Psychiatric Comorbidities Modify the Association Between Childhood ADHD and Risk for Suicidality: A Population-Based Longitudinal Study

Abstract

Objective: The objective of this study was to evaluate the effect of psychiatric comorbidities on the association between childhood ADHD and suicidality among adults. Method: Subjects were recruited from a population-based birth cohort. Participating adult subjects with childhood ADHD and non-ADHD controls were administered a structured psychiatric interview to assess suicidality and psychiatric comorbidities. Associations were assessed using logistic regression. Results: Compared with controls, ADHD cases were significantly more likely to meet criteria for suicidality. Subjects with childhood ADHD who met criteria for generalized anxiety disorder had a higher than expected risk of suicidality with an observed odds ratio of 10.94 (95% confidence interval [4.97, 24.08]) compared with an expected odds ratio of 4.86, consistent with a synergistic interaction effect. Significant synergistic interactions were also observed for hypomanic episode and substance-related disorders. Conclusion: Childhood ADHD is significantly associated with adult suicidal risk. Comorbidity between ADHD and some psychiatric disorders is associated with a higher suicidal risk than expected.

Keywords

ADHD suicidality adults psychiatric comorbidity epidemiology

Introduction

ADHD has been reported to be frequently accompanied by a broad range of psychiatric comorbidities throughout the life span period from childhood to adulthood (Adler, Spencer, & Wilens, 2015; Biederman et al., 2005; Cortese et al., 2013; Hesson & Fowler, 2015; Kessler et al., 2006; Kooij et al., 2012; Newcorn, Weiss, & Stein, 2007; Rucklidge, Downs-Woolley, Taylor, Brown, & Harrow, 2016; von Gontard & Equit, 2015; Williamson & Johnston, 2015; Yoshimasu et al., 2012). Recent studies have highlighted the strong association between ADHD and elevated risks of completed suicide, suicide attempt, suicidal ideation, and self-harm behaviors during childhood, adolescence, and adulthood (Balazs, Miklósi, Keresztény, Dallos, & Gádoros, 2014; Barbaresi et al., 2013; Furczyk & Thome, 2014; Impey & Heun, 2012; Ljung, Chen, Lichtenstein, & Larsson, 2014). However, as the majority of the previous studies on this issue has been conducted on completed suicide (Impey & Heun, 2012; James, Lai, & Dahl, 2004; Ljung et al., 2014), there is relatively little evidence regarding the associations between ADHD and suicidality, more broadly defined (e.g., suicide attempt and suicidal ideation).

Although ADHD is accompanied by a variety of psychiatric disorders (PDs), it is still unclear to what extent the association between ADHD and suicidality is moderated or mediated by such comorbid disorders. Almost all PDs accompanied by ADHD have themselves been shown to be associated with suicidal risk (Wasserman, 2016). While some studies indicate that the association between ADHD and suicidality is partially mediated by such comorbidities (James et al., 2004; Ljung et al., 2014), other studies showed that the association was fully mediated by comorbid disorders such as mood, anxiety, and substance use disorders among clinic-referred children and adolescents (Balazs et al., 2014), and that linkages between self-harm behavior and ADHD symptomatology might be due primarily to comorbid mental disorders and “emotion-focused coping” (Taylor, Boden, & Rucklidge, 2014). The relationship between suicidal behaviors and depressive mood was also shown to be moderated by ADHD symptoms among college students (Patros et al., 2013).

Furthermore, a history of PDs was shown to modify the gender ratio of completed suicide (Liu, Chen, Cheung, & Yip, 2009). Although undiagnosed ADHD was shown to be a potential risk factor for self-harming behaviors among those below age 21 years, males had more serious medical outcomes (death or major effect) but females were more likely to be admitted to a critical care unit, suggesting a need for gender-specific suicide prevention strategies (Sheikh et al., 2015). However, to our knowledge, no studies examined gender differences on the association between suicidality and childhood ADHD among adults, while accounting for the effects of comorbid PDs.

We have conducted an epidemiological study regarding ADHD and comorbid PDs in a population-based birth cohort (Yoshimasu et al., 2012). The purposes of the current, longitudinal study are to evaluate the extent to which the association between childhood ADHD and suicidality is mediated or moderated by adult psychiatric comorbidities, overall and separately by gender.

Method

Study Setting and Data Source

The capacity for population-based epidemiologic research on psychiatric comorbidity of ADHD in Rochester, Minnesota, is the result of a unique set of circumstances that we have described previously (Katusic et al., 2017; Kurland & Molgaard, 1981; Melton, 1996; Yoshimasu et al., 2012). The study was approved by the Institutional Review Boards of Mayo Clinic and Olmsted Medical Center.

Birth Cohort

Our birth cohort consisted of all children born between January 1, 1976, and December 31, 1982, to mothers who still lived in Rochester at or after the age of 5 years (n = 5,718; 2,956 boys, 2,762 girls) (Katusic et al., 2005a, 2005b). The steps and resources used for identification and follow-up of this birth cohort have been reported (Katusic et al., 2017).

Childhood Identified ADHD Incidence Cases—Identification and Case Definition

Our strategy in identifying childhood ADHD incidence cases consisted of several steps, used multiple sources of information, and relied on recorded history of symptoms, and individual assessment results. Details regarding those research criteria as well as the identification process for childhood ADHD cases (n = 379) were described elsewhere (Barbaresi et al., 2002; Katusic et al., 2005a).

Recruitment of Childhood ADHD Cases and Controls for the Adult Outcome Assessment

The detailed procedures for recruiting childhood ADHD cases and controls for the adult outcome phase of our study have been previously described (Barbaresi et al., 2013). From among the 379 childhood ADHD cases, 232 agreed to participate.

To ensure enrollment of sufficient non-ADHD controls, a random sample of 801 adults from the same birth cohort was also invited to participate; 335 non-ADHD controls (42.1%) participated.

