Research

Original Investigation

Depression Outcomes Associated With an Intervention

Implemented in Employment Training Programs
for Low-Income Adolescents and Young Adults
S. Darius Tandon, PhD; Amanda D. Latimore, PhD; Eric Clay, BS; Lois Mitchell, MSW;
Margaret Tucker, MSPH; Freya L. Sonenstein, PhD

Supplemental content at
IMPORTANCE Recent estimates indicate that 6.5 million adolescents and young adults in the jamapsychiatry.com
United States are neither in school nor working. These youth have significant mental health
concerns that require intervention.

OBJECTIVE To determine whether a mental health intervention, integrated into an

employment training program that serves adolescents and young adults disconnected from
school and work, can reduce depressive symptoms and improve engaged coping strategies.

DESIGN, SETTING, AND PARTICIPANTS A quasi-experimental study was conducted; 512

adolescents and young adults newly enrolling in one employment training program site were
intervention participants, while 270 youth from a second program site were enrolled as
controls. Participants were aged 16 to 23 years and not in foster care. Study recruitment took
place from September 1, 2008, to May 31, 2011, with follow-up data collection occurring for
12 months after recruitment. Propensity score matching adjusted for observed baseline
differences between the intervention and control groups.

MAIN OUTCOMES AND MEASURES Depressive symptoms measured on a Center for

Epidemiologic Studies Depression Scale (CES-D) and engaged coping strategies.

RESULTS The mean age of participants was 19 years, 93.7% were African American, and
49.4% were male. Six- and 12-month follow-up rates were 61.0% (n = 477) and 56.8%
(n = 444), respectively. Males in the intervention group with high baseline depressive
symptoms exhibited a statistically significant decrease in depressive symptoms at 12 months
(5.64-point reduction in CES-D score; 95% CI, –10.30 to –0.96; P = .02) compared with
similar males in the control group. A dosage effect was observed at 12 months after the
intervention, whereby males with greater intervention exposure showed greater
improvement in depressive symptoms compared with similar males with lower intervention
doses (effect on mean change in CES-D score, −3.37; 95% CI, –6.72 to –0.09; P = .049). Males
and females in the intervention group were more likely than participants in the control group
to increase their engaged coping skills, with statistically significant differences found for
males (effect on mean change in CES-D score, 0.32; 95% CI, 0.14-0.50; P = .001) and females
(effect on mean change in CES-D score, 0.19; 95% CI, 0.01-0.37; P = .047) at 12 months.

Author Affiliations: Northwestern

CONCLUSIONS AND RELEVANCE Given the growing number of adolescents and young adults University Feinberg School of
using employment training programs and the mental health needs of this population, Medicine, Chicago, Illinois (Tandon);
increased efforts should be made to deliver mental health interventions in these settings that Johns Hopkins University Bloomberg
School of Public Health, Baltimore,
usually focus primarily on academic and job skills. Ways to extend the effect of intervention
Maryland (Latimore, Sonenstein);
for females and those with lower levels of depressive symptoms should be explored. Historic East Baltimore Community
Action Coalition, Baltimore, Maryland
(Clay, Mitchell); Insight Policy
Research Inc, Arlington, Virginia
(Tucker).
Corresponding Author: S. Darius
Tandon, PhD, Northwestern
University Feinberg School of
Medicine, 750 N Lake Shore Dr,
JAMA Psychiatry. 2015;72(1):31-39. doi:10.1001/jamapsychiatry.2014.2022 10th Floor, Chicago, IL 60611
Published online November 12, 2014. (dtandon@northwestern.edu).

31

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a Universite Laval User on 04/25/2016

