CAOG Papers www. AJOG.

org

The pattern of depression screening results

across successive pregnancies
Laura M. La Porte, MA; J. Jo Kim, PhD; Marci Adams, MPH; Hongyan Du, MS; Richard K. Silver, MD

OBJECTIVE: Determine whether depression screen results are consis-
tent across successive pregnancies.
STUDY DESIGN: The Edinburgh Postnatal Depression Scale was ad-
ministered in 2 successive pregnancies to 2116 women. A woman was
“screen-positive” if she scored ⱖ12 at 24-28 weeks’ or 6-weeks’
postpartum. Screen-positive women were assessed by telephone and
triaged by mental health professionals.
RESULTS: Most women (87.9%) were screen-negative in both preg-
nancies; 1.7% screened successively positive, 5.9% screened pos-
itive in only the first pregnancy; 4.5% screened positive in only the
second pregnancy. Unpartnered, nonwhite, and publicly insured
women were each likelier to screen positive in either or both preg-
nancies (P ⬍ .0001). Gestational age at delivery was significantly
greater in women who never screened positive (P ⬍ .05). A majority
(63%) of screen-positive women in both pregnancies reported no
history of mood disorder.
CONCLUSION: There is sufficient variability in depression screening re-
sults between successive gestations to warrant screening during each
pregnancy.
Key words: Edinburgh Postnatal Depression Scale, perinatal
depression, screening

Cite this article as: La Porte LM, Kim JJ, Adams M, et al. The pattern of depression screening results across successive pregnancies. Am J Obstet Gynecol
2012;206:261.e1-4.

D epression is considered a common

complication of pregnancy.1,2 Its
prevalence, coupled with the availability
postpartum timeframes.8 Other limita-
tions in this literature include underrep-
resentation of minority subjects and
ing screening patterns across 2 successive
gestations in the same patients.

of easily administered and validated
screening tools has led to support for
universal screening among obstetri-
cians3,4 and legislatures.5-7 However,
routine screening and systematic triage
are not yet commonplace. As a result,
most studies of depression risk have ex-
amined unselected groups of individual
pregnancies and their corresponding
From the Departments of Obstetrics and
Gynecology (Ms La Porte, Drs Kim and
Silver, and Ms Adams) and the Center for
Clinical and Research Informatics (Ms Du),
NorthShore University HealthSystem,
Evanston, and the University of Chicago
Pritzker School of Medicine (Drs Kim and
Silver), Chicago, IL.
Received Sept. 1, 2011; accepted Dec. 11,
2011.
The authors report no conflicts of interest.
Presented in part at the Central Association of
Obstetricians and Gynecologists 78th annual
meeting, Nassau, Bahamas, Oct. 26-29, 2011.
Reprints: Richard K. Silver, MD, 2650 Ridge
Ave., Suite 1507, Evanston, IL 60201.
rsilver@northshore.org
0002-9378/$36.00
© 2012 Mosby, Inc. All rights reserved.
doi: 10.1016/j.ajog.2011.12.005
those individuals who are uninsured or
Medicaid enrollees. A systematic review
by the Agency for Healthcare Research
and Quality1 called for studies to parse
out screening differences using these de-
mographic variables to identify relevant
cohorts and their unique attributes.
Depression screens analyzed in the
current study were administered in the
context of a comprehensive program for
universal perinatal depression screening
and mental health referral. Since 2003,
the program has included centralized
processing of all screening, a 24/7 crisis
hotline, a network of community-based
mental health professionals and an edu-
cational program for obstetricians and
nurse midwives.9 In a prior study10 re-
porting on 1584 patients screened once
during pregnancy and again at 6-weeks’
postpartum, we observed that unique
screen-positive cohorts were identified
before and after delivery, suggesting that
screening status was not necessarily
static over time in a single pregnancy. A
related question is whether depressive
risk is sustained between pregnancies
and our program data provided the op-
portunity to address this issue by study-
M ATERIALS AND M ETHODS
The Edinburgh Postnatal Depression
Scale (EPDS) was chosen for its validity
in both the antepartum and postpartum
timeframes.11 The EDPS is a self-report
instrument that does not provide a clin-
ical diagnosis of a perinatal mood disor-
der but does identify at-risk patients. A
validated version of the EPDS was ad-
ministered to patients at 24-28 weeks’
gestation and 6-weeks’ postpartum.
These screening times were based on the
potential onset of perinatal mood disor-
ders8,12 and also to coincide with routine
clinical activities (ie, glucose tolerance
testing and the postpartum check-up).
Completed screens were e-faxed from
the outpatient office to a confidential
email account that was monitored daily.
Scored screens were inputted into the
electronic medical record and linked to
the corresponding demographic and
perinatal data. EPDS screens were desig-
nated “positive” for scores ⱖ12 or for a
response other than “never” to the ques-
tion describing thoughts of harming
oneself based on recommendations for
use of the EPDS.13-15 Screen-positive pa-
tients received a telephone call from a
mental health professional and were re-

