714776

research-article2017
LDXXXX10.1177/0022219417714776Journal of Learning DisabilitiesFuller-Thomson et al.

Article
Journal of Learning Disabilities

Suicide Attempts Among Individuals

1­–10
© Hammill Institute on Disabilities 2017
Reprints and permissions:
With Specific Learning Disorders: An sagepub.com/journalsPermissions.nav
DOI: 10.1177/0022219417714776
https://doi.org/10.1177/0022219417714776

Underrecognized Issue journaloflearningdisabilities.sagepub.com

Esme Fuller-Thomson, PhD1, Samara Z. Carroll, MA, MSW1,

and Wook Yang, MEd1

Abstract
Several studies have linked specific learning disorders (SLDs) with suicidal ideation, but less is known about the disorders’
association with suicide attempts. This gap in the literature is addressed via the 2012 nationally representative Canadian
Community Health Survey (n = 21,744). The prevalence of lifetime suicide attempts among those with an SLD was much
higher than those without (11.1% vs. 2.7%, p < .001). In comparison with their peers without SLDs, adults with SLDs
had 46% higher odds of having ever attempted suicide, even after adjusting for most known risk factors (e.g., childhood
adversities, history of mental illness and substance abuse, sociodemographics; odds ratio = 1.46, 95% CI [1.05, 2.04]).
The largest attenuation in the association between SLD and suicidal attempts was accounted for by adverse childhood
experiences. Among those with SLDs (n = 745), a history of witnessing chronic parental domestic violence and ever having
had a major depressive disorder were associated with substantially higher odds of suicide attempts.

Keywords
adverse childhood experiences, specific learning disorders, suicide attempts, Canadian Community Health Survey–Mental
Health

The overall prevalence of specific learning disorders neurodevelopmental disorder first manifests during the years
(SLDs) in children aged 0 to 17 years has been estimated at of formal schooling and is characterized by persistent and
9.7% in the United States (Altarac & Saroha, 2007). SLDs, impairing difficulties with learning foundational academic
such as dyspraxia, dysgraphia, and dyscalculia, are typi- skills in reading, writing, and/or math. (p. 32)
cally diagnosed among school-age children around the age
of 11 to 12 (Wilson, Deri Armstrong, Furrie, & Walcot, According to the manual, SLDs feature challenges in learn-
2009). These disabilities are lifelong conditions (Wilson ing principal academic skills, academic performance that is
et al., 2009) that may develop due to a number of factors, well below average for age, and learning difficulties in the
including genetics, problems during pregnancy and birth, as early school years, and these learning difficulties are not
well as prenatal exposure to maternal alcohol or drug use global but specific. Approximately 4% of adults appear to
(Margai & Henry, 2003; National Institute of Child Health have SLDs (American Psychiatric Association, 2013).
& Human Development, n.d.). SLDs are characterized by Research suggests that the negative impact of SLDs con-
impairments in perceiving, thinking, organizing, and/or tinues well into adulthood. Adults with these disabilities
processing of information (Kozey & Siegel, 2008; Walcot- often have lower emotional well-being and face challenges
Gayda, 2004) and an individual’s underachievement in pursuing higher education, maintaining a job, and living
comparison with one’s intelligence level (Fletcher, Morris, independently (Hoy, Gregg, Wisenbaker, & Manglitz, 1997;
& Lyon, 2004). SLDs fall on a spectrum from mild to severe Klassen, Tze, & Hannok, 2013). One particularly disturbing
(Walcot-Gayda, 2004). The Diagnostic and Statistical potential outcome of SLDs is suicidal behaviors (Hayes &
Manual of Mental Disorders–Fifth Edition (American
Psychiatric Association, 2013) provides the following defi- 1
University of Toronto, Canada
nition of SLDs: Corresponding Author:
Esme Fuller-Thomson, PhD, Factor-Inwentash Faculty of Social Work,
Specific learning disorder, as the name implies, is diagnosed University of Toronto, 246 Bloor Street West, Toronto, ON M5S 1V4,
when there are specific deficits in an individual’s ability to Canada.
perceive or process information efficiently and accurately. This Email: [email protected]
2 Journal of Learning Disabilities 00(0)

