Health Psychology Copyright 2008 by the American Psychological Association

2008, Vol. 27, No. 5, 604 615 0278-6133/08/$12.00 DOI: 10.1037/a0013242

Pregnancy-Specific Stress, Prenatal Health Behaviors, and Birth Outcomes

Marci Lobel and Dolores Lacey Cannella Jennifer E. Graham

Stony Brook University Pennsylvania State University

Carla DeVincent and Jayne Schneider Bruce A. Meyer

Stony Brook University University of Texas Southwestern Medical School

Objective: Stress in pregnancy predicts earlier birth and lower birth weight. The authors investigated
whether pregnancy-specific stress contributes uniquely to birth outcomes compared with general stress,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

and whether prenatal health behaviors explain this association. Design: Three structured prenatal
This document is copyrighted by the American Psychological Association or one of its allied publishers.

interviews (N 279) assessing state anxiety, perceived stress, life events, pregnancy-specific stress, and
health behaviors. Main Outcome Measures: Gestational age at delivery, birth weight, preterm delivery
(37 weeks), and low birth weight (2,500 g). Results: A latent pregnancy-specific stress factor
predicted birth outcomes better than latent factors representing state anxiety, perceived stress, or life
event stress, and than a latent factor constructed from all stress measures. Controlling for obstetric risk,
pregnancy-specific stress was associated with smoking, caffeine consumption, and unhealthy eating, and
inversely associated with healthy eating, vitamin use, exercise, and gestational age at delivery. Cigarette
smoking predicted lower birth weight. Clinically-defined birth outcomes were predicted by cigarette
smoking and pregnancy-specific stress. Conclusion: Pregnancy-specific stress contributed directly to
preterm delivery and indirectly to low birth weight through its association with smoking. Pregnancy-
specific stress may be a more powerful contributor to birth outcomes than general stress.

Keywords: pregnancy, stress, birth weight, prematurity, health behaviors

Research over the last two decades establishes that women who
experience high stress during pregnancy deliver infants earlier, and
deliver infants who weigh less at birth (Copper et al., 1996; Davis
et al., 2005; Dole et al., 2003; Hedegaard, Henriksen, Secher,
Hatch, & Sabroe, 1996; Hoffman & Hatch, 1996; Lobel, Dunkel-
Schetter, & Scrimshaw, 1992; Lobel, 1994; Wadhwa, Sandman,
Porto, Dunkel-Schetter, & Garite, 1993). Preterm delivery (37
weeks) and low birth weight (2,500g) are leading causes of
infant mortality, infant morbidity, and health problems that may
persist into childhood, adolescence and adulthood (Barker, 2007;
Goldenberg et al., 1996; Hack & Merkatz, 1995; Mathews &
MacDorman, 2006; Newnham, 1998; OConnor, Heron, & Glover,
2002).
Prenatal maternal stress (PNMS) has been operationalized in a
variety of ways (Lobel, 1994). Much of the research has assessed
Marci Lobel and Dolores Lacey Cannella, Department of Psychology,
Stony Brook University; Jennifer E. Graham, Department of Biobehavioral
Health, Pennsylvania State University; Carla DeVincent, Department of
Pediatrics, Stony Brook University; Jayne Schneider, Department of Psy-
chiatry, Stony Brook University; Bruce A. Meyer, Department of Obstet-
rics and Gynecology, University of Texas Southwestern Medical School.
This research was funded by National Institutes of Health Grant
NR03443. We are grateful to the staff of the Stony Brook Pregnancy
Project, the staff of the prenatal care facility, and all of the women who
participated in this study. We also thank Dr. Judith A. Stein for statistical
consultation.
Correspondence concerning this article should be addressed to Marci
Lobel, Department of Psychology, Stony Brook University, Stony Brook,
NY 11794-2500. E-mail: marci.lobel@stonybrook.edu
stress that is general or nonspecific to pregnancy. For example,
two longstanding approaches involve the assessment of emotions
during pregnancy, especially anxiety, and assessment of stressful
prenatal stimuli or conditions, especially major life events (Lobel
& Dunkel-Schetter, 1990). These approaches to PNMS have meth-
odological and conceptual shortcomings that may help explain
why such assessments have yielded inconsistent associations with
birth outcomes across studies (Dunkel-Schetter, 1998; Hogue,
Hoffman, & Hatch, 2001; Lobel, 1994; Stanton, Lobel, Sears, &
DeLuca, 2002). For example, focusing on state anxiety alone or
counting the number of life events that occur during pregnancy
does not reveal what women are anxious about or whether they
experience life events as stressful. In other words, these ap-
proaches fail to consider womens perceptions or appraisals of
their experiences (Lobel, 1994). Several research teams (e.g.,
Molfese et al., 1987; Norbeck & Tilden, 1983) have operationally
defined PNMS as the aggregate of negative emotional state and
stressful events or conditions during pregnancy, and in some cases
have also included pregnant womens perceptions or appraisals of
their circumstances (e.g., Lobel et al., 1992; Rini, Dunkel-Schetter,
Wadhwa, & Sandman, 1999). Such multivariable operationaliza-
tions have shown more consistent associations with birth outcomes
(Lobel, 1994; Roesch, Dunkel-Schetter, Woo, & Hobel, 2004).
A more recent approach to prenatal stress measurement focuses
on pregnancy-specific stress. Pregnant women experience stress
originating from a variety of pregnancy-specific issues, including
physical symptoms, parenting concerns, relationship strains,
bodily changes, anxiety about labor and delivery, and concerns
about the babys health (Affonso, Liu-Chiang, & Mayberry, 1999;

