Suicide, Alcohol Intoxication and Age among Whites and American Indians/Alaskan Natives

Abstract

Background:

Among American Indians/Alaskan Natives (AI/ANs), suicides are disproportionately high among those younger than 40 years of age. This paper examines suicide and alcohol intoxication (post-mortem BAC ≥ 0.08 g/dl) by age among Whites and AI/ANs to better understand the reasons for the high rate of suicide among AI/ANs for those younger than 40.

Methods:

Data come from the restricted 2003–2016 National Violent Death Reporting System (NVDRS), with post-mortem information on 79,150 White and AI/AN suicide decedents of both genders who had a BAC test in 32 states of the U.S.

Results:

Among Whites, 39.3% of decedents legally intoxicated are younger than 40 years of age while among AI/ANs the proportion is 72.9% (p<.001). Multivariable logistic regression with data divided by age show that in the 18–39 age group AI/ANs are about 2 times more likely than Whites to have a postmortem BAC≥.08. Veteran status compared to non-veteran, and history of alcohol problems prior to suicide were also associated with BAC≥.08. Suicide methods other than by firearm, and a report of the presence of 2 or more suicide precipitating circumstances were protective against BAC≥.08. Results for the age group 40 years of age and older mirror those for the younger group with one exception: Race/ethnicity was not associated with BAC level.

Conclusions:

The proportion of suicide decedents with a BAC≥.08 is higher among AI/ANs than Whites, especially among those 18–39 years of age. However, acute alcohol intoxication does not fully explain differences in suicide age structure between AI/ANs and Whites.

Introduction

Suicide remains an important public health problem in the U.S., ranking as the 10th leading cause of death in 2016 (Heron, 2018), when nearly 45,000 persons died by suicide (Centers for Disease Control and Prevention, 2018a). Trend data for age adjusted suicide rates show that rates among American Indians/Alaskan Natives (AI/AN) have been continuously higher than that among all U.S. races and U.S. Whites since 1972 (U.S. Department of Health and Human Services, 2017). AI/AN suicides were highest for years 1972–1974 (29.4/100,000) falling to 15.9/100,000 in 1984–1986, and then increased again continuously. In 2016, suicide rates were highest among AI/ANs (17/100,000), followed by Whites, (15/100,000), followed by Blacks, Asians/Pacific Islanders and Hispanics (5, 6, and 6/100,000, respectively).

Suicide rates vary considerably across AI/AN tribes and U.S. regions (Alcántara and Gone, 2007; Gone and Trimble, 2012; O’Keefe and Reget, 2017). For instance, the rate of suicide for the White Mountain Apache Tribe for the age groups 15–19, 20–24, and 25–34 was 6.3, 7.9, and 5.1 times higher than the rate for all AI/AN (Mullany et al., 2009). Age-adjusted to the 2000 U.S. standard population, suicide rates across Indian Health Service (IHS) regions based on 1999–2009 data varied from 42.5/100,000 in Alaska to 8.4/100,000 in the East IHS region (Herne et al., 2014). Suicidal ideation was lower in a Southwest tribe and 2 Northern Plains tribes than in the U.S. general population, but Northwest Plains tribes were almost two times more likely than the Southwest one to have attempted suicide in their lifetime (Bolton et al., 2014).

Many risk factors for suicide among AI/ANs have been identified in the literature. Among adults, suicides seem to be associated with male gender, alcohol, illicit drug use, death by firearm, loose tribal social integration, rapid socioeconomic change, physical and sexual abuse (Olson and Wahab, 2006; Gone and Trimble, 2012). Among adolescents, factors of risk are physical and sexual abuse, family disruption, previous suicide attempts, lack of religious identification, depression, anxiety, high acculturation pressure, overweight, and social isolation (Olson and Wahab, 2006; Zamora-Kapoor et al., 2016). Alcántara and Gone (2007), however, caution that many studies of suicidal behavior among AI/ANs are based on reservation or near-reservation samples, thus excluding urban populations. This latter group of AI/Ans may have different risk factors than those on reservations. For instance, AI/AN adolescents who grew up in urban areas (2/3 time spent in the area) had lower rates of suicidal ideation, conduct disorder, substance dependence/abuse, gang involvement, friends who attempted or completed suicide than those who grew up mostly on reservations (Freedenthal and Stiffman, 2004).

