Introduction

Top Abstract Introduction Participants and methods Results Results Discussion References

In England and Wales 0.5% of pregnancies end in stillbirth after 24 weeks' gestation.¹ Parents inevitably suffer a process of grief and mourning which may last a year or more.² There has been debate about when women should conceive again, with some clinicians arguing that they need time to recover emotionally before embarking on another pregnancy ^{3 4} but with little systematic evidence to support this view. Studies consistently report that about 50% of women become pregnant within 12 months after loss.² There are descriptive reports of the psychological difficulties of pregnancy after stillbirth, ^{5 6} but the limited research is from small numbers⁷ or specialised groups.⁸ Recent findings that maternal anxiety and depression during and after pregnancy have deleterious effects on mother and infant reinforce the clinical relevance of the issue.

Maternal anxiety in pregnancy is associated with earlier births and lower birthweights ^{9 10} and impairment of fetal brain development.¹¹ Mediating mechanisms may include abnormal uterine blood flow¹² and increased cortisol transfer from mother to fetus.¹³ Antenatal depression is associated with poor clinic attendance and poor self care,¹⁴ and postnatal depression is associated with infant cognitive delay and emotional problems,¹⁵ which may have long term effects.¹⁶

We followed 60 pregnant women who had had stillbirth or late miscarriage and 60 matched control women and evaluated depression and anxiety and the relation between time since loss and symptom levels. There were four assessments between the third trimester and 12 months post partum.

	Participants and methods

Top Abstract Introduction Participants and methods Results Results Discussion References

Participants
Criteria for inclusion were women whose previous pregnancy ended in loss after 18 weeks' gestation, had no live children, were aged over 20 years, had a partner, had a singleton pregnancy, spoke English, and were progressing satisfactorily in the present pregnancy. We excluded women receiving treatment for acute physical or mental illness and those with previous termination for abnormality.

Instruments
We used four measures to assess depression and anxiety: an observer rated demographic questionnaire; the Edinburgh postnatal depression scale (a 10 item self report scale developed for postnatal use¹⁷ and now validated for use during pregnancy¹⁸; for dichotomous analysis the 14/15 cut off was used antenatally (as recommended by Murray and Cox because of the high levels of dysphoria in pregnancy¹⁸) and for measurement at 6 weeks the 12/13 cut off was used); the Beck depression inventory (10 item self report questionnaire)¹⁹; and the Spielberger state-trait inventory (a 40 item questionnaire measuring anxiety at time of testing (state) and general tendency to anxiety (trait)).²⁰

Procedure
About 30 000 sets of case notes were screened at three district general hospitals. Controls were simultaneously identified in the same antenatal clinics by using casenotes to find primigravida matched on ethnicity, age, and socioeconomic status. A letter was sent to all identified women in the third trimester of the current pregnancy inviting them to take part. Four assessments were done: during the third trimester and at 6 weeks, 26 weeks, and 12 months after the birth. Assessment included a demographic questionnaire, Edinburgh postnatal depression scale (antenatal, 6 weeks), Beck depression inventory (6, 12 months), and Spielberger state-trait inventory (third trimester, 12 months). Interviews were carried out at the participant's home or in the outpatient departments. The procedure had approval from all local ethics committees.

Statistical analyses
Groups were compared at the four time points by using paired tests on all measures. Data for depression scores were analysed with both continuous scores and dichotomous scores with a cut off 14/15 (antenatal) and 12/13 (6 weeks postnatal) for the Edinburgh scale and 10/11 for Beck's inventory. Continuous data deviated from Gaussian distribution in one group or the other on skewness, kurtosis, or Komolgorov-Smirnov test for all variables except Spielberger trait anxiety at baseline. Hence non-parametric Wilcoxon tests are reported for paired comparisons. All analyses are reported for pairs with complete data on the dependent variable, hence n varies. Analyses to compare time since loss were conducted for gaps of 9, 12, 15, and 18 months between loss and conception. Parametric 95% confidence intervals are reported for all comparisons. Comparisons of prevalences for dichotomised variables are expressed as 95% confidence intervals for relative risk.

	Results

Top Abstract Introduction Participants and methods Results Results Discussion References

Participants included and lost
Of 96 subjects who met the inclusion criteria, 82 (86%) agreed to participate. Thirteen (14%) delivered before the first interview, and 69 (72%) had a third trimester interview. Four were excluded after the birth because of illness or injury; data from five who were interviewed were unmatched or very incomplete. Seven dropped out later (three moved abroad, four refused to continue).

Characteristics of sample
The mean (range) age of participants was 30 (20-46) years for women who had experienced stillbirth and 29 (20-43) years for control women. There were 39 pairs of white women, seven pairs of Afro-Caribbean women, nine pairs of Indian or Pakistani women, four pairs of African women, and one pair of Chinese women. There was no significant difference between groups on educational level or material situation. Four subjects and two control women reported previous physical illness and four subjects and six control women previous mental illness; none had been admitted to hospital. No women reported that they had received psychiatric treatment after the stillbirth. Six women reported that the stillborn infant had a congenital abnormality; the remainder had been told the infant was normal. Ten subjects who had undergone previous termination of pregnancy showed no significant differences from other subjects on baseline variables. Thirty one women (52%) conceived less than 12 (range 2-11; median 5) months after loss and 29 (48%) more than 12 (12-180; 23) months after loss.