604
 PREGNANCY-SPECIFIC STRESS AND BIRTH OUTCOMES
Halman, Oakley, & Lederman, 1995; Levin, 1991; Lobel, 1998;
Misra, OCampo, & Strobino, 2001; Yali & Lobel, 1999). Evi-
dence from several research teams distinguishes pregnancy-
specific stress from more general types of stress in pregnancy
(DiPietro, Ghera, Costigan, & Hawkins, 2004; Huizink, Mulder,
Robles de Medina, Visser, & Buitelaar, 2004; Lobel, DeVincent,
Kaminer, & Meyer, 2000; Rini et al., 1999; Roesch et al., 2004;
Wadhwa et al., 1993; Yali & Lobel, 1999). For example, DiPietro
and colleagues (DiPietro, Hilton, Hawkins, Costigan, & Pressman
2002; DiPietro et al., 2004) developed a measure that focuses on
pregnancy-specific stressors, namely prenatal hassles and uplifts.
Roesch et al.s (2004) focus was on pregnancy-specific anxiety.
They devised a 4-item measure assessing how anxious, concerned,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
afraid, and panicky women were about being pregnant (see also
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Gurung, Dunkel-Schetter, Collins, Rini, & Hobel, 2005).
An important issue that arises from this work concerns the
independent ability of pregnancy-specific stress to predict birth
outcomes relative to general stress. Some investigators have sug-
gested that pregnancy-specific stress may be a more powerful
contributor to birth outcomes than stress from sources unrelated to
pregnancy (DiPietro et al., 2002, 2004; Huizink et al., 2004;
Roesch et al., 2004). Roesch et al. found that pregnancy anxiety
predicted younger gestational age at birth, whereas its correlate,
general anxiety, did not (general perceived stress and life events
also did not predict gestational age in this study). Similarly,
examining fetal behavior rather than birth outcomes, DiPietro et al.
(2002) found that their measure of pregnancy-specific hassles and
uplifts was a better predictor than general stressors. However,
neither DiPietro et al. nor Roesch et al. compared the predictive
validity of their pregnancy-specific measure to a multivariable
aggregate of general, nonspecific PNMS such as those that have
shown excellent success in predicting birth outcomes in prior
research (Lobel et al., 1992; Rini et al., 1999).
An important reason to investigate the predictive validity of
pregnancy-specific stress vis-a`-vis more elaborate measures of
PNMS is to evaluate whether it is a more powerful contributor to
birth outcomes than general stress. A second reason involves
practical utility. While they may offer conceptual and psychomet-
ric strength, multivariable operationalizations of PNMS are prob-
ably unwieldy for use in clinical settings, where a briefer and
simpler means of measuring stress may be needed. Thus, evaluat-
ing the ability of a brief assessment of pregnancy-specific stress to
predict birth outcomes also has important clinical relevance.
We used a measure of pregnancy-specific stress adapted from
our prior research on PNMS (Lobel et al., 2000; Yali & Lobel,
1999) and examined its ability to predict birth outcomes compared
to a multivariable operational definition of PNMS that included
pregnancy-specific stress and general stress measures that have
been investigated in prior studies, including state anxiety, per-
ceived stress, and life events. Additionally, we compared the
predictive power of the pregnancy-specific stress measure to each
of the individual general stress measures that comprised the mul-
tivariable operational definition (state anxiety, perceived stress,
and life events). These comparisons offered a direct test of the
independent contribution of pregnancy-specific stress, and an op-
portunity to determine whether it is a stronger contributor to birth
outcome than general stress.
605
Behavioral Mechanisms
An additional goal was to examine prenatal health behaviors
that may help explain the impact of PNMS on birth outcomes.
Identifying behavioral mechanisms can help guide interventions
against the effects of PNMS. Highly stressed people engage in
unhealthful behaviors (Ng & Jeffery, 2003; Stetson, Rahn, Dub-
bert, Wilner, & Mercury, 1997) that are likely to contribute to poor
birth outcomes (Dunkel-Schetter, Gurung, Lobel, & Wadhwa,
2001). Stress has been shown to be related to poor nutrition,
inadequate physical activity, cigarette smoking, and alcohol and
other substance use in pregnant women (Barnet, Duggan, Wilson,
& Joffe, 1995; Borrelli, Bock, King, Pinto, & Marcus, 1996;
Bresnahan, Zuckerman, & Cabral, 1992; Chomitz, Cheung, &
Lieberman, 1995; Hutchins & DiPietro, 1997; McCormick et al.,
1990; Pritchard, 1994; Rodriguez, Bohlin, & Lindmark, 2000;
Zuckerman, Amaro, Bauchner, & Cabral, 1989). These studies
provide a basis for proposing that health behaviors during preg-
nancy help account for the impact of PNMS on birth outcomes.
Method
Overview
All aspects of the research were approved by an Institutional
Review Board. The study was conducted at a public, university-
affiliated prenatal care facility in a suburban area of the North-
eastern United States. PNMS and health behaviors were assessed
at three prenatal time points through structured interviews con-
ducted by trained research assistants with volunteer participants.
Medical records were abstracted after birth by obstetric research
nurses. Structural equation modeling (SEM) was used for data
analyses.
Participants
Participants were required to speak English fluently, to be less
than 20 weeks pregnant at recruitment, and to be a minimum of 18
years old. The gestational age requirement was later relaxed to 25
weeks to meet recruitment goals. Approximately 73% of eligible
women approached consented to participate, and 57% of these
(339 women) completed study measures. Others did not complete
the study for reasons including miscarriage, referral to other health
care providers, and relocation. No differences were found between
study completers and noncompleters, with two exceptions. Non-
completers were 47% non-White and 22% had household incomes
under $10,000, compared with 34% and 13% of study completers
( ps .05). Of the 339 study completers, 7 participants with
multiple pregnancies and 27 who had substantial missing data (i.e.,
entire measures missing) were deleted from the sample. A negli-
gible amount of data was missing for other participants (0.1% of
all study items); these were replaced using a linear interpolation
replacement technique in SPSS, which is superior to other missing
data replacement techniques (McKnight, McKnight, Sidani, &
Figueredo, 2007).
After replacement, we followed systematic procedures to eval-
uate multivariate normality (Breckler, 1990; Byrne, 2001). Al-
though sometimes overlooked, multivariate non-normality has im-
portant consequences for SEM results, resulting in inflated chi
square values, underestimated fit indices, and spuriously low stan-
 606 LOBEL ET AL.

dard errors that can produce apparently significant regression
coefficients that may not be true of the population (Breckler, 1990;
Byrne, 2001). Mardias coefficient was greater than 1.96, indicat-
ing non-normality. We therefore ran the first model and examined
Mahalanobis distance (to assess robustness of the departure from
normality) and its two associated p values. Since both p values
were statistically significant ( .05), we removed the participant
farthest from the centroid. The model was rerun, and Mahalanobis
distance was re-examined. We continued to remove one participant
at a time until at least one of the two p values associated with
Mahalanobis distance was no longer statistically significant, as
recommended (Arbuckle, 2003). Using this iterative procedure, 26
multivariate outliers were deleted, resulting in a total sample size
of 279 women.1
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
non-White categories and investigated whether they differed from
self-identified White women on all study variables. There were no
ethnic differences in any of the stress variables at the three time
points, in medical risk or other maternal characteristics such as
age, income, or education, nor in the birth outcome variables.
There were some differences in health behaviors: White women
smoked, consumed caffeine, experienced physical strain, and ate
unhealthy more often than non-White women (all ps .05).
Although prior studies have often reported ethnic differences in
birth outcomes, these tend to be confounded with socioeconomic
status, which was not the case in the present study.
Procedures

The first interview was conducted between 10 and 20 weeks of

This document is copyrighted by the American Psychological Association or one of its allied publishers.

Participant characteristics are presented in Table 1. The sample

was heterogeneous with respect to sociodemographic variables and pregnancy (later extended to 25 weeks after relaxation of the
parity. The rate of low birth weight and preterm delivery was lower eligibility requirement described earlier), M 16.6 (SD 4.4)
than population norms, but this is a likely result of the samples weeks. The second interview was conducted between 21 and 30
restriction to adults with singleton pregnancies initiating prenatal weeks (later, extended to 35 weeks), M 26.1 (SD 3.8) weeks,
care in the first half (approximately) of pregnancy. Each of these and the third interview was conducted after 30 weeks, M 34.1
characteristics is associated with lower prevalence of adverse birth (SD 2.4) weeks. A minimum of 2 weeks was maintained
outcomes. A number of maternal characteristics were correlated. between participant interviews, and in no case did a participant
For example, older women had more prior pregnancies and births, complete a second or third interview prior to completing the
greater education and income, and were more likely to be married preceding one(s).
or partnered (all ps .05).
Because of the small number of women in each ethnic group, we Assessment of PNMS and Prenatal Health Behaviors
combined all participants who identified themselves in any of the
PNMS was assessed with multiple measures: pregnancy-
Table 1 specific stress, stressful life events, state anxiety, and perceived
Characteristics of Study Participants (N 279) stress. Each measure has been used previously and possesses good
psychometric properties, including testretest reliability and con-
Frequency % vergent and predictive validity (Hamilton & Lobel, 2008; Lobel &
Dunkel-Schetter, 1990; Lobel et al., 1992; 2000; Yali & Lobel,
Birthweight
2500g 11 3.9
2002). Life events since conception were assessed in the third
2500 g 268 96.1 interview; events occurring between the third interview and birth
Weeks gestation at birth were assessed in a telephone interview 1 month after birth. The
Preterm ( 37 weeks) 21 7.5 remaining assessments were administered in all three prenatal
Full-term (37 weeks) 258 92.5 interviews. Each measure is described below.
Age
M 27.0; SD 5.9 A revised version of the Prenatal Distress Questionnaire (PDQ;
Ethnicity Yali & Lobel, 1999) assesses stress originating from issues com-
White 182 65.2 mon in pregnancy (see Appendix). The revised PDQ includes 9
African American or Black 33 11.8 items administered at each interview, plus unique items added to
Latino or Hispanic 33 11.8
Multi-ethnic 23 8.2
the second and third interviews to assess issues that become more
Asian or Pacific Islander 6 2.2 relevant as pregnancy progresses (e.g., concerns about caring for a
Native American 2 .7 newborn). Respondents indicate the extent to which they are
Married or partnered feeling bothered, upset, or worried at this point about issues
Yes 177 63.4 including medical care, physical symptoms, parenting, bodily
No 102 36.6
Parity changes, and the infants health. Responses are on a 3-point scale
Primiparous 60 21.5 ranging from 0 (not at all) to 2 (very much). Average pregnancy-
Multiparous 219 78.5 specific distress scores were calculated for each time point. As
Annual household income indicated by Cronbachs alphas (see Table 2), items comprising the
$10,000 32 11.5
$10,000$30,000 145 52.0
revised PDQ were modestly intercorrelated at each time point,
$30,000$50,000 71 25.0 reflecting the expected independence of some of the items. This
$50,000 31 11.0
Education
Some high school 38 14.0 1
To ensure that study results would not have been substantially different
High school graduate 111 39.8 if departures from multivariate normality were ignored, we reran some of
Some college 104 37.3 the main models, retaining the outliers. The pattern of findings was
College graduate 23 8.2 essentially the same, although as expected, there were some differences in
Graduate degree 3 1.1
the value of fit statistics.
 PREGNANCY-SPECIFIC STRESS AND BIRTH OUTCOMES 607