Age is one of the major risk factors for suicide, and recently there has been concern about the increase in suicide rate among younger age groups (e.g. Millennials, roughly the age group 23 to 38 years of age in 2019) (Trust for America’s Health, 2019). Such concern is warranted, but a high rate of suicide in young age groups in the U.S. is not a recent phenomenon. Epidemiological evidence on suicide rates has shown for quite some time that the age structure of both suicides and suicide attempts is quite different across U.S. ethnic groups. The major differences are between Whites and AI/ANs, Hispanics and Blacks. Data from 2003–2009 show that among Whites, 33.4% of suicides occurred among those younger than 30 years of age (Caetano et al., 2013). Among AI/ANs, Blacks and Hispanics that proportion was 69.8%, 59.1%, and 61.8%, respectively. Comparisons between AI/ANs and Whites on suicide rates per 100,000 between 1999–2009 show that among those younger than 25 the AI/AN rate was 21.3, while among Whites it was 5.3. In the 25–44 age group the AI/AN and White rates were 42.2 and 17.4, respectively (Herne et al., 2014). Data on suicide decedents from 2003–2014 indicated that while 35.7% of suicides among AI/ANs occurred in the 10–24 years of age group, among Whites the proportion was 11.1% (Leavitt et al., 2018). Regarding suicide ideation the risk among AI/AN adolescents was 31% higher than among Whites (Zamora-Kapoor et al., 2016).

Alcohol is another important risk factor for suicidal behavior (Cherpitel et al., 2004; Kaplan et al., 2012). Alcohol is detected in about a third of all suicide decedents, with proportions ranging from 45.5% among AI/ANs to 22.9% among Asian/Pacific Islanders (Centers for Disease Control and Prevention, 2009). Acute use of alcohol confers risk over and above the risk attributed to chronic use (Kaplan et al., 2012). AI/AN suicide decedents are almost three times more likely than White decedents to have consumed alcohol in the hours preceding death, and almost two times more likely to have had an alcohol problem while alive (Leavitt et al., 2018). Kaplan et al. (2016) found near gender parity in the prevalence of alcohol use among suicide decedents. Among AI/AN 42.5% of men and 35% of women were intoxicated at the time of their death by suicide. Also, AI/AN male and female suicide decedents were, respectively, 12.0 and 10.8 times more likely than a living comparison to have used alcohol heavily immediately prior to their death. The proportion of suicide decedents with blood alcohol concentration (BAC) ≥.08 is particularly high among younger AI/AN: 32.5% among those 21–29 and 30% among those 30–39 (Caetano et al., 2013).

Given this background of differences in age structure and alcohol-related suicide rates between White and AI/AN suicide decedents, the objective of this paper is to examine U.S. suicide data from 2006 to 2016 for Whites and AI/ANs with a focus on analyses by age and acute alcohol intoxication as determined by a BAC≥.08 g/dl Although the legal age for drinking in the United States is 21, the age analysis herein includes suicide decedents 18 years of age and older. This is because much of alcohol epidemiology in the United States uses age 18 as a cut-off point to differentiate adolescents from adults. Also, given the differences in suicide decedents’ age structure between White and AI/ANs, data are analyzed in two large age groups: 18–39 and 40 and more. The overall aim is to gain additional knowledge about why among AI/ANs relative to Whites, suicide and alcohol-related suicide affect disproportionately those below age 40. More specifically, it is important to know to what extent alcohol intoxication immediately preceding suicide among younger AI/ANs contributes to suicide disparities by age in these two racial/ethnic groups. Based on previous cross-ethnic analyses of the National Violent death reporting System (NVDRS) and the general alcohol epidemiology literature on drinking by Whites and AI/ANs (Gone and Trimble, 2012; Caetano et al., 2013; Caetano et al., 2015; Dawson et al., 2015), the hypotheses guiding the analyses are: a) Suicide decedents will be younger among AI/ANs than Whites; b) the proportion of suicide decedents with post mortem BAC ≥ 0.08 g/dl will be higher among AI/ANs than among Whites, especially in the younger age group; c) the number of circumstances preceding suicide will be higher among Whites; d) AI/ANs will have a higher proportion of decedents with a current alcohol problem.

Methods

Data source

The National Violent Death Reporting System (NVDRS) is a Centers for Disease Control and Prevention (CDC) state-based surveillance system that presently provides information on violent deaths occurring in 32 participating states. Until 2013 only 17 states participated in the NVDRS (Alaska, Colorado, Georgia, Kentucky, Maryland, Massachusetts, New Jersey, New Mexico, North Carolina, Ohio, Oklahoma, Oregon, South Carolina, Rhode Island, Utah, Virginia, and Wisconsin) (Caetano et al., 2013). Between 2014 and 2016, given an increase in funding, 15 more states were included (Arizona, Connecticut, Hawaii, Illinois, Indiana, Iowa, Kansas, Maine, Michigan, Minnesota, New Hampshire, New York, Pennsylvania, Vermont, Washington) for a total of 32 participating states. Data are gathered from coroner/medical examiner (C/ME) records including toxicology reports, police reports, death certificates, crime laboratories, and Alcohol, Tobacco, Firearms and Explosives (ATF) firearm trace reports. Information on acute alcohol use is based on toxicological analyses of decedents as part of the C/ME investigation. Suicide decedents were identified as those with death certificates that listed International Classification of Diseases, 10th Revision codes X60–84 or Y87.0. The analytical sample for the analyses herein includes AI/AN and White suicide decedents 18 years of age and older from the 32 NVDRS states from 2003 to 2016. Of these, 54.7% (N=76,972) have blood alcohol test results and constitute the analytical sample.

