Vulnerability to Major Depression following Childbirth: personality as a risk factor Dr Shivi Jaggi Senior House Officer, Luton NE CMHT Bedfordshire & Luton Mental Health & Social Care Partnership NHS Trust Introduction You have to go about with that big smile on your face and people saying, Aren t you lucky to have such a dear little baby, and you feel utter despair. (Welburn 1980) Postpartum depression refers to depression that occurs specifically after childbirth, there being practically no significant difference in symptoms and their prevalence with non-postpartum depression. In fact, some of literature on postpartum depression is confusing and there have been controversies about the discrete nature and definition of postnatal depression. In their metaanalysis of 59 studies, O Hara & Swain (1996) reported that it may be better to use relatively short time frames for the determination of postpartum depression. They reported that associations between risk factors and postpartum depression were strongest in those studies that had used one or two weeks as the length of the postpartum period under evaluation. It is therefore considered better to define postpartum depression where the onset occurs within the first few weeks following childbirth. The common and most frequently encountered symptoms and features of postpartum depression are fluctuations in mood, mood lability, and preoccupation with infant well-being (American Psychiatric Association 2000), although it is important to realise that these symptoms are not necessarily specific to postpartum depression. O Hara et al (1996) reported an average prevalence rate for postpartum depression of 13%, which is similar to the prevalence rate for mild to moderate depression at any other time in a woman s life (Johnstone et al 2004). However, serious postpartum depression that requires admission to hospital is clearly more prevalent (Robertson et al 2004). Postnatal depression does not just affect the mother, but also other family members, including her baby and spouse, and it presents specific challenges for professionals. The serious consequences of postnatal depression are well-documented, which can include a disturbed mother-infant relationship, adverse effects upon child development, and a higher rate of depression in the spouses of postnatally depressed women, that can have the effect of reducing the available compensatory supports for the child (Johnstone et al 2004). 58
Volume 4: Number 2 (Spring / Summer 2007) ISSN 1743-1611 (On-line) Having an accurate knowledge of risk factors for post-partum depression is therefore clearly likely to assist it s prevention, recognition and early treatment. These risk factors have been the subject of two meta-analyses, which identified a past history of depression, depression / anxiety during pregnancy, life stress, marital discord and a lack of social support as the strongest predictors of postnatal depression. However, there are some differences in the major risk factors for postnatal depression across cultures, with the female gender of the baby being a strong predictor of postpartum depression in some cultures, but not in western countries. The aim of this paper is to explore whether or not personality is a risk factor for vulnerability to depression following childbirth, for which a search of the contemporary literature was conducted. Risk Factors for Postnatal Depression Although any woman can develop postnatal depression, women who have certain risk factors have a significantly higher risk of developing depression following childbirth. Having an accurate knowledge of risk factors is therefore clearly very important in recognising those at risk and either preventing or ensuring the early traetment of postnatal depression, which may involve the provision of additional support by primary healthcare workers, such as health visitors, that has been shown to be effective in reducing depression (Johnstone et al 2004). In identifying the risk factors for postpartum depression, a literature search was conducted. Two major metaanalyses, one of which included an update, were found (O'Hara and Swain 1996, Beck 1996, Beck 2001), which considered a large number of studies and reported effect sizes (if Cohen s d = 0.2, this indicates a small relationship; if d = 0.4, this indicates a moderate relationship; and if d = 0.8, this indicates a strong relationship) (O Hara et al 1996). The known risk factors for postnatal depression are summarised in Table 1. As shown in Table 1, O Hara et al (1996) reported that the strongest risk factors for postpartum depression are: depression during pregnancy; anxiety during pregnancy; social support; stressful life events; baby s father s support; and, previous history of depression. Risk factors, which were weak