The psychological consequences of childbirth

Abstract

It is becoming increasingly recognised that some women develop post traumatic stress disorder (PTSD) after childbirth. This study aimed to determine whether women experienced symptoms of PTSD and depression at 6–12 months post‐partum; and what factors predict the development of psychopathology. This was a retrospective postal study of women who have given birth in the previous 6–12 months. A total of 102 women who delivered in hospital completed measures of PTSD, depression, perceptions of labour and delivery and provided clinical and demographic information at 6–12 months post‐partum (mean = 9.5 months). A total of 3.9% showed questionnaire responses suggesting clinically significant levels of PTSD. A further 19.6% women reported sub‐clinical symptoms. Regression analysis showed that higher depression scores, fear for the baby and unexpectedness of procedures during labour predicted higher scores on the PTSD measure. A total of 21.5% of women were depressed. A history of mental health problems and the presence of PTSD symptomatology predicted higher depression scores. This study demonstrates that a significant number of women continue to experience some level of PTSD and depression at 9.5 months post‐partum. It is important to be aware that births involving unexpected obstetric procedures may precipitate PTSD symptoms in some women.

Keywords:

• post traumatic stress disorder
• childbirth
• depression

Introduction

In recent years, it has become increasingly recognised that a proportion of women can develop post traumatic stress disorder (PTSD) following Ayers & Pickering, 2001; Bailham & Joseph, 2003; Beck, 2004; Czarnocka & Slade, 2000; Lyons, 1998; childbirth (Wimja, Soderquist & Wimja, 1997). PTSD is characterised by persistent re‐experiencing of the event, avoidance of stimuli associated with the traumatic event, numbing of general responsiveness and symptoms of increased arousal (American Psychiatric Association, 1994). This can occur following exposure to an event which is perceived to be traumatic. Prior to the changes in the Diagnostic and Statistical Manual for Mental Disorders in 1994 (DSM; American Psychiatric Association, 1994), a traumatic event was defined as ‘something that would evoke symptoms of distress in most people and was outside the range of usual human experience’. Given this, it was unlikely that childbirth would have been recognised by clinicians or researchers as a sufficiently ‘traumatic event’ to fulfil diagnostic criteria. It was, therefore, less likely that clinicians would attribute postnatal psychological distress to PTSD. The DSM‐IV criteria (American Psychiatric Association, 1994) redefined the traumatic event as ‘an experience in which the person witnessed or confronted serious physical threat or injury to themselves or others, and in which the person responded with feelings of fear, helplessness or horror’. The broadening of the diagnostic criteria allowed clinicians and researchers to consider the possibility that PTSD can occur as a consequence of childbirth, and a number of published studies subsequently appeared in the literature (Ayers & Pickering, 2001; Bailham & Joseph, 2003; Beck, 2004; Czarnocka & Slade, 2000; Lyons, 1998; Wimja et al., 1997).

Current studies estimate that approximately 1.7–5.6% of women have PTSD at around 6–8 weeks after birth, and that this figure is somewhat less at 6 months (Ayers & Pickering, 2001; Bailham & Joseph, 2003; Wimja et al., 1997). With the exception of two studies (Ayers & Pickering, 2001; Soet, Brack, & Dilorio, 2003), other studies have not accurately determined whether PTSD symptomatology is specifically related to childbirth, and may be present as a consequence of exposure to a traumatic event other than childbirth. This may mean that current estimates of postnatal PTSD are incorrectly reported. Furthermore, the majority of the research has been carried out within 4–6 weeks of the birth, which is likely to be due to the fact that a diagnosis of PTSD can only be made after a period of 4 weeks, and this would be the earliest opportunity for researchers to approach women and to maximise consent rates. Presently, little is known about recovery rates in women who develop PTSD as a consequence of childbirth (Ayers, 2003; McKenzie‐McHarg, 2004).

Current research suggests that women are more likely to develop postnatal PTSD if particular vulnerability factors are present. The following have been associated with PTSD after childbirth: high trait anxiety (Ayers & Pickering, 2001; Czarnocka & Slade, 2000; Soet et al., 2003); pre‐existing fear of childbirth (tokophobia)(Soderquist, Wimja, & Wimja, 2002); the presence of additional physical and/or mental health problems during the pregnancy (Moleman, van der Hart, & van der Kalk, 1992; Soderquist et al., 2004); a previous traumatic birth experience (Hoffman & Brockington, 2000) and a history of child sexual abuse (Rhodes & Hutchinson, 1994; Roussillon, 1998; Soet et al., 2003).

