Suicide Rates Among 5–29-Year-Old African Americans,1990

Abstract

Suicide is the 10th leading cause of death across all ages in the USA, and the 2nd leading cause for ages 10–34. The rate of suicide for ages 10–34 has been increasing from 1999 to 2017 for males and females, although this time-trend varied across subpopulations defined by gender and age. This study analyzed Centers for Disease Control and Prevention (CDC) data from 1990 to 2018 on suicide among African Americans aged 5–29 years. Suicide incidence in this population was greater in older groups and in males than females. For 9 of the 10 gender/age subpopulations, the time-trend of suicide over the three decades was U-shaped—decreasing during the first decade and increasing over the last decade of the study period. The exception was the increasing trend from 1990 to 2018 for Black females in the 10–14 age-group. These results reinforce the need for analyses of the data from specific subpopulations in order to design adequate suicide prevention programs for these African American subpopulations.

Keywords

African American suicide rates disaggregated data time-trends

Suicide is a major public health issue among adolescents and transitional age youth in America. Hedegaard et al. (2018), reporting on national data, found that suicide rates increased for both males and females during the period 1999–2017, and that the rate was higher in 2017 than in 1999 for all age-groups for both males and females. Similarly, Miron et al. (2019) examined suicide data from 2000–2017 for the age-groups 15–19 and 20–24 years old. Among the 15–19 age-group, they found a decreasing trend in the suicide rate among males from 2000 to 2007, an increasing trend from 2007 to 2015, and a larger increasing trend from 2015 to 2017. Among females, there was no trend from 2000 to 2010 but an increasing trend from 2010 to 2017. Among the age-group 20–24 years, there was an increasing trend from 2000 to 2013 and a larger increasing trend from 2013 to 2017 among males, while among females there was an increasing trend from 2000 to 2017. A comparison of the findings by Hedegaard et al. (2018) and Miron et al. (2019) illustrates an important general point; the description of trends in suicide rate depends on the level of aggregation of the relevant data––with respect to how the target subpopulations are defined, and how time is treated. Hedegaard et al. (2020) presented data for four broad age-groups, that is, 10–24, 25–44, 45–64, 65–74, and over 75, which included gender but did not include race in the presentation of the data. These studies presented general data for the total population and did not provide any detailed analysis of suicide among the Black population. In seeking to describe suicide among the Black population, authors have used different levels of aggregation that would limit comparability across studies. For example, Price et al. (2019) looked at the age-group 13–19 for males and females for the period 2001–2017, while Ramchand et al. (2021) examined males for the age-group 15–24 for the period 1999–2019. The gap in the literature is the lack of consistent data disaggregated by 5-year age-groups, gender, and over the same period of time. In these studies, subpopulations were defined by age-group, gender, and urbanization; and time was treated as a sequence of consecutive years after 1999. In the present study, I focused on the population of African Americans 5–29 years old and included data from the years before 1999, so that the descriptions of suicide rate would, as expected, differ from those given for the general population. In 2006, a special edition on Black suicide was published in the Journal of Black Psychology (JBP). In the introduction to that publication, Crosby and Molock (2006) stated, “African American adolescents and young adults have the highest number and the highest rate of suicide of any age-group of African Americans” (p. 2). In that special issue, Joe (2006) reported on the scarcity of literature on Black suicide and called for “cohort-period analyses that are specific to the U.S. Black experience and that provide a greater understanding of the nature of the changes in Blacks’ patterns of suicide” (p. 263). Crosby and Molock (2006) described trends in suicide rate that were different from those offered by Hedegaard et al. (2018) and Miron et al. (2019), albeit for a different time period. Crosby and Molock (2006) found among Non-Hispanic African Americans, suicide rate decreased from 1990 to about 2003. Joe (2006) used a cohort-based analysis of data from 1981 to 2002 to show that the suicide rate increased across the 21 years; that the rate for Black females was relatively flat and lower than that for Black males across the time period; and that there was a noticeably higher rate of suicide among the 20–24 and 25–29 age-groups than among the older age-groups of 75–79 and 80–84 years. Other authors in that special JBP issue examined suicidal risk factors and ideation (Durant et al., 2006), explanation of gender differences in completed suicides (Griffin-Fennel et al., 2006), suicidal attempts (Kaslow et al., 2006), support systems and coping strategies as buffers against suicide attempts (Molock et al., 2006) and beliefs about suicide (Walker et al., 2006). These studies are relevant to the design of suicide reduction programs, a topic to which I will return in the Discussion section.¹

Nock et al. (2008), in examining rates of suicide (number of suicides per 100,000 population) from 1997 to 2007, showed that the rate of suicide was consistently higher for males than females across all age-groups. Nock et al. (2008) presented cross-sectional data for 2005 disaggregated by race, gender, and 5-year age-groups. In the cross-sectional comparison among age-groups, Nock et al. (2008) reported that the suicide rate for young Black males increased with age, from about 2 per 1,000,000 at ages 10–14 to about 9 per 1,000,000 at ages 15–19, and to about 17 per 1,000,000 for ages 20–24. However, in the trend analysis for the period 1990–2005, they only presented data by race and gender, and across four broad age-groups, that is, 10–24, 25–44, 45–64, and over 65.

Bridge et al. (2015) published one of the first studies to include children as young as 5 years old. They combined the data for the ages 5–9 and 10–11 into a 5–11 group, and examined trends across four 5-year periods starting with the period, 1993–1997, and ending in 2008–2012. They found that there was a significant increase in the suicide rate among Black children in this age-group, and a significant decline in the suicide rate among White children.

The Need for Disaggregated Data

As noted above, the description of trends in suicide depends on how data are aggregated. This issue of aggregated data is particularly important for the youngest age-group (i.e., 5–9 years). For example, it is inappropriate to group children 5–9 years old with older children ages 10–11 years, and then attribute to the younger age-group the same risk factors as the older age-group.