MARCH 2012 American Journal of Obstetrics & Gynecology 261.e1

CAOG Papers
FIGURE
Distribution of antepartum and postpartum Edinburgh Postnatal
Depression Scale scores
Distribution of EPDS screen scores of women who were screened at 24-28 weeks of gestation and
again at 6-weeks postpartum in 2 pregnancies (total of 5723 screens).
La Porte. Depression screening results across successive pregnancies. Am J Obstet Gynecol 2012.
ferred to mental health resources as indi-
cated.14 An interpretative language tele-
phone line allowed for evaluation in each
woman’s primary language and women
were reminded about the availability of
the crisis hotline. Institutional review
board (IRB) approval was obtained for
this study. We followed our IRB-ap-
proved protocol that specified present-
ing all screened women the opportunity
to give written permission to be ap-
proached about research participation at
a future date. In this study, we did not
rely on additional patient contact and
the data were deidentified before analy-
sis. Therefore, the IRB exempted us from
further consent requirements for this
protocol.
Women who were screened at least
once in 2 successive pregnancies (yield-
ing a total of 2, 3, or 4 screens per subject)
formed the basis for this study. A woman
was defined as “positive” within each
pregnancy if she screened positive at
24-28 weeks’ gestation and/or 6-weeks’
postpartum. Women were further cate-
gorized into concordant cases across 2
pregnancies (negative/negative or posi-
tive/positive) and discordant cases (neg-
261.e2 American Journal of Obstetrics & Gynecology MARCH 2012
ative/positive or positive/negative). Cor-
responding demographic, perinatal, and
mental health variables were analyzed to
determine factors that might be associ-
ated with the serial screening results. Sta-
tistical analysis was conducted using SAS
9.2 (SAS Institute, Cary, NC). A ␹2 was
used to assess association between cate-
gorical covariates and the 4 screening
phenotypes. Post hoc pairwise compari-
sons were Bonferroni corrected. Non-
parametric methods (Wilcoxon rank
sum test and Kruskal-Wallis test) were
used to assess between-group difference
with respect to continuous covariates,
and post hoc pairwise comparisons were
adjusted via permutation test.
R ESULTS
We screened 2116 women in 2 successive
pregnancies between 2003 and 2010. The
Table shows the demographic charac-
teristics of this sample. At the time of
the first screen, the mean age was 31.1
years, ⫾ 4.6. Nearly half (48.3%) of the
sample were primigravidas, and the re-
maining were multigravida (2, 3, and ⱖ4
in 27.4%, 13.3%, and 11%, respectively).
www.AJOG.org
Average gestational age at the time of an-
tepartum screening was 29.0 weeks and
6.6 weeks for postpartum screening. A
total of 949, 843, and 324 women had 2,
3, and 4 screens available for analysis for
a total of 5723 screens studied among
4232 individual pregnancies. Of women
who were positive in 1 or both pregnan-
cies, 12.1% were positive using our ad-
justed definition of “positive” as scoring
positive at any time in a pregnancy. The
frequency distribution of EPDS scores is
similar to that observed in prior publica-
tions from our center (Figure).9,10
Most women in our sample (89.6%)
screened concordantly across pregnan-
cies. A majority of cases (87.9%) were
negative/negative; 1.7% screened posi-
tive in both pregnancies. Of discordant
cases, 5.9% screened positive in the first
pregnancy (positive/negative) and 4.5%
screened positive in the second preg-
nancy (negative/positive).
In comparing demographic character-
istics of 4 cohorts, maternal age was not
different between groups. We observed
significant differences in successive
screening results by marital status, race
distribution, and insurance category. Of
women with known marital status (n ⫽
1702), partnered women were more
likely to screen negative in both pregnan-
cies (89.3% vs 74.8%, P ⬍ .0001), were
less likely to screen positive in the first
pregnancy (5.3% vs 12.2%, P ⬍ .05), and
less likely to screen positive in both preg-
nancies (1.2% vs 6.1%, P ⬍ .05).
Of women whose race was known
(n ⫽ 1669), white women were signifi-
cantly more likely to screen negative in
both pregnancies when compared with
an aggregate of African American, Asian,
American Indian, and Alaska Native
women (91.03% vs 82.14%, adjusted
P ⬍ .0001). White women were also less
likely to screen positive in the second
pregnancy (3.3% vs 6.4%, P ⬍ .05).
Of women whose insurance status was
known (n ⫽ 1962), privately insured
women were more likely to screen nega-
tive in both pregnancies (89.4% vs
74.3%, P ⬍ .0001). Privately insured
women were less likely to screen positive
in the first pregnancy (5.14% vs 14.16%,
P ⬍ .05) and less likely to screen positive
www.AJOG.org CAOG Papers