Sloat, 1988; Wilson et al., 2009). The emerging literature on education and socioeconomic status are more likely to
SLDs and suicidal behaviors has tended to focus on adoles- attempt suicide (Zhang, Mckeown, Hussey, Thompson, &
cents (Daniel et al., 2006; McBride & Siegel, 1997; Svetaz, Woods, 2005).
Ireland, & Blum, 2000). Research based on the National Adults with SLDs are more likely to be single and often
Longitudinal Survey of Adolescent Health revealed that a struggle with integrating into the workforce and other
higher prevalence of males and females with SLDs reported aspects of adult life (Hoy et al., 1997). Among individuals
attempting suicide than their peers without SLDs (Svetaz with SLDs, those who report low social support often expe-
et al., 2000). Several smaller studies support these findings. rience feelings of poor social competence, which is a risk
One study found that adolescents with poor reading abilities factor for suicide. People with SLDs were only half as likely
were more likely to report suicidal ideation and attempts to be married as those without SLDs (Wilson et al., 2009).
than their peers with normal reading ability (Daniel et al., Importantly, social support is a significant contributing fac-
2006). A study that analyzed suicide notes for errors in tor to successful outcomes among those with SLDs (Wilson
spelling and writing found that 89% of the 27 adolescents et al., 2009).
who completed suicide had problems in spelling and hand- A population-based study reported that one-third of
writing consistent with SLDs (McBride & Siegel, 1997). adults with dyslexia had experienced physical abuse in
Increased suicidal ideation among people with SLDs childhood or adolescence, in comparison with 7.2% of
appears to continue into adulthood. One large population- those without dyslexia, a difference that was statistically
based study of 15 to 44 year olds found that individuals with significant (Fuller-Thomson & Hooper, 2015). Felitti and
self-reported SLDs had 3 times higher odds of suicidal ide- colleagues (1998) reported that adults with four or more
ation in comparison with their counterparts without SLDs adverse childhood experiences (ACEs) had 12-times-higher
(Wilson et al., 2009). The study indicated that adults aged odds of attempting suicide. Those who, during their child-
30 to 44 years with self-reported SLDs had 1.6-times-higher hood, were sexually abused, physically abused (Brezo et
odds of suicidal ideation when compared with persons with al., 2008; Fergusson, Boden, & Horwood, 2008) and/or
SLDs who were aged 15 to 21 years. exposed to parental domestic violence (Fuller-Thomson,
It is important to take into account that the association Baird, Dhrodia, & Brennenstuhl, 2016) are more likely to
between SLDs and suicidal behaviors may be explained by have attempted suicide. Strong risk factors for suicide
other factors that are associated with both SLDs and suicidal attempts among adolescents with SLDs include having wit-
behaviors, such as poor mental health, negative health behav- nessed and/or been a victim of a violent act, each associated
iors (e.g., cannabis and tobacco smoking), low socioeconomic with a tripling of the odds of suicide attempts among ado-
status, low social support, and adverse childhood experiences lescents with SLDs (Svetaz et al., 2000).
(ACEs). These factors are reviewed in brief below. Males are approximately 2 to 3 times more likely than
Research indicates that individuals with SLDs are more females to be diagnosed with SLDs (Coutinho & Oswald,
likely to engage in negative health behaviors, such as smok- 2005). Yet, female adolescents with SLDs are more likely
ing tobacco and cannabis, when compared with their peers than adolescent males with SLDs to have attempted suicide
without learning difficulties, although those with SLDs do (9% vs. 5%), and both genders are more likely to have
not seem to have an increased risk of alcohol use (Finn, attempted suicide when compared with their peers without
Lopata, & Marable, 2010; Maag, Irvin, Reid, & Vasa, 1994). SLDs (Svetaz et al., 2000). Although women in the general
People who attempt suicide often exhibit a range of sub- population are at higher risk for attempting suicide than
stance problems (Borges, Walters, & Kessler, 2000), includ- men, men are more likely to complete suicide (Baca-Garcia
ing smoking (Bronisch, Höfler, & Lieb, 2008) and cannabis et al., 2010; Mościcki, 1994).
abuse (Beautrais, Joyce, & Mulder, 1999). The prevalence of SLDs diagnoses varies substantially by
SLDs are associated with poorer mental health outcomes race (Shifrer et al., 2011). African Americans, Native
(Wilson et al., 2009), including depression (Gallegos, Americans, and Hispanics are more likely than Asians and
Langley, & Villegas, 2012; Huntington & Bender, 1993) Caucasians to be placed in special education programs for spe-
and anxiety disorders (Gallegos et al., 2012; Hoy et al., cific learning disabilities (Office of Special Education
1997). In the general population, depression and anxiety are Programs, 2007). Native Americans (Baca-Garcia et al., 2010)
strongly associated with suicidal behaviors (Nepon, Belik, and First Nation Canadians (Kirmayer, 1994) have a particu-
Bolton, & Sareen, 2010; Nock, Hwang, Sampson, & larly high prevalence of suicide attempts. Suicide attempts are
Kessler, 2010). more prevalent among non-Hispanic White Americans than
People with SLDs come disproportionately from low- among African Americans (Nock et al., 2008).
income homes (Shifrer, Muller, & Callahan, 2011), and Although earlier cycles of the Canadian Community
people with SLDs are vulnerable to poorer employment Health Survey have been used to identify the prevalence of
prospects and poverty in adulthood (Hoy et al., 1997; suicidal thoughts among individuals with SLDs in Canada
Wilson et al., 2009). In turn, those with lower levels of (Wilson et al., 2009), there is no information available on
Fuller-Thomson et al. 3