Table 2
Correlations of Study Variables

Variable name 1 2 3 4 5 6 7 8 9 10 11 12

1. Pregnancy-specific stress T1
2. Pregnancy-specific stress T2 .65
3. Pregnancy-specific stress T3 .64 .72
4. Perceived stress T1 .51 .50 .47
5. Perceived stress T2 .34 .52 .44 .58
6. Perceived stress T3 .40 .50 .45 .59 .68
7. State anxiety T1 .47 .45 .43 .52 .38 .41
8. State anxiety T2 .42 .57 .51 .47 .49 .50 .53
9. State anxiety T3 .41 .48 .52 .38 .42 .57 .49 .56
10. Prenatal life events .33 .36 .44 .37 .36 .31 .25 .26 .26
11. Life event distress .24 .24 .30 .23 .28 .27 .22 .17 .26 .33
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

12. Obstetric risk .07 .09 .03 .03 .05 .03 .12 .08 .07 .09 .13
This document is copyrighted by the American Psychological Association or one of its allied publishers.

13. Cigarettes .20 .20 .15 .20 .11 .19 .12 .11 .12 .07 .08 .08
14. Caffeine .18 .13 .11 .13 .08 .16 .06 .06 .13 .06 .06 .02
15. Unhealthy eating .27 .25 .21 .24 .22 .19 .23 .22 .21 .11 .07 .08
16. Physical strain .07 .08 .11 .13 .14 .23 .08 .06 .10 .12 .12 .04
17. Healthy eating .13 .12 .07 .25 .26 .25 .26 .20 .23 .01 .09 .09
18. Vitamins .15 .16 .15 .22 .10 .14 .11 .21 .20 .06 .05 .06
19. Exercise .13 .14 .09 .20 .15 .11 .21 .17 .19 .01 .10 .09
20. Gestational age (weeks) .15 .20 .16 .05 .02 .05 .13 .09 .14 .03 .12 .27
21. Birth weight (grams) .07 .10 .12 .01 .01 .09 .08 .07 .12 .03 .02 .17
22. 37 weeks (0 No; 1 Yes) .17 .14 .14 .07 .03 .13 .10 .12 .13 .10 .06 .10
23. 2,500 grams (0 No; 1 Yes) .05 .01 .03 .10 .09 .01 .01 .06 .04 .01 .02 .13
Meana .72 .59 .55 6.38 6.16 5.39 9.57 7.57 7.65 5.09 2.78 2.08
SD .35 .32 .30 3.30 3.06 3.13 6.65 5.93 5.72 3.37 .93 1.50
Alpha .59 .71 .79 .78 .79 .82 .88 .89 .87 .68 .71 .31

Variable name 13 14 15 16 17 18 19 20 21 22 23

13. Cigarettes
14. Caffeine .36
15. Unhealthy eating .06 .23
16. Physical strain .14 .20 .17
17. Healthy eating .19 .14 .15 .13
18. Vitamins .02 .13 .16 .17 .31
19. Exercise .08 .04 .17 .15 .33 .03
20. Gestational age (weeks) .12 .01 .08 .06 .04 .02 .09
21. Birth weight (grams) .21 .06 .01 .03 .01 .03 .03 .57
22. 37 weeks (0 No; 1 Yes) .13 .02 .03 .04 .07 .08 .01 .47 .32
23. 2,500 grams (0 No; 1 Yes) .15 .02 .12 .06 .08 .11 .01 .33 .38 .57
Meana .80 1.73 1.69 1.69 2.90 3.38 1.26 38.90 3,357 1.92 1.96
SD 1.27 1.03 .63 .74 .63 .99 .91 1.60 501 .26 .19
Alpha .61 .73 .71 .77
a
Means for pregnancy-specific stress and health behavior variables are item means; means for other stress variables and obstetric risk are scale means.
*
p .05, p .01.