The young child (5–9 years old) has not yet developed the cognitive ability and reasoning skills to make informed decisions about taking one’s life. Child development specialists and researchers, including Piaget (1936), Kohlberg (1958), and Fu-Kiau and Lukondo-Wamba (2000) have emphasized the different stages of mental and moral development that characterize the childhood years. Because of the developmental level of children under 9 years old, research on suicide has usually excluded this group of children. Besides cognitive-developmental reasons for separating the 5–9 age-group from the 10–11 age-group, it is likely that the trends seen in the 5–11 age-group are largely determined by the 10–11 age-group because the older age-group almost certainly has more suicides than the younger age-group. The 5–9 age-group was included in the present study because, even though suicide is relatively rare in this age-group, the risk factors that have been found in the older age-groups may be present in this younger age-group, albeit at a lower level.

Aggregation level across years is another possible contributor to differences in apparent trends in suicide. In the present study, I used annual data rather than data pooled into, for example, 4- or 5-year periods, as used by Bridge et al. (2015). I used annual data because pooling across years often masks or eliminates the nuances of annual data. Finally, I included data from 1990 to 2018, because, as reported above, the decreasing trends reported by Crosby and Molock (2006) for the last decade of the 20th century are different from the increasing trends reported by Hedegard et al. (2018) and Miron et al. (2019) for the first two decades of the 21st century.

Purpose of Study

The purpose of this study is to fill the gaps in the literature about the trends in suicide over time within specific age-groups (5–29 years) in the Black population. More specifically, I examined the time-trends in suicide within the Black population disaggregated by age-group, year, and gender from 1990 to 2018. The time span of the data to the years 1990–2018 was extended because earlier research on Non-Hispanic African Americans has shown a decrease in suicide rate from 1990 to 2003 (Crosby & Molock, 2006). Also, data of children aged 5–9 years were included as well as 4 other age-groups, 10–14, 15–19, 20–24, and 25–29, so as to obtain a fine-grained description of age-group effects. Gender was nested within each age-group to examine differences between females and males. Restricting the study to the Black population removed the use of a racial comparative approach, but provided a more fine-tuned analysis of suicide among the Black population 5–29 years old.

Close attention was paid to the apparent conflict in reports of the time-trend in suicides as (a) increasing (e.g., Hedegaard et al., 2018; Joe, 2006) and (b) decreasing (e.g., Crosby & Molock, 2006.) Further, the data for Black children were re-examined by separating the data for the 5–9 age group from that of older children (10–11) to explore the trend of suicide in this young age-group over time. Finally, I extended the analysis of the data for children by including transitional age youth (20–24 years old) and young adults (25–29 years old) to explore the age at which suicide rates could possibly begin to decline.

Method

Data Source

Annual suicide data were obtained from the Web-based Injury Statistics Query and Reporting System (WISQARS) of the Centers for Disease Control and Prevention, National Center for Injury Prevention and Control for the period 1990–2018 (CDC, 2019a; 2019b). I extracted the number of suicides for each year (1990–2018) in 5-year age-groups (5–9, 10–14, 15–19, 20–24, and 25–29) for Non-Hispanic Black males and females. Population level data were used for the Non-Hispanic Black population. Intercensal population estimates were obtained from the 1990–2018 online database with bridged-race estimates of the July 1 resident population from the Vintage 2018 postcensal series, the revised 2000–2009 intercensal series, and the 1990–1999 intercensal series. On the 1990–2018 query page, Vintage 2018 postcensal estimates of the July 1 resident population are available for year 2010 and later; intercensal estimates of the July 1 resident population are available for the years 1990–1999 and 2000–2009 (USDHHS, 2019).

It should be noted that the present study is based on census-level population data, not on primary data from a sample of participants. In this regard, this study differs from studies of ideational and behavioral factors that may be associated with suicide (cf., (Hollingsworth et al., 2016; Morrison & Hopkins, 2019; Wilton et al., 2018))

Measures

The variables used from the databases were gender and age-groups for Blacks who had committed suicide. Total suicides for the Non-Hispanic Black population were organized by 5-year age-groups and gender for the period 1990–2018. Annual suicide rates per 100,000 population were computed for the four oldest age-groups, 10–14, 15–19, 20–24, and 25–29 years old over the time span of the study. Actual numbers of suicide deaths were analyzed for the youngest age-group 5–9 years old as the numbers were too small to allow for computation of rates and were susceptible to errors in reporting.

Procedures

The National Center for Health Statistics released bridged-race population estimates annually of the July 1 resident population of the United States for use in calculating vital rates. These annual estimates resulted from “bridging” the 31 race categories used in Census 2000, as specified in the 1997 Office of Management and Budget (OMB) standards for the collection of data on race and ethnicity, to the four race categories specified under the 1977 standards (Asian or Pacific Islander, Black or African American, American Indian or Alaska Native, White). The process of race bridging made data collected from two different methods of race classification compatible. In this way, data were created that would enable estimation and analysis of race-specific statistics. There were no missing data points or outliers, and the bridged-data provided a more consistent data set throughout the study period 1990–2018. For this study, I extracted population data for males and females for each year (1990–2018) in 5-year age-groups (5–9, 10–14, 15–19, 20–24, and 25–29).

Statistical Methodology

First, the use of annual data was contrasted with the choice made by other writers (cf., Bridge et al., 2015; Nock et al., 2008) to pool data into 5-year periods, and then to examine trends across periods. This choice was used because pooling data across years can obscure short-term fluctuations that may be helpful in understanding the data. Second, in disaggregating the data by age-group, it is important to analyze the data of the 5–9 age-group separately from that of the 10–11 age-group because the time-trends observed in the 5–11 age-group reported by Bridge et al. (2015) could be dominated by that of the 10–11 age-group and hence, would not accurately reflect trends in the 5–9 age-group.