Measures

Blood Alcohol Content (BAC)

Overall, 55.0% of decedents had blood aclohol test results, including 54.7% (n=75,884) of Whites and 54.9% (n=1,088) of AI/ANs. Medical examiners and coroner officers were only instructed to enter toxicology information for substances that tested positive. They were not required to note that alcohol was tested if it tested negative. Therefore, the true proportion tested for alcohol is likely higher than the proportion with BAC results (Centers for Disease Control and Prevention, 2016). Blood alcohol levels were coded in terms of weight by volume, and then classified as < 0.08 g/dl or ≥ 0.08 g/dl. Blood alcohol levels in the NVDRS data were distributed continuously from 0 to .998 g/dl, and then jumped to 10.0 g/dl. Given this discontinuity, blood alcohol levels above .998 g/dl were excluded from the analysis, which resulted in the exclusion of 4 decedents

Circumstances prior to suicide:

The circumstances preceding suicide were derived from death investigations assembled by the coroner/medical examiner and law enforcement reports, based on on-scene investigations; information from next of kin and witnesses; medical records; autopsy examinations; postmortem toxicological testing; and, in some cases, contact with health care providers (Centers for Disease Control and Prevention, 2016). Most medical examiner and coroner offices use trained lay investigators who follow national investigative guidelines (National Institute of Justice, 2011). Each circumstance was coded as either 0 (not present, not available, unknown) or 1 (yes, present). This structure is intended to prevent abstractors from making assumptions about circumstances in cases without sufficient evidence (Centers for Disease Control and Prevention, 2018b). However, it also makes it impossible to separate decedents who did not have the circumstance present from those for whom the information about the circumstance was not available. Twelve special circumstances that preceded the suicide were analyzed: depressed mood, current mental health problem, current treatment for mental illness, substance problem other than alcohol, disclosed intent to commit suicide, history of suicide attempts, experienced a crisis, financial problem, physical health problem, job problem, alcohol problem, and intimate partner problem.

Race/Ethnicity:

The analytical sample consists of Non-Hispanic White and Non-Hispanic American Indian/Alaska Native (AI/AN) decedents. Race/ethnicity was obtained from death certificates.

Sociodemographic variables.

Also obtained from death certificates and included in the analyses were gender, age group (18–39, 40+), marital status, veteran status and metropolitan status. Counties of death were assigned a rural-urban continuum code based on a classification system developed by the U.S. Department of Agriculture Economic Research Service (2003). The original 9 categories were collapsed into 2 categories: metropolitan (codes 1 through 3) and non-metropolitan (codes 4 through 9).

Method of suicide:

The type of weapon that caused the death was recoded by C/ME and law enforcement. In cases where multiple weapons were used, only the weapon that was most likely to have caused the fatal injury was listed. Methods of suicide are divided into four categories: firearm, hanging/suffocation, poison, and “other” (biological weapons, blunt instrument, drowning, explosives, falls, fire or burns, motor vehicle, non-powder gun, and sharp instruments).

Statistical Analysis

In Table 1, Pearson chi-square was used to test differences between Whites and AI/ANs within BAC level in the proportion of decedents with a specific sociodemographic characteristic. In Table 2, Pearson chi-square was used to test differences in proportions within BAC level and age group between Whites and AI/ANs. T-tests were used to assess statistical differences between BAC means. In Table 3, chi-square was used to test statistical significance in differences in the distribution of proportions of decedents with a specific method of suicide across age and BAC groups. In Table 4, the association between the presence of a precipitating circumstance and race/ethnicity within BAC level and age group was assessed with logistic regression testing the effect of each specific circumstance on a binary outcome representing race/ethnicity coded as 1 for AI/ANs and 0 for Whites (reference group). In Table 5, multivariable logistic regression analysis was used to examine the independent effect of the number of precipitating circumstances on BAC (BAC ≥ 0.08 g/dl) controlling for sociodemographic factors, alcohol problems and suicide method. The levels of significance in the analyses in Tables 4 and 5 were corrected using the Bonferroni method.

Table 1.

Sociodemographic Characteristics of Suicide Decedents 18 and older by BAC (g/dl) and Ethnicity: National Violent Death Reporting System 2003 to 2016 (n=76,972)

	BAC < 0.08		BAC ≥ 0.08
	White (54,926)	AI/AN (567)	White (20,958)	AI/AN (521)
Gender
Male	74.6	72.0ns	79.9	77.2ns
Female	25.4	28.0	20.1	22.8
Age
18–39	31.5	60.2***	39.3	72.9***
40 and more	68.5	39.9	60.7	27.1
Marital status
Not married	63.0	72.1***	65.4	81.2***
Married	37.0	27.9	34.6	18.8
Veteran status
Nonveteran	79.0	91.9***	82.8	91.9***
Veteran	21.1	8.1	17.2	8.1
Urban-rural continuum
Metropolitan	80.3	44.3***	80.9	44.2***
Nonmetropolitan	19.7	55.7	19.1	55.8

Sample size in parentheses; BAC, blood alcohol content; AI/AN, American Indian/Alaska Native.