In addition to these, a number of other factors have also been associated with the development of PTSD after childbirth, such as: extreme pain and inadequate pain relief during labour (Allen, 1998; Creedy, Shocket, & Horsfall, 2000); higher levels of obstetric intervention, such as assisted delivery using forceps, ventouse or emergency caesarean section (Creedy et al., 2000; Soet et al., 2003); fear of death or physical harm to self or baby (Allen, 1998; Beck, 2004); feelings of loss of control and powerlessness during labour (Ballard, Stanley, & Brockington, 1995; Czarnocka & Slade, 2000; Reynolds, 1997). The development of PTSD after childbirth appears to be complex and is clearly dependent upon a number of vulnerability and environmental factors.

Postnatal depression and PTSD after childbirth

It has been reported that depression is often present in individuals who develop PTSD after exposure to any traumatic event (Green, 1994). Given that both PTSD and depression share a number of clinical features, this is not surprising. Two studies have reported that depression is present in a significant proportion of women who are reported to have PTSD after childbirth (Ballard et al., 1995; Czarnocka & Slade, 2000). However, the two disorders are distinct and require a different treatment approach. Thus, it is important to determine the presence of either or both disorders in women presenting with psychological distress after childbirth. Currently, no studies have investigated whether or not women continue to experience PTSD and depression in the longer term, and whether the presence of one disorder inhibits the natural recovery of another.

Given that postnatal PTSD has been recognised by professionals relatively recently, there remain a number of unanswered questions, as already outlined. Although a number of studies have reported prevalence rates for PTSD after childbirth, only two have screened women for a history of previous trauma (Ayers & Pickering, 2001; Soet et al., 2003). In addition, there is little knowledge about recovery rates in the longer term, i.e. up to 12 months after the birth. This study aims to address both of these issues by investigating whether women have experienced birth related trauma at 6–12 months post‐partum. A previous study has reported that a significant proportion of women experience a degree of PTSD symptomatology, but would not necessarily meet diagnostic criteria for PTSD (Czarnocka & Slade, 2000). The concept of partial PTSD has received some attention in the PTSD literature in recent years (Mylle & Maes, 2004). Thus, this study will also establish how many women continue to experience partial PTSD, as well as determining how many women report PTSD after childbirth.

Since few studies have investigated comorbidity in women who develop PTSD after childbirth, this study aims to determine how many women are also affected by postnatal depression. Finally, this research aims to determine which factors might predict levels of psychopathology in women 6–12 months after the birth of a baby.

Methodology

Design

This was a retrospective, self‐report study that examined women at 6–12 months postpartum (mean = 9.5 months; SD = 2.4; range = 6–13 months). This time frame exceeds the 4 week minimum period required by diagnostic criteria (DSM‐IV) to determine the presence of PTSD.

Participants

The participants were women who had given birth to a baby in the period October 2003–March 2004 at a district general hospital in North Wales. The following exclusion criteria were applied: stillbirth or subsequent death of baby; severe maternal mental health problems (e.g. psychosis/severe postnatal depression); women judged by their General Practitioner (GP) to object/become distressed by being approached by a researcher. In addition, care was taken to ensure that families who had experienced a maternal death were not approached.

Procedure

Approximately 1000 women were identified as having given birth in the above time frame, of whom 500 were selected by means of alternate randomisation by a Senior Information Analyst employed by the NHS Trust. An electronic check of patient records was carried out to ensure that, as far as the trust were aware, the women and their babies were still alive. Following this, the names, addresses and details of the General Practitioner of each woman were forwarded on to the trust audit office. The audit department contacted the GP of every woman, on behalf of the researcher, giving them an opportunity to decide whether it was appropriate for their patients being contacted to participate in the research, and further ensuring that women who may be distressed by being approached were not contacted. A total of 21 women were lost at this stage, for various reasons (no longer in the area; severe postnatal depression; infant death in another hospital; babies born with multiple disabilities; patient removed from GP list).

