ORIGINAL PAPER

Stress and development of depression and heavy drinking in adulthood: moderating effects of childhood trauma

Ian Colman • Yasmin Garad • Yiye Zeng • Kiyuri Naicker •

Murray Weeks • Scott B. Patten • Peter B. Jones •

Angus H. Thompson • T. Cameron Wild

Received: 30 November 2011 / Accepted: 23 May 2012 / Published online: 9 June 2012

� Springer-Verlag 2012

Abstract

Purpose Studies suggest that childhood trauma is linked

to both depression and heavy drinking in adulthood, and

may create a lifelong vulnerability to stress. Few studies

have explored the effects of stress sensitization on the

development of depression or heavy drinking among those

who have experienced traumatic childhood events. This

study aimed to determine the effect of childhood trauma on

the odds of experiencing depression or heavy drinking in

the face of an adult life stressor, using a large population-

based Canadian cohort.

Methods A total of 3,930 participants were included from

the National Population Health Survey. The associations

among childhood trauma, recent stress and depression/heavy

drinking from 1994/1995 to 2008/2009 were explored using

logistic regression, as were interactions between childhood

trauma and recent stress. A generalized linear mixed model

was used to determine the effects of childhood trauma and

stressful events on depression/heavy drinking. Analyses

were stratified by sex.

Results Childhood trauma significantly increased the

odds of becoming depressed (following 1 event: OR =

1.66; 95 %CI 1.01, 2.71; 2? events, OR = 3.89; 95 %CI

2.44, 6.22) and drinking heavily (2? events: OR = 1.79;

95 %CI 1.03, 3.13). Recent stressful events were associ-

ated with depression, but not heavy drinking. While most

interaction terms were not significant, in 2004/2005 the

association between recent stress and depression was

stronger in those who reported childhood trauma compared

to those with no childhood trauma.

Conclusions Childhood trauma increases risk for both

depression and heavy drinking. Trauma may moderate the

effect of stress on depression; the relationship among

trauma, stress and heavy drinking is less clear.

Keywords Childhood trauma � Stress � Depression � Alcohol abuse � Epidemiology

Introduction

Several studies show that childhood trauma is associated

with the development of depression [1–3] and heavy

drinking [1, 4–7] in adulthood. Traumatic events vary in

their frequency and severity, but many have been associ-

ated with depression and heavy drinking later in life,

including physical and sexual abuse [4–8], parental divorce

[9–11] and exposure to violence [8, 12].

Childhood trauma may lead to depression through its

effect on the stress response. The stress sensitization model

I. Colman � Y. Garad � Y. Zeng � K. Naicker � T. C. Wild

School of Public Health, University of Alberta,

Edmonton, Canada

I. Colman (&) � M. Weeks

Department of Epidemiology and Community Medicine,

University of Ottawa, 451 Smyth Road, RGN 3230C,

Ottawa, ON K1H 8M5, Canada

e-mail: [email protected]

S. B. Patten

Departments of Psychiatry and Community Health Sciences,

University of Calgary, Calgary, Canada

P. B. Jones

Department of Psychiatry, University of Cambridge,

Cambridge, UK

A. H. Thompson

Institute of Health Economics, Edmonton, Canada

123

Soc Psychiatry Psychiatr Epidemiol (2013) 48:265–274

DOI 10.1007/s00127-012-0531-8

suggests that individuals who experience traumatic events

in childhood have a lower tolerance for stress in adulthood,

and consequently may be more likely to develop mental

illness after stressful life events compared to those who do

not experience trauma in childhood [13]. Biological evi-

dence supports this model; both animal [14, 15] and human

studies [16] have shown changes in stress reactivity after

exposure to early trauma. Persistent sensitization of the

stress response alters the hypothalamic–pituitary–adrenal

(HPA) axis, which is linked to the development of

depression [17, 18]. Elevated levels of catecholamine and

cortisol that are consequent to trauma may have a neuro-

degenerative impact on the developing brain [17]. Simi-

larly, the link between childhood trauma and alcohol abuse

may be mediated by stress reactivity. In Rhesus monkeys,

parental separation leads to increased cortisol levels and

excessive alcohol consumption [19]. It has been suggested

that the link between childhood trauma and alcohol abuse

is due to individuals using alcohol as a means to reduce the

effects of a dysregulated biological stress response system

or reduce the symptoms of depression [17].

