Psychosocial stress moderates the relationships between oxytocin, perinatal depression, and maternal behavior

Hormones and Behavior 66 (2014) 351–360

Contents lists available at ScienceDirect

Hormones and Behavior journal homepage: www.elsevier.com/locate/yhbeh

Regular article

Psychosocial stress moderates the relationships between oxytocin, perinatal depression, and maternal behavior Phyllis Zelkowitz a,b,c,⁎, Ian Gold d, Nancy Feeley e,i, Barbara Hayton b, C. Sue Carter f, Togas Tulandi g, Haim A. Abenhaim h, Pavel Levin b,e,i a

Department of Psychiatry, McGill University, Montreal, Canada Department of Psychiatry, Jewish General Hospital, Montreal, Canada Lady Davis Institute, Jewish General Hospital, Montreal, Canada d Department of Philosophy, McGill University, Montreal, Canada e Centre for Nursing Research, McGill University, Jewish General Hospital, Montreal, Canada f Department of Psychiatry, University of North Carolina, Chapel Hill, NC, USA g Department of Obstetrics and Gynecology, McGill University, Montreal, Canada h Department of Obstetrics and Gynecology, Jewish General Hospital, Montreal, Canada i Ingram School of Nursing, McGill University, Montreal, Canada b c

a r t i c l e

i n f o

Article history: Received 8 November 2013 Received in revised form 13 June 2014 Accepted 14 June 2014 Available online 20 June 2014 Keywords: Oxytocin Depression Psychosocial stress Pregnancy Postpartum Maternal behavior

a b s t r a c t The hormone oxytocin (OT) is of particular interest in the study of childbearing women, as it has a role in the onset and course of labor and breastfeeding. Recent research has linked OT to maternal caregiving behavior towards her infant, and to postpartum depressive symptomatology. There is also evidence that psychosocial adversity affects the oxytocin system. The present study investigated the relationship of endogenous OT in women during pregnancy and at 8 weeks postpartum to psychosocial stress, maternal symptoms of depression, and maternal sensitive behavior. It was hypothesized that OT would mediate the effects of maternal depressive symptoms on maternal interactive behavior. We also tested the hypothesis that psychosocial stress would moderate the relationship between OT and maternal depressive symptoms and sensitive behavior. A community sample of 287 women was assessed at 12–14 weeks of gestation, 32–34 weeks of gestation, and 7–9 weeks postpartum. We measured plasma OT, maternal symptoms of depression and psychosocial stress. At the postpartum home visit, maternal behavior in interaction with the infant was videotaped, and then coded to assess sensitivity. In the sample as a whole, OT was not related to maternal depressive symptoms or to sensitive maternal behavior. However, among women who reported high levels of psychosocial stress, higher levels of plasma OT were associated with fewer depressive symptoms and more sensitive maternal behavior. These results suggest that endogenous OT may act as a buffer against the deleterious effects of stress, thereby protecting high risk women from developing depressive symptoms and promoting more sensitive maternal interactive behavior. © 2014 Elsevier Inc. All rights reserved.

Introduction Community studies of depression during pregnancy and postpartum report prevalence rates of 6.5–13% (Gavin et al., 2005). Depressive symptoms affect not only the woman herself, but also the mother– infant relationship and infant development. Depressed mothers are less likely than other mothers to engage in maternal behaviors such as face-to-face visual contact, “baby talk”, affectionate touch, sensitivity and contingent responsivity that are associated with optimal infant cognitive and socioemotional development (Cohn et al., 1990; Feldman and Eidelman, 2007; Field, 2010). While maternal psychological distress ⁎ Corresponding author at: Institute of Community and Family Psychiatry, Jewish General Hospital, 4333 Cote Ste Catherine Road, Montreal, QC H3T 1E4, Canada. E-mail address: [email protected] (P. Zelkowitz).

http://dx.doi.org/10.1016/j.yhbeh.2014.06.014 0018-506X/© 2014 Elsevier Inc. All rights reserved.

may be a proximal cause of parenting behavior, a mother's own history of poor parenting, experiences of abuse and neglect and current stressors may be distal influences on her parenting capacity (Barrett and Fleming, 2011). Recent research has examined neuroendocrine function as it relates to maternal experiences of adversity, depressive symptomatology, and interactive behavior. The hormone oxytocin (OT) is of particular interest in the study of childbearing women since it is implicated in labor and birth through stimulation of uterine contractions (Zeeman et al., 1997) as well as milk letdown (Brunton and Russell, 2008). OT may have a role in the regulation of both mammalian social behaviors and emotional reactivity (Carter and Keverne, 2002). It is synthesized within the paraventricular and supraoptic magnocellular nuclei of the hypothalamus, where it is transported axonally to the posterior pituitary gland, and also acts on brain regions involved in emotions and social

352

P. Zelkowitz et al. / Hormones and Behavior 66 (2014) 351–360

cognition (Neumann and Landgraf, 2012). OT can modulate stress reactivity by attenuating activation of the HPA axis (Smith and Wang, 2012), and can down-regulate the response to stressors and the reactivity of the autonomic nervous system (Neumann, 2008). There is conflicting evidence regarding the association between OT and depression. Some studies have found lower levels of plasma and salivary OT in depressed patients as compared to non-depressed controls (Apter-Levy et al., 2013; Zetzsche et al., 1996), with OT plasma concentrations decreasing with more elevated levels of symptomatology (Scantamburlo et al., 2007). In contrast, other research reports higher levels of plasma OT in depressed patients as compared to healthy controls (Parker et al., 2010). Yet another study found no difference between patients and controls in absolute levels of plasma OT, but reported a dysregulated pattern of OT pulsatile release in depressed patients under stressful conditions (Cyranowski et al., 2008). Lower OT levels during pregnancy have been associated with postpartum depression, though there was no concurrent association between OT and depressive symptoms during pregnancy (Skrundz et al., 2011). These discrepant findings may be due in part to a small sample size (often fewer than 25), and different timings, frequencies and methodologies of OT assays. Psychosocial stress may be related to dysregulation of the OT system. There is very limited and conflicting evidence concerning the effects of childhood traumatic experiences on the OT system, with some research finding lower levels of OT in women with a history of childhood maltreatment compared to healthy controls (Heim et al., 2009), others finding higher levels in response to a stressor (Seltzer et al., 2014) and yet others finding no such difference (Pierrehumbert et al., 2010). Experimental research has shown that the effects of intranasal OT administration are moderated by early experience (Riem et al., 2014; Simeon et al., 2011), enhancing social cognition and attenuating stress reactivity only in those with a history of early adversity. Studies of concurrent stress are also contradictory: higher plasma OT levels have been associated with marital quality (Gouin et al., 2010) but also with perceptions of interpersonal distress (Taylor et al., 2010; Turner et al., 1999). There is a need to consider how psychosocial stress may moderate the effects of endogenous OT on perinatal depression and maternal behavior. There is evidence in both the animal and the human literature implicating OT and its receptor in the mother–infant relationship. In rats, lactating females who exhibit low licking and grooming (LG) and their pups have lower OT receptor (OTR) binding in the medial preoptic area (MPOA) than high LG mothers and their offspring (Champagne et al., 2001; Francis et al., 2000). In human mothers, plasma OT is related to maternal mental representations of attachment (Levine et al., 2007), and to maternal behaviors such as gaze, vocalizations, positive affect, and affectionate touch (Feldman et al., 2007). Variation in the OT receptor gene (OXTR) has been associated with sensitive parenting behavior (Bakermans-Kranenburg and van IJzendoorn, 2008). These lines of evidence suggest that the OT system may represent a potential mechanism through which maternal mental health influences the development of the mother–infant relationship. Psychosocial stress early in life, as well as later adverse life circumstances, may be associated with dysregulation of the OT system, which in depressed mothers may affect their ability to manage stress, and provide sensitive and responsive care. The present study investigated the relationship of endogenous OT levels in women during pregnancy and at 8 weeks postpartum to psychosocial stress, maternal symptoms of depression, and maternal sensitive behavior. We examined changes over time in OT levels and in depressive symptoms, and whether such changes might be related. We tested the hypothesis that psychosocial stress moderates the association of OT with depressive symptoms and maternal sensitive behavior. We also tested the mediational hypothesis that mothers with more depressive symptoms have lower levels of endogenous OT, which in turn is associated with less sensitive maternal behavior.