Non-significant: Chi2 gender differences BAC<0.08, White x AI/AN: NS; Chi2 gender differences BAC≥0.08, White x AI/AN

^***P<0.001: Chi2 age, marital status, veteran status, urban-rural, BAC<0.08, White x AI/AN; Chi2 age, marital status, veteran status, urban-rural, BAC≥0.08, White x AI/AN

Table 2.

Suicide decedents with BAC positive, with BAC ≥ 0.08, and mean positive BAC by race/ethnicity and age

	18–39 years		40 years and over
	White	AI/AN	White	AI/AN
	(17,304)	(341)	(37,616)	(226)
BAC positive (%)	42.6	59.3***	34.8	47.1***
BAC≥ .08 (%)	32.2	52.7***	25.3	38.4***
Mean BAC (95% CI)1,2	.15 (.145– .149)	.19 (.179–.201)***	.15 (.149–.152)	.18 (.165–.198)***

Sample size in parentheses; BAC, blood alcohol content; AI/AN, American Indian/Alaskan Native.

^***p<0.001: Chi2 BAC positive, BAC ≥ 0.08, 18–39 years, White x AI/AN; Chi2 BAC positive, BAC ≥ 0.08 40+ years, White x AI/AN;

^1.two-sided, independent t-test for unequal variance testing the difference in mean BAC between racial/ethnic group within 18–39 years category: p<0.0001;

^2.two-sided, independent student’s t-test testing the difference in mean BAC between racial/ethnic group within 40+ years category: p<0.0003

Table 3.

Proportions of Decedents with Specific Methods of Suicide by BAC (g/dl), Age and Race/Ethnicity

	BAC < 0.08				BAC ≥ 0.08
	18–39 years***		40 years and over*		18–39 years***		40 years and over***
	White	AI/AN	White	AI/AN	White	AI/AN	White	AI/AN
	(17,304)	(341)	(37,616)	(226)	(8,229)	(380)	(12,726)	(141)
Method
Firearm	42.3	29.0	50.9	42.9	52.7	42.6	54.9	39.7
Hanging/suffocation	31.7	47.5	17.0	22.1	32.2	42.9	19.9	29.8
Poison	19.0	16.4	25.2	24.8	10.1	5.5	20.4	19.9
Other	7.0	7.0	7.0	10.2	5.0	9.0	4.9	10.6

Sample size in parentheses; BAC, blood alcohol content; AI/AN, American Indian/Alaskan Native.

^*p<0.05: Chi2 method BAC<0.08 and 40+ years, White x AI/AN

^***p<0.001: Chi2 method BAC<0.08 and 18–39 years, White x AI/AN; Chi2 method, BAC≥0.08 and 18–39 years, White x AI/AN: Chi2 method, BAC≥0.08 and 40+ years, White x AI/AN

Table 4.

Proportions of Suicide Decedents with Specific Types of Precipitating Circumstances by BAC (g/dl), Age and Race/Ethnicity

	BAC < 0.08
	18–39 years		40 years and over
	White	AI/AN	White	AI/AN
	(17,306)	(341)	(37,620)	(226)
Circumstance1
Depressed mood	38.3	34.6	42.1	28.3
Current mental health problem	49.1	36.7	49.8	32.3
Current treatment for mental illness	35.7	22.6	37.7	21.2
Disclosed intent to commit suicide	27.5	29.6	24.4	23.0
History of suicide attempts	25.0	22.0	18.9	12.8
Substance problem other than alcohol	27.0	25.2	12.2	17.3
Alcohol problem	11.5	15.8**	11.7	20.8***
Experienced a crisis	34.3	30.2	26.1	26.6
Financial problem	9.5	5.3	12.5	7.5
Physical health problem	8.6	6.7	32.8	30.1
Job problem	12.3	7.6	11.7	5.3
Intimate partner problem	37.2	37.2	20.7	24.8
Mean number of circumstances (95% CI)2	3.16 (3.13–3.19)	2.74 (2.53–2.94)***	3.00 (2.99–3.03)	2.50 (2.28–2.72)***
	BAC≥0.08
	18–39 years		40 years and over
	White	AI/AN	White	AI/AN
	(8,230)	(380)	(12,728)	(141)
Circumstance1
Depressed mood	38.3	34.0	40.8	31.2
Current mental health problem	42.0	23.7***	45.0	29.8
Current treatment for mental illness	26.2	12.4	30.5	17.7
Disclosed intent to commit suicide	30.1	30.8	29.1	34.0
History of suicide attempts	22.1	25.8***	19.0	16.3
Substance problem other than alcohol	23.3	17.6**	14.7	12.8
Alcohol problem	39.5	45.8***	47.6	53.2
Experienced a crisis	35.7	29.5	31.1	33.3
Financial problem	9.8	6.6	13.5	6.4
Physical health problem	5.4	4.0	17.9	17.0
Job problem	13.9	7.1**	15.7	7.8
Intimate partner problem	47.5	48.4	36.3	35.5
Mean number of circumstances (95% CI)2	3.3 (3.29–3.38)	2.8 (2.66–3.05)***	3.4 (3.38–3.45)	2.9 (2.62–3.28)