the prevalence of suicide attempts in this population. The attempted suicide or tried to take your own life?” The
current study used the 2012 Canadian Community Health respondents were asked to provide a yes-or-no answer (i.e.,
Survey–Mental Health (CCHS-MH) to determine (1) the never vs. ever). The frequency of attempts was not assessed.
prevalence of suicide attempts among those with and without The section on childhood adversities was introduced
SLDs and (2) the degree to which the association between with the statement “The next few questions are about things
SLDs and suicide attempts is attenuated by attention-deficit/ that may have happened to you before you were 16 in your
hyperactivity disorder (ADHD), sex, race, age, household school, in your neighbourhood, or in your family.” Individuals
income and education, three childhood adversities (chronic who reported that they had seen or heard one of their “parents,
parental domestic violence, childhood sexual abuse, and step-parents or guardians hit each other or another adult” (≥18
childhood physical abuse), and lifetime history of depen- years) in the home >10 times were coded as having been
dence on and/or abuse of drugs or alcohol and (3) to identify exposed to chronic parental domestic violence. Childhood
which of these factors are associated with suicide attempts sexual abuse was coded as never versus ever. This was mea-
among those with SLDs. Based on the literature on suicide sured by the question “How many times did an adult force
ideation, the current study hypothesizes that people with you or attempt to force you into any unwanted sexual activ-
SLDs will show a higher prevalence of suicide attempts ity, by threatening you, holding you down or hurting you in
when compared with those without SLDs. In addition, the some way?” People who reported that an adult had at least
current study predicts that adjusting for ACEs, substance once kicked, bit, punch, choked, burned, or physically
abuse, and mental health will substantially attenuate the attacked them were identified as having been physically
association between SLDs and suicide attempts. We also abused in childhood.
anticipate that among those with SLDs, those with a history Sociodemographic variables include self-identified race
of mental illness, substance abuse, and childhood adversi- (non-Aboriginal White vs. visible minority and/or Aboriginal)
ties will be more likely to have attempted suicide at some and age (in decades). Socioeconomic status was measured by
point in their lives. education level (less than high school, high school graduate,
some postsecondary school vs. postsecondary diploma or uni-
versity degree). In addition, income was based on Statistics
Methods
Canada’s measure of household income as a ratio related to the
As has been described elsewhere (Fuller-Thomson et al., national low-income cutoff, divided into deciles. This variable
2016), the current study used data from the 2012 Canadian takes into account the number of people in the household and
Community Health Survey–Mental Health (CCHS-MH). the size of the community. Three health behaviors were also
This cross-sectional survey is representative of Canadians measured: smoking, obesity, and physical activity level.
who are living in the 10 provinces and are ≥15 years old Smoking status was measured with a derived variable created
(Statistics Canada, 2013). by Statistics Canada that we divided into three categories
The CCHS-MH relied upon a multistage cluster design (never smoked, ever smoked but never a daily smoker, and
to identify its sample. The overall response rate was 68.9% ever a daily smoker, which includes current and former daily
in this survey. Only respondents ≥18 years old were asked smokers). Obesity was assessed as a body mass index ≥30 and
questions about ACEs. Therefore, the sample for our study was based on self-reported height and weight. A missing cate-
was restricted to adult respondents (≥18 years old) with gory was also included because of the high nonresponse for the
complete data on all of investigated variables. The final weight variable. Physical activity level was measured by ask-
sample size for the analyses for Objectives 1 and 2 consists ing whether or not the respondent had participated in any mod-
of 10,032 men and 11,712 women. For Objective 3, the erate or vigorous physical activity in the past 7 days.
analysis was restricted to the 745 respondents who reported Substance abuse/dependence was based on the World
that they had ever been diagnosed with SLDs. Health Organization’s version of the Composite
International Diagnostic Interview (WHO-CIDI), which
was created by combining the derived variables created by
Measures
Statistics Canada: “drug abuse or dependence (including
The following preamble was read to participants in the cannabis)” and “alcohol abuse or dependence.” Both were
CCHS-MH: “We are interested in conditions diagnosed by derived with the lifetime algorithm. To be defined as a sub-
a health professional [which] are expected to last or have stance abuser, at least one of the following symptoms must
already lasted 6 months or more.” After which, they were be present while using drugs or alcohol:
asked about several conditions, including “Do you have a
learning disability?” If they responded affirmatively, they 1) Recurrent drug or alcohol use resulting in a failure to fulfill
were classified as having SLDs. major role obligations at work, school, or home (e.g., repeated
Lifetime suicide attempt was measured by an individu- absences or poor work performance related to drug or alcohol
al’s affirmative response to the question “[Have you ever] use; drug or alcohol-related absences, suspensions, or
4 Journal of Learning Disabilities 00(0)

expulsions from school; neglect of children or household); 2) (1994) summary of CIDI reliability, there is a very high per-
Recurrent drug or alcohol use in situations in which it is centage agreement between test and retest scores for the
physically hazardous (e.g., driving an automobile or operating subscales used in this study: depression (92%), anxiety dis-
a machine when impaired by substance use); 3) Recurrent drug order (89%), alcohol abuse/dependence (91%), and sub-
or alcohol-related legal problems (e.g., arrests for drug-related
stance abuse/dependence (93%). These scales also show to
disorderly conduct); 4) Continued drug or alcohol use despite
have high validity (Andrews & Peters, 1998; Wittchen,
having persistent or recurrent social or interpersonal problems
caused or exacerbated by the effects of drugs or alcohol (e.g., 1994). The concordance between clinical checklists and the
arguments with spouse about consequences of intoxication, CIDI subscales were very strong, with kappas of 0.84 for
physical fights). (Statistics Canada, 2013, p. 84) depressive disorders, 0.76 for anxiety/phobic disorders, and
0.83 for substance use (Wittchen, 1994). For more informa-
To be defined as substance dependent, the respondent tion on these variables, see the derived variable specifica-
must report a maladaptive pattern of drug or alcohol abuse tions of the CCHS-MH (Statistics Canada, 2013).
defined by at least three of the following symptoms: toler-
ance, withdrawal, increased consumption, attempts to quit, Statistical Analyses
time lost, reduced activities, continued drinking. The
Psychoactive Substance Use Disorders Scales had a con- A chi-square test was conducted of SLDs by lifetime sui-
cordance of 0.83 with clinical interviews (Janca, Robins, cide attempt (ever, never) for the complete sample and for
Cottler, & Early, 1992), and the test-rest reliability was men and women separately. Two sets of logistic regression
>90% (Wittchen, 1994). Mental health factors include life- analyses were conducted. For Question 1, which included
time anxiety and depression. Individuals were classified as those with and without SLDs (total n = 21,744), SLD was
having an anxiety disorder if they met the WHO-CIDI life- the exposure variable, and suicide attempt was the outcome
time criteria for generalized anxiety disorder at some point variable. The first model of this series included learning dis-
in their lives. Respondents who met the criteria for lifetime abilities, ADHD, sex, race, age, household income, and
generalized anxiety disorder reported education. The second set of logistic regression analyses
included Model 1 variables in addition to ACEs (i.e.,
1) excessive anxiety and worry and anxiety about at least two chronic domestic violence, physical abuse, and sexual
different events or activities that lasted at least six months; 2) abuse). The third model included Model 1 variables and a
finding it difficult to control the worry; 3) the anxiety and the lifetime history of dependence on and/or abuse of drugs or
worry were associated with three or more of the symptoms alcohol. The fourth model included a lifetime history of
associated with anxiety; 4) the focus of the anxiety and worry generalized anxiety disorder or major depressive disorders
was not confined to features of an Axis 1 disorder; and 5) the in addition to Model 1 variables. The fifth and final model
anxiety, worry, or physical symptoms caused clinically
included all the aforementioned variables.
significant distress or significant impairment in social,
The second series of logistic regression analyses also
occupational, or other important areas of functioning. (Statistics
Canada, 2013, p. 84) had lifetime suicide attempts as the outcome variable but
was restricted to the subsample of those who reported that
Respondents were categorized as having a depressive they had been diagnosed with SLDs (n = 745). The first
disorder if they met the WHO-CIDI lifetime criteria for model included only the three ACEs. The second and final
major depressive episode (Statistics Canada, 2013, p. 84). model included ACEs, sex, race, age, education and house-
hold income, lifetime history of substance abuse, depres-
sion or anxiety disorders, chronic pain, and ADHD. The
Respondents who met the criteria reported: 1. two weeks or
longer of depressed mood or loss of interest or pleasure and at Nagelkerke R2 value provided information on the percent-
least five symptoms associated with depression which represent age of the variability in suicide attempts associated with
a change in functioning; 2. that symptoms cause clinically each cluster of risk factors.
significant distress or impairment in social, occupational or All analyses are based on data weighted to account for
other important areas of functioning; and 3. that symptoms are the probability of selection and nonresponse; however, the
not better accounted for by bereavement or symptoms last sample sizes are described in their crude, unweighted form.
more than two months or the symptoms are characterised by a All analyses were conducted with SPSS Statistics (version
marked functional impairment, preoccupation with 23; IBM, Chicago, IL).
worthlessness, suicidal ideation, or psychomotor retardation.
(Statistics Canada, 2013, pp. 52–53)
Results
The WHO-CIDI measures have strong psychometric The first research question was to determine the prevalence
properties and excellent reliability and validity (Andrews & of suicide attempts among those with and without SLDs in
Peters, 1998; Wittchen, 1994). According to Wittchen’s the general population (n = 21,744). Our hypothesis—that
Fuller-Thomson et al. 5