Methods Sample A community sample of women was recruited from obstetrical practices and prenatal clinics at a general hospital and a birthing center in Montreal, Quebec, during the first routine prenatal examination. In order to be eligible for this study, women had to be 18 years of age or over, within 12–14 weeks of gestation, and expecting to deliver a single infant. An additional inclusion criterion was the ability to respond to questionnaires in either English or French. Women were excluded from the sample if they delivered preterm (before 36 weeks of gestation), and their infants were admitted to neonatal intensive care. A total of 341 women were recruited, of whom 29 were ultimately excluded due to miscarriage or preterm birth. The number of women followed to 8 weeks postpartum was 287, for a retention rate of 92%; 25 women did not complete the study because they were no longer interested, unavailable, did not wish to have a blood draw, or were asked by their partner to withdraw. This sample of 287 women was used in the analyses reported here. Dropouts did not differ from those retained in the study in terms of demographic variables, depressive symptoms and plasma OT levels at baseline. Demographic characteristics of the sample are shown in Table 1. Almost all the participants (93%) reported that they were breastfeeding their infants at 7–9 weeks postpartum. Previously published data from a subsample of the participants in this study showed that parity was associated with OT levels in late pregnancy, which were in turn associated with a more negative selfreported experience of labor and an increased likelihood of having epidural anesthesia during labor (Prevost et al., 2014). In addition, the amount of synthetic OT (syntocinon) administered during labor was associated with postpartum OT levels. These results do not alter the interpretation of the data presented in this report. Procedures Participants were assessed at 3 points in time: 12–14 weeks of gestation (T1), 32–34 weeks of gestation (T2), and 7–9 weeks postpartum (T3). The first two assessments took place at the hospital or birthing center, at regularly scheduled obstetrical care appointments. The postpartum assessment was conducted at the home of the participant. At each time point, participants completed a self-report questionnaire assessing symptoms of depression, and provided a 10 ml sample of blood, to measure circulating plasma OT levels. The postpartum blood sampling took place at least 30 min following breastfeeding. The measure of psychosocial stress was obtained at the first prenatal assessment. Table 1 Sample characteristics. Variable Maternal characteristics Age (years) Years of schooling Primiparous mothers Married or living with partner Infant gender (male) Psychosocial stress (ANRQ score) Endogenous oxytocin (pg/ml) In early pregnancy In late pregnancy Early postpartum Maternal depressive symptoms EPDS score in early pregnancy EPDS score in late pregnancy EPDS score early postpartum Maternal behavior Sensitivity

Mean (SD) or N (%) 31.48 (4.56) 16.50 (3.05) 137 (47.74%) 270 (94.08%) 152 (52.96%) 18.95 (10.28) 311.91 (283.81) 395.21 (278.31) 283.96 (269.89) 6.08 (4.22) 5.69 (4.27) 4.73 (4.00) 3.55 (0.62)

ANRQ = Antenatal Risk Questionnaire; EPDS = Edinburgh Postnatal Depression Scale; and SD = standard deviation.

P. Zelkowitz et al. / Hormones and Behavior 66 (2014) 351–360

353

Table 2 Correlates of depressive symptoms and maternal sensitivity.

Pearson correlations Maternal age Years of schooling OTa ANRQ score EPDS at T1 Mean differences Child gender Male (N = 152) Female (N = 135) Student's t Primiparous mother Yes (N = 150) No (N = 137) Student's t Partnered Yes (N = 270) No (N = 17) Student's t

EPDS at T1

EPDS at T2

EPDS at T3

Maternal sensitivity

−0.113 −0.155⁎⁎ 0.020 0.284⁎⁎⁎ N/A

−0.044 −0.149⁎ 0.064 0.315⁎⁎⁎ 0.476⁎⁎⁎

0.009 −0.078 −0.115 0.223⁎⁎⁎ 0.409⁎⁎⁎

0.138⁎ 0.113 0.083 0.034 −0.139⁎

5.72 (3.85) 6.50 (4.60) 1.566

5.74 (4.31) 5.63 (4.24) −0.222

4.57 (3.92) 4.92 (4.10) 0.744

3.50 (0.63) 3.60 (0.61) 1.21

5.95 (4.33) 6.23 (4.10) 0.575

5.82 (4.38) 5.54 (4.16) −0.557

4.35 (3.55) 5.15 (4.43) 1.680

3.61 (0.63) 3.48 (0.61) 0.620

5.95 (4.18) 8.24 (4.42) 2.183⁎

5.53 (4.30) 8.38 (2.58) 2.617⁎⁎

4.60 (3.94) 6.88 (4.51) 2.300⁎

3.56 (0.62) 3.27 (0.59) −1.803

EPDS = Edinburgh Postnatal Depression Scale; T1 = early pregnancy; T2 = late pregnancy; T3 = postpartum; and OT = oxytocin. Mean differences are reported with standard deviations in parentheses. a Log-transformed, concurrent for EPDS, T1 for maternal sensitivity. ⁎ p b 0.05. ⁎⁎ p b 0.01. ⁎⁎⁎ p b 0.001.

At the postpartum assessment, a brief observation of mother–infant interaction was videotaped for later coding. To measure circulating oxytocin, whole blood was obtained from participants in a heparinized tube, and stored on ice until centrifuged at 1600 ×g for 15 min at 4 °C. Plasma was pipetted off, aliquoted into 2 ml samples, then stored in vials and frozen at −80 °C. Samples were shipped to the laboratory of Dr. Sue Carter, where plasma levels of oxytocin were measured using an enzyme-linked immunosorbent assay (EIA) obtained from Enzo Life Sciences. Direct measurement of plasma was done following the manufacturers' protocols and samples were diluted to 1:2 or 1:4 to obtain values on the most sensitive part of the standard curve. Samples were assayed without extraction in duplicate or triplicate; the minimum detection limit for OT was 15.6 pg/ml, with inter- and intra-assay coefficients of variation (CVs) of less than 10%. The OT assay was chosen based on its sensitivity and specificity, and has been extensively validated (Carter, 2007; Kramer et al., 2004). The protocol used here has been employed in dozens of published studies relating plasma levels of OT to behavior (see review by Ebstein et al., 2012). Using this protocol, individual values in nonpregnant women and men, taken at intervals of one month or more, have been shown to be stable in both our laboratory (Gouin et al., 2010) and by other groups (Weisman et al., 2013). Values obtained in the present study are in the range that is typically reported in unextracted plasma samples in laboratories following this protocol. Measures Depressive symptomatology was assessed using the Edinburgh Postnatal Depression Scale (EPDS) (Cox et al., 1987). This widely used

