This longitudinal study was carried out in Trieste (Italy), and based at the maternity hospital, where more than 95% of the births in the city take place. From January to April 2004, women were approached on the post-natal ward, and asked to participate in a study on health during pregnancy; confidentiality of the responses was assured. Women who refused were asked to respond to a few questions (nationality, age, type of birth). Women who accepted were interviewed with a questionnaire, and asked if they would agree to respond to a follow-up telephone interview 8 months later. The women gave oral informed consent.
Well-trained female research staff carried out the interviews. Contacts were established with the hospital social worker and with the local Women’s Shelter, in case a woman requested a referral. The study was approved by the Ethics Committee of the Hospital.
We used two different questionnaires. Questions in the post-partum questionnaire covered: social and demographic characteristics, pregnancy intendedness, woman’s health and health behaviour during and before the pregnancy, childbirth, health of the baby and experience of violence.
The second questionnaire, 8 months post-partum, covered: baby’s health, breast-feeding, resumption of sexual activity, physical and mental health, working situation, couple relationship and past and present violence. The main reason for having the follow-up interview at 8 months post-partum was to be able to explore the role of paid work in new mothers’ mental health. In Italy, most women who resume work after birth, do so between 6 and 9 months post-partum [16
There are no specific instruments to investigate violence against women in the post-partum period. To assess violence occurring after the birth, we developed a detailed series of 28 questions, adapted from the international literature and from previous studies on violence against women in Italy [18
], covering four areas of violence: (1) domination and control, (2) verbal abuse (considered both as psychological violence), (3) physical and (4) sexual violence. Some items (like degrading the woman’s competence as a mother), were developed for the specific situation of a new mother. Possible answers were: not at all, occasionally or more than occasionally.
Women were asked to respond to the questions regarding violence, first asking about partner’s behaviour and then about the behaviour of other people, whom the woman was asked to specify. In the context of a couple relationship, the classification of some of these items as violence, such as pressures not to have another baby, taken in isolation, could be ambiguous. Therefore, women were considered to have experienced ‘present partner violence’ when they answered positively to at least two abusive items. As the other perpetrators mentioned by women were exclusively family members, we constructed the variable ‘present family violence’, including women who had answered positively at a least one abusive item. These questions were used to construct a synthetic variable, ‘any present violence’, including ‘present partner’ or ‘present family’ violence at the time of the post-partum interview.
Three questions investigated past violence. Women were asked whether, as a child or an adolescent, they had experienced physical, psychological or sexual violence, and by whom. As most of this violence was perpetrated by family members, we constructed a synthetic variable called ‘past family violence’, including all women who had experienced any violence by a family member in the past.
Present psychological distress was measured with the General Health Questionnaire (GHQ), in its 12-item version, an internationally validated screening instrument [19
], covering anxiety, depression and self-esteem experienced in the last month. It has been widely used with women in the perinatal period [20
] and in studies on the mental health impact of violence, in both cases also with Italian samples [13
]. Although a cut-off point of > 2 positive answers is generally used as a screening measure, a cut-off point of > 5 was chosen for this study to select a group of more seriously distressed women. In the analyses, a dichotomous variable was used: GHQ score 0–5 versus
GHQ score 6–12.
To analyse women’s previous mental health, we used two questions from the first questionnaire regarding ‘frequent feelings’ of depression and ‘frequent feelings’ of anxiety. In the present analyses, we combined these two variables into one: reporting of frequent feelings of depression or anxiety before pregnancy.
In the first questionnaire, we asked whether the pregnancy was: wanted in the same way by the woman and her partner; unwanted in the same way; she wanted it more; he wanted it more; she had almost forced the pregnancy on him; or he had almost forced the pregnancy on her. The question was re-coded in two categories: ‘both wanted to the same extent’, including the first answer; and ‘other’, including all the other responses. Another question concerned couple decision-making on contraception before the pregnancy. Answers were: contraception was (1) mostly decided by the woman, (2) mostly decided by the man, (3) decided together, (4) disagreed on or (5) unnecessary because they wanted a baby. The question was re-coded into two categories. One category, ‘decided together’ included answers 3 and 5; the category ‘other’, included the other answers.
In the second questionnaire, we asked women about the congruence between their wishes and reality concerning current work situation. Answers were: (1) I am at home but would prefer to be working, (2) I am at home and happy with that, (3) I am working and happy with that, (4) I am working but would prefer to stay at home and (5) I am working full time, but would prefer a part-time job. Answers 2 and 3 were recoded into one category, referred to as ‘satisfied’. Answers 1, 4 and 5 were recoded as: ‘dissatisfied’.
After initial examination of bi-variate relationships between demographic, violence and other psychosocial variables and high psychological distress in the post-partum period, multivariate logistic regression was used to examine the relative importance of violence, controlling for other important predictors of distress.