 Research Original Investigation
O
ne-fourth of US adolescents and young adults (aged
16-24 years) will experience a depressive episode by
age 24—the highest incidence rate of any age group.1,2
Depression during this period has been shown to be associ-
ated with substance abuse, interpersonal problems, delin-
quency, academic and workplace difficulties, and suicide
attempts.3-6 Moreover, individuals experiencing a depressive
episode during adolescence or young adulthood are likely to
experience another episode in later adulthood.7,8 Provision of
mental health services to adolescents and young adults has
proved challenging, however. National studies indicate that only
25% of African American young adults (aged 18-24 years) re-
ceive needed services for depression, with this percentage fall-
ing to 19% among unemployed African American young adults.9
The number of African American adolescents and young
adults neither connected to school nor the workforce has risen
dramatically in recent years, with an estimated 5.8 million
youth10 now falling into this category of disconnected youth.11
There is, however, a growing public and private commit-
ment12,13 to developing employment training programs to
place these adolescents and young adults on a trajectory
toward educational advancement and employment. These
employment training programs can also serve as sites where
mental health services and supports can be provided. This
focus on addressing mental health concerns of adolescents
and young adults in employment training programs is par-
ticularly salient given research showing higher rates of health
problems among this population than among their school-
aged counterparts14 and policy briefs highlighting the preva-
lence of mental health concerns among this population and
the challenges faced by employment training programs in
addressing these concerns.15,16
This study is the first, to our knowledge, to rigorously as-
sess the effectiveness of a mental health intervention in em-
ployment training programs for adolescents and young adults.
Our primary aim was to examine the effectiveness of a multi-
component mental health intervention aimed at reducing de-
pressive symptoms and improving engaged coping strategies
integrated into an employment training program.17 Our inter-
vention was consistent with the Institute of Medicine’s defi-
nition of a universal preventive intervention as we targeted all
adolescents and young adults within the employment train-
i ng p ro g r a m . S i m i l a r to s o m e p re v i o u s u n ive r s a l
interventions,18,19 we also explored the effect of the interven-
tion on the population of adolescents and young adults ex-
hibiting higher levels of depressive symptoms at baseline. We
also examined whether intervention effects varied by sex given
the exploratory nature of this study. Our intervention strate-
gies were based on cognitive-behavioral and interpersonal ap-
proaches that have been demonstrated to be effective in pre-
venting and treating adolescent and young adult depression.20
Methods
Study Design and Treatment Conditions
We conducted a quasi-experimental study with 2 employ-
ment training programs in Baltimore, Maryland. Youth Op-
32 JAMA Psychiatry January 2015 Volume 72, Number 1
Copyright 2015 American Medical Association. All rights reserved.
Downloaded From: http://archpsyc.jamanetwork.com/ by a Universite Laval User on 04/25/2016
Depression Outcomes in Teenagers and Young Adults
portunities (YO) is a national employment training model, with
2 sites (Eastside YO and Westside YO) using identical program
models. These centers provide comprehensive social and edu-
cational services, including General Education Development
certification classes, support for college enrollment, resume
building, career development resources, and job placement.
Initially funded by the US Department of Labor, YO centers
serve communities with pervasive poverty, high unemploy-
ment rates, and general distress characterized by high drop-
out rates and several other negative social, health, and eco-
nomic indicators.
Eastside YO program enrollees received a multicompo-
nent intervention aimed at improving mental health status—
Healthy Minds at Work, which consisted of (1) mental health
training for YO program staff; (2) audio computer-assisted self-
interview (ACASI) mental health screening at the time of pro-
gram enrollment to determine the level of need for mental
health services; (3) psychoeducational workshops (eg, anger
management, coping with stress) both integrated into the em-
ployment training curriculum and conducted as freestanding
sessions; and (4) on-site mental health services provided by 2
full-time licensed clinical social workers and, as needed, a psy-
chiatrist to provide medication management. Results from the
ACASI screening were shared with the on-site clinical social
worker, and an initial appointment with the clinical social
worker was scheduled for each individual completing an ACASI.
During this initial visit, the clinical social worker reviewed
ACASI results, asked clarifying and probing questions to bet-
ter assess participants’ mental health needs, and developed a
preliminary case plan. All YO enrollees, regardless of baseline
mental health status, were eligible for psychoeducational work-
shops and on-site mental health services. Enrollees with mod-
erate (Center for Epidemiologic Studies Depression Scale21
[CES-D] score of 10-26) depressive symptoms were eligible for
a peer-led depression prevention group, while enrollees with
moderate and high (CES-D score >26) depressive symptoms
were recommended to engage in a minimum of 8 one-on-one
cognitive behavioral therapy (CBT) sessions with the on-site
clinical social worker. These CBT sessions lasted 45 minutes
and focused on enhancing understanding of how behavior,
thoughts, and emotions act in concert and providing effec-
tive ways of behaving and thinking in response to stressful situ-
ations. Intervention group engagement in Healthy Minds at
Work mental health services is reported in eTable 1 in the
Supplement. Westside YO enrollees received the ACASI screen-
ing and initial visit with an on-site mental health clinical so-
cial worker. However, the Westside clinical social worker
worked only 20 hours per week, limiting availability for fol-
low-up visits.
Sample
New YO enrollees aged 16 to 23 years and not in foster care
were eligible for research participation. A total of 782 youth
were enrolled across the comparison (n = 270) and interven-
tion (n = 512) samples between September 1, 2008, and May
31, 2011, representing 91.0% of total YO enrollees during the
recruitment period. The mean age of participants was 19
years, 93.7% were African American, 49.4% were male, and
jamapsychiatry.com
 Depression Outcomes in Teenagers and Young Adults
82.9% entered the program without a General Education
Development certification or high school diploma. A total of
37.4% had moderate to high depressive symptom levels
(CES-D score ≥10). Given the unstable housing and highly
mobile nature of our study population, intensive efforts
were undertaken to minimize participant loss to follow-up,
including the use of social networking sites and in-home vis-
its. Among the 782 participants observed at baseline, 477
(61.0%) completed follow-up assessments at 6 months and
444 (56.8%) at 12 months. A missingness analysis was also
conducted to assess the characteristics of those lost to
follow-up (see eAppendix in the Supplement). The final
sample for this intent-to-treat analysis includes 473 partici-
pants (307 intervention and 166 control participants) at 6
months and 441 participants (275 intervention and 166 con-
trol participants) at 12 months for whom no missing values
existed on the matching covariates.
Data Collection
Baseline ACASIs were conducted at the time of program en-
rollment as part of both YO programs’ standard enrollment pro-
cedure. A research assistant introduced the study to new en-
rollees, obtained informed written consent, and set them up
to complete the baseline ACASI. For participants younger than
18 years, parental written consent was also required before con-
ducting the ACASI. Six- and 12-month follow-up ACASIs were
done at a location and time most convenient for study partici-
pants; most follow-up assessments were conducted at the YO
program offices. Incarcerated youth were not eligible for fol-
low-up assessments. Individuals not completing or not eli-
gible for the 6-month ACASI were contacted for the 12-month
assessment. Participants were given $20 cash for completing
the 6- and 12-month follow-ups. The Johns Hopkins Univer-
sity School of Medicine institutional review board approved
all study procedures.
Measures
Outcome Variables
The CES-D,21 a 20-item self-report instrument widely used in
depression research with adolescents and young adults,18
was used to measure depressive symptoms. The Children’s
Coping Strategies Checklist–Revision 122 assessed domains of
engaged coping: active coping (eg, trying to figure out why
things like this happen), support seeking (eg, telling people
how you feel about a problem), and distraction (eg, listening
to music).
Matching Variables
Propensity scores were used to achieve balance between the in-
tervention and control groups. Participants were matched on
the following baseline characteristics: age (16-17, 18-19, 20-23),
sex, homelessness (yes or no), self-reported employment sta-
tus (employed full- or part-time or not employed), highest aca-
demic grade attained, clinically significant depressive symp-
toms as measured via the CES-D (score ≥16, which is the cutoff
for clinically significant depressive symptoms), level of finan-
cial and emotional support received from one’s mother,23 stig-
matized depression24 (yes or no), moderate to severe anxiety
jamapsychiatry.com
Copyright 2015 American Medical Association. All rights reserved.
Downloaded From: http://archpsyc.jamanetwork.com/ by a Universite Laval User on 04/25/2016
Original Investigation Research
as measured via the Beck Anxiety Inventory25 (score ≥16), and
number of life stressors experienced in the past 6 months as
measured via the Life Events Scale26 (above or below median),
as well as interactions between these variables and the follow-
ing potential effect modifiers: age, sex, employment status, and
baseline depressive symptoms.
Interaction Terms
Enrollment at the intervention or comparison site served as
the treatment variable. Given the exploratory nature of this
study, analyses were stratified by sex. Age group and a di-
chotomous variable for baseline depression were also tested
as interaction terms with the treatment variable. Established
cutoffs distinguished those with lower (CES-D <16) and higher
(CES-D ≥16) levels of depressive symptoms.21
Statistical Analysis
Propensity Score Matching
Intervention and comparison sites differed on some pre-
treatment characteristics even though the neighborhoods
served by the program were thought to be similarly dis-
tressed (Table 1). Covariates for the propensity score model
were required to have no missingness and be associated
with both treatment assignment and either of the 2 primary
outcomes. Stuart 27 explains that there is a low cost to
including variables not associated with the treatment but a
higher cost to excluding variables associated with the out-
come. Therefore, variables were also included that had no
association with the treatment but a theoretical association
with the outcome. Garber’s28 review enumerates 11 factors
associated with adolescent depression, 6 of which were
available in the current data with limited missingness and
included in our propensity score model: sex, anxiety, sub-
syndromal depression, negative cognition, stressors (ie, life
events), and interpersonal relationships (ie, social support).
Analyses were conducted to assess intervention effect sen-
sitivity to unobserved confounders. The magnitude of the
bias needed to alter our conclusions was such that it is
unlikely that the effect we observed was due to unmeasured
confounders.29
The propensity score was estimated using a multivari-
able logistic regression model in which the dependent vari-
able was a binary indicator of group assignment. Covariates
predicting the probability of intervention assignment, less the
interaction terms, are described in Table 2. Full matching meth-
ods were used in which all participants in the data set were re-
tained using MatchIt (R v2.15).30,31 Standardized mean bias and
propensity score distribution overlap assessed performance
of the matching technique. Propensity score matching was
considered successful at balancing intervention and control
groups when each covariate, including interaction terms,
achieved a standardized bias less than 0.25.32 Full matching
on propensity scores reduced the mean standardized bias by
99.1% and 98.7% at 6 and 12 months, respectively (Table 2).
Distributions of propensity scores for the intervention and con-
trol groups were also appropriately similar; therefore, match-
ing was deemed successful in improving the balance across im-
portant covariates.
JAMA Psychiatry January 2015 Volume 72, Number 1 33
 Research Original Investigation Depression Outcomes in Teenagers and Young Adults