measure has been validated for use in pregnancy (Murray and Cox, 1990). The 10 items ask women to report on symptoms during the past 7 days. A score of 12 or higher has optimal sensitivity and specificity in relationship to a diagnosis of major depression. The Antenatal Risk Questionnaire (ANRQ) (Austin et al., 2013) provided a measure of early and ongoing psychosocial stress. This instrument measures psychosocial stressors associated with perinatal depression (Austin et al., 2005). These include emotional support from the mother in childhood, past history of mental health problems, stressful life events in the past 12 months, social support, marital relations, and history of physical and sexual abuse. Scores range from 5 to 67; a cutoff score of 23 is used to identify high stress women who are at risk for perinatal depression, with the receiver operating characteristic (ROC) area under the curve of 0.69 (acceptable) at this cutoff point. In a study of more than 2100 pregnant women in Sydney, Australia, 28% of women scored above the cutoff score (Priest et al., 2008). In the present study, 33% of the participants (n = 94) scored above the cutoff score, and these women were at higher risk of scoring above the cutoff score of 12 on the EPDS at the postpartum assessment (10.6% vs. 3.6% of women who scored below 23 on the ANRQ, χ2 (1) = 5.40, p b .05). Our sample of high stress women did not differ from low stress women in terms of marital status, years of schooling, parity, or infant gender. The Global Ratings Scales (GRS) (Murray et al., 1996) assess mother– infant interaction. A 5 min, face-to-face interaction between mother and infant is videotaped, and rated by a trained coder who is blind to all other data that have been obtained from the mother, thereby providing a measure of maternal behavior that does not rely on maternal selfreport. The maternal sensitivity scale of the GRS is comprised of 5 items, each rated on a scale from 1 to 5, with higher scores reflecting

Table 3 Depressive symptoms and maternal sensitivity in high and low stress groups.

EPDS at T1 EPDS at T2 EPDS at T3 Maternal sensitivity

High stress groupa N = 94

Low stress groupa N = 190

95% CI for difference

Cohen's d

7.18 (4.11) 7.34 (4.47) 5.84 (4.68) 3.54 (0.61)

5.53 (4.21) 4.91 (3.95) 4.23 (3.53) 3.54 (0.63)

[−2.69, −0.61] [−3.46, −1.39] [−2.69, −0.53] [−1.64, 1.67]

0.39 0.59 0.41 0.00

EPDS = Edinburgh Postnatal Depression Scale; T1 = early pregnancy; T2 = late pregnancy; T3 = postpartum; and CI = confidence interval. High psychosocial stress: ≥23 on ANRQ; and low psychosocial stress: b23 on ANRQ (reference category). a Depression and sensitivity scores are expressed as mean (SD).

P. Zelkowitz et al. / Hormones and Behavior 66 (2014) 351–360

EPDS Score

354 8.5 8 7.5 7 6.5 6 5.5 5 4.5 4 3.5

High stress group Low stress group

T1

T2

T3

T2

T3

550

Oxytocin, pg/ml

500 450 400 350 300 250 200 T1

Time Fig. 1. Changes in average plasma OT and depressive symptoms at three time points: T1 = early pregnancy; T2 = late pregnancy; and T3 = early postpartum. The low stress group scored below 23 on the Antenatal Risk Questionnaire (ANRQ), and the high stress group scored 23 or higher on the ANRQ. The error bars indicate 95% bootstrapped confidence intervals for means. EPDS = Edinburgh Postnatal Depression Scale.

optimal behavior: warm/positive (measures physical and verbal expressions of affection and endearment), accepting (measures mother's ability to follow the interests of the infant without criticism or disappointment), responsive (measures awareness of and response to the infant's signals), non-demanding (measures the extent to which the mother requires the infant to perform in accordance with her expectations), and sensitive (measures mother's ability to empathize with her infant and respond appropriately to the infant's cues). This measure has been used cross-culturally (Gunning et al., 2004), and in studies of mother–infant interaction in the context of maternal mental illness (Mantymaa et al., 2004; Riordan et al., 1999). We evaluated inter-rater reliability by having two coders rate 20 videotapes; they achieved an intra-class correlation coefficient of .69, which is considered to be a satisfactory agreement.

discrete categories were assessed through analysis of variance (ANOVA). In order to examine effect sizes in our statistical analyses, we computed Cohen's ds for comparisons between high- and low stress groups. We examined changes in OT and depressive symptoms across the 3 time points of the study by performing multivariate analysis of variance (MANOVA) of transformed OT values and of EPDS scores with time as a within-subject covariate (Davis, 2002). Significance was determined using Pillai's trace statistic for within-subject effects, and using a multivariate F-test for between-subject effects. We assessed change over time by testing the terms of polynomial within-subject contrasts. Post-hoc comparisons between different time points were done with Bonferroni corrections. In our multivariate linear models the effect sizes were computed using partial eta squared. To address the question of how changes in plasma OT levels might be related to changes in depressive symptoms we computed the absolute changes in transformed OT and in EPDS scores between T1 and T2, and between T2 and T3 by simple subtraction. We then correlated corresponding values in order to test linear relationships. Multiple regression analysis was used to determine the predictors of maternal depressive symptoms and maternal sensitive behavior. Linear regression models were fitted using the standard least-squares approach. In order to test the moderation hypothesis, we used a hierarchical linear regression framework (Aiken and West, 1991). In particular, for each model, an F-test with one degree of freedom was used to

Data analysis All data analyses were performed with R statistical software (R Core Team, 2012). Before the analyses, OT values were log-transformed. Plasma concentrations tend to have heavily skewed distributions, and the logarithmic transformation is an acceptable way to correct for this. Indeed, after the transformation our OT values were approximately normally distributed. Bivariate correlations between continuous measures were computed using Pearson product–moment correlation coefficients. Differences in the means of continuous measures across

Table 4 Correlations between oxytocinc and depressive symptoms by psychosocial stress group. EPDS, T1

EPDS, T2 a

c

OT , T1 OTc, T2 OTc, T3

b

EPDS, T3 b

Maternal sensitivity

High stress

Low stress

High stress

Low stress

High stress

Low stress

High stressa

Low stressb

−.208⁎ −.152 −.229⁎

.123 .170⁎

.080 −.003 .038

.097 .050 .061

−.195 −.227⁎ −.367⁎⁎⁎

.092 .054 .052

.319⁎⁎ .117 .141

−.050 −.085 −.114

.123

a

a

b

ANRQ = Antenatal Risk Questionnaire; OT = oxytocin; T1 = early pregnancy; T2 = late pregnancy; T3 = postpartum; and EPDS = Edinburgh Postnatal Depression Scale. a High psychosocial stress: ≥23 on ANRQ. b Low psychosocial stress: b23 on ANRQ. c Log-transformed. ⁎ p b 0.05. ⁎⁎ p b 0.01. ⁎⁎⁎ p b 0.001.

