672
Suicide and Life-Threatening Behavior 42(6) December 2012
2012 The American Association of Suicidology
DOI: 10.1111/j.1943-278X.2012.00121.x
Latina Adolescent Suicide Ideations and
Attempts: Associations with Connectedness to
Parents, Peers, and Teachers
SUSAN M. DE LUCA, PHD, PETER WYMAN, PHD, AND KEITH WARREN, PHD
Associations between suicidal behavior and social-ecological variables were
examined among 1,618 Latina high school students (mean age = 15) from the
nationally representative Add Health sample (68% were U.S.-born). Ideations
were associated with having a suicidal friend, lower perceived father support, and
overall parental caring. Attempts were associated with having a suicidal friend,
and lower perceived teacher and parental support. Peer and mother relationship
variables were not predictors of ideations or attempts. The protective role of
father and teacher support has not previously been emphasized in the literature.
Strengthening connections to parents and teachers may reduce suicidal behavior
in adolescent Latinas.
Suicide is the third leading cause of death
for 10 to 24 year olds in the United States,
and Latina adolescents had the highest rates
of suicidal ideation (21%) and attempts
(14%) compared with their peers in 2011
(e.g., 17% and 18% of African American
and Caucasian adolescent females, respectively, reported suicidal ideation, while 9%
of African American and 8% of Caucasian
adolescent females, respectively, reported
making an attempt in the past 12 months)
(Centers for Disease Control [CDC], 2012).
The public health importance of Latina
adolescent suicidal behavior is apparent
when considering Latinos are the largest
SUSAN M. DE LUCA, Assistant Professor,
School of Social Work, The University of Texas
at Austin, Austin, TX, USA; Peter Wyman, Professor, Dept. of Psychiatry, University of Rochester Medical Center, Rochester, NY, USA; and
Keith Warren, Associate Professor, College of
Social Work, The Ohio State University, Columbus, OH, USA.
Address correspondence to Susan M. De
Luca, The University of Texas at Austin, School
of Social Work, 1925 San Jacinto Blvd., Stop
D3500, Austin, TX 78712; E-mail: [email protected]
nonmajority ethnic group in the United
States, projected to comprise 24% of the
population by 2050 (U.S. Census, 2002).
Mental health problems, including
depression, anxiety, and substance abuse,
increase risk of suicidal behavior for Latina
adolescents and for other adolescents
(Duarte-Velez & Bernal, 2007; Foley, Goldston, Costello, & Angold, 2006; Gould et al.,
2004). In addition to individual-level factors,
evidence is increasing that risk of suicidal
behavior is influenced by social-ecological
factors.
The socioecological paradigm (i.e.,
Systems Theory), which also served as a conceptual framework for this study, posits that
individual development is an outcome of
interaction among the micro-, meso-, and
macrosystems (Hawley, 1950). Focusing on
relationships involving an individual within
these settings, this paradigm emphasizes
influences of diverse settings (i.e., family,
community, and school/work) on a variety of
physical and mental health outcomes (Mayberry, Espelage, & Koenig, 2009). Through
these multiple systems, social-ecological theory posits that qualities such as interdepen-
DE L UCA, WYMAN, AND WARREN
dence and negative feedback describe the
dynamic relationships people have in their
environment and influence risk of psychopathology (Stokols, 1992). Racial/ethnic status
and gender can also influence risk and protective processes across settings by affecting
levels of perceived stress and access to support (Goebert et al., 2010).
Congruent with a social-ecological
perspective, an adolescent’s relationships
with peers, family, and adults outside of the
family are emerging as factors influencing
risk of suicidal thoughts and behaviors. For
example, attempt rates are higher in adolescents who report lower peer emotional support (Borowsky, Resnick, Ireland, & Blum,
1999). Having fewer social networking
opportunities and weaker friendship ties have
also been linked to suicidal behaviors, particularly for female adolescents (Bearman &
Moody, 2004; Roberts & Chen, 1995). Differences between boys and girls in suicidal
behavior may be attributable to divergence
in socialization processes, including with
peers. For example, higher expectations for
peer relationships may heighten females’
vulnerability to relationship losses (AllgoodMerten, Lewinsohn, & Hops, 1990). In addition to peer network characteristics, peer
behaviors are also influential in adolescents’
risk of suicidal behavior. Suicidal behavior
of peers, in particular, is associated with
elevated risk of suicidal ideation and behavior. For both male and female adolescents,
studies in the literature report that if a member of an adolescents’ peer friendship group
attempted suicide, the remaining peers have
an increased likelihood to attempt as well
(Bearman & Moody, 2004; Joiner, 2002), and
the relative risk of dying by suicide after a suicide death in one’s social sphere has been
reported as two to four times higher among
15 to 19 year olds than for other age groups
(Gould, Wallenstein, & Kleinman, 1990).