Adult PDs

All participants in the prospective phase of the adult outcomes study were administered the Mini International Neuropsychiatric Interview (M.I.N.I.) by two interviewers who were trained by a child psychologist and developmental pediatrician. The M.I.N.I. is a brief structured diagnostic interview for Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; American Psychiatric Association, 1994) and International Classification of Diseases, Tenth Revision, (ICD-10) mental disorders that includes17 Axis-I disorders (Sheehan et al., 1998; Sheehan et al., 1997). The M.I.N.I. was specifically designed to meet the need for an accurate structured psychiatric interview without heavy burden to the study participants because required time for the interview is very short. In validation studies of the M.I.N.I., comparing it to the Structured Clinical Interview for DSM-III-R, Patient Version (SCID-P) and the Composite International Diagnostic Interview (CIDI), excellent kappa values, sensitivity, and specificity were reported for 17 Axis-I disorders (Sheehan et al., 1998; Sheehan et al., 1997). Of the 17 Axis-I disorders included in the M.I.N.I., we chose to evaluate 11 as potential candidate disorders associated with increased risk of suicide: major depressive disorder, dysthymia, posttraumatic stress disorder, social phobia, obsessive-compulsive disorder, hypomanic episode, generalized anxiety disorder, antisocial personality disorder, alcohol-related disorders, panic disorder, and substance-related disorders. These disorders are conventionally classified into two categories; that is, internalizing and externalizing (including hypomanic) disorders (Kotov et al., 2011).

Assessment of Suicidality

Suicide risk was measured by six items included in the M.I.N.I., five of which were concerned with suicidal thoughts or behaviors within the previous 1 month, while one dealt with lifetime experiences of suicide attempts. Point values (1, 2, 4, 6, or 10) were assigned for each positive responses to each of the six questions (see the Appendix), in accordance with M.I.N.I. 5.0.0 scoring system (January 1, 2004). The calculated score reflected the severity of the current suicidal risk for each subject. The severity of the suicide risk was divided into four strata based on the total score; that is, none (0), low (1-5), moderate (6-9), or high (10+). Because previous suicidal attempt was shown to be very strongly associated with an increased risk of completed suicide (odds ratio [OR] = 16; Yoshimasu, Kiyohara, Miyashita, & The Stress Research Group of the Japanese Society for Hygiene, 2008), 10 points were allotted for reported suicide attempt within the previous month and those subjects are considered to have high suicidal risk. Subjects with a score of at least 1 were regarded as having some degree of suicidality, because in the epidemiological study using M.I.N.I., those with “any” suicidal risk were shown to be rare in nonclinical populations (Takemura, Yoshimasu, Fukumoto, Yamamoto, & Miyashita, 2011; Yoshimasu et al., 2011).

Data Analysis

All analyses were performed using the SAS version 9.3 software package (SAS Institute, Inc., Cary, NC). Demographic and perinatal factors were compared between research-identified childhood ADHD cases and non-ADHD controls (and between participants and nonparticipants in the adult follow-up study) using the two-sample t test for age, Wilcoxon rank sum test for ordinal variables, and the chi-square test for all other variables. Logistic regression models were fit to assess the mediating and moderating effect of each of the 11 comorbid PDs on the association between ADHD (intervening variable) and suicidality (dependent variable). To assess for mediation, the direct effect of ADHD on suicidality was assessed by evaluating the parameter coefficient for ADHD after adjusting for each of the 11 different comorbid PDs in the model one at a time. This direct effect was contrasted with the total effect of ADHD on suicidality. There has been general consensus in the epidemiological community that measuring interaction on the additive (as opposed to multiplicative) scale is most appropriate for assessing the public health importance of interactions (Rothman, 1986). Therefore, for each comorbid PD we evaluated the risk of suicidality within the 4 strata defined by presence of childhood ADHD (yes vs. no) and presence of comorbid PD (yes vs. no). OR_AB, OR_A, and OR_B are the corresponding odds ratios (OR) for those with both childhood ADHD and the PD, childhood ADHD only, and PD only, respectively. Assuming childhood ADHD and PD are independent, the expected joint effect on an additive scale would be OR_A + OR_B – 1. A synergistic interaction effect exists if the observed joint effect, OR_AB, exceeds this expected joint effect. Additional models were fit to examine the mediator and moderator effects after adjusting for the following potential confounders: age at participation, gender, maternal age, maternal education, and marital status at the subject’s birth.

Results

The following results are based on 232 subjects with childhood ADHD and 335 non-ADHD referents who participated in the adult prospective phase of the study. Participating subjects with childhood ADHD were younger (27.0 vs. 28.6 years; p < .001), more likely to be male (p = .021), have mothers with fewer years of education (p = .010), and have younger mothers at birth (p = .027), compared with participating non-ADHD referents (Table 1).

Table 1.

Characteristics of Participants.

Characteristic	Childhood ADHD (n = 232)	Non-ADHD (n = 335)	p ^a
Age at participation, M (SD)	27.0 (2.6)	28.6 (2.2)	<.001
Male gender, n (%)	167 (72.0)	210 (62.7)	.021
Caucasian race, n (%)	231 (99.6)	329 (98.2)	.15
Birth weight <2,500 g, n (%)	10/232 (4.3)	16/334 (4.8)	.79
Parental characteristics at birth
Maternal age, M (SD)	26.0 (4.6)	26.8 (4.4)	.027
Paternal age, M (SD)	28.6 (5.0)	29.0 (5.5)	.35
Maternal years of education, n (%)			.010
<12	13/212 (6.1)	17/302 (5.6)
12	88/212 (41.5)	96/302 (31.8)
13-15	74/212 (34.9)	110/302 (36.4)
16+	37/212 (17.5)	79/302 (26.2)
Paternal years of education, n (%)			.016
<12	10/199 (5.0)	13/290 (4.5)
12	74/199 (37.2)	78/290 (26.9)
13-15	51/199 (25.6)	80/290 (27.6)
16+	64/199 (32.2)	119/290 (41.0)
Married at birth, n (%)	212 (91.4)	313 (93.4)	.36
Current status assessment of participant based on the M.I.N.I.
Suicidality, n (%)			<.001
None (0 points)	181 (78.0)	300 (89.6)
Low (1-5 points)	35 (15.1)	21 (6.3)
Moderate (6-9 points)	7 (3.0)	9 (2.7)
High (10+ points)	9 (3.9)	5 (1.5)

Note. M.I.N.I. = Mini International Neuropsychiatric Interview.