Table 1. Intervention and Comparison Group Differences at Baseline

Comparison Group Intervention Group
Variable (n = 270) (n = 512) P Value
Matching variables
Age group, y, No. (%)
16-17 48 (17.8) 162 (31.6) <.001
18-19 135 (50.0) 193 (37.7) <.001
20-23 87 (32.2) 157 (30.7) <.001
Male sex, No. (%)a 114 (42.2) 272 (53.1) .004
Highest grade attained, mean (SD) 10.4 (1.58) 9.8 (1.37) <.001
Employed part-time or full-time, No. (%) 45 (16.7) 47 (9.2) .008
Ever been homeless, No. (%) 110 (40.7) 154 (30.1) .003
Moderate to severe depressive symptoms (CES-D score ≥16) (n = 780), No. (%)a 98/270 (36.3) 187/510 (36.7) .92
Received emotional and material support from mother (2 = “not helpful”, 6 = “a great 4.4 (1.4) 4.7 (1.3) .004
deal helpful”), mean (SD)
Ranked depression as more stigmatizing than using a wheelchair or having asthma or 17/268 (6.3) 56/512 (10.9) .03
AIDS (n = 780), No. (%)
Had above the median number (9) of life stressors, No. (%)b 115 (42.6) 236 (46.1) .35
Symptomatic elevated anxiety, No. (%) 35 (13.0) 57 (11.1) .45
Other demographic variables
Had at least 1 job before baseline (n = 225), No. (%) 27/99 (27.3) 8/126 (6.4) <.001
High school diploma or GED obtained before enrollment, among those ≥17 y 71/256 (27.7) 42/474 (8.8) <.001
(n = 730), No. (%)
Been physically hurt by a boyfriend or girlfriend 44/268 (16.4) 51/509 (10.0) .01
(n = 777), No. (%)
Ever drink in the morning (n = 540), No. (%) 12/185 (6.5) 45/355 (12.7) .03
Used contraception at last sexual intercourse (n = 718), No. (%) 153/249 (61.4) 326/469 (69.5) .03
Used >1 method of contraception at last sexual intercourse (n = 478), No. (%) 33/151 (21.8) 109/327 (33.3) .01
Used birth control pill at last sexual intercourse (n = 478), No. (%) 17/151 (11.3) 61/327 (18.6) .04
No. of sexual partners in the past 90 d (n = 624), mean (SD) [No. of participants] 2.2 (2.4) [n = 220] 3.1 (6.9) [n = 404] .08
Condom used with casual sexual partners in the past 90 d: always/every single time 60/102 (58.8) 156/217 (71.9) .02
(n = 319), No. (%)
Men who have sex with another man in lifetime (n = 383), No. (%) 7/113 (6.2) 4/270 (1.5) .01
Ever gave birth or fathered a child (n = 458), No. (%) 74/158 (46.8) 171/300 (57.0) .04
Problems with behaviors, feelings, drugs, or alcohol, No. (%)
Stayed overnight in an alcohol or drug treatment unit (n = 220) 11/68 (16.2) 11/152 (7.2) .04
Stayed overnight in a detention center, prison, or jail (n = 221) 39/68 (57.4) 121/153 (79.1) .001
Received outpatient services from an emergency department (n = 779) 20/269 (7.4) 64/510 (12.6) .03
Received outpatient services from probation officer or court counselor (n = 777) 30/269 (11.2) 86/508 (16.9) .03
No. of mental health services used in the past 12 mo (n = 771), mean (SD) [No. of 0.66 (1.5) [n = 267] 0.76 (1.5) [n = 504] .02
participants]
Life events, No. (%)
People moved in and out of house (n = 777) 43/269 (16.0) 56/508 (11.0) .048
Had difficulty with boyfriend or girlfriend (n = 779) 116/270 (43.0) 267/509 (52.5) .01
A friend needed help because of a relationship (n = 779) 104/269 (38.7) 237/510 (46.5) .04
Number of times emotional and material support received from a close group of 3.7 (1.1) 4.0 (1.1) .001
friends, mean (SD)
People living in house with serious drug or alcohol problems (n = 779), No. (%) 34/270 (12.6) 36/509 (7.1) .01
Baseline reading level (TABE score) (n = 586), mean (SD) [No. of participants] 7.8 (3.0) [n = 140] 6.3 (2.9) [n = 446] <.001
Baseline math level (TABE score) (n = 588), mean (SD) [No. of participants] 6.9 (2.5) [n = 142] 5.5 (2.3) [n = 446] <.001
Posttraumatic stress disorder (n = 768), No. (%) 25/266 (9.4) 75/502 (14.9) .03
Depressive symptom category (CES-D) (n = 780), No. (%)
Low 100/270 (37.0) 178/510 (35.0) .08
Moderate 146/270 (54.1) 258/510 (50.6) .08
High 24/270 (8.9) 74/510 (14.5) .08