‘‘Connectedness’’ (Whitlock, 2006)
describes linkages that encompass dyadic
forms of relationships between individuals,
belonging and reciprocal positive regard, and
explains risk and protective mechanisms for
suicidal behavior within social systems (Wy-
673
man et al., 2010). Family serves as a key
source of connectedness for children and
adolescents, and research points to the
importance of family as a source of protection and risk in regard to suicidal behavior. A
number of studies have linked adolescents’
perceived connectedness to family with
reduced risk of suicidal behavior (Borowsky
et al., 1999; Kerr, Preuss, & King, 2006)
including among Latinos (Guiao & Esparza,
1995). Relatedly, positive parenting in mothers of Latina adolescents is also linked to
lower depression as well as to reduced suicidal behaviors in their daughters (Turner,
Kaplan, Zayas, & Ross, 2002). Conversely,
Latinas who had conflict with parents are
more likely to consider suicide or make an
attempt (O’Donnell, O’Donnell, Wardlaw,
& Stueve, 2004; Zayas & Pilat, 2008).
Latina adolescents’ risk of suicide may
also be influenced by their families’ cultural
status, which encompasses factors such as
language use at home, place of birth, and
citizenship (Rogler, Malgady, & Rodriguez,
1990; Tortolero & Roberts, 2001), and is an
important antecedent to a range of psychosocial outcomes (Marin & Flores, 1994). For
example, Latinos born in the United States
are at an increased risk of depression and
drug use compared with native-born Latinos
(Escobar, Hoyos, & Gara, 2000). Similarly,
positive relationships have been found
between acculturation and suicidality (Zayas,
Bright, Alvarez-Sanchez, & Cabassa, 2009).
As a Latina acculturates, cultural protections
including close familial bonds, spirituality,
and collectivism tend to weaken, and it has
been posited that her risk of adverse mental
health outcomes including suicidal ideation
and attempts increase (Turner et al., 2002;
Zayas & Pilat, 2008). Thus, family cultural
status is intertwined with other protective
and risk factors within the family system.
Although prior studies point to the
importance of factors across social-ecological
systems in influencing risk of suicide among
Latina adolescents, a number of questions
remain unanswered. First, although the quality of Latina adolescents’ relationships with
mothers, and with parents as an aggregate,
674
LATINA ADOLESCENT SUICIDE
have been linked to risk of suicidal behavior
(Turner et al., 2002; Zayas et al., 2009), connectedness with mothers and with fathers
individually has not been examined.
Although mother–daughter relationships
have been studied extensively, father–daughter connectedness has received minimal
attention. However, it may be argued that
perceived father support could be a protective factor for Latinas because of evidence
that acculturation strain increases conflict
with parents, particularly mothers (Turner et
al., 2002). Second, the role of ties to peers
and nonparental adults, which may compensate for weakened ties with parents associated
with acculturation strain, has received little
focus with Latina adolescents. Latinas at risk
for attempts reported rejection by their peers
and teachers (Turner et al., 2002); however,
the role of peer and adult support in school
has not previously been examined along with
support from parents.
In the present study we examined the
contribution of the aforementioned social
influences on Latinas’ risk of ideation and
attempts. The study was guided by the following hypotheses: suicide proclivity (ideations and attempts) would be associated with
lower self-reported social integration and
perceived support, as evidenced by fewer
positive connections with parents as an
aggregate, with father and mother separately,
and with peers and teachers. In addition, suicidal peers were expected to increase a Latina’s risk of suicide ideations and attempts.