For all age-groups except the 5–9 age-group, the index of suicide used in the analyses was the number of suicides per 100,000 of population, that is, the suicide rate, for each year. For the 5–9 age-group, the index used was based on the raw number of suicides for each year. Recent analyses of suicide have used the JoinPoint Regression Program (National Cancer Institute, 2010), in which the time-trend in suicide rate is modeled as a sequence of piecewise linear segments (usually beginning, middle, and ending segments) that intersect at the so-called join-points (e.g., Hedegaard et al., 2018; Miron et al., 2019). The program output includes estimates of the join-points, as well as of the annual percent change (APC) during each segment, so that, for example, for a trend consisting of 3 linear segments, 6 (2 [parameters] * 3 [segments]) parameter estimates are returned. However, in the present study, I applied the more familiar polynomial regression models for two reasons. First, for all of the trends examined, a model, in which Time (indexing “year”) is assumed possibly to have both linear and quadratic effects, provides a satisfactory description of the observed trends. These time-trends are approximately linear for a few groups, but are U-shaped for most groups (see Figures 1, 2, and 3 in the Results Section below). Second, to account for a U-shaped trend, the JoinPoint Regression Program estimates 6 structural parameters, as already noted, while the quadratic regression model estimates only 3 parameters. Therefore, the quadratic regression model was the more parsimonious choice given the present purpose of description. The statistical analysis relied on the regression functions in R, lm() for the data on suicide rate in the older groups, and glm(), for the data on number of suicides in the 5–9 age-group (R Core Team, 2020).

Figure 1.

Probability of 1 or more suicides versus year, for males and females in the 5–9 age-group. Note. N = 58 data points. For Black children aged 5–9, the probability of 1 or more suicides decreases from 1990 to about 2000, and then increases thereafter.

Figure 2.

Suicide rates by year for females in the 10–14, 15–19, 20–24, and 25–29 cohorts. Note. N = 116 data points. For the youngest cohort (10–14), suicide rate increases from 1990 to 2018. For the older cohorts, suicide rate decreases for the first 8–12 years of the period, and increases thereafter.

Figure 3.

Suicide rates by year for males in the 10–14, 15–19, 20–24, and 25–29 cohorts. Note. N = 116 data points. For all cohorts, suicide rate decreases from 1990 to about 2007, and then increases thereafter.

Results

The results for the 5–9 age-group are presented separately from those of the older groups. In any given year, there are few suicides within this youngest group and the available CDC data consist of the absolute number of suicides each year, as opposed to the rate per 100,000 people of suicides among the older groups. Accordingly, for this youngest group, logistic regression was used to analyze the data. The familiar regression functions in R were used, lm() for the older groups, and glm() for the 5–9 age-group (R Core Team, 2020). These functions generated significance tests for the regression coefficients, and then complementary R functions were used to assess the practical importance of the observed effects, that is, their “effect sizes.” (For an accessible discussion of “effect size,” see, e.g., Howell, 2010, pp. 159–165. Also, see Mittlböck & Schemper [1996] for a more technical discussion). Given the linear regression results from lm(), effect size was measured as the “proportion of variance explained,” that is, as an R² type of statistic. The function, effect.size(), within the R package, “yhat,” provides these statistics (Nimon et al., 2021). Given the logistic regression results from glm(), there is no agreed upon measure of effect size that is analogous to R², although there are many competing measures, each with limitations. These measures include the so-called pseudo-R² statistics that are based on deviance, a variable that is analogous to variance and is an inverse measure of the goodness-of-fit of the logistic model to the outcome data. McFadden’s pseudo-R² (McFadden, 1974) was used as the index of effect size, primarily because there is some agreement that values of this index from “.2 to .4 represent excellent fit. (see the Stack Exchange discussion of pseudo-R² at https://stats.stackexchange.com/questions/82105/mcfaddens-pseudo-r2-interpretation).

Number of Suicides Among the 5–9 Age-Group

As just stated, the available data contained the absolute number of suicides each year for this age-group. During the 29-year period 1990–2018, there were 10 suicides among Black females and 35 suicides among Black males. There was a total of 14 suicides during the first 15 years (1990–2004), and 31 suicides in the remaining 14 years. Also, there were 6 years in which there were 1 or more female suicides, and 7 years in which there were no male suicides. To analyze these data, a categorical variable, Sui5-9, was defined, separately for males and females, such that Sui5-9 = 0, if there were no suicides in a given year, and Sui5-9 = 1, if there were 1 or more suicides that year. This coding generated a time-series of 29 0’s and 1’s for Black females and a similar one for Black males, for a total of 58 data points. Logistic regression was then applied to these 58 data points, with the outcome variable being the logarithm of the odds that Sui5-9 = 1, and the predictor variables being Gender, a linear effect of Time and a quadratic effect of Time. Time was set at 1 for 1990, 2 for 1991, …, and 29 for 2018; and the quadratic effect, TimeSq, was defined as (Time – 15)². The Gender by Time interactions were not statistically significant (p’s > .8), and we reported the results of the regression without these interactions. Table 1 presents the regression coefficients, and the associated standard errors, z-scores, and significance levels.

Table 1.

Logistic Regression for suicide among the 5-9 age group by Time and Gender.

Coefficient	Estimate	Std. error	z-value
(Intercept)	−4.933	1.421	−3.472***
Time	.131	.052	2.505*
TimeSq	.016	.007	2.345*
Gender male	3.446	.904	3.811***

Note. Results for the logistic regression of whether or not there is a suicide among the 5–9 age-group in a year on Time (1 = 1990, 2 = 1991, …, 29 = 2018) and Gender. The regression coefficients in this Table were used to generate the curves in Figure 1. N = 58 data points. Predictor variables: Gender Male (1 = male, 0 = female), Time, and TimeSq (defined as (Time – 15)²).

Summary Statistics: Null deviance = 80.336, 57 degrees of freedom; and Residual deviance = 47.440, 54 degrees of freedom. McFadden’s pseudo-R² = 1 – (47.44/80.336) = .41.

*p < .05, **p < .01, ***p < .001.

The odds of 1 or more suicides were greater for Black males than Black females (p < .001), reflecting there were more than 3 times as many suicides among males (35) as among females (10) between 1990 and 2018. The linear effect of Time was statistically significant (p = .012), reflecting the doubling in the number of suicides from 14 during the first 15 years of the period to 31 during the next 14 years. However, this increase was not consistent over the years. For Black males and females, the odds of a suicide decreased from 1990 to about 2000, and increased in the latter years of the period (see Figure 1). This U-shaped trend was captured in the statistically significant (p = .019) quadratic effect of Time. In Figure 1, the points (i.e., circles and triangles) represent the observed values of Sui5-9 each year for males and females. The curves in Figure 1 show the probability of 1 or more suicides for each year, as estimated by logistic regression.

