Social Psychiatry and Psychiatric Epidemiology

, Volume 43, Issue 4, pp 266–272

Adverse parenting as a risk factor in the occurrence of anxiety disorders

A study in six European countries

Authors

    • Health Economics Research Unit, Dept. of PsychiatryUniversity of Leipzig
  • Herbert Matschinger
    • Dept. of PsychiatryUniversity of Leipzig
  • Sebastian Bernert
    • Dept. of PsychiatryUniversity of Leipzig
  • Jordi Alonso
    • Health Services Research UnitInstitut Municipal d’Investigació Mèdica
  • Traolach S. Brugha
    • Dept. of PsychiatryUniversity of Leicesterm
  • Ronny Bruffaerts
    • Dept. of PsychiatryUniversity Hospital Gasthuisberg
  • Giovanni de Girolamo
    • Dept. of Mental HealthAUSL Città di Bologna
  • Sandra Dietrich
    • Dept. of PsychiatryUniversity of Leipzig
  • Matthias C. Angermeyer
    • Dept. of PsychiatryUniversity of Leipzig
  • The ESEMeD/MHEDEA 2000 investigators
ORIGINAL PAPER

DOI: 10.1007/s00127-007-0302-0

Cite this article as:
Heider, D., Matschinger, H., Bernert, S. et al. Soc Psychiat Epidemiol (2008) 43: 266. doi:10.1007/s00127-007-0302-0

Abstract

Objective

The aim of the study was to test the homogeneity of the association between adverse parenting and anxiety disorders within these disorders as well as among six European countries.

Method

Based on data from 8,232 respondents (part II sample) originating from the European study of the epidemiology of mental disorders (ESEMeD), we examined the association between three dimensions of parental rearing (care, overprotection, authoritarianism) measured by a short form of the parental bonding instrument (PBI) and anxiety disorders by computing one logistic regression model per disorder.

Results

A similar pattern of recalled parenting behaviour across the four anxiety disorders assessed was found, with care and overprotection having the strongest associations. There were only minor country-specific variations of this pattern.

Conclusion

Our results suggest an association between adverse parenting and the risk of anxiety disorders in particular as well as psychiatric disorders in general that is rather non-disorder specific.

Key words

anxiety disordersparental bonding instrumentepidemiological study

Introduction

Much evidence supports the important role psychosocial factors play apart from biological factors in the aetiology of anxiety disorders [14, 15]. Among these factors, the way a person has grown up is considered a potential vulnerability factor [34]. The parental bonding instrument (PBI) [32] has made systematic research on this topic possible.

Most of the studies using the PBI to assess parental rearing styles have explored the association between parenting and the occurrence of major depression in adulthood [9, 20, 27, 29, 33, 35, 37, 38]. In contrast, the association between parenting style and the occurrence of anxiety disorders was examined in only a few studies, with most of them focusing exclusively on panic disorder (PD) or panic disorder with and without agoraphobia (PDAG) [10, 25, 39, 43]. Three studies examined the relationship between additional anxiety disorders (generalized anxiety disorder (GAD), social phobia (SOP) and simple phobia also denominated as specific phobia (SP) according to DSM-IV) and parenting [9, 16, 19].

Conducted in different countries, the results of the studies are somewhat inconsistent. With regard to PD, the reported care scores for both parents were significantly lower than for healthy control subjects. Exceptions to this were Silove et al. [39], Manicavasagar et al. [23] and Turgeon et al. [41]. Silove et al. [39] and Faravelli et al. [10] found higher maternal overprotection, while Leon and Leon [19] reported only paternal overprotection relevant for the occurrence of PD in their Columbian study. Manicavasagar et al. [23] and Pachierotti et al. [25] completely failed to find higher overprotection scores for both parents of PD-patients. The remaining studies found positive associations between overprotection for both parents and subsequent occurrence of PD. Concerning the association of the PBI with anxiety disorders other than PD, Parker [26] reported lower care and higher overprotection scores for both parents of SOP patients. Leon and Leon [19] found GAD patients to score both of their parents as uncaring and overprotective. Kendler et al. [16] used a three-factor PBI model with the additional dimension of authoritarianism and failed to find significant associations between this particular parenting style and GAD as well as phobia, when including all of the three parenting dimensions in their logistic regression model. Enns et al. [9] found significant associations between SP and the following: paternal care, maternal overprotection in men and maternal care in women.