Sample size in parentheses. BAC, blood alcohol content; AI/AN, American Indian/Alaskan Native.

^1.Analysis used logistic regression to test differences in proportion of circumstances between racial/ethnic group within BAC and age category and alpha was adjusted for multiple comparison using Bonferroni method (p < 0.004).

^2.Analysis used two-sided, independent t-tests to test differences in mean number of circumstances between racial/ethnic group within BAC and age category.

^**P<0.004

^***P<0.001

Table 5.

Odds Ratios and 95% Confidence Limits from Multivariate Logistic Regression of BAC Level ≥0.08 g/dl on Sociodemographic Factors, Suicide Methods, and Number of Precipitating Circumstances by Age Group, OR (95%)

	18–39 years	40 years and over
	(24,160)	(46,358)
Race/ethnicity (ref: White)
Native American/Alaskan Native	2.18 (1.84–2.57) ***	1.41 (1.11–1.80)
Alcohol problem (ref: no alcohol problem)
Yes	5.66 (5.29–6.07) ***	7.40 (7.04–7.79) ***
Number of circumstances (ref: no circumstances reported)
1 circumstance	.85 (.76–.96)	.80 (.73–.88) ***
2 circumstances	.80 (.72–.90) ***	.75 (.69–.82) ***
3 or more circumstances	.64 (.58–.71) ***	.64 (.59–.70) ***
Gender (ref: female)
Male	1.11 (1.03–1.20)	1.09 (1.03–1.16)
Marital status (ref: married)
Not married	1.03 (.96–1.10)	.99 (.95–1.04)
Veteran status (ref: not a veteran)
Veteran	1.17 (1.07–1.29) **	.72 (.68–.76) ***
Urban-rural continuum (ref: not metropolitan)
Metropolitan	1.07 (1.00–1.15)	1.02 (.97–1.08)
Method of suicide (ref: firearm)
Hanging, strangulation, suffocation	.80 (.74–.85) ***	.97 (.91–1.03)
Poisoning	.41 (.37–.45) ***	.68 (.64–.73) ***
Other	.59 (.52–.67) ***	.57 (.52–.64) ***

Sample size in parentheses. BAC, blood alcohol content. Alpha adjusted for multiple comparison using Bonferroni method (p< 0.004);

^**p< 0.004

^***p< 0.001

Results

Suicide decedents’ sociodemographic characteristics by BAC level and race/ethnicity

There are no differences in gender distribution between Whites and AI/ANs independent of BAC level (Table 1). Differences in the distribution of White and AI/AN decedents by age, marital status, veteran status, and urban-rural continuum are all significant (p<.001) both in the younger and the older group, i.e., independent of BAC level. Among those with BAC<.08 and BAC≥.08, Whites are significantly older than AI/ANs, with a larger proportion of suicides happening among those 40 and older. For instance, among Whites with BAC<.08, a third of the suicides occur among those 18–39 years old while among AI/ANs that proportion is about two times higher. The same is true among decedents with a BAC≥.08. Among Whites, almost two fifths of legally intoxicated decedents are in the 18–39 age group while among AI/ANs the proportion is almost 2 times larger. BAC level appears to have a similar effect in the age distribution of suicides among Whites and AI/Ans by being associated with an increase in the proportion of suicides. Differences in marital status, veteran status and urban-rural continuum distributions indicate that among both BAC levels, a significantly higher proportion of Whites are married, are veterans, and live in metropolitan areas.

Proportion of decedents BAC positive, proportion with BAC ≥ 0.08, and mean positive BAC by race/ethnicity and age

The proportion of White and AI/ANs decedents with a positive post mortem BAC are 37.4% and 55.1%, respectively (p<.0001; data not shown). The proportion of decedents with a BAC ≥ 0.08 is higher among AI/ANs than Whites (47.9% x 27.6%; p<.0001). So, the majority of White (70%) and AI/ANs (86%) decedents who had a positive post mortem BAC had a BAC≥ 0.08. Data by age show that the proportion of AI/ANs with a positive BAC was significantly higher than among Whites in both age groups (Table 2). The same is true for the proportion of AI/AN with a BAC ≥ 0.08. Mean BAC was about two times higher than the legal intoxication level among both younger and older decedents and among both Whites and AI/ANs. Among decedents 18–39 years of age and among those 40 years of age and older, mean BAC was significantly higher (p<.001) among AI/AN than among Whites.