Table 1. Logistic Regression of Lifetime Suicide Attempts Among Those With Specific Learning Disorders (SLDs) Versus Those
Without (n = 21,744).

Odds Ratio [95% CI]

SLDs Model 1 Model 2 Model 3 Model 4 Model 5

Yesa 2.20 [1.63, 2.97] 1.57 [1.14, 2.16] 1.94 [1.43, 2.64] 1.93 [1.41, 2.63] 1.46 [1.05, 2.04]

Note. Model 1 includes specific learning disorders, attention-deficit/hyperactivity disorder, sex, race, age, household income, and education. Model
2 includes all the variables in Model 1 and three childhood adversities (chronic parental domestic violence, childhood sexual abuse, and childhood
physical abuse). Model 3 includes all the variables in Model 1 and lifetime history of dependence on and/or abuse of drugs or alcohol. Model 4 includes
all the variables in Model 1 and history of generalized anxiety disorder or major depressive disorders. Model 5 includes all the aforementioned
variables (fully adjusted). CI = confidence interval; SLDs = specific learning disorders.
a
Not diagnosed with SLDs (reference).

individuals with SLDs would show a higher prevalence suicide attempt included having had witnessed domestic
of suicide attempts when compared with those without violence and experiencing childhood sexual abuse. The
SLDs—was supported. The prevalence of lifetime suicide Nagelkerke R2 statistic indicates that the three early adversi-
attempts among those with am SLD was much higher than ties alone explain 6.5% of the variability in suicide attempts.
those without the disorder among men (7.7% vs. 2.1%, p < Once all the other variables were added to the model, child-
.001), among women (16.6% vs. 3.3%, p < .001), and in the hood sexual abuse became nonsignificant, declining from
total sample (11.1% vs. 2.7%, p < .001). 2.71 (95% CI [1.49, 4.96]) in Model 1 to 1.97 (95% CI
The second research question was to determine the [0.97, 3.99]). Only two factors that remained significant in
degree to which the association between SLDs and suicide the final model were lifetime depression, with 7-fold the
attempts was attenuated by ADHD, sex, race, age, house- odds of suicide attempts in comparison with those who
hold income and education, three childhood adversities never had a major depressive disorder (OR = 7.58, 95% CI
(chronic parental domestic violence, childhood sexual [4.33, 13.28]), and having had witnessed chronic parental
abuse, and childhood physical abuse), and lifetime history domestic violence, with double the odds of suicide attempts
of dependence and/or addiction on drugs and/or alcohol. in comparison with those without exposure to this early
Our hypothesis—that adjusting for these factors would sub- adversity (OR = 2.39, 95% CI [1.11, 5.13]). The final model
stantially attenuate the association between specific leaning explained 27.7% of the variability in suicide attempts
disorders and suicide attempts—was supported. As shown among those with SLDs.
in Table 1, when adjusting for sociodemographic variables
and ADHD, we found that people with an SLD have more
Discussion
than twice the odds of a suicide attempt when compared
with those without such a disability (odds ratio [OR] = 2.20, This large representative survey of community-dwelling
95% CI [1.63, 2.97]). These odds of suicide attempts among Canadians showed that one in every six women with an
those with SLDs declined but remained statistically signifi- SLD had attempted suicide, as had one in every nine men.
cant when each of the following groups of variables was Even after controlling for many of the known risk factors
entered in the model separately: ACEs (Model 2: OR = for suicide attempts, those with SLDs had 46% higher odds
1.57, 95% CI [1.14, 2.16]), substance abuse/dependence of having ever attempted suicide in comparison with their
(OR = 1.94, 95% CI [1.43, 2.64]), and lifetime depression peers without SLDs. These findings on suicide attempts
and anxiety (OR = 1.93, 95% CI [1.41, 2.63]). Among these parallel earlier work documenting a link between SLDs and
factors, ACEs account for the largest attenuation of the rela- suicidal thoughts (Daniel et al., 2006; Wilson et al., 2009).
tionship between SLDs and suicide attempt. In the final
model, with all the aforementioned variables, the odds of
ACEs and Suicide Attempts
suicide attempts among those with SLDs was 46% higher
than those without SLDs (OR = 1.46, 95% CI [1.05, 2.04]). In this study, ACEs provided the largest attenuation in the
The third objective was to identify which of the investi- association between SLDs and suicidal attempts. Adding
gated factors are associated with suicide attempts among ACEs to the sociodemographically adjusted model resulted
those with SLDs (n = 745). The findings supported our in a decline in the odds of suicide attempts among those
hypothesis that people with SLDs who had a history of with SLDs, from 2.20 to 1.57. The addition of all the
mental illness and childhood adversities would be more remaining variables examined (e.g., depression, anxiety
likely to have attempted suicide at some point in their lives. disorders, substance abuse) resulted in only a modest fur-
As shown in Table 2 (Model 1), significant correlates of ther decline in the odds of suicide attempt among those with
6 Journal of Learning Disabilities 00(0)

