
.
8
Hentig H. The Criminal and His Victim. New Haven: Yale University Press,
1948
.
9
Holinger P, Klemen E. Violent deaths in the United States, 1900–1975. Soc Sci Med
1982
;
16
:
1929
–38.
10
Wolfgang M. Patterns in Criminal Homicide. Philadelphia: University of Pennsylvania Press,
1958
.
11
Holinger P. Violent deaths as a leading cause of mortality: an epidemiological study of suicide, homicide and accidents.
Psychiatry
 
1980
;
137
:
472
–76.
12
Brenner M. Times Series Analysis of Relationships Between Selected Economic and Social Indicators. Springfield, VA: National Technical Information Service,
1971
.
13
Klebba A, Dolman A. Comparability of mortality statistics for the seventh and eighth revisions of the international classification of diseases, United States.
Vital Health Stat 1
 
1975
;
2
:
1
–93.
14
Lester D. Patterns of Homicide and Suicide in the World. Commack, NY: Nova Science Publishers, Inc.,
1996
.
15
Reza A, Mercy J, Krug E. Epidemiology of violent deaths in the world.
Inj Prev
 
2001
;
7
:
104
–11.
16
Palmer S. Characteristics of suicide in 54 nonliterate societies.
Suicide Life Threat Behav
 
1971
;
1
:
178
–83.
17
Unnithan P, Corzine J, Huff-Corzine L, Whitt H. The Currents of Lethal Violence: An Integrated Model of Suicide and Homicide. Albany, NY: State University of New York Press,
1994
.
18
Wu B. Testing the stream analogy for lethal violence: A macro study of suicide and homicide.
Western Criminology Review
 
2003
;
4
:
215
–25.
19
Batton C. The Stream Analogy: A Historical Study of Lethal Violence Rates from the Perspective of the Integrated Homicide-Suicide Model. Dissertation. Nashville: Vanderbilt University,
1999
.
20
World Health Organization. World Report on Violence and Health. Geneva: WHO,
2002
.
21
Peers I. Statistical Analysis for Education and Psychology Researchers. Washington, DC: The Falmer Press,
1996
.
22
Lester D. Regional variation in suicide and homicide.
Suicide Life Threat Behav
 
1985
;
15
:
110
–16.
23
Rosenfeld R, Messner S. The social sources of homicide in different types of societies.
Sociological Forum
 
1991
;
6
:
51
–70.
24
Quinney R. Suicide, homicide, and economic development.
Soc Forces
 
1965
;
43
:
401
–06.
25
Desjarlais R, Eisenberg L, Good B, Kleinman A. World Mental Health: Problems and Priorities in Low-Income Countries. New York: Oxford University Press,
1995
.
26
Marshall J. Political integration and the effect of war on suicide: United States, 1933–76.
Soc Forces
 
1981
;
59
:
771
–85.
27
Archer D, Gartner R. Violent acts and violent times: a comparative approach to postwar homicide rates.
Am Sociol Rev
 
1976
;
41
:
937
–63.
28
Pinguet M. Voluntary Death in Japan. (trans. Morris R) Cambridge, MA: Polity Press,
1993
.
29
Holinger PC. Violent deaths among the young: Recent trends in suicide, homicide, and accidents.
Am J Psychiatry
 
1979
;
136
:
1144
–47.
30
Holinger PC. Suicide and homicide in the United States: an epidemiological study of violent death, population changes, and the potential for prediction.
Am J Psychiatry
 
1987
;
144
:
215
–19.
31
Griffith EEH, Bell CC. Recent trends in suicide and homicide among blacks.
JAMA
 
1989
;
262
:
2265
–69.
32
Diekstra RFW. The epidemiology of suicide and parasuicide.
Acta Psychiatrica Scandinavica
 
1993
;
371
(Suppl.):
9
–20.
33
Jenkins J, Sainsbury P. The accuracy of officially reported suicide statistics for purposes of epidemiological research.
J Epidemiol Community Health
 
1982
;
36
:
43
–48.
34
Pickering W, Walford G. Durkheim's Suicide: A Century of Research and Debate. New York: Routledge,
2000
.