Comparisons evaluated using the two-sample t test for age, the Wilcoxon rank sum test for level of education and suicidality, and the chi-square test for all other variables.

A comparison of the baseline characteristics between participating and nonparticipating subjects with childhood ADHD has been previously described (Barbaresi et al., 2013), showing no significant differences except for higher rates of high school graduation among participants versus nonparticipants (84.3% vs. 64.8%, respectively). There were no significant differences in these factors among participating and nonparticipating non-ADHD controls except that the proportion of males (62.7% vs. 74.2%) and those with congenital anomalies (0% vs. 1.7%) were significantly higher in nonparticipants than participants.

Fifty-one (22.0%) of the childhood ADHD cases met the criteria for suicidality on the M.I.N.I., compared with 35 (10.4%) of the non-ADHD referents (OR = 2.42; 95% confidence interval [CI] [1.51, 3.86]). The majority of subjects meeting the suicidality criteria had severity scores in the low range (35/51 childhood ADHD cases and 21/35 non-ADHD controls), as opposed to moderate and high combined (16/51 childhood ADHD cases and 14/35 non-ADHD controls). We observed a statistically significant direct effect of ADHD on suicidality after controlling for each of the 11 different comorbid PDs one at time, suggesting that that the relationship is not fully mediated by the comorbid disorders (Table 2). These direct effects were minimally impacted after further adjusting for age at participation, gender, and maternal age, education, and marital status at the subject’s birth. Furthermore, upon calculating each direct effect as a percent change of the total effect, we estimated that major depressive disorder accounted for 25.7% of the effect of ADHD on suicidality, followed by occurrence of a hypomanic episode (18.3%) and dysthymia (18.1%).

Table 2.

Evaluation of the Role of Comorbid Psychiatric Disorders as a Mediator of the Association Between Childhood ADHD and Suicidality.

Association between ADHD and suicidality	Unadjusted for covariates			Covariate-adjusted^a
Association between ADHD and suicidality	OR (95% CI)	p	% of the total effect mediated by the comorbid PD	OR (95% CI)	p
Unadjusted	2.42 [1.51, 3.86]	<.001	—	2.58 [1.54, 4.31]	<.001
Adjusted for the presence/absence of each of the comorbid PD separately:
Major depressive disorder	1.93 [1.18, 3.15]	.009	25.7	1.90 [1.10, 3.26]	.021
Dysthymia	2.06 [1.27, 3.35]	.004	18.1	2.12 [1.25, 3.61]	.006
Social phobia	2.12 [1.31, 3.45]	.002	14.7	2.19 [1.29, 3.72]	.004
Obsessive-compulsive disorder	2.12 [1.31, 3.44]	.002	14.7	2.18 [1.28, 3.71]	.004
Generalized anxiety disorder	2.27 [1.40, 3.67]	<.001	7.1	2.27 [1.34, 3.85]	.002
Panic disorder	2.38 [1.47, 3.84]	<.001	1.7	2.50 [1.47, 4.23]	<.001
PTSD	2.28 [1.41, 3.69]	<.001	6.6	2.37 [1.40, 4.01]	.001
Hypomanic episode	2.06 [1.27, 3.34]	.004	18.3	2.23 [1.31, 3.78]	.003
Substance-related disorders	2.16 [1.34, 3.48]	.002	12.7	2.36 [1.40, 4.00]	.001
Antisocial personality disorder	2.20 [1.36, 3.55]	.001	10.8	2.38 [1.41, 4.02]	.001
Alcohol dependence/abuse	2.37 [1.48, 3.79]	<.001	2.4	2.50 [1.49, 4.20]	<.001
Any of the above PDs	1.94 [1.19, 3.15]	.007	24.9	2.01 [1.18, 3.42]	.011

Note. OR = hazard ratio; CI = confidence interval; PD = psychiatric disorders; PTSD = posttraumatic stress disorder.

Multivariable model adjusted for age at participation, gender, and maternal age, education, and marital status at the subject’s birth.

Table 3 demonstrates the role of internalizing comorbid PDs as effect modifiers of the association between childhood ADHD and suicidality. Compared with those with neither ADHD nor Generalized Anxiety Disorder (GAD), the odds ratios of suicidality were 2.1 for patients with ADHD alone, and 3.7 for patients with GAD alone. The observed joint effect of ADHD and GAD (10.94; 95% CI [4.97, 24.08]) significantly exceeded the joint effect of ADHD and GAD on an additive scale (4.86). This synergistic interaction effect was attenuated in the covariate-adjusted model. Similar patterns were also observed for major depressive disorder, obsessive-compulsive disorder, and panic disorder. However, a significant synergistic interaction was only observed for the internalizing disorder of GAD.

Table 3.

Evaluation of the Role of Internalizing Comorbid Psychiatric Disorders as an Effect Modifier of the Association Between Childhood ADHD and Suicidality.

