Thirty-six healthy pregnant women between the ages of 24 and 39 years were recruited via public maternity health care units in Uppsala County and through local newspaper advertisement. The inclusion criteria were gestational length more than 35 weeks, normal singleton pregnancy, and a planned vaginal delivery. Exclusion criteria were serious pregnancy-related complications (pre-eclampsia, intrauterine growth restriction, gestational diabetes), treatment with psychotropic drugs (including serotonin reuptake inhibitors), and ongoing anxiety and/or depressive disorders during pregnancy. Ongoing psychiatric disorders were evaluated using the Swedish version of the Mini International Neuropsychiatric Interview, a structured interview based on DSM-IV and ICD-10 (Sheehan et al. ¹⁹⁹⁸). In addition, exclusion criteria for the postpartum visit were severe delivery or postpartum complications (mother and/or fetus), and more than 50% non-responses at the first test session. Four women were non-responders, and one woman was excluded due to an obstetric complication. Hence, at the postpartum test session, 31 women with valid startle data and no severe obstetric complications remained eligible and consented to a second visit. The excluded women did not differ from the remaining study population in terms of age, parity, BMI, MADRS-S, and STAI-S scores (excluded 30.8 ± 3.4 vs. included 30.1 ± 3.8 years, 0.4 ± 0.9 vs. 0.4 ± 0.7 prior deliveries, 24.9 ± 4.7 vs. 23.9 ± 4.2 kg/m²; MADRS-S, 6.2 ± 3.2 vs. 6.3 ± 4.8; STAI-S, 32.4 ± 3.4 vs. 29.1 ± 6.4).

The study procedures were in accordance with ethical standards for human experimentation, and the study was approved by the Regional Ethical Review Board, Uppsala University, Sweden. Written informed consent was obtained from all subjects before inclusion.

All tests were carried out at the Department of Women’s and Children’s Health, Uppsala University between 0800 and 1800 hours. The first test session was scheduled approximately 14 days before the predicted day of delivery, based on routine ultrasound examination in gestational weeks 16–17. The second test session took place 4 to 6 weeks postpartum and was scheduled based on the date of delivery. All visits were made between January and August 2009. One researcher conducted all test sessions.

At the pregnancy test session, subjects were interviewed about obstetric history, medication during the preceding 3 months, and tobacco and alcohol use. At both visits, all subjects filled out the Montgomery-Åsberg Depression Rating Scale–Self-rated version, MADRS-S (Svanborg and Asberg ¹⁹⁹⁴) and the Spielberger State–Trait Anxiety Inventory–State version, STAI-S (Spielberger ¹⁹⁸³). Subjects also filled out the Edinburgh Postnatal Depression Scale, EPDS (Cox et al. ¹⁹⁸⁷) at the postpartum test session.

The acoustic startle response was assessed by electromyographic (EMG) recording of the right musculus orbicularis oculi. The acoustic startle probes and the recording of the eye blink responses were controlled by a commercial startle system (SR-HLAB, San Diego Instruments, San Diego, CA, USA). Acoustic startle probes were delivered binaurally by headphones (TDH-39-P, Maico, Minneapolis, MN, USA). The sound was calibrated with a Quest Electronics meter (model 210 Quest Technologies, Oconomov, WI, USA). After the skin was scarified with alcohol, two miniature silver/silver chloride electrodes (In Vivo Metric, Healdsburg, CA, USA) with a small amount of electrode gel (Sigma gel, In Vivo Metric, CA, USA), were positioned below the subject’s right eye, over the orbicularis oculi muscle. A ground electrode was placed in the center of the forehead (Blumenthal et al. ²⁰⁰⁵). Electrode impedances were confirmed to be less than 5 kOhm. The EMG was filtered (100–1,000 Hz), digitized at 1 kHz and analyzed by the EMG startle response software which rectified and smoothed the EMG response with a 10-ms time constant.