suggests that the measure differentiates concerns about specific
aspects of pregnancy.
The Prenatal Life Events Scale (PLES) was adapted from prior
versions (Lobel et al., 1992, 2000). Respondents report which of
28 events they experienced during pregnancy (e.g., being robbed,
being involved in a serious accident, having someone close die).
For each event endorsed, respondents report how undesirable or
negative the event was on a scale from 0 (not at all) to 3 (very
much). Two indices were computed: number of life events and
mean life event distress.
The 10-item State Anxiety subscale of the State-Trait Person-
ality Inventory (STPI; Spielberger, 1995) assesses how anxious
respondents feel currently. Participants rated the applicability of
items such as I feel nervous, and I feel calm (reverse scored)
on a 4-point scale from 0 (not at all) to 3 (very much). The STPI
is highly correlated with its 20-item parent measure, the State-Trait
Anxiety Inventory, and possesses equivalently strong psychomet-
ric properties. A 4-item version of the Perceived Stress Scale
(Cohen & Williamson, 1988) assessed appraisals of nonspecific
stress. Each item is rated on a 5-point scale from 0 (never) to 4
(very often).
Self-reported health behaviors over the prior two weeks were
assessed using the Prenatal Health Behaviors Scale (DeLuca &
Lobel, 1995). This measure examines a range of healthful and
unhealthful behaviors including nutrition, exercise, smoking, and
substance abuse. Responses are provided on a 5-point scale from
0 (never) to 4 (very often). Seven subscales were identified through
factor analyses of the measure: cigarette smoking (1 item), caffeine
consumption (1 item), healthy eating (3 items calcium, high-
fiber foods, & fruits and vegetables), prenatal vitamin use (1 item),
 608 LOBEL ET AL.
exercise (2 itemsstretching or calisthenics & exercise), physical
strain (3 itemslifting heavy objects, overstretching, & standing
for long periods), and unhealthy eating (3 itemssnack foods,
fatty or oily foods, & overeating). Prior studies using this measure
(DeLuca & Lobel, 1995; Lobel et al., 2000) have found predicted
associations with outcomes. Although it is likely that social desir-
ability concerns affect pregnant womens reports of their health
behaviors, recent research finds that self-report measurement of
cigarette smoking during pregnancy is reliable and accurate, and
highly correlated with serum cotinine levels (McDonald, Perkins,
& Walker, 2005; Okah, Cai, Dew, & Hoff, 2005). Similarly,
physical activity recall measures show strong correlations with
objective measures of activity in pregnant women (Lindseth &
Vari, 2005; Timperio, Salmon, & Crawford, 2003) and there is
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
some evidence that pregnant women reliably report their eating
behaviors (Verbeke & De Bourdeaudhuij, 2007).
Assessment of Obstetric Risk and Birth Outcomes
Obstetric risk was statistically controlled because it is a predic-
tor of birth outcomes, typically associated with health behaviors,
and sometimes with PNMS (Lobel, 1994). A 36-item index
adapted from the Problem Oriented Perinatal Risk Assessment
System (Hobel, 1979) and shown to have excellent predictive
validity in prior studies (Lobel et al., 2000; Wadhwa et al., 1993)
was used to assign risk scores. It assesses six categories: unusual
conditions of pregnancy, gynecological and obstetric history, com-
plications of past pregnancies, family history, medical history, and
current pregnancy complaints. Items are dichotomously coded
(present/absent), then summed.
The primary birth outcome variables were continuously coded:
gestational age at delivery (weeks) and birth weight (g). Because
rates of clinically defined adverse birth outcomes (age 37 weeks
and weight 2,500 g) have low incidence, large samples are often
needed to assess the impact of predictors on these. Also, although
dichotomous variables may have more immediate clinical inter-
pretability, associations with childhood and adult outcomes exist
even across the normal range of birth weight and gestational age
(Breslau, Chilcoat, DelDotto, Andreski, & Brown, 1996; Matte,
Bresnahan, Begg, & Susser, 2001; Raikkonen et al., 2007; Rich-
ards, Hardy, Kuh, & Wadsworth, 2001; Sorensen et al., 1997).
Thus, we focused on continuous birth outcomes in this study, but
also examined the impact of PNMS on dichotomous birth out-
comes: preterm delivery (yes/no) and low birth weight (yes/no).
Statistical Analysis
AMOS 5.0 software for SEM (Arbuckle, 2003; Byrne, 2001)
was used to construct a single latent variable from the four mea-
sures of PNMS, to compare the ability of pregnancy-specific stress
to predict gestational age and birth weight relative to the composite
latent variable and to the individual measures of general stress, and
to examine whether health behaviors explain an association be-
tween PNMS and birth outcomes. Any maternal characteristic
significantly correlated with birth outcomes was to be included in
models predicting birth outcomes. SEM is ideal for use in this
study because of its ability to accommodate multiple, correlated
outcomes (i.e., gestational age and birth weight) and its facility
with latent variables. Results are comparable to analyzing several
regression equations simultaneously (Byrne, 2001).2
Results
Sample Description
Correlation coefficients, means, standard deviations, and inter-
nal consistency coefficients appear in Table 2. As compared with
prior samples of pregnant women studied by our research team,
participants in the present study had similar scores on the measures
of perceived stress and life event distress, slightly higher state
anxiety scores, and reported more prenatal life events (Lobel et al.,
1992; 2000). Perceived stress and state anxiety scores in this and
our previous studies of pregnant women are higher than population
norms reported for nonpregnant women (Cohen & Williamson,
1988; Spielberger, 1983, 1995). With respect to pregnancy-
specific stress, average item responses correspond to a rating
between not at all and somewhat. Approximately 22% of
women reported smoking sometimes, fairly often, or very
often. On average, they reported exercising almost never, and
sometimes consuming caffeine, eating unhealthily, and experi-
encing physical strain. Participants reported eating healthy meals
fairly often and taking prenatal vitamins between fairly and
very often. Finally, 60% of the sample had two or more obstetric
risk factors. The low internal consistency of the obstetric risk
measure confirms that items comprising this index are substan-
tially independent and that scores are not artifactually elevated.
No maternal characteristic other than obstetric risk was associ-
ated with birth outcomes, so only this variable was added to the
structural equation models. As expected, all of the stress variables
were significantly intercorrelated, underscoring the appropriate-
ness of modeling these as a latent factor. There were also signif-
icant intercorrelations among some of the health behaviors, such as
between vitamin use and healthy eating. Finally, as expected, the
two birth outcomes were highly intercorrelated, corroborating that
gestational age should be modeled as a predictor of birth weight.
Model Testing: PNMS, Gestational Age, and Birth Weight
The first model confirmed that the four measures of PNMS
(pregnancy-specific stress, perceived stress, state anxiety, and life
events) could be represented by a single, higher-order latent factor,
replicating prior studies establishing the coherence of a PNMS
factor based on these components. The component factors repre-
2
A variety of fit indices were examined. A nonsignificant chi-square
indicates no difference between model and data, but because this fit index
is sensitive to sample size and to violation of normality assumptions (the
value in both cases tends to be large), we also used the Comparative Fit
Index (CFI), with scores closer to 1.0 indicating good fit, and the Root
Mean Square Error of Approximation (RMSEA), for which values of less
than .10 are desired. Lagrange Multiplier test values were also calculated;
only conceptually consistent model improvements with a value above 4.0
were considered (Byrne, 2001). For model comparison, the Akaike Infor-
mation Criterion (AIC; Akaike, 1987) and the Bayesian Information Cri-
terion (BIC; Raftery, 1993; Schwarz, 1978) were used; smaller values
indicate better fit. These indices are affected by the number of parameters
in models and therefore are only appropriate when comparing models with
equivalent degree of complexity.
 PREGNANCY-SPECIFIC STRESS AND BIRTH OUTCOMES
senting pregnancy-specific stress, perceived stress, and state anx-
iety were comprised of the average scores on these measures at the
three prenatal time points; the life events latent factor consisted of
number of events and average life event distress. All path coeffi-
cients were statistically significant ( ps .05). This model of the
higher-order latent PNMS factor exhibited good fit, 2 70.83,
df 38, p .001; CFI .98; RMSEA .06 (.03.08). Based on
Lagrange Multiplier test values, we allowed correlations between
the residuals of perceived stress and state anxiety measured at the
first time point and at the third time point.
A model examining the impact of this latent PNMS factor on
birth outcomes was then examined. Results indicated that the path
from the latent stress factor to gestational age (B .13, p .04)
was significant, but the latent factor did not predict birth weight
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
(see Figure 1). Obstetric risk was also a predictor of gestational
age, which in turn predicted birth weight. The model explained
7.2% of variance in gestational age and 31% of variance in birth
weight.
To examine whether pregnancy-specific stress was an indepen-
dent predictor of gestational age, we removed it from the latent
PNMS factor and examined its effects separately, retaining its
correlation with the PNMS latent factor (now comprised of the
remaining three general stress variables). The resulting model
confirmed that pregnancy-specific stress is a significant, indepen-
dent predictor of gestational age (B .16, p .008). With
pregnancy-specific stress predicting gestational age, the latent
PNMS factor no longer predicted gestational age, so we removed
this path from the model. The model (see Figure 2) explained 8%
of variance in weeks gestation and 31.6% of variance in birth
weight. A nonsignificant 2 difference test comparing the models
with and without a path from the latent PNMS factor to gestational
PDQ1 .75
.87 Pregnancy-Specific
PDQ2
Stress
PDQ3 .84
PSS1 .74
.81 Perceived
PSS2
Stress
PSS3 .82
S ANX1 .67
.78
S ANX2 State Anxiety
S ANX3 .72
TOT LE .67
.50 Prenatal
LE DIS Life Events
Figure 1. Structural equation model examining the impact of the latent PNMS factor and obstetric risk on birth
weight and gestational age at delivery, 2 116.81, df 71, p .001; CFI .97; RMSEA .05 (.03.06).
All path coefficients displayed are statistically significant ( p .05). For simplicity, error values and correlations
among some of the measured stress indicators are not shown.
609
age indicated that the model was not improved by retaining the
path, diff
2
2.50, df 1, p .05. Furthermore, imposing an
equality constraint on the two path coefficients representing asso-
ciations with gestational age from pregnancy-specific stress and
from latent PNMS produced a significant 2 difference test, indi-
cating significant departure from a model with these coefficients
unconstrained, diff
2
4.69, df 1, p .05, and signifying that
these coefficients are not of equal magnitude. These tests confirm
the validity of removing the path from the latent PNMS factor to
gestational age.
We used the same method of model-testing for each of the
remaining components of the PNMS latent factor: removing the
component from the latent factor, allowing it to correlate with the
latent factor (comprised of the three remaining stress variables),
and examining whether the component itself predicted birth out-
comes. Each of these models was a good fit, but they indicated that
perceived stress and life events were neither significant predictors
of gestational age nor birth weight. However, state anxiety did
predict gestational age (B .13, p .04) when it was separated
from (but still correlated with) the latent PNMS factor. This model
explained 7.2% of the variance in weeks gestation and 31.6% of
variance in birth weight. We conducted a 2 difference test com-
paring this model with and without adding a path from the latent
PNMS factor to gestational age. Results indicated that the model
was not improved by including the path from latent PNMS. Im-
posing an equality constraint on the two path coefficients repre-
senting associations with gestational age from state anxiety and
from latent PNMS produced a nonsignificant 2 difference test,
indicating that the coefficients for these paths are not significantly
different. These tests suggest that the path from state anxiety to
gestational age is not a vital component of the model.
Obstetric
.89
Risk
-.23
.82
Prenatal Maternal -.13 Weeks
Stress Gestation
.89
.56
.72
Birthweight
 610 LOBEL ET AL.