Internalizing comorbid psychiatric disorder	Unadjusted				Covariate-Adjusted^a
Internalizing comorbid psychiatric disorder	No. (%) with suicidality	OR (95% CI)	p	Expected OR^b	OR (95% CI)	p	Expected OR^b
MDD
No MDD, No ADHD (n = 326)	31 (9.5)	1.00			1.00
No MDD, ADHD (n = 202)	35 (17.3)	1.99 [1.19, 3.35]	.009		2.02 [1.14, 3.58]	.016
MDD, No ADHD (n = 9)	4 (44.4)	7.61 [1.94, 29.84]	.004		9.28 [2.23, 38.55]	.002
Joint effect: MDD, ADHD (n = 30)	16 (53.3)	10.88 [4.85, 24.38]	<.001	8.60	10.30 [4.48, 23.71]	<.001	10.30
DYS
No DYS, No ADHD (n = 331)	32 (9.7)	1.00
No DYS, ADHD (n = 216)	41 (19.0)	2.19 [1.33, 3.60]	.002		2.32 [1.34, 4.01]	.003
DYS, No ADHD (n = 4)	3 (75.0)	28.03 [2.83. 277.45]	.004		31.76 [3.00, 335.77]	.004
Joint effect: DYS, ADHD (n = 16)	10 (62.5)	15.57 [5.31, 45.66]	<.001	29.22	12.23 [4.05, 36.90]	<.001	33.08
SP
No SP, No ADHD (n = 331)	32 (9.7)	1.00			1.00
No SP, ADHD (n = 219)	42 (19.2)	2.22 [1.35, 3.64]	.002		2.34 [1.36, 4.04]	.002
SP, No ADHD (n = 4)	3 (75.0)	28.03 [2.83, 277.46]	.004		35.55 [3.36, 376.01]	.003
Joint effect: SP, ADHD (n = 13)	9 (69.2)	21.02 [6.13, 72.14]	<.001	29.25	18.97 [5.31, 67.76]	<.001	36.89
OCD
No OCD, No ADHD (n = 327)	32 (9.8)	1.00
No OCD, ADHD (n = 213)	39 (18.3)	2.07 [1.25, 3.42]	.005		2.17 [1.25, 3.79]	.006
OCD, No ADHD (n = 8)	3 (37.5)	5.53 [1.26, 24.23]	.023		7.46 [1.63, 34.19]	.010
Joint effect: OCD, ADHD (n = 19)	12 (63.2)	15.80 [5.81, 43.00]	<.001	6.60	16.94 [5.99, 47.87]	<.001	8.63
GAD
No GAD, No ADHD (n = 305)	27 (8.9)	1.00			1.00
No GAD, ADHD (n = 199)	34 (17.1)	2.12 [1.24, 3.64]	.006		2.17 [1.20, 3.91]	.010
GAD, No ADHD (n = 30)	8 (26.7)	3.74 [1.52, 9.21]	.004		3.71 [1.46, 9.48]	.006
Joint effect: GAD, ADHD (n = 33)	17 (51.5)	10.94 [4.97, 24.08]	<.001	4.86^c	9.95 [4.40, 22.47]	<.001	4.88
PD
No PD, No ADHD (n = 318)	28 (8.8)	1.00
No PD, ADHD (n = 215)	42 (19.5)	2.51 [1.50, 4.20]	<.001		2.66 [1.52, 4.64]	<.001
PD, No ADHD (n = 17)	7 (41.2)	7.25 [2.56, 20.53]	<.001		6.81 [2.33, 19.94]	<.001
Joint effect: PD, ADHD (n = 17)	9 (52.9)	11.65 [4.17, 32.58]	<.001	8.76	10.51 [3.60, 30.70]	<.001	8.47
PTSD
No PTSD, No ADHD (n = 332)	34 (10.2)	1.00			1.00
No PTSD, ADHD (n = 225)	44 (19.6)	2.12 [1.31, 3.44]	.002		2.18 [1.29, 3.67]	.004
PTSD, No ADHD (n = 3)	1 (33.3)	5.20 [0.49, 55.22]	.17		4.52 [0.36, 57.18]	.25
Joint effect: PTSD, ADHD (n = 7)	7 (100.0)	129.78 [5.96, infinity]	.002	6.32	90.36 [4.26, infinity]	.004	5.70

Note. OR = hazard ratio; CI = confidence interval; MDD = major depressive disorder; DYS = dysthymia; SP = social phobia; OCD = obsessive-compulsive disorder; GAD = generalized anxiety disorder; PD = panic disorder; PTSD = posttraumatic stress disorder.

Multivariable model adjusted for age at participation, gender, and maternal age, education, and marital status at the subject’s birth.

The expected odds ratio under the assumption of additivity. If this estimate is contained within the 95% CI for the observed joint effect of each psychiatric disorder and ADHD, then there is not sufficient evidence to suggest a moderating effect on an additive scale.

Statistically significant (p < .05) synergistic interaction. The observed joint effect of generalized anxiety disorder and ADHD exceeded the expected joint effect on an additive scale. Subjects with childhood ADHD who met the criteria for generalized anxiety disorder on the Mini International Neuropsychiatric Interview (M.I.N.I.) have an unexpectedly high risk of suicide.

The role of externalizing comorbid PDs as effect modifiers of the association between childhood ADHD and suicidality is presented in Table 4. The joint effect of childhood ADHD and hypomanic episode had a higher than expected risk of suicidality with an observed odds ratio of 7.40 (95% CI [3.48, 15.77]) compared with an expected odds ratio of 2.66. This was also true for the joint effect of childhood ADHD and substance-related disorders with an observed odds ratio of 6.62 (95% CI [3.15, 13.91]) compared with an expected odds ratio of 3.02. These synergistic interaction effects remained statistically significant in multivariate analysis.

Table 4.

Evaluation of the Role of Externalizing Comorbid Psychiatric Disorders as an Effect Modifier of the Association Between Childhood ADHD and Suicidality.

Externalizing comorbid psychiatric disorder	Unadjusted				Covariate-Adjusted^a
Externalizing comorbid psychiatric disorder	No. (%) with suicidality	OR (95% CI)	p	Expected OR^b	OR (95% CI)	p	Expected OR^b
HME
No HME, No ADHD (n = 323)	33 (10.2)	1.00			1.00
No HME, ADHD (n = 197)	35 (17.8)	1.90 [1.14, 3.17]	.014		2.01 [1.15, 3.50]	.014
HME, No ADHD (n = 12)	2 (16.7)	1.76 [0.37, 8.37]	.48		1.27 [0.26, 6.34]	.77
Joint effect: HME, ADHD (n = 35)	16 (45.7)	7.40 [3.48, 15.77]	<.001	2.66^c	7.55 [3.39, 16.83]	<.001	2.28^c
SRD
No SRD, No ADHD (n = 313)	31 (9.9)	1.00			1.00
No SRD, ADHD (n = 194)	35 (18.0)	2.00 [1.19, 3.37]	.009		2.20 [1.25, 3.86]	.006
SRD, No ADHD (n = 22)	4 (18.2)	2.02 [0.64, 6.35]	.23		2.63 [0.79, 8.70]	.11
Joint effect: SRD, ADHD (n = 38)	16 (42.1)	6.62 [3.15, 13.91]	<.001	3.02^c	9.52 [4.10, 22.09]	<.001	3.83^c
APD
No APD, No ADHD (n = 322)	33 (10.3)	1.00			1.00
No APD, ADHD (n = 193)	38 (19.7)	2.15 [1.30, 3.56]	.003		2.29 [1.33, 3.97]	.003
APD, No ADHD (n = 13)	2 (15.4)	1.59 [0.34, 7.50]	.56		1.43 [0.29, 7.09]	.66
Joint effect: APD, ADHD (n = 39)	13 (33.3)	4.38 [2.05, 9.34]	<.001	2.74	4.85 [2.12, 11.07]	<.001	2.72
ADA
No ADA, No ADHD (n = 284)	30 (10.6)	1.00			1.00
No ADA, ADHD (n = 171)	35 (20.5)	2.18 [1.28, 3.70]	.004		2.31 [1.30, 4.08]	.004
ADA, No ADHD (n = 51)	5 (9.8)	0.92 [0.34, 2.50]	.87		1.14 [0.41, 3.17]	.81
Joint effect: ADA, ADHD (n = 61)	16 (26.2)	3.01 [1.52, 5.97]	.002	2.10	3.96 [1.85, 8.48]	.002	2.45

