Social Psychiatry and Psychiatric Epidemiology

, Volume 46, Issue 5, pp 413–423

Social dynamics of postpartum depression: a population-based screening in South-Eastern Hungary

Authors

    • Women and Children’s Division, Department of Obstetrics and GynaecologyOslo University Hospital, Ullevaal
  • Robert B. Dudas
    • Department of PsychiatryUniversity of Cambridge, Box 189, Level 4, Addenbrooke’s Hospital
  • Sarolta Csatordai
    • Faculty of Health SciencesUniversity of Szeged
  • Iván Devosa
    • Section of Behaviour ScienceUniversity of Szeged
  • Éva Tóth
    • Faculty of Arts and Sciences, Institute of PsychologyUniversity of Pécs
  • Dávid Szabó
    • Department of Obstetrics and GynaecologyUniversity of Szeged
  • János Sikovanyecz
    • Department of Obstetrics and GynaecologyUniversity of Szeged
  • János Zádori
    • Centre for Assisted Reproduction, Kaáli InstituteUniversity of Szeged
  • Katalin Barabás
    • Section of Behaviour ScienceUniversity of Szeged
  • Attila Pál
    • Department of Obstetrics and GynaecologyUniversity of Szeged
    • Centre for Assisted Reproduction, Kaáli InstituteUniversity of Szeged
Original Paper

DOI: 10.1007/s00127-010-0206-2

Cite this article as:
Kozinszky, Z., Dudas, R.B., Csatordai, S. et al. Soc Psychiatry Psychiatr Epidemiol (2011) 46: 413. doi:10.1007/s00127-010-0206-2

Abstract

Purpose

To determine contributing psychosocial factors to postnatal depression (PND) in Hungary in 1996 and in 2006.

Methods

In 1996 and 2006, a total of 2,333 and 1,619 women, respectively, were screened for PND in South-Eastern Hungary, based on a Leverton questionnaire (LQ) score of ≥12 at 6–10 weeks after delivery.

Results

The LQ scores indicated an increase in PND from 15.0% in 1996 to 17.4% in 2006. The best predictors for PND in a multiple regression analysis were living in an urban environment [adjusted odds ratio (AOR) = 11.26], unstable relationship (AOR = 3.1) and a perceived lack of social support from partner (AOR = 3.65) in 1996, and recent major life events (AOR = 3.38), unstable relationship (AOR = 3.84), self-reported low income (AOR = 1.82), and intention to return to work soon after delivery (AOR = 0.47) in 2006.

Conclusions

A self-defined low socioeconomic status and an intention to return to work have become significant factors in the development of PND. Besides the family factors recognized as salient variables in 1996, economic features came into prominence as newly identified main predictive factors for PND in 2006.

Keywords

Leverton questionnairePostpartum depressionMultiple logistic regressionEffects of social changesHungary

Introduction

Pregnancy and the puerperal period involve particularly dramatic changes in the lives of women, and can lead to mood disorders [32]. Thus, various psychiatric symptoms readily appear in the postpartum period, which is a critical time as regards the development of postnatal depression (PND). PND is a complex entity with an extensively growing body of literature concerning the occurrence and risk factors. In a wide-ranging literature review, Robertson et al. found that the strongest predictors of PND were past history of depression, depressive symptoms or anxiety during pregnancy, experiencing stressful life events in the previous year, and low levels of social support [6, 25, 33]. Mothers satisfied with their family life appear more protected from PND [16, 2527, 36, 37]. Those with employment problems or on chronically low income are at a greater risk of developing PND [8, 13, 38]. Unplanned or unwanted pregnancy can likewise be a risk factor [8, 26, 27, 33], as recognized recently.

PND affects 5–25% of women worldwide [5], but the prevalence varies depending on the method used for diagnosis or screening and on whether the study is a prospective or retrospective one [5, 14]. Importantly, an international study using basically the same methodology (although adapted to local conditions) described significant differences between countries [1]. Most epidemiological studies of PND in Western societies have reported a prevalence of 10–15% at 6 weeks postpartum [5]. The accurate identification of high-risk women necessitates targeted screening. Although epidemiological investigations and comprehensive meta-analyses of predictive studies have led to the development of predictive models whereby one can identify high-risk women on the basis of maternal mood 6 weeks postpartum, antenatal screening demonstrated limited predictive validity [5, 14, 17, 34]. It is of practical relevance that a meta-analysis has revealed that the absolute difference in estimates between self-report assessments and diagnostic interviews is small [33].