PDQ1 .75
.88
.87 Pregnancy-Specific
PDQ2
Stress

PDQ3 .84

Obstetric
PSS1 .74 Risk
.81 Perceived -.16
PSS2
Stress -.23
.82
PSS3 .82
Prenatal Maternal Weeks
Stress Gestation
.67 .89
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

S ANX1
This document is copyrighted by the American Psychological Association or one of its allied publishers.

.78 .56
S ANX2 State Anxiety
.72
S ANX3 .72
Birthweight

TOT LE .67
.50 Prenatal
LE DIS Life Events

Figure 2. Structural equation model examining the impact of pregnancy-specific stress and obstetric risk on
birth weight and gestational age at delivery, including the correlation between latent PNMS and pregnancy-
specific stress, 2 114.03, df 71, p .001; CFI .97; RMSEA .05 (.03.06). All path coefficients
displayed are statistically significant ( p .05). For simplicity, error values and correlations among some of the
measured stress indicators are not shown.

AIC and BIC2 were calculated to compare the above models
examining pregnancy-specific stress and state anxiety as unique
predictors of birth outcome. AIC and BIC were 182.03 and 305.49,
respectively, for the model examining pregnancy-specific stress,
and 184.81 and 308.27, respectively, for the model examining state
anxiety, suggesting that the former is a slightly better-fitting
model.
Finally, we examined two pared-down models to examine the
independent predictive validity of state anxiety and pregnancy-
specific stress without other stress variables in the model. In the
first, the state anxiety latent factor was modeled as a predictor of
birth outcomes, without pregnancy-specific stress, perceived
stress, or life events in the model. This model fit well, 2 9.84,
df 9, p .36; CFI 1.0; RMSEA .02 (.00 .07). However,
state anxiety only marginally predicted weeks gestation, B
.12, p .06. The second model, examining pregnancy-specific
stress and birth outcomes without state anxiety, perceived stress, or
life events (see Figure 3), also fit well. Pregnancy-specific stress
was a significant predictor of gestational age in this model, B
.18, p .004. The model explained 9% of variance in gesta-
tional age and 32% of variance in birth weight.
The Role of Health Behaviors
The next set of structural equation models examined whether
pregnancy-specific stress was associated with prenatal health be-
haviors and if so, whether any of the latter predicted birth weight
or gestational age at delivery. Bivariate correlations (see Table 2)
indicated that six of the seven health behaviors (i.e., all but
physical strain) were associated with pregnancy-specific stress at
two or three time points. Therefore, these six health behavior
variables were included in the structural equation model. The
resulting model, shown in Figure 4, indicates that pregnancy-
specific stress significantly predicts cigarette smoking, caffeine
consumption, and unhealthy eating, and is inversely associated
with healthy eating, prenatal vitamin use, and exercise. Addition-
ally, cigarette smoking predicts lower birth weight. The direct
association between pregnancy-specific stress and gestational age
(B .18) remains significant in this model. The model fits well.
There are also significant correlations among some of the health
behaviors. All of these are in predictable directions (e.g., healthy
eating is associated with prenatal vitamin use). This model ac-
counts for 10% of the variance in gestational age at delivery and
33% of the variance in birth weight.
Predicting Dichotomous Birth Outcomes
We replaced the continuous birth outcome variables in the
preceding model with dichotomous outcomes (normal/low birth
weight and full-/preterm delivery). Obstetric risk did not signifi-
cantly predict full-/preterm delivery (B .11), so this path was
removed from the model. The resulting model fit well (see Figure
5). Comparing Figures 4 and 5 reveals that the magnitude of path
coefficients is essentially the same; pregnancy-specific stress is as
strongly associated with gestational age (B .18) as with
full-term delivery (B .18). However, as compared with the
model predicting continuous outcomes, this model does not ac-
count for as much variance in the timing of delivery (R2 .03 for
full-/preterm delivery vs. .10 for continuously measured gesta-
tional age). Both models account for the same portion (33%) of
 PREGNANCY-SPECIFIC STRESS AND BIRTH OUTCOMES 611

Obstetric
Risk

-.23

PDQ1 .76

.86 Pregnancy-Specific -.18 Weeks

PDQ2
Stress Gestation

PDQ3 .84
.56

Birthweight
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Figure 3. Structural equation model examining the impact of pregnancy-specific stress and obstetric risk on
birth weight and gestational age at delivery, 2 10.29, df 9, p .33; CFI 1.0; RMSEA .02 (.00 .07).

variance in birth weightwhether continuously measured or di- Discussion

chotomized as normal/low.
Finally, to quantify the impact of pregnancy-specific stress and
obstetric risk, we examined the gestational age at delivery of
infants born to women highest and lowest in pregnancy-specific
stress and highest and lowest in obstetric medical risk (splitting
these variables at their median). Women with low stress and low
risk delivered at M 276 days (SD 8.4), low stress/high risk
women delivered at M 271 days (SD 11.2), low risk/high
stress women delivered at M 272 days (SD 11.2), and women
with high obstetric risk and high stress delivered at M 267 days
(SD 12.6).
Results of this study indicate that pregnancy-specific stress is a
better predictor of birth outcome than state anxiety, perceived
stress, or life event stress, or than a higher-order latent factor
constructed from all four types of stress. Pregnancy-specific stress
was associated with earlier delivery and with clinically defined
preterm delivery in an ethnically and socioeconomically diverse
sample of women. Although the majority of women did not deliver
preterm, delivery was nine days earlier for those with high obstet-
ric risk and high pregnancy-specific stress than for women low in
risk and stress. Pregnancy-specific stress was also associated with