An investigation of the association between parental child rearing styles and anxiety via a statistical account of the cultural dependence of this linkage could give some clarification on the varying results stemming from different countries with different cultural backgrounds (Australia, United States, Italy, Canada (French part), Columbia SA). Related to this issue is a study by Carter et al. [5]. The authors compared the association between maternal rearing style—measured by the PBI—and anxiety as well as depression in African and European American college students, finding identical levels of care and overprotection in both groups. While anxiety and depression were positively associated with overprotection in European Americans, they found no significant association, however, between anxiety and depression on the one hand and overprotection among African Americans on the other hand.

Aims of the study

We desired to investigate the association between parental child rearing styles measured by a short form of the PBI and four anxiety disorders: SOP, SP, GAD and PDAG. The present study, using data from the ESEMeD, aims to answer two main questions:
  1. 1.

    Does this association vary with respect to particular anxiety disorders in specific countries?

     
  2. 2.

    Do we find any cross-cultural variations of this association in any specific disorder?

     

Materials and methods

Recruitment procedure

ESEMeD is a cross-sectional study in a stratified multi-stage random sample of 21,425 adult respondents (aged 18 years and older) living in non-institutional settings in six European countries (Belgium N = 2,419, France N = 2,894, Germany N = 3,555, Italy N = 4,712, Netherlands N = 2,372, Spain N = 5,473) [1, 8]. Using computer-assisted interview (CAPI) techniques, the respondents were interviewed at their homes, using a revised version of the Composite International Diagnostic Interview (WMH-CIDI) [18] which is subdivided into 38 different sections. The interview was divided into two parts. Including screening questions for some specific disorders of anxiety and mood, Part I was administered to all 21,425 respondents. All participants responding positively to any of the screening questions had to complete the CIDI section of the specific disorder prompted by that question according to DSM-IV criteria. Part II was administered to Part I respondents meeting the lifetime criteria for any core disorder and to a probability sub-sample (25%) of other Part I respondents not meeting any criteria [8].

The parental bonding instrument

The original PBI [32] was intended to measure the perceived parental rearing styles of overprotection and care as remembered by the respondents during their first 16 years of life. It consists of 25 items to be assessed separately for mother and father. In a number of studies the PBI proved to be a reliable and valid measure of actual and not merely perceived parenting [21, 28, 30, 35]. The long-term stability of the PBI has been demonstrated by Wilhelm et al. [44] in a non-clinical cohort study. Recently some studies have used confirmatory factor analyses to show the pre-eminence of three-factor models of the PBI with the additional dimension ‘Authoritarianism’ [7, 17, 24, 36]. In a previous study the authors were also able to confirm this three-factor structure for the nine-item short form of the PBI used in the ESEMeD [13]. Table 1 provides an overview of the nine items and their assignment to the three PBI-factors.
Table 1

PBI-items and their assignment to the PBI factors of maternal/paternal care, overprotection and authoritarianism (further details in [13])

Maternal/paternal care

How much did she/he understand your problems and worries?

How much could you confide in her/his about things that were bothering you?

How much time and attention did she/he give you when you needed it?

How much love and affection did she/he give you?

Maternal/paternal overprotection

How overprotective was she/he?

How much did she/he baby you?

Maternal/paternal authoritarianism

How strict was she/he with her/his rules for you?

How consistent was she/he about the rules?

How much did she/he expect you to do your best in everything you did?