To consider the practical importance of the Gender and Time effects, the McFadden pseudo-R² index that is applicable to logistic regression was applied. The deviance statistic is computed for a null model that included no predictor variables, and again for the fitted model with all the predictors. D₀ and D₁ denoted these deviances, respectively, so that the reduction in deviance due to the predictors was D₀ − D₁. Then the McFadden pseudo-R² index was defined as the ratio of this reduction in deviance to the null deviance, namely, (D₀ − D₁)/D₀. In this sense, this index is analogous to R², the “proportion of explained variance” attributable to the predictors in a linear regression. In Table 1, McFadden’s pseudo-R² is calculated to be .41, which indicated that the present logistic model offered a satisfactory fit to the data.

In using the fitted logistic model to predict the time-course of suicide probability for a few years beyond 2018, it can be seen in Figure 1 that the probability for males was already at the asymptote of 1, whereas that for females in 2018 was still in a steep part of the logistic curve. Therefore, even though the number of suicides would be likely to have the same gradual increase for males and females during the next 2 or 3 years, the probability of 1 or more suicides would be expected to increase more sharply for females than males in this age cohort.

For the 10–14 and older age-groups, analyses were conducted for the suicide rates. The number of suicides in the 10–11 age-group for the period 1990–2018 was 238. This total was compared to that of the 5–9 age-group, namely, 45, to verify that pooling the 5–9 and 10–11 age-groups into a 5–11 age-group (as did Bridge et al., 2015) yielded an age-group that was heavily skewed (84%) toward the 10–11 age-group. This finding supported my decision to exclude those aged 10–11 when assessing suicide trends in very young children.

Suicide Rates Among Older Cohorts

The suicide rates over years for the 10–14, 15–19, 20–24, and 25–29 age-groups (or cohorts) are plotted for Black females in Figure 2, and Black males in Figure 3. The observed means are shown as points, and the values fitted by multiple regression are shown as curves. As noted in these Figures, suicide rates were higher for males than females, and Time (i.e., Year) appeared to have linear and quadratic effects, as is seen in Figure 1. These time-trends were also consistent with those observed by Miron et al. (2019) among adolescents and young adults in the USA. Further, Figures 2 and 3 have shown that the differences in suicide rate among age cohorts were greater for males than females. This latter observation suggested that the Gender * Cohort interaction was an important predictor of suicide rate in older cohorts.

Accordingly, a linear regression model was fitted in which Gender was included as a predictor that interacted with Cohort, Time, and TimeSq (the quadratic effect of Time). Table 2 presents the resulting regression coefficients, and the associated standard errors, t-scores, and significance levels. As seen in Table 2, this model accounted for 96.9% of the variance in suicide rate, and all the effects were statistically significant. The R package, “yhat,” was used to decompose this total explained variance into components associated with each possible combination of predictor variables. It was found that 6 of the 127 possible combinations of predictors accounted for 94.8% of the variance in suicide rate, and all of these combinations included the Gender:Cohort interaction term.¹ While the full implications of this decomposition were unclear, it was concluded that the Gender * Cohort interaction was the most important source of explained variance and more analysis is required.

Table 2.

Linear regression of suicide rate among the 10–14, 15–19, 20–24, and 25–29 cohorts for the years 1990–2018.

Coefficients	Estimate	Std. error	t-value
(Intercept)	−.310	.369	−.842
Gender male	2.734	.521	5.246***
Cohort 15–19	1.316	.358	3.673***
Cohort 20–24	2.190	.358	6.114***
Cohort 25–29	2.394	.358	6.684***
Time	.039	.015	2.565*
TimeSq	.006	.002	2.973**
Gender male: Cohort 15–19	6.316	.507	12.467***
Gender male: Cohort 20–24	15.333	.507	30.266***
Gender male: Cohort 25–29	14.889	.507	29.390***
Gender male: Time	−.155	.021	−7.231***
Gender male: TimeSq	.011	.003	3.935***

Note. N = 232 data points. Predictor variables: Gender Male (1 = male, 0 = female), Cohort, Time, TimeSq (defined as (Time – 15)²), and the 2-way interactions involving Gender Male.

Summary Statistics: Residual standard error = 1.364, DF = 220.

Multiple R² = .9689, Adjusted R² = .9674. F(11, 220) = 624***.

*p < .05, **p < .01, ***p < .001.

The practical importance of the Gender * Cohort interaction can be assessed in a more straightforward way by comparing the fits of nested regression models. In particular, the model, in which Time is omitted and the only “predictor” is the Gender * Cohort interaction, explained 94.8% of the variance in suicide rates. This finding suggested that the effects involving Time accounted for only (96.9%–94.8%) = 2.1%. Further, the model, in which Time and the Gender:Cohort predictors were omitted and the only predictors were the Gender and Cohort main effects, explained 76.9% of the variance in suicide rates. Thus, the unique effect of the Gender:Cohort predictor was (94.8%–76.9%) = 17.9%, consistent with the results yielded by “yhat.” In sum, this comparison of nested models suggested that the interaction between Gender and Cohort was practically important, whereas the effects of Time were relatively unimportant.

Discussion

The purpose of this study was to examine the incidence of suicide across time (1990–2018) within the 5- to 29-year-old Black population disaggregated by age-group (or cohort), year, and gender. In designing the study, three critical choices were made so as to facilitate a discussion of some discrepancies in the literature (cf., Crosby & Molock [2006], Hedegard et al. [2018] and Miron et al. [2019]). First, the chosen time-period was longer than those used in earlier studies (cf., Hedegard et al. [2018] and Miron et al. [2019]). Second, the 5–9 age-group was separated out from the 10–11 age-group because it was expected that the incidence of suicide would be much greater in the latter age-group (cf., Bridge et al., 2015). Finally, young adults 20 years-old and older were included to see whether suicide rates declined or stayed steady across the older cohorts (cf., Wang et al., 2016).