P. Zelkowitz et al. / Hormones and Behavior 66 (2014) 351–360

355

Table 5 Regression analysis predicting depressive symptoms in early and late pregnancy and postpartum. SE B

β

t

p

−0.055 −0.084 −1.085 0.475 4.246 −1.943

−0.086 −0.124 −0.095 0.106 1.331 −1.171

−1.455 −2.046 −1.608 1.438 2.828 −2.459

0.147 0.042 0.109 0.152 0.005 0.015

Late pregnancy Age 0.018 Years of schooling −0.142 Marital status −1.246 EPDS (T1) 0.449 −0.226 OTa (T2) −1.177 Stress groupb 0.484 OTa × stress group interaction 2 2 N = 280, R = .281 (ΔR = .001), F(1, 268) = 0.389 (p = .533)

0.050 0.077 1.010 0.056 0.452 4.570 0.776

0.019 −0.100 −0.066 0.434 −0.033 −0.129 0.314

0.358 −1.835 −1.234 7.970 −0.501 −0.258 0.624

0.721 0.068 0.218 0.000 0.617 0.797 0.533

Postpartum Age .057 Years of schooling −.075 Marital status −1.114 EPDS (T1) .341 −.031 OTa (T3) 12.112 Stress groupb −2.047 OTa × stress group interaction N = 276, R2 = .241 (ΔR2 = .024), F(1, 266) = 8.403 (p = .004)

.048 .075 .948 .053 .439 3.879 .706

.065 −.057 −.065 .363 −.005 1.423 −1.327

1.188 −1.001 −1.175 6.429 −.070 3.122 −2.899

.236 .318 .241 .000 .944 .002 .004

B Early pregnancy Age −0.079 Years of schooling −0.172 Marital status −1.745 a 0.682 OT (T1) b 12.005 Stress group a OT × stress group interaction −1.876 N = 282, R2 = .096 (ΔR2 = .020), F(1, 275) = 6.045 (p = .015)

OT = oxytocin; EPDS = Edinburgh Postnatal Depression Scale; T1 = early pregnancy; T2 = late pregnancy; and T3 = postpartum. a Log-transformed. b High psychosocial stress: ≥23 on ANRQ; and low psychosocial stress: b23 on ANRQ (reference category).

determine whether a linear model with the interaction “product” term could account for significantly more variance than the main effects alone. For brevity, we only report the coefficients for the final models together with the R2 changes and F statistics associated with addition of the interaction terms. Possible mediation effects were analyzed through linear regression (Baron and Kenny, 1986) and the Sobel test (Sobel, 1982). We tested the mediation hypothesis by looking at the regression coefficient of EPDS predicting maternal sensitivity, the coefficient of EPDS predicting OT (the hypothesized mediator), the coefficient of OT predicting maternal sensitivity, and finally the coefficients of both OT and EPDS on maternal sensitivity when entered in the equation together. In addition we used the Sobel test to see the extent to which the effect of EPDS on maternal sensitivity is reduced after inclusion of OT as mediator. The maternal sensitivity variable had the most missing data points (38 cases, 13.2%), largely due to the fact that some infants fell asleep during the time of the home visit. The rate of missing cases in other variables ranged between 0 and 2.8%. To account for possible effects of missing data we performed multiple imputation (Rubin, 1987) with the “Multivariate Imputation by Chained Equations” library for R (van Buuren and Groothuis-Oudshoorn, 2011). The software used regression techniques to impute 10 complete copies of the data, analyzed each of them, and combined the results using Rubin's (1987) rules. The procedure showed that there was no bias due to missing data. We did not include the output of the multiple imputation analysis as it did not change the results. Results Association of psychosocial stress and OT during pregnancy and postpartum We found a small but statistically significant correlation between psychosocial stress, as measured by the ANRQ, and OT during pregnancy (Pearson r = .11, p b .05 in early pregnancy, and Pearson r = .13, p b .05

in late pregnancy). The association at 8 weeks postpartum was not statistically significant (Pearson r = .08, p = .20). Correlates of depressive symptoms during pregnancy and postpartum There were no consistent demographic correlates of depressive symptoms during pregnancy and the postpartum period (see Table 2). During pregnancy, years of education was negatively associated with EPDS scores. Single mothers reported more depressive symptoms than mothers who were partnered (i.e., married or living in common-law relationships). Parity and infant gender were unrelated to depressive symptoms at all 3 time points. There was moderate stability of depressive symptomatology across the 3 time points, with correlations of .48 (p b .01) between measurements at T1 and T2, .41 (p b .01) between T1 and T3, and .46 (p b .01) between T2 and T3. Psychosocial stress, as measured by the ANRQ, was significantly and positively correlated with depressive symptoms at all 3 time points (see Table 2). There were no statistically significant associations of plasma OT with EPDS scores across the 3 time points. Changes in OT and EPDS from pregnancy to postpartum We examined the trajectories of OT levels and EPDS scores over time (see Fig. 1), using a MANOVA repeated measures design, with maternal stress as a between-subjects factor. For OT levels, we found a significant within-subject effect of time (p b 0.01, η2 = .30), but not of stress (p = .18, η2 = .01) or time × stress interaction (p = .45, η2 = .00). Post-hoc analysis showed that mean OT increased from 311.91 to 395.21 (p b 0.01) from T1 to T2 and then decreased to 283.96 at T3 (p b 0.01). The difference in OT between T1 and T3 was not statistically significant (p = 0.12). A test of polynomial within-subject contrasts showed a significant quadratic term (p b 0.01 η2 = .29), confirming that the trajectories of OT were non-linear in time between early pregnancy and early postpartum (an increase followed by a decrease). These results show that stress did not appear to influence the change in OT over time.

356

P. Zelkowitz et al. / Hormones and Behavior 66 (2014) 351–360

B) T2

A) T1

C) T3

25

25

25

EPDS Score

Low stress group High stress group

20

20

20

15

15

15

10

10

10

5

5

5

0

0 100

0 100

1000

1000

100

1000

Oxytocin, pg/ml Fig. 2. The moderating effect of psychosocial stress on the association between oxytocin and depressive symptoms at three time points: T1 = early pregnancy; T2 = late pregnancy; and T3 = early postpartum. The low stress group scored below 23 on the Antenatal Risk Questionnaire (ANRQ), and the high stress group scored 23 or higher on the ANRQ. EPDS = Edinburgh Postnatal Depression Scale.

Analysis of within-subject effects of time on EPDS scores shows a statistically significant linear decrease in EPDS scores across the 3 time points (p b .01, η 2 = .09). Post-hoc analysis showed that T3 was statistically significantly lower than both T1 and T2 (p b .01 in both cases), however we cannot conclude that T1 was different from T2 (p = .27). As expected, the quadratic term of the polynomial contrasts was not statistically significant suggesting a simple linear decrease in EPDS values across the three time points. Maternal stress had a significant between-subjects effect on EPDS levels (p b .01, η 2 = .07), since it was associated with depression at all three time points (see Table 2), however the time × stress interaction was not significant (p = .27, η2 = .01). The linear correlations between changes in plasma OT with the changes in EPDS scores were not statistically significant, with correlations ranging from −0.10 to −0.07, p N .05. OT and depressive symptoms as a function of psychosocial stress We divided our sample into high and low stress groups, based on the cutoff score of 23 on the ANRQ. Women in the high stress group had higher EPDS scores (see Table 3). When we examined the associations of OT and depressive symptoms in the context of psychosocial stress, we found support for the moderation hypothesis. Higher levels of plasma OT were associated with less depressive symptomatology in

mothers who reported high levels of psychosocial stress (see Table 4). This was the case in early pregnancy and postpartum, but not in late pregnancy. In the low stress group, more depressive symptoms in early pregnancy were associated with higher OT levels in late pregnancy. There were no other statistically significant associations between OT and depressive symptoms in the low stress group. Multiple regression analysis was used to determine the significant predictors of depressive symptoms at each time point (see Table 5). The following variables were included in the regression models: age, marital status, years of education, concurrent OT level, and stress group. In early pregnancy, maternal education remained significantly associated with depressive symptoms, controlling for stress group, concurrent OT levels, and marital status. We also found a significant interaction effect of OT by psychosocial stress group (high vs. low), in that mothers with high psychosocial stress and lower levels of OT reported more depressive symptoms (see Fig. 2). The same interaction effect held in relation to depressive symptoms at 8 weeks postpartum, when depressive symptoms at T1 were added to the model. In the analysis predicting depressive symptoms in late pregnancy, only depressive symptoms at T1 were a significant predictor. Adding the OT × stress group interaction term to the main effects models, as a test of the moderation hypothesis, significantly improves the overall fit at T1 (ΔR2 = .020, p = .01) and at T3 (ΔR2 = .024, p = .00), but not at T2 (ΔR2 = .001, p = .53). Correlates of maternal sensitivity