Sample

The PBI was administered to all Part II respondents plus those with problems because of drinking or drugs. While 8,813 respondents completed either the paternal or the maternal scale, there were responses from 8,255 respondents who answered all nine questions of the PBI short scale for both of their parents completely. Because data for the education variable were missing for 23 of them, data from 8,232 respondents (Belgium N = 937, France N = 1,375, Germany N = 1,217, Italy N = 1,685, Netherlands N = 1,007, Spain N = 2,011) remained for analysis. The socio-demographic characteristics of the sample are shown in Table 2.
Table 2

Socio-demographic characteristics of the ESEMeD respondents by country (N = 8232)

 

Total sample n = 8232

Belgium n = 937

France n = 1375

Germany n = 1217

Italy n = 1685

The Netherlands n = 1007

Spain n = 2011

Age: mean (SD)

47.3 (16.8)

47.3 (15.9)

45.8 (15.9)

47.1 (16.2)

46.8 (17.1)

47.1 (15.3)

49.0 (18.4)

Age categories: n (%)

 18–24

632 (7.7)

56 (6.0)

113 (8.2)

94 (7.7)

152 (9.0)

48 (4.8)

169 (8.4)

 25–34

1540 (18.7)

174 (18.6)

266 (19.4)

208 (17.1)

327 (19.4)

197 (19.6)

368 (18.3)

 35–49

2543 (30.9)

308 (32.9)

453 (33.0)

415 (34.1)

494 (29.3)

334 (33.2)

539 (26.8)

 50–64

2003 (24.3)

240 (25.6)

346 (25.2)

301 (24.7)

413 (24.5)

277 (27.5)

426 (21.2)

 65+

1514 (18.4)

159 (17.0)

197 (14.3)

199 (16.4)

299 (17.7)

151 (15.0)

509 (25.3)

Gender: n (%)

 Male

3529 (42.9)

434 (46.3)

582 (42.3)

534 (43.9)

774 (45.9)

409 (40.6)

796 (39.6)

 Female

4703 (57.1)

503 (53.7)

793 (57.7)

683 (56.1)

911 (54.1)

598 (59.4)

1215 (60.4)

Education categories: n (%)

 0–4

473 (5.8)

30 (3.2)

4 (0.3)

6 (0.5)

106 (6.3)

3 (0.3)

324 (16.1)

 5–8

1747 (21.2)

118 (12.6)

*

282 (23.2)

629 (37.3)

133 (13.2)

585 (29.1)

 9–12

2895 (35.2)

372 (39.7)

888 (64.6)

625 (51.4)

288 (13.5)

340 (33.8)

442 (22.0)

 13+ years

3117 (37.9)

417 (44.5)

483 (35.1)

304 (25.0)

722 (42.9)

531 (52.7)

660 (32.8)

Marital status categories: n (%)

 Married/ living with someone

5391 (65.5)

625 (66.7)

873 (63.5)

789 (64.8)

1136 (67.4)

653 (64.9)

1315 (65.4)

 Previously married

1244 (15.1)

164 (17.5)

228 (16.6)

180 (14.8)

145 (8.6)

195 (19.4)

332 (16.5)

 Never married

1597 (19.4)

148 (15.8)

274 (19.9)

248 (20.4)

404 (24.0)

159 (15.8)

364 (18.1)

Living arrangement: n (%)

Not living with someone

1501 (18.2)

199 (21.2)

310 (22.6)

262 (21.5)

173 (10.3)

266 (26.4)

291 (14.5)

Living with someone

6731 (81.8)

738 (78.8)

1065 (77.4)

955 (78.5)

1512 (89.7)

741 (73.6)

1720 (85.5)

Geographical area: n (%)

 Rural (<10000)

2401 (29.2)

156 (16.7)

729 (53.0)

373 (30.7)

609 (36.1)

39 (3.9)

495 (24.6)

 Midsize urban (10,000–100,000)

3569 (43.4)

684 (73.0)

422 (30.7)

409 (33.6)

658 (39.1)

661 (65.6)

735 (36.6)

