Archives of Women's Mental Health

, Volume 16, Issue 1, pp 1–10

The impact of subjective birth experiences on post-traumatic stress symptoms: a longitudinal study

Authors

    • Department of Psychosomatics and Health BehaviourNorwegian Institute of Public Health
    • Health Services Research CentreAkershus University Hospital
  • Tilmann von Soest
    • Institute of PsychologyUniversity of Oslo
  • Margarete E. Vollrath
    • Department of Psychosomatics and Health BehaviourNorwegian Institute of Public Health
    • Institute of PsychologyUniversity of Oslo
  • Malin Eberhard-Gran
    • Department of Psychosomatics and Health BehaviourNorwegian Institute of Public Health
    • Health Services Research CentreAkershus University Hospital
Original Article

DOI: 10.1007/s00737-012-0301-3

Cite this article as:
Garthus-Niegel, S., von Soest, T., Vollrath, M.E. et al. Arch Womens Ment Health (2013) 16: 1. doi:10.1007/s00737-012-0301-3

Abstract

The aim of this prospective study was to examine the etiology of post-traumatic stress symptoms following childbirth within a transactional framework of stress. Participants were women (N = 1,499) from the Akershus Birth Cohort. These women were followed from pregnancy to 8 weeks postpartum. We modeled predisposing factors (e.g., fear of childbirth) and precipitating factors (subjective and objective birth experiences) as predictors of post-traumatic stress symptoms. Post-traumatic stress symptoms were measured by means of the Impact of Event Scale, objective birth experiences by means of birth journals, and subjective birth experiences by means of three questions. A structural equation model showed that subjective birth experiences had the highest association with post-traumatic stress symptoms. Moreover, they mediated the effect of predisposing factors and objective birth experiences. The results suggest that women’s subjective birth experiences are the most important factor in the development of post-traumatic stress symptoms following childbirth.

Keywords

Post-traumatic stress symptoms following childbirthPredisposing factorsPrecipitating factorsSubjective birth experiencesAkershus Birth Cohort

Introduction

Traumatic reactions to childbirth are an important public health issue, and symptoms of post-traumatic stress are known to occur after childbirth (Olde et al. 2006a; Slade 2006). Up to one-third of women view their labor and delivery as traumatic. An estimated 2–6 % of women experience the full constellation of symptoms of post-traumatic stress disorder (PTSD) and qualify for a clinical diagnosis (Slade 2006). The Diagnostic and Statistical Manual for Mental Disorders (DSM-IV) categorizes symptoms of PTSD into three symptom clusters: reexperiencing symptoms (e.g., nightmares about the delivery, flashbacks of the event), avoidance symptoms (e.g., avoiding getting pregnant again, amnesia for the event), and increased arousal symptoms (e.g., irritability, concentration problems) (American Psychiatric 1994; Olde et al. 2006a).

Most research studies studying predictors of PTSD use retrospective designs and investigate people after the traumatic event, since traumatic stressors are rarely predictable (Ayers and Pickering 2001). Childbirth, however, is a naturally occurring and predictable event. Investigating childbirth provides an opportunity to study the development of post-traumatic stress reactions and to examine the role of different etiological variables that are involved prospectively (Ayers and Pickering 2001; Olde et al. 2006a). This study uses just such prospective data to examine predictors of post-traumatic stress symptoms following childbirth.

Although a substantial number of women experience a difficult delivery, only a fraction of them develop PTSD (Ayers et al. 2008). Therefore, research has focused on detecting women at greatest risk. Numerous predictors of post-traumatic stress symptoms following childbirth have been identified (Cohen et al. 2004; Creedy et al. 2000; Czarnocka and Slade 2000; Keogh et al. 2002; Olde et al. 2005; Söderquist et al. 2002, 2004; Soet et al. 2003; Wijma et al. 1997). However, a clear conceptual framework organizing these findings has been lacking. Recently, Slade (2006) suggested a two-dimensional conceptual framework for the development of post-traumatic stress symptoms following childbirth. One dimension relates to a time frame of both predisposing and precipitating factors. The second dimension relates to whether these factors are internal (subjective) or external (objective) or constitute interaction between the two. In this study, we apply Slade’s conceptual framework to examine potential mechanisms leading to post-traumatic stress symptoms.

Predisposing factors

Predisposing factors are often pregnancy-related or may have been present already before conception, increasing the risk of post-traumatic stress symptoms. Several predisposing factors have been suggested.