Note. OR = hazard ratio; CI = confidence interval; HME = hypomanic episode; SRD = substance-related disorder; APD = antisocial personality disorder; ADA = alcohol dependence/abuse.

Multivariable model adjusted for age at participation, gender, and maternal age, education, and marital status at the subject’s birth.

Statistically significant (p < .05) synergistic interaction. The observed joint effect of hypomanic episode and ADHD exceeded the expected joint effect on an additive scale. Subjects with childhood ADHD who met the criteria for hypomanic episode on the Mini International Neuropsychiatric Interview (M.I.N.I.) have an unexpectedly high risk of suicide. Similar results were observed for substance use disorder.

The impact of gender on the three comorbid PDs with synergistic interactions were further explored in Table 5. Significant synergistic interaction effects were observed for generalized anxiety disorder and hypomanic episode in men, but not in women. These interactions retained statistical significance among males in covariate-adjusted models (data not shown).

Table 5.

Evaluation of the Role of Comorbid Psychiatric Disorders as an Effect Modifier of the Association Between Childhood ADHD and Suicidality, by Gender (Unadjusted Models).

Comorbid PD	Males				Females
Comorbid PD	No. (%) with suicidality	Unadjusted OR (95% CI)	p	Expected OR^a	No. (%) with suicidality	Unadjusted OR (95% CI)	p	Expected OR^a
GAD
No GAD, No ADHD	12/191 (6.3)	1.00			15/114 (13.2)	1.00
No GAD, ADHD	21/149 (14.1)	2.44 [1.16, 5.15]	.019		13/50 (26.0)	2.32 [1.01, 5.34]	.048
GAD, No ADHD	2/19 (10.5)	1.76 [0.36, 8.50]	.48		6/11 (54.6)	7.92 [2.15, 29.21]	.002
Joint effect: GAD, ADHD	11/18 (61.1)	23.44 [7.70, 71.37]	<.001	3.20^b	6/15 (40.0)	4.40 [1.37, 14.14]	.013	9.24
HME
No HME, No ADHD	14/204 (6.9)	1.00			19/119 (16.0)	1.00
No HME, ADHD	22/144 (15.3)	2.41 [1.20, 4.86]	.014		13/53 (24.5)	1.71 [0.77, 3.79]	.19
HME, No ADHD	0/6 (0.0)	1.01 [0.04, 23.65]	.99		2/6 (33.3)	2.63 [0.45, 15.40]	.28
Joint effect: HME, ADHD	10/23 (43.5)	10.22 [3.83, 27.29]	<.001	2.42^b	6/12 (50.0)	5.26 [1.53, 18.07]	.008	3.34
SRD
No SRD, No ADHD	11/191 (5.8)	1.00			20/122 (16.4)	1.00
No SRD, ADHD	19/133 (14.3)	2.73 [1.25, 5.94]	.012		16/61 (26.2)	1.81 [0.86, 3.82]	.12
SRD, No ADHD	3/19 (15.8)	3.07 [0.78, 12.14]	.11		1/3 (33.3)	2.55 [0.22, 29.49]	.45
Joint effect: SRD, ADHD	13/34 (38.2)	10.13 [4.03, 25.45]	<.001	4.80	3/4 (75.0)	15.30 [1.51, 154.64]	.021	3.36
Any PD
No PD, No ADHD	8/139 (5.8)	1.00			10/97 (10.3)	1.00
No PD, ADHD	9/82 (11.0)	2.02 [0.75, 5.46]	.17		4/31 (12.9)	1.29 [0.37, 4.44]	.69
PD, No ADHD	6/71 (8.5)	1.51 [0.50, 4.54]	.46		11/28 (39.3)	5.63 [2.07, 15.33]	<.001
Joint effect: PD, ADHD	23/85 (27.1)	6.08 [2.57, 14.34]	<.001	2.53^b	15/34 (44.2)	6.87 [2.68, 17.61]	<.001	5.92

Note. PD = psychiatric disorder; OR = hazard ratio; CI = confidence interval; GAD = generalized anxiety disorder; HME = hypomanic episode; SRD = substance-related disorder.

Statistically significant (p < .05) synergistic interaction.

Discussion

This epidemiological study employed a population-based birth cohort, with carefully defined, research-identified childhood ADHD cases and their comorbid adult psychiatric conditions including suicidality as assessed by the M.I.N.I. These data add significantly to our understanding of the role of psychiatric comorbidities on the association between childhood ADHD and suicidal risk in adulthood.

Our study clearly indicates a strong association between childhood ADHD and suicidality, which is largely consistent with previous studies (Balazs & Kereszteny, 2017; Impey & Heun, 2012; James et al., 2004). This association was only partially mediated by comorbid mental disorders such as mood disorders or substance-related disorders, which is also consistent with previous findings (Hurtig, Taanila, Moilanen, Nordström, & Ebeling, 2012; Ljung et al., 2014), while majority of the previous studies have highlighted the mediating role of psychiatric comorbidities (Balazs & Kereszteny, 2017). Although 29.3% of our childhood ADHD subjects fulfilled the diagnostic criteria for adult ADHD, some symptoms of ADHD are considered to persist into adulthood for 60% to 75% even if subjects do not meet the full diagnostic criteria for adult ADHD (Wilens, Biederman, & Spencer, 2002). Thus, comorbid PDs combined with ADHD symptoms may play an important role with respect to the risk for suicidality even among adults whose residual ADHD symptoms are not sufficient to reach the diagnostic threshold for adult ADHD.