to moderate predictors of postpartum depression included: neuroticism; obstetric and pregnancy complications; cognitive attributions; relationship with spouse; and, income. Similarly, Beck (2001) reported the following risk factors for postpartum depression as having a moderate effect size: prenatal depression (.44 to.46); self esteem (.45 to. 47); childcare stress (.45 to.46); prenatal anxiety (.41 to.45); life stress (.38 to.40); social support (.36 to.41); marital relationship (.38 to.39); history of previous depression (.38 to.39); infant temperament; (.33 to.34) and, maternity blues (.25 to.31). Risk factors with only a small effect size included: marital status (.21 to.35); socioeconomic status (.19 to.22); and, unplanned/unwanted pregnancy (.14 to.17). Although there is a big difference in the effect sizes for individual risk factors between the two meta-analyses, the risk factors for vulnerability to postpartum depression are not very different. Overall, the strongest predictors of postnatal depression are past history of depression, depression/anxiety during pregnancy, life stress, marital discord and a lack of social 59
Table 1: Risk Factors for Postnatal Depression (O Hara et al 1996) Risk Factor Number of Studies Cohen s d 95% Confidence Interval (CI) Effect Size (relationship to postpartum depression) Depression during 12 0.7491 0.67 / 0.83 Strong pregnancy Anxiety during pregnancy 6 0.6799 0.55 / 0.81 Moderate to strong Social support 4-0.6299-0.75 / -0.51 Moderate to strong Stressful life events 14 0.6041 0.54 / 0.67 Moderate to strong Baby s father s support 4-0.5265-0.67 / -0.39 Moderate to strong Previous history of 13 0.5704 0.49 / 0.65 Moderate to strong depression Neuroticism 5 0.3892 0.21 / 0.57 Weak to moderate Obstetric complications 13 0.2633 0.19 / 0.34 Weak to moderate Cognitive attributions 8 0.2444 0.18 / 0.31 Weak Relationship with spouse 7-0.2437-0.39 / -0.10 Weak Occupation 4-0.1460-0.25 / -0.04 None to weak Income 13-0.1405-0.21 / -0.08 None to weak Dyadic Adjustment Scale 5-0.1327-0.20 / -0.06 None to weak Age 22 0.0227-0.02 / 0.07 None Education 10-0.0428-0.11 / 0.03 None Working through 6-0.0219-0.10 / 0.05 None pregnancy Parity 5 0.0964 0.00 / 0.19 None Number of children 6 0.0286 -.05 / 0.10 None Marital status 10 0.0528-0.02 / 0.12 None Duration of relationship 5 0.0101-0.11 / 0.13 None Depressed relative 7 0.0506-0.06 / 0.16 None support. However, studies undertaken in Nigeria (Adewuya et al 2005), China (Lee et al 2000) and India (Patel et al 2002) identified that the female gender of the baby is a strong predictor of postpartum depression in these cultures, which may be the consequence of the new mother perhaps being subjected to antipathy, criticism, and even hostility from her husband and extended family if a female child is born (Lee et al 2000). Similar studies within minority ethnic groups in the UK would need to be conducted in establishing whether or not this is a risk factor. Influence of Personality as a Risk Factor As personality is revealed by a person s characteristic behaviour, researchers have derived various personality factors from well-known personality traits (for example: anxiety, energy, flexibility, hostility, impulsiveness, moodiness, orderliness and self-reliance) in producing a more scientific set of categories that can be measured through the use of questionnaires (for example, see Gelder et al 2001). As various personalities may predispose to psychiatric disorders, O Hara et al (1996) concluded that neuroticism is a weak to moderate predictor of postnatal depression, having assessed neuroticism as a risk factor in 60
Volume 4: Number 2 (Spring / Summer 2007) ISSN 1743-1611 (On-line) Table 2: Neuroticism and Postpartum Depression (O Hara et al 1996) Study Cohen s d 95 % Confidence Comments Interval (CI) Kitamura et al 0.2236-0.27 / 0.71 Very wide confidence interval Kumar & Robson 0.3315-0.22 / 0.88 Very wide confidence interval (1984) Watson et al (1984) 0.4964 0.25 / 0.74 Mild to strong relationship Boyce et al (1991) - 0.3431-0.86 / 0.17 Very wide confidence interval Pitt (1968) 0.9149 0.39 / 1.44 Moderate to strong relationship five studies, as shown in Table 2. However, it must be noted that these studies were old, with the latest being published in 1991. Review of Literature In response to this weakness, following a further search of the literature, a series of more contemporary studies were reviewed in exploring whether or not personality is a risk factor for vulnerability to postnatal depression. A brief review of the first six studies which were identified is presented. Study 1: Lee et al (2000) conducted a longitudinal (cohort) study, aimed at identifying psychosocial risk factors for postpartum depression among Hong Kong Chinese females. During the