Table 2. Logistic Regression of Suicide Attempts Among Those With Specific Learning Disorders (n = 745).

Odds Ratio [95% CI]

Model 1: ACEs only Model 2: Fully adjusted

ACEs
Domestic violence
  Witnessed ≥11 times 2.54 [1.29, 4.99] 2.39 [1.11, 5.13]
   Never witnessed or witnessed <10 times (ref) 1.00 1.00
Physical abuse
  Abused 0.93 [0.49, 1.74] 0.87 [0.43, 1.77]
  None (ref) 1.00 1.00
Sexual abuse
  Abused 2.71 [1.49, 4.96] 1.97 [0.97, 3.99]
  None (ref) 1.00 1.00
Demographics
Sex
  Female — 1.66 [0.94, 2.91]
  Male (ref) — 1.00
Race
  White only — 0.50 [0.27, 0.92]
   Visible minority or aboriginal (ref) — 1.00
Age by decade — 0.93 [0.75, 1.14]
Socioeconomic status
Education
   No high school graduation (ref) — 1.00
   High school graduate — 1.19 [0.59, 2.41]
  Postsecondary graduate — 0.75 [0.37, 1.52]
Household income by decile — 0.98 [0.88, 1.10]
Drug or alcohol abuse or dependence
Addicted or dependent on drugs and/or alcohol — 1.16 [0.67, 2.02]
No drug or alcohol dependence or abuse (ref) — 1.00
Mental health
Anxiety
   Lifetime anxiety disorder — 1.16 [0.63, 2.13]
   No anxiety disorder (ref) — 1.00
Depression
   Lifetime depressive disorder — 7.58 [4.33, 13.28]
   No depression (ref) — 1.00
Pain
Moderate or severe — 1.15 [0.63, 2.10]
Less than moderate (ref) — 1.00
ADHD
Yes — 1.17 [0.68, 2.02]
No (ref) — 1.00
Nagelkerke R2 0.065 0.277
–2 Log likelihood 495.7 408.9

Note. ACE = adverse childhood experience; ref = reference; ADHD = attention-deficit/hyperactivity disorder.

SLDs, to 1.46. A recent study in the general population et al., 2016). That study’s results indicate that depression,
found that ACEs, including childhood sexual abuse and wit- anxiety, substance abuse, and chronic pain partially medi-
nessing domestic violence, are associated with a much ated the association between childhood adversities and sui-
higher prevalence of suicide attempts (Fuller-Thomson cide attempts in the general population. Thus, the minimal
Fuller-Thomson et al. 7