A total of 479 women were invited to take part in the research. Women willing to participate were asked to return their consent forms to the university. Up to this point, the principal researcher was not aware of the identity of the women, thus maintaining patient confidentiality. A total of 128 women agreed to take part in the research, and 102 anonymously completed questionnaires were returned.

This research was approved by: Ethics committee of the School of Psychology, University of Wales, Bangor; Local Research Ethics Committee & Research & Development Committee; Conwy & Denbighshire NHS Trust.

Measures

Clinical, demographic and birth characteristics

The following clinical and demographic information was collected: age; marital status; social class (derived from occupation; Hollingshead,1975); parity; number of children; age of baby; previous history of mental health problems; type of delivery,1 presence of physical health problems since delivery.

Screening for history of traumatic events

In order to determine whether women had experienced an event which may be perceived to be traumatic, they were asked to indicate (yes/no) if they had ever experienced any of the following: robbery/mugging; physical attack/beaten up; forced into sex by threat of force; involved in a serious motor vehicle accident in which someone was killed or seriously injured; death of a loved one due to an accident, murder or suicide; injury or property damage due to fire, weather or natural/man‐made disaster or some other ‘shocking/terrifying experience’.

Perinatal Post Traumatic Stress Disorder Questionnaire (PPQ)

The PPQ (Quinnell & Hynan, 1999) is a 14‐item self‐report scale that parallels diagnostic criteria for re‐experiencing/intrusions, avoidance and increased arousal (Criteria B, C and D) in DSM‐III‐R2 (APA, 1987). This questionnaire was designed to quantify symptoms of PTSD specifically related to childbirth. The PPQ has been validated in populations of women who have given birth to premature infants (deMier et al., 2000; Quinnell & Hynan, 1999). The PPQ has demonstrated high levels of internal consistency (Cronbach's alpha = 0.85); test re‐test reliability (r = 0.92) and correlates significantly with other well‐validated measures of PTSD (r = 0.61, p = 0.0001). In this study, the PPQ correlated significantly with the Post Traumatic Stress Disorder Checklist (r = 0.73; p = 0.001).

According to DSM‐IV (APA, 1994), a minimum number of items are required on criterion of re‐experiencing, avoidance and increased arousal. To meet diagnostic criteria for PTSD, individuals have to report at least one symptom for re‐experiencing; three or more symptoms for avoidance and at least two symptoms on increased arousal. On the PPQ, a scale of 0 to 4 is applied—rated from ‘not at all’ (0) to ‘often, for more than once a month’ (4); scores of 4 were considered clinically significant in this study. To determine whether women were symptomatic in any of the three domains of the PPQ, individual responses were scrutinised.

Women whose scores on the PPQ were clinically significant in all three areas of re‐experiencing (one item), avoidance (three items) and increased arousal (two items) were labelled fully symptomatic (S). Women whose scores were clinically significant in one area were labelled partially symptomatic (PS). Women whose scores were not clinically significant were labelled non‐symptomatic (NS). This method of determining partial PTSD has been reported in a recent review (Mylle & Maes, 2004).

The Cronbach's alpha for the PPQ in this study was 0.89.

Post traumatic stress disorder checklist (PCL)

The PCL (Weathers, Litz, Huska, & Keane, 1994) is a 17‐item self‐report scale that parallels diagnostic criteria for re‐experiencing, avoidance and increased arousal in DSM‐IV (APA, 1994). This questionnaire prompts individuals to endorse the level of distress that has co‐occurred with each reported PTSD symptom over the prior 30 days. The PCL has been validated in a number of veteran and civilian populations (Cronbach's alpha = 0.97; test–re‐test reliability = 0.93; Spearman correlation with other PTSD scales, r = 0.93), with differing cut off points (cut point of 50 yielded sensitivity of 0.60 and specificity of 0.99; positive predictive power = 0.75 (Andrykowski, Cordova, Stuidts, & Miller, 1998; Blanchard, Jones‐Alexander, Buckley, Forneris, 1996; Ruggiero, Del Ben, Scotti, & Rabalais, 2003). To date, the PCL has not been used to determine PTSD symptomatology in this population, but was employed in this study as a validity check for the PPQ.

Cronbach's alpha for the PCL in this study was 0.93.