study period, 330 women were admitted to the postnatal ward of the Department of Obstetrics and Gynaecology at the Prince of Wales Hospital, of whom 220 (66.7%) agreed to participate in the study, which excluded non-chinese. Recruited into the study on the second day after delivery, a research nurse collected basic demographic, obstetric and psychiatric data from participants, which included completion of the Beck Depression Inventory (BDI) and the 30-item General Health Questionnaire (GHQ) at entry point and at six weeks postpartum. These women were also assessed for psychiatric disorder at six weeks postpartum using the Chinese translation of the non-patient version of the Structured Clinical Interview for the DSM III [SCID- NP], which was modified in making a sixweek, instead of four-week, diagnosis. Neuroticism was measured during the baseline assessment by using Eysenck Personality Questionnaire. The mean age of these women was 29 years (ranging from 16 to 42 years). Only 145/220 (65.9%) women attended for the follow-up appointment at six weeks postpartum, 17 (11.7%) of whom were diagnosed to be suffering from postnatal depression. All seventy five (34.1%) participants who did not attend for a follow up appointment completed the GHQ and BDI questionnaires by telephone. The findings indicated that women who scored high on neuroticism had an odds ratio (OR) of 1.3 for the development of postnatal depression (95% CI = 1.2 1.5; p<0.001). These women were thus 1.3 times more likely to develop postnatal depression as compared to women who scored low on neuroticism. In this study, neuroticism was measured at baseline, before the development of depression, minimising reverse causality and recall bias. The follow up process was good and complete, with reminder 61
phone calls being given and all participants completing the study. Furthermore, the small 95% confidence interval and p value highlight confidence in the study findings. However, this study would have benefited from a larger sample population recruited from several hospitals, which would have increased the generalisation of findings (Lee et al 2000). Furthermore, as only Chinese women were recruited in this study, it is difficult to say whether the results would apply to the UK population. There was no adjustment for confounders and whether or not the women with neuroticism were the same in all other aspects, such as history of depression and other sociodemographic characteristics, was not reported. In addition, the completion of survey via telephone in 75 cases may have limited the validity of the data (Lee et al 2000). Study 2: Dennis & Boyce (2004) conducted a longitudinal (cohort) study near Vancouver in British Columbia, measuring personality using the Vulnerable Personality Style Questionnaire (VPSQ), which was specifically developed to detect personalities that have been identified as a risk factor for the development of postnatal depression. 166 women were recruited antenatally (>32 weeks pregnant) and 667 were recruited postnatally (within the first week postpartum), with participants in both groups being mailed the postpartum questionnaires at one, four and eight weeks. 115/166 (69%) of those women recruited antenatally and 479/667 (72%) of those recruited postnatally returned the one-week postpartum questionnaire. Those who did not return the one-week questionnaire were then excluded from the study. Overall, 594 women competed the one-week postpartum questionnaire, of whom 535 (90%) completed the fourweek questionnaire and 498 (84%) completed the eight-week postpartum questionnaire. Depression was measured using a cut-off score of >12 on the Edinburgh Postnatal Depression Scale (EPDS). Dennis & Boyce (2004) concluded that the odds ratio for developing postnatal depression, for women with a vulnerable personality at one, four and eight weeks, was 1.35 (95% CI = 1.25 1.46), 1.33 (95% CI = 1.24 1.42), and 1.35 (95% CI = 1.26 1.47) respectively, which means that for an increase in one point on the VPSQ, women were approximately 1.35 times more likely to develop postnatal depression. Again, as a longitudinal (cohort) study, the chances of reverse causality and recall bias were minimised and this study benefited from having a large sample size which was representative of the general population. Study subjects were followed up for eight weeks, which may be viewed as an adequate follow-up period with regard to postnatal depression, and the assessment phase was conducted at three time points postpartum. Multiple regression was conducted to adjust for confounders. However, no information was given on the reasons for those dropping out of the study and whether or not there were any differences between completers and drop-outs. A further weakness concerns the use of the VPSQ, as there may have been an observer bias as the outcome of interest is known. Furthermore, assessing women at one week after childbirth may be somewhat debatable as the findings may have been confounded by experience of the maternity blues (Dennis et al 2004), and for a diagnosis of postnatal depression there is a requirement that symptoms are present for, at the very least, one week 62