Suicide methods by BAC level, age, and race-ethnicity

Suicide methods vary significantly across Whites and AI/ANs within BAC levels and age groups (Table 3). Among decedents with a post mortem BAC<.08, Whites 18–39 years of age and those 40 years and older have a significantly higher proportion of decedents who died by firearm (18–39: p<.001; 40 +: p<.05) than AI/ANs. This difference between Whites and AI/ANs is particularly noticeable in the younger age group. Among AI/ANs 18–39 years of age, a higher proportion of decedents died by hanging/suffocation than by any other method. Among decedents with a BAC≥.08, the preferred method of suicide is still by firearm. However, among AI/AN in the 18–39 group with BAC≥.08, suicide by hanging/suffocation is as common as by firearm. The effect of BAC≥.08 therefore is seen only among AI/ANs in the younger age group and it is associated with an increase in the use of firearm as a suicide method.

Types of precipitating circumstances by BAC level, age and race/ethnicity

Most decedents independent of BAC level, age, and race/ethnicity had at least one precipitating circumstance (data not shown). The proportion of decedents with at least one precipitating circumstance varied from 88.4% among AI/ANs 18–39 years of age and BAC ≥.08 to 92.5% among White decedents 40 years of age and older with BAC equal to .08 or higher. Among decedents with BAC<.08, the mean number of precipitating circumstances is significantly higher among Whites than among AI/ANs, especially among those 40 years of age and older (top half, Table 4). Among these decedents, the proportion of AI/AN suicide decedents with a history of alcohol problems prior to suicide is higher than the proportion among White decedents both in the 18–39 and the 40 and older age groups.

Among decedents with BAC<.08, the mean number of precipitating circumstances is significantly higher among Whites than among AI/ANs in the 18–39 and 40 years of age and older group (upper half, last row, Table 4). Among decedents with BAC≥.08, the mean number of precipitating circumstances is significantly higher among Whites than among AI/ANs in the 18–39 years of age group only (lower half, last row, Table 4). Among those with a BAC≥.08, AI/ANs 18–39 years of age have a lower proportion of decedents with current mental health and job problems, but a higher proportion of decedents with a history of suicide attempts and a history of alcohol problems compared to Whites in the same BAC and age group.

Association between BAC level and sociodemographic factors, suicide methods and the number of precipitating circumstances

This analysis was conducted with multivariable logistic regression. The first model was run on all data and had all the variables shown in Table 5 plus a main effect for race-ethnicity, a main effect for age, and an interaction term between race/ethnicity and age. Looking at the race-ethnicity main effect, in this model AI/ANs were more likely than Whites to have BAC≥ .08 (AOR:1.46; 95%CI:1.144–1.851; p<.004). Regarding the age main effect, suicide decedents in the18–39 age group were more likely than those 40 years of age and older to have a post mortem BAC≥.08 (AOR: 1.43; 95%CI: 1.374–1.486; p<.000). The interaction term suggested that AI/ANs 18–39 years of age were the most likely to have post mortem BAC≥.08 whereas Whites 40 years of age and older were the least likely to have post mortem BAC ≥ .08 (AOR:1.46; 95%CI:1.091–1.955; p<.05).

Given the focus of the paper on age differences in suicide rates between Whites and AI/ANs and the above results for the initial regression, the next step in the analysis estimated AORs using multivariable logistic regression with data divided by age, 18–39 and 40 and more (Table 5). One regression model was run for each age group, with significance levels adjusted by the Bonferroni method. In the 18–39 age group AI/ANs are about 2 times more likely than Whites to have a postmortem BAC≥.08. The presence of alcohol problems prior to suicide is strongly related to BAC ≥.08 in both age groups. The presence of 2 or more suicide precipitating circumstances was protective against BAC≥.08 in the 18–39 age group, while in the older age group the presence of any circumstance was protective against BAC≥.08. Veteran status compared to non-veteran was also associated with BAC≥.08, but in different ways across age groups. It is a factor of risk among those 18–39 but it is a protective factor in the older group. Suicide methods other than by firearm are almost all protective against BAC≥.08. Other results for the age group 40 years of age and older mirrored those for the younger group with one important exception: Bonferroni corrected statistical significance showed that race/ethnicity was not associated with BAC level among older decedents.

Discussion

Of the four hypotheses tested in this paper three were fully confirmed and one was confirmed for specific age groups. First, suicide decedents are indeed younger among AI/ANs than Whites and, second, the proportion of suicide decedents with post mortem BAC ≥ 0.08 g/dl is higher among AI/ANs than among Whites. Third, the mean number of suicide precipitating circumstances was confirmed as higher among Whites. Fourth, the hypothesis that AI/ANs would have a higher proportion of decedents with a current alcohol problem was not confirmed among decedents 40 years of age and older with a BAC≥.08.