(continued)

34 JAMA Psychiatry January 2015 Volume 72, Number 1 jamapsychiatry.com

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a Universite Laval User on 04/25/2016

 Depression Outcomes in Teenagers and Young Adults Original Investigation Research

Table 1. Intervention and Comparison Group Differences at Baseline (continued)

Comparison Group Intervention Group
Variable (n = 270) (n = 512) P Value
Negative attitudes toward depression treatment 40/270 (14.8) 105/509 (20.6) .047
(n = 779), No. (%)
If you are feeling depressed: definitely/probably not okay to wait and get over it 127/270 (47.0) 185/510 (36.3) .004
(n = 780), No. (%)
Outcome variables
No. of depressive symptoms (CES-D) (n = 780), mean (SD) [No. of participants] 14.2 (9.4) [n = 270] 15.4 (10.2) [n = 510) .20
Coping, mean (SD) [No. of participants]c
Active (n = 780) 2.6 (0.7) [(n = 270] 2.6 (0.7) [n = 510] .45
Distraction (n = 780) 2.2 (0.6) [n = 270] 2.2 (0.6) [n = 511] .87
Support-seeking (n = 780) 2.2 (0.8) [n = 270] 2.2 (0.8) [n = 510] .77
b
Abbreviations: CES-D, Center for Epidemiologic Studies Depression Scale; GED, Above the median number of life stressors experienced in the past 6 months
General Education Development certification; TABE, Tests of Adult Basic as measured via the Life Events Scale.26
Education. c
Use of coping skills, where 1 indicates never; 2, sometimes; 3, often; and 4,
a
Interaction terms. most of the time.

Table 2. Standardized Biases for Baseline Matching Covariates

6-mo Sample 12-mo Sample

Matching
Characteristic Before After Before After
Total participants, No. 473 473 441 441
Intervention:control, No. 307:166 307:166 275:166 275:166
Sex −.227 .172 −.240 −.063
Age −.210 .101 −.234 −.063
Homelessness −.198 −.044 −.214 −.078
Employment status −.381 .074 −.392 .045
Highest school grade attained −.454 .049 −.401 .056
Depressive symptoms (low/high) .084 .147 .007 .006
Anxiety (low/high) −.026 .0004 −.064 −.070
Stigmatized depression .225 −.004 .206 .012
Emotional/financial support from mother .225 −.100 .219 −.074
Above/below median number of life stressors (interaction terms not listed) .020 −.071 .059 .070
Mean standardized bias 1.030 .009 1.029 .014
Balance improvement 99.1 98.7

Analysis With Matched Data

Weights for matched data were exported to Stata, version 12.1
(StataCorp LP),33 and intervention effects on depressive symp-
toms and coping skills were estimated using sex-specific lin-
ear regression models. Covariates used in matching were
entered into the final regression models for doubly robust co-
variate adjustments. We used a regressor approach (Ytime 2 re-
gressed on Ytime 1 and X) to assess scores at our 2 time points
due to the causal effects of the baseline depressive symp-
toms on depressive symptoms at 6 and 12 months.34 To as-
sess intervention dose effects, the CES-D score change (base-
line to 6 and 12 months) was regressed on a dichotomous dose
variable (high vs low) using a sample of only intervention par-
ticipants matched on dose. Matching with doses was accom-
plished using nbpmatching in R (v2.15) and successfully mini-
mized differences between matched pairs on key baseline
variables while maximizing the difference between interven-
tion dose.35
Results
Intervention Exposure
Intervention dose ranged from 0 to 59 intervention services
and is represented by the total number of psychoeducational
workshops and clinical social worker sessions attended dur-
ing the 1-year intervention period among intervention partici-
pants only. Males were no different than females in their lev-
els of engagement in the intervention, with 87.5% of males and
88.8% of females using at least 1 intervention service, averag-
ing 4.5 (males) and 4.9 (females) services among those with
any engagement. Males (mean [SD], 6.18 [6.80]) and females
(mean [SD], 6.20 [7.67]) with moderate to severe baseline de-
pressive symptoms had significantly greater levels of inter-
vention engagement than males (mean [SD], 3.67 [4.87]) and
females (mean [SD], 3.93 [4.25]) with low depressive symp-
toms (males, P = .001; females, P = .007).

jamapsychiatry.com JAMA Psychiatry January 2015 Volume 72, Number 1 35

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a Universite Laval User on 04/25/2016

 Research Original Investigation Depression Outcomes in Teenagers and Young Adults

Table 3. Intervention Effects by Sex and Depressive Symptom Category at 6 and 12 Months