An approach involving the identification of the sociocultural factors most relevant to the sufferer’s own experience makes it possible for any intervention to be more closely targeted to the individual case [30]. Although PND has even been thought of as a culture-bound syndrome of Western societies [7], research on the effects of social changes on the predictors of PND has been sparse. An increasing prevalence of affective disorders during the past century has been well documented [9]. Although it would be of paramount importance to identify and treat depression during and after pregnancy, especially among women at heightened risk, in Eastern Europe there is a relative dearth of research on the epidemiology of PND. The only large study conducted in Eastern Europe revealed an incidence of 31% of maternity blues, while the frequency of depressive symptoms was 10–17% among mothers in Poland during the first week postpartum [28], and risk factors were not looked at in that study at all. To our knowledge, our study is the first one to look into the prevalence rates and sociodemographic, economic and psychological predictors of PND and the effect of the fundamental political and social changes on these during the transition period in the former Eastern block.

Aims

The main goals of this study were to perform a large-scale survey to establish the incidence of PND in South-Eastern Hungary and to elucidate the extent to which certain sociodemographic, economic, and psychological risk factors influence PND, enabling a comparison with the situation in other countries. A secondary objective was to evaluate the effect of social changes on the prevalence of maternal depressive symptoms as captured by the Leverton questionnaire (LQ) in South-Eastern Hungary in 1996 and in 2006. The early 1990s were characterized by profound changes in the political and economic systems in Hungary and we were specifically interested to know whether the social changes that followed in the wake of these would have an impact on the prevalence of PND over a 10-year period. We believed that an understanding of the risk factors involved in PND would provide clues to the identification of mothers at risk.

We were specifically interested in three main groups of variables:
  1. 1.

    sociodemographic and obstetric (age, type of residence, marital status, education, number of children, primiparity, unplanned pregnancy and unwanted pregnancy),

     
  2. 2.

    economic (self-reported low financial income and intention to return to work after the postpartum period), and

     
  3. 3.

    psychological [unstable relationship, perceived lack of support by partner and perceived lack of support by the family, a history of major depressive episode(s), history of major life events over the last 12 months, and an independent style of management of everyday life problems].

     

Our hypothesis was that these sociodemographic, economic, and psychological variables will have a complex and significant influence on the LQ score and the likelihood of scoring above the cut-off for PND.

Over the studied 10-year period, there was a change at population level in several of the sociodemographic variables (e.g. marital status, number of children, etc.) [20, 24] relevant to our study and we predicted that this had an influence on the prevalence of PND.

The economic situation of the average Hungarian family got worse from 1996 to 2006 and we predicted that economic variables would have a stronger influence on PND in 2006 than they did 10 years earlier. The poverty rate in the entire population increased from 9.8 to 14.3% [21]. In 2006, in those households where the mother returned to work soon after delivery the poverty rate was less than half of that in households where the mother remained at home. These proportions were more favourable 10 years earlier (5 vs. 15%) [21].

We wanted to study the interactions of the effects of the above factors with psychological factors, such as the stability of the relationship, social support [22, 25, 38], and stressful life events [22, 31].

Methods

Subjects and procedure

The study was carried out in all the 62 pregnancy-care centres of South-Eastern Hungary between 1 January and 20 November 1996, and between 1 January and 20 November 2006. All postpartum women living in the four counties of South-Eastern Hungary were invited into the study. The total population of the region is around 1.8 million people. In Hungary, the gynaecologists and specialist health-care staff serving in the pregnancy-care centres provide high-level pre- and postpartum care and reach almost 100% of all mothers in the postpartum period [20]. Every pregnant woman is invited to monthly visits where a trained midwife carries out a comprehensive health check, including blood pressure, urine test, physical signs of progress of pregnancy, and performs a cardiotocography if needed. On four occasions, a gynaecologist reviews the patient and performs an ultrasound scan and reviews the serological test results. It is customary for young mothers to present in these centres for a review by a gynaecologist at 6 weeks postpartum. Mothers are seen by health visitors trained at college level over the first year after delivery as frequently as necessary up to once a week.

After an explanation of the objectives of the study, a trained health visitor conducted a questionnaire interview, guaranteeing the anonymity and the protection of the subjects’ privacy. The subjects were attending the routine check-up at 6 weeks postpartum. Those who had a stillbirth/perinatal death or who had other reasons for depression (e.g. depression related to mourning or organic causes, depression due to chronic physical illness not related to the pregnancy or delivery, or other neurological/psychiatric problems such as epilepsy) were excluded from the study in both years. Illiterate women were also excluded from the study.