Significant or marginally significant synergistic interactions with ADHD on the risk for suicidality were observed for specific psychiatric comorbidities including GAD, hypomanic episode, and substance-related disorders, which indicates that subjects with childhood ADHD who met the criteria for these disorders have higher risks of suicide than expected from the combined effects of ADHD alone or each of these disorders alone. This tendency was more apparent in men than women, which may be due to the smaller number of women in our cohort of childhood ADHD cases. Interestingly, one hospital-based epidemiological study (Perugi et al., 2013) employing the M.I.N.I. and DSM-IV criteria reported that patients with borderline personality disorder (BPD) and bipolar disorder reported significantly more comorbid substance abuse, anxiety disorders, ADHD, and history of suicide attempt than those without BPD. Another study using DSM-5 (American Psychiatric Association, 2013) and research-based diagnostic criteria reported that depressive mixed state, defined as the presence of three or more manic/hypomanic features, was associated with ADHD, lifetime suicidal attempt, and substance-related disorders (Perugi et al., 2015). Our finding of increased risk for suicidality among males with “any” psychiatric comorbidity reflects the overall moderating effect of these comorbidities on the association between ADHD and suicidality.

Another study using structural equation modeling showed that an association between offspring suicidal ideation and maternal depression was independently mediated by offspring major depressive disorder and GAD (Hammerton et al., 2015). Results were similar for offspring suicide attempt except for additional evidence of an indirect effect through offspring ADHD symptoms (Hammerton et al., 2015). Although the reason is unclear, these findings suggest that ADHD, GAD, suicidality, substance-related disorders and hypomanic episode may be psychopathologically or genetically linked each other.

We found gender differences for the modifier effects of psychiatric comorbidities on the association between suicidality and ADHD, suggesting that ADHD combined with comorbid GAD or hypomanic episode as well as the presence of “any” mental disorder might be associated with a higher than expected risk of suicidality among men, but not women. Our contrasting findings for men and women with childhood ADHD may be due to limited statistical power because of the fewer number of women in our study. Alternatively, men with ADHD may be more likely to manifest the impulsivity associated with suicidal thinking and acts. This latter hypothesis is consistent with the fact that that male/female ratio was reported to be 5:1 for U.S. 15- to 19-year-old suicide completers (Safer, 1997), and male gender was shown to be a significant predictor of suicide among people with a history of self-harm (Chan et al., 2016).

Limitations

The relatively small sample size of women with ADHD and comorbid mental disorders limits statistical power to identify relationships among ADHD, comorbid PDs, and risk for suicidality among women, which might obscure the effects of psychiatric comorbidities on suicidality among females with ADHD. Furthermore, we did not have sufficient information to evaluate the severity of the psychiatric comorbidities. Although significant synergistic interactions were not observed for some PDs, these might have been observed if subjects consisted exclusively of individuals with more severe PDs. Therefore, potential influence of comorbidities on suicidality might be masked for some disorders. The M.I.N.I. does not include questions that address factors such as quality of life or activities of daily living that would permit some assessment of the severity of comorbid PDs; however, the M.I.N.I. has been shown to be efficient and accurate for diagnosing psychiatric patients in clinical practice (van Vliet & de Beurs, 2007). Finally, it is possible that potential subjects with more severe PDs were less likely to respond to requests to participate in the study, limiting our ability to evaluate these associations.

Conclusion and Future Direction

Our findings demonstrate that childhood ADHD is significantly associated with elevated suicidal risk in adulthood. In addition, the risk for suicidality is higher than expected on an additive scale when adults with a history of childhood ADHD have comorbid generalized anxiety disorder, hypomanic episode, or substance-related disorders. Adults with a history of childhood ADHD and such specific comorbid adult PDs should be regarded as a high-risk population for suicide beyond the adverse effects of each disorder. Adults with a history of childhood ADHD and later psychiatric comorbidities require not only treatment for their PDs, but also recognition of increased risk for suicide related to the presence of multiple PDs.

Footnotes

Appendix

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.

Author Biographies

Kouichi Yoshimasu, MD, PhD, is a psychiatrist and an associate professor at Wakayama Medical University (Japan) as well as research collaborator at Mayo Clinic. His research interests are epidemiology and etiology of mental disorders, especially neurodevelopmental disorders.

William J. Barbaresi, MD, is director of the Developmental Medicine Center and Associate Chief of the Division of Developmental Medicine at Boston Children’s Hospital and professor of pediatrics at Harvard Medical School. He obtained a MD from the University at Buffalo School of Medicine and Biomedical Sciences in 1984. His research includes long-term, population studies of ADHD, the epidemiology of autism spectrum disorder, and learning disorders.

Robert C. Colligan, PhD, (deceased) was a former elementary and junior high school teacher, and consulting school psychologist. Later, he served as Head of Section—Psychology, Department of Psychology & Psychiatry at Mayo Clinic and was recipient of their Distinguished Career Award, and received a Lifetime Achievement award from the Minnesota Psychological Association. Most recently, he was an Emeritus Consultant to the Mayo Clinic Department of Psychology & Psychiatry. He was Diplomate of the American Board of Professional Psychology (ABPP), Fellow of the Societies for Personality Assessment, Clinical Child and Adolescent Psychology, Clinical Psychology and Child and Family Policy. He has served as co-investigator on grants from several organizations, including the National Institutes of Mental Health and the National Institute of Health.

Robert G. Voigt, MD, is a professor of pediatrics at Baylor College of Medicine. He is a developmental pediatrician who is board-certified in both developmental-behavioral pediatrics and neurodevelopmental disabilities. He is the director of the Meyer Center for Developmental Pediatrics and the Autism Center at Texas Children’s Hospital, and he is also program director for the fellowship program in developmental-behavioral pediatrics at Baylor College of Medicine.