In general, the time-trends observed in the present study were consistent with those observed by Price and Khubchandani (2019) among African American adolescents, and Miron et al. (2019) among adolescents and young adults in the USA. However, the findings of the study also illustrated how the description of time-trends in suicide depends on the level of aggregation in the data. For example, Bridge et al. (2015) combined the data for the 5–9 and 10–11 age-groups into a 5–11 age-group and pooled data for the years 1993–2012 into four 5-year periods starting with the period 1993–1997 and ending with 2008–2012. They found a significant increase in the suicide rate among Black children aged 5–11, and a significant decline in the suicide rate among White children. By separating the 5–9-year group from the 10–11-year group, the number of suicides was found to be higher in the 10–11 group compared to the 5–9 group (a ratio of about 5 to 1). What this result indicates is that the time-trend observed by Bridge et al. (2015) was largely determined by the 10–11 age-group and says little about the suicide trend within the 5–9 age-group. Also, this finding suggests the need to reexamine suicide among young Black children. Although the number of suicides among the 5–9 age-group was small in absolute terms, suicide in this group is still an important issue that warrants attention.

Turning now to the subpopulations examined in the present study, the specific age-groups included were Black children (5–9 years), pre-teen and adolescents (10–14 years), transitional age youth (15–19 and 20–24 years), and young adults (25–29 years) for males and females. I found that, as in the general population, suicide was more prevalent in older than younger groups, and in males than females. For 9 of the 10 subpopulations defined by age-group and gender, the time-trend of suicide was U-shaped—decreasing during the first decade or so of the study period, and increasing over the last decade or so. The exception was the increasing trend from 1990 to 2018 for females aged 10–14. The finding is important because it sheds light on a hitherto unknown aspect of suicide among the Black population. While it has been generally accepted that suicide among females has been consistently lower than among males, there is no study, to the best of my knowledge, that has documented in detail the actual rate of suicide among Black females in specific age-groups. In documenting the increasing trend in suicide among young, Black females, this study has pinpointed a vulnerable group that can benefit from targeted intervention activities. This finding provides clear evidence of the need to analyze suicide rates among the Black population stratified by age-group and gender. Without this type of analysis, it would be possible to miss this critical information showing the need to engage in intervention activities in specific age-groups of the Black population, rather than to adopt the same general approach for all subpopulations.

The U-shaped trends described above allow for the integration of two apparently contradictory findings, namely, the decrease in suicide rate from 1990 to about 2003 among Non-Hispanic African Americans reported by Crosby and Molock (2006), and the increase in suicide rate in the last 10–13 years in the general population reported by Miron et al. (2019). These opposite trends are best viewed as part of the same U-shaped trend observed over the past 30 years. Without the use of this longer time-period, it is easy to misunderstand the time-course of suicide rate in the Black population.

Among the older cohorts in the study, the data showed that the suicide rate for Black males was the highest among transitional age youth (20–24 years old), with the rate among young adults (25–29 years old) being slightly lower, reflecting the point at which the suicide rate among Black males starts to decline. Among Black females, the suicide rate was highest among young adults (aged 25–29), indicating that suicide rate among Black females increases across age cohorts. These findings are similar to those reported by Wang et al. (2016) for the period 1983–2012. Wang et al. (2016) reported that the risk for suicide among Black males increased with age, peaking in the age-group 20–24; whereas among Black females, the risk peaked in the age-group 25–29.

The data also showed that the suicide rate among Black females has been increasing since 2008 for all the age-groups in this study, continuing past adolescence, transition age, and well into young adulthood (25–29 years). This finding supports the need for a more fine-tuned analysis of suicide among the Black population as presented in this study. Without this detailed time-trend analysis, one could easily miss information that is necessary to guide public policy, research activities, and intervention programs. This information can be critical, because the design of appropriate suicide prevention programs for a subpopulation is likely to be different when the suicide rate is decreasing from when it is increasing.

Although no systematic analysis of racial differences in suicide rate was conducted, a comparison with data from Miron et al. (2019) for 2015 showed that the Black suicide rate for both males and females in the 15–19 and 20–24 age-groups was lower than the national average. In seeking to explain the pattern of suicide within the Black population, one might consider certain structural economic factors. For example, during the period 2001–2007, the American economy suffered from the collapse of the dot.com economy, the financial crisis of the banking and monetary system, and the collapse of the housing market (Mian & Sufi, 2015). Economic recovery started in 2008 with the passage of the Emergency Economic Stabilization Act and the American Recovery and Reinvestment Act in 2009, and continued through 2018 (Wallach, 2015). Paradoxically, however, suicide rates among African Americans were relatively constant during this period of economic recession but increased during the period of recovery. Further analysis is required to explore what structural factors may impact suicide behavior among African Americans in the 10–29 age-range.

The increase in suicide in recent years is particularly troubling among Black females, a group traditionally not included in any large extent in the research on suicide in the Black community. As shown in the data, Black females have had generally low rates of suicide, but the rates have been increasing since 2004. Young Black women are exposed to several stressors and strains that put them at higher risk of suicidal behavior. Although many of the risk factors associated with suicide behavior seem to be similar across racial and cultural groups, there are additional risk factors that are unique to the Black population. Lamis et al. (2017) found that childhood sexual abuse and intimate partner violence were significantly related to suicidal ideation among low-income Black women aged 18–56. In a study on colorism among Black female adolescents, Abrams et al. (2020) indicated that the intersectionality of racism and sexism, as manifested in colorism beliefs about body image that are internalized, can have negative effects on Black females’ social and psychological well-being. In this vein, colorism becomes an additional risk factor for suicidal behavior that Black females have to confront as part of their lived experience. These findings are also critical in understanding the way in which music has influenced suicidal ideation and behavior. Kresovich et al. (2020), in an “analysis of 125 of the most popular rap songs in the United States during the period 1998–2018, found a statistically significant increase in the proportion of rap songs referencing suicide, depression, and metaphors suggesting mental health struggles” (p. 4). They further stated, “all of the references to suicide or suicidal ideation in our sample were found among the popular songs taken from 2008 to 2018” (Kresovich et al., 2020, p. 4). It is important to note that this increase in suicide ideation and mental health issues in rap music coincided with the general increase in suicide among Black male and female adolescents during the time period 2004–2018. With or without adult supervision and control, rap music can become a subtle way of shaping and influencing behavior.