-.15
Caffeine Cigarettes Birthweight
.16

Unhealthy
Eating
.29
.22 Obstetric .55
Risk
Healthy
-.26
Eating -.14

Prenatal -.19 Pregnancy-Specific -.18 Gestational

Vitamins Stress Age

-.15 .76 .83

.86

Exercise
PDQ 1 PDQ 2 PDQ 3

Figure 4. Structural equation model examining the impact of pregnancy-specific stress, health behaviors, and
obstetric risk on birth weight and gestational age at delivery, 2 64.40, df 49, p .07; CFI .98;
RMSEA .03 (.00 .05). All path coefficients displayed are statistically significant ( p .05). For simplicity,
error values and correlations among some of the health behaviors are not shown.
 612
Caffeine
.16
Unhealthy
Eating
.29
Healthy
Eating -.14
Prenatal -.19
Vitamins
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
-.15 .77
Exercise
PDQ 1
Figure 5. Structural equation model examining the impact of pregnancy-specific stress and health behaviors on
normal birth weight and full-term delivery, 2 52.88, df 39, p .07; CFI .98; RMSEA .04 (.00 .06).
All path coefficients displayed are statistically significant ( p .05). For simplicity, error values and correlations
among some of the health behaviors are not shown.
a variety of prenatal health behaviors involving eating, exercise,
and cigarette smoking. Of these, cigarette smoking predicted lower
birth weight. These results suggest that pregnancy-specific stress
may be a direct contributor to early delivery, and through its
association with cigarette smoking, an indirect contributor to low
birth weight. Furthermore, associations of pregnancy-specific
stress with birth outcome exist after controlling for obstetric risk,
illustrating that the impact of womens psychological condition is
independent of their medical condition during pregnancy.
State anxiety, which has been used in many prior studies to
operationally define prenatal stress, predicted gestational age in
this study when its correlation with the latent stress factor was
modeled, but only to a small degree, and post hoc tests indicated
that this association was a nonsignificant contributor to the model.
Similarly, when it was the only stress variable in the model, state
anxiety only marginally predicted gestational age. In contrast,
pregnancy-specific stress was a significant predictor of gestational
age in several models, including the model in which its correlation
with other stress variables was controlled, and when it was mod-
eled as the sole stress variable predicting gestational age. These
findings suggest that the impact of pregnancy-specific stress on
birth outcome is stronger than and independent of state anxiety and
the other general types of stress measured in this study.
Why might pregnancy-specific stress be a better predictor of
birth outcome than general stress? Pregnancy-specific stress may
be a more potent type of stress. Other researchers have suggested,
for example, that pregnancy-specific stress triggers greater physi-
ological arousal than general stress (DiPietro et al., 2002, 2004;
Huizink et al., 2004). Unlike general stress, pregnancy-specific
stress involves stressors and responses that are context specific.
Therefore, pregnancy-specific stress may have more proximal
impact on the outcome of pregnancy. Measures of pregnancy-
specific stress may also be a more valid means of assessing what
LOBEL ET AL.
-.12
Cigarettes 2500+ grams
.22 .56
Pregnancy-Specific -.18
37+ Weeks
Stress
.83
.86
PDQ 2 PDQ 3
women are experiencing during pregnancy than measures which
capture general emotional state or conditions.
Results of this study offer some insight into the processes
through which pregnancy-specific stress may affect birth outcome.
The direct associations observed between pregnancy-specific
stress and gestational age are most likely attributable to vascular,
immunological, and neuroendocrine changes that have been im-
plicated by prior research. For example, prenatal stress has been
linked in several studies with higher levels of corticotropin-
releasing hormone (Austin & Leader, 2000; Hobel, Dunkel-
Schetter, Roesch, Castro, & Arora, 1999; Mancuso, Dunkel-
Schetter, Rini, Roesch, & Hobel, 2004; Wadhwa et al., 2004)
which triggers release of prostaglandins, facilitates the actions of
oxytocin, stimulates contractions of the myometrium during labor,
and through these mechanisms may contribute to earlier delivery
(Hobel et al., 1999). Associations between pregnancy-specific
stress and birth weight in this study may be better explained by
changes in health behaviors, namely cigarette smoking, which
affects birth weight by causing uterine blood vessel spasm and
reducing uterine blood flow, thereby retarding fetal growth (Scott,
2003). Highly stressed women may resort to smoking as a means
of coping. Study findings may underestimate the impact of smok-
ing, however, if participants underreported their smoking fre-
quency due to social desirability concerns. Although recent evi-
dence suggests that pregnant women accurately report the amount
they smoke (McDonald et al., 2005), self-report measures of
cigarette smoking and other health behaviors are not ideal. This
study provides a foundation for further research examining the
association of prenatal stress with objective indicators of health
behaviors such as serum cotinine and fitness tests.
The present study did not examine whether the impact of
pregnancy-specific stress and general stress differ according to
when they occur during pregnancy, but this question is worthy of
 PREGNANCY-SPECIFIC STRESS AND BIRTH OUTCOMES
further investigation, particularly given recent evidence suggesting
that responses to stress may change across pregnancy (Glynn,
Wadhwa, Dunkel-Schetter, Chicz-Demet, & Sandman, 2001;
Glynn, Dunkel-Schetter, Wadhwa, & Sandman, 2004). Another
caveat to the results of the present study is that women excluded
from analyses because they did not complete study measures were
more likely to be poor and to be non-White than women included,
raising the possibility that findings are not generalizable to other
groups of pregnant women. Furthermore, although a model was
successfully fit that implicated pregnancy-specific stress as a pre-
dictor of clinically defined adverse birth outcomes, the sample size
did not permit a strong test of the impact of PNMS on these
outcomes, which have low prevalence. Larger samples may be
needed to definitively assess the impact of pregnancy-specific
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
stress on preterm delivery and low birth weight. However, because
small differences in gestational age and birth weight, even within
normal ranges, have been shown to predict later outcomes such as
IQ, cognitive function, and adult depression (e.g., Behrman &
Butler, 2007; Matte et al., 2001; Raikkonen et al., 2007; Sorensen
et al., 1997), the stress effects on continuously measured birth
outcomes observed in this study are probably not inconsequential.
Study results underscore the value of further research to deter-
mine whether a brief measure of pregnancy-specific stress such as
the PDQ can be used in clinical settings to successfully identify
and intervene with pregnant women at greatest risk of delivering
early or of engaging in behaviors that reduce birth weight. Inter-
ventions might focus on educating such women about pregnancy,
providing support, and enhancing self-care (e.g., Ickovics et al.,
2007). Educating pregnant women about the symptoms and
changes that they can expect, and about other aspects of preg-
nancy, birth, and parenting is likely to decrease pregnancy-specific
stress. We found that knowledge was associated with lower stress
in another study of pregnant women (Lobel, Dias, & Meyer, 2005).
Results of the present research suggest that alleviating pregnancy-
specific stress may reduce womens risk of preterm delivery and
the chances that they will smoke, thereby decreasing their risk of
low birth weight. Coupled with the results of prior studies, these
findings demonstrate that the opportunity to protect childrens
health likely begins in utero by improving the psychosocial well-
being of their mothers.
References
Affonso, D. D., Liu-Chiang, C. Y., & Mayberry, L. J. (1999). Worry:
Conceptual dimensions and relevance to childbearing women. Health
Care Women International, 20, 227236.
Akaike, H. (1987). Factor analysis and AIC. Psychometrika, 52, 317332.
Arbuckle, J. L. (2003). AMOS (Version 5.0): Small Waters Corp.
Austin, M. P., & Leader, L. (2000). Maternal stress and obstetric and infant
outcomes: Epidemiological findings and neuroendocrine mechanisms.
Australian and New Zealand Journal of Obstetrics and Gynaecology,
40, 331337.
Barker, D. J. P. (2007). The origins of the developmental origins theory.
Journal of Internal Medicine, 261, 412 417.
Barnet, B., Duggan, A. K., Wilson, M. D., & Joffe, A. (1995). Association
between postpartum substance use and depressive symptoms, stress, and
social support in adolescent mothers. Pediatrics, 96, 659 666.
Behrman, R. E., & Butler, A. S. (Eds.). (2007). Preterm birth: Causes,
consequences, and prevention. Washington, DC: National Academies
Press.
613
Borrelli, B., Bock, B., King, T., Pinto, B., & Marcus, B. H. (1996). The
impact of depression on smoking cessation in women. American Journal
of Preventative Medicine, 12, 378 387.
Breckler, S. J. (1990). Applications of covariance structure modeling in
psychology: Cause for concern? Psychological Bulletin, 52, 260 271.
Breslau, N., Chilcoat, H., DelDotto, J., Andreski, P., & Brown, G. (1996).
Low birth weight and neurocognitive status at six years of age. Biolog-
ical Psychiatry, 40, 389 397.
Bresnahan, K., Zuckerman, B., & Cabral, H. (1992). Psychosocial corre-
lates of drug and heavy alcohol use among pregnant women at risk for
drug use. Obstetrics and Gynecology, 80, 976 980.
Byrne, B. M. (2001). Structural equation modeling with AMOS. Mahwah,
NJ: Erlbaum.
Chomitz, V. R., Cheung, L. W., & Lieberman, E. (1995). The role of
lifestyle in preventing low birth weight. Future of Children, 5, 121138.
Cohen, S., & Williamson, G. M. (1988). Perceived stress in a probability
sample of the United States. In S. Spacapan & S. Oskamp (Eds.), The
social psychology of health (pp. 31 67). Newbury Park, CA: Sage.
Copper, R. L., Goldenberg, R. L., Das, A., Elder, N., Swain, M., Norman,
G., et al. (1996). The preterm prediction study: Maternal stress is
associated with spontaneous preterm birth at less than thirty-five weeks