First, women that developed PTSD following a traumatic event prior to or during pregnancy are at greater risk of post-traumatic stress symptoms following childbirth (Ayers et al. 2009; Cohen et al. 2004). Most of the research studies in the field have not evaluated the presence of PTSD prior to or during pregnancy (Ayers 2004). If earlier PTSD is not taken into account, it is difficult to determine whether postpartum post-traumatic stress symptoms are birth-related or were present beforehand (Ayers et al. 2008).

Second, fear of childbirth before delivery is another important risk factor for post-traumatic symptoms after childbirth (Söderquist et al. 2002, 2004; Wijma et al. 1997). Women with fear of childbirth appraise the forthcoming delivery more negatively than others, considering it as a potentially traumatic event. Thus, their appraisal may be as important as the event itself for the development of traumatic stress reactions (Söderquist et al. 2004).

Third, a history of psychiatric problems in the woman is associated with severe traumatic stress responses (Ayers 2004). In particular, depressive symptoms during pregnancy and symptoms of anxiety appear to predict the development of post-traumatic stress symptoms after childbirth (Cohen et al. 2004; Czarnocka and Slade 2000; Keogh et al. 2002; Soet et al. 2003).

Precipitating factors

Precipitating factors act together with predisposing factors to determine if a woman develops post-traumatic stress after giving birth (Ayers 2004). Precipitating factors are birth-related and comprise both objective and subjective birth experiences. They are more proximal to the birth and may mediate the effects of the distal and static predisposing factors (Ozer et al. 2003; Slade 2006). Most studies on precipitating factors have focused on objective birth experiences, such as unplanned instrumental delivery (i.e., emergency cesarean section or instrumental vaginal delivery), long labor, extensive blood loss, or negative child outcomes. These studies found that objective birth experiences are all associated with experiencing the birth as traumatic (Ayers 2004; Cohen et al. 2004; Creedy et al. 2000; Kersting et al. 2004; Ryding et al. 1998; Söderquist et al. 2002; Soet et al. 2003; Turton et al. 2001).

The importance of subjective appraisals for the experience outcome is a cornerstone of the transactional stress theory by Lazarus and collaborators (Lazarus 1966; Lazarus and Folkman 1984). Only events appraised as being harmful or threatening to the self will cause negative emotional consequences. This emphasis on subjective experiences is reflected in the very definition of post-traumatic stress. Thus, the subjective birth experience would be central in mediating subsequent development of stress symptoms, whereas objective characteristics of the birth would be of lesser importance. Some women perceive giving birth as life-threatening and develop post-traumatic stress symptoms even if there are no serious medical complications. Other women may undergo an objectively life-threatening delivery without developing post-traumatic stress symptoms (Ayers et al. 2008). For instance, even though Söderquist et al. (2002) found an association between instrumental delivery and traumatic stress, most women in their study with a PTSD symptom profile after birth had a normal vaginal delivery (Söderquist et al. 2002). Similarly, a meta-analysis of studies on PTSD found that the subjective experience of the traumatic event was one of the most important predictors of post-traumatic stress symptoms (Ozer et al. 2003). In a related vein, in the childbirth literature, it was found that the perception of intrapartum care and perceived support from the staff or a woman’s partner predicted traumatic stress (Ayers 2004; Creedy et al. 2000).

In sum, research shows that both predisposing and precipitating factors are important in predicting which women will experience symptoms of PTSD after childbirth. Among these factors, subjective birth experiences appear to be crucial (Ayers et al. 2008). However, a complex interplay of different factors over time is likely, and no previous study has taken all of these factors into account simultaneously.

Aim

To gain a more systematic understanding of the etiology of post-traumatic stress symptoms following childbirth, we based our study on the conceptual framework suggested by Slade (2006). Consequently, we designed a large-scale, representative prospective study that followed women from pregnancy through 8 weeks postpartum. The prospective relationships between the predisposing, precipitating, and outcome variables were modeled taking into account both objective and subjective factors. Figure 1 depicts the conceptual model of the study.
https://static-content.springer.com/image/art%3A10.1007%2Fs00737-012-0301-3/MediaObjects/737_2012_301_Fig1_HTML.gif
Fig. 1

Conceptual model of the study. The different pathways represent hypotheses 1–4

Specifically, we aimed to test the following hypotheses:
  1. Hypothesis 1.