Edinburgh Postnatal Depression Questionnaire (EPDS)

The EPDS (Cox, Holden, & Sagovsky, 1987) is a self‐rated questionnaire that asks women to rate their feelings over the previous week. The EPDS has become the most universal tool for research in this population. It has been found to have satisfactory levels of validity (split half reliability = 0.88; Cronbach's alpha = 0.87) and was acceptable to mothers and primary care workers (Cox & Holden, 2002; Cox et al., 1987). The 10‐item version was used in this study, where a score of ⩾13 indicates the presence of depressive symptomatology. The original validation study reported that using a cut‐off point of ⩾13 yielded 85% sensitivity and 77% specificity, with an overall positive predictive value of 83% (Cox et al., 1987).

Cronbach's alpha for the EPDS in this study was 0.88.

Perceptions of Labour and Delivery Questionnaire

This measure is a set of 24 questions that asks participants about their perceptions of their experience of labour and delivery (Czarnocka & Slade, 2000). Participants were asked to rate their perceptions of labour and delivery on a visual analogue scale, using a scale of 1–10 regarding items considering: severity of pain, amount of distress associated with the experience, satisfaction and confidence with coping, preparedness, fear for self and baby, unexpectedness of procedures and outcome, support from partner and staff, perceived control, feeling informed and listened to by staff, responsibilities for difficulties experienced, and evaluation of personal performance in labour in comparison with others.

Given that this is a visual analogue scale, the alpha values of this measure are not relevant.

Statistical analysis

The clinical and demographic information recorded was summarised by descriptive statistics (mean and standard deviation).

One sample Kolmogorov–Smirnov tests and visual examination of the data confirmed that the scores on the PPQ (z = 1.74–5.09; p = 0.0001), PCL (z = 2.5–4.36; p = 0.0001), and EPDS (z = 2.3–5.1; p = 0.0001) measures all had poor approximations to normal distributions. Thus, the Mann–Whitney non‐parametric test was used to determine differences between groups.

Spearman correlation was used to test for strength of linear association.

Linear multiple regression analysis was used to determine predictors of PTSD symptomatology. Whilst it is acknowledged that multiple regression analysis should normally be used with parametric data; it is also recognised that the parametric assumptions can be violated where there are moderate/large sample sizes. In this study, the use of multiple regression analysis can be justified in the presence of a large sample of 102 participants. Although there is considerable debate about what might constitute a large enough sample when conducting regression analysis, it has been reported that much smaller sample sizes (50 + 8 × number of predictors) may be used when conducting general regression analysis (Green, Salkind, & Akey, 2000; Osborne, 2000; Tabachnick & Fidell, 1996).

Statistical significance was defined as p<0.002. Alpha adjustment procedures were applied to the data, yielding a new probability value (see page 116).

Results were analysed by the Statistical Package for Social Sciences (SPSS) version 10.0.7 software package (SPSS, 2001).

Results

A total of 102 women completed the questionnaires, thus giving an overall response rate of 21%. The clinical and demographic characteristics of the participants are given in Table 1. The mean scores and standard deviations on the questionnaires are given in Table 2.

Table 1. Clinical and demographic characteristics of participants, according to clinical group (NS = non‐symptomatic; PS = partial PTSD symptomatology; FS = full range of PTSD symptomatology).

	Mean
	NS n = 78	PS n = 20	FS n = 4
Age^a	30.1 (5.5)	30.7 (5.3)	26.2 (8.7)
Social class^b†:
Professional	17 (21.7%)	7 (35%)	0 (0%)
Skilled non‐manual	21 (26.9%)	7 (35%)	1 (25%)
Skilled manual	14 (17.9%)	0 (0%)	1 (25%)
Semi‐skilled	14 (17.9%)	7 (35%)	1 (25%)
Manual	4 (5.1%)	2 (10%)	1 (25%)
Marital status†:
Single	5 (6.4%)	2 (10%)	2 (50%)
Married/cohabiting	72 (92.3%)	14 (70%)	2 (50%)
Parental status:
Primiparas	31 (39.7%)	10 (50%)	2 (50%)
Multiparas	47 (60.2%)	10 (50%)	2 (50%)
Type of delivery:
Normal vaginal delivery	47 (60.2%)	9 (45%)	2 (50%)
Assisted delivery (ventouse/forceps)	7 (8.9%)	4 (20%)	0 (0%)
Elective caesarean section	16 (20.5%)	1 (5%)	0 (0%)
Emergency caesarean section	8 (10.2%)	6 (30%)	2 (50%)
Previous mental health problem
Yes	14 (17.9%)	6 (30%)	0 (0%)
No	64 (82.1%)	14 (70%)	4 (100%)
† Any variation in numbers is due to missing data.^a = mean figures given; ^b = actual figures given; (% given in parenthesis).