Table 6 Regression analysis predicting maternal sensitivity. Variable

B

SE B

β

t

p

Age EPDS (T1) OTa (T1) Stress groupb OT × stress group interaction

0.015 −0.015 −0.037 −1.460 0.266

0.009 0.009 0.075 0.673 0.121

0.110 −0.103 −0.040 −1.111 1.141

1.740 −1.586 −0.485 −2.169 2.206

0.083 0.114 0.628 0.031 0.028

N = 246, R2 = .059 (ΔR2 = .019), F(1, 240) = 4.866 (p = .028). EPDS = Edinburgh Postnatal Depression Scale; T1 = early pregnancy; T3 = postpartum; and OT = oxytocin. a Log-transformed. b High psychosocial stress: ≥23 on ANRQ; and low psychosocial stress: b23 on ANRQ (reference category).

The correlates of maternal sensitivity are presented in Table 2. Maternal age was significantly and positively associated with maternal sensitive behavior. Years of schooling, parity, marital status and infant gender were not related to sensitivity. Depressive symptoms in early pregnancy were associated with less sensitive maternal behavior. In the sample as a whole, we found no association between plasma OT levels, measured during pregnancy and at 8 weeks postpartum, with maternal behavior. However, when we divided the sample into high and low stress groups, we found that OT in early pregnancy was associated with maternal sensitivity in the high stress group only (see Table 4). Moreover, we found that the association of depressive symptoms during pregnancy with sensitivity held only in the high stress

P. Zelkowitz et al. / Hormones and Behavior 66 (2014) 351–360

357

5 Low stress group High stress group

4.5

Maternal sensitivity

4

3.5

3

2.5

2

1.5

100

1000

Oxytocin, pg/ml Fig. 3. The moderating effect of psychosocial stress on the association between oxytocin and sensitive maternal behavior in the early postpartum. The low stress group scored below 23 on the Antenatal Risk Questionnaire (ANRQ), and the high stress group scored 23 or higher on the ANRQ. T1 = early pregnancy.

group (r = −0.24, p = 0.03); there was no association in the low stress group (r = −0.10, p = 0.23). Multiple regression analysis of the predictors of maternal sensitivity (see Table 6) included the following independent variables: maternal age, depressive symptoms in early pregnancy, Time 1 OT level, and risk group. The results again support the moderation hypothesis, in that the interaction of psychosocial stress levels and endogenous OT during pregnancy was associated with maternal sensitivity at 8 weeks postpartum (ΔR2 = .024, p = .00). High stress mothers with higher levels of OT during early pregnancy were more sensitive in interaction with their infants (see Fig. 3). OT as a mediator of the association between depressive symptoms and maternal behavior Our hypothesis that OT may function as a mediator (Baron and Kenny, 1986) between EPDS and maternal sensitivity is not supported by the data, since EPDS does not predict OT (β = 0.003, p = .73) and the main effect of OT on sensitivity is not significant when also controlling for EPDS (β = 0.076, p = .19). Furthermore, the 95% confidence interval for the product of path coefficients from EPDS to OT and from OT to maternal sensitivity (which quantifies the variance due to EPDS through OT) includes zero ([−0.64, 0.64]), indicating that there is not enough evidence to support the indirect effect of EPDS on sensitivity through OT (Sobel, 1982). Discussion The association between psychosocial stress and maternal mental health during pregnancy and the postpartum period has been well established (Robertson et al., 2004). There is also evidence that OT acts to attenuate psychological and physiological responses to psychosocial stress (Heinrichs et al., 2003). The results of the present study suggest that neuroendocrine factors may have a role in determining which mothers are likely to experience depressive symptoms both pre- and postpartum. Our hypothesis that psychosocial stress moderates the relationship between OT and depressive symptoms, and OT and maternal behavior, was supported. It appears that endogenous OT

may act as a buffer against the deleterious effects of stress, thereby protecting high stress women from developing depressive symptoms and promoting more sensitive maternal interactive behavior. In contrast, we found no evidence of an association between OT and mood, or between OT and sensitive maternal behavior, in women with low levels of psychosocial stress. These findings are consistent with experimental research showing that intranasal OT administration may have differential effects on attitudes and behavior as a function of interpersonal history or psychiatric diagnosis (Bartz et al., 2011). The presence of psychosocial stress was associated with somewhat higher levels of OT during pregnancy; however, there was no significant association at the postpartum assessment. Psychosocial stress was not associated with changes in OT levels over time. The experience of psychosocial stress may be associated with increased neuroendocrine response to stress (Heim et al., 2008), including the release of OT. In animal models, chronic stress during pregnancy prevents the normal rise in hypothalamic OT mRNA that occurs at the end of pregnancy (Hillerer et al., 2011). While psychosocial stress was associated with depressive symptomatology in the sample as a whole, depressive symptoms were attenuated in women at high psychosocial risk who exhibited elevated levels of OT. There is evidence of increased HPA axis reactivity in women suffering from symptoms of depression or posttraumatic stress disorder (Brand et al., 2010; Heim et al., 2008), and OT has been shown to reduce activation of the HPA axis (Ditzen et al., 2009; Heinrichs et al., 2003). Early adversity has been found to interact with polymorphisms of the OT receptor gene to predict postpartum depressive symptomatology (Jonas et al., 2013), suggesting greater vulnerability to the deleterious effects of poor quality care experiences among women with particular genetic and neuroendocrine profiles. Consistent with other research (Levine et al., 2007), we found wide variability in endogenous OT levels between individuals. However, while there have been reports of high individual stability in OT levels (Levine et al., 2007) over time, in this sample of childbearing women we found only moderate stability across time points, and an overall pattern of increasing OT from early to late pregnancy, followed by a decline in OT at 8 weeks postpartum. The pattern of attenuated symptomatology among high risk women was found in early pregnancy

