Abstract
The growth of an infant in a loving, healthy relationship with a primary carer, usually the mother, helps to set a positive developmental trajectory for the child and the family. Maternal depression has an unfortunate adverse effect on this relationship [1]. On average, 13% of women develop major depression in the post-partum period [2]. Despite the importance of early detection and treatment, many cases of postnatal depression (PND) are missed by primary health-care and obstetric care providers [3]. Early recognition of PND and the development and evaluation of preventative interventions are currently a major focus of research attention [4–7].
There has been increased interest in the use of antenatal screening to identify women at high risk for PND, allowing for preventive interventions, and close postnatal follow up [7]. A review of studies involving almost 24 000 women concluded that the strongest predictors of PND are depression or anxiety during pregnancy, a past history of psychiatric illness, recent stressful life events, and perceived low levels of social support [8]. Neuroticism and marital problems were identified as moderate risk factors, and socioeconomic deprivation and obstetric complications as low-level risk factors.
Antenatal screening for psychosocial risk factors and depression therefore has the potential to improve detection of PND. There has been considerable support for the concept of introducing antenatal screening in Australia [9], and two Australian studies have found that screening is well accepted by women attending antenatal clinics [10], [11].
Generally, EPDS [12] is used as the gold standard to determine the presence of PND, although comparison between studies is limited by the use of a range of different cut-off scores to indicate caseness on the EPDS [13]. Further, Pallant et al. have demonstrated that an eight-item EPDS scale, rather than the 10-item scale currently in use, would have better psychometric properties [14].
The EPDS has been used antenatally to identify women at risk for PND. Rubertsson et al., in a study of 2430 women, found that 37% of women who had an EPDS score of ≥12, which was taken to indicate depression, also reached this criterion postnatally [15]. Josefsson et al., using an EPDS cut-off score ≥10 found that 33% of women who had antenatal depression went on to develop PND [16]. The results suggest that the EPDS, administered antenatally, will identify approximately one-third of women who will develop PND.
It would therefore make sense to use knowledge about risk factors for PND to develop instruments with better predictive validity. The APQ, developed in Sydney, Australia, includes questions about practical and emotional support, recent life events and major stresses, self confidence, obsessional traits, present or past depression or anxiety, childhood abuse, past or recent domestic violence, thoughts of self harm and substance use [17]. The predictive validity of this instrument has not been reported previously, and the aim of the present study was to assess the utility of the APQ in identifying women who subsequently developed PND.
The Pregnancy Risk Questionnaire (PRQ), also developed in Sydney [18], has some similarities to the APQ, but in addition includes items from the Parenting Bonding Instrument [19]. Unlike the PRQ it generates a single score. A cut off score of ≥46 has been found to be optimal, but only 23.5% of women scoring above this cut-off when assessed antenatally were found to have PND on follow up.
While PND is especially important because of the potential impact on the infant, and in severe cases because of the risks of neglect or harm to the infant as well as the mother, research following women through the perinatal period has consistently found that depression is more common during pregnancy than after the birth of the baby [20]. Recently anxiety, in particular worry, has also been identified as a common problem during pregnancy and as a predictor of PND [21].
Socioeconomic deprivation is a risk factor for PND [22], and also has numerous indirect effects. The catchment area for the Lyell McEwin Health Service, where the present study was carried out, includes four of the six most disadvantaged areas in urban South Australia. This region has very high ratings on health and social indicators ranging from low-birthweight babies to substantiated child abuse and neglect. In 2001 more than 65% of children in this region lived in families that were dependent on Government benefits [23].
The present study was especially concerned with the welfare of these women during the perinatal period. We aimed to investigate whether the APQ might be a useful instrument in identifying women at risk of PND within this disadvantaged population.
Method
Participants
All women presenting for antenatal care at the Lyell McEwin Health Service between June 2003 and December 2003 were offered the opportunity to participate in the study. Of the 431 eligible women, 421 provided complete antenatal data and consented to postnatal follow up. The women's date of confinement was obtained from the birthing records of the hospital and the EPDS was posted to them, with a reply paid envelope, 6 weeks after delivery. One hundred and fifty-four women responded, a response rate of 36.6%. Respondents had a mean age of 26.3±5.8 years (range 16–43 years), and 64 (43.8%) were having their first baby. There was no significant difference between responders and non responders on age, number of children or antenatal EPDS score. The present study was approved by the Ethics Committee of the North West Adelaide Health Service and all women gave written informed consent.
Measures
Study participants completed the 13-item APQ and the 10-item EPDS following their initial triage interview with a midwife. Depressive symptoms were measured using the EPDS, a self-report scale with favourable internal consistency [24] previously validated in Australian women [25]. We took EPDS scores ≥10 to indicate depression [26], [27]. This threshold of depressive symptoms has been shown to have optimal sensitivity (86.7%) and specificity (91.5%) among pregnant women [26], [28].
Statistical analysis
Results are reported as mean±SD. Antenatal and postnatal EPDS scores were compared using paired t-tests to determine whether there was significant change in the mean score over time. Correlations between antenatal and postnatal EPDS scores were determined using Pearson correlations. On answers to individual APQ questions, ‘yes’ and ‘sometimes’ were both taken to indicate a positive response. The association between scores on individual APQ items and EPDS caseness was evaluated using the χ2 test. Stepwise logistic regression was performed to examine the relationship between caseness on the postnatal EPDS and each individual APQ risk factor. p<0.05 was taken to indicate significance. SPSS Version 14.0 (SPSS, Chicago, IL, USA) was used for all analyses.