Table 2. Mean scores and standard deviations for questionnaires used to determine the presence of psychopathology (n = 102).

	Mean	(SD/%)
Previous traumatic event^b:	60	(58.8%)
Yes	42	(41.2%)
No
Nature of traumatic event†:
Robbery/mugging	6	(5.9%)
Beaten up/attacked	31	(30.4%)
Forced into sex by threat of force	13	(12.7%)
Serious motor vehicle accident severe enough to cause injury to one or more passengers	12	(11.8%)
Loved one died by accident, murder or suicide	17	(16.7%)
Injury or property damage due to severe weather, natural or man made disaster	6	(5.9%)
Other terrifying/shocking experience (not specified)	29	(28.4%)
Perinatal Post Traumatic Stress Disorder Questionnaire (PPQ)^a:	13.3	(11.5)
Post Traumatic Stress Disorder Checklist (PCL)^a:	31.1	(13.9)
Edinburgh Postnatal Depression Questionnaire^a:	8.4	(5.6)
† = According to the Previous Traumatic Events Questionnaire.^a = mean figures given; ^b = actual figures given; (% given in parenthesis).

It can be seen from Table 2 that a considerable proportion (58.8%) of the sample under study reported having experienced one (or more) events which may be perceived to be traumatic: robbery/mugging (5.8%); beaten up/attacked (30%); forced into sex by threat of force (12.7%); serious car accident (11.7%); loved one died by accident, suicide or murder (16.6%); severe injury or property damage (5.8%); other terrifying/shocking experience—not specified (28.4%).

Rates of PTSD symptomatology at 6–12 months post‐partum

From the sample of 102 women (see Table 3), a total of 4 (3.9%) showed the number, frequency and duration (at least 4 weeks) of symptoms suggestive of clinically significant levels of re‐experiencing, avoidance and increased arousal. This group will be labelled fully symptomatic (FS).

Table 3. Rates of PTSD symptoms according to PPQ.

Outcomes	No. (%)
PTSD symptoms (total sample = 102)
No symptoms	78 (76.4%)
Partially symptomatic*	20 (19.6%)
Fully symptomatic	4 (3.9%)
Types of symptoms (PPTSD)†:
Re‐experiencing	8 (7.8%)
Avoidance	4 (3.9%)
Increased arousal	8 (7.8%)
Partial PTSD – one area	14 (13.7%)
Partial PTSD – two areas	6 (5.8%)
* In at least one domain of re‐experiencing; increased arousal or avoidance.† Any variation in n is due to symptom overlap.

Of the remaining 98 participants, 8 (7.8%) women reported re‐experiencing, 8 (7.8%) increased arousal and 4 (3.9%) avoidance. In all, a total of 203 (19.6%) women in this population are defined as partially symptomatic (PS). Of these, 14 (13.7%) reported PTSD symptomatology in one domain and 6 (5.8%) reported PTSD symptomatology in two domains. Combined with the FS group, 24/102 (23.5%) of women may be conceptualised as experiencing clinically significant levels of PTSD symptomatology at 9.5 months post‐partum. This group will be labelled symptomatic (S). The remaining 78 women will form the non‐symptomatic group (NS).

The mean score for the PCL was 31.1 and is reported in Table 2.

Further analysis was undertaken to determine whether the experience of PTSD was associated with previous traumatic experiences. Mann–Whitney analysis showed that there were no significant differences between groups (S vs. NS) on the number of previous traumatic events experienced by this population of women (z = −0.48; p = 0.62). In addition, the number of previous traumatic experiences did not correlate significantly with total scores on the PPQ (r = 0.14; p = 0.15), although a moderate correlation between the number of traumatic events and total score on the PCL was observed (r = 0.41; p = 0.0001).