 Large urban (>100,000)

2262 (27.5)

97 (10.4)

224 (16.3)

435 (35.7)

418 (24.8)

307 (30.5)

781 (38.8)

Employment: n (%)

 Paid employment

4362 (53.0)

529 (56.5)

832 (60.5)

661 (54.3)

912 (54.1)

581 (57.7)

847 (42.1)

 Unemployed

561 (6.8)

68 (7.3)

98 (7.1)

92 (7.6)

125 (7.4)

38 (3.8)

140 (7.0)

 Retired

1701 (20.7)

205 (21.9)

282 (20.5)

273 (22.4)

355 (21.1)

145 (14.4)

441 (21.9)

 Homemaker

904 (11)

60 (6.4)

68 (5.0)

90 (7.4)

203 (12.1)

127 (12.6)

356 (17.7)

 Student

177 (2.2)

12 (1.3)

31 (2.3)

32 (2.6)

24 (1.4)

14 (1.4)

64 (3.2)

 Maternity leave

51 (0.6)

3 (0.3)

13 (1.0)

25 (2.1)

1 (0.1)

1 (0.1)

8 (0.4)

 Illness leave

91 (1.1)

14 (1.5)

19 (1.4)

9 (0.7)

4 (0.2)

12 (1.2)

33 (1.6)

 Disabled

309 (3.8)

38 (4.1)

21 (1.5)

18 (1.5)

46 (2.7)

88 (8.7)

98 (4.9)

 Other

33 (0.4)

2 (0.2)

5 (0.4)

8 (0.7)

10 (0.6)

1 (0.1)

7 (0.4)

 Don’t know or refused

43 (0.5)

6 (0.6)

6 (0.4)

9 (0.7)

5 (0.3)

0 (0.0)

17 (0.9)

*In France, education category 9–12 should be understood as 0–12 (not possible to create smaller groups here)

Statistical analysis

Since the three PBI factors of Care, Overprotection and Authoritarianism had been extracted using only parts of the sample in a previous study, the factor analysis was repeated using the entire data at hand. Being almost identical to the previous study, the results will not be presented in this work (for results and detailed information about the items used see [13]). The 2 (maternal and paternal) × 3 (child rearing styles) factor scores of the PBI, centred at their mean, were used as explanatory variables in multiple logistic regression models with each anxiety disorder (SOP, SP, GAD, PDAG) as outcome variable. Additionally, each of these models contains a set of important socio-demographic characteristics and the respondents’ citizenship. In order to test for observable heterogeneity across the single countries, all interaction terms between countries and PBI-scores were added to the regression model. This was done since even non-significant effects of the PBI-scores may exhibit considerable variance. Adjusted Wald-tests determined if the interaction terms as a whole differ significantly from zero per disorder and if there are differences between maternal and paternal parental rearing styles. The effects of the explanatory variables are presented by means of odds ratios.

As for the testing of the country effects, effect coding was chosen over the more popular dummy coding. By introducing a new reference category, this form of contrasting allows one to test the deviation of each country from an “average European” effect, the so-called grand mean. Because the grand mean functions as a reference category, one of the six country effects does not have to be omitted during the presentation of the computed logistic regression models, as dummy coding would stipulate.

It seemed accurate to present the results of the logistic regression models in two separate tables. Table 3 shows the OR’s of the socio-demographic variables and the country effects; the effects of the PBI dimensions and their interaction terms with the country codes are presented in Table 4. The effects reported in both tables stem from one logistic regression model per disorder. Therefore, the socio-demographic and country effects presented are adjusted by the PBI dimensions and their interactions with the countries, and visa versa. To better understand country variations, we chose an unconventional presentation of the effects in Table 4. We multiplied the OR’s of the single PBI dimensions with their corresponding interaction terms, such as: care mother times care mother-Belgium. This leads to OR’s that can be interpreted in a more simple and straightforward manner. Usually one would interpret in a first step whether the (average) effects of the PBI dimensions differ significantly from zero. In a second step interpreting the interaction terms between the PBI dimensions and the country codes would test whether the effects of the PBI dimensions in each country differ from the average PBI effects. Interpreting both effects on their own is very informative to the reader; thus they were also included in Table 4. However, the combined information of both effects multiplied by their product provides the reader with information about the effect of a single PBI dimension in a specific country at a glance.
Table 3

