Abstract
Women are at greater risk of developing a mental illness following childbirth than at any other time and the risk is further increased for first time mothers [1, 2]. Severe mental illness in the mother is likely to have an adverse effect on the infant who is already genetically vulnerable to mental illness [3]. The child may also be at risk as a result of abnormalities in mother–infant interaction [4] or physical abuse [5] including infanticide [6]. Although information is available on possible long-term adverse effects of postnatal depression on children [7, 8], the impact of severe maternal mental illness in the mother is, in general, poorly understood. M. SALMON, K. ABEL, L. CORDINGLEY, T. FRIEDMAN, L. APPLEBY 557
Joint admission for mother and child in a psychiatric setting was proposed as early as 1948 by Main [9] who believed that both infant and mother would benefit in the long-term. Contemporaries agreed that joint admissions had a positive effect on the mental health of the mother, and also encouraged earlier discharge [10, 11]. Douglas [12] and Glaser [13] proposed that separation of mother and infant should be avoided, as feelings of guilt at her apparent inability to look after her child added to the mother's distress.
During the last four decades at least 12 designated mother and baby units (MBU) for the joint admission of mothers with mental health problems and their babies have been established in the UK [14]. In addition, facilities for joint mother–baby admission are provided in a number of general psychiatric inpatient units [15]. Both units and facilities are primarily for the care and treatment of women with severe post partum psychiatric disorder, but also admit mothers with their infants aged up to 1 year. In recent years they have admitted women with chronic mental disorders for the assessment of parenting skills [16].
Mother and baby units are not unique to the UK. They have developed in other European countries, in particular France, but in some parts of the world such as the US, perinatal provision does not include MBUs. In Australia and New Zealand, MBUs have become an important aspect of perinatal care and a number of studies have been published. Milgrom et al. [17] describe specialist programmes provided by MBUs for women with perinatal disorders, having reviewed the characteristics and outcomes for 36 women admitted to the MBU over a 6 month period. Snellen et al. [18] and Mowry and Lennon [19] report on mothers with schizophrenia and other psychotic disorders detailing their level of interaction with their infants and how this may affect the legal status of the infants at discharge. However, there have been no major clinical trials of the effectiveness of mother and baby units, mainly because of the practical difficulty of randomising acutely ill women to a control group. Although case series have been published [20, 21], there has been no previous study of the determinants of outcome.
In the mid-1990s a number of mother and baby units in the UK agreed to take part in nationwide data collection under the auspices of the Marcé Society, the international association for the study of mental disorders in childbearing women. This paper is the first published report of findings from this project. We report on four aspects of maternal outcome as assessed by clinical staff using the Marcé checklist, namely clinical recovery, practical parenting skills, emotional responsiveness to infant, and risk of harm to infant.
Methods
Eight mother and baby units in the United Kingdom submitted data on joint mother–baby admissions to a central database between 1994 and 2000. During the study period, the number of units nationally varied, but was usually 11 or 12. An additional three psychiatric units where facilities for joint mother–baby admission existed also provided data on admissions. The starting date for submissions varied a little between units, but thereafter each submitted a consecutive case series.
Subjects
The subjects were 1081 inpatients of mother and baby units/facilities admitted with their infants (Table 1) during the study period.
Demographic/social and clinical data on 1081 joint mother and baby admissions
Data collection questionnaire
The data consisted of clinical and social information recorded on the Marcé Clinical Checklist. The Checklist is made up of 10 sections and is completed by senior clinical staff. The first part, completed at or soon after admission, covers demography, psychiatric history and details of referral. The second part, covering symptoms, diagnosis, treatment and outcome, is completed around the time of discharge.
The four outcome items (items 10.3 and 10.8 in the checklist) that follow are the subject of this report. Maternal clinical outcome can be recorded as: (i) symptom free; (ii) considerably improved/symptoms persist; (iii) slightly improved/symptoms persist; (iv) no change or worse. The other outcome items grouped under the heading ‘Maternal skills’ are significant practical problems in parenting, significant problems of emotional response towards infant, and significant risk of harm to infant. These three are rated as present or absent.
Standardization of ratings
Standardization was established for behavioural disturbance (aggressive or odd behaviour, item 7.1), clinical outcome (10.3) and three maternal skills items (10.8), five variables in total. These items were chosen to establish a standard for ratings as they are based on subjective rather than objective assessment. The exercise consisted of requesting staff from the participating MBUs to assess, then rate, a video of four mother/interviewer and staff/interviewer scenarios using the checklist. The proportion of ratings in agreement with our ratings for each variable were as follows: behavioural disturbance 89%; clinical outcome 68%; practical problems in child care 75%; problems of emotional responsiveness 93%; and risk of harm 86%. The overall level of agreement between ourselves and the other raters for all five variables was 82%.
Statistical analysis
The analysis was carried out in three stages. First, descriptive statistics were produced highlighting the main clinical and social variables, including the four outcome variables. Secondly, eight demographic/ social and eight clinical variables (Tables 2–5) were selected as being those most likely to be associated with maternal and/or parenting outcome. Each of the four outcomes (see list above) was then examined in 558 OUTCOMES FOLLOWING MOTHER–BABY ADMISSION relation to each of these potential predictor variables. Thirdly, logistic regression analysis was carried out for each of the four outcomes, with all 16 predictor variables in the model. In both the univariate and multivariate analyses, the four possible responses for the variable of ‘clinical outcome’ were reduced to two: good (symptom free or greatly improved) and poor (slightly improved or no improvement/worse). The number of missing cases for each variable varied between 20 and 40, hence for some variables different totals are used for calculations.