Women's perception of their childbirth experience (Perceptions of Labour and Delivery Questionnaire)

Analysis was carried out to determine whether there were significant differences between the two groups on a number of independent factors experienced during labour and delivery, such as: perceptions of pain, pleasure, distress, elements of personal coping, preparedness, fear for self and baby, unexpectedness of procedures, perceived support, control, blame and whether women felt informed during labour. To control for multiple measurement, alpha adjustment procedures were applied to the data (existing probability value/number of items in measure = 0.05/24), yielding a new probability value (p = 0.002)(Cohen & Cohen, 1983). Results for the analysis using Mann–Whitney test are shown in Table 4. From this, it can be seen that there are several significant differences between the two groups. In comparison to the non‐symptomatic group, the symptomatic group reported higher levels of distress and fear for their baby; felt less prepared; reported lower perceived levels of control and reported that the procedures carried out were less expected.

Table 4. Mann–Whitney analysis showing significant differences (< 0.5) between groups on the Perceptions of labour and Delivery Questionnaire (symptomatic group (FS+PS), n = 24; non symptomatic group, n = 78).

Questionnaire item (Range 1–10)	Mean score	(SD) NS	Mann–Whitney U FS	p value PS
Overall rating of
Pleasure in labour	5.2 (2.7)	3.4 (2.5)	567.5	0.004
Overall distress	4.59 (2.9)	7.1 (2.6)	476	0.0001*
Preparedness	7.36 (2.6)	5.1 (3.2)	538	0.002*
Worst level of fear for self	4.68 (3.04)	6.6 (3.1)	594	0.008
Worst level of fear for baby	5.3 (3.2)	8.2 (2.1)	434	0.0001*
Average rating of fear for self	4 (2.6)	5.6 (2.9)	622	0.01
Average rating of fear for baby	4.8 (3.01)	7.7 (2.2)	423	0.0001*
Unexpectedness of procedures	4.1 (3.1)	7.5 (2.9)	411.5	0.0001*
Confidence during labour/delivery	6.7 (2.7)	5.5 (2.6)	673	0.05
Levels of control	5.9 (3.07)	3.5 (2.9)	516	0.001*
Overall rating of feeling informed	8.2 (1.9)	6.4 (3.1)	619.5	0.01
Wishes listened to by staff	8.3 (2.05)	6.9 (2.8)	608	0.01
Labour worse than expected	3.5 (2.9)	6 (3.7)	591	0.006
Labour better than expected	5.4 (3.1)	3.7 (3.1)	599	0.03
Overall rating of coping skills	8.9 (2.03)	7.4 (3.1)	707.5	0.05
* Significant at 0.002 level.

Rates of depression at 6–12 months post‐partum

A total of 22/102 (21.5%) women obtained a score of⩾13 on the EPDS, indicating clinical levels of depression. Furthermore, 6/22 (27.2%) also had some level of PTSD (5 = PS; 1 = FS), suggesting a significant amount of co‐morbidity in this group of women.

Predictors of PTSD symptomatology

A correlation matrix was generated between clinical and demographic factors and outcome variables (PPQ and EPDS scores). To maximise predictive power, only the demographic, clinical and outcome variables which correlated significantly at 0.2 (p<0.05) and above were entered into a multiple regression analysis. Given that this study is essentially exploratory in nature, this choice of method was made as there is insufficient evidence on which to base forced entry models (Tabachnick & Fidell, 1996). In addition, items from the perceptions of labour and delivery questionnaire, on which significant differences between the two groups were observed, were also entered into the regression model.

Within the final model predicting PPQ scores, 48% of the variance is accounted for by the 10 variables entered, as shown in Table 5 (only significant predictors are reported). It can be seen from this that EDPS scores, fear for the baby and the unexpectedness of procedures predicted higher scores on the PPQ.

Table 5. Results of linear multiple regression analysis showing significant predictors of higher scores on the Perinatal PTSD Questionnaire (n = 102).

Variable	Standardised
	B	SE β	β	t	p
EPDS scores	0.87	0.17	0.42	5.0	0.0001*
Being single	5.3	3.1	0.13	1.67	0.09
*Worst level of fear for baby	1.23	0.51	0.34	2.4	0.01*
*Unexpectedness of procedures	1.15	0.33	0.34	3.4	0.001*
Note: model was significant (F = 10.02); p = 0.0001; adjusted R 2 = 0.48.* These items are from the Perceptions of Labour and Delivery Questionnaire, which correlated significantly at 0.2 (p<0.05).