CES-D Score
Low High Across Baseline
Intervention Effect Intervention Effect Intervention Effect
Characteristic (95% CI) P Value (95% CI) P Value (95% CI) P Value
Males
6 mo (n = 256)
Depression −1.51 (−3.56 to 0.52) .14 0.99 (−0.25 to 4.52) .58 −0.90 (−2.68 to 0.88) .32
Coping overall −0.06 (−0.22 to 0.11) .52 0.04 (−0.24 to 0.33) .76 −0.03 (−0.18 to 0.12) .69
Coping active −0.04 (−0.24 to 0.15) .66 0.12 (−0.21 to 0.45) .47 −0.01 (−0.17 to 0.16) .99
Coping distraction 0.05 (−0.13 to 0.22) .59 −0.05 (−0.35 to 0.25) .74 0.02 (−0.12 to 0.17) .77
Coping support-seeking −0.04 (−0.26 to 0.18) .72 0.06 (−0.33 to 0.44) .77 −0.02 (−0.21 to 0.18) .88
12 mo (n = 209)
Depression 1.31 (−1.54 to 4.17) .37 −5.64 (−10.30 to −0.96) .02 −0.78 (−3.53 to 1.98) .58
Coping overall 0.10 (−0.11 to 0.31) .34 0.82 (0.51 to 1.14) <.001 0.32 (0.14 to 0.50) .001
Coping active 0.17 (−0.08 to 0.41) .18 0.97 (0.60 to 1.34) <.001 0.41 (0.20 to 0.62) <.001
Coping distraction 0.11 (−0.10 to 0.32) .29 0.71 (0.38 to 1.03) <.001 0.29 (0.11 to 0.47) .01
Coping support-seeking −0.03 (−0.27 to 0.21) .83 0.91 (0.54 to 1.28) <.001 0.25 (0.04 to 0.46) .02
Females
6 mo (n = 216)
Depression −2.69 (−6.34 to 0.96) .15 1.70 (−2.45 to 5.86) .42 −0.54 (−3.04 to 1.95) .67
Coping overall 0.17 (−0.05 to 0.39) .14 −0.03 (−0.28 to 0.22) .82 0.08 (−0.09 to 0.25) .34
Coping active 0.43 (0.28 to 0.58) <.001 −0.10 (−0.39 to 0.18) .48 0.10 (−0.10 to 0.29) .32
Coping distraction 0.03 (−0.17 to 0.24) .75 −0.07 (−0.30 to 0.17) .57 −0.01 (−0.17 to 0.14) .89
Coping support-seeking 0.18 (−0.10 to 0.46) .21 −0.04 (−0.36 to 0.27) .78 0.08 (−0.13 to 0.29) .45
12 mo (n = 231)
Depression 0.36 (−2.98 to 3.70) .83 10.08 (6.21 to 13.96) <.001 4.51 (1.92 to 7.10) .001
Coping overall 0.19 (0.01 to 0.37) .047 −0.15 (−0.36 to 0.06) .15 0.04 (−0.10 to 0.18) .59
Coping active 0.31 (0.08 to 0.54) .008 −0.02 (−0.51 to 0.01) .05 0.06 (−0.12 to 0.23) .50
Coping distraction −0.05 (−0.21 to 0.12) .58 −0.11 (−0.30 to 0.08) .25 −0.07 (−0.20 to 0.05) .24
Coping support-seeking 0.26 (0.02 to 0.49) .03 −0.41 (−0.69 to −0.14) .003 −0.03 (−0.22 to 0.15) .73
Males and Females
6 mo (n = 473)
Depression −1.96 (−3.85 to −0.06) .04 1.15 (−1.55 to 3.87) .40 −0.93 (−2.49 to 0.62) .24
Coping overall 0.02 (−0.12 to 0.15) .80 −0.01 (−0.19 to 0.18) .94 0.01 (−0.10 to 0.12) .87
Coping active 0.06 (−0.09 to 0.21) .43 −0.01 (−0.22 to 0.20) .92 0.04 (−0.09 to 0.16) .56
Coping distraction 0.02 (−0.11 to 0.16) .72 −0.68 (−0.25 to 0.12) .47 −0.01 (−0.11 to 0.10) .90
Coping support-seeking 0.04 (−0.13 to 0.22) .61 −0.01 (−0.25 to 0.23) .93 0.02 (−0.11 to 0.16) .72
12 mo (n = 441)
Depression 0.76 (−1.46 to 2.98) .50 5.12 (2.15 to 8.09) .001 2.33 (0.54 to 4.11) .01
Coping overall 0.15 (0.01 to 0.30) .045 0.23 (0.04 to 0.42) .02 0.18 (0.06 to 0.30) .01
Coping active 0.25 (0.07 to 0.42) .006 0.21 (−0.02 to 0.44) .07 0.23 (0.09 to 0.37) .001
Coping distraction 0.04 (−0.10 to 0.17) .62 0.22 (0.04 to 0.40) .02 0.10 (−0.01 to 0.21) .06
Coping support-seeking 0.12 (0.06 to 0.30) .19 0.11 (−0.13 to 0.34) .37 0.11 (−0.03 to 0.26) .12

Abbreviation: CES-D, Center for Epidemiologic Studies Depression Scale.

Depressive Symptoms at 6 and 12 Months
While no effect on depressive symptoms was observed at 6
months, statistically significant effects of the intervention
were observed at 12 months, which were modified by sex
and baseline depressive symptoms (Table 3). Males with
moderate to severe depressive symptoms at baseline (CES-D
score ≥16) in the intervention group showed a 5.64-point
reduction (95% CI, –10.30 to –0.96; P = .02) in depressive
symptoms compared with similar males in the comparison
group at 12 months. Among females with moderate to
severe baseline depressive symptoms, the decrease in the
control group was significantly greater than in the interven-
tion group (–10.08, 95% CI, -6.21 to –13.96; P < .001). There
were no statistically significant differences between the
intervention and control groups at 12 months among those
with low baseline depressive symptoms, regardless of sex.
Effects of geographic clustering were examined at 6 and 12
months, and no significant geographical clustering was

36 JAMA Psychiatry January 2015 Volume 72, Number 1 jamapsychiatry.com

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a Universite Laval User on 04/25/2016