Initially, in 1996, 2,774 postpartum mothers were invited to participate in the study, out of whom 319 (11.5%) declined to participate and 226 (8.1%) were excluded. The questionnaire was answered by 2,229 mothers, i.e. the response rate was 80.3%. In 2006, the primary sample comprised 1,936 mothers; 212 (10.9%) did not wish to participate in the study and 111 (5.7%) were excluded, resulting in a response rate of 83.3%. According to the information at the pregnancy-care centers, those who did not respond did not differ in mean age, parity, and residential area from those who did.

Leverton et al. [18] created their 24-item questionnaire (LQ) tapping into symptoms of major and minor depression to detect PND. Although, the LQ does not have an extensive coverage in the scientific literature, we had previously confirmed that it was a valid and reliable scale measurement for screening for depression [12] and this was the only test validated in Hungarian available at the time of designing our study. The cut-off value of the LQ was determined in our previous study at 11/12 for PND with a sensitivity of 88.0% and a specificity of 94.4% [12]. The interview in this study included the LQ, and additional structured questions exploring sociodemographic, economic, and psychological risk factors. A specially designed training course was organized for all health visitors on the interview methodology, minor and major depression, and postnatal physiology before the study. Participants filled in the questionnaire on their own and a health visitor went through their responses with them to clarify and explain items, as necessary.

The additional structured questions were based on a consensus among the authors following a review and discussion of the available literature. Some questions derived from worldwide findings regarding major risk factors for PND, whereas others were created in an attempt to study the effects of sociodemographic changes specific to Hungary. All of these questions were ‘yes/no’ questions and corresponded to a variable in our analysis, except for the variable of major life events in the past year for which a minimum of two items from a list of ten questions selected from established life events scales [31] was required to be answered positively for the variable to be coded as positive [33]. Ten checklist items were selected from established life event scales [31]: (1) being separated or divorced, (2) serious problems in marriage or cohabiting relationship, (3) serious problems or conflicts with family, friends, or neighbours, (4) problems at work or in place of education, (5) economic problems, (6) serious illness or injury, (7) serious illness or injury in close family, (8) road traffic accident, fire, or theft, (9) loss of a close relative, and (10) other difficulties [31]. As described by several authors, the occurrence of two or more stressful major life events in the previous year is a strong predictor of PND [18, 33]. A history of major depression independent of childbirth or prior postpartum depression has a strong influence on PND [18, 33], and this too was examined. Self-defined low income and the intention to return to work after the postpartum period (6–12 months after delivery) also appeared of interest based on the previous literature [13, 26, 38]. Emerging evidence suggests that subjective evaluation of one’s own economic status is a better predictor of future psychiatric morbidity than unemployment or objectively defined poverty [39]. Intention to return to work was an important variable, as, although the relevant laws and state provision had not changed, there seemed to be a changing trend in mothers to return to work as early as possible (it is not entirely possible here to distinguish between the effect of a career-oriented lifestyle and financial necessity) and this had not been examined in previous studies. Previously, it used to be typical for mothers to stay at home with the baby for the full length of 3 years of state provision for mothers. We further wished to examine associations with different aspects of satisfaction with family life, perceived support by partner and family, relationship stability, and an independent style of management of problems of everyday life [18, 36, 37].

The study protocol and the questionnaire were approved by the Clinical Research Ethics Committee of the University of Szeged and by the Medical Ethics Committees in the four counties involved.

Statistical analysis

Computations were carried out with the SPSS 14.0 software (SPSS Inc., Chicago, IL). Most continuous variables did not show normal distribution in our subject groups; therefore, non-parametric tests were used.

The correlations between binary categorical variables in cases and non-cases were analysed by Chi-square tests in each year. The resultant odds ratios were then compared between the two distinct years using the Mantel–Haenszel test, offering an estimation of the changes in these characteristics. To test between group differences, the Kruskal–Wallis analysis of variance was used.

Also, multiple logistic regression analysis was applied to assess demographic and psychosocial characteristics of PND in both 1996 and 2006 in a simultaneous fashion. All bivariate analyses were weighted to age and residential area representative for South-Eastern Hungary [29]. Each variable included into the final model was highly significant and the most robust model was chosen in both years [19].