Within the final model predicting EPDS scores, 32% of the variance is explained by the three variables entered, as shown in Table 6. Again, it can be seen from Table 6 that PPQ scores and the presence of previous mental health problems predicted higher scores on the EPDS. Being single did not predict higher scores on the EPDS.

Table 6. Results of forward stepwise multiple regression analysis showing significant predictors of higher scores on the Edinburgh Postnatal Depression Questionnaire (n = 102).

Variable	Standardised
	B	SE β	β	t	p
PPQ scores	0.23	0.04	0.48	5.59	0.0001
Previous mental health problem	−2.6	1.17	−0.18	−2.23	0.02
Note: model was significant (F = 16.7); p = 0.0001; adjusted R 2 = 0.32.

Discussion

Of the women in this study, 4/102 (3.9%) showed significant levels of all three dimensions of PTSD, as measured by the PPQ, at 9.5 months post‐partum. This figure is higher than other reports at 6 months post‐partum, where a small number of studies have reported the prevalence to be around 1.5% (Ayers et al., 2001; Wimja et al., 1997). However, Ayers et al., (2001) reported a confidence interval of 0–3% and suggested that the actual incidence of PTSD might well exceed 1.5%. A number of other studies have also suggested that current estimates of postnatal PTSD are conservative, which may be partly attributed to methodological variation across studies (Ayers et al., 2001; Bailham & Joseph, 2003; Creedy et al., 2000).

A further 20/102 (19.6%) of the sample showed significant symptoms on at least one domain of the PPQ (re‐experiencing, increased arousal or avoidance). Other studies have also reported that a substantial proportion of women report some degree of PTSD (Czarnocka & Slade, 2000; Soet et al., 2003) in the first month after childbirth. This appears to be the first study that has ascertained that a considerable proportion of women (23.5%) continue to experience some level of PTSD at 9.5 months post‐partum.

Ayers et al. (2001) suggest that after 6 months' duration, there is less spontaneous recovery of PTSD symptoms and women are likely to continue as chronic cases and therefore require treatment (Ayers et al., 2001).

Previous traumatic event

A considerable proportion of women (58.8%) in this study reported that they had experienced a stressful event which may be perceived to have been traumatic. However, this cannot be assumed to generate PTSD and may well be in line with population norms. What is interesting is that the number of stressful events correlated with the PCL and not the PPQ. The PTSD measure selected for use in this study, the PPQ, asks women to report symptoms that are specifically related to the childbirth experience. Given this, it is unlikely that women were reporting symptoms which were related to an event other than childbirth. Furthermore, participants in the symptomatic group did not experience significantly more traumatic events than the non‐symptomatic group, suggesting that exposure to a previous stressful event does not account for their PTSD symptomatology in this study.

Women's perception of their childbirth experience

There were significant differences between the symptomatic and non‐symptomatic groups on a number of factors relating to the childbirth experience. The symptomatic group reported higher levels of distress and fear for their baby; felt less prepared; reported lower perceived levels of control and reported that the procedures carried out were less expected, in comparison to the non‐symptomatic group.

These findings are consistent with other studies, which have suggested that the fear of physical harm or death to the self/baby and the loss of control have precipitated the development of PTSD after childbirth (Allen, 1998; Ballard et al., 1995; Beck, 2004; Czarnocka & Slade, 2000; Reynolds, 1997). These findings have a number of potential implications for health care professionals involved in the antental/intrapartum care of women. It is likely that education during pregnancy may have an important preventative role to play, such as that provided in antenatal classes, ensuring that women are better prepared for childbirth. It was not known in this study which women had attended such classes, although future research might like to consider this.

Postnatal depression

A significant proportion (21.5%) of women in this study reported depressive symptomatology at nine months post‐partum. This does not appear to be an unusual finding, given that, although the majority of depressive episodes after childbirth spontaneously resolve within 3–6 months, a sizeable proportion remain depressed beyond the first postnatal year (McMahon, Barnett, Kowalenko, & Tennant, 2005). Studies have found that 20–25% of mothers of toddlers report clinically elevated symptoms of depression (Campbell & Cohn, 1997; McLennan, Kotelchuck, & Cho, 2001). Of the 22/102 women with elevated scores on the EPDS in this study, 6/22 also had symptomatic levels of PTSD. This is not surprising, given that there are a number of shared clinical features in both PTSD and depression. These findings also highlight the fact that although there is some overlap between the two conditions, a significant proportion of women with PTSD symptomatology would not be detected by routine screening for postnatal depression, as measures of depression would not capture symptoms of PTSD (re‐experiencing, avoidance and increased arousal in DSM‐IV; APA, 1994).