Owing to well-established associations
between depression and substance use levels
(Pena et al., 2008) and cultural status (Zayas
& Pilat, 2008) with risk of suicidal behaviors,
the influence of depression, substance use,
and several cultural status variables were
controlled for in testing relationships
between social-ecological variables and suicidal ideation and behaviors.
METHODS
A nationally representative sample of
adolescents was obtained from the Add
Health (National Longitudinal Study of
Adolescent Health) Wave I dataset (Harris et
al., 2009) compiled from school and in-home
questionnaires. This study used the first of
multiple waves of Add Health for several reasons. First, because prior studies with the
dataset had not examined social-ecological
risk and protective factors specifically for Latinas, a simpler, cross-sectional design was
deemed the most appropriate starting point
rather than a complex longitudinal design.
Second, suicide ideation and attempts are
frequently acute issues, and shorter time
scales may be more relevant. A sample of
1,618 Latinas were selected for the analysis.
The complex sample design, with oversampling of Puerto Rican and Cuban participants, required the use of weighted analysis
in STATA 10.0.
Measures
Suicidal Ideations and Attempts. The
ideation measure was dichotomous (‘‘past
12 months, did you ever seriously think
about committing suicide?’’). We use the
term ‘‘ideations’’ throughout this article
because individuals rarely have one single
thought concerning suicide. Attempts were
measured using a question as follows: ‘‘past
12 months, how many times did you actually
attempt suicide?’’ Both variables have been
used with Add Health data (Bearman &
Moody, 2004; Borowsky et al., 1999), and
higher values indicated presence of ideation
or attempts.
Family and Adult Support. The
mother and father indices included three
questions based on Haynie, South, and
Bose’s (2006) study measuring parental
closeness with Latino adolescents’ suicidal
risk. Mothers, fathers, and parents collectively were examined separately to ascertain
any differences.
Parental support was measured by the
following: (1) How close do you feel toward
your mother/father? (2) How much do you
think she/he cares about you? (3) Most of the
time your mother/father is warm and loving
toward you. The three aforementioned ques-
DE L UCA, WYMAN, AND WARREN
tions created two separate variables for
mother and father support. ‘‘Parents care’’
was measured by the item, ‘‘How much do
you feel that your parents care about you?’’
‘‘Teachers care’’ was measured by the item,
‘‘How much do you feel that your teachers
care about you?’’ All measures were based on
a Likert scale (1 = not at all; 2 = very little;
3 = somewhat; 4 = quite a bit; 5 = very much).
There were minimal missing data (less than
20%) on mothers, fathers, and parents.
Suicidal Friends. Latinas were asked
‘‘Have any of your friends tried to kill themselves in the past 12 months?’’ This question
has been used extensively in the adolescent suicide literature (e.g., Russell & Joyner, 2001).
The variable is the affirmative of its label.
Peer Support. Three variables evaluated friend support: ‘‘How much do feel that
your friends care about you?’’ coded 1 (not at
all) to 5 (very much); ‘‘I feel socially accepted’’
coded 1 (strongly disagree) to 5 (strongly agree),
and ‘‘During the past week, how many times
did you just hang out with your friends?’’
(0 = not at all, 1 = 1–2 times, 2 = 3–4 times,
and 3 = 5+ times). These measures were used
in Kidd et al.’s (2006) work on adolescent
attempters.
Reciprocated Friendships. Respondents
nominated their female friends and best
female friend. On both variables (friends,
best friend), Latinas who nominated friends,
or a best friend, and also received nominations by those same friends received a positive reciprocity score (1). A zero score
indicated no reciprocity (i.e., Latina nominated a friend but received no nominations
in return from that friend). Bearman and
Moody (2004) used these variables in their
study on adolescent suicidality.
In addition to the preceding predictors, the following measures were used as
control variables.
Cultural Status. The composite variable consisted of four questions: mother’s,
father’s, and child’s place of birth and language preference (‘‘English’’ predominately
spoken at home). All variables were dichotomous and the affirmative of their label.
A value of ‘‘4’’ would indicate higher accul-
675
turation level (i.e., Latina and parents were
U.S.-born and spoke English in the home).