358

P. Zelkowitz et al. / Hormones and Behavior 66 (2014) 351–360

and postpartum, when OT levels are similar, but not in late pregnancy, when OT levels are significantly higher than at the other two time points. There is evidence that the HPA axis declines in responsiveness at the end of pregnancy, perhaps in part to protect the developing fetus from the deleterious effects of maternal stress (Brunton and Russell, 2011). It is possible that the increasing numbers of OT receptors in late pregnancy (Fuchs et al., 1982) and higher levels of endogenous OT in the third trimester of pregnancy may affect the action of OT on the HPA axis. These factors may have affected our results, and suggest the importance of assessing OT and other hormones at multiple time points during pregnancy. The role of other hormones, such as estrogen, progesterone, prolactin, and vasopressin, which may vary widely during pregnancy and lactation, also needs to be considered (Stuebe et al., 2012). With respect to maternal interactive behavior, our results point to the role of OT in promoting sensitive maternal behavior, but only among women at high psychosocial risk. We were not able to confirm the hypothesis that OT mediates the relationship between maternal depressive symptoms and sensitive interactive behavior. Moreover, we found that it was OT measured in early pregnancy that was associated with more sensitive maternal behavior among high risk women. It is possible that the measurement of OT in early pregnancy constitutes a better approximation of a mother's baseline OT, since hormonal changes increase as pregnancy progresses. In addition, the administration of synthetic OT to induce or augment labor, may affect maternal endogenous OT levels in the postpartum period (Jonas et al., 2009; Prevost et al., 2014). Lactation may also affect endogenous OT. In fact, 93% of the mothers in our study were breastfeeding at the time of the postpartum assessment. In order to mitigate the effect of lactation, we obtained the blood samplings for the OT assay at least 30 min after a feeding. There is conflicting evidence regarding the association of certain polymorphisms of the OT receptor gene with maternal sensitivity. One study (Bakermans-Kranenburg and van IJzendoorn, 2008) demonstrated such a relationship, while another (Mileva-Seitz et al., 2013) failed to show this association. The extent to which epigenetic changes associated with psychosocial adversity may have an effect on the OT receptor gene, and ultimately on maternal sensitive behavior, would be an important direction for future research. The present study has some limitations. Because this was a community sample of mothers, levels of depressive symptomatology were generally low, with only about 10% of mothers scoring above the clinical cutoff for our depression measure in early pregnancy, and about 6% postpartum. Other research has shown that psychological symptoms decline between pregnancy and the postpartum period (Onoye et al., 2013; Sidebottom et al., 2014). Our longitudinal analyses confirmed that depressive symptoms were significantly lower at 8 weeks postpartum than at the other 2 time points of the study. This may help to account for the finding that, like other studies with community samples (Mileva-Seitz et al., 2013), we did not find an association between concurrent maternal depressive symptoms and sensitivity. It is possible that it is not depressive symptomatology per se but other factors, including socioeconomic status (van Doesum et al., 2007) and personality pathology (Conroy et al., 2010) that may be more closely related to the depressed mother's capacity for sensitive behavior. The fact that prenatal depressive symptoms were associated with maternal sensitivity is consistent with a growing literature on the independent effects of prenatal depression, even controlling for subsequent depressive episodes, on the mother–infant relationship and long-term developmental outcomes (Field, 2011; Hayes et al., 2012; Pearson et al., 2013). Another limitation involves our measure of psychosocial stress, which includes both concurrent and past stressors. The ANRQ assesses factors that have been demonstrated to predict postpartum depression, including a past history of depression, stressful life events and a lack of social support (Robertson et al., 2004). This would help to account for the small but statistically significant association between ANRQ and EPDS scores, which assess current levels of depressive symptoms. An

examination of the effects of early life adversity on maternal OT would require a more detailed assessment of the mother's history. However, measures of cumulative risk do have value, in that they have been shown to have stronger effects on postpartum depression and developmental outcome than individual risk factors (Evans et al., 2013; Liu and Tronick, 2013). Our study is also limited in that we assessed only circulating OT, and did not evaluate the possible effects of other hormones that might affect the function of OT or OT levels. For example, OT gene expression is affected by estrogen and progesterone (Hashimoto et al., 2012). Other physiological changes during pregnancy, such as the drop in albumin, which is the main carrier protein for OT, may also affect the detection of circulating OT, particularly during the third trimester (Abduljalil et al., 2012). Future studies must address the extent to which these factors may affect plasma OT levels. The extent to which peripheral measurement of OT (in plasma) corresponds to central levels of OT in brain areas affecting affect and behavior is difficult to determine (MacDonald and MacDonald, 2010). To date, the body of human research on the effects of OT on mood and the mother–infant relationship has relied on peripheral measures of OT, and the present study has employed the same measurement strategies. There is also debate regarding the accuracy of OT assays, with questions surrounding the use of extracted versus unextracted samples (Szeto et al., 2011), and radioimmunoassay versus enzyme immunoassay (Kramer et al., 2004). However, ongoing studies using mass spectrometry indicate that extraction removes much of the OT in plasma (Martin and Carter, 2013), and for this reason unextracted samples are used here. The development of more sensitive assays may help to resolve some of the discrepancies in research findings in this field. The present study suggests the importance of considering psychosocial risk factors in understanding the relationship between OT, mood, and maternal behavior. The identification of such risk factors may provide us with a better understanding of how environmental and hormonal factors may interact to affect the mother–infant relationship.

Acknowledgments This research was supported by a grant from the Canadian Institutes of Health Research (GTA-91755) Institute of Gender and Health. Nancy Feeley is supported by a Research Scholar Award from the Fonds de la recherche du Québec (FRQS). The authors would like to acknowledge the contribution of Stephanie Robins in coordinating the data collection and analysis, Hossein Nazarloo, who did the oxytocin assays, and the many research assistants involved in recruitment and data collection.

References Abduljalil, K., Furness, P., Johnson, T.N., Rostami-Hodjegan, A., Soltani, H., 2012. Anatomical, physiological and metabolic changes with gestational age during normal pregnancy: a database for parameters required in physiologically based pharmacokinetic modelling. Clin. Pharmacokinet 51, 365–396. Aiken, L.S., West, S.G., 1991. Multiple Regression: Testing and Interpreting Interactions. Sage, Thousand Oaks, CA. Apter-Levy, Y., Feldman, M., Vakart, A., Ebstein, R.P., Feldman, R., 2013. Impact of maternal depression across the first 6 years of life on the child's mental health, social engagement, and empathy: the moderating role of oxytocin. Am. J. Psychiatr. 170, 1161–1168. Austin, M.P., Hadzi-Pavlovic, D., Saint, K., Parker, G., 2005. Antenatal screening for the prediction of postnatal depression: validation of a psychosocial Pregnancy Risk Questionnaire. Acta Psychiatr. Scand. 112, 310–317. Austin, M.P., Colton, J., Priest, S., Reilly, N., Hadzi-Pavlovic, D., 2013. The antenatal risk questionnaire (ANRQ): acceptability and use for psychosocial risk assessment in the maternity setting. Women Birth 26, 17–25. Bakermans-Kranenburg, M.J., van IJzendoorn, M.H., 2008. Oxytocin receptor (OXTR) and serotonin transporter (5-HTT) genes associated with observed parenting. Soc. Cogn. Affect. Neurosci. 3, 128–134. Baron, R.M., Kenny, D.A., 1986. The moderator–mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J. Pers. Soc. Psychol. 51, 1173–1182.