    In line with earlier research, we hypothesized that predisposing factors such as prior nonspecific PTSD, fear of childbirth, and symptoms of depression and anxiety are related to post-traumatic stress symptoms following childbirth.

     
  2. Hypothesis 2.

    We also hypothesized that precipitating factors such as subjective and objective birth experiences are related to post-traumatic stress symptoms following childbirth.

     
  3. Hypothesis 3.

    In addition, we postulated that the effects of the predisposing factors on the post-traumatic stress symptoms will be mediated by precipitating factors.

     
  4. Hypothesis 4.

    Given the importance of subjective perception, we further hypothesized that effects of the objective birth experiences on the post-traumatic stress symptoms will be mediated by the subjective experiences of the birth.

     

Material and methods

Study population

This study was part of the Akershus Birth Cohort (ABC), which targeted all women scheduled to give birth at Akershus University Hospital, which is located near Oslo, the capital of Norway, and serves approximately 350,000 people from both urban and rural areas. On average, 4,200 women give birth at the hospital’s maternity ward each year.

Recruitment was from November 2008 until April 2010. Mothers were recruited for the study when they underwent their routine fetal ultrasound examination performed around gestational week 17. Of the eligible women (able to complete a questionnaire in Norwegian), 79 % (N = 3,751) agreed to participate and returned the first questionnaire. Participants also completed questionnaires at 32 weeks gestation and 8 weeks postpartum, with response rates of 81 % (2,936 out of 3,620) and 66 % (2,217 out of 3,380), respectively. The number of eligible women dropped somewhat at 32 weeks gestation and 8 weeks postpartum, because some women had moved or had experienced perinatal death of their infant or other complications. For this study, we used information from all three questionnaires and from the hospital’s birth journal. The birth journal is completed by the hospital staff and contains sociodemographic and medical information about the mother, child, pregnancy, and birth. We chose to exclude women who underwent an elective cesarean section (6.1 %) from the analyses, as they present a very selected group with regard to both their subjective and objective birth experiences. The ABC study obtained ethical approval from the Regional Committees for Medical and Health Research Ethics, and all participants gave a written informed consent.

Measures

Predisposing factors were measured during pregnancy, at gestation week 17 (symptoms of PTSD), and gestation week 32 (fear of childbirth, symptoms of depression, and symptoms of anxiety). Information on objective birth experiences was obtained during and right after birth. Post-traumatic stress symptoms following childbirth as well as subjective birth experiences were measured at 8 weeks postpartum.

Post-traumatic stress symptoms following childbirth

The Impact of Event Scale was used to measure post-traumatic stress symptoms (Horowitz et al. 1979). The Impact of Event Scale is a self-rating scale that measures symptoms of intrusion (seven items) and avoidance (eight items). The scale has four response categories (0 = not at all, 1 = rarely, 3 = sometimes, and 5 = often). Higher scores reflect a higher degree of post-traumatic stress. In our multivariate analyses, we conducted a confirmatory factor analysis instead of using sum scores. In these analyses, items were handled as ordered categorical variables to reflect uneven intervals between the response options of the items (see “Statistical analyses”).

Precipitating factors

Three questions were used to form a latent factor for subjective birth experiences: (1) “How frightened were you during the birth?” The response scale was numeric, ranging from a minimum score of 0 (“not frightened at all”) to a maximum score of 10 (“extremely frightened”). (2) “What was your overall experience of the birth?” This also was scored on a scale ranging from 0 (“very good”) to 10 (“extremely bad”). (3) “To what degree did you feel taken care of during the birth?” The response scale ranged from a minimum score of 1 (“very good”) to a maximum score of 4 (“very bad”).

Using information on labor complications registered in the electronic birth journal, we constructed an index of negative objective birth experiences. Each potential complication was treated as a dichotomous variable, depending on whether it had occurred or not. In the next step, we created a sum score of the number of complications, resulting in an index ranging from 0 (no labor complication) to 11 (presence of 11 labor complications). The 11 complications were the following: (1) unplanned instrumental delivery (emergency cesarean section or instrumental vaginal delivery), (2) placental abruption, (3) shoulder dystocia, (4) eclampsia during labor, (5) maternal infection during labor, (6) active phase of labor lasted longer than 12 h, (7) severe vaginal tears (degrees 3 and 4), (8) extensive blood loss (≥1,000 ml), (9) umbilical cord complications (umbilical cord prolapse, nuchal cord compressing the child’s neck, short cord, other complications not further specified), (10) intrapartum asphyxia, and (11) low neonate Apgar score at 5 min (<7, not due to intrapartum asphyxia).