gestation. National Institute of Child Health and Human Development
Maternal-Fetal Medicine Units Network. American Journal of Obstet-
rics and Gynecology, 175, 1286 1292.
Davis, E. P., Glynn, L. M., Dunkel-Schetter, C., Hobel, C., Chicz-Demet,
A., & Sandman, C. A. (2005). Corticotropin-releasing hormone during
pregnancy is associated with infant temperament. Developmental Neu-
roscience, 27, 299 305.
DeLuca, R. S., & Lobel, M. (1995). Conception, commitment, and health
behavior practices in medically high-risk pregnant women. Womens
Health: Research on Gender, Behavior, and Policy, 1, 257271.
DiPietro, J. A., Ghera, M. M., Costigan, K., & Hawkins, M. (2004).
Measuring the ups and downs of pregnancy stress. Journal of Psycho-
somatic Obstetrics & Gynecology, 25, 189 201.
DiPietro, J. A., Hilton, S. C., Hawkins, M., Costigan, K. A., & Pressman,
E. K. (2002). Maternal stress and affect influence fetal neurobehavioral
development. Developmental Psychology, 38, 659 668.
Dole, N., Savitz, D. A., Hertz-Picciotto, I., Siega-Riz, A. M., McMahon,
M. J., & Buekens, P. (2003). Maternal stress and preterm birth. Amer-
ican Journal of Epidemiology, 157, 14 24.
Dunkel-Schetter, C. (1998). Maternal stress and preterm delivery. Prenatal
and Neonatal Medicine, 3, 39 42.
Dunkel-Schetter, C., Gurung, R. A. R., Lobel, M., & Wadhwa, P. D.
(2001). Stress processes in pregnancy and birth: Psychological, biolog-
ical, and sociocultural influences. In A. Baum, T. A. Revenson, & J. E.
Singer (Eds.), Handbook of health psychology (pp. 495518). Hillsdale,
NJ: Erlbaum.
Glynn, L. M., Dunkel-Schetter, C., Wadhwa, P. D., & Sandman, C. A.
(2004). Pregnancy affects appraisal of negative life events. Journal of
Psychosomatic Research, 56, 4752.
Glynn, L. M., Wadhwa, P. D., Dunkel-Schetter, C., Chicz-Demet, A., &
Sandman, C. A. (2001). When stress happens matters: Effects of earth-
quake timing on stress responsivity in pregnancy. American Journal of
Obstetrics and Gynecology, 184, 637 642.
Goldenberg, R. L., DuBard, M. B., Cliver, S. P., Nelson, K. G., Blankson,
K., Ramey, S. L., et al. (1996). Pregnancy outcome and intelligence at
age five years. American Journal of Obstetrics and Gynecology, 175,
15111515.
Gurung, R. A. R., Dunkel-Schetter, C., Collins, N., Rini, C., & Hobel, C. J.
(2005). Psychosocial predictors of prenatal anxiety. Journal of Social
and Clinical Psychology, 24, 497519.
Hack, M., & Merkatz, I. R. (1995). Preterm delivery and low birth weighta
dire legacy. New England Journal of Medicine, 333, 17721774.
Halman, L. J., Oakley, D., & Lederman, R. (1995). Adaptation to preg-
 614 LOBEL ET AL.
nancy and motherhood among subfecund and fecund primiparous
women. Maternal-Child Nursing Journal, 23, 90 100.
Hamilton, J. G., & Lobel, M. (2008). Types, patterns, and predictors of
coping with stress during pregnancy: Examination of the Revised Pre-
natal Coping Inventory in a diverse sample. Journal of Psychosomatic
Obstetrics & Gynecology, 29, 97104.
Hedegaard, M., Henriksen, T. B., Secher, N. J., Hatch, M. C., & Sabroe, S.
(1996). Do stressful life events affect duration of gestation and risk of
preterm delivery? Epidemiology, 7, 339 345.
Hobel, C. J. (1979). Assessment of the high risk fetus. Clinics in Obstetrics
and Gynaecology, 6, 367377.
Hobel, C. J., Dunkel-Schetter, C., Roesch, S. C., Castro, L. C., & Arora,
C. P. (1999). Maternal plasma corticotropin-releasing hormone associ-
ated with stress at 20 weeks gestation in pregnancies ending in preterm
delivery. American Journal of Obstetrics and Gynecology, 180, S257
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
S263.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Hoffman, S., & Hatch, M. C. (1996). Stress, social support and pregnancy
outcome: A reassessment based on recent research. Paediatric and
Perinatal Epidemiology, 10, 380 405.
Hogue, C. J. R., Hoffman, S., & Hatch, M. C. (2001). Stress and preterm
delivery: A conceptual framework. Paediatric and Perinatal Epidemi-
ology, 15. 30 40.
Huizink, A. C., Mulder, E. J., Robles de Medina, P. G., Visser, G. H., &
Buitelaar, J. K. (2004). Is pregnancy anxiety a distinctive syndrome?
Early Human Development, 79, 8191.
Hutchins, E., & DiPietro, J. (1997). Psychosocial risk factors associated
with cocaine use during pregnancy: A case-control study. Obstetrics and
Gynecology, 90, 142147.
Ickovics, J. R., Kershaw, T. S., Westdahl, C., Magriples, U., Massey, Z.,
Reynolds, H., et al. (2007). Group prenatal care and perinatal outcomes: A
randomized controlled trial. Obstetrics and Gynecology, 110, 330 339.
Levin, J. S. (1991). The factor structure of the pregnancy anxiety scale.
Journal of Health and Social Behavior, 32, 368 381.
Lindseth, G., & Vari, P. (2005). Measuring physical activity during preg-
nancy. Western Journal of Nursing Research, 27, 722734.
Lobel, M. (1994). Conceptualizations, measurement, and effects of prena-
tal maternal stress on birth outcomes. Journal of Behavioral Medicine,
17, 225272.
Lobel, M. (1998). Pregnancy and mental health. In H. Friedman (Ed.),
Encyclopedia of mental health (pp. 229 238). San Diego, CA: Aca-
demic Press.
Lobel, M., DeVincent, C. J., Kaminer, A., & Meyer, B. A. (2000). The
impact of prenatal maternal stress and optimistic disposition on birth
outcomes in medically high-risk women. Health Psychology, 19, 544
553.
Lobel, M., Dias, L., & Meyer, B. A. (2005). Distress associated with
prenatal screening for fetal abnormality. Journal of Behavioral Medi-
cine, 28, 6576.
Lobel, M., & Dunkel-Schetter, C. (1990). Conceptualizing stress to study
effects on health: Environmental, perceptual, and emotional compo-
nents. Anxiety Research, 3, 213230.
Lobel, M., Dunkel-Schetter, C., & Scrimshaw, S. C. (1992). Prenatal
maternal stress and prematurity: A prospective study of socioeconomi-
cally disadvantaged women. Health Psychology, 11, 32 40.
Mancuso, R. A., Dunkel-Schetter, C., Rini, C. M., Roesch, S. C., & Hobel,
C. J. (2004). Maternal prenatal anxiety and corticotropin-releasing hor-
mone associated with timing of delivery. Psychosomatic Medicine, 66,
762769.
Mathews, T. J., & MacDorman, M. F. (2006). Infant mortality statistics
from the 2003 period linked birth/infant death data set. National Vital
Statistics Reports, 54, 132.
Matte, T. D., Bresnahan, M., Begg, M. D., & Susser, E. (2001). Influence
of variation in birth weight within normal range and within sibships on
IQ at age 7 years: Cohort study. British Medical Journal, 323, 310 314.
McCormick, M. C., Brooks-Gunn, J., Shorter, T., Holmes, J. H., Wallace,
C. Y., & Heagarty, M. C. (1990). Factors associated with smoking in
low-income pregnant women: Relationship to birth weight, stressful life
events, social support, health behaviors and mental distress. Journal of
Clinical Epidemiology, 43, 441 448.
McDonald, S. D., Perkins, S. L., & Walker, M. C. (2005). Correlation
between self-reported smoking status and serum cotinine during preg-
nancy. Addictive Behaviors, 30, 853 857.
McKnight, P. E., McKnight, K. M., Sidani, S., & Figueredo, A. J. (2007).
Missing data: A gentle introduction. New York: Guilford Press.
Misra, D. P., OCampo, P., & Strobino, D. (2001). Testing a sociomedical
model for preterm delivery. Paediatric and Perinatal Epidemiology, 15,
110 122.
Molfese, V. J., Bricker, M. C., Manion, L. G., Beadnell, B., Yaple, K., &
Moires, K. A. (1987). Anxiety, depression and stress in pregnancy: A
multivariate model of intra-partum risks and pregnancy outcomes. Jour-
nal of Psychosomatic Obstetrics & Gynecology, 7, 7792.
Newnham, J. (1998). Consequences of fetal growth restriction. Current
Opinion in Obstetrics and Gynecology, 10, 145149.
Ng, D. M., & Jeffery, R. W. (2003). Relationships between perceived stress
and health behaviors in a sample of working adults. Health Psychology,
22, 638 642.
Norbeck, J. S., & Tilden, V. P. (1983). Life stress, social support and
emotional disequilibrium in complications of pregnancy: A prospective,
multivariate study. Journal of Health and Social Behavior, 24, 30 46.
OConnor, T. G., Heron, J., & Glover, V. (2002). Antenatal anxiety
predicts child behavioral/emotional problems independently of postnatal
depression. Journal of the American Academy of Child and Adolescent
Psychiatry, 41, 1470 1477.
Okah, F. A., Cai, J., Dew, P. C., & Hoff, G. L. (2005). Are fewer women
smoking during pregnancy? American Journal of Health Behavior, 29,
456 461.
Pritchard, C. W. (1994). Depression and smoking in pregnancy in Scotland.
Journal of Epidemiology and Community Health, 48, 377382.
Raftery, A. E. (1993). Bayesian model selection in structural equation
models. In K. A. Bollen & J. S. Long (Eds.), Testing structural equation
models (pp. 163180). Newbury Park, CA: Sage.
Raikkonen, K., Pesonen, A-K., Kajantie, E., Heinonen, K., Forsen, T.,
Phillips, D. I. W., et al. (2007). Length of gestation and depressive
symptoms at age 60 years. British Journal of Psychiatry, 190, 469 474.
Richards, M., Hardy, R., Kuh, D., & Wadsworth, M. (2001). Birthweight
and cognitive function in the British 1946 birth cohort: Longitudinal
population-based study. British Medical Journal, 322, 199 203.
Rini, C. K., Dunkel-Schetter, C., Wadhwa, P. D., & Sandman, C. A.
(1999). Psychological adaptation and birth outcomes: The role of per-
sonal resources, stress, and sociocultural context in pregnancy. Health
Psychology, 18, 333345.
Rodriguez, A., Bohlin, G., & Lindmark, G. (2000). Psychosocial predictors
of smoking and exercise during pregnancy. Journal of Reproductive and
Infant Psychology, 18, 203223.
Roesch, S. C., Dunkel-Schetter, C., Woo, G., & Hobel, C. J. (2004).
Modeling the types and timing of stress in pregnancy. Anxiety, Stress &
Coping: An International Journal, 17, 87102.
Schwarz, G. (1978). Estimating the dimensions of a model. Annals of
Statistics, 6, 461 464.
Scott, J. R. (Ed.). (2003). Danforths obstetrics and gynecology (9 Ed.).
Philadelphia PA: Lippincott Williams & Wilkins.
Sorensen, H. T., Sabroe, S., Olsen, J., Rothman, K. J., Gillman, M. W., &
Fischer, P. (1997). Birth weight and cognitive function in young adult
life: Historical cohort study. British Medical Journal, 315, 401 403.
Spielberger, C. D. (1983). Manual for the State-Trait Anxiety Inventory.
Palo Alto, CA: Consulting Psychologists Press.
Spielberger, C. D. (1995). Preliminary manual for the State-Trait Person-
 PREGNANCY-SPECIFIC STRESS AND BIRTH OUTCOMES 615

