Anxiety, depression and relationship satisfaction in the pregnancy following stillbirth and after the birth of a live-born baby: a prospective study

Depression and anxiety was measured using short versions of the Hopkins Symptom Checklist (SCL) [28] shown to correlate highly with the total score of the original scale, and to have good psychometric properties [29, 30]. We used two 4-item subscales measuring anxiety and depression during the previous 2 weeks (SCL-4a and SCL-4d). A combined score was used in pregnancy week 17. Items were scored on a Likert scale ranging from 1 (“not at all bothered”) to 4 (“very much bothered”). We defined a mean score > 1.75 on SCL-4a and/or SCL-4d as presence of anxiety and/or depression [31]. Cronbach’s alpha of internal consistency ranged from 0.69–0.80 for the anxiety subscale and 0.77–0.81 for the depression subscale.

A five-item version of the Relationship Satisfaction Scale (RS) was used to assess maternal relationship satisfaction among married/cohabiting women [32]. Developed for the MoBa study, the RS is based on core items from previously developed measures of marital satisfaction and relationship quality [33‐35]. The RS correlates 0.92 with the Quality of Marriage index [36] and has a high ability to predict future break-up/divorce and life satisfaction [32, 37, 38]. The abbreviated five-item version (RS5) correlates 0.97 with the full 10-item version [32]. Each item is rated on a 6-point (1–6) Likert scale, and the total score is the mean score of all items. An average score below 4.0 implies a relatively high risk of break-up (11–15%) [32] and a score ≥ 4 was applied as cut-off to denote relationship satisfaction in this study. Cronbach’s alpha ranged from 0.87 to 0.90.

Sociodemographic, health related and obstetrical history factors were considered as potential confounders for the association between having experienced a stillbirth and anxiety or depression in the subsequent pregnancy. Maternal age at the time of the MoBa delivery (whole years) was retrieved from the MBRN. Co-morbidity was defined as having at least one of the following previous medical problems reported in the MBRN: Asthma, hypertension, recurrent urinary tract infections, kidney disease, rheumatoid arthritis, heart disease, epilepsy, diabetes mellitus, and/or thyroid disease. Other covariates were questionnaire data obtained at gestational week 17 or 30 and included parental status at first assessment (married/cohabiting), native language other than Norwegian, pre-pregnancy daily smoking, high pre-pregnancy body mass index (BMI ≥ 25), low education (high school or less), low income (< 200,000 Norwegian kroner/year) and previous termination(s) of pregnancy or miscarriage(s). Stressful life events were defined as having at least one of the following experiences during the last 12 months: 1) Problems at work or study place, 2) financial problems, 3) divorce/separation/relationship break-up, 4) conflicts with family or friends, 5) serious injury or illness to the woman herself or a loved one, or 6) involvement in a serious accident, fire or robbery.

Information on gestational age at the time of stillbirth and inter-pregnancy interval was retrieved from the MBRN. Gestational age at stillbirth (based on last menstrual period and/or ultrasound measurement) was categorised as ≤ 30 weeks or > 30 weeks. Inter-pregnancy interval was defined as number of months between the date of stillbirth and the subsequent conception (estimated by ultrasound measurements) and categorised as < 12 months or ≥ 12 months.

Categorical data were reported as proportions and compared between groups using chi-square tests. Age at the time of the MoBa delivery was reported as mean years and compared between groups using independent samples t-test. To reduce potential sample distortion caused by missing values, the Estimation-Maximation procedure in SPSS was used to impute missing values on SCL-4a, SCL-4d and RS5 if at least 50% of items were present. This resulted in 0.4% missing on SCL-4a, 0.4% missing on SCL-4d and 1.9% missing on RS5 at first assessment.

For the stillbirth group, separate binary regression models were used to test if gestational age at stillbirth or inter-pregnancy interval were significant predictors for anxiety or depression in the subsequent pregnancy. To preserve power and reduce the number of comparisons, we combined anxiety and depression in the subgroup analyses. Covariates and anxiety/depression in the third trimester were compared between participants completing all five questionnaires and participants who dropped out at any point after 30 gestational weeks.

In the third trimester of pregnancy (30 gestational weeks), women with a previous stillbirth more often experienced anxiety (22.5%) and depression (19.7%) compared with women with previous live births (4.4% and 10.3% respectively) and previously nulliparous women (5.5% and 9.9% respectively) (Table 2).