relying on prior screening results to as-

TABLE sess risk for perinatal mood symptoms,
Characteristics of the sample population (n ⴝ 2116a) these data support repeat depression
at the time of their first depression screen screening in each subsequent pregnancy.
Characteristic Frequency Percent Our data also suggest that demo-
Marital status graphic variables correlate with positive
.....................................................................................................................................................................................................................................
screens, whereas most perinatal variables
Married 1555 91.4
..................................................................................................................................................................................................................................... do not differentiate between screening
Single 144 8.5 status over time. The positive screening
.....................................................................................................................................................................................................................................
Divorced 2 0.1
.....................................................................................................................................................................................................................................
propensity across pregnancies of unpart-
Widowed 1 0.1 nered, nonwhite, and publicly insured
.....................................................................................................................................................................................................................................
women, regardless of their age, supports
Total 1702 100.0
.............................................................................................................................................................................................................................................. existing literature on the relatively
Race higher prevalence of perinatal mood dis-
.....................................................................................................................................................................................................................................
White 1182 70.8
.....................................................................................................................................................................................................................................
orders among women with these at-risk
Other 325 19.5 demographics.16,17 For obstetric offices
.....................................................................................................................................................................................................................................
that do not systematically screen pa-
Asian 91 5.5
..................................................................................................................................................................................................................................... tients, demographic variables may be
African American 67 4.0 used as a starting point in identifying
.....................................................................................................................................................................................................................................
American Indian or Alaska Native 4 0.2 women with potentially heightened risk.
.....................................................................................................................................................................................................................................
Total 1669 100.0 We could not determine whether a
..............................................................................................................................................................................................................................................
change in insurance or marital status be-
Insurance
..................................................................................................................................................................................................................................... tween pregnancies influenced sequential
Private 1849 94.2 screening results because too few women
.....................................................................................................................................................................................................................................
Public 113 5.8 experienced such changes.
.....................................................................................................................................................................................................................................
Total 1962 100.0 Women who screened positive in 2
..............................................................................................................................................................................................................................................
a
pregnancies also had high rates of self-
Demographic categories contained missing data not included in descriptive statistics.
La Porte. Depression screening across successive pregnancies. Am J Obstet Gynecol 2012.
reported mood disorders, but most dis-
orders had not been diagnosed at the
time of the first screen assessment. These
in both pregnancies (1.4% vs 5.3%, The telephonic mental health assess- data support prior research suggesting
P ⬍ .05). ments of positive/positive women con- that pregnancy may uncover an underly-
We also compared the 4 cohorts on ducted by the program’s licensed mental ing depressive propensity, not unlike
several perinatal variables specific to health professionals were retrieved and gestational diabetes is a risk factor for
each pregnancy. Most of these variables analyzed. Seventy-one mental health as- adult onset disease thereafter.18 We rec-
(delivery route, perineal laceration, in- sessments were conducted in this cohort ognize that the accuracy of screening in
duction of labor, prior cesarean, ces- over the 2 pregnancies. It was deter- the second pregnancy may have been in-
arean indication, delivery anesthesia, mined that 63% of these patients re- fluenced by the experience or the results
and neonatal intensive care admission) ported no history of a mood disorder or of initial pregnancy screening, but the
prior treatment at the time of the first average interval of 2.5 years between
were not significantly different between
assessment, but did report a mood disor- pregnancies should effectively dampen
groups for either pregnancy. Initial anal-
der diagnosis during the assessment that influence. Although a history of de-
ysis demonstrated that gestational age
completed in the second pregnancy. pression19 and previous perinatal de-
and birthweight varied between groups.
pression20 are considered strong predic-
In both pregnancies, the negative/nega- tors of future perinatal depression, our
tive subset delivered at a more advanced C OMMENT data suggest that women without a re-
gestational age and had higher birth- A major strength of this study is the op- ported mental health history can still be
weight infants. However, a post hoc, portunity to analyze successive pregnan- at risk.
pairwise comparison revealed that only cies in the same system using the same One potential limitation of this study
gestational age at delivery in the second screening methodology and the same is the classification of depressive risk in
pregnancy was significantly greater in triage algorithm in more than 2000 pregnancy. Our definition of a “positive”
the negative/negative and the positive/ women. These data showed that unique woman allowed for her to screen nega-
negative groups compared with the cohorts of women screened positive tive at 1 of the 2 screening time points. As
other cohorts with positive second pre- in different pregnancies. Furthermore, a result, the concordant and discordant
gnancy screens (Bonferroni adjusted most positive women were positive in cohorts in successive pregnancies may
P ⬍ .05). only 1 pregnancy, not both. Instead of not be homogeneous. We contemplated