Volume 4: Number 2 (Spring / Summer 2007) ISSN 1743-1611 (On-line) (O Hara et al 1996). Study 3: Boyce & Hickey (2005) conducted their longitudinal (cohort) study over a period of four months at Nepean Hospital, Penrith, New South Wales, Australia, recruiting women who had delivered a healthy infant during the study period at the hospital. Of a total of 723 eligible participants, 522 were recruited into the study 79 were discharged early and 122 declined to participate. Of the 522 participants, 97 then missed two or more follow-up appointments, leaving 425 women in the final study sample. A baseline assessment was conducted with each participant within two days after childbirth, following which postal questionnaires were distributed at six, 12, 18 and 24 weeks postpartum. Postnatal depression was diagnosed for those with an EPDS score >12 on two occasions, and confirmed through the use of the depression and anxiety sections of the Structured Clinical Interview for DSM-III-R (SCID). Personality was measured using VPSQ and it was found that women with a vulnerable personality had a significantly increased risk of experiencing postpartum depression (OR = 3.5; 95% CI = 2.80 11.36). The adjusted odds ratio for vulnerable personality was 2.82 (95% CI = 1.06 7.45), meaning that having a vulnerable personality is a risk factor for postpartum depression at both the univariate and multivariate level. This study benefited from a large study sample, and although this was not a true consecutive sample, it was representative of general population. A hierarchical logistic regression analysis was performed to examine the independent relationship between vulnerable personality and postpartum depression, which adjusts for confounders. However, 97 women dropped out of the study and it was noted that the study group was significantly older than those who dropped out. As the baseline assessment was completed within two days postpartum, antenatal depression may have been underestimated, which could have confounded the findings. Furthermore, detailed measures in the form of full semi structured interviews were not used to measure psychosocial variables, thus potentially missing specific elements that may increase risk. Study 4: Johnstone et al (2001) conducted a prospective (cohort) study, in which women were recruited from each of four participating hospitals in New South Wales, Australia. Their exclusion criteria included: having a history of psychotic illness; drug/alcohol dependence; significant medical problems with the infant; inadequate English language skills; age < 16 years; or, a history of repeated deliberate self-harm. As 14 women were unable to complete the EPDS, the final sample size was 490, with a mean age of 28.0 years (range: 16.0 to 42.8 years). The VPSQ was used to measure personality within one week postpartum and the EPDS was used to identify the women with postnatal depression at eight weeks postpartum: 64 women were found to be suffering from postnatal depression (PND) at eight weeks postpartum. Their findings showed that the chances of experiencing postnatal depression were increased if a woman defined herself as being either nervy (OR = 3.25; CI = 1.38 7.56), shy/ self-conscious (OR = 4.13; CI = 2.09 8.20), obsessional (OR =3.44; CI = 1.68 7.03), angry (OR = 2.18; CI = 1.01 4.63) or a worrier (OR = 4.40; CI = 2.07 9.57). Although women who rated themselves as being timid (i.e. unable to assert yourself) also had higher odds of 63
developing postnatal depression, the confidence interval was very wide (OR = 1.94; CI = 0.80 4.61). Subjects who had mid high scores on the VPSQ also had higher odds of developing postnatal depression (OR = 3.86; CI = 1.44 11.23). This study had a number of strengths: the response rate was high, with approximately 98% of women completing the study; determination of a statistically significant relationship between an independent and the dependent variable was based on the Chi square statistic (Mantel-Haenzel weighted), which adjusted for confounding; the study sample was taken from four hospitals in two different regions of New South Wales, Australia, thus increasing the generalisation of the findings; and, hierarchical logistic regression analysis was also performed. However, depression was only assessed at eight weeks postpartum and would have benefited from assessment at two different time points