In spite of a wealth of biological evidence suggesting

that childhood trauma can create a lifelong vulnerability to

stress [13, 20], there is little evidence from population-

based human studies demonstrating this effect outside

laboratory conditions. A notable exception is a recent

paper, which found that individuals with multiple child-

hood traumas were significantly more likely to develop

major depression, post-traumatic stress disorder and anxi-

ety disorders after major stressful events, when compared

to individuals with no childhood trauma [21]. The primary

objectives of the current study were to: use data from a

Canadian longitudinal cohort study to replicate the finding

that individuals who suffer from traumatic childhood

events are more likely to become depressed after stressful

events in adulthood compared to those without childhood

trauma, using repeated measures over a 16-year period; to

investigate whether stress sensitization effects also exist in

the relationship between childhood trauma and heavy

drinking in adulthood. We hypothesized that those who

have experienced traumatic childhood events would be

significantly more likely to become depressed or drink

heavily after stressful events in adulthood compared to

those without childhood trauma.

Methods

Sample

The National Population Health Survey (NPHS) is a

nationwide longitudinal study conducted by Statistics

Canada, which started in 1994/1995 and included health

and other health-related information, such as economic,

social, demographic, occupational and environmental data.

At study inception, 17,276 individuals were randomly

selected using a stratified two-stage sample design. The

cohort is representative of the Canadian population, and

has been followed up every 2 years. The first cycle of data

collection (1994/1995) had a response rate of 83.6 %, and

63.6 % of the original cohort were still participating in

2006/2007 [22]. A total of 14,117 members of the NPHS

aged 18 years or above in 1994/1995 were eligible for this

study, while 13,020 members completed traumatic events

questions. Among these, 7,275 individuals completed

traumatic events questions again in 2006/2007. Among

those with missing data, 26.70 % was due to survey

member death, 1.74 % due to institutionalization, 54.60 %

did not respond to the survey in 2006/2007, while the

remainder responded to the survey but did not answer all

seven traumatic event questions (see Fig. 1). Only indi-

viduals who responded to all childhood traumatic events

questions and who also consistently recalled traumatic

events were included, in order to ensure that individuals

were not selectively reporting childhood trauma according

to their current mental state [23]. The final study sample

included 3,930 participants.

NPHS 1994/95 (n=17276)

1994/95 Age<18 (n=3159)

1994/95 Age>=18 (n=14117)

1994/95 Completed all 7 traumatic

events questions (n=13020)

2006/07 Completed all 7 traumatic events questions (n=7275)

2006/07: Deceased (n=1871)

2006/07: Institutionalized (n=138)

2006/07: Didn’t respond (n=2778)

2006/07: Responded but did not complete all 7 traumatic events questions (n=958)

Study sample: Consistently recalled all 7

traumatic events questions in 1994/95 and 2006/07

(n=3930)

Fig. 1 Study sample inclusion and exclusion

266 Soc Psychiatry Psychiatr Epidemiol (2013) 48:265–274

123

Outcome: depression or heavy drinking

Two main outcomes were considered, the occurrence of

major depression or heavy drinking, from data collection

cycles from 1996/1997 to 2008/2009. Major depression in

the NPHS is captured by the Composite International

Diagnostic Interview-Short Form (CIDI-SF). The CIDI-SF,

a 10-min interview, has been found to have 90 % sensitivity

and 94 % specificity in identifying a major depressive

episode compared with the full CIDI [24], an hour-long

interview that can identify depressive episodes consistent

with the Diagnostic and Statistical Manual of Mental

Disorders (DSM-IV) [25]. For each individual, a score of 5

or higher on a 0–8 scale was considered as having major

depression [24]. This corresponds to DSM-IV criteria for a

major depressive episode: five of nine depressive symp-

toms in a 2-week period during the past year, including

either loss of interest or depressed mood [25].

Alcohol consumption was assessed at each cycle of the

NPHS. Individuals were considered to be drinking heavily

if they reported consuming an average of more than 17

drinks per week for a man and more than 12 drinks per

week for a woman over the previous year [26].