Predisposing factors

The women in our study reported whether they at any time in their life had been involved in or had experienced a dramatic and terrifying event. If this was the case, they reported whether they had suffered from eight potential symptoms related to that event during the last month. The symptoms were based on the questions regarding PTSD included in the Mini-International Neuropsychiatric Interview (M.I.N.I.), which is designed for epidemiological studies and clinical trials. The M.I.N.I is a short structured clinical interview which enables researchers to make diagnoses of psychiatric disorders according to DSM-IV or ICD-10 (Sheehan et al. 1998). The symptoms we measured were “During the last month I… (1) “reexperienced the event (e.g., in dreams, nightmares, intense memories, or flashbacks),” (2) “avoided thinking or talking about the event,” (3) “had problems remembering the event,” (4) “felt distant,” (5) “had problems sleeping,” (6) “had problems concentrating,” (7) “have been nervous,” and (8) “ have been considerably disturbed by the event in my work and in social activities.” Depending on whether the symptom was present, a score was given or not. This resulted in a symptom score ranging from 0 (no symptoms) to 8 (maximum number of symptoms).

Fear of childbirth was assessed by the Wijma Delivery Expectancy/Experience Questionnaire (W-DEQ, version A) (Wijma et al. 1998). This is the most frequently used instrument to measure fear of childbirth. The W-DEQ, version A, measures fear of childbirth as operationalized by the cognitive appraisal of the approaching delivery. The 33-item rating scale has six response categories ranging from 0 to 5. Sum scores can have a minimum score of 0 and a maximum of 165, with higher scores reflecting a greater degree of fear of childbirth.

Symptoms of depression during the past week were measured by the Edinburgh Postnatal Depression Scale (Cox et al. 1987). The Edinburgh Postnatal Depression Scale is a ten-item self-rating scale designed to identify postnatal depression. The scale has four response categories ranging from 0 to 3; thus, sum scores can range from 0 to 30. Higher scores reflect higher levels of depression. The scale has also been validated for use in pregnancy as well as with nonpostnatal mothers (Eberhard-Gran et al. 2001; Murray and Cox 1990).

Ten items from the Hopkins Symptom Check List were used to evaluate for anxiety symptoms during the previous week. The Hopkins Symptom Check List is a widely used self-rating scale, and the first ten items comprise the anxiety score (SCL-anxiety). The scale has four response categories ranging from 1 to 4. Consequently, sum scores range from 10 to 40, with higher scores indicating higher levels of anxiety (Eberhard-Gran et al. 2003; Nettelbladt et al. 1993).

Statistical analyses

Analyses were conducted in the framework of structural equation modeling (Bollen 1989), and we used the statistical program Mplus 6 for all analyses. A robust weighted least squares estimator (WLSMV) was employed, because some of the items included in the analyses were considered to be ordered categorical variables (Muthén and Muthén 2010).

The percentage of missing data for the variables included in the study was 3.5 % for post-traumatic stress symptoms following childbirth, 1.7 % for symptoms of depression, 1.3 % for symptoms of anxiety, and 1.8 % for earlier PTSD symptoms. Moreover, fear of childbirth and objective birth experiences had a somewhat higher percentage of missing data, with 12.8 and 14.9 %, respectively. Missing data were accounted for by the missing routines for WLSMV in Mplus, which are based on pairwise present analysis (Muthén and Muthén 2010). Our final sample consisted of 1,499 women.

First, we conducted a confirmatory factor analysis of the Impact of Event Scale, constructing a latent factor for the post-traumatic symptoms. We also conducted a confirmatory factor analysis for subjective birth experiences, constructing a latent factor. As conducting confirmatory factor analyses on all scales would have led to an excessively complex model, we chose to construct latent variables only for the most important psychological variables, i.e., the mediators and the outcome. The remaining variables were treated as manifest variables. Despite of representing a mediator, objective birth experiences were as well treated as a manifest variable, since this variable does not represent a psychological variable which corresponds to an underlying latent construct. Then, we conducted correlation analyses to study the bivariate associations. We further ran a series of regression analyses, regressing the trauma factor on the precipitating and predisposing factors. Finally, we estimated the entire mediation model, including the estimation of direct and indirect effects. Tests of mediation were conducted examining whether the indirect effect involving the putative mediator was statistically significant or not (Hayes 2009).