The findings about the increase in suicide rates among the Black population in recent years are important to consider given the current COVID-19 pandemic. The pandemic has exposed many racial and gender disparities in normal healthcare delivery in the USA and other countries (cf., Khan et al., 2020; Patrick et al., 2020). Also, the activities related to the control of the pandemic (e.g., physical distancing and stay-at-home orders) are difficult to maintain and these difficulties are associated with mental health conditions, such as anxiety, stress, and social isolation (cf., Lee, 2020; Usher & Bhullar, 2020). These health conditions, in turn, are risk factors for suicide and, in this way, the pandemic has been the likely source of an increase in suicidal ideation and suicides. For example, Czeisler et al. (2020) found in a study of 5,412 people over 18 years old that more than 40% of respondents reported at least one adverse mental or behavioral health condition. These conditions included anxiety, depression, and increased substance abuse to cope with stress or emotions related to COVID-19. This figure was higher among younger respondents, with 74.9% of respondents aged 18–24, and 51.9% of respondents aged 25–44 having at least one adverse mental or behavioral health condition. Further, they noted that 15% of Black respondents reported having seriously considered suicide in the month preceding the survey, and that suicidal ideation was more prevalent among males than among females.

In one of the few studies on suicide mortality during the COVID-19 pandemic, Bray et al. (2021) calculated the number of suicide deaths per day for three periods, January 1–March 4, March 5–May 7, and May 8–July 7 in 2020. They described these periods as pre-COVID-19, progressive closure, and progressive reopening, respectively. For comparison, they looked at the same three periods for the years 2017–2019 and, for each period, computed the mean suicide rate across these 3 years. For the pre-COVID-19 period, Bray et al. reported that “daily suicide mortality in 2020 did not differ from the means in 2017 to 2019 for either race” (p. 3). However, for the period of progressive closure (March 5 to May 7), when deaths due to COVID-19 peaked and Maryland was in lock-down, they found that suicide mortality increased (.34 per day in 2020 vs. .18 per day for 2017–2019) among Blacks in Maryland, and decreased among Whites (.67 per day in 2020 vs. 1.22 per day in 2017–2019). During the reopening phase in Maryland (May 8 to July 7), there was no difference from the historical values for Blacks (.23 per day in 2020 vs. .29 per day in 2017–2019), but a decrease among Whites (.79 per day in 2020 vs. 1.13 per day in 2017–2019). Although this report did not provide age-group or gender data, the overall trend is of concern. Another observation that is particularly worrisome is a recent report that the number of suicides among women in Japan in 2020 was “nearly 15 percent more than in 2019” (Rich & Hida, 2021, p. 1).

Limitations

The major limitation of this study is the quality of the national suicide data. The validity and reliability of suicide statistics have been questioned for quite some time. According to Warshauer and Monk (1978), suicide among Blacks tended to be underreported, partly as a result of the change in the classification of suicide from the seventh to eighth revision of the International Classification of Deaths. Other researchers have focused on issues with the cause of death listed on the death certificate (Sehdev & Hutchins, 2001). Several scholars (e.g., Mohler & Earls, 2001; Rocket et al., 2006, 2010; Rockett, 2017) have indicated that misclassification of cause of death is the major factor affecting the validity and reliability of suicide statistics; that suicide is undercounted and underreported; and that this undercount is more common for Blacks than for Whites and for females than males. However, these concerns about data quality apply to most research on suicide because the national data set is the best available.

Another limitation of this study is that only national statistics were analyzed and not data at the state or urban/rural levels. This type of detailed information is important for public health agencies and community-based organizations that work at the local level to develop and implement targeted suicide prevention and intervention activities for specific groups. National suicide data are insufficient for these types of local activities.

Future Research Directions

The findings in this study open a wide array of public health issues and research opportunities on suicide in the Black population, especially among those older than 29 years, the cutoff in the present study. While information about time-trends in suicide rate is important, the present calculations of “effect size” suggest that, even more important, would be the examinations of the many ways in which Gender and Cohort (i.e., Age) interact with other risk factors, like colorism and body image, to influence suicide ideation and suicide rate. Another topic for future research by psychologists and suicidologists could be the influence of rap music and other social media influences on suicide-related beliefs, attitudes, and behavior, especially among Black females.

Protective factors have traditionally insulated young Black females from the risk factors associated with suicide. However, the recent increase of suicidal rates requires further analysis to determine if and how traditional protective factors for Black females may be eroding or how suicidal ideation may be having a greater impact among young Black females (10–14 and 15–19) in ways that could result in increasing suicide rates. Additional research is needed to examine whether protective factors have declined, and what specific aspects of the protective factors have declined to account for the rise in suicide rates in recent years.

Further, the COVID-19 pandemic represents a perfect storm for increases in risk factors that are associated with suicide ideation and behavior. Therefore, researchers and policy-makers are urged to cooperate in trying to understand and address the short- and long-term effects of the pandemic on suicide rates in all demographic groups.

Implications for Intervention

The general increase in suicide among the Black population in recent years highlights the need for intervention and prevention services to address this public health issue. New and creative intervention services should be developed to enhance protective factors that traditionally insulated young, Black females from the risk factors associated with suicide and suicidal behavior (e.g., Hollingsworth et al., 2016; Morrison & Hopkins, 2019; Wadsworth et al., 2013). These new interventions should be grounded in African and African American cultural orientation and the science of Black human functioning (i.e., Black Psychology). The logic of African-centered programs is based on the principle of introducing and supplementing traditional African American cultural values, like cooperation, interdependence, communalism, spirituality, and collective responsibility that have historically protected African Americans. These values serve to increase resiliency and positive sense of self-worth and can serve as protective factors.