Results

Subject attributes in 1996 and 2006

The sociodemographic circumstances of the participants were assessed in the study groups in 1996 (N = 2,333) and 2006 (N = 1,613) (Table 1). A group of 333 (15.0%) subjects exhibited an LQ score of ≥12 (PND) in 1996, whereas there were significantly more mothers with an LQ score indicative of PND in 2006 (17.4%). The PND group were significantly younger than the non-depressed mothers (non-PND) in 1996 but not in 2006. In 1996, a higher proportion of those with PND lived in towns (79.9%) than those without depression (65.5%). Less than one-third of all participants lived in a rural area in 1996, but somewhat more did so in 2006, and the significant difference in residential area between the depressed and non-PND groups detected in 1996 was not observed in 2006. Those without depression were significantly more likely to be married both in 1996 and in 2006.
Table 1

Selected sociodemographic and obstetric anamnestic data of the study groups in 1996 (N = 2,229) and in 2006 (N = 1,613)

 

Participants in 1996 (N = 2,229)

P value***

OR (95% CI) ***

Participants in 2006 (N = 1,613)

P value***

OR (95% CI) ***

P value****

Non-depressed mothers (<12 points in the LQ) (N = 1,896)

Mothers with PPD (≥12 points in the LQ) (N = 333)

Non-depressed mothers (<12 points in the LQ) (N = 1,332)

Mothers with PPD (≥12 points in the LQ) (N = 281)

n

%

n

%

n

%

n

%

Age (year) (mean ± SD)*

24.6 ± 4.75

23.8 ± 4.37

0.003

 

27.8 ± 4.32

27.8 ± 5.27

NS*****

 

Type of residence

 Town

1,242

65.5

266

79.9

<0.001

 

868

65.2

166

59.1

NS*****

  

 Village

595

31.4

65

19.6

 

406

30.5

96

34.2

  

 Outlying area

59

3.1

2

0.6

 

58

4.4

19

6.8

  

 Married

1,496

78.9

226

67.9

<0.001

0.56 (0.44–0.73)

900

67.6

165

58.7

0.005

0.68 (0.52–0.89)

0.04

Educational level

 Primary

322

17.0

59

17.7

NS*****

 

173

13.0

36

12.8

NS*****

  

 Secondary

800

42.2

147

44.1

 

622

46.7

120

42.7

  

 High school or university

774

40.8

127

38.1

 

534

40.3

125

44.5

  

Number of children (mean ± SD)*,**

1.79 ± 0.64

1.64 ± 0.66

<0.001

 

1.72 ± 0.99

1.53 ± 1.34

<0.001

  

Primiparity

599

31.7

121

36.3

NS*****

1.24 (0.97–1.58)

793

59.5

157

55.9

NS****

0.86 (0.66–1.12)

<0.001

Unplanned pregnancy

219

11.6

60

18.0

0.002

1.68 (1.23–2.30)

265

19.9

125

44.5

<0.001

3.22 (2.46–4.23)

<0.001

Unwanted pregnancy

68

3.6

18

5.4

NS*****

1.54 (0.90–2.62)

30

2.3

0

0

0.006

0.98 (0.97–0.98)

NS****

* Continuous variables displayed as means ± standard deviation (SD)

** Comparison of continuous data in two distinct years with Mann–Whitney U test

*** P value, odds ratio and 95% confidence interval of comparison of categorical data with Fisher exact test or Chi-square test

**** P value for the Mantel–Haenszel test

***** Statistically not significant

A slightly lower educational attainment was observed in PND mothers in 1996, whereas in 2006 the distribution of primary, secondary and tertiary education was almost the same in the PND and the non-PND groups. There was practically no difference in the rates of primiparity between the PND and non-PND groups in either year; however, there was an increase in rates of primiparity of about 20% in the total sample in 2006 compared to 1996. Importantly, pregnancy planning was a factor highly distinctive between the PND and non-PND mothers both in 1996 (rate of unplanned pregnancy: 18.0 vs. 11.6%, respectively) and in 2006 (44.5 vs. 19.9%, respectively). As expected, the occurrence of giving birth following an unwanted pregnancy was relatively small, as were the differences observed between the PND and non-PND groups in this regard; however, in 2006, this difference became significant.

From the prevalences of a variable of interest in the depressed and the non-PND groups, an odds ratio can be calculated for 1996 and another one for 2006. These two odds ratios can then be compared by the Mantel–Haenszel test (Table 1). Although mothers with PND were generally less likely to be married relative to those without, this was more so in 1996 than in 2006. In 1996, the PND mothers were more likely than those without PND to be primiparous (OR = 1.24), whereas in 2006, the likelihood seemed to have changed direction (OR = 0.86), a difference which was statistically significant. The increase in the odds for mothers with PND of having had an unplanned pregnancy as compared with those without PND was smaller in 1996 than in 2006. Relative to a wanted pregnancy, the OR of PND after an unwanted pregnancy was 1.54 in 1996, whereas in 2006 it was 0.98, although this was not a significant difference.