Behavioral Health Status. Nineteen of
20 questions from the widely used CES-D
(Shrier, Harris, Sternberg, & Beardslee,
2001) measured severity of depression symptoms in the past week on a 4-point scale
(0 = never; 1 = sometimes; 2 = a lot of the time;
3 = most/all of the time). The study designers
of Add Health deleted a question about
excessive sleep from the survey because of
concerns about its appropriateness for adolescents. The item average score was used.
The measure had adequate internal reliability (Cronbach’s a = .74). A substance use
index was based on reported consumption
during past 12 months of alcohol and illegal
drugs (e.g., LSD, PCP, ecstasy). Respondents reported the number of times they
used alcohol and illicit substances. These
measures were used in adolescent suicide
attempt studies (Granillo, Jones-Rodriguez,
& Carvajal, 2005; Pena et al., 2008).
Data Analysis
Procedures. Multiple imputation was
used to address missing values for the two
reciprocity variables (46% missing data for
naming friends and 36% for naming best
friend; Rubin, 1987). No other variables had
missing data above 20% of the cases. Multiple imputation is a Monte Carlo technique
that replaces missing data through simulation and is common practice in the field of
epidemiology (Rubin, 1987).
Owing to the distribution of the suicide
attempt variable being highly kurtotic, negative
binomial regression technique was employed
(Hilbe, 2007). Negative binomial regression is
an extension of the poisson regression (Cameron & Trivedi, 1998; Hilbe, 2007). Many of the
assumptions of the poisson are essentially the
same as the negative binomial regression,
although one important deviation is the negative binomial regression assumes the mean and
variance are not equal (Hilbe, 2007). Negative
binomial regression calculates count models
when the poisson estimation is deemed unsuitable as a result of overdispersion.
LATINA ADOLESCENT SUICIDE
676
Logistic regression was used only for
models that examined the dichotomous
outcome variable of ideations and does not
assume a normal distribution.
Two separate models were created for
both ideations and attempts as the dependent
variables (Model 1 and Model 2). Model 1
included the measures of mother and father
support separately, while Model 2 included
only the measure of self-reported level of
care provided by parents. The development
of prediction models for ideation and
attempts utilized a forward selection procedure, and each variable was entered individually. First, all four cultural status variables
were regressed on the outcome to detect any
possible connections between culture and
ideation or attempts. Afterward the measures
representing depression, substance use, suicidal friends, parental supports, and a variety
of peer supports and contacts from their network were included. Variables that were
entered but did not reach statistical significance were not included in subsequent models. Upon inclusion and removal of variables,
changes in coefficients were also observed to
detect any possible confounders.
Logistic models were tested for goodness-of-fit using the Hosmer and Lemenshow (2000) statistic. Odds ratios were
reported to interpret the effect size of the
significant results in the logistic models.
Negative binomial models do not have a
goodness-of-fit test that is agreed on because
of their characteristic overdispersion of the
outcome variable (Kvalseth, 1985), but effect
sizes were reported using the incidence rate
ratio (IRR; Cummings, 2009).
RESULTS
Sample Description
Descriptive analyses of variables
included in both models are presented in
Table 1. The mean age was 15.0 years (range
of 12–19 years old) for this culturally diverse
group. Seventeen percent (275/1,618) of
the Latinas reported ideation in the past
year, and 6% reported at least one suicide
attempt.
The item-average mean for depression
symptom (CES-D) questions was 2.03, with
a median of 1.81. The distribution of depression scores indicated that 46% of the Latinas
reported an item average of 2.0, meaning
that these girls experienced symptoms of
depression either ‘‘some or a lot of the time’’
in the past week.
The substance abuse index resulted in
a range of responses (0–1,020 times of use).
The average number of times of use was 8.22
for any drugs in the past year, with a median
of 0.20, indicating that use was highly
skewed to the right with many Latinas
reporting no use.
The mean for the father support measure was 1.40, with the median of 2.00, and
the mother support measure had a mean of
2.00 and a median of 2.20. Over one third
(38%) of Latinas reported that their father’s
level of support was not supportive at all and
33% of the Latinas reported that their mothers were supportive ‘‘very little.’’
Latina Suicide Attempts. Results from
regression analyses predicting suicide
ideations and attempts are summarized in
Table 2.