Jill M. Killian, BS, is a statistical programmer analyst in the Department of Health Sciences Research at Mayo Clinic, Rochester, MN. She primarily collaborates with investigators in the Division of Epidemiology and is involved in data preparation, programming, and statistical analysis.

Amy L. Weaver, MS, is a master’s-level statistician in the Department of Health Sciences Research at Mayo Clinic, Rochester, MN. She primarily collaborates with investigators in the Departments of Pediatrics and Adolescent Medicine and Obstetrics and Gynecology and is involved in all aspects of research studies including study concept and design, statistical analysis, interpretation, and manuscript writing.

Slavica K. Katusic, MD, is associate professor of epidemiology and pediatrics. As a medical doctor and epidemiologist, she has more than 25 years of experience in assembling huge population-based birth cohorts, maintaining a long-standing research relationship with 41 public and private schools, and in getting the access to detailed birth certificate data. All of this, together with comprehensive medical records data available to every member of the birth cohort, allowed her and her team to study childhood development disorders (e.g., ADHD) and their adult outcomes.

References

Adler

L. A.

Spencer

T. J.

Wilens

T. E.

(2015). Attention-Deficit Hyperactivity Disorder in adults and children. Cambridge, UK: Cambridge University Press.

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Arlington, VA: American Psychiatric Publishing.

Balazs

Kereszteny

(2017). Attention-deficit/hyperactivity disorder and suicide: A systematic review. World Journal of Psychiatry, 7, 44-59.

Balazs

Miklósi

Keresztény

Dallos

Gádoros

(2014). Attention-deficit hyperactivity disorder and suicidality in a treatment naïve sample of children and adolescents. Journal of Affective Disorders, 152-154, 282-287.

Barbaresi

W. J.

Colligan

R. C.

Weaver

A. L.

Voigt

R. G.

Killian

J. M.

Katusic

S. K.

(2013). Mortality, ADHD, and psychosocial adversity in adults with childhood ADHD: A prospective study. Pediatrics, 131, 637-644.

Barbaresi

W. J.

Katusic

S. K.

Colligan

R. C.

Pankratz

V. S.

Weaver

A. L.

Weber

K. J.

. . . Jacobsen

S. J.

(2002). How common is attention-deficit/hyperactivity disorder? Incidence in a population-based birth cohort in Rochester, Minn. Archives of Pediatrics & Adolescent Medicine, 156, 217-224.

Biederman

Kwon

Aleardi

Chouinard

V. A.

Marino

Cole

. . . Faraone

S. V.

(2005). Absence of gender effects on attention deficit hyperactivity disorder: Findings in nonreferred subjects. The American Journal of Psychiatry, 162, 1083-1089.

Chan

M. K.

Bhatti

Meader

Stockton

Evans

O’Connor

R. C.

. . . Kendall

(2016). Predicting suicide following self-harm: Systematic review of risk factors and risk scales. The British Journal of Psychiatry, 209, 277-283.

10.

Cortese

Brown

T. E.

Corkum

Gruber

O’Brien

L. M.

Stein

. . . Owens

(2013). Assessment and management of sleep problems in youths with attention-deficit/hyperactivity disorder. Journal of the American Academy of Child & Adolescent Psychiatry, 52, 784-796.

11.

Furczyk

Thome

(2014). Adult ADHD and suicide. ADHD Attention Deficit and Hyperactivity Disorders, 6, 153-158.

12.

Hammerton

Zammit

Mahedy

Pearson

R. M.

Sellers

Thapar

. . . Collishaw

(2015). Pathways to suicide-related behavior in offspring of mothers with depression: The role of offspring psychopathology. Journal of the American Academy of Child & Adolescent Psychiatry, 54, 385-393.

13.

Hesson

Fowler

(2015). Prevalence and correlates of self-reported ADD/ADHD in a large national sample of Canadian adults. Journal of Attention Disorders. Advance online publication. doi:10.1177/1087054715573992

14.

Hurtig

Taanila

Moilanen

Nordström

Ebeling

(2012). Suicidal and self-harm behaviour associated with adolescent attention deficit hyperactivity disorder—A study in the Northern Finland Birth Cohort 1986. Nordic Journal of Psychiatry, 66, 320-328.

15.

Impey

Heun

(2012). Completed suicide, ideation and attempt in attention deficit hyperactivity disorder. Acta Psychiatrica Scandinavica, 125, 93-102.

16.

James

Lai

F. H.

Dahl

(2004). Attention deficit hyperactivity disorder and suicide: A review of possible associations. Acta Psychiatrica Scandinavica, 110, 408-415.

17.

Katusic

S. K.

Barbaresi

W. J.

Colligan

R. C.

Weaver

A. L.

Leibson

C. L.

Jacobsen

S. J.

(2005a). Case definition in epidemiologic studies of AD/HD. Annals of Epidemiology, 15, 430-437.

18.

Katusic

S. K.

Barbaresi

W. J.

Colligan

R. C.

Weaver

A. L.

Leibson

C. L.

Jacobsen

S. J.

(2005b). Psychostimulant treatment and risk for substance abuse among young adults with a history of attention-deficit/hyperactivity disorder: A population-based, birth cohort study. Journal of Child and Adolescent Psychopharmacology, 15, 764-776.

19.

Katusic

S. K.

Colligan

R. C.

Myers

S. M.

Voigt

R. G.

Yoshimasu

Stoeckel

R. E.

. . . Weaver

A. L.

(2017). What can large population-based birth cohort study ask about past, present and future of children with disorders of development, learning, and behavior? Journal of Epidemiology and Community Health, 71, 410-416.

20.

Kessler

R. C.

Akiskal

H. S.

Ames

Birnbaum

Greenberg

Hirschfeld

R. M.

. . . Wang

P. S.

(2006). Prevalence and effects of mood disorders on work performance in a nationally representative sample of U.S. workers. American Journal of Psychiatry, 163, 1561-1568.

21.