Maternal clinical outcome
Significant practical problems in parenting
Significant problems in emotional response towards infant
Significant risk of harm to infant
Results
The main social and clinical characteristics of 1081 admissions are shown in Table 1. The mean age of mothers was 29 years (range 23–35 years). All diagnoses were determined at discharge by the clinical team and consultant in charge according to ICD-10 criteria. The diagnostic groups included in the study are: schizophrenia n = 224; bipolar affective disorder n = 155; depressive illness n = 411; anxiety/ phobic/panic disorder n = 29; obsessive-compulsive disorders n = 15; alcohol/substance dependence n = 9; personality disorder n = 35; learning disability n = 4; and other n = 74. In the majority of cases the clinical and parenting skills outcomes were good: 78% showing ‘good’ on the clinical outcome variable and at least 80% on each of the three parenting outcomes having no significant problem, according to staff.
Tables 2–5 show the results of the regression analyses for the four outcomes (see above list). A number of the social and clinical variables were associated with each outcome in the univariate analyses. In the logistic regression analyses, the independent predictors of poor outcome were similar for each outcome variable. A primary diagnosis of schizophrenia and behavioural disturbance during the current illness were associated with poor outcome on all four ‘outcome’ variables. Low social class (semiskilled, unskilled, never employed) was associated with poor clinical outcome, practical problems in baby care, and poor emotional responsiveness, but not risk of harm to the baby. Psychiatric illness in the woman's partner was associated with practical problems in baby care, poor emotional responsiveness and risk of harm to the baby. The absence of a good relationship with partner was associated with poor clinical outcome and poor emotional responsiveness. The absence of good relationships with people other than the partner, for example parents, was associated with practical problems in baby care.
Poor outcome on one variable was associated with poor outcome on others. Of the four outcome variables (clinical outcome plus the three parenting variables), 187 (17%) women were reported to have poor outcome on at least one variable; 83 (8%) had poor outcome on two; 64 (6%) on three; and 35 (3%) on four. Of these 35, 23 (66%) had a primary diagnosis of schizophrenia. Fifty-two (5%) women were reported to have a poor outcome on all three parenting skill outcome variables. Of these 52, 34 (65%) had schizophrenia.
Discussion
This is the largest reported sample of joint mother– baby psychiatric admissions, and the sample size is able to support detailed analysis. The results show that for most women admitted with their babies, the clinical outcome is good, and they leave hospital without significant parenting problems. However, in almost one-third of admissions, outcome is poor on one or more clinical or parenting measure. Risk of harm (as perceived by clinical staff) to the infant at the time of discharge is unusual though not rare, occurring in 9%. Poor clinical or parenting outcome is associated with a diagnosis of schizophrenia, behavioural disturbance, low social class, psychiatric illness in the woman's partner and the absence of a good relationship with partner or others.
These findings are consistent with those of Kumar et al. [21], who reported that, in a case-series of women admitted to a mother and baby unit, the mothers with schizophrenia were found to have the poorest outcome, 50% being separated from their babies at discharge. Follow-up studies of mothers with schizophrenia reported that these women are also more likely than those with affective disorders to experience further psychiatric episodes, thus further endangering the relationship between mother and child [22]. Hipwell [23] followed 25 mothers with schizophrenia at 12 months postnatally, and found that the majority of women were not fully recovered until 3–6 months after discharge.
There are a number of limitations to this study. First, because not all mother and baby units in England participated, the sample may be biased. However, as most units carry out similar work, it seems likely that the sample is reasonably representative of joint mother– baby admissions nationally. Secondly, the rating of outcomes may have been affected by a response set with the effect that poor outcome on one variable would lead to a rating of poor outcome on another. However, around half of those with poor outcome on one variable had no other poor outcomes. It would also be expected that poor outcomes on different variables would cluster in the same individuals. Thirdly, although the study demonstrates a number of associations with poor outcomes, we do not know whether these are causal.
Despite these limitations, our findings highlight an association between poor clinical and parenting outcome and partner/husband factors. The occurrence of postnatal depression and of depression in women in the community has previously been linked to the lack of a confiding relationship with the partner [24, 25]. However, in this study, the illnesses were severe enough to require hospital admission and the focus of the study was outcome during hospital stay. The partners of women admitted to mother and baby units have been shown to have a high rate of mental disorder [26]. These results raise the possibility that greater involvement of a woman's partner in the plan of care and treatment could be beneficial, either by improving the relationship through marital counselling or improving the mental health of the partner by treating him in his own right.
The approach to the treatment of severe psychiatric illness in women with young infants now requires further study through clinical trials and we believe that a number of service developments are needed. These include the development of interviews that include the partners of the mentally ill; this strategy has been shown to be beneficial in the treatment of depression [27]. The findings highlight the need for community-based, multiagency perinatal mental health teams [28], allowing the early identification, support and treatment of families with a mentally ill parent. We also need to develop standardized assessments of parenting in the MBU setting. For mothers who have a chronic psychiatric illness, as well as those with an acute illness, there should be access to safe and suitable inpatient resources, which are geographically close to the mother's social and family network [28].
Clinical implications
Maternal clinical outcome and parenting outcome is usually good after joint mother–baby admission.
Women with schizophrenia or behavioural disturbance have poorer outcome and may need more intensive clinical input.
Improving the women's relationship with her partner or giving further consideration to his mental health may improve these outcomes.
Limitations
This was not a controlled trial; good outcome may have occurred without mother–baby admission.
The predictors of poor outcome may not be causal, and any implications for treatment are preliminary at this stage.
Footnotes
Acknowledgements
The project received funding from the Marcé Society, South Manchester University Hospitals Trust (1996– 2000) and Manchester Mental Health and Social Care Trust (2000–2002).