 Depression Outcomes in Teenagers and Young Adults
found by zip code (within correlation, –0.001 and –0.019,
respectively) or community statistical area (within correla-
tion, –0.025 and 0.0002, respectively).36
Intervention Dose Effects on Depressive Symptoms
at 12 Months
Given the statistically significant intervention effects ob-
served at 12 months, additional propensity score analyses de-
termined the dose-response relationship between level of en-
gagement in the intervention and depressive symptoms at 12
months as represented by changes in CES-D score from base-
line to 12 months. Of the 444 participants observed at 12
months, 278 were in the intervention group, 264 of whom were
included in the matching-by-dose analysis, resulting in 132
matched pairs. Intervention participants with low-dose inter-
vention service use (mean number of intervention services, 2.8;
range, 0-23) were matched with intervention participants with
high-dose intervention service use (mean number of inter-
vention services, 8.0; range, 1-59), with a mean dose differ-
ence of 5.8 (range, 1-33) services. Additional details of the analy-
sis can be found in eTables 2 through 5 in the Supplement.
Males with higher doses of the intervention had a greater re-
duction in depressive symptoms between baseline and 12
months (effect on mean change in CES-D score, –3.37; 95% CI,
–6.72 to –0.09; P = .049) compared with similar males with
lower doses of the intervention (Figure).
Coping Skills at 6 and 12 Months
At 12 months, females in the intervention group with lower de-
pressive symptoms at baseline demonstrated a greater im-
provement in the average frequency of using active coping (95%
CI, 0.08-0.54; P = .008), support-seeking (95% CI, 0.02-0.49;
P = .03), and overall coping (95% CI, 0.01-0.37; P = .047) strat-
egies than did females in the control group (Table 3). Overall
coping increases were also significantly greater at 12 months
for females in the intervention group with lower depressive
symptoms at baseline compared with females in the control
group. Intervention effects were found at 12 months for males
with moderate to severe depressive symptoms for all 3 en-
gaged coping strategies—active coping (95% CI, 0.20-0.62;
P < .001), distraction coping (95% CI, 0.11-0.47; P = .01), and
support seeking (95% CI, 0.04-0.46; P = .02)—as well as over-
all use of coping strategies (95% CI, 0.14-0.50; P = .001)
(Table 3).
Discussion
This is the first known study, to our knowledge, to examine
the effects of a mental health intervention for impoverished
African American adolescents and young adults in employ-
ment training programs—a highly vulnerable population who
cannot be accessed via school- or workforce-based interven-
tions. Male adolescents and young adults receiving our inter-
vention who entered the employment training program with
greater depressive symptoms showed greater reduction in de-
pressive symptoms and greater use of engaged coping strate-
gies compared with male participants not receiving mental
jamapsychiatry.com
Copyright 2015 American Medical Association. All rights reserved.
Downloaded From: http://archpsyc.jamanetwork.com/ by a Universite Laval User on 04/25/2016
Original Investigation Research
Figure. Dose Effect on Depressive Symptoms (Center for Epidemiologic
Studies Depression Scale [CES-D] Score) at 12 Months
Men Women
Average Effect: −3.37, Average Effect: 0.88,
P =.049 P =.63
2
1.65
1.51
1
0.63
CES-D Score
0
–1
–1.72
–2
Low High Low High
A dose effect was observed for the effect of the intervention on depressive
symptoms at 12 months. Among males, those with higher engagement in
intervention services had significantly more reductions in depressive symptoms
than those with lower service engagement. No significant dose effect was
observed among females.
health services. These data indicating that the effect of our in-
tervention was sex-specific and more effective for partici-
pants with higher baseline depressive symptoms are consis-
tent with data from previous studies demonstrating greater
effectiveness for targeted interventions than for universal pro-
grams, as well as sex differences.37,38 While we did not con-
duct mediation analyses, we believe that changes in depres-
sive symptoms were at least partially influenced by increased
use of engaged coping strategies among study participants.
Contrary to some previous psychotherapy research that shows
more immediate intervention benefits on the use of engaged
coping skills and depressive symptoms, this study showed ben-
efits largely at our 12-month follow-up. We believe this out-
come may be due to the amount of time it took for study par-
ticipants to master skills they received during the Healthy
Minds at Work intervention and the cumulative effect of using
these skills over time.
Although female intervention participants entering with
higher depressive symptoms also exhibited a decline in de-
pressive symptoms, the magnitude of change among females
in the intervention group entering with higher symptoms was
significantly smaller than in similar females in our control
group. These sex-specific findings are not due to sex differ-
ences in intervention engagement or type of services re-
ceived. One possible explanation for these sex differences is
that our intervention’s largely CBT focus may have been bet-
ter suited for males, as there has been some debate whether
male and female adolescents and young adults may respond
more favorably to cognitive-behavioral or interpersonal ap-
proaches, respectively.18
A unique aspect of this study was its intentional effort to
deliver a mental health intervention via different modalities—
via on-site clinical social workers, psychoeducational ses-
sions, and staff mental health training—within a setting, an ap-
proach found to improve program efficacy.39 Although our
JAMA Psychiatry January 2015 Volume 72, Number 1 37
 Research Original Investigation Depression Outcomes in Teenagers and Young Adults

dosage analyses did not specifically identify what constella-
tion of mental health services and supports were associated
with varied outcomes, it appears that greater intervention ex-
posure resulted in greater reductions in depressive symp-
toms for male participants. This finding may be particularly
important for future depression prevention trials conducted
outside of school settings, where it may be more difficult to
deliver intervention content to a large number of individuals
at the same time.
Some limitations to our study exist. Our propensity
score analysis adjusted only for observed covariates. Differ-
ences between intervention and control groups may still
have existed due to unobserved confounders even after bal-
ance on observed covariates was achieved, although our
sensitivity analysis suggests that these unobserved vari-
ables were not likely to have an effect on the study’s inter-
nal validity. Differences in the sample were observed across
time, making longitudinal analyses across all 3 time points
unsuitable. Challenges in sustaining intervention partici-
pants’ contact with on-site clinical social workers may have
prevented us from achieving a greater intervention effect.
Although embedding mental health services in an employ-
ment training program was presumed to minimize many
barriers (eg, transportation, stigma) to accessing such ser-
vices in other community or clinical settings, many inter-
vention participants still had several barriers that limited
their engagement in intervention activities. Most notably,
participants had difficulty regularly attending scheduled
clinical social worker visits or psychoeducational work-
shops given the varied and sometimes unanticipated
demands they faced in caretaking for younger siblings
and/or older family members. We anticipated that many of
these barriers might interfere with regular receipt of CBT
sessions delivered by on-site clinical social workers. Thus,
our program model recommended 8 CBT sessions for indi-
viduals with high baseline depressive symptoms, which is a
shorter duration than many other CBT-based interventions
to treat adolescent depression.40 While it would have likely
been challenging to engage YO members in CBT for a longer
duration given the logistical issues noted above, we may
have seen a larger effect on depression and coping out-
comes had we recommended a greater number of CBT ses-
sions. Finally, caution should be used in generalizing find-
ings to other populations of adolescents and young adults
who are not in school or the workforce, both within and
outside employment training programs.
Conclusions
Given the enormous need for mental health services and in-
terventions among youth in employment training programs
and the growing number of adolescents and young adults en-
gaged in such programs, further efforts to meet the needs of
this population are highly warranted. Subsequent work should
carefully consider whether a universal approach is cost effec-
tive or if resources should be focused on youth with greater
depressive symptoms upon program enrollment. While this
study reports solely on mental health outcomes, our research
team is also examining the effect of the intervention on edu-
cation, employment, and incarceration outcomes, which may
aid in making decisions about universal vs indicated ap-
proaches. Future work should also determine whether cur-
rent intervention approaches are insufficient or inadequately
designed to improve mental health outcomes for female ado-
lescents and young adults.