The factors potentially associated with PND are listed in Table 2. There was a significant association between a past history of a major depressive episode and PND both in 1996 and in 2006. Interestingly, the association between major life events and PND was only significant in 2006 when the likelihood of developing PND was nearly four times higher in those with a major life event during the year prior to testing. It seems that an unstable relationship made PND more likely in both years; however, this was significantly more so in 2006. Regarding the perceived lack of support by the partner, the difference between the 2 years was even more pronounced (OR = 26.48 vs. 6.66, in 1996 and 2006, respectively). The lack of family support seemed consistently to increase the risk of PND.
Table 2

Psychosocial determinants concerning PPD in the study groups in 1996 (N = 2,229) and in 2006 (N = 1,613)

 

Participants in 1996 (N = 2,229)

P value*

OR (95% CI)*

Participants in 2006 (N = 1,613)

P value*

OR (95% CI)*

P value**

Non-depressed mothers (<12 points in the LQ) (N = 1,896)

Mothers with PPD (≥12 points in the LQ) (N = 333)

Non-depressed mothers (<12 points in the LQ) (N = 1,332)

Mothers with PPD (≥12 points in the LQ) (N = 281)

n

%

n

%

n

%

n

%

History of major depression

99

5.2

41

12.3

<0.001

2.55 (1.73–3.74)

75

5.6

26

9.3

0.029

1.71 (1.07–2.72)

<0.001

Major life events in past year

325

17.1

53

15.9

NS**

0.92 (0.66–1.26)

288

21.6

141

50.2

<0.001

3.65 (2.79–4.77)

<0.001

Self-reported low financial income

338

17.8

63

18.9

NS**

1.08 (0.80–1.45)

132

9.9

81

28.8

<0.001

3.68 (2.69–5.04)

<0.001

Unstable relationship

95

5.0

44

13.2

<0.001

2.89 (1.98–4.21)

29

2.2

48

17.1

<0.001

9.26 (5.72–14.98)

<0.001

Lack of support by partner

38

2.0

117

35.1

<0.001

26.48 (17.89–39.2)

39

2.9

47

16.7

<0.001

6.66 (4.26–10.41)

<0.001

Lack of support by family

333

17.6

85

25.5

0.001

1.61 (1.22–2.11)

517

38.8

142

50.5

<0.001

1.61 (1.24–2.09)

<0.001

Intention to return to work after postpartum period

802

42.3

127

38.1

NS**

0.84 (0.66–1.07)

639

48.0

100

35.6

<0.001

0.60 (0.46–0.78)

<0.001

Independent style of management of problems of everyday life

954

50.3

204

61.3

<0.001

1.56 (1.23–1.98)

768

57.8

185

65.8

0.011

1.41 (1.08–1.84)

<0.001

P value, odds ratio and 95% confidence interval of comparison of categorical data with Fisher exact test or Chi-square test

** P value for Mantel–Haenszel test

*** Statistically not significant

A poor relationship between the mother and her partner was a risk factor for developing PND in both years but in 2006 only to a lesser extent than in 1996, and, similarly, the impact of a perceived lack of support from the partner in the postpartum period, although still a risk factor, appeared less harmful in 2006 than 10 years earlier. Perceived support by the family appeared to decrease the likelihood of PND in both 1996 and 2006, and although the level of perceived support reported by all mothers fell by about 20–25% over this period, the proportions of those who felt supported within the PND and healthy mothers did not change significantly. A self-reported low income, similar to recent major life events, only seemed to have an influence on the prevalence of PND in 2006. Relative to non-depressed mothers, significantly more mothers had an independent style of management of everyday life problems in the PND group in both years of interest, with the Mantel–Haenszel test suggesting a less powerful effect in 2006 relative to 1996.