Results
The mean antenatal EPDS score (7.55±5.37) was significantly higher than the mean postnatal EPDS score (6.58±5.304; t = 2.45, df = 145, p = 0.015). There was a significant positive correlation between antenatal and postnatal EPDS scores (R = 0.592, p < 0.001).
Forty-four women (30.1%) met criteria for antenatal depression and 33 women (22.6%) met criteria for postnatal depression. Twenty-one women (14.4%) were depressed both before and after delivery of their baby.
The results of the APQ showed that 100 women (33.4%) had been abused as children. Of these, 10 (6.5%) were emotionally abused, nine (5.8%) were sexually abused, two (1.3%) were physically abused, and 13 (8.4%) reported experiencing several types of abuse. Fifty-six women (36.3%) had experienced major stresses, losses or worries in the previous 12 months. The only missing data related to questions about substance use. Forty-one women were smoking cigarettes during pregnancy (27% of 149 respondents), 32 were drinking alcohol (25.4% of 126 respondents) and six were using illicit drugs (4.6% of 129 respondents).
Women who met EPDS caseness criteria for antenatal depression were significantly more likely to also meet criteria for postnatal depression (χ2=22.72, p = 0.000). The women who met criteria for antenatal depression were significantly younger than women who did not meet criteria for antenatal depression (24.56±6.2 years compared to 27.1±5.5 years; t(144) = 2.47, p = 0.015), but there was no significant association between age and PND. Women who met EPDS criteria for PND were more likely to agree that they had felt anxious, miserable, worried or depressed for more than a couple of weeks (χ2=17.56, df = 2, p < 0.001) and more likely to report thoughts of self-harm (χ2=16.45, df = 2, p < 0.001).
Stepwise logistic regression was performed to examine the relationship between EPDS caseness for PND and the various risk factors identified antenatally by the APQ. The only antenatal risk factor found to predict PND was childhood emotional abuse (odds ratio = 15.24, 95% confidence interval = 1.36–171.32).
Discussion
We found high rates of both antenatal and postnatal depression in study participants, which may reflect their experiences of past and present adversity. Austin and Lumley observed that the prevalence of depression after birth ranges from 5.5% to 31.5% [29]. Comparison between studies was limited by the variation in EPDS cut-off scores, which ranged from >9 to >14. Josefsson et al. used the same EPDS cut-off as the present study when they followed 1192 women through the perinatal period [16]. They found that 12.7% of their subjects had PND, a considerably lower prevalence than that found in the present study.
Our study supported previous findings that depression is more common antenatally than postnatally [20]. Many of the present subjects were young women, almost half of whom were having their first baby, and they may well have been facing problems adapting to the changed life circumstances that accompany pregnancy. In addition, more than one-third had experienced major life stresses in the previous year. There was a significant drop in mean EPDS scores postnatally, with a correlation between antenatal and postnatal scores. However, although antenatal depression was associated with PND, more than half the women who were depressed antenatally did not remain depressed after delivery. Conversely, one-third of the women who had PND were new cases. This means that in the present study population, if the EPDS had been used antenatally to screen for PND, half of the women identified would not actually develop PND, and one-third of cases of PND would not be detected.
Individual risk factors on the APQ were generally unhelpful in predicting PND. Emotional abuse was the only predictor, and it is difficult to account for the lack of an effect for other forms of childhood abuse. Women with PND were more likely to have endorsed items concerning a past history of anxiety or depression. The APQ does not enable differentiation of anxiety and depression, and also does not specify when these feelings occurred, so it is unclear whether this finding reflects current anxiety and/or depression, or a past history of anxiety and/or depression. The endorsement of thoughts of self-harm is likely to reflect current depression. These items therefore appear to reflect past or current depression or anxiety, so it is not surprising that they tended to be associated with PND.
Interestingly we had a very good response to all questions except those dealing with substance abuse, where some women chose not to answer specific questions. We found high rates of tobacco and alcohol use compared to a Sydney study in which only 1.76% of women acknowledged smoking cigarettes and 0.89% admitted to drinking alcohol during pregnancy [18]. The high rates in the present study are probably an underestimate, because it is likely that the non-responders were in fact using these substances. A recent SA Department of Health report on inequality in South Australia found that in 2001, 27–40% of women in the hospital catchment area smoked during pregnancy, compared to a state average of 21.9% [23]. Overall, the proportion of women smoking during pregnancy increased with increasing socioeconomic disadvantage. Tobacco and alcohol are known to be harmful to the developing fetus so these findings are very concerning, suggesting that specific interventions to target these problems are needed.
The present study was limited by the low response rate, considerably less than that reported by Austin et al. studying a group of predominantly middle class women [18]. Further development of antenatal psychosocial risk instruments is needed. The PRQ [18] does show promise but more evaluation of this instrument is needed.
In conclusion, antenatal screening for psychosocial risk factors and depression is helpful in identifying women who would benefit from antenatal intervention, and providing data about issues such as tobacco and alcohol use. The APQ was limited by the lack of a single numeric score, or set of scores, and the PRQ may be more helpful in this regard. However, based on the present data at least one-third of cases of PND will be missed by antenatal screening, so there is no substitute for screening by general practitioners and other health workers caring for women after the birth of their baby. Further studies in women who are socioeconomically disadvantaged are needed to ensure that the extreme challenges facing some of these women are recognized, and the need for appropriate, targeted support and intervention is a priority for service delivery.
Footnotes
Acknowledgements
The authors thank the midwives at the LMHS Maternity Service for their help. We also thank Professor Bryanne Barnett, of Southwest Sydney Area Health Service, for her help and support.