postpartum. Personality was measured one week postpartum, which might be confounded by postpartum blues. Furthermore, the confidence interval was very wide for all types of personality, and specially for being timid, which may in fact act as a protective factor, reducing the chance of experiencing postnatal depression. Study 5: Matthey et al (2000) conducted a longitudinal (cohort) study, examining the course of postnatal depression in firsttime mothers and fathers with an emphasis on the role of personality and parental relationships as risk factors. Women were recruited early in the second trimester, with assessments being conducted at that time and at six weeks, four months and twelve months postpartum. Following a screening and matching process, 283 women were invited to participate in the study with their partners, of whom 166 (59%) accepted. A further nine couples were excluded as their infants were born prematurely (N=8) or were severely disabled (N=1). Of 157 participating couples, another 11 (7%) dropped out, six of whom due to moving home. The Beck Depression Inventory (BDI) was used to assess depression in these women antenatally and at four and 12 months postpartum, while the EPDS was used to assess depression at six weeks postpartum. Personality was measured using Eysenck Personality Questionnaire (1964 version). Matthey et al (2000) concluded that the level of neuroticism in women was associated with development of postnatal depression at each of the four assessment points with correlations ranging from 0.28 to 0.42. Again, as a longitudinal study, reverse causality was minimised. The total number of subjects dropping out of the study was small. However, the study sample had a relatively high level of education, thus limiting the generalisation of the findings to other groups. Furthermore, a follow-up period of 12- months for postnatal depression could be considered too long. Study 6: In the final study, Verkerk et al (2005) conducted a longitudinal (cohort) study in which subjects were recruited during mid-pregnancy, when visiting the obstetrician or midwife for antenatal care. Initially, 1,618 subjects were referred by the midwife or obstetrician, although only 1,031 were deemed eligible (being Dutch speaking, 20 30 weeks pregnant, living in the vicinity of Tilburg and Eindhoven, having returned a fully completed questionnaire, and having consented to participate in a follow-up study during pregnancy and the postpartum). 64
Volume 4: Number 2 (Spring / Summer 2007) ISSN 1743-1611 (On-line) Screening questionnaires were numbered in the order in which they were received, and odd numbers were selected in identifying a group of 339 women who were assessed during pregnancy and continued their participation through postpartum. 45 (13%) dropped out of the study and data was incomplete for a further 17 women, who were thus excluded from the study. A total of 277 women therefore participated in the present study. Personality was assessed using the Dutch Personality Questionnaire (a Dutch version of the California Psychological Inventory). Assessment was conducted for neuroticism and introversion, with high neuroticism being indicated if scoring O 13 on the neuroticism scale and high introversion being indicated if scoring O 12 on the introversion scale. Participants were then classified into four personality types: high neuroticism and high introversion (N=44; 16%); highneuroticism and low-introversion (N=38; 14%); low-neuroticism and highintroversion (N=48; 17%); and, low neuroticism and low introversion (N=147; 53%). High neuroticism was found to be a risk factor for the development of postnatal depression and women with the high neuroticism and high introversion personality type had a higher risk of developing depression during the first year postpartum, even when controlling for depression during pregnancy. The odds ratio for the development of postpartum depression among women with high neuroticism and high introversion was 3.08 at three months (95% CI = 1.10-8.63), 4.64 at six months (CI = 1.65-13.16), and 6.83 at twelve months postpartum (CI = 1.97-23.74). Considering the strengths of this study: subjects were randomly selected and representative of the general population; neuroticism and introversion were measured at 32 weeks pregnancy, minimising reverse causality; multiple logistic regression analyses were conducted to adjust for confounders; and, subjects were stratified by personal history of depression and high-risk personality. However, assessment for postpartum depression was completed at three, six and twelve months after childbirth, although postpartum is considered as depression that occurs during the first six to eight weeks postpartum. Furthermore, whilst clinical depression was diagnosed using Research Diagnostic