Simultaneous risk factors of PND as determined by logistic regression analysis

Table 3 presents the logistic regression models, including parameters and corresponding adjusted odds ratios (AORs) for PND as compared with the non-PND women in 1996 and 2006.
Table 3

Stepwise multiple logistic regression analysis for evaluating postpartum depression in the study groups in 1996 (N = 2,229) and in 2006 (N = 1,613)

Variable

Participants in 1996 (N = 2,229)

Variable

Participants in 2006 (N = 1,613)

P value

Adjusted OR**

95% CI

B***

P value

Adjusted OR**

95% CI

B***

Age

<0.001

0.957

0.90–0.98

−0.066

Primiparity

0.01

1.49

1.10–2.03

0.400

Married marital status

0.001

1.83

1.32–2.65

0.58

Unstable relationship

<0.001

3.84

2.22–7.14

1.275

Type of residence*

    

Lack of support by partner

<0.001

1.72

1.28–2.32

1.04

 Town

0.019

11.26

7.44–12.20

2.51

Independent management of problems in everyday life

0.014

1.46

1.08–1.97

0.378

 Village

0.05

6.45

3.90–9.12

2.02

Unplanned pregnancy

<0.001

2.04

1.47–2.86

1.01

Unstable relationship

<0.001

3.10

2.65–3.85

1.67

Major life events

<0.001

3.38

2.49–4.51

1.20

Lack of support by partner

<0.001

3.65

1.22–5.21

1.34

Self-reported low financial income

0.001

1.82

1.22–2.76

0.602

Lack of support by family

<0.001

4.34

2.70–6.13

2.54

Intention to return to work after postpartum period

<0.001

0.47

0.34–0.65

0.714

Unplanned pregnancy

<0.001

1.71

1.13–2.32

2.00

     

History of major depression

0.017

1.79

1.11–2.90

0.585

     

Independent management of problems in everyday life

<0.001

1.59

1.22–2.10

0.52

     

95% CI 95% confidence interval

* Compared to residency in outlying area

** Adjusted odds ratio

*** B coefficient

We first turn our attention to our findings in 1996. Regarding the factors that simultaneously affect PND, there was no significant association between PND and the number of the children of the mother, despite the difference experienced in the single comparison in 1996. The number of children is a confounder, with a high correlation with other variables in the model, such as age and marital status, which did not remain in the model in a stepwise multiple logistic regression analysis. Higher maternal age had a small but significant risk-reducing effect on PND. It is of note that married women also had higher odds of PND, which is contrary to what we found in our univariate analyses. Although, in general, higher age and higher number of children were associated with lower risk of PND, in married women the direction of these relationships was the opposite. Mothers with PND were much more likely to live in a town or city or in a rural area, than on remote farms outside a community. Like in the univariate analyses, an unstable relationship and the lack of perceived support from the partner or the family had strong effects on PND in the combined analysis, with increased odds ratios. An independent style of management of the everyday problems and a history of depression posed a risk for PND, although their effect was smaller than that of the previous variables. Unplanned pregnancy seemed to increase the odds of PND to a relatively small extent. There were no collinear variables in the analysis.

Now moving onto 2006, an unstable relationship appeared to be the strongest predictor in a multiple regression analysis 10 years later. The AOR of major life events within the last 12 months for PND was remarkably high at 3.38. Unplanned pregnancy also appeared to contribute to a higher likelihood of PND. A perceived lack of partner support and independent management of everyday problems increased the likelihood of PND. Importantly, in 2006, a self-reported low financial income appeared to have become a significant predictor of PND. The risk of PND among mothers who wanted to go back to work soon after the postpartum period was about half that of those who did not wish to do so. It is of note that although the number of primiparous mothers was not different in the PND and non-PND groups, in the multivariate analysis primiparity did emerge as a significant risk factor in 2006. Primiparity was a stronger factor than the number of children and was the variable that remained in the model out of these two. Being married which was associated with a lower PND risk in a simple correlation analysis did not remain in the model. It is worth noting that in 2006 the proportion of those cohabiting as opposed to being married at the time of delivery was higher. Also, women tended to have their children at an older age and the difference in the average ages of the PND and non-PND groups disappeared. Out of the effect of the lack of support by the partner and an unstable relationship, the former was a confounder of the latter and got excluded from the model. Unwanted pregnancy was excluded from the model due to the influence of an unplanned pregnancy. The slight univariate difference between the PND mothers and those without PND concerning a previous depression history persisted in 2006, but when it was entered into the final multiple regression model, it did not prove to be a simultaneous risk factor. There were no collinear variables in the model in 2006.

Discussion

Methodological issues

Our study has some strengths that deserve mentioning. As far as we are aware, this is the first study to use a logistic regression analysis model to evaluate the determinants of PND in a large community in Eastern Europe, with a longitudinal approach. The duration of the sampling periods was the same in 1996 and in 2006. There was however a decline in the number of live births in Hungary over this 10-year period, and the smaller number of women in our 2006 group, due to our sampling method of including every woman giving birth in our region during the sampling period, reflected this. Robust, simultaneous estimates of multiple risk factors were possible due to the fairly extensive data set collected from each participant. Conducting a personal interview enabled us to reduce the likelihood of misestimated PND and the overestimation of the observed risk factors relative to the results of self-reported questionnaire surveys [5, 10]. The interview conducted by trained health visitors reduced the chances of participants not being completely honest while giving their answers [2, 3]. Moreover, the participants were spared from reading the questionnaire which might have otherwise caused distress to some of them, especially those ‘at risk’ of depression [36].