Model 1—Mother and Father Support
Model 1 included the separate measures mother and father support. Higher levels of depressive symptoms, lower levels of
perceived teacher support, and having a suicidal friend were associated with attempts.
The IRR indicated for every point higher on
the depression scale a Latina reported, her
likelihood of attempts increased 11%, and
for every point higher on level teachers caring the likelihood of attempts decreased by
26%. Having a suicidal friend (regardless if
friend reported reciprocity) increased the
likelihood of attempts by 209%.
Model 2—Parents Care
Higher reported depression levels,
lower perceived parental caring and teacher
DE L UCA, WYMAN, AND WARREN
677
TABLE 1
Descriptive Statistics: Suicide Behavior, Risk, and Protective Factors
Variable
Outcome measures
Suicidal friend
Reciprocity
Cultural status
Adult level of support
Peer connectedness
Control variables
%
Suicide ideations
Suicide attempts
Never attempted
Attempted once
2–3 times
4–5 times
6 or more times
Had a suicidal friend
Friend reciprocity
Best friend reciprocity
English proficient in home
Latina adolescent born in United States
Mother born in United States
Father born in United States
Mother support
Father support
Parents care
Teachers care
Hung out with friends this week
Feel socially accepted by peers
Friends care
Depression index
Substance abuse index
caring, and having a suicidal friend predicted
attempts. The IRR indicated for every point
higher on the depression scale a Latina
reported, her likelihood of attempts
increased by 10%; for every level increase in
parents and teachers caring, the likelihood of
an attempt decreased 41% and 23%, respectively. Having a suicidal friend increased
likelihood of attempts by 235%.
Peer support variables, substance use,
social acceptance, having friends who cared,
hanging out with friends, and having a best
friend or friend reciprocate did not predict
attempts in either model.
Latina Suicide Ideations. Model 1 used
the previously mentioned forward selection
procedure that included mother and father
support, while the second model included
whether parents cared (Table 2).
17.0
94.0
3.5
1.5
0.5
0.5
22.0
68.0
39.0
46.0
62.0
4.0
15.0
M
SD
Median
2.00
1.35
4.77
3.46
1.78
3.96
4.26
2.03
8.22
2.93
4.89
0.63
1.09
0.49
0.77
0.73
0.61
4.16
2.20
1.80
5.00
3.00
2.00
4.00
4.00
1.81
0.20
Model 1—Mother and Father Support
Higher depression, less support from
fathers, and having a suicidal friend predicted ideation. There was a 12% increase
in risk of ideation for every level increase
in depression; a 5% decrease for every
level increase in perceived father support;
and a 120% increase if she had a suicidal
friend.
Model 2—Parents Care
Higher depression, lower perceived
parental caring, and having a suicidal friend
predicted ideation. Odds ratios indicated a
12% increase in ideation for each level
increase in depression, a 28% decrease for
each level increase in parents caring, and a
678
LATINA ADOLESCENT SUICIDE
TABLE 2
Summary of Regression Analysis Predicting Suicide Attempts and Suicide Ideations (N = 1,618)
Dependent variable = Suicide attempts
Model 1
Depression
Teachers care
Suicidal friends
Model 2
Depression
Parents care
Teachers care
Suicidal friends
Dependent variable = Suicide Ideations
Model 1
Depression
Father support
Suicidal friends
Model 2
Depression
Parents care
Suicidal friends
B
SE
p value
B 95% CI
0.11
)0.31
1.13
0.02
0.12
0.35
£.001
0.01
0.01
0.07–0.15
)0.54 to )0.08
1.04–1.83
0.10
)0.53
)0.26
1.21
0.02
0.17
0.11
0.31
£.001
0.01
0.01
£.001
0.07–0.14
)0.86 to )0.20
)0.48 to )0.04
1.09–1.83
1.12
0.95
2.20
0.01
0.02
0.52
£.001
0.01
0.01
1.09–1.16
0.91–0.98
1.38–3.50
1.12
0.72
2.27
0.02
0.09
0.52
£.001
0.01
£.001
1.08–1.16
0.56–0.93
1.45–3.56
Note. All models adjust for depression and cultural status.