additional attenuation between SLDs and suicide attempts adolescents with SLDs, including impulsivity, hostility, social
seen in the current study when depression, anxiety disor- withdrawal, lack of ability to relate causes and consequences,
ders, substance abuse were added to the ACEs (e.g., Table poor problem-solving skills, and interpersonal difficulties.
2: Model 5 vs. Model 2) suggests that these factors were These personality traits may predispose adolescents to higher
somewhat subsumed in the ACEs. Although our study does suicide risk, independent of their levels of depression.
not provide insight into the mechanism through which Many students with SLDs experience increased stress at
ACEs affect suicide attempts, Joiner (2005) has suggested school due to their difficulties with academic tasks.
that childhood exposure to physical pain may result in a Researchers have hypothesized that these difficulties in
predisposition to later self-injury. The current study, to our academic work have a negative impact on their self-esteem,
knowledge, is the first nationally representative study to which may increase the risk of depression and suicide
show that ACEs play a key role in attenuating the link attempts (Huntington & Bender, 1993). “Escape theory”
between SLDs and suicide attempts. proposes that as youth with reading disabilities face chal-
When we restricted our analyses to those with SLDs, lenges in school and perform below the standards, they
individuals who experienced chronic domestic violence begin to view themselves in negative ways, blaming them-
exposure or childhood sexual abuse were approximately selves and labeling themselves as lazy or stupid (Baumeister,
twice as likely to attempt suicide. These early adversities 1990). They also tend to compare themselves with youth
explained 6.5% of the variability in suicide attempts. without SLDs and eventually move into a “numb” state,
Perplexingly, our results also showed that childhood physi- looking for an escape. This escape is often in the form of
cal abuse was not associated with attempted suicide when their dropping out of school and, in its extreme form,
these other ACEs were included in the analysis. In our final attempting suicide (Daniel et al., 2006). It is possible that
model with full adjustment, chronic parental domestic vio- such a negative self-view may extend into adulthood, per-
lence was the only ACE that remained significant. This haps reinforced by reading problems or underachievement
relationship could flow in either direction. We speculate in the workplace. Potential important buffers to such a neg-
that parental violent conflict could be an indicator of poor ative self-view include family connectedness, school con-
childhood circumstances (low socioeconomic status, lack nectedness, and an overall sense of belonging (Svetaz et al.,
of reading in the home, disorganized household, lack of 2000).
social supports, etc.) which may increase the likelihood of Previous research by Wilson and colleagues (2009),
SLDs. The higher stress levels in these homes may under- using the 2003 Canadian Community Health Survey, found
mine children’s ability to focus or ask for help, thereby that adults with SLDs had higher odds of suicidal thoughts.
impairing learning. Alternatively, a child’s scholastic under- Their analysis controlled for income, marital status, educa-
performance may cause parental conflict, which may esca- tion, social support, and physical health. The current study
late into violence. differs from this because of our focus on suicidal attempts
rather than suicidal thoughts. Moreover, the current study
had additional control variables. Specifically, we found that
Additional Factors Associated With Suicide
adults with SLDs were found to have 46% higher odds of
Attempts Among Adults With SLDs having attempted suicide at some point in their lives, even
Mental illness plays a large role in suicide attempts. For women when a variety of potential confounders were taken into
and men in the general population, histories of depression and account. Wilson and colleagues revealed that the largest
anxiety were each associated with substantially increased odds prevalence of mental health problems was among the oldest
of suicide attempts, as found in other studies (e.g., Fuller- population with SLDs. When the analyses were restricted to
Thomson et al., 2016). Furthermore, among adults with SLDs, adults with SLDs, the current study found that a history of
those with a history of depression had much higher odds of anxiety disorders was not statistically significantly associ-
suicide attempts. People with SLDs who had experienced ated with lifetime suicide attempts among adults with SLDs.
depression reported 7.5-fold the odds of suicide attempts in This was in sharp contrast to a study of suicide attempts in
comparison with their peers without a history of depression. A the general population, which found that those with a his-
recent study indicated that children with SLDs are more prone tory of anxiety disorders had 4 times the odds of suicide
to depression (Gallegos et al., 2012), which may be partially attempts (Fuller-Thomson et al., 2016).
due to difficulties processing social information (Bauminger, Surprisingly, drug and alcohol abuse was also not associ-
Schorr Edelsztein, & Morash, 2005). However, our analyses ated with suicide attempts among those with learning disor-
indicated that adults with SLDs had approximately twice the ders. In contrast, studies of the general population suggest
odds of suicide attempts even when lifetime depression and that substance abuse doubles the odds of suicide attempts
sociodemographics were taken into account (see Table 1: (Fuller-Thomson et al., 2016). Future research is warranted
Model 4). Bender, Rosenkrans, and Crane (1999) posit that to understand why adults with SLDs have such a different
there are personality traits that are more common among profile of risk factors from that of their peers without SLDs.
8 Journal of Learning Disabilities 00(0)

However, our results clearly indicated that the strongest SLDs. At a minimum, assessment of children and adults
predictors of suicide attempts were lifetime depression and with SLDs for mental health problems and suicidality is
chronic parental domestic violence. Although there is a warranted (Svetaz et al., 2000). Additionally, those with
strong body of research on the relationship between suicide SLDs are more resilient if they come from close families
and SLDs in the adolescent population, there is a limited (Svetaz et al., 2000). Perhaps intervention at the family
amount of research on adults with SLDs. More efforts to level would promote better adjustment in adulthood.
address the mental health problems of adults with SLDs The results from our study indicate that it is important to
should be made to better assist this vulnerable population. recognize that people with SLDs are vulnerable to suicide
attempts. The current study revealed that childhood adversi-
ties are important factors that are associated with suicide
Limitations attempts for those with SLDs. This information can be uti-
This study has several limitations. In the current study, lized to develop targeted intervention programs to help
respondents with SLDs were identified by their self-report reduce the suicide attempts for those with SLDs who have
of a medical diagnosis of SLD. Although we recognize the experienced early adversity.
limitations that are associated with self-reporting, previous
research has shown that self-report is highly correlated with Acknowledgments
psychometric tests (Schulte-Körne, Deimel, & Remschmidt,
The analysis presented in this paper was conducted at the Toronto
1997), so the possibility of false positives is not large. Research Data Centre, which is part of the Canadian Research
However, there is a possibility of underreporting (e.g., false Data Centre Network (CRDCN). We are grateful to Statistics
negatives), particularly among older adults, because a diag- Canada for permission to access the 2012 Canadian Community
nosis of SLDs was relatively rare before 1970 (Wilson et Health Survey–Mental Health via the Research Data Center at the
al., 2009). Such underreporting would bias the findings University of Toronto and to the staff at the centre for their sup-
toward the null. Furthermore, due to the stigma associated port. The services and activities provided by the Toronto RDC are
with suicide attempts, there might be underreporting on this made possible by the financial or in-kind support of the SSHRC,
variable as well. In future research, a review of medical the CIHR, the CFI, Statistics Canada, and University of Toronto.
charts for a history of suicide attempts and a diagnosis of The opinions expressed do not represent the views of Statistics
SLDs would provide more accurate results. Canada.