Predictors of PTSD and depressive symptomatology

Regression analysis found that depression scores, fear for the baby and the unexpectedness of procedures predicted higher scores on the PPQ. Surprisingly, a previous mental health problem did not appear to be predictive of higher scores on the PPQ. Other studies have suggested that higher levels of obstetric intervention, which may not have been anticipated, have been associated with greater PTSD symptomatology. Women who undergo an emergency caesarean section appear to find childbirth particularly traumatic (Bailham & Joseph, 2003; Creedy et al., 2000; Soet et al., 2003). It is interesting to observe that, in this study, one‐third of women in the symptomatic group had undergone emergency caesarean section, in comparison to 10% of the non‐symptomatic group.

As previously mentioned, other studies have reported that levels of fear for either the self or baby have precipitated PTSD symptoms in women (Allen, 1998; Ballard et al,. 1995; Beck, 2004; Czarnocka & Slade, 2000; Reynolds, 1997). In particular, Allen (1998) found that women's fear that their baby will be harmed was related to the emergency nature of medical interventions. The fact that fear for the baby was a predictor of higher PPQ scores is consistent with DSM‐IV (APA, 1994) criteria, which suggests that PTSD stressors relate to threatened death, injury or threatened physical integrity to self or other people.

Further regression analysis found that PPQ scores and a history of previous mental health problems predicted higher scores on the EPDS. Given that a number of participants experienced both PTSD and depressive symptomatology; and the overlap between the two conditions, it is not surprising that the PPQ scores emerged as a predictor of depressive symptomatology. In addition, individuals who have previously experienced mental health problems, particularly depression, are more likely to become depressed after the birth of a baby and studies have reported a high rate of relapse/reoccurrence in approximately 50% of cases (McMahon et al., 2005; O‘Hara, 2002).

Clinical implications

This study has demonstrated that PTSD symptomatology persists well into the first postnatal year in a significant number of women, and highlights the need for accurate detection of symptoms. Given that around a quarter of women also report depressive symptoms, it is likely that a significant number of women experiencing PTSD symptoms are going undetected during routine screening for postnatal depression. Furthermore, the findings of this study highlight the role of obstetric care in the development of PTSD symptoms, in that the unexpectedness of procedures predicted higher scores on the PPQ. In addition, many women with symptoms of PTSD reported high levels of distress and fear, and had a perceived lack of control during the birth. There are clear implications for health care professionals working with women in the antenatal/postnatal period; in that they need to be aware of psychological responses to childbirth, be able to identify women who are more at risk of developing PTSD, and to ensure that the birth experience involves excellent communication and optimal control. Furthermore, healthcare professionals need to be able to identify women presenting with PTSD symptomatology and, where relevant, refer on to mental health services.

Future studies might also like to investigate whether PTSD occurs in women who opt to give birth at home.

Limitations of this study

There are often limitations when carrying out retrospective research, such as the sample being less representative and consisting of those sufficiently motivated to take part. Thus, in this study, women with psychological problems may well have been less likely to participate. Ideally, women would have been recruited prospectively, which may have increased overall response rates, which was relatively low at 21%. Thus, it is difficult to generalise the findings of this study, given the bias in the sample, as mentioned above. It is also possible that other aspects of the study design impacted upon response rates. None of the participants had direct contact with the researchers, which may have made it less likely that they would participate in the research. Therefore, future research might like to consider a prospective design where the researcher recruits women via face‐to‐face contact.

Acknowledgements

We would like to thank Dr David Daley, Senior Research Tutor, for his statistical advice and Dr Matthew Kimble for his assistance in the planning of this research.

Notes

1. Type of delivery is hierarchical: 1 = normal vaginal delivery; 2 = elective caesarean section; 3 = assisted delivery; 4 = emergency caesarean section.

2. It is important to note that the only major difference between DSM IIIR and DSM‐IV in relation to PTSD is the definition of the traumatic event as outlined on page 3.

3. Any variation in n is due to symptom overlap.