Results

Descriptive findings

Mean age at birth was 30.7 years (standard deviation (SD) 4.9 years, range 17.2–45.5 years). Approximately half (49.8 %) of the women were first-time mothers, and the remainder (50.2 %) had one or more previous deliveries. Sixty-five percent of women experienced no labor complications during their delivery, 25 % experienced one labor complication, and 10 % experienced two or more labor complications. The proportion of each labor complication was comparable with the data reported in the Medical Birth Registry of Norway. However, the participants were less often smokers, and there were fewer younger women and fewer single women compared to the national data from the Medical Birth Registry of Norway from 2008 (Norwegian Institute of Public Health website 2012). In line with earlier research (Slade 2006), we found that 1.9 % of the participants had a high risk for suffering from PTSD, i.e., scored above 34 on the Impact of Event Scale (Neal et al. 1994).

Table 1 shows means and SDs for all variables. For the descriptive analyses, we kept the suggested weighting of the items of the Impact of Event Scale and coded them 0, 1, 3, or 5 (Horowitz et al. 1979). We also report the means and SDs of the three items assessing subjective birth experiences separately, as they do not share the same metric.
Table 1

Mean scores and standard deviations of all variables

Measures

Mean (SD)

Post-traumatic stress symptoms

Impact of Event Scale, total score

6.86 (8.06)

Impact of Event Scale, intrusion score

4.38 (4.90)

Impact of Event Scale, avoidance score

2.55 (4.04)

Precipitating factors

Subjective birth experiences

 

 “How frightened”

3.03 (3.63)

 “Overall experience”

2.93 (2.72)

 “Taken care of”

1.37 (0.66)

Objective birth experiences

0.49 (0.75)

Predisposing factors

Symptoms of PTSD

0.28 (0.79)

Fear of childbirth

57.2 (19.95)

Symptoms of depression

5.05 (4.23)

Symptoms of anxiety

12.89 (3.17)

SD standard deviation, PTSD post-traumatic stress disorder

On average, participants scored relatively low on postpartum post-traumatic stress symptoms following childbirth. The mean intrusion score was higher than the avoidance score, even though avoidance symptoms were represented by one more item than the intrusion symptoms were.

The subjective birth experiences scores tended to be low. On average, the women were not very frightened during birth, rated their birth as a good overall experience, and felt well taken care of. Likewise, prepartum symptom levels of PTSD, depression, and anxiety were relatively low. The symptom level of fear of childbirth was somewhat higher. This is in accordance with earlier research, as fear of childbirth is normative to a certain degree and expected to be more or less normally distributed (Salomonsson et al. 2010).

Bivariate results

The confirmatory factor analysis of the Impact of Event Scale resulted in a latent factor with a good fit (root mean square error of approximation (RMSEA) = .06, comparative fit index (CFI) = .95, Tucker–Lewis Index (TLI) = .95). The factor for subjective birth experiences included only three indicators. Consequently, the factor model was just identified and yielded a perfect fit (RMSEA = .00, CFI = 1.00, TLI = 1.00). Even though the fit indices in this case do not provide information about the adequacy of the items to measure subjective birth experiences, such a latent factor was constructed to remove measurement error from the multivariate model presented below.

Hypotheses 1 and 2

Table 2 shows bivariate correlations with post-traumatic stress symptoms and subjective and objective birth experiences. As expected, all predisposing and precipitating factors significantly predicted post-traumatic stress symptoms. Not only subjective birth experiences were particularly highly related to post-traumatic stress symptoms but also fear of childbirth and symptoms of depression and anxiety were relatively highly correlated with the outcome.
Table 2

Bivariate correlations with post-traumatic stress symptoms, subjective birth experiences, and objective birth experiences

 

Post-traumatic stress symptoms

Subjective birth experiences

Objective birth experiences

Post-traumatic stress symptoms

1.00

  

Subjective birth experiences

.65*

1.00

 

Objective birth experiences

.22*

.26*

1.00

Symptoms of PTSD

.19*

.13*

.00 n.s.

Fear of childbirth

.37*

.43*

.07*

Symptoms of depression

.40*

.31*

.01 n.s.