The subjects sat upright in an armchair and were instructed to watch a 14.1-inch computer monitor positioned approximately 1 m in front of the subject. The participants were aware that they were going to see pleasant and unpleasant pictures and hear unpleasant but harmless sound pulses throughout the experiment. The lights were switched off and the researcher left the room during testing. The test session began with a 5-min acclimation period without startle probes or images. Following the acclimation period, a 10-min slide show was displayed on the monitor while semi-randomized startle probes, 105-dB 40-ms broadband white noise with instantaneous rise time, were delivered. The startle modulation paradigm in this study is analogous to an image anticipation task previously used in a number of imaging studies of neural circuit activation across pleasant and unpleasant anticipation (Simmons et al. ²⁰⁰⁹; Aupperle et al. ²⁰¹¹). Each session consisted of 34 blocks containing three different startle conditions: (1) a black screen during which baseline startle response was measured (control condition), (2) a red or green screen as the negative or positive anticipation stimulus, (3) an unpleasant or pleasant image. A red screen always preceded unpleasant images and a green screen always preceded pleasant images. Across the 34 blocks, a total number of 48 startle probes were delivered; 20 during the control condition, seven during positive anticipation stimuli, seven during negative anticipation stimuli, seven during pleasant image stimuli, and seven during unpleasant image stimuli. The pleasant and unpleasant image blocks were counterbalanced. The control condition was presented for variable times, ranging between 8 and 13 s, while the anticipation stimuli were presented for 6 s and the image stimuli were presented for 2 s. Startle probes were delivered 4 to 10 s after the onset of the black screen (control condition), 1.5 to 4.5 s after the onset of the anticipation stimuli, and 0.5 to 1.5 s after onset of the image stimuli. To enable exclusion of trials with excessive background noise, 22 amplitude recordings were made when no startle probe was delivered (non-stimulus recordings).

The images were obtained from the International Affective Picture System (IAPS) (Lang et al. ²⁰⁰⁵). The pictures were selected to be pleasant or unpleasant, with 34 pictures from each category (specific numbers are listed in Table 1) divided in two series (A and B) with equal normative ratings of valence and degree of arousal according to the IAPS manual (Lang et al. ²⁰⁰⁵). Half of the study group was presented with image series A during pregnancy and series B on their postpartum assessment, while the other half saw the image series in opposite order. Previous studies have shown that no habituation in startle modulation, either by images or by anticipation, occurs when the affective images are changed between sessions (Larson et al. ²⁰⁰⁰; Bannbers et al. ²⁰¹¹). After every session, the subjects were asked to rate the image valence and arousal. Valence was rated on 10-cm visual analogue scales ranging from positive (0.0 cm) to negative (10.0 cm). Arousal ratings ranged between calm (0.0 cm) and aroused (10.0 cm).

Venous blood samples were collected after the startle sessions. The samples were centrifuged at 1,500 × g for 10 min and stored at −20°C within an hour after sampling. The steroid hormone analyses were carried out by competitive immunometry electrochemistry luminescence detection at the Department of Clinical Chemistry, Uppsala University hospital. The samples were run on a Roche Cobas e601 with Cobas Elecsys estradiol and progesterone reagent kits, respectively (Roche Diagnostics, Bromma, Sweden). For progesterone, the measurement interval was 0.1–191 nM and for estradiol 18.4–15,781 pM. Thus, the serum samples taken in late pregnancy were diluted 1:100. Progesterone intra-assay coefficient of variation was 2.2% at 2.39 nmol/L and 2.8% at 31.56 nmol/L. The total coefficient of variation was 4.8% at 2.52 nmol/L and 2.0% at 112 nmol/L. Estradiol intra-assay coefficient of variation was 6.8% at 85.5 pmol/L and 2.8% at 1,640 pmol/L. The total coefficient of variation was 4.7% at 120 pmol/L and 2.6% at 12,935 pmol/L.

Peak startle amplitudes were measured automatically within 20–150 ms following the onset of the startle probe. A zero response score was given if no response was detected according to the default criteria provided by the software: (1) the peak startle response occurred outside the 20–150-ms time frame (2) a baseline shift exceeded 40 arbitrary units, or (3) a startle response was 25 arbitrary amplitude units or less. Of the responses, 3.0% were scored as zero responses, and they were evenly distributed between the pregnant and postpartum states. Startle magnitude was defined as the total amplitude of all trials with response/total number of trials.