Childhood traumatic events

All survey members above 18 years were asked in

1994/1995 whether they had experienced any of the fol-

lowing events as a child or teenager: (1) 2 weeks or longer

in hospital; (2) parental divorce; (3) parental unemploy-

ment; (4) a frightening experience that was thought about

for years after; (5) being sent away from home for

wrongdoing; (6) family problems due to parental substance

abuse; and (7) physical abuse by someone ‘‘close’’ [27].

The same questions were repeated in 2006/2007. Only

individuals who consistently recalled all seven questions

were included in the study. All individuals in this study

were assigned to three groups: those who experienced none

of these events during childhood, those who experienced

one of these events and those who experienced two or more

events.

Recent stressors

Eight indices of recent stress were considered: marital

disruption, recent unemployment, poor health, household

financial problems, injury, decreased social support, high

work stress and high chronic stress. Marital disruption was

defined by a change from single/married/partnered to

divorced/widowed/separate since last interview. Recent

unemployment was defined as being employed 2 years

previously and currently unemployed or not in the labor

force. Poor physical health was defined as either develop-

ing a chronic illness or a decrease in self-rated health from

good/very good/excellent to fair/poor during the last

2 years. Household financial problems were defined as a

drop below Statistic Canada’s low income cutoff (LICO)

since the last interview. The LICO score takes into account

an individual’s income relative to the community in which

an individual lives and the size of the family [28]. Injury

was defined as suffering from a new injury in the last

2 years. Social support was measured by four items: having

someone to confide in or talk to about private feelings or

concerns, having someone to really count on to help out in

a crisis situation, having someone to really count on to give

advice when making important personal decisions and

having someone who makes you feel loved and cared for

[27]. A drop from having three or four positive answers

2 years ago to having one or zero positive answer was

considered to be a recent decrease in social support. Work

stress is measured in the NPHS by 13 questions that assess

job security, autonomy, conflict and satisfaction [27].

Those above the 90th percentile on this scale were con-

sidered to have high work stress. Chronic stress is mea-

sured by 18 questions that assess stress in personal life,

with a primary focus on relationship and family strife [27].

Those above the 90th percentile on this scale were con-

sidered to have high chronic stress. Work stress and

chronic stress were measured only from 2000/2001, while

the other six items were measured at all cycles. Conse-

quently, a recent stressful life event index was calculated

based on six items in 1996/1997 and 1998/1999 (a 0–6

scale) and eight items (a 0–8 scale) from 2000/2001 to

2008/2009. Occurrence of recent stressful life events was

treated as a three-category variable (0, 1 or 2? events).

Statistical analysis

In the first stage of the analysis, logistic regression was

used to investigate the association between childhood

trauma, recent stress and depression/heavy drinking in

eight reporting cycles from 1994/1995 to 2008/2009.

Model covariates included age and gender. Interactions

between childhood trauma and recent stress were explored

to identify whether individuals who reported childhood

trauma were more likely to be depressed or drinking

heavily in the face of recent stress. The standard errors for

all estimates were calculated using the bootstrap method

[29]. All estimates were weighted to adjust for unequal

selection probabilities and cluster sampling, ensuring that

results were representative of the Canadian population.

To account for the repeated measures, the effects of

childhood traumatic events and stressful life events on

depression or heavy drinking were explored using the

generalized linear mixed model (GLMMIX) with logit link

Soc Psychiatry Psychiatr Epidemiol (2013) 48:265–274 267

123

in the second stage of the analysis. The mixed model can

handle missing data due to individuals dropping out of the

study or selectively completing questionnaires (i.e., can

allow for partially complete data) under a missing at ran-

dom assumption. In the mixed model, stressful life events

were considered as a time-dependent factor, while age,

sex and childhood traumatic events were time-independent

variables. All estimates were weighted to adjust for unequal

selection probabilities and cluster sampling.

Previous reports have identified differing effects of

childhood trauma and stressful life events on depression by

sex [21]. Consequently, all analyses were stratified by sex

after initial combined analyses. Interactions by sex were

also explored. In addition, we explored whether the effects

of childhood trauma and recent stress were modified by

age. Interaction terms between childhood trauma and age,

recent stress and age, and a three-way interaction between

childhood trauma, recent stress and age were explored.