Symptoms of anxiety

.35*

.28*

.00 n.s.

n.s. not statistically significant

*p ≤ 0.001

Fear of childbirth was, in turn, relatively highly related to subjective birth experiences. Objective birth experiences, on the other hand, were not significantly related to any of the predisposing factors except fear of childbirth. Yet this was a rather low association.

Multivariate results

A structural equation model was constructed to test the mediation hypotheses stated in “Introduction” (hypotheses 3 and 4). Figure 2 shows the results of the full mediation model. Only the statistically significant pathways are depicted. The multiple mediation model showed an acceptable fit (RMSEA = .04, CFI = .91, TLI = .90) and was therefore deemed to be an appropriate model.
https://static-content.springer.com/image/art%3A10.1007%2Fs00737-012-0301-3/MediaObjects/737_2012_301_Fig2_HTML.gif
Fig. 2

Multivariate results. Only the statistically significant pathways are shown

Hypothesis 3

Confirming our hypothesis, including subjective and objective birth experiences in the mediation model reduced the significant relationship between fear of childbirth and post-traumatic stress symptoms following childbirth to nonsignificance. These findings indicate that the effect of fear of childbirth on post-traumatic stress symptoms was fully mediated by the precipitating factors. Moreover, the analyses showed that subjective birth experiences explained most of the relationship between fear of childbirth and post-traumatic stress symptoms (standardized indirect effect through subjective birth experiences = .19, p < 0.001), whereas objective birth experiences accounted for only a small part of the mediated effect (standardized indirect effect through objective birth experiences = .01, p < 0.05).

Furthermore, subjective birth experiences functioned as a partial mediator for the relationship of symptoms of depression and anxiety on the one side and post-traumatic stress symptoms on the other (standardized indirect effect through subjective birth experiences = .05, p = 0.01 and .06, p = 0.001, respectively). Objective birth experiences could not function as a mediator for symptoms of depression and anxiety, as they did not correlate with them in the first place (see Table 2) (Baron and Kenny 1986).

Contrary to our hypothesis, the relationship between earlier symptoms of PTSD and post-traumatic stress symptoms following childbirth was not mediated by either subjective or objective birth experiences, since earlier symptoms of PTSD were not related to either subjective or objective birth experiences.

Hypothesis 4

In line with our hypothesis, the effects of objective birth experiences on post-traumatic stress symptoms were partially mediated by the subjective experiences (standardized indirect effect through subjective birth experiences = .14, p < 0.001). That is to say, the relationship between objective birth experiences and post-traumatic stress symptoms following childbirth remained significant. Nonetheless, a major portion of the effect of the objective birth experiences on post-traumatic stress symptoms was accounted for by the effect of the subjective birth experiences.

Discussion

The aim of the study was to gain a more systematic understanding of the etiology of post-traumatic stress symptoms following childbirth. The key findings of this study are as follows: First, post-traumatic stress symptoms following childbirth were related to all predisposing and precipitating factors. However, they were correlated by far the highest with subjective birth experiences. Second, subjective birth experiences not only were strongly related to post-traumatic stress symptoms directly but also mediated much of the effect of the other factors on post-traumatic stress symptoms. Earlier PTSD symptoms were the only factor that was not mediated by subjective birth experiences.

In order to understand variations among individuals under comparable stress conditions, we must take into account the subjective appraisal that mediates between the stress experience and the subsequent emotional reaction (Lazarus and Folkman 1984). Applied in our context, this means that whether or not a woman develops post-traumatic stress symptoms following childbirth is strongly influenced by her subjective birth experience.

Given this background, our results support earlier studies underlining the particular importance of subjective experience regarding the traumatic event (Ayers 2004; Ayers et al. 2008; Creedy et al. 2000; Ozer et al. 2003; Slade 2006), as subjective experience is both an independent contributor to and a mediator for developing post-traumatic stress symptoms. The finding that the association between fear of childbirth and post-traumatic stress symptoms was to a major part mediated by the precipitating factors can thus explain why fear of childbirth results in trauma responses. In addition, women who reported fear of childbirth at gestation week 32 were also more likely to have objective birth experiences that in turn influenced negative subjective birth experiences. Together, objective and subjective birth experiences fully mediated the association between fear of childbirth and post-traumatic stress symptoms.