These differences remained significant in the multivariate analyses. The aOR for anxiety was 5.47 compared with the previous live birth group (95% CI 2.90–10.32, p <  0.001, adjusted for age, education, pre-pregnancy smoking and stressful life events) and 4.97 compared with the previously nulliparous group (95% CI 2.68–9.24, p <  0.001, adjusted for age and education). The aOR for depression was 1.91 compared with the previous live birth group (95% CI 1.11–3.27, p = 0.019, adjusted for age, pre-pregnancy smoking, BMI and stressful life events) and 1.91 compared with the previously nulliparous group (95% CI 1.11–3.36, p = 0.026, adjusted for age, education and BMI). There were no interactions present in any of the final models.

The prevalence of anxiety and depression decreased significantly from first assessment to 6 months postpartum among women with a previous stillbirth (p < 0.001 for anxiety and p = 0.031 for depression). By six and 18 months postpartum, respectively, the prevalence of depression and anxiety was not significantly different between groups (Table 2).

From six to 36 months postpartum, the prevalence of anxiety and depression increased significantly in the stillbirth group (p = 0.039 and 0.035 respectively) and the prevalence of anxiety, but not depression, increased significantly in the nulliparous group (p = 0.039) (Fig. 2). At 36 months postpartum, the prevalence of anxiety and depression was higher among women with a previous stillbirth compared with women with a previous live birth, but not compared with previously nulliparous women (Table 2).

The prevalence of anxiety and depression in the third trimester differed among women with a previous stillbirth completing all five questionnaires compared with drop-outs at any point after 30 gestational weeks (for anxiety 15.2 vs 32.4%, respectively, p = 0.007 and for depression 12.1 vs 29.7% respectively, p = 0.004). No such differences were observed when comparing drop-outs with responders among women with a previous live birth or previously nulliparous women.

The mean gestational age at stillbirth was 33.5 weeks (95% CI 32.5–34.6, range 20.4 to 42.6) and 115 women (68%) lost their baby at gestational age more than 30 weeks. The median number of months between stillbirth and the subsequent conception was 6 (range 1 to 183 months) and the majority of the women (n = 122, 70.5%) became pregnant within 12 months after the stillbirth. Inter-pregnancy interval shorter than 12 months between stillbirth and the next conception or gestational age at stillbirth > 30 weeks was not significantly associated with higher odds of anxiety and/or depression in the third trimester of the subsequent pregnancy (OR 1.49, 95% CI 0.70–3.16 p = 0.301 and OR 2.11, 95% CI 0.98–4.55, p = 0.056 respectively).

The frequency of relationship satisfaction among married/cohabiting women decreased slightly in all three groups from the first assessment to 36 months postpartum (p = 0.012 for women with a previous stillbirth, 0.049 for women with a previous live birth, and < 0.001 for previously nulliparous women). There was no significant difference between women with a previous stillbirth and the reference groups at any point (Table 2).

Women with a previous stillbirth had higher prevalence of anxiety and depression in the subsequent pregnancy compared with women with previous live births and previously nulliparous women. The prevalence decreased considerably after the birth of a live-born baby, and was not significantly different from the reference groups by 6 months postpartum for depression, and 18 months postpartum for anxiety. However, 36 months postpartum the prevalence of anxiety and depression had increased and was again significantly higher compared with women with a previous live birth, but not compared with previously nulliparous women. Relationship satisfaction was not significantly different between groups at any time point. Having experienced a late stillbirth (> 30 weeks) or a short interval between stillbirth and the subsequent conception (< 12 months) was not significantly related to anxiety and/or depression in the subsequent pregnancy.

Allthough symptom levels have been studied previously, to our knowledge this is the first study to estimate the prevalence of anxiety and depression contemporaneously among women pregnant after stillbirth.

We are also the first to assess relationship satisfaction in this setting. The data is derived from a large national cohort and our sample size is larger than the majority of previous studies in this field. The prospective design of the present study minimised reporting bias and enabled a long follow-up period. Previous studies have typically made comparisons solely to a control group consisting of either women with previous live births or primigravidas. Applying two reference groups to further explore the psychological impact of stillbirth makes this study unique.