Declaration of Conflicting Interests

Implications
The author(s) declared no potential conflicts of interest with
To address the links among depression, suicide attempts, respect to the research, authorship, and/or publication of this
and SLDs, there must be more emphasis placed on imple- article.
menting protective factors at home and at school and ways
for students with SLDs to build resiliency. Understanding Funding
effective interventions for children with SLDs is crucial
The author(s) disclosed receipt of the following financial support
(Wilson et al., 2009). However, interventions are also
for the research, authorship, and/or publication of this article:
needed for adults with SLDs, recognizing that effects of Esme Fuller-Thomson gratefully acknowledges support received
SLDs are lifelong and the odds of suicidal behaviors remain from the Sandra Rotman Endowed Chair in Social Work at the
high (Wilson et al., 2009). Effective strategies are needed to University of Toronto.
work with these populations (Wilson et al., 2009).
Since depression is a major predictor of suicide attempts
References
among those with SLDs, detecting signs of depression at a
young age could be helpful. Teachers should be trained to Altarac, M., & Saroha, E. (2007). Lifetime prevalence of learning
detect early signs of depression and refer to educational inter- disability among US children. Pediatrics, 119(2), S77.
ventions and counseling. For adolescents with SLDs, their American Psychiatric Association. (2013). Diagnostic and statis-
tical manual of mental disorders: DSM-5. Washington, DC:
school environment can be a place of risk and/or resiliency.
American Psychiatric Association.
Many students with SLDs experience high levels of stress at Andrews, G., & Peters, L. (1998). The psychometric properties
school. However, with targeted interventions, students can of the composite international diagnostic interview. Social
experience academic success. Such success can serve as a pro- Psychiatry and Psychiatric Epidemiology, 33(2), 80–88.
tective factor that may minimize the prevalence of depression Baca-Garcia, E., Perez-Rodriguez, M., Keyes, K. M., Oquendo, M.
among students with SLDs (Bender et al., 1999). A., Hasin, D. S., Grant, B. F., & Blanco, C. (2010). Suicidal
Factors to increase resilience should be fostered, which ideation and suicide attempts in the United States: 1991–1992
may include assessment and mentoring of the child with and 2001–2002. Molecular Psychiatry, 15(3), 250–259.
Fuller-Thomson et al. 9

Baumeister, R. F. (1990). Suicide as escape from self. Fuller-Thomson, E., & Hooper, S. (2015). The association between
Psychological Review, 97(1), 90–113. doi:10.1037//0033- childhood physical abuse and dyslexia: Findings from a popu-
295X.97.1.90 lation-based study. Journal of Interpersonal Violence, 30(9),
Bauminger, N., Schorr Edelsztein, H., & Morash, J. (2005). Social 1583–1592. doi:10.1177/0886260514540808
information processing and emotional understanding in chil- Gallegos, J., Langley, A., & Villegas, D. (2012). Anxiety,
dren with LD. Journal of Learning Disabilities, 38(1), 45–61. depression, and coping skills among Mexican school chil-
doi:10.1177/00222194050380010401 dren: A comparison of students with and without learning
Beautrais, A. L., Joyce, P. R., & Mulder, R. T. (1999). Personality disabilities. Learning Disability Quarterly, 35(1), 54–61.
traits and cognitive styles as risk factors for serious suicide doi:10.1177/0731948711428772
attempts among young people. Suicide & Life-Threatening Hayes, M., & Sloat, R. (1988). Learning disability and sui-
Behavior, 29(1), 37–47. cide. Intervention in School and Clinic, 23(5), 469–475.
Bender, W. N., Rosenkrans, C. B., & Crane, M. (1999). Stress, doi:10.1177/105345128802300503
depression, and suicide among students with learning disabili- Hoy, C., Gregg, N., Wisenbaker, J., & Manglitz, E. (1997).
ties: Assessing the risk. Learning Disability Quarterly, 22(2), Depression and anxiety in two groups of adults with learning
143–156. doi:10.2307/1511272 disabilities. Learning Disability Quarterly, 20(4), 280–291.
Borges, G., Walters, E. E., & Kessler, R. C. (2000). Associations Huntington, D., & Bender, W. (1993). Adolescents with learn-
of substance use, abuse, and dependence with subsequent ing disabilities at risk? Emotional well-being, depression,
suicidal behavior. American Journal of Epidemiology, 151(8), suicide. Journal of Learning Disabilities, 26(3), 159–166.
781–789. doi:10.1177/002221949302600303
Brezo, J., Paris, J., Vitaro, F., Hebert, M., Tremblay, R. E., & Janca, A., Robins, L. N., Cottler, L. B., & Early, T. S. (1992).
Turecki, G. (2008). Predicting suicide attempts in young Clinical observation of assessment using the Composite
adults with histories of childhood abuse. The British Journal of Internation Diagnostic Interview (CIDI): An analysis of the
Psychiatry: The Journal of Mental Science, 193(2), 134–139. CIDI Field Trials–Wave II at the St. Louis site. The British
doi:10.1192/bjp.bp.107.037994; 10.1192/bjp.bp.107.037994 Journal of Psychiatry, 160(6), 815–818.
Bronisch, T., Höfler, M., & Lieb, R. (2008). Smoking predicts Joiner, T. E. (2005). Why people die by suicide. Cambridge, MA:
suicidality: Findings from a prospective community study. Harvard University Press.
Journal of Affective Disorders, 108(1), 135–145. Kirmayer, L. (1994). Suicide among Canadian aboriginal peo-
Coutinho, M., & Oswald, D. (2005). State variation in gender dis- ples. Transcultural Psychiatry, 31(1), 3–58. doi:10.1177/
proportionality in special education. Remedial and Special 136346159403100101
Education, 26(1), 7–15. doi:10.1177/074193250502600102 Klassen, R., Tze, V., & Hannok, W. (2013). Internalizing problems
01 of adults with learning disabilities: A meta-analysis. Journal
Daniel, S., Walsh, A., Goldston, D., Arnold, E., Reboussin, B., of Learning Disabilities, 46(4), 317–327. doi:10.1177/002221
& Wood, F. (2006). Suicidality, school dropout, and read- 9411422260
ing problems among adolescents. Journal of Learning Kozey, M., & Siegel, L. S. (2008). Definitions of learning disabili-
Disabilities, 39(6), 507–514. doi:10.1177/00222194060390 ties in Canadian provinces and territories. Canadian Psychology,
060301 49(2), 162–171. doi:10.1037/0708-5591.49.2.162
Felitti, V. J., Anda, R. F., Nordenberg, D., Williamson, D., Spitz, Maag, J., Irvin, D., Reid, R., & Vasa, S. (1994). Prevalence and
A. M., Edwards, V., . . . Marks, J. S. (1998). Relationship predictors of substance use. Journal of Learning Disabilities,
of childhood abuse and household dysfunction to many of 27(4), 223–234. doi:10.1177/002221949402700404
the leading causes of death in adults. American Journal of Margai, F., & Henry, N. (2003). A community-based assessment
Preventive Medicine, 14(4), 245–258. doi:10.1016/S0749- of learning disabilities using environmental and contextual
3797(98)00017-8 risk factors. Social Science & Medicine, 56(5), 1073–1085.
Fergusson, D. M., Boden, J. M., & Horwood, L. J. (2008). McBride, H., & Siegel, L. (1997). Learning disabilities and ado-
Exposure to childhood sexual and physical abuse and adjust- lescent suicide. Journal of Learning Disabilities, 30(6), 652–
ment in early adulthood. Child Abuse & Neglect, 32(6), 607– 659. doi:10.1177/002221949703000609
619. Mościcki, E. K. (1994). Gender differences in completed and
Finn, K. V., Lopata, C., & Marable, M. (2010). Marijuana use in attempted suicides. Annals of Epidemiology, 4(2), 152–158.
suburban schools among students with learning disabilities. National Institute of Child Health & Human Development. (n.d.)
The Educational Forum, 74(4), 278–288. What causes learning disabilities? [Web page]. Retrieved
Fletcher, J. M., Morris, R. D., & Lyon, R. D. (2004). Classification from https://www.nichd.nih.gov/health/topics/learning/con-
and definition of learning disabilities: An integrative perspec- ditioninfo/pages/causes.aspx
tive. In H. L. Swanson, K. R. Harris, & S. Graham (Eds.), Nepon, J., Belik, S., Bolton, J., & Sareen, J. (2010). The relation-
Handbook of learning disabilities (pp. 30–56). New York, ship between anxiety disorders and suicide attempts: Findings
NY: Guilford. from the National Epidemiologic Survey on Alcohol and
Fuller-Thomson, E., Baird, S., Dhrodia, R., & Brennenstuhl, S. Related Conditions. Depression and Anxiety, 27(9), 791–798.
(2016). The association between adverse childhood expe- doi:10.1002/da.20674
riences (ACEs) suicide attempts in a population-based Nock, M. K., Borges, G., Bromet, E. J., Cha, C. B., Kessler, R. C.,
study. Child: Care, Health and Development, 42, 725–734. & Lee, S. (2008). Suicide and suicidal behavior. Epidemiologic
doi:10.1111/cch.12351 Reviews, 30(1), 133–154. doi:10.1093/epirev/mxn002
10 Journal of Learning Disabilities 00(0)