There is one previous study showing that fear of childbirth predicts negative obstetric outcomes (Andersson et al. 2004). Yet the reverse causal direction is conceivable, too. Women who have experienced greater birth complications in previous pregnancies may have greater fear of childbirth as a consequence. Nonetheless, the major part of the effect was accounted for by the subjective, and not the objective birth experiences. Also, the bivariate association between fear of childbirth and objective birth experiences was rather low (Table 2).

The other predisposing factors maintained direct effects on post-traumatic stress symptoms following childbirth. However, some of the effects of both symptoms of depression and anxiety on post-traumatic stress symptoms were accounted for by the subjective birth experiences. This implies that symptoms of depression and anxiety color the subjective birth experiences and thereby predict post-traumatic stress symptoms. Still, they have an independent detrimental effect, albeit a relatively small one.

Earlier PTSD symptoms constituted an independent risk factor for the development of post-traumatic stress symptoms following childbirth. However, the effect size was small. In line with earlier findings, new post-traumatic stress symptoms may emerge postnatally, even if earlier PTSD symptoms are taken into account (Ayers and Pickering 2001). Post-traumatic stress symptoms following childbirth seem therefore to be primarily due to other factors. On the other hand, earlier PTSD symptoms were measured more distally (at gestation week 17) than for example symptoms of depression and anxiety (at gestation week 32) which also may have contributed to a relatively low association.

Subjective birth experiences also explained in part how objective birth experiences (i.e., medical complications) lead to trauma responses. A number of studies found that objective birth experiences and complications can predict post-traumatic stress symptoms (Ayers 2004; Cohen et al. 2004; Creedy et al. 2000; Kersting et al. 2004; Ryding et al. 1998; Söderquist et al. 2002; Soet et al. 2003; Turton et al. 2001). We confirmed this. However, when the effect of subjective birth experiences is taken into account, the situation is apparently different. As described in other studies, even though objective birth experiences do contribute to a potential trauma, the subjective appraisal of the delivery is much more important. It is, of course, influenced by the objective birth experiences (Ayers 2004; Ayers et al. 2008; Creedy et al. 2000; Ozer et al. 2003; Slade 2006).

Strengths and limitations

This study fills an important gap in the literature. To our knowledge, this is the largest population study investigating the etiology of post-traumatic stress symptoms following childbirth employing a prospective design. Previous studies have often been confined in examining singular risk factors of post-traumatic stress (Slade 2006). In contrast, using sophisticated methods and having enough power, we tested a more comprehensive model. Earlier studies have underlined the likely importance of subjective birth experiences and called for more research in this field (Ayers et al. 2008). However, the main focus often remains on the role of obstetric variables or the predisposing factors. Our results contribute to a more holistic conceptualization of the development of post-traumatic stress symptoms following childbirth by elucidating some of the mechanisms and highlighting the key role of subjective birth experiences.

Readers should also note some limitations to our findings. First, currently we do not have data on maintaining factors of post-traumatic stress symptoms following childbirth. Our model would have been even more complete if we would have been able to look at postnatal aspects which may influence the development and duration of symptoms (Slade 2006). However, we are in the process of collecting these data and are planning to examine this issue in the future.

Second, the Impact of Event Scale does not contain items tapping the third PTSD symptom cluster “hyperarousal,” which has led some researchers to criticize its content validity as a measure of PTSD (Joseph 2000). However, childbirth differs from other traumatic events in many ways and involves large physiological changes (Ayers et al. 2009). Symptoms of arousal could be affected by normal physiological changes and fatigue after birth. Thus, hyperarousal must not necessarily be indicative of a traumatic birth. It is possible that heightened arousal is a common and potentially adaptive adjustment following the birth of a helpless and dependent infant (Ayers et al. 2009; Slade 2006). In fact, hyperarousal is the dimension of PTSD that typically features with the highest frequency among women who have just given birth (Slade 2006). Moreover, one study showed that many women who showed significant hyperarousal had not rated their births as traumatic (Soet et al. 2003). What is more, in a recent study, the Impact of Event Scale was compared with the Impact of Event Scale-Revised which also covers symptoms of hyperarousal (Olde et al. 2006b). According to the study, adding hyperarousal symptoms did not improve the scale’s quality but rather decreased its psychometric properties. Consequently, the authors concluded that the 15-item Impact of Event Scale was still the most suitable instrument for assessing traumatic stress responses following childbirth (Olde et al. 2006b).

