Arch Womens Ment Health
DOI 10.1007/s00737-017-0728-7
LETTER TO THE EDITOR
Associated symptoms of depression: patterns
of change during pregnancy
Rita T. Amiel Castro 1,2 & Claudia Pinard Anderman 3 & Vivette Glover 1 &
Thomas G. OConnor 4 & Ulrike Ehlert 2 & Martin Kammerer 1
Received: 18 April 2017 / Accepted: 3 May 2017
# Springer-Verlag Wien 2017
Dear Editor,
We thank the authors of the letter for their interest in and
comments on our article, BAssociated symptoms of depres-
sion: patterns of change during pregnancy.^ The aim of our
paper was to examine the natural course of specific depressive
symptoms throughout pregnancy in a non-clinic sample. The
findings indicate that, for some symptoms, there are sizable
and non-linear changes in symptom frequency across the 9-
month period of study. Whether or not these symptom-level
changes comport with biological and social changes in preg-
nancy is not yet clear, but we hope that the findings may
encourage further research that adopts an assessment method
that is more attentive to the evident detailed changes by symp-
tom and stage of pregnancy.
In their letter, the authors offered several helpful com-
ments; we will address these in turn. The first concerns the
variation in p values in Table 2, which displays the associa-
tions between the core symptom of depressed mood and the
eight additional symptoms and features of depression. We
Rita T. Amiel Castro and Claudia Pinard Anderman contributed equally
to this work.
* Martin Kammerer
m.kammerer@imperial.ac.uk
1
Faculty of Medicine, Institute of Reproductive and Developmental
Biology, Imperial College London, London, UK
2
Institute of Psychology, Department of Clinical Psychology and
Psychotherapy, University of Zurich, Zurich, Switzerland
3
Department of Applied Psychology, University of Applied Sciences
Zurich, Zurich, Switzerland
4
Department of Psychiatry, University of Rochester Medical Centre,
Rochester, NY, USA
agree that close attention to the statistical significance is im-
portant, perhaps particularly where multiple tests are conduct-
ed. In that regard, and using the authors example of self-
esteem, it would certainly be possible to consider p values
adjusted for the number of tests, i.e., by each month of gesta-
tion, or .05/9. That alteration would result in modest changes
in p values in the Table. We did not formally do this for two
reasons. One technical reason, which we note in the text, is
that this analysis is exploratory because we had no a priori
expectations (as it was a novel analysis). The second and more
substantive point is that the key feature of Table 2 is not the p
value but the effect size. That is, what is important about
Table 2 is the degree to which the target symptom covaries
with depressed mood across pregnancy and may (or may not)
constitute a Bcore^ feature of depression. In the Discussion
and conclusion section, we note that the patterns of overlap
does not easily match what is known about biological changes
in pregnancy (that is especially so where quadratic patterns are
detected).
A second issue highlighted by the authors is the need for
greater research to identify the sources and causes of depres-
sion in pregnancy. We agree with that. This paper was princi-
pally concerned with patterns over time, however, and so our
analyses of etiology were limited. A further comment was
about the nature of the sample. As we note in the paper, we
excluded patients using psychotropic medication and/or in
psychological treatment to avoid the confounds of differential
effect on symptoms. No exclusions were made about symp-
tom severity without regard to treatment exposure: women
who fulfilled the SCID criteria and were not in use of psycho-
tropic medication and/or in a psychological treatment at re-
cruitment were included (these women composed about 4% of
our sample).
The authors also raise the important point about recall bias
and queried if we were sufficiently attentive to that matter. We

do make very clear in the paper why this is an important issue,
and how it affects study design and results interpretation.
Lastly, the authors wondered whether trimester rather
than monthly assessments would have been preferable.
We are not aware of any empirical evidence relevant to
that decision. Our focus on monthly assessments derived
from our interest in capturing a fine-grained develop-
mental picture of mood throughout pregnancy. In this
context, we are aware of several research groups using
ecological momentary assessments in pregnant women
and look forward to their results on the degree to which
mood and features of depression change, and the rea-
sons why, throughout pregnancy. Our findings and theirs
may be instructive for refining biopsychosocial hypoth-
eses for understanding depression in pregnancy and at
other time periods in the life course, and that will ben-
efit treatments.
Amiel Castro R.T. et al.
In summary, we believe our results support our conclusion
that both psychological and somatic symptoms change during
pregnancy, but in different patterns, and that the discrepancy
between the patterns of depressed mood and many somatic
and psychological symptoms suggest complex interactions
and potentially important implications for assessment and
monitoring treatment.
We also note typographic errors in Table 2. The phi
coefficient between Depressed Mood and Decreased ener-
gy at month 1 should be .086; phi coefficient between
Depressed Mood and Lack of concentration at month 8
should be .085; phi coefficient between Depressed Mood
and Guilt at month 4 should be .040; phi coefficient
between Depressed Mood and Guilt at month 6 should
be .031. Moreover, we also note a duplicated sentence
in the paragraph before the last paragraph of our paper
(Discussion and conclusion section).