	Results

Top Abstract Introduction Participants and methods Results Results Discussion References

Third trimester
Stillbirth and control groups differed significantly in depression scores (mean for stillbirth group 10.8 v 8.2 in control group; Wilcoxon test P=0.004; 95% confidence interval for difference 0.7 to 4.5). With 14/15 cut off for the Edinburgh scale¹⁸ 17 (28%) of the stillbirth group and 5 (8%) of the control group scored high (McNemar P=0.01; 95% confidence interval for relative risk 1.3 to 16), with a significant difference on state anxiety (39.8 v 32.8; Wilcoxon P=0.003; 95% confidence interval for difference 2.6 to 11.2) but not trait anxiety (38.4 v 35.8; Wilcoxon P=0.11; 1.1 to 6.3).

Further assessments
At 6 weeks there was no significant difference between groups on Edinburgh depression scores (7.3 v 7.0; Wilcoxon P=0.72; 1.4 to 1.9). With 12/13 cut off, six (10%) in the stillbirth group and three (5%) in the control group scored high (McNemar P=0.51; 95% confidence interval for relative risk 0.431 to 12.4). At 26 weeks there was no significant difference between groups on depression scores (Beck 5.9 v 6.3; Wilcoxon P=0.59; 95% confidence interval for difference 2.8 to 2.0). With 10/11 cut off, four (7%) in the stillbirth group and seven (12 %) in the control group scored high (McNemar P=0.32; 95% confidence interval for relative risk 0.4 to 12.4). There was no significant difference between anxiety and depressive symptoms at 1 year (Beck 6.0 v 5.1; Wilcoxon P=0.44; 95% confidence interval for difference 1.1 to 2.9; Spielberger state 31.7 v 32.5; P=0.57; 5.1 to 3.6; trait 34.8 v 36.4; P=0.46; 6.0 to 2.7). Dichotomy of the Beck score at 10/11 showed a trend towards higher symptoms in the index group (11 (19%) v 5 (8%); McNemar P=0.39; 95% confidence interval for relative risk 0.1 to 1.9).

Relation between time from loss to conception and symptom levels
The table shows the results of the investigation of the realtions with delay to conception. Depression and anxiety in women waiting 12 months before next conception were similar to those in the control women at all points, with less trait anxiety (95% confidence interval for difference 0.5 to 13.0; P=0.04) 1 year after birth, while women who next conceived less than 12 months after loss had more symptoms than their control women at all points. Differences between those waiting 12 months before conception and those who conceived sooner were significant in the third trimester for state anxiety (P=0.02). At 1 year there were significant differences in state and trait anxiety (both P=0.02) and depression (P=0.01; table).

Antenatal scores as a predictor of later depression
Scores on the Edinburgh postnatal depression scale during the third trimester strongly predicted 12 month follow up scores on Beck's depression inventory (Spearman r=0.37; 0.22 to 0.54; P<0.0005; n=109). Correlation was notably higher in subjects (r=0.5; 0.27 to 0.68; P<0.0005; n=55) than controls (r=0.23; 0.08 to 0.47; P=0.10; n=53).

	Discussion

Top Abstract Introduction Participants and methods Results Results Discussion References

We have shown that women whose previous pregnancy ended in stillbirth have significantly higher levels of depression and state anxiety in the third trimester of the next pregnancy than control women, are no different from controls at 6 and 26 weeks post partum, and show a trend towards greater depression 1 year after the next birth. When we included time since loss in the analysis, the psychopathology was found to be among women who conceived in the 12 months after stillbirth whereas women who had a gap of 12 or more months between loss and conception were not different from their controls at any point.

One explanation is that women may need a year to recover at least partially from the bereavement. Women who lost a child less than 18 months earlier (that is, before the third trimester assessment) are understandably still grieving. Although this explains the higher symptoms in the third trimester, however, it does not entirely explain the pattern of recovery after the next birth and later increase in depression and anxiety. In the women who conceived after a longer time since loss, those who were 18 or more months after loss when assessed in the third trimester were no more depressed or anxious than control women; yet in women who conceived quickly, 12 months after the birththat is, 23 to 32 months after lossthey were more depressed and anxious than controls. This gives some support to Lewis's contention that pregnancy interferes with the normal process of mourning.³ The findings of this study suggest that depression and anxiety may be prolonged for some women who become pregnant within 12 months after stillbirth and at a time when they would most wish to be well.

An alternative explanation is that choosing to become pregnant relatively soon after the loss is partly determined by personality. The lower trait anxiety among slower to conceive women 1 year after the next birth offers some support for this view. It suggests either that those who do not conceive quickly are less anxious than those who conceive sooner or that the longer process of mourning has left mothers who wait less anxious than before.

Women mentioned various determinants in their timing of the next pregnancy, including the longing of one or both parents to replace the lost child as a way of reducing distress, advice from relatives or health professionals, maternal age, and other social factors. Some mothers did not get pregnant soon after loss because the first pregnancy had been unwanted or because they no longer had a partner or because they thought that they needed time to recover physically and psychologically from the trauma. Few current pregnancies were unplanned, and those few were failure to use contraception rather than failure of contraception.

Advice to parents
The decision to have another child after late loss in pregnancy is a personal one, but parents often seek professional advice about timing the next pregnancy, although they resent being told categorically when to go ahead.²¹ We have shown that conception within a year after stillbirth was associated with higher levels of depression and anxiety in pregnancy compared with conception later. Most women recovered after the birth, but those who had conceived more quickly were again at higher risk of depression and anxiety when their baby was a year old. Personality could be the common determinant both of conceiving sooner and of higher psychopathology, but the need for adequate time to grieve is a common sense explanation and likely contributor.