SAS 9.2 (SAS Institute Inc., Cary, North Carolina,

USA) was used for all analyses.

This study was approved by the Health Research Ethics

Board of the University of Alberta. Written informed

consent was obtained from survey members by Statistics

Canada.

Results

The prevalence of 12-month major depression in the

baseline sample was 3.7 %; the prevalence of heavy

drinking was slightly higher (see Table 1). Prevalence rates

were similar for the subsequent cycles. Depression was

more common among females compared to males, while

the prevalence of heavy drinking was higher among males

compared to females; 26.63 % of the participants reported

they had at least one childhood traumatic event. The most

commonly reported event was being in hospital for

2 weeks or more (8.9 %), while the least common event

was being sent away from home (0.5 %). In each cycle,

approximately 35 % of the sample reported a recent

stressor.

Several differences between those who were eligible for

the study and those who were not were observed (see

Table 1). Notably, those who were inconsistent in their

reporting of childhood traumatic events (i.e., reported a

childhood event at one time point, but did not report the

same event 12 years later), those who did not complete all

questions on traumatic events, and those who dropped out

before the end of the study were more likely to be

depressed, drink heavily, rate their health poorly, report

high levels of stress and report a higher number of child-

hood traumatic events at baseline than the final study

sample.

Both childhood trauma and recent stressors were sig-

nificantly associated with major depression (Table 2).

Reporting two or more childhood traumatic events was

consistently associated with adult depression; the associa-

tion between one childhood trauma and adult depression

was less consistent. A similar gradient was observed for the

relationship between recent stressors and depression. For

both childhood trauma and recent stressors, there was on

numerous occasions a significant association with depres-

sion for females but not males, although this difference

between genders was not statistically significant (i.e., no

significant interaction effects were found). An investigation

of the interactions between childhood trauma and recent

stress yielded mixed results. While most interaction terms

were not significant, in 2004/2005 the association between

recent stress and depression was stronger in those who

reported childhood trauma compared to those with no

childhood trauma.

The association between childhood trauma and heavy

drinking was inconsistent. In 2000/2001, 2004/2005,

2006/2007 and 2008/2009, individuals who reported two or

more traumatic events in childhood were more likely to be

drinking heavily (Table 3). Recent stress was not associ-

ated with alcohol abuse, with the exception that those with

two or more recent stressors were more likely to be

drinking heavily in 2000/2001 compared to those with no

recent stress. There were no significant differences by sex,

and interactions between childhood trauma and recent

stress were not significant.

Age did not modify the effect of childhood trauma,

recent stress, or the interaction between childhood trauma

and recent stress on either depression or heavy drinking

(interaction terms non-significant).

Results from the GLMMIX mixed models were con-

sistent with findings from the logistic regression models

described above (Tables 2, 3).

Discussion

This is the first study to our knowledge to assess the impact

of childhood trauma on the associations among stressful

life events and depression and heavy drinking using a large

population-based sample. Our findings support the hypo-

thesis that those who experienced traumatic childhood

events may be more likely to develop depression in adult-

hood following recent stressful life events, compared to

those who had no traumatic experiences in childhood. Our

findings do not support this equivalent hypothesis in the

prediction of heavy drinking.

The gender differences observed in the prevalences of

these outcomes, with women twice as likely to become

depressed and men experiencing a higher frequency of

268 Soc Psychiatry Psychiatr Epidemiol (2013) 48:265–274

123

T a

b le

1 B

as el

in e

ch ar

ac te

ri st

ic s

o f

st u

d y

p ar

ti ci

p an

ts in

1 9

9 4

/1 9

9 5

% P

re v

al en

ce o

r m

ea n

(S D

)

F em

al es

(n =

2 ,2

2 1

)

M al

es

(n =

1 ,7

0 9

)

T o

ta l

st u

d y

sa m

p le

(N =

3 ,9

3 0

)

In co

n si

st en

t re

p o

rt s

o f

tr au

m a

(n =

3 ,3

4 5

)

D ec

ea se

d

(n =

1 ,8

7 1

)

In st

it u

ti o

n al

iz ed

(n =

1 3

8 )

N o

re sp

o n

se in

2 0

0 6

/2 0

0 7

(n