These hypotheses partially agree with previous results showing higher suicide rates and alcohol-related suicide rates in younger age groups among AI/ANs than among Whites (Centers for Disease Control and Prevention, 2009; Caetano et al., 2013; Keyes et al., 2013; Herne et al., 2014; Leavitt et al., 2018). Results from the multivariable regression analyses show that, irrespective of ethnicity, suicide decedents 18–39 years of age are about 40% more likely than those 40 years of age and older to have a post mortem BAC≥.08. Further, AI/ANs in this younger age group are more than 2 times more likely than Whites matched on age to have a BAC≥.08. These results reflect in part U.S. alcohol epidemiology, which shows that alcohol consumption and associated problems are higher among those in their twenties and thirties than in older groups (Dawson et al., 2015; Grant et al., 2017), and are also higher among AI/ANs than Whites, although there is considerable variation across tribes in drinking (Dawson et al., 2015; Vaeth et al., 2017).

However, the difference in suicide age structure between Whites and AI/ANs is present independent of whether post mortem BAC shows acute alcohol intoxication (BAC≥.08) or not: Results in Table 1 show that post mortem BAC≥.08 is associated only with a slight increase in the proportion of suicides in the younger age group, and this happens both among Whites and AI/ANs. Results also show that AI/AN decedents have a significantly higher mean BAC than Whites in both age groups under analysis, corroborating previous findings on mean post mortem BAC across ethnic groups (Centers for Disease Control and Prevention, 2009; Caetano et al., 2013).

The choice of suicide method results from complex interactions between individual characteristics, suicide risk factors, the availability of means (e.g., access to firearms), and sociocultural ideas about suicide (Kaplan et al., 2012). Results in Table 3 show that death by firearm is the preferred suicide method both among Whites and AI/ANs, with exception of AI/ANs in the 18–39 age group. In this latter group the preferred suicide method is hanging/suffocation. Suicide by suffocation is the 5^th leading cause of death in the 15–34 age group in the U.S. (National Center for Health Statistics, 2017). Lack of access to firearms in the younger age groups may be the main reason why suicide by hanging and suffocation are more common. Caetano et al. (2013) reported similar findings for AI/ANs, although their data were not broken down by age group. Also, in contrast to suicide by firearm, all the other suicide methods in the analysis are associated with a significantly smaller likelihood of a post mortem BAC≥.08. A previous analysis of NVDRS data on suicide and BAC level did not show an association between these two factors among AI/ANs as well as among other ethnic groups, indicating that the pattern is not specific to AI/ANs (Caetano et al., 2013). Those analyses were not age-specific as in this paper and were also conducted with data covering a different set of years (2003 to 2009). The increased proportion of suicide by firearms in the older age group compared to the younger one has been described before, especially among men (Kaplan et al., 2009; Kaplan et al., 2012). In fact, the best predictor of death by firearm is a health problem at the time of death, which are more common among older men (Kaplan et al., 2012).

Results on suicide precipitating circumstances are somewhat puzzling. Frist, the regression analysis shows that the presence of precipitating circumstances decreased the likelihood that post mortem BAC will be .08 or higher. It is possible that decedents with strong adverse circumstances affecting their lives do not need to drink alcohol as a suicide facilitator. Also, the analysis of circumstances does not present a uniform picture across BAC levels and age groups in this paper. Given that AI/ANs constitute a socioeconomically disadvantaged group in the U.S., also affected by ethnic discrimination and historical trauma, one would expect that a higher proportion of suicide decedents among AI/ANs relative to Whites would have one or more suicide precipitating circumstances among those listed in the NVDRS. However, as in a previous analysis (Caetano et al., 2015), AI/ANs have a mean number of circumstances lower than Whites at both BAC levels and age groups in Table 4, with exception of those with BAC≥0.08 and 40 years of age and older among whom the mean number of circumstances is not different between AI/ANs and Whites. In fact, results in Table 4 show significant differences between Whites and AI/ANs only among decedents 18–39 years of age and with a BAC≥0.08. In this group, AI/ANs have, unexpectedly, a lower proportion of decedents with a current mental health problem, a substance abuse problem, and a job problem. This was recently reported by Leavitt et al. (2018) as well, who suggested that this difference between AI/ANs and Whites could be due to AI/ANs’ reduced access to mental health services and thus lack of identification of such problems, especially in rural areas.