Effects of socio-demographic variables and countries on the four anxiety disorders, applying one logistic regression model per disorder

Variable

Social phobia (SOP)

Specific phobia (SP)

Generalized anxiety disorder (GAD)

Panic disorder with/without Agoraphobia (PDAG)

 

OR

OR

OR

OR

Sex

1.72**

2.05***

1.46*

1.80***

Age (centered at mean)

0.98***

0.99**

0.99*

0.99**

Income

1.00

1.00

1.00

1.00

Education

1.04*

0.97

1.01

0.97

Belgium

0.54*

0.90

1.22

1.04

France

1.91**

1.90***

1.95***

1.52**

Germany

1.76*

1.95***

0.50*

0.93

Italy

1.01

0.70**

0.82

0.74*

Netherlands

0.88

0.92

1.43*

1.24

Spain

0.61*

0.47***

0.72*

0.74*

*P < 0.05, **P < 0.01, ***P < 0.001

Table 4

Average effects, country specific effects and interactions between countries and PBI dimensions of the four logistic regression models

 

Belgium

France

Germany

Italy

Nether-lands

Spain

Average (Grand Mean)

SOP

 Care mother

0.47** (0.72)

0.69** (1.07)

0.78 (1.21)

0.93 (1.44)

0.49*** (0.76)

0.63 (0.97)

0.65***

 Care father

0.75 (0.99)

0.64 (0.84)

0.88 (1.15)

0.75 (0.98)

0.89 (1.17)

0.69 (0.91)

0.76**

 Overprotection mother

2.17 (1.51)

1.41 (0.98)

1.04 (0.72**)

1.78** (1.23)

1.13 (0.79)

1.39 (0.97)

1.44**

 Overprotection father

0.53 (0.57)

1.33 (1.42*)

1.22 (1.30)

0.77 (0.81)

1.05 (1.12)

0.99 (1.05)

0.94

 Authoritarianism mother

1.30 (1.36)

0.98 (1.02)

0.99 (1.04)

0.89 (.93)

0.83 (0.87)

0.82 (0.86)

0.96

 Authoritarianism father

1.04 (1.03)

1.09 (1.08)

1.00 (0.99)

0.96 (0.95)

1.19 (1.17)

0.85 (0.83)

1.02

SP

 Care mother

1.08 (1.35)

0.89 (1.11)

0.88 (1.10)

0.62** (0.77)

0.66** (0.83)

0.77 (0.96)

0.80**

 Care father

0.82 (0.86)

1.19 (1.26)

0.95 (1.00)

1.17 (1.23)

0.94 (0.99)

0.71 (0.75)

0.95

 Overprotection mother

1.24 (0.97)

1.51** (1.18)

1.04 (0.81)

1.55** (1.21)

1.17 (0.91)

1.25 (0.97)

1.28**

 Overprotection father

0.83 (0.79)

0.76 (0.72)

1.36 (1.29)

1.02 (0.97)

0.94 (0.89)

1.64 (1.56)

1.05

 Authoritarianism mother

1.24 (1.16)

0.97 (0.91)

0.88 (0.82)

1.21 (1.12)

1.14 (1.06)

1.05 (0.98)

1.07

 Authoritarianism father

1.10 (1.15)

1.13 (1.18)

1.11 (1.15)

0.75 (0.78)

0.99 (1.03)

0.77 (0.80)

0.96

GAD

 Care mother

0.73 (0.93)

0.83 (1.06)

0.83 (1.05)

1.07 (1.36)

0.61*** (0.78)

0.73 (0.92)