Model 1 included measures of mother and father support separately.
Model 2 included a single measure of parental caring.
127% increase of ideations if she had a suicidal friend.
Both models fit well using the Hosmer
and Lemenshow (2000) statistic. Peer support variables fell below significance and
were ultimately removed from the final
model as was the ‘‘hung out with friends’’
and substance abuse variables.
DISCUSSION
In this study we utilized a social-ecological model to test predictors of adolescent
Latinas’ risk of suicidal attempts and ideation. Above and beyond the well-established
risk factor for elevated depression, support
from both parents and teachers was associated with fewer suicide attempts. In addition,
Latinas who perceived their fathers as more
supportive, as well as their parents as caring
for them, were less likely to have suicide
ideations. Having a suicidal friend was asso-
ciated with elevated risk of both ideation and
attempts. Latina’s self-reported relationships
with peers and mothers were not associated
with either ideation or attempts, nor were
substance use or measures of cultural status.
Overall, our findings support the contention
of social-ecological theory that relationships
between individuals and the systems that
comprise their environment are associated
with aspects of positive and maladaptive
development, including risk of suicidal
behavior, although not all relationship systems were equivalent in conferring risk of
suicidal behaviors. Our findings in particular
point to parents and teachers as potentially
important protective factors for Latina adolescents and to suicidal peers as important
risk factors.
Traditional Latino culture stresses
respect for elders (Trepper, Nelson, McCollum, & McAvoy, 1997), and in that context,
our finding that greater parental caring was
associated with reduced risk of attempts and
DE L UCA, WYMAN, AND WARREN
ideation was expected. In contrast, our finding that perceived caring from teachers was
associated with lower risk of attempts was
less expected and is more novel. Latinas in
this study came largely from immigrant families and turning to nonfamily members has
been viewed a counter to cultural norms
(Berdahl & Torres-Stone, 2009). However,
the benefits of perceived caring from teachers may be manifold. In addition to the
salutary psychological benefits of perceived
connectedness with teachers, teacher caring
may also reflect an adolescents’ positive
engagement to school, which is a key developmental competency and associated with
many benefits to well-being (Whitlock,
2006).
Prior research has shown that mother–
daughter mutuality reduces Latinas’ risk of
suicide attempts (Turner et al., 2002); our
finding that perceived caring by mothers
was not a significant protective factor was
less expected. In contrast, our finding
that greater connectedness to their father
decreased risk of ideation is novel. Prior
research has found that many Latino fathers
are absent or have minimal emotional interactions with their daughters and may have
minimal influence on their child’s mental
health (Yu et al., 2008). The potential
influence of fathers on Latina adolescents’
well-being and risk of suicide requires
further inquiry.
Among Latinas in the present Add
Health sample, 38% were first-generation
immigrants and less than 20% were from
second-generation (or higher) immigrant
families. In other words, the majority were
recent immigrants. Daughter–parent relationships for Latinas in our study may differ
compared with those in more acculturated
households. Acculturative stress is common
between traditional mothers and their
daughters who are exposed in school or via
media to mainstream U.S. views about
women’s roles and norms for behavior,
prompting adolescents’ frustration by their
mother’s adherence to traditional gender
roles. Consequently, Latinas might experience less acculturative strain with their
679
father, who is more likely to work outside of
the home and has possibly adopted more
mainstream views compared with her
mother. In this context, a father’s support
might impact well-being differently. Further
inquiry is needed to explore father–daughter
relationships including samples that have
more acculturated Latinas. In addition, prevention programs that encourage father
involvement with daughters in school activities might profitably be explored.
First-generation Latina adolescents
may also have different types of relationships
outside the home than more acculturated Latinas. For example, role conflict may occur
when Latinas become cultural brokers
between family members and adults (Buriel,
Perez, DeMent, Chavez, & Moran, 1998),
and in this context, her peer relationships
might include teachers and other adults,
more so than same-age peers in some
instances. Therefore, the importance of peer
support might not have the same protective
benefits for Latinas from immigrant homes
compared with other youth.