P. Zelkowitz et al. / Hormones and Behavior 66 (2014) 351–360 Barrett, J., Fleming, A.S., 2011. Annual research review: all mothers are not created equal: neural and psychobiological perspectives on mothering and the importance of individual differences. J. Child Psychol. Psychiatry 52, 368–397. Bartz, J., Simeon, D., Hamilton, H., Kim, S., Crystal, S., Braun, A., Vicens, V., Hollander, E., 2011. Oxytocin can hinder trust and cooperation in borderline personality disorder. Soc. Cogn. Affect. Neurosci. 6, 556–563. Brand, S.R., Brennan, P.A., Newport, D.J., Smith, A.K., Weiss, T., Stowe, Z.N., 2010. The impact of maternal childhood abuse on maternal and infant HPA axis function in the postpartum period. Psychoneuroendocrinology 35, 686–693. Brunton, P.J., Russell, J.A., 2008. The expectant brain: adapting for motherhood. Nature reviews. Neuroscience 9, 11–25. Brunton, P.J., Russell, J.A., 2011. Neuroendocrine control of maternal stress responses and fetal programming by stress in pregnancy. Prog. Neuropsychopharmacol. Biol. Psychiatry 35, 1178–1191. Carter, C.S., 2007. Sex differences in oxytocin and vasopressin: implications for autism spectrum disorders? Behav. Brain Res. 176, 170–186. Carter, C.S., Keverne, E.B., 2002. The neurobiology of social affiliation and pair bonding. In: Pfaff, D.W. (Ed.), Hormones, Brain, and Behavior. Academic Press, San Diego, CA, pp. 299–337. Champagne, F., Diorio, J., Sharma, S., Meaney, M.J., 2001. Naturally occurring variations in maternal behavior in the rat are associated with differences in estrogen-inducible central oxytocin receptors. Proc. Natl. Acad. Sci. U. S. A. 98, 12736–12741. Cohn, J.F., Campbell, S.B., Matias, R., Hopkins, J., 1990. Face-to-face interactions of postpartum depressed and non-depressed mother–infant pairs at two months. Dev. Psychol. 26, 15–23. Conroy, S., Marks, M., Schacht, R., Davies, H., Moran, P., 2010. The impact of maternal depression and personality disorder on early infant care. Soc. Psychiatry Psychiatr. Epidemiol. 45, 285–292. Cox, J.L., Holden, J.M., Sagovsky, R., 1987. Detection of postnatal depression: development of the 10-item Edinburgh Postnatal Depression Scale. Br. J. Psychiatry 150, 782–786. Cyranowski, J.M., Hofkens, T.L., Frank, E., Seltman, H., Cai, H.-M., Amico, J.A., 2008. Evidence of dysregulated peripheral oxytocin release among depressed women. Psychosom. Med. 70, 967–975. Davis, C.S., 2002. Statistical Methods for the Analysis of Repeated Measurements. Springer, New York. Ditzen, B., Schaer, M., Gabriel, B., Bodenmann, G., Ehlert, U., Heinrichs, M., 2009. Intranasal oxytocin increases positive communication and reduces cortisol levels during couple conflict. Biol. Psychiatry 65, 728–731. Ebstein, R.P., Knafo, A., Mankuta, D., Chew, S.H., Lai, P.S., 2012. The contributions of oxytocin and vasopressin pathway genes to human behavior. Horm. Behav. 61, 359–379. Evans, G.W., Li, D., Whipple, S.S., 2013. Cumulative risk and child development. Psychol. Bull. 139, 1342–1396. Feldman, R., Eidelman, A.I., 2007. Maternal postpartum behavior and the emergence of infant–mother and infant–father synchrony in preterm and full-term infants: the role of neonatal vagal tone. Dev. Psychobiol. 49, 290–302. Feldman, R., Weller, A., Zagoory-Sharon, O., Levine, A., 2007. Evidence for a neuroendocrinological foundation of human affiliation: plasma oxytocin levels across pregnancy and the postpartum period predict mother–infant bonding. Psychol. Sci. 18, 965–970. Field, T., 2010. Postpartum depression effects on early interactions, parenting, and safety practices: a review. Infant Behav. Ddev. 33, 1–6. Field, T., 2011. Prenatal depression effects on early development: a review. Infant Behav. Dev. 34, 1–14. Francis, D.D., Champagne, F.C., Meaney, M.J., 2000. Variations in maternal behaviour are associated with differences in oxytocin receptor levels in the rat. J. Neuroendocrinol. 12, 1145–1148. Fuchs, A.R., Fuchs, F., Husslein, P., Soloff, M.S., Fernstrom, M.J., 1982. Oxytocin receptors and human parturition: a dual role for oxytocin in the initiation of labor. Science 215, 1396–1398. Gavin, N.I., Gaynes, B.N., Lohr, K.N., Meltzer-Brody, S., Gartlehner, G., Swinson, T., 2005. Perinatal depression: a systematic review of prevalence and incidence. Obstet. Gynecol. 106, 1083. Gouin, J.P., Carter, C.S., Pournajafi-Nazarloo, H., Glaser, R., Malarkey, W.B., Loving, T.J., Stowell, J., Kiecolt-Glaser, J.K., 2010. Marital behavior, oxytocin, vasopressin, and wound healing. Psychoneuroendocrinology 35, 1082–1090. Gunning, M., Conroy, S., Valoriani, V., Figueiredo, B., Kammerer, M.H., Muzik, M., GlatignyDallay, E., Murray, L., 2004. Measurement of mother–infant interactions and the home environment in a European setting: preliminary results from a cross-cultural study. Br. J. Psychiatry 184, s38–s44. Hashimoto, H., Uezono, Y., Ueta, Y., 2012. Pathophysiological function of oxytocin secreted by neuropeptides: a mini review. Pathophysiology 19, 283–298. Hayes, L.J., Goodman, S.H., Carlson, E., 2012. Maternal antenatal depression and infant disorganized attachment at 12 months. Attach Hum. Dev. 1–21. Heim, C., Newport, D.J., Mletzko, T., Miller, A.H., Nemeroff, C.B., 2008. The link between childhood trauma and depression: insights from HPA axis studies in humans. Psychoneuroendocrinology 33, 693–710. Heim, C., Young, L.J., Newport, D.J., Mletzko, T., Miller, A.H., Nemeroff, C.B., 2009. Lower CSF oxytocin concentrations in women with a history of child abuse. Mol. Psychiatry 14, 954–958. Heinrichs, M., Baumgartner, T., Kirschbaum, C., Ehlert, U., 2003. Social support and oxytocin interact to suppress cortisol and subjective responses to psychosocial stress. Biol. Psychiatry 54, 1389–1398. Hillerer, K.M., Reber, S.O., Neumann, I.D., Slattery, D.A., 2011. Exposure to chronic pregnancy stress reverses peripartum-associated adaptations: implications for postpartum anxiety and mood disorders. Endocrinology 152, 3930–3940.