0.79**

 Care father

0.54* (0.84)

0.78 (1.20)

0.45* (0.69**)

0.68 (1.05)

0.78 (1.20)

0.74 (1.14)

0.65***

 Overprotection mother

1.55 (1.32)

1.39* (1.19)

1.42 (1.22)

0.85 (0.73)

1.33 (1.14)

0.74 (0.63*)

1.17

 Overprotection father

1.18 (1.11)

0.72 (0.68)

0.87 (0.82)

1.02 (0.97)

1.14 (1.07)

1.65* (1.56*)

1.06

 Authoritarianism mother

0.82 (0.80)

1.14 (1.11)

1.07 (1.04)

1.25 (1.22)

1.02 (0.99)

0.93 (0.90)

1.03

 Authoritarianism father

1.12 (0.99)

1.08 (0.95)

1.49 (1.32)

1.23 (1.08)

0.87 (0.76)

1.11 (0.98)

1.14

PDAG

 Care mother

1.06 (1.26)

0.93 (1.11)

0.79 (0.95)

0.76* (0.91)

0.80 (0.95)

0.73 (0.87)

0.84*

 Care father

0.56* (0.73)

0.71 (0.93)

0.76 (0.99)

0.87 (1.14)

0.82 (1.07)

0.92 (1.21)

0.76**

 Overprotection mother

1.34 (1.06)

1.17 (0.92)

1.52* (1.20)

1.43* (1.13)

1.49* (1.18)

0.81 (0.64*)

1.26*

 Overprotection father

1.37 (1.15)

1.37 (1.15)

1.27 (1.07)

0.93 (0.79)

1.15 (0.97)

1.10 (0.93)

1.19

 Authoritarianism mother

0.80 (0.77)

1.02 (0.98)

1.03 (1.00)

1.30 (1.26)

1.20 (1.16)

0.94 (0.91)

1.04

 Authoritarianism father

1.06 (1.11)

1.22 (1.28)

0.98 (1.03)

0.76 (0.80)

0.81 (0.85)

0.94 (0.99)

0.95

The right column shows the average OR’s for the single parenting styles across the countries. The other columns show (in parenthesis) the country-specific differences from the average (interactions), as well as the country specific OR’s for the parenting styles (average times interaction)

Finally the effects of the single PBI-dimensions across the four models were compared by utilizing a procedure known as seemingly unrelated estimation (SUEST) [6, 42]. This procedure, included in the statistical software STATA 8.2 [40], facilitates testing cross-model hypotheses by estimating the covariance matrix of the parameters resulting from several different models. Thus it provides us information on disorder specific differences and how specific parenting styles influence the emergence of the four single anxiety disorders.

Results

Influence of socio-demographic variables on the prevalence of anxiety disorders

With regard to socio-demographic variables, sex and age are significantly associated with all four disorders (Table 3), thus indicating a higher risk for women (two-times higher in case of SP) and a decreasing risk for older respondents of developing an anxiety disorder. The occurrence of the four disorders is not related to respondents’ income. The risk for SOP increased significantly for respondents with higher educational levels.

Variation of anxiety disorder prevalence rates across countries

There were some significant variations of the anxiety prevalence rates among the six countries. In particular, the risk nearly doubled for SOP, SP and GAD in France and for SP in Germany, and the risk was only half as great for GAD in Germany, SP in Spain and SOP in Belgium.

Variation of the association between parental child rearing styles and anxiety across single disorders

An inspection of the average effects of the six PBI-scores in the last column of Table 4 reveals—regardless of the variations between the single countries—‘maternal care’ showing significant inverse associations with the occurrence of all four anxiety disorders. Similar associations are to be observed between SOP, GAD and PDAG on the one hand and ‘paternal care’ on the other. With the exception of GAD, there are significant positive associations between ‘maternal overprotection’ and anxiety. ‘Paternal overprotection’ as well as ‘maternal’ and ‘paternal authoritarianism’ demonstrate no significant average effects at all.