Another strength of our report is that we performed prospective repeated cross-sectional measurement to acquire the data. Eberhard-Gran et al. [15] pointed out that many published predicted value estimates are exaggerated, because they are measured in higher-prevalence populations, such as specialist clinics. We studied a large representative sample of the South-East Hungarian postpartum population [20], which was not significantly different from the national average in terms of socio-demographic variables.

It is important to note that, relative to similar studies, only a small proportion of the eligible women declined to participate in the study (19.6 and 16.7% in 1996 and in 2006, respectively). Also, in contrast to other studies, we surveyed a chosen set of data exploring the effect on the mothers of their intention to return to work, their financial difficulties, or their perceived support from their social environment.

It was a limitation of our study that we did not conduct diagnostic clinical interviews with psychiatrists to establish a clinical diagnosis of depression [10]; however, our previous study had proved that the LQ had excellent sensitivity and specificity for PND when compared against a full psychiatric assessment.

Risk factors of PND in the light of social and economic changes in Hungary

Our findings suggest that the prevalence of PND among Hungarian women is not higher than that in most Western populations [1, 35, 16]. Nevertheless, we think it is important to draw attention to the rising trend in Hungary (from 15.0% in 1996 to 17.4% in 2006) and relate our results to the social and economic context. Among the few relevant studies that have been reported, Chinese investigations dealing with the effect of socioeconomic transformation indicated a significantly lower prevalence of PND [31] prior to 2005 [38]. Although the direction of this change is consistent with our findings, concerns have been raised [31] about comparing new data with that from several decades ago due to methodological differences [25].

After the disappearance of the ‘iron curtain’, marked sociocultural changes occurred in Hungary. These included access to a wider repertoire of contraceptive methods, which induced a long-term, stable decrease in reproductive figures [23]. The overall number of pregnancy terminations has declined in parallel with the number of live births during the past decade [20, 23]. Similar to Western European countries, there has been a shift in Hungary towards delaying childbirth, due to the more career-oriented lifestyle of women, besides other social and economic reasons [20, 24]. In the middle of the 1990s, around 80% of unintended (unplanned and unwanted together) pregnancies were prevented either by modern contraception or terminated by abortion in Hungary. Merely 2% of live births could be classified as unwanted pregnancy in 2002 [20]. Our corresponding data for unwanted pregnancy were 3.9% in 1996 and 1.8% in 2006. It is of note that we detected a marked increase in the rate of unplanned pregnancy, from 12.5 to 24.2%, which is in line with the 17% reported for 2002 [20].

Although the overall purchasing power parity values for Hungary increased from 1996 to 2006 [9,323 vs. 18,250 USD; http://indexmundi.com/hungary/gdp_per_capita_(ppp).html], in 2006 more families fell into the lower economic bracket than 10 years earlier and more mothers were compelled to return to work after giving birth due to the unfavourable socioeconomic circumstances [21]. Clearly, there is a growing gap between the well-to-do and the poor and to understand the complex relationship between the changes caused by this and prevalence of PND would merit another study.

The family restructuring trends mean that families now tend to raise fewer children, at a later age, after a more risky pregnancy [20, 24], which can give rise to PND more easily [11, 33]. It is of relevance that an average of 1.6 children is born in families in Hungary and the mother is at least 27 years of age at the first childbearing [20]. The mothers in our sample were significantly older in 2006 (27.8 ± 5.01 years) than in 1996 (24.2 ± 4.64 years), with fewer children (1.71 ± 0.64 in 1996 vs. 1.65 ± 1.24 in 2006). Apart from the increasing jeopardy of the instability of relationships, the decrease in fertility with increasing age could also be a real risk [24].

10 years later, primiparity seemed to be a more important predictive factor for PND. Mothers who have undergone severe episodes of a PND may be less likely to have further children and the initial transition to parenthood is a greater psychosocial stressor than subsequent deliveries [4, 18, 33, 37]. A first pregnancy at an advanced age is more likely to result in delivery complications and consequently lead to PND [24].