Because mean baseline startle response differed by a factor of 10 between individuals, z-scores standardized to the control condition within each subject and test session for pleasant and unpleasant images, and positive and negative anticipation were used in the analyses (magnitude of each startle − mean magnitude of control startles)/standard deviation of magnitude control startles. The startle magnitude during the four different stimuli was compared between the pregnancy and postpartum states by use of two-way ANOVA with repeated measures. Within subjects factors in the ANOVA analyses were state (pregnancy vs. postpartum period) and stimulus (positive anticipation stimuli vs. negative anticipation stimuli or pleasant images vs. unpleasant images). Differences between pregnant and postpartum states in baseline startle response (control condition), MADRS-S scores and STAI-S were compared with Mann–Whitney U tests and chi-square tests. All values in the text are displayed as mean ± SD, unless otherwise stated. The SPSS statistical package was used for the analyses. P values of less than 0.05 were considered to be statistically significant.

Startle magnitude during the control condition did not differ between the pregnant and postpartum state, Table 2. A significant main effect of stimulus was found (F(3,28) = 9.50; p < 0.001). Post hoc test revealed significant decreases in startle response between the control condition and the pleasant image stimuli (F(1,30) = 7.40; p < 0.05) and the positive anticipation stimuli (F(1,30) = 26.05; p < 0.001), whereas there was no difference between the control condition and the negative image stimuli (F(1,30) = 1.72; p = 0.20) and negative anticipation stimuli (F(1,30) = 2.41; p = 0.13), respectively, Table 2. Startle magnitude differed between the positive and negative anticipation stimuli (F(1,30) = 25.19; p < 0.001), but not between the pleasant and unpleasant image stimuli (F(1,30) = 3.98; p = 0.055). No difference in startle modulation was observed between image series A and B and the non-stimulus recordings revealed no significant background noise.

No difference in startle magnitude response to pleasant and unpleasant image stimuli across the pregnant and postpartum states was evident (main effect of state F(1,30) = 1.99; p = 0.169, state by stimulus interaction F(1,30) = 0.48; p = 0.492), Fig. 1.

As depicted in Fig. 2, a significant pregnancy/postpartum state by stimulus interaction for the positive and negative anticipation stimuli was found (F(1,30) = 4.86; p = 0.035), illustrated by a significant difference between the positive and negative anticipation conditions during late pregnancy (F(1,30) = 36.91; p < 0.001) but not at the postpartum assessment (F(1,30) = 4.06), Fig. 2.

As expected, the serum concentrations of estradiol and progesterone declined between the pregnancy and postpartum states (estradiol: pregnant, 108,063 ± 36,057 pmol/L; postpartum, 76 ± 56 pmol/L; progesterone: pregnant, 692.8 ± 240.6 nmol/L; postpartum, 1.0 ± 0.5 nmol/L). The difference between positive and negative anticipation stimuli (Δ–anticipation startle response) was significantly and positively correlated with progesterone levels in the postpartum state (r = 0.405, p = 0.024) but there was no correlation with estradiol levels (r = 0.01, p = 0.97). No correlations between progesterone- or estradiol serum concentrations and Δ–anticipation startle response were found during the pregnant state (progesterone: r = 0.09, p = 0.61; estradiol: r = 0.17, p = 0.35).

Mean valence and arousal scores are displayed in Table 3. A significant interaction was found between pregnancy/postpartum states and valence ratings: F(1,29) = 10.11; p < 0.01. The women rated the unpleasant images more negative and the pleasant images more positive at the postpartum visit than during pregnancy (p < 0.05, respectively). There was also a significant interaction between pregnant/postpartum states and arousal ratings: F(1,29) = 12.91; p = 0.001. The women indicated increased arousal scores for unpleasant images at their postpartum visit (p < 0.01), whereas there was no difference in arousal scores between pregnancy and the postpartum period for the pleasant images.

The authors declare that they have no conflict of interest.

This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.

AU arbitrary units

Data is presented as mean ± S.D. Valence was rated on 10-cm visual analogue scales ranging from positive (0.0 cm) to negative (10.0 cm). Arousal ratings were made from calm (0.0 cm) to aroused (10.0 cm)

*p < 0.05; **p < 0.01, compared with late pregnancy