In terms of implications for prevention, our findings regarding the potential risk
factor for friends’ suicide attempts suggest
that heightened awareness of the maladaptive
behavior of a Latina’s friends may contribute
to early detection of suicidal risk, particularly
if her friend’s suicidal behavior supports a
perceived norm that suicide is an acceptable
response to distress. In this study, Latinas
with suicidal friends were twice as likely to
report ideation compared with Latinas who
did not have a suicidal friend. Given the
higher risk of suicide contagion among adolescents, parents and teachers should be
keenly aware of a Latina’s friendship network
(Insel & Gould, 2008).
Consistent with previous studies (e.g.,
Bearman & Moody, 2004; Russell & Joyner,
2001; Winterrowd, Canetto, & Chavez,
2010), having a suicidal friend was associated
with increased risk of suicidal ideation and
attempts in this sample of Latinas. This link
is well-established in community samples of
adolescents that combine different race/ethnic groups. To the best of our knowledge,
680
this study is one of the first to show this relationship is also reliable with Latina adolescents for both suicidal ideation and suicide
attempts, above and beyond the effects of
depression and parent support. Because Latinas in this study predominately came from
immigrant households, and because suicide
is a rare occurrence in their parents’ homelands (World Health Organization, 2010),
many immigrant parents might not be aware
of the associations between suicidal friends
and their own children. Educational programs to increase parent awareness of links
between peer group suicidal behavior and
risk of suicide in other teens may be one
profitable prevention focus, such as through
school or parent–teacher meetings on the
topic of suicide prevention.
Our finding that a Latina’s level of
perceived caring from her teachers was associated with reduced risk of suicide attempts
may also have implications for prevention.
Greater school engagement among female
adolescents has been associated with effective
suicide prevention efforts (Wyman et al.,
2010), but programs have rarely targeted
Latinas or addressed specific barriers to
school connectedness for this group. A number of school-based suicide prevention programs have been developed which focus on
strengthening coping skills and help-seeking
from adults to aid suicidal peers (Wyman
et al., 2010) and train students, teachers, and
staff to become more aware of the potential
warning signs of suicide (Wyman et al.,
2008). Programs targeting the total population of students in a school may help to
reduce the stigma of asking for help and
enable students to learn prosocial behaviors
as a collective. Latinos often use family
members for emotional support, therefore
identifying approaches that engage Latinas
in school interventions while at the same
time respecting cultural values emphasizing
family support is imperative.
We did not include in this study Latinas who had dropped out of school, a group
that may be at a high risk for suicide because
of the lack of protective factors from school
and relationships with peers and adults who
LATINA ADOLESCENT SUICIDE
introduce them to risk factors associated with
ideation and attempts (i.e., drug use; Epstein,
Botvin, Diaz, & Schinke, 1995). Promoting
community interventions via neighborhood
centers and spiritual leaders may be important considerations for Latinas not attending
school.
Limitations and Strengths
One limitation, common with selfreport questionnaires, is the risk of social
desirability bias. However, to protect against
this threat in Add Health, respondents used
earphones for sensitive questions and entered
their answers in provided computers. Children who were absent the day the questionnaire was distributed, or who had dropped
out of school, were not included in the study.
Latinos have the highest dropout rates in the
country (U.S. Census, 2003) and this could
have affected our results. Another limitation
was the measurement of many constructs by
single items. Although ‘‘face valid,’’ these
items may have limited reliability. Our analytic models were also limited in several
instances by missing data for peer and friendship network variables. In addition, Add
Health was designed to ascertain as representative a sample as possible. However, a large
number of Latinas reported having at least
one immigrant parent, and whether the sample was representative of all Latino families
in terms of immigrant status is difficult to
determine. As a result, our findings should
be observed with some caution. Additionally,
it should be noted that the data were collected between 1994 and 1995, and risk and
protective factors could have changed
because of the increased number of Latinos
in the United States and changes in other
cultural factors.
We focused on family, adult, and
peer relationship factors among Latina
adolescents to determine associations with
suicidal risk. We examined familial supports in two specific ways, parents collectively and mother and fathers separately.
Although the parents care and teachers
care measures were based on one question,
DE L UCA, WYMAN, AND WARREN
it lays a foundation for further inquiry
regarding these constructs.
POLICY AND RESEARCH
IMPLICATIONS
Latinas hav