Nock, M. K., Hwang, I., Sampson, N. A., & Kessler, R. C. Svetaz, M. V., Ireland, M., & Blum, R. (2000). Adolescents
(2010). Mental disorders, comorbidity and suicidal behavior: with learning disabilities: Risk and protective factors
Results from the National Comorbidity Survey Replication. associated with emotional well-being. Findings from the
Molecular Psychiatry, 15(8), 868–876. National Longitudinal Study of Adolescent Health. Journal
Office of Special Education Programs. (2007). 27th annual report of Adolescent Health, 27(5), 340–348. doi:10.1016/S1054-
to Congress on the implementation of the Individuals with 139X(00)00170-1
Disabilities Education Act (Vol. 1). Washington, DC: U.S. Walcot-Gayda, E. (2004). Understanding learning disabilities.
Department of Education. Education Canada, 44(1), 36–39.
Schulte-Körne, G., Deimel, W., & Remschmidt, H. (1997). Can Wilson, A., Deri Armstrong, C., Furrie, A., & Walcot, E. (2009).
self-report data on deficits in reading and spelling predict The mental health of Canadians with self-reported learning
spelling disability as defined by psychometric tests? Reading disabilities. Journal of Learning Disabilities, 42(1), 24–40.
and Writing, 9(1), 55–63. doi:10.1177/0022219408326216
Shifrer, D., Muller, C., & Callahan, R. (2011). Disproportionality Wittchen, H. U. (1994). Reliability and validity studies of the
and learning disabilities: Parsing apart race, socioeconomic WHO–Composite International Diagnostic Interview (CIDI):
status, and language. Journal of Learning Disabilities, 44(3), A critical review. Journal of Psychiatric Research, 28(1),
246–257. doi:10.1177/0022219410374236 57–84.
Statistics Canada. (2003). Canadian Community Health Survey, Zhang, J., Mckeown, R. E., Hussey, J. R., Thompson, S. J., &
Cycle 1.2 mental health and well-being. Ottawa, Canada: Woods, J. R. (2005). Gender differences in risk factors for
Ministry of Industry. attempted suicide among young adults: Findings from the
Statistics Canada. (2013). Canadian Community Health Survey– third national health and nutrition examination survey. Annals
Mental Health (CCHS). Ottawa, Canada: Ministry of of Epidemiology, 15(2), 167–174. doi:10.1016/j.annepidem
Industry. .2004.07.095