ARTICLE INFORMATION Foundation, The Annie E. Casey Foundation, Aaron REFERENCES

Submitted for Publication: January 20, 2014; final
revision received May 9, 2014; accepted June 17, 2014.
Published Online: November 12, 2014.
doi:10.1001/jamapsychiatry.2014.2022.
Author Contributions: Dr Latimore had full access
to all the data in the study and takes responsibility
for the integrity of the data and the accuracy of the
data analysis.
Study concept and design: Tandon, Clay, Mitchell,
Sonenstein.
Acquisition, analysis, or interpretation of data:
Tandon, Latimore, Tucker.
Drafting of the manuscript: Tandon, Latimore,
Mitchell, Sonenstein.
Critical revision of the manuscript for important
intellectual content: Clay, Tucker, Sonenstein.
Statistical analysis: Tandon, Latimore.
Obtained funding: Tandon, Mitchell, Sonenstein.
Administrative, technical, or material support:
Latimore, Clay, Mitchell, Tucker.
Studysupervision:Tandon,Mitchell,Tucker,Sonenstein.
Conflict of Interest Disclosures: None reported.
Funding/Support: Funding for the implementation
of the Healthy Minds at Work intervention came
from the Robert Wood Johnson Foundation, the
Jacob and Hilda Blaustein Foundation, The Abell
Foundation, the Leonard and Helen R. Stulman
Straus and Lillie Straus Foundation, and the
France-Merrick Foundation. The research portion of
Healthy Minds at Work was developed as the core
research project of the Johns Hopkins Center for
Adolescent Health, a prevention research center
funded by the Centers for Disease Control and
Prevention (grant no. 1-U48-DP-000040).
Role of the Funder/Sponsor: The funding sources
had no role in the design and conduct of the study;
collection, management, analysis, and
interpretation of the data; preparation, review, or
approval of the manuscript; and decision to submit
the manuscript for publication.
Additional Contributions: We thank the Eastside
and Westside YO programs for their ongoing
collaboration on this study, including the dedicated
mental health clinicians working at the Eastside YO
site. Elizabeth Stuart, PhD, provided valuable
guidance on the propensity score analyses used in
the manuscript. She was not compensated for her
time. Trevor Arnett, MA, Meron Solomon, BA,
Danielle Feldman, MSPH, Amanda Manning, BA,
Irisha Burrell, BA, and Courtney Hegadorn, BA,
assisted with collection of baseline and follow-up
data; they were staff members who were
compensated for their work on the project.
Correction: This article was corrected online
January 7, 2015, to fix a reference.
1. Kessler RC, Berglund P, Demler O, Jin R,
Merikangas KR, Walters EE. Lifetime prevalence
and age-of-onset distributions of DSM-IV disorders
in the National Comorbidity Survey Replication
[published correction appears in Arch Gen
Psychiatry. 2005;62(7):768]. Arch Gen Psychiatry.
2005;62(6):593-602.
2. Kessler RC, Walters EE. Epidemiology of
DSM-III-R major depression and minor depression
among adolescents and young adults in the
National Comorbidity Survey. Depress Anxiety.
1998;7(1):3-14.
3. Lewinsohn PM, Rohde P, Seeley JR, Klein DN,
Gotlib IH. The consequences of adolescent major
depressive disorder on young adults. In: Joiner TE,
Brown JS, Kistner J, eds. The Interpersonal,
Cognitive, and Social Nature of Depression. Mahwah,
NJ: Lawrence Erlbaum Associates; 2006:43-68.
4. Lewinsohn PM, Rohde P, Seeley JR, Klein DN,
Gotlib IH. Psychosocial functioning of young adults
who have experienced and recovered from major
depressive disorder during adolescence. J Abnorm
Psychol. 2003;112(3):353-363.
5. Bridge JA, Goldstein TR, Brent DA. Adolescent
suicide and suicidal behavior. J Child Psychol
Psychiatry. 2006;47(3-4):372-394.

38 JAMA Psychiatry January 2015 Volume 72, Number 1 jamapsychiatry.com

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a Universite Laval User on 04/25/2016