Conclusion

In this study, I examined time-trends for suicide among the Black population, 5–29 years old, disaggregated by 5-year age-groups and gender. I found that suicide rates were generally higher for Black males than Black females in all age-groups studied. Also, the number of suicides among the 5–9 age-group was small, and that the majority of the suicides in the age-group 5–11 came from the 10–11 age-group. One of the critical findings was the on-going increase in suicide among Black females, particularly in the 10–14 age-group, for which the rate has been increasing since 1990. Suggestions are provided for research on risk factors and protective factors associated with suicide. In addition, implications for intervention practices that may help alleviate the rising rate of suicide, particularly among Black females, are provided.

Footnotes

Acknowledgments

I would like to thank Ewart Thomas for conducting the statistical analysis of the data for this study and general consultation.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Lawford L. Goddard

Note

References

Abrams

J. A.

Belgrave

F. Z.

Williams

C. D.

Maxwell

M. L.

(2020). African American adolescent girls’ beliefs about skin tone and colorism, Journal of Black Psychology, 46(2-3), 169–194. https://doi.org/10.1177/0095798420928194

Bray

M. J. C.

Daneshvari

N. O.

Radhakrishnan

Cubbage

Eagle

Southall

Nestadt

P. S.

(2021). Racial differences in statewide suicide mortality trends in Maryland during the coronavirus disease 2019 (COVID-19) Pandemic. JAMA Psychiatry, 78(4), 444–447. https://doi.org/10.1001/jamapsychiatry.2020.3938

Bridge

J. A.

Asti

Horowitz

L. M.

Greenhouse

J. B.

Fontanella

C. A.

Sheftall

A. H.

Campo

J. V.

(2015). Suicide trends among elementary school–aged children in the United States from 1993 to 2012. JAMA Pediatrics, 169(7), 673–677. https://doi.org/10.1001/jamapediatrics.2015.0465

Centers for Disease Control and Prevention (CDC) . (2019a). Web-based injury statistics query and reporting system. Fatal Injury Reports, National, Regional and State, 2013- 2017. https://www.cdc.gov/injury/wisqars/index.html

Centers for Disease Control and Prevention . (2019b). Web-based injury statistics query and reporting system: Leading causes of death reports, 1981-2019. https://www.cdc.gov/injury/wisqars/index.html

Crosby

A. E.

Molock

S. D.

(2006). Introduction: Suicidal behaviors in the African American community, Journal of Black Psychology, 32(3), 1–9. https://doi.org/10.1177/0095798406290552

Czeisler

M. É.

Lane

R. I.

Petrosky

Wiley

J. F.

Christensen

Njai

Weaver

M. D.

Robbins

Facer-Childs

E. R.

Barger

L. K.

Czeisler

C. A.

Howard

Rajaratnam

S. M. W.

(2020). Mental health, substance use, and suicidal ideation during the COVID-19 pandemic — United States, June 24–30, 2020. CDC Morbidity and Mortality Weekly Report (MMWR), 69(32), 1049–1057. http://doi.org/10.15585/mmwr.mm6932a1externalicon

Durant

Mercy

Kresbnow

Simon

Potter

Hammond

W. R.

(2006). Racial differences in hopelessness as a risk factor for a nearly lethal suicide attempt. Journal of Black Psychology, 32(3), 285–302. https://doi.org/10.1177/0095798406290468

Fu-Kiau

K. K. B.

Lukondo-Wamba

A. K.

(2000). Kindezi :The Kongo art of babysiting. Black Classic Press.

10.

Griffin-Fennell

Williams

(2006). Examining the complexities of suicidal behavior in the African American community. Journal of Black Psychology, 32(3), 303–319. https://doi.org/10.1177/0095798406290469

11.

Hedegaard

Curtin

S. C.

Warner

(2018). Suicide mortality in the United States. NCHS data brief, No 330. National Center for Health Statistics. https://www.cdc.gov/nchs/products/index.htm

12.

Hedegaard

Curtin

S. C.

Warner

(2020). Increase in suicide mortality in the United States. NCHS data brief, No 362. National Center for Health Statistics. https://www.cdc.gov/nchs/products/index.htm

13.

Hollingsworth

D. W.

Wingate

L. R.

Tucker

R. P.

O’Keefe

V. M.

Ashley

(2016). Hope as a moderator of the relationship between interpersonal predictors of suicide and suicidal thinking in African Americans. Journal of Black Psychology, 42(2), 175–190. https://doi.org/10.1177/0095798414563748

14.

Joe

(2006). Explaining changes in the patterns of Black suicide in the United States from 1981 to 2002: An age, cohort, and period analysis. Journal of Black Psychology, 32(3), 262–284. https://doi.org/10.1177/0095798406290552

15.

Kaslow

N. J.

Jacobs

C. H.

Young

S. L.

Cook

(2006). Suicidal behavior among low-income African American women: A comparison of first-time and repeat suicide attempters. Journal of Black Psychology, 32(3), 349–365. https://doi.org/10.1177/0095798406290459

16.

Khan

K. S.

Mamun

M. A.

Griffiths

M. D.

Ullah

(2020). The mental health impact of the COVID-19 pandemic across different cohorts. International Journal of Mental Health Addiction, 20, 380–386. https://doi.org/10.1007/s11469-020-00367-0

17.

Kohlberg

(1958). The developmental modes of thinking and choices in years 10-16. University of Chicago.

18.

Kresovich

Collins

M. K. R. C.

Riffe

Carpentier

F. R. D.

(2020). A content analysis of mental health discourse in popular rap music. JAMA Pediatrics.

19.

Lamis

D. A.

Cavanaugh

C. E.

Anastasiades

M. H.

Garcia-Williams

Anderson

Kaslow

N. J.

(2017). Intimate partner sexual coercion mediates the childhood sexual abuse–suicidal ideation link among African American women. Journal of Black Psychology, 43(3) 305–324. https://doi.org/10.1177/0095798416644885

20.

Lee

(2020). Mental health effects of school closures during COVID-19. Lancet Child Adolescence Health, 4(6), Article 421. https://doi.org/10.1016/S2352-4642(20)30109-7

21.

McFadden

(1974). The measurement of urban travel demand. Journal of Public Economics, 3, 303–328. https://doi.org/10.1016/0047-2727(74)90003-6

22.