This lower level of precipitating circumstances among AI/ANs could be due to several factors. Authors have called attention to cultural differences between Whites and AI/ANs in reasons for and reactions to suicide, which among AI/ANs would be more associated with group and cultural related events such as historical trauma. Wexler and Gone (2012) and Wexler et al. (2015) argue that the conceptualization of suicide as a personal act that indicates serious psychological problems that should be treated and/or prevented by a specialized professional is not readily acceptable to AI/AN communities. In these communities, suicide is not a person-centered response but is seen as resulting from oppression, cultural loss and historical trauma, which means that frequently the organization of behavioral health services that respond to suicide in AI/AN communities is incompatible with AI/ANs’ cultural values (Mullany et al., 2009; Wexler, 2011). NVDRS data collection on suicide precipitating circumstances does not cover these group-level factors, focusing on individual risk factors only. It is also possible that for cultural reasons there is less disclosure of suicide precipitating circumstances factors among AI/ANs during the NVDRS suicide investigations.

The proportion of AI/AN decedents with alcohol problems prior to suicide is higher than among Whites, with exception of those with BAC≥0.08 and 40 year of age and older. Also, and as expected, results from the regression analysis show that the presence of alcohol problems prior to suicide is the strongest predictor of post mortem BAC≥0.08. The higher rate of alcohol problems among AI/ANs relative to Whites is present in the existing epidemiological literature. For instance, a recent national survey, the 2012–2013 National Epidemiologic Survey on Alcohol and Related Conditions (NESARC III), showed that among all U.S. ethnic groups, Native Americans had the highest rate of 12-month high risk drinking (17% versus 12% among Whites) (Grant et al., 2017). Native Americans also had the highest rate of 12-month (19.2% versus 14% for Whites) and lifetime Diagnostic and Statistical Manual (Fifth Edition) alcohol use disorder (AUD) (43.4% versus 32.6% for Whites), and the highest rate of 12-month (7.7% versus 3.2% for Whites) and lifetime (27.2% versus 15.8% for Whites) severe AUD (Grant et al., 2015).

A history of previous suicide attempts is also more common among 18–39 year old AI/ANs with a post mortem BAC≥0.08. Similar levels of previous suicide attempts are present in the same age group with a BAC<.08, but that result is not statistically significant. The presence of a history of prior suicide attempts in roughly a fifth to a quarter of suicide decedents in this age group is not surprising given that suicide is the 2^nd leading cause of death in the 15–34 age group in the U.S. (National Center for Health Statistics, 2017), and given the continuous trend for higher suicide rates among AI/ANs compared to Whites (Gone and Trimble, 2012; U.S. Department of Health and Human Services, 2017). Also, suicides in this younger age group have been associated with a higher level of impulsivity, which could also contribute to a more common history of previous attempts (McGirr et al., 2008; Gvion and Apter, 2011). Finally, veteran status is positively associated with having a post mortem BAC≥0.08 among younger decedents, and negatively associated with this outcome among older decedents. Kaplan et al. (2012) reported similar findings using NVDRS data, showing a higher rate of acute alcohol intoxication among younger than older veterans. Caetano et al. (2013) also analyzing NVDRS data reported null findings for the association between veteran status and BAC level. These variations in results could be due to differences in years of NVDRS data analyzed across these studies, with perhaps a mixture of veterans with various war and post-war adaptation experiences in their return home. More generally, analysis of suicides rates among U.S. veterans have yielded conflicting results showing an increased risk of suicide among all veterans (Kaplan et al., 2007; Maynard and Boyko, 2008; McCarthy et al., 2009), in younger age groups (Kaplan et al., 2009; Kaplan et al., 2012), and no association between veteran status and suicide (Miller et al., 2009).

Strengths and Limitations

The NVDRS is the only surveillance system for violent deaths in the United States. Presently it collects data in 32 states, although in 15 of these states data collection began 3–5 years ago. Data collection comes from various sources, including medical examiner offices, police, and family and friends. Toxicological examination covers a comprehensive set of legal and illegal drugs. The NVDRS also has limitations. First, not all suicide decedents are tested for BAC across states. In addition, because medical examiners and coroners offices are not required to report negative BAC results, the number of decedents with a negative BAC may be underestimated. Some states that do not participate in the NVDRS have sizable AIAN populations, such as North and South Dakota, Montana, and Wyoming, and the results of this study may not be generalizable to AI/AN throughout the U.S. Decedents’ ethnicity is determined from information provided by family members, by direct observation by the coroner or medical examiner’s office, by police officers, and by directors of funerary agencies. Second, ethnicity can be misidentified in death certificates, being more substantially so for AI/ANs than other ethnic groups (Stehr-Green et al., 2002; Arias et al., 2008). Third, the validity of the information about precipitating circumstances is dependent not only on the quality of the post-mortem interview with friends and family members, but also on the extent to which friends and family members can be contacted and are willing to be interviewed. Fourth, the data do not include a non-suicide (living) comparison group.

In summary, suicide decedents among AI/ANs are younger than Whites, have a lower mean number of precipitating circumstances, a higher proportion of decedents with a post mortem BAC indicating acute alcohol intoxication, and a higher post mortem BAC mean. However, as a risk factor that is more prevalent among younger AI/AN decedents, acute alcohol intoxication does not fully explain differences in suicide age structure between AI/ANs and Whites.