MARCH 2012 American Journal of Obstetrics & Gynecology 261.e3

CAOG Papers
restricting the focus of the study only to
women who screened persistently posi-
tive over the course of a pregnancy epi-
sode (antepartum and postpartum), but
too few women had 4 screens (2 per
pregnancy) to form the basis of this
study. We believe that whether depres-
sive symptoms have their onset during
pregnancy or after delivery, the clinical
implications are likely similar.
Our study also lacked verifiable base-
line mental health history in the nega-
tive/negative screening cohort because
we did not routinely contact negatively
screening women. This limited the op-
portunity to compare the likelihood of a
subsequent mood disorder diagnosis
with the rate observed in the positive-
screening cohort.
Finally, it is the case that our tele-
phonic mental health assessment and tri-
age may influence the outcome of the
subsequent pregnancy screen; reflecting
the fact that our subjects are studied in a
real clinical environment such that with-
holding triage would be inappropriate.
This confounder might influence the
women who screen positive in the first
pregnancy and negative subsequently,
but the additional subset that screened
persistently positive underscores that
any enduring effect of our intervention is
not uniform across all at-risk women. f
261.e4 American Journal of Obstetrics & Gynecology MARCH 2012
REFERENCES
1. Gaynes BN, Gavin N, Meltzer-Brody S, et al.
Perinatal depression: prevalence, screening ac-
curacy, and screening outcomes. Rockville,
MD: Agency for Healthcare Research and Qual-
ity; 2005.
2. O’Hara MW, Swain AM. Rates and risk of
postpartum depression—a meta-analysis. Int
Rev Psychiatry 1996;8:37-54.
3. Dietrich AJ, Williams JW, Ciotti MC, et al. De-
pression care attitudes and practices of newer
obstetrician-gynecologists: a national survey.
Am J Obstet Gynecol 2003;189:267-73.
4. LaRocco-Cockburn A, Melville J, Bell M, Ka-
ton W. Depression screening attitudes and
practices among obstetrician-gynecologists.
Obstet Gynecol 2003;101:892-98.
5. Patient Protection and Affordable Care Act of
2010, Pub. L. No. 111-148, 124 Stat. 119
through 124 Stat. (Mar. 23, 2010).
6. An Act Concerning Post-partum Depression
and Amending, Pub. L. No. 2000, C.26:2-176
(April 13, 2006).