 Depression Outcomes in Teenagers and Young Adults
6. Hammen C, Brennan PA, Keenan-Miller D,
Herr NR. Early onset recurrent subtype of
adolescent depression: clinical and psychosocial
correlates. J Child Psychol Psychiatry. 2008;49(4):
433-440.
7. Lewinsohn PM, Rohde P, Klein DN, Seeley JR.
Natural course of adolescent major depressive
disorder, I: continuity into young adulthood. J Am
Acad Child Adolesc Psychiatry. 1999;38(1):56-63.
8. Jonsson U, Bohman H, von Knorring L, Olsson G,
Paaren A, von Knorring AL. Mental health outcome
of long-term and episodic adolescent depression:
15-year follow-up of a community sample. J Affect
Disord. 2011;130(3):395-404.
9. Substance Abuse and Mental Health Services
Administration. Results From the 2012 National
Survey on Drug Use and Health: Mental Health
Findings, NSDUH Series H-47, HHS Publication No.
(SMA) 13-4805. Rockville, MD: Substance Abuse and
Mental Health Services Administration; 2013.
10. Lewis K, Burd-Sharps S. Halve the Gap by 2030:
Youth Disconnection in America’s Cities. New York, NY:
Social Science Research Council; 2013.
11. Annie E. Casey Foundation. Kids Count Message
Highlights Perils of Ignoring “Disconnected Youth.”
Baltimore, MD: Annie E. Casey Foundation; 2004.
12. Fernandes A, Gabe T. Disconnected Youth:
A Look at 16- to 24-Year Olds Who Are Not Working
or in School. Washington, DC: Congressional Research
Service; 2009.
13. The Aspen Institute. Aspen Forum for
Community Solutions and Opportunity Youth
Incentive Fund. http://aspencommunitysolutions
.org/the-fund/. Accessed May 9, 2014.
14. Tandon SD, Marshall B, Templeman AJ,
Sonenstein FL. Health access and status of
adolescents and young adults using youth
employment and training programs in an urban
environment. J Adolesc Health. 2008;43(1):30-37.
15. Harris L. What’s a Youngster to Do? The
Education and Labor Market Plight of Youth in High
Poverty Communities. Washington, DC: Center for
Law and Social Policy; 2005.
16. Government Accountability Office.
Disconnected Youth: Federal Action Could Address
Some of the Challenges Faced by Local Program
That Reconnect Youth to Education and
Employment. Washington, DC: Government
Accountability Office; 2008.
jamapsychiatry.com
Copyright 2015 American Medical Association. All rights reserved.
Downloaded From: http://archpsyc.jamanetwork.com/ by a Universite Laval User on 04/25/2016
17. Compas BE, Connor-Smith JK, Saltzman H,
Thomsen AH, Wadsworth ME. Coping with stress
during childhood and adolescence: problems,
progress, and potential in theory and research.
Psychol Bull. 2001;127(1):87-127.
18. Horowitz JL, Garber J. The prevention of
depressive symptoms in children and adolescents:
a meta-analytic review. J Consult Clin Psychol.
2006;74(3):401-415.
19. Weare K, Nind M. Mental health promotion and
problem prevention in schools: what does the
evidence say? Health Promot Int. 2011;26(suppl 1):
i29-i69.
20. Gladstone TRG, Beardslee WR, O’Connor EE.
The prevention of adolescent depression. Psychiatr
Clin North Am. 2011;34(1):35-52.
21. Radloff LS. The CES-D scale: a self-report
depression scale for research in the general
population. Appl Psychol Meas. 1977;1(3):385-401.
doi:10.1177/014662167700100306.
22. Ayers TS, Sandler IN, West SG, Roosa MW.
A dispositional and situational assessment of
children’s coping: testing alternative models of
coping. J Pers. 1996;64(4):923-958.
23. Seidman E, Allen L, Aber JL, et al. Development
and validation of adolescent-perceived
microsystem scales: social support, daily hassles,
and involvement. Am J Community Psychol. 1995;
23(3):355-388.
24. Jaycox LH, Asarnow JR, Sherbourne CD, Rea
MM, LaBorde AP, Wells KB. Adolescent primary care
patients’ preferences for depression treatment.
Adm Policy Ment Health. 2006;33(2):198-207.
25. Beck AT, Epstein N, Brown G, Steer RA. An
inventory for measuring clinical anxiety:
psychometric properties. J Consult Clin Psychol.
1988;56(6):893-897.
26. D’Imperio RL, Dubow EF, Ippolito MF. Resilient
and stress-affected adolescents in an urban setting.
J Clin Child Psychol. 2000;29(1):129-142.
27. Stuart EA. Matching methods for causal
inference: a review and a look forward. Stat Sci.
2010;25(1):1-21.
28. Garber J. Depression in children and
adolescents: linking risk research and prevention.
Am J Prev Med. 2006;31(6, suppl 1):S104-S125.
29. Rosenbaum PR. Observational study. In: Everitt
BS, Howell D, eds. Encyclopedia of Statistics in
JAMA Psychiatry January 2015 Volume 72, Number 1
Original Investigation Research
Behavioral Science. Chichester, West Sussex: John
Wiley & Sons; 2005;3:1451-1462.
30. Ho D, Imai K, King G, Stuart E. Matching as
nonparametric preprocessing for reducing model
dependence in parametric causal inference. Polit Anal.
2007;15(3):199-236. doi:10.1093/pan/mpl013.
31. Ho D, Imai K, King G, Stuart E. MatchIt:
nonparametric preprocessing for parametric causal
inference. J Stat Softw. 2011;42(8):1-28.
http://www.jstatsoft.org/v42/i08. Accessed October
1, 2014.
32. Rubin DB. Using propensity scores to help
design observational studies: application to the
tobacco litigation. Health Serv Outcomes Res
Methodol. 2001;2(3-4):169-188. doi:10.1023/A:
1020363010465.
33. StataCorp. Stata Statistical Software: Release
12. College Station, TX: StataCorp LP; 2011.
34. Allison PD. Change scores as dependent
variables in regression analysis. Sociol Methodol.
1990;20:93-114. doi:10.2307/271083.
35. Lu B, Greevy R, Xu X, Beck C. Optimal
nonbipartite matching and its statistical
applications. Am Stat. 2011;65(1):21-30.
36. Baltimore Neighborhood Indicators Alliance.
Baltimore Community Statistical Areas (CSAs) 2010
[data file]. http://bniajfi.org/mapping-resources/.
Accessed May 9, 2014.
37. Spence SH, Shortt AL. Research review: can we
justify the widespread dissemination of universal,
school-based interventions for the prevention of
depression among children and adolescents? J Child
Psychol Psychiatry. 2007;48(6):526-542.
38. Merry S, McDowell H, Hetrick S, Bir J, Muller N.
Psychological and/or educational interventions for
the prevention of depression in children and
adolescents. Cochrane Database Syst Rev. 2004;12
(1):CD003380.
39. Jané-Llopis E, Hosman C, Jenkins R, Anderson
P. Predictors of efficacy in depression prevention
programmes: meta-analysis. Br J Psychiatry. 2003;
183:384-397.
40. Reinecke MA, Curry JF, March JS. Findings
from the Treatment for Adolescents with
Depression Study (TADS): what have we learned?
what do we need to know? J Clin Child Adolesc
Psychol. 2009;38(6):761-767.
39