Every third Hungarian child was born out of wedlock in 2006, whereas this rate was only 22.6% in 1996, Hungarian women nowadays being less likely to be married. Nowadays, more than half (52.3%) of marriages end with divorce. The mothers in our survey reflected this tendency (the rate of being married at the time of delivery was 77.25% in 1996 and 66% in 2006). Our univariate analyses demonstrated that economic factors (reported low income and intention to return to work after the postpartum period) became significant determinants of PND risk and factors previously associated (e.g. age and type of residence) ceased to be important. Our findings regarding income was in keeping with that of Segre et al. [35] who had found that income was the strongest predictor of PND out of the following variables: income, occupational prestige, marital status, and number of children. In accord with previous studies, we observed that the presence of a previous minor/major depressive disorder was significantly associated with postpartum depression [3, 18, 33]. The relevance of independent management of everyday problems has decreased but still remains significant [39].

When the features were examined simultaneously (multivariate analyses), family-related risk factors (e.g. an unstable relationship, and lack of partner support) appeared important in both years despite the above-described changes in marital status of the mothers, findings consistent with previous research [6].

Family factors

To our great surprise, in our study, a married status played an important role in triggering rather than protecting from PND, whereas others concluded that it is living alone that is associated with a higher chance of PND [3, 14, 18]. A possible explanation for this paradox is the trend of raising a child in a cohabitant relationship, which does not necessarily mean that the relationship is unstable. An unstable relationship with the partner had a detrimental effect in terms of PND risk in both years [18, 25, 33, 3638], and lack of support by the partner fell out from the logistic model in 2006. It is evident that a lack of support may force a mother to try to overcome her everyday problems alone, which makes it necessary for the mother to put her feelings in her pockets. With an increasing trend in Europe for couples not to get married, we predict that the stability and quality of the relationship will be likely to take over the protective quality against PND of being married which has often been reported in the literature, as not being married is becoming more common and probably less of a sign of relationship problems and a risk factor for depression. Based on our results, it is possible to predict that marital status in itself will be a less influential predictor of PND in modern society.

Research has consistently demonstrated in China that rituals such as ‘peiyue’ (the mother-in-law moves in with the family after the birth of the baby to help the mother with household chores) has a strong influence on PND [25], whereas the formerly supposed but lately disproved ‘doing the month’ (the traditional rest period of the mother after childbirth) does not have any effect on PND [38]. Our study also showed that perceived support from the family environment can help the mother, a finding also consistent with Robertson et al. [33]. However, it is difficult to assess causal relations, because from our data it is not possible to determine whether PND is caused by a lack of support or whether PND distorts a depressed mother’s perception of her environment’s attitude.

Conclusion

This study draws attention to a likely increase in the risk of PND over the studied 10-year period in Hungary, but we are far from completely unravelling the complex social, economic, and psychological background of PND with its public-health consequences. This increase in our Hungarian sample raises the question of the need for a different approach for the detection of PND risk in Eastern Europe relative to Western societies. The prevalence rate in our study in 2006 is similar to that reported in a Polish survey [28]. Overall, it appears that in Hungary, socioeconomic effects are beginning to play a significant role, besides biological effects, and this might be important for health-care providers screening vulnerable mothers and attempting to prevent PND. This study is the first and only wide-ranging survey in Hungary, and there was no systematic screening and therapy for PND before. The size of the problem and the increasing trends strongly argue in favour of universal screening in the postpartum population. The LQ administered by trained health visitors and nurses could be a suitable instrument [12].

Improvement of family functioning can be promoted by family support for mothers in the postpartum period. Training among public-health nurses offers an opportunity for detecting mothers at risk or mothers with PND for counselling, other psychological interventions, or treatment with antidepressant medications. Preventive efforts need to be focused on the enhancement of family relations and protection from unplanned pregnancy. Improving the financial security of mothers in the postpartum period, irrespective of marital status, appears to be desirable.

Future studies should perhaps concentrate on three social factors and their influence on PND: (a) the narrowing reproductive potential and the trend towards later childbearing, leading to an unfavourable obstetric outcome [24] and the rising rate of infertility [20, 24]; (b) the growing instability of partner and family relationships, resulting in declining support for the mother and (c) the socioeconomic need to return to work of women after the postpartum period, in association with a lower income. We suggest that future epidemiologic studies focus on multilevel determinants involving social, demographic, economic, and obstetric variables in Eastern Europe.

Acknowledgments

We would like to express our appreciation to Professor Péter Szeverényi for his most helpful comments on earlier drafts of this paper.

Copyright information

© Springer-Verlag 2010