By applying the SUEST-procedure, we found that ‘paternal care’ was the only parenting dimension whose association with anxiety differed significantly across the four disorders F(3, 8,229) = 3.78, Prob  > F = 0.0101.

Variation of the association between parental child rearing styles and anxiety disorders across countries

The adjusted Wald testing of the interaction terms between country-dummies and the PBI-scores showed that these interactions as a whole differ significantly from zero only with GAD (F(30, 8,202) = 1.54, Prob  > F = 0.0309). When looking at the interaction effects between parenting styles and countries in Table 4, it turns out that there were three significant deviations—‘paternal care’ in Germany, ‘maternal’ and ‘paternal overprotection’ in Spain—from the average effect in the logistic regression model using GAD as a dependent variable. With ‘maternal overprotection’ in Germany and ‘paternal overprotection’ in France, there were two significant deviations of SOP. Only one significant deviation (‘maternal overprotection’ in Spain) was found in the PDAG model. Furthermore, there are observable significant country specific effects of single parenting styles only when there is also a significant average effect of this parenting style, as observable, for instance, with maternal care in Belgium, France and the Netherlands. These effects also always worked in the same direction as their corresponding average effects. One exception was paternal overprotection, which shows a significant effect in the GAD model in Spain, but no significant average effect. There were no significant country specific effects of maternal and paternal authoritarianism.

Differences between maternal and paternal parenting

By applying Wald tests no significant differences were found between maternal and paternal parenting dimensions that affected one of the four anxiety disorders. This also holds for the specific dimension of overprotection that increased the risk for anxiety only in cases of maternal overprotection. This does not apply to the country specific effect of paternal overprotection on the occurrence of GAD in Spain.

Discussion

Concerning our first question on variations of the relationship between parenting and anxiety in particular anxiety disorders, we found disorder-specific differences solely in the relationship between paternal care and the four disorders. This finding could stem from the non-significant average effect of paternal care on SP, since the other three disorders are significantly affected by that parenting style.

Our second question addressed differences in the relationship between the three parenting styles and the occurrence of anxiety across the countries under study. With SOP, SP and PDAG non-significant adjusted Wald tests clearly indicate the existence of a consistent relationship between parenting styles and these three disorders across the six European countries. Only the significance of the Wald test with GAD points toward a rather inconsistent relationship between parenting and subsequent GAD across the single countries.

The finding of non-significant differences between maternal and paternal parenting dimensions is complies with Kendler et al. [16] but contrasts with Enns et al. [9], who found maternal parenting generally more consistently associated with psychopathology. Since maternal overprotection was always significantly associated with anxiety and paternal overprotection was not, we suggest interpreting our study’s results in a more differentiated way with differences only in the particular dimension of overprotection. Therefore we must comment that a parenting pattern comprising ‘lack of parental care’ and ‘maternal overprotection’ is mostly homogeneous across anxiety disorders and across the countries under study.

The pattern of associations between parenting styles and each anxiety disorder appears very similar to the one between parenting styles and depression [e.g. 37]. Therefore, we wonder whether the delineated pattern is of a more general nature, making it valid for psychiatric disorders in general. After examining the relevance of three PBI dimensions as potential risk factors for 13 psychiatric disorders in the National comorbidity study (NCS), Enns et al. [9] have suggested that the increased risk of adult disorders, caused particularly due to “lack of care”, is generally non-disorder specific. Kendler et al. [16] point in the same direction. Assessing a sample of female twin pairs and their parents, this study found a pattern of co-morbidity suggesting that: “…much of the impact of parenting on risk for psychiatric and substance use disorders in women may be largely mediated through an increase in risk for major depression and two common anxiety disorders: phobia and GADs”. Parental lack of care and maternal overprotection may therefore represent a general psycho-social vulnerability factor.