7. Perinatal Mental Health Disorders Prevention
and Treatment Act of 2008, Pub. L. No. 95-469
(Jan. 1, 2008).
8. Gavin N, Bradley G, Lohr K, Meltzer-Brody S,
Gartlehner G, Swinson T. Perinatal depression:
a systematic review of prevalence and inci-
dence. Obstet Gynecol 2005;106:1071-83.
9. Gordon TEJ, Cardone IA, Kim JJ, Gordon
SM, Silver RK. Universal perinatal depression
screening in an academic medical center. Ob-
stet Gynecol 2006;107:342-7.
10. Kim JJ, Gordon TEJ, La Porte LM, Adams
M, Kuendig JM, Silver RK. The utility of maternal
depression screening in the third trimester.
Am J Obstet Gynecol 2008;199:509.e5.
11. Murray D, Cox J. Screening for depression
during pregnancy with the Edinburgh Depres-
www.AJOG.org
sion Scale (EPDS). J Reprod Infant Psychol
1990;8:99-107.
12. Stowe ZN, Hostetter AL, Newport DJ. The
onset of postpartum depression: implications
for clinical screening in obstetrical and primary
care. Am J Obstet Gynecol 2005;192:522-6.
13. Cox J, Holden J. Perinatal mental health: a
guide to the Edinburgh Postnatal Depression
Scale. London: Gaskell; 2003.
14. Cardone IA, Kim JJ, Gordon TEJ, Gordon
SM, Silver RK. Psychosocial assessment by
phone for high-scoring patients taking the Ed-
inburgh Postnatal Depression Scale: communi-
cation pathways and strategies. Arch Womens
Ment Health 2006;9:87-94.
15. Murray L, Carothers AD. The validation of
the Edinburgh Postnatal Depression Scale on a
community sample. Br J Psychiatry 1990;157:
288-90.
16. Grote NK, Bledsoe SE, Swartz HA, Frank E.
Culturally relevant psychotherapy for prenatal
depression in low-income ob/gyn patients. Clin
Soc Work J 2004;32:327-47.
17. Chaudron LH, Szilagyi PG, Tang W, et al.
Accuracy of depression screening tools for
identifying postpartum depression among ur-
ban mothers. Pediatrics 2010;125:E609-17.
18. Cooper PJ, Murray L. Course and recur-
rence of postnatal depression: evidence for the
specificity of the diagnostic concept. Br J Psy-
chiatry 1995;166:826-7.
19. Leigh B, Milgrom J. Risk factors for antena-
tal depression, postnatal depression and par-
enting stress. BMJ 2008;8:24.
20. Dietz PM, Williams SB, Callaghan WM,
Bachman DJ, Whitlock EP, Hornbrook MC.
Clinically identified maternal depression be-
fore, during and after pregnancies ending
in live births. Am J Psychiatry 2007;164:
1515-20.