ality Inventory (STPI). Tampa, FL: Center for Research in Behavioral
Medicine and Health Psychology, University of South Florida.
Stanton, A. L., Lobel, M., Sears, S., & DeLuca, R. S. (2002). Psychosocial aspects
of selected issues in womens reproductive health: Current status and future
directions. Journal of Consulting and Clinical Psychology, 70, 751770.
Stetson, B. A., Rahn, J. M., Dubbert, P. M., Wilner, B. I., & Mercury,
M. G. (1997). Prospective evaluation of the effects of stress on
exercise adherence in community-residing women. Health Psychol-
ogy, 16, 515520.
Timperio, A., Salmon, J., & Crawford, D. (2003). Validity and reliability of a
physical activity recall instrument among overweight and non-overweight men
and women. Journal of Science and Medicine in Sport, 6, 477491.
Verbeke, W., & De Bourdeaudhuij, I. (2007). Dietary behaviour of preg-
nant versus non-pregnant women. Appetite, 48, 78 86.
Wadhwa, P. D., Garite, T. J., Porto, M., Glynn, L., Chicz-DeMet, A.,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
prospective investigation. American Journal of Obstetrics and Gynecol-
ogy, 191, 10631069.
Wadhwa, P. D., Sandman, C. A., Porto, M., Dunkel-Schetter, C., &
Garite, T. J. (1993). The association between prenatal stress and
infant birth weight and gestational age at birth: A prospective inves-
tigation. American Journal of Obstetrics and Gynecology, 169, 858
865.
Yali, A. M., & Lobel, M. (1999). Coping and distress in pregnancy: An
investigation of medically high risk women. Journal of Psychosomatic
Obstetrics and Gynaecology, 20, 39 52.
Yali, A. M., & Lobel, M. (2002). Stress resistance resources and coping in
pregnancy. Anxiety, Stress & Coping: An International Journal, 15,
289 309.
Zuckerman, B., Amaro, H., Bauchner, H., & Cabral, H. (1989). De-
pressive symptoms during pregnancy: Relationship to poor health

Dunkel-Schetter, C., et al. (2004). Placental corticotropin-releasing hor- behaviors. American Journal of Obstetrics and Gynecology, 160,
This document is copyrighted by the American Psychological Association or one of its allied publishers.

mone (CRH), spontaneous preterm birth, and fetal growth restriction: A 11071111.

Appendix

Revised Prenatal Distress Questionnaire Items

Are you feeling bothered, upset, or worried at this point in your pregnancy:

Repeated at each timepoint:

. . .about the effect of ongoing health problems such as high blood pressure or diabetes on your pregnancy?
. . .about feeling tired and having low energy during your pregnancy?
. . .about paying for your medical care during pregnancy?
. . .about changes in your weight and body shape during pregnancy?
. . .about whether you might have an unhealthy baby?
. . .about physical symptoms of pregnancy such as vomiting, swollen feet, or backaches?
. . .about the quality of your medical care during pregnancy?
. . .about working or caring for your family during your pregnancy?
. . .about whether the baby might be affected by alcohol, cigarettes, or drugs that you have taken?
Added at second and third timepoint:
. . .about whether the baby might come too early?
. . .about changes in your relationships with other people due to having a baby?
. . .about paying for the babys clothes, food, or medical care?
Added at third timepoint:
. . .about taking care of a newborn baby?
. . .about pain during labor and delivery?
. . .about what will happen during labor and delivery?
. . .about working at a job after the baby comes?
. . .about getting day care, babysitters, or other help to watch the baby after it comes?

Note. Response scale is Not at all (0), Somewhat (1), or Very much (2).