Limitations

The cross-sectional nature of these data does not allow a causal interpretation of the results. The comparability with studies using the original 25-item PBI scale is also limited, because an abbreviated PBI-scale was used in ESEMeD. Generally, shortening scales by dropping items decreases reliability. The resulting increase of measurement error and reduced variance of the abbreviated PBI-scale compared to the original 25-item scale has likely led to an attenuation of the observed relationship between PBI and anxiety disorders in our study, along with a possible underestimation of potential country-specific effects and differences between the four anxiety disorders. It is further necessary to emphasize that the PBI is designed for retrospective recall of behavior and therefore is only capable of measuring the perceived parental rearing style. Although it has been shown [31] that the current mood state of respondents does not lead to biased responses to the PBI questionnaire, idealisation or anger against parents tends to influence the recall of parental child rearing. This holds especially for clinical samples where sub-optimal childhood experiences and idealisation or anger against parents appear more often [22]. Nevertheless, these matters will also affect respondents’ answers to a lesser extent in non-clinical samples like the ESEMeD.

Implications

Disregarding possible differences between countries, a similar parenting pattern across the four anxiety disorders was found. With respect to mood disorders, the same pattern has been observed in a number of previous studies. Therefore, the present results are in agreement with Enns et al. [9] and Kendler et al. [16], suggesting that the association between adverse parenting and the risk of psychiatric disorders is rather non-disorder specific. Nevertheless, only with GAD could minor country-specific variations of the parenting pattern be found. Future studies including measurements of culturally varying mediators of the relationship between parental bonding and anxiety into future studies would be helpful, making culture-specific variations of the association between these two constructs causally explicable and more understandable.

As stated by Feldner et al. [11] in their review article, prevention programs for anxiety disorders focused in the past on the modification of existing treatment strategies rather than targeting specific risk factors. Understanding risk factors for anxiety disorders would allow us to develop more targeted interventions. This would also likely increase the cost effectiveness of such programs. The goal should be to strengthen parental care associated with anxiety and other psychiatric disorders, preferably in high risk groups, by means of public mental health programs. These might include intervention programs on an individual level aimed at sharpening parents’ understanding of the basic needs of their offspring, like emotional warmth and protection. For example, the latest efforts of the German Federal Ministry of family, senior citizens, women and young people to establish ambulant help for parenting aim in this direction. In accordance with article 19 of the United Nations convention on the rights of the child [12], this ambulant help is being established to protect children from violence in the form of neglect. Furthermore, strengthening families’ material resources as well as improving the integration of families into society seem appropriate social-political means for the medium-term creation of a caring family environment and thus a long-term reduction of psychiatric disorders. We would argue that only a combination of complementary strategies can suffice to reduce the vulnerability to social disorders by activating the protective effects of parental care. Adjustments of single strategies and their mixture in accordance with cultural differences will be necessary for a successful intervention [2, 11].

Acknowledgments

This project was funded by the European Commission (Contract QLG5-1999-01042); the Piemont Region (Italy), Fondo de Investigacion Sanitaria, Instituto de Salud Carlos III, Spain (FIS 00/0028), Ministerio de Ciencia Y Tecnologia, Spain (SAF 2000-158-CE), Department de Sanitat, Generalitat de Catalunya, Spain, other local agencies and by an unrestricted educational grant from GlaxoSmithKline. ESEMeD is carried out in conjunction with the World Health Organization World Mental Health (WMH) Survey initiative. We thank the WMH staff for assistance with instrumentation fieldwork, and data analysis. These activities were supported by the John D. and Catherine T. MacArthur Foundation, the Pfizer Foundation, the US Public Health Service (R13-MH066849, R01-MH069864, and R01 DA016558), the Fogarty International Center (FIRCA R01-TW006481), the Pan American Health Organization, Eli Lilly and Company, Ortho-McNeil Pharmaceutical, Inc., GlaxoSmithKline, and Bristol-Myers Squibb. A complete list of WMH publications can be found at http://www.hcp.med.harvard.edu/wmh/.

Copyright information

© Springer-Verlag 2008