The participation rate of 40.6% at first assessment is a weakness, but as expected for population-based studies [39]. A study investigating selection bias in the MoBa study found that there was an under-representation of participants with a number of exposure variables, including previous stillbirth [40]. The same study found that prevalence estimates of exposures or outcomes may be biased due to self-selection, but that self-selection is not a problem in studies of exposure-outcome associations. We therefore argue that our findings can be generalised to other women pregnant after stillbirth. However, we cannot rule out that women with greater psychological distress after a previous stillbirth more often declined participation than women coping better after the incident. Neither can we rule out that women with mild psychological distress may have been less motivated to participate. Further, the data reported was collected over a relatively long time period (from 1999 to 2012) and changes in practice and support may have influenced our findings.

Due to ethical limitations regarding linking the MoBa data to the MBRN, the study was approved only to use a limited number of reference women instead of using the entire birth cohort as a reference. However, the prevalence of anxiety and depression among the two reference groups was similar to a control group of women without epilepsy in a previous MoBa sub-study [41].

Although the dropout rate was comparable to other studies of perinatal depression [42], missing data in the follow-up period is a concern regarding the ability of this study to make conclusions about mental health outcomes from 6 months to 3 years postpartum. As anxiety and/or depression in the subsequent pregnancy after stillbirth was more prevalent among drop-outs, anxiety and depression at follow-up is probably underestimated.

Unfortunately, we do not have reliable data regarding the prevalence of anxiety and depression before the occurrence of stillbirth. It would also be interesting to compare these women to their nonpregnant counterparts in order to assess whether the prevalence of anxiety and depression are indeed associated with being pregnant.

The estimates for anxiety and depression in our study relied on self-reporting using short-form versions of validated screening tools. Even though short-form versions affect the measurement precision, it often remains sufficient for epidemiological purposes [43]. Psychiatric symptoms may be more correctly reported in an anonymous questionnaire than in a clinical interview [44] and questionnaire-based screening tools are often used to estimate the proportion at risk of having a mental disorder in a population. However, it is important to highlight that the screening tools are not suited to make formal diagnoses.

The sample size required that data on anxiety and depression were combined in the subgroup analyses on gestational age at stillbirth and inter-pregnancy interval, limiting the generalizability of these analyses.

Our findings confirm that anxiety and depression is prevalent in the pregnancy following stillbirth. Hughes et al. [8] found that, compared with primi-gravida, women who were pregnant subsequent to a stillbirth had significantly higher levels of depression and state anxiety during pregnancy, but did not differ significantly from controls in the postpartum period and 12 months postpartum. Armstrong et al. similarily reported decreased levels of depressive symptoms and anxiety three and 8 months after the birth of a subsequently healthy infant among women with a history of perinatal death [17]. This is in accordance with our findings. However, as the follow-up period in our study extends further, we found that the prevalence of anxiety and depression increased again by 36 months post-partum. This may indicate that the subsequent birth of a live-born baby is only temporarily relieving for the psychiatric morbidity associated with stillbirth. Blackmore et al. reported that depression and anxiety associated with a previous prenatal death show a persisting pattern up to 33 months after the birth of a healthy child [18]. The latter study included mainly miscarriages, and only few stillbirths, and it is not specified whether the pregnancy is directly subsequent to the loss. It is therefore not comparable to ours.

Contrary to Hughes et al. [8], we did not find that becoming pregnant within 12 months after stillbirth significantly increases the risk of anxiety and/or depression in the subsequent pregnancy [45]. One explanation for the discrepancy may be that the majority of the women in our study conceived within a year after the loss. A Swedish study demonstrated that mothers whose baby had died in utero were given different kinds of advice concerning a suitable time for a subsequent pregnancy. The best advice seems to be that the mother should wait until she, herself, feels ready [43].

In our study, early (23–30 weeks) compared with late stillbirth (> 30 weeks) was not significantly associated with anxiety and/or depression in the subsequent pregnancy. However, the p-value was just slightly above the significance level. While the duration of the pregnancy could be relevant for the risk of psychological distress after a loss [19], this is probably of diminishing importance in pregnancies that have advanced beyond 22 gestational weeks.

In accordance with findings by Rådestad et al. [23], we found that a previous stillbirth did not affect satisfaction with partner relationship. Relationship satisfaction decreased slightly in all study groups and the explanation may be that having a child is by itself associated with a decline in relationship satisfaction [37, 46, 47].