Third, subjective birth experiences were measured retrospectively, at 8 weeks postpartum. Ideally, the subjective birth experiences should be measured as soon as possible after delivery. However, it is often not feasible and appropriate to distribute questionnaires to mothers immediately after delivery. Measuring subjective birth experiences and post-traumatic stress symptoms simultaneously may obscure the causal direction between these two constructs. Emerging post-traumatic stress symptoms could influence subsequent memories and thus color ratings of subjective birth experiences. However, simultaneous measurement has been used in a previous study (Ford et al. 2010). This seems reasonable, as the transactional stress model assumes that the cognitive appraisal influences the stress response and not vice versa (Lazarus and Folkman 1984).

Fourth, no established and validated instrument such as the Perceptions of Labour and Delivery Scale (Bailham et al. 2004) or the Childbirth Perception Questionnaire (Padawer et al. 1988) was used in the present study to measure subjective birth experiences. Instead, we formed our own latent factor based on three questions. This factor functioned well and had a high predictive value. Nevertheless, the measurement of subjective birth experiences can certainly be improved. Using a more comprehensive and validated measurement instrument might provide further insights into the role of subjective birth experiences for the development of post-traumatic stress symptoms. In a similar vein, even though our measure for earlier PTSD symptoms is based on the established and validated M.I.N.I. (Sheehan et al. 1998), most but not the entire number of symptoms of PTSD were included.

Finally, generalizability of the results is limited by the fact that only Norwegian-speaking women were included, resulting in a relatively homogeneous, mainly Caucasian sample. Different results might be obtained for other ethnic groups. Also, with regard to labor complications, our study population seems to be representative of the total population of women in Norway. However, with regard to sociodemographic characteristics, there is a reason to believe that the women in the study were somewhat more resourceful than the general birthing population in Norway.

Clinical implications

Our results emphasize the need to focus on the subjective experience of the birth to prevent initial traumatization. Recognizing the importance of the subjective birth experiences opens up important opportunities for prevention. Women with symptoms of depression and anxiety and women who fear giving birth are at increased risk of developing post-traumatic stress symptoms (Cohen et al. 2004; Czarnocka and Slade 2000; Keogh et al. 2002; Söderquist et al. 2002, 2004; Soet et al. 2003; Wijma et al. 1997). Our results indicate that subjective birth experiences play an important role in that process. Therefore, these women ought to be handled with special care and made to feel safe during birth to avoid traumatization. Negative objective birth experiences are also a risk factor (Ayers 2004; Cohen et al. 2004; Creedy et al. 2000; Kersting et al. 2004; Ryding et al. 1998; Söderquist et al. 2002; Soet et al. 2003; Turton et al. 2001). However, medical complications cannot always be avoided. Thus, if they occur, it is important to make women feel safe and well taken care of.

Prevention of post-traumatic stress symptoms following childbirth is of great importance, as consequences for the partnership relationship can be potentially detrimental (Ayers et al. 2006). It has also been suggested that childbirth-related post-traumatic stress symptoms may adversely influence the mother–child relationship and maternal perception of the child (Ayers et al. 2006; Creedy et al. 2000; Slade 2006). Recent studies however indicate that to some degree, this association is rather due to comorbid depression symptoms (Davies et al. 2008; McDonald et al. 2011). Moreover, traumatized women have a higher likelihood of not having future children or delaying subsequent pregnancy (Czarnocka and Slade 2000; Gottvall and Waldenstrom 2002; Slade 2006).

This large-scale population study gives valuable insights into possible modes of prevention. Also, on a more general note, childbirth offers a special research opportunity for researchers to study the development of post-traumatic stress reactions prospectively (Ayers and Pickering 2001; Olde et al. 2006a). In this way, our prospective study contributes to the general PTSD literature as well, highlighting the importance of the subjective experience of a traumatic event.

Conclusions

In conclusion, the results of our study indicate that the most important factor in the development of post-traumatic stress symptoms following childbirth is women’s subjective birth experiences. Subjective birth experiences are both an independent contributor to and a mediator for developing post-traumatic stress symptoms. Being aware of the essential role of subjective birth experiences offers a unique opportunity to prevent traumatization.

Future studies should further explore the role of maintaining factors of post-traumatic stress symptoms that are part of the framework suggested by Slade (2006). For this reason, we plan to collect follow-up data from this research project that will make it possible to examine this issue in the future.

Copyright information

© Springer-Verlag 2012