Mian

Sufi

(2015). House of debt: How they (and you) caused the great recession, and how we can prevent it from happening again. University of Chicago Press.

23.

Miron

Wilf-Miron

Kohane

I. S.

(2019). Suicide rates among adolescents and young adults in the United States, 2000-2017. JAMA, 321(23), 2362–2364.

24.

Mittlböck

Schemper

(1996). Explained variation for logistic regression. Statistics in Medicine, 15, 1987–1997.

25.

Mohler

Earls

(2001). Trends in adolescent suicide: Misclassification bias? American Journal of Public Health, 91(1) 150–153. https://doi.org/10.2105/AJPH.91.1.150

26.

Molock

S. D.

Puri

Matlin

Barksdale

(2006). Relationship between religious coping and suicidal behaviors among African American adolescents. Journal of Black Psychology, 36(3) 366–389. https://doi.org/10.1177/0095798406290466

27.

Morrison

Hopkins

(2019). Cultural identity, Africultural coping strategies, and depression as predictors of suicidal ideations and attempts among African American female college students. Journal of Black Psychology, 45(1), 3–25. https://doi.org/10.1177/0095798418813511

28.

National Cancer Institute (2010). Joinpoint regression program.

29.

Nimon

Oswald

Roberts

(2021). yhat: Interpreting regression effects. R package version 2.0-3. https://CRAN.R-project.org/package=yhat

30.

Nock

M. K.

Borges

Bromet

E. J.

Cha

C. B.

Kessler

R. C.

Lee

(2008). Suicide and suicidal behavior. Epidemiologic Reviews, 30. https://doi.org/10.1093/epirev/mxn002

31.

Patrick

S. W.

Henkhaus

L. E.

Zickafoose

J. S.

Lovell

Halvorson

Loch

Letterie

Davis

(2020). Well-being of parents and children during the COVID-19 pandemic: A national survey. Pediatrics, 146(4): Article e2020016824.

32.

Piaget

(1936). The origins of intelligence in children. International Universities Press.

33.

Price

J. H.

Khubchandani

(2019). The changing characteristics of African-American adolescent suicides, 2001–2017. J Community Health 44, 756–763. https://doi.org/10.1007/s10900-019-00678-x

34.

Ramchand

Gordon

J. A.

Pearson

J. L.

(2021). Trends in suicide rates by race and ethnicity in the United States. JAMA Network Open, 4(5), Article e2111563. https://doi:10.1001/jamanetworkopen.2021.11563

35.

R Core Team . (2020). R: A language and environment for statistical computing. R foundation for statistical computing.

36.

Rich

Hida

(2021). As pandemic took hold, suicide rose among Japanese women. New York Times. https://www.nytimes.com/2021/02/23/world/as-the-pandemic-took-hold-suicide-rose-among-japanese-women.html?searchResultPosition=1

37.

Rockett

I. R. H.

(2017). The gender suicide gap and differential misclassification: A research autobiography. International Review of Modern Sociology, 43(1) 5–32.

38.

Rockett

I. R. H.

Samora

J. B.

Coben

J. H.

(2006). The black-white suicide paradox: Possible effects of misclassification. Social Science & Medicine, 63, 2165–2175. https://doi.org/10.1016/j.socscimed.2006.05.017

39.

Rockett

I. R. H.

Wang

S. S.

Steven

D. D.

Frost

J. L.

Ducatman

A. M.

Walker

R. L.

Kapusta

N. D.

(2010). Race/ethnicity and potential suicide misclassification: Window on a minority suicide paradox? BMC Psychiatry. 10(35). https://doi.org/10.1186/1471-244X-10-35

40.

Sehdev

A. E. S.

Hutchins

G. M.

(2001), Problems with proper completion and accuracy of the cause-of-death statement. Archives Intern Med, 161(2), 277–284.

41.

United States Department of Health and Human Services (US DHHS), Centers for Disease Control and Prevention (CDC), National Center for Health Statistics (NCHS) . (2019). Bridged-race population estimates, United States July 1st resident population by state, county, age, sex, bridged-race, and Hispanic origin. http://www.cdc.gov/nchs/nvss/bridged_race.htm

42.

Usher

Bhullar

(2020). The COVID‐19 pandemic and mental health impacts. Int J Ment Health Nurs, 29(3), 315–318.

43.

Wadsworth

Kubrin

C. E.

Herting

J. R.

(2013). Investigating the rise (and fall) of young Black male suicide in the United States, 1982–2001. Journal of African American Studies, 18(1), 72–91. https://doi.org/10.1007/s12111-013-9256-3

44.

Walker

R. L.

Lester

Joe

(2006). Lay theories of suicide: An examination of culturally relevant suicide beliefs and attributions among African Americans and European Americans. Journal of Black Psychology, 32(3), 320–334. https://doi.org/10.1177/0095798406290467

45.

Wang

Bao

Gao

Xiang

(2016). Age-period-cohort analysis of suicide mortality by gender among White and Black Americans, 1983–2012. International Journal for Equity in Health, 15(107). https://doi.org/10.1186/s12939-016-0400-2

46.

Warshauer

M. E.

Monk

(1978). Problems in suicide statistics for Whites and Blacks. American Journal of Public Health, 68(4), 383–388. https://doi.org/10.2105/AJPH.68.4.383

47.

Wilton

Lelutiu-Weinberger

Frye

Hirshfield

Hoover

D. R.

Downing

M. J.

Lucy

Usher

Koblin

(2018). Characteristics and correlates of lifetime suicidal thoughts and attempts among young Black men who have sex with men (MSM) and transgender women. Journal of Black Psychology, 44(3), 273–289. https://doi.org/10.1177/0095798418771819

Suicide Rates Among 5–29-Year-Old African Americans,1990–2018

Abstract

Keywords

The Need for Disaggregated Data

Purpose of Study

Method

Data Source

Measures

Procedures

Statistical Methodology

Results

Number of Suicides Among the 5–9 Age-Group

Suicide Rates Among Older Cohorts

Discussion

Limitations

Future Research Directions

Implications for Intervention

Conclusion

Footnotes

Acknowledgments

Declaration of Conflicting Interests

Funding

ORCID iD

Note

References