Criteria, it is difficult to generalise these findings to the more current DSM-IV or ICD-10 criteria. Although it was noted that the 95% confidence interval was very wide, it still showed positive correlation. Conclusion All of the studies reviewed show a clear relationship between certain types of personality and the development of postpartum depression, even though all have particular limitations and weaknesses. As shown, all were longitudinal (cohort) studies, even though this was not a criterion for their selection. However, in identifying a relationship between personality and postpartum depression, this is probably the best study design, as it minimises reverse causality and recall bias. There was no general consensus on the measurement scales used to assess personality as the Eysenck Personality Questionnaire was used in two studies, the Vulnerable Personality Style Questionnaire was used in three studies and the Dutch Personality Questionnaire was used in the sixth study. As these personality questionnaires are different, the personality factors found to be associated with postpartum depression were also 65
different. The time period for both baseline assessment and follow-up also differed within these studies, ranging from six weeks to one year. Nevertheless, as an outcome of this literature review, it is possible to conclude that personality is a risk factor for vulnerability to the episodes of depression after childbirth. The personality types and factors that were found to be associated with postpartum depression are high neuroticism, high introversion (in combination with high neuroticism) and vulnerable personality (as measured using the VPSQ). What is less clear is whether there is any specific relationship between personality and postpartum depression over and above the well-established relationship of personality with nonpostpartum depression. Further research needs to be conducted comparing neuroticism scores for women with major depression who have and who have not suffered with episodes of postpartum depression, taking account of and minimising the weaknesses of those studies reviewed in this paper. References Adewuya AO, Fatoye FO, Ola BA, Ijaodola OR & Ibigbami SM (2005) Sociodemographic and obstetric risk factors for postpartum depressive symptoms in Nigerian women. Journal of Psychiatric Practice 11 (5): 353-358. American Psychiatric Association (2000) Diagnostic & Statistical Manual of Mental Disorders DSMIV-TR (Fourth edition). Washington: American Psychiatric Association, 422-423. Beck CT (1996) A meta-analysis of predictors of postpartum depression. Nursing Research 45 (5): 297-303. Beck CT (2001) Predictors of postpartum depression: an update. Nursing Research 50 (5): 275-285. Boyce P & Hickey A (2005) Psychosocial risk factors to major depression after childbirth. Social Psychiatry and Psychiatric Epidemiology 40: 605-612. Dennis CL & Boyce P (2004) Further psychometric testing of a brief personality scale to measure vulnerability to postpartum depression. Journal of Psychosomatic Obstetrics & Gynecology 25: 305-311. Gelder M, Mayou R & Cowen P (2001) Shorter Oxford Textbook of Psychiatry (Fourth edition). Oxford: Oxford University Press, 157-160. Johnstone EC, Owens DGC, Lawrie SM, Sharpe M & Freeman CPL (2004) Companion to Psychiatric Studies (Seventh edition). London: Churchill Livingstone, 748-750. Johnstone SJ, Boyce PM, Hickey AR, Morris- Yates AD & Harris MG (2001) Obstetric risk factors for postnatal depression in urban and rural community samples. Australian and New Zealand Journal of Psychiatry 35: 69-74. Lee DTS, Yip ASK, Leung TYS & Chung TKH (2000) Identifying women at risk of postnatal depression: prospective longitudinal study. Hong Kong Medical Journal 6 (4): 349-354. Matthey S, Barnett B, Ungerer J & Waters B (2000) Paternal and maternal depressed mood during the transition to parenthood. Journal of Affective Disorders 60: 75 85. O Hara MW & Swain AM (1996) Rates and risk of postpartum depression a meta-analysis. International Review of Psychiatry 8: 37-54. Parry BL & Haynes P (2000) Mood disorders and the reproductive cycle. Journal of Gender Specific Medicine 3: 53 58. Robertson E, Grace S, Wallington T & Stewart DE (2004) Antenatal risk factors for postpartum depression: a synthesis of recent literature. General Hospital Psychiatry 26 (4): 289-295. Verkerk GJM, Denollet J, Heck GLV, Son MJMV & Pop VJM (2005) Personality factors as determinants of depression in postpartum women: a prospective 1-year follow-up study. Journal of Psychosomatic Medicine 67: 632-637. Welburn V (1980) Postnatal depression. London: Fontana/Collins, cover page. World Health Organization (2002) The ICD-10 classification of mental and behavioural disorders. Geneva: World Health Organization, 195. 66