359

Jonas, K., Johansson, L.M., Nissen, E., Ejdeback, M., Ransjo-Arvidson, A.B., Uvnas-Moberg, K., 2009. Effects of intrapartum oxytocin administration and epidural analgesia on the concentration of plasma oxytocin and prolactin, in response to suckling during the second day postpartum. Breastfeed. Med. 4, 71–82. Jonas, W., Mileva-Seitz, V., Girard, A.W., Bisceglia, R., Kennedy, J.L., Sokolowski, M., Meaney, M.J., Fleming, A.S., Steiner, M., M.R.T., 2013. Genetic variation in oxytocin rs2740210 and early adversity associated with postpartum depression and breastfeeding duration. Genes Brain Behav. 12, 681–694. Kramer, K.M., Cushing, B.S., Carter, C.S., Wu, J., Ottinger, M.A., 2004. Sex and species differences in plasma oxytocin using an enzyme immunoassay. Can. J. Zool. 82, 1194–1200. Levine, A., Zagoory-Sharon, O., Feldman, R., Weller, A., 2007. Oxytocin during pregnancy and early postpartum: individual patterns and maternal–fetal attachment. Peptides 28, 1162–1169. Liu, C.H., Tronick, E., 2013. Re-conceptualising prenatal life stressors in predicting postpartum depression: cumulative-, specific-, and domain-specific approaches to calculating risk. Paediatr. Perinat. Epidemiol. 27, 481–490. MacDonald, K., MacDonald, T.M., 2010. The peptide that binds: a systematic review of oxytocin and its prosocial effects in humans. Harvard Rev. Psychiatry 18, 1–21. Mantymaa, M., Puura, K., Luoma, I., Salmelin, R.K., Tamminen, T., 2004. Early mother–infant interaction, parental mental health and symptoms of behavioral and emotional problems in toddlers. Infant Behav. Dev. 27, 134–149. Martin, W.L., Carter, C.S., 2013. Oxytocin and vasopressin are sequestered in plasma. 10th World Congress of Neurohypophyseal Hormones Bristol, England. Mileva-Seitz, V., Steiner, M., Atkinson, L., Meaney, M.J., Levitan, R., Kennedy, J.L., Sokolowski, M.B., Fleming, A.S., 2013. Interaction between oxytocin genotypes and early experience predicts quality of mothering and postpartum mood. PLoS ONE 8, e61443. Murray, D., Cox, J.L., 1990. Screening for depression during pregnancy with the Edinburgh Depression Scale (EPDS). J. Reprod. Infant Psychol. 8, 99–107. Murray, L., Fiori-Cowley, A., Hooper, R., Cooper, P., 1996. The impact of postnatal depression and associated adversity on early mother–infant interactions and later infant outcome. Child Dev. 67, 2512–2526. Neumann, I.D., 2008. Brain oxytocin: a key regulator of emotional and social behaviours in both females and males. J. Neuroendocrinol. 20, 858–865. Neumann, I.D., Landgraf, R., 2012. Balance of brain oxytocin and vasopressin: implications for anxiety, depression, and social behaviors. Trends Neurosci. 35, 649–659. Onoye, J., Shafer, L., Goebert, D., Morland, L., Matsu, C., Hamagami, F., 2013. Changes in PTSD symptomatology and mental health during pregnancy and postpartum. Arch. Women's Ment. Health 1–11. Parker, K.J., Kenna, H.A., Zeitzer, J.M., Keller, J., Blasey, C.M., Amico, J.A., Schatzberg, A.F., 2010. Preliminary evidence that plasma oxytocin levels are elevated in major depression. Psychiatry Res. 178, 359–362. Pearson, R.M., Evans, J., Kounali, D., et al., 2013. Maternal depression during pregnancy and the postnatal period: risks and possible mechanisms for offspring depression at age 18 years. JAMA Psychiatry 70, 1312–1319. Pierrehumbert, B., Torrisi, R., Laufer, D., Halfon, O., Ansermet, F., Beck Popovic, M., 2010. Oxytocin response to an experimental psychosocial challenge in adults exposed to traumatic experiences during childhood or adolescence. Neuroscience 166, 168–177. Prevost, M., Zelkowitz, P., Tulandi, T., Hayton, B., Feeley, N., Carter, C.S., Joseph, L., Pournajafi-Nazarloo, H., Yong Ping, E., Abenhaim, H., Gold, I., 2014. Oxytocin in pregnancy and the postpartum: relations to labor and its management. Front. Publ. Health 2. Priest, S.R., Austin, M.P., Barnett, B.B., Buist, A.E., 2008. A psychosocial risk assessment model (PRAM) for use with pregnant and postpartum women in primary care settings. Arch. Women's Ment. Health 11, 307–317. R Core Team, 2012. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. Riem, M.M., Bakermans-Kranenburg, M.J., Voorthuis, A., van Ijzendoorn, M.H., 2014. Oxytocin effects on mind-reading are moderated by experiences of maternal love withdrawal: an fMRI study. Prog. Neuropsychopharmacol. Biol. Psychiatry 51C, 105–112. Riordan, D., Appleby, L., Faragher, B., 1999. Mother–infant interaction in post-partum women with schizophrenia and affective disorders. Psychol. Med. 29, 991–995. Robertson, E., Grace, S., Wallington, T., Stewart, D.E., 2004. Antenatal risk factors for postpartum depression: a synthesis of recent literature. Gen. Hosp. Psychiatry 26, 289–295. Rubin, D.B., 1987. Multiple Imputation for Nonresponse in Surveys. John Wiley, New York. Scantamburlo, G., Hansenne, M., Fuchs, S., Pitchot, W., MarÈchal, P., Pequeux, C., Ansseau, M., Legros, J.J., 2007. Plasma oxytocin levels and anxiety in patients with major depression. Psychoneuroendocrinology 32, 407–410. Seltzer, L.J., Ziegler, T., Connolly, M.J., Prososki, A.R., Pollak, S.D., 2014. Stress-induced elevation of oxytocin in maltreated children: evolution, neurodevelopment, and social behavior. Child Dev. 85, 501–512. Sidebottom, A.C., Hellerstedt, W.L., Harrison, P.A., Hennrikus, D., 2014. An examination of prenatal and postpartum depressive symptoms among women served by urban community health centers. Arch. Womens Ment. Health 17, 27–40. Simeon, D., Bartz, J., Hamilton, H., Crystal, S., Braun, A., Ketay, S., Hollander, E., 2011. Oxytocin administration attenuates stress reactivity in borderline personality disorder: a pilot study. Psychoneuroendocrinology 36, 1418–1421. Skrundz, M., Bolten, M., Nast, I., Hellhammer, D.H., Meinlschmidt, G., 2011. Plasma oxytocin concentration during pregnancy is associated with development of postpartum depression. Neuropsychopharmacology 36, 1886–1893. Smith, A.S., Wang, Z., 2012. Salubrious effects of oxytocin on social stress-induced deficits. Horm. Behav. 61, 320–330. Sobel, M.E., 1982. Asymptotic confidence intervals for indirect effects in structural equation models. Sociol. Methodol. 13, 290–312.

360

P. Zelkowitz et al. / Hormones and Behavior 66 (2014) 351–360

Stuebe, A.M., Grewen, K., Pedersen, C.A., Propper, C., Meltzer-Brody, S., 2012. Failed lactation and perinatal depression: common problems with shared neuroendocrine mechanisms? J. Womens Health (Larchmt) 21, 264–272. Szeto, A., McCabe, P.M., Nation, D.A., Tabak, B.A., Rossetti, M.A., McCullough, M.E., Schneiderman, N., Mendez, A.J., 2011. Evaluation of enzyme immunoassay and radioimmunoassay methods for the measurement of plasma oxytocin. Psychosom. Med. 73, 393–400. Taylor, S.E., Saphire-Bernstein, S., Seeman, T.E., 2010. Are plasma oxytocin in women and plasma vasopressin in men biomarkers of distressed pair-bond relationships? Psychol. Sci. 21, 3–7. Turner, R.A., Altemus, M., Enos, T., Cooper, B., McGuinness, T., 1999. Preliminary research on plasma oxytocin in normal cycling women: investigating emotion and interpersonal distress. Psychiatry 62, 97–113.

van Buuren, S., Groothuis-Oudshoorn, 2011. mice: Multivariate Imputation by Chained Equations in R. Journal of Statistical Software, 45. van Doesum, K.T.M., Hosman, C.M.H., Riksen-Walraven, J.M., Hoefnagels, C., 2007. Correlates of depressed mothers' sensitivity toward their infants: the role of maternal, child, and contextual characteristics. J. Am. Acad. Child Adolesc. Psychiatry 46, 747–756. Weisman, O., Zagoory-Sharon, O., Schneiderman, I., Gordon, I., Feldman, R., 2013. Plasma oxytocin distributions in a large cohort of women and men and their gender-specific associations with anxiety. Psychoneuroendocrinology 38, 694–701. Zeeman, G.G., Khan-Dawood, F.S., Dawood, M.Y., 1997. Oxytocin and its receptor in pregnancy and parturition: current concepts and clinical implications. Obstet. Gynecol. 89, 873–883. Zetzsche, T., Frasch, A., Jirikowski, G.F., Murck, H., Steiger, A., 1996. Nocturnal oxytocin secretion is reduced in major depression. Biol. Psychiatry 39, 584.