Kooij

J. J.

Huss

Asherson

Akehurst

Beusterien

French

. . . Hodgkins

(2012). Distinguishing comorbidity and successful management of adult ADHD. Journal of Attention Disorders, 16(5, Suppl.), 3S-19S.

22.

Kotov

Ruggero

C. J.

Krueger

R. F.

Watson

Yuan

Zimmerman

(2011). New dimensions in the quantitative classification of mental illness. Archives of General Psychiatry, 68, 1003-1011.

23.

Kurland

L. T.

Molgaard

C. A.

(1981). The patient record in epidemiology. Scientific American, 245(4), 54-63.

24.

Liu

K. Y.

Chen

E. Y.

Cheung

A. S.

Yip

P. S.

(2009). Psychiatric history modifies the gender ratio of suicide: An East and West comparison. Social Psychiatry and Psychiatric Epidemiology, 44, 130-134.

25.

Ljung

Chen

Lichtenstein

Larsson

(2014). Common etiological factors of attention-deficit/hyperactivity disorder and suicidal behavior: A population-based study in Sweden. JAMA Psychiatry, 71, 958-964.

26.

Melton

L. J.

3rd . (1996). History of the Rochester Epidemiology Project. Mayo Clinic Proceedings, 71, 266-274.

27.

Newcorn

J. H.

Weiss

Stein

M. A.

(2007). The complexity of ADHD: Diagnosis and treatment of the adult patient with comorbidities. CNS Spectrums, 12(8, Suppl. 12), 1-14, quiz 15-16.

28.

Patros

C. H.

Hudec

K. L.

Alderson

R. M.

Kasper

L. J.

Davidson

Wingate

L. R.

(2013). Symptoms of attention-deficit/hyperactivity disorder (ADHD) moderate suicidal behaviors in college students with depressed mood. Journal of Clinical Psychology, 69, 980-993.

29.

Perugi

Angst

Azorin

J. M.

Bowden

Vieta

Young

A. H.

. . . Group

B. S.

(2013). Is comorbid borderline personality disorder in patients with major depressive episode and bipolarity a developmental subtype? Findings from the international BRIDGE study. Journal of Affective Disorders, 144, 72-78.

30.

Perugi

Angst

Azorin

J. M.

Bowden

C. L.

Mosolov

Reis

, . . . BRIDGE-II-Mix Study Group. (2015). Mixed features in patients with a major depressive episode: The BRIDGE-II-MIX study. The Journal of Clinical Psychiatry, 76, e351-358.

31.

Rothman

K. J.

(1986). Modern epidemiology (1st ed.). Boston, MA: Little, Brown.

32.

Rucklidge

J. J.

Downs-Woolley

Taylor

Brown

J. A.

Harrow

S. E.

(2016). Psychiatric comorbidities in a New Zealand sample of adults with ADHD. Journal of Attention Disorders, 20, 1030-1038.

33.

Safer

D. J.

(1997). Adolescent/adult differences in suicidal behavior and outcome. Annals of Clinical Psychiatry, 9, 61-66.

34.

Sheehan

D. V.

Lecrubier

Sheehan

K. H.

Amorim

Janavs

Weiller

. . . Dunbar

G. C.

(1998). The Mini-International Neuropsychiatric Interview (M.I.N.I.): The development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. The Journal of Clinical Psychiatry, 59(Suppl. 20), 22-33, quiz 34-57.

35.

Sheehan

D. V.

Lecrubier

Sheehan

K. H.

Janavs

Weiller

Keskiner

. . . Dunbar

G. C.

(1997). The validity of the Mini International Neuropsychiatric Interview (MINI) according to the SCID-P and its reliability. European Psychiatry, 12, 232-241.

36.

Sheikh

Hendry

Lynch

Kalynych

C. J.

Aldridge

Kraemer

(2015). Poisoning with suicidal intent aged 0-21 years reported to poison centers 2003-12. The Western Journal of Emergency Medicine, 16, 497-502.

37.

Takemura

Yoshimasu

Fukumoto

Yamamoto

Miyashita

(2011). Subjective symptoms related to depression and suicidal risk in a Japanese community: A cross-sectional study. Open Journal of Psychiatry, 1, 40-48.

38.

Taylor

M. R.

Boden

J. M.

Rucklidge

J. J.

(2014). The relationship between ADHD symptomatology and self-harm, suicidal ideation, and suicidal behaviours in adults: A pilot study. Attention Deficit and Hyperactivity Disorders, 6, 303-312.

39.

van Vliet

I. M.

de Beurs

. (2007). [The MINI-International Neuropsychiatric Interview. A brief structured diagnostic psychiatric interview for DSM-IV en ICD-10 psychiatric disorders]. Tijdschrift voor Psychiatrie, 49(6):393-7.

40.

von Gontard

Equit

. (2015). Comorbidity of ADHD and incontinence in children. European Child & Adolescent Psychiatry, 24, 127-140.

41.

Wasserman

(2016). Suicide an unnecessary death (2nd ed.). Oxford, UK: Oxford University Press.

42.

Wilens

T. E.

Biederman

Spencer

T. J.

(2002). Attention deficit/hyperactivity disorder across the lifespan. Annual Review of Medicine, 53, 113-131.

43.

Williamson

Johnston

(2015). Gender differences in adults with attention-deficit/hyperactivity disorder: A narrative review. Clinical Psychology Review, 40, 15-27.

44.

Yoshimasu

Barbaresi

W. J.

Colligan

R. C.

Voigt

R. G.

Killian

J. M.

Weaver

A. L.

. . . Katusic

S. K.

(2012). Childhood ADHD is strongly associated with a broad range of psychiatric disorders during adolescence: A population-based birth cohort study. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 53, 1036-1043.

45.

Yoshimasu

Fukumoto

Takemura

Shiozaki

Yamamoto

Miyashita

(2011). Subjective symptoms related to suicide risk in Japanese male police officers. Suicidology Online, 2, 38-47.

46.

Yoshimasu

Kiyohara

Miyashita

, & The Stress Research Group of the Japanese Society for Hygiene. (2008). Suicidal risk factors and completed suicide: Meta-analyses based on psychological autopsy studies. Environmental Health and Preventative Medicine, 13, 243-256.