Abstract
Research into postpartum psychiatric disorders was sparse for most of the twentieth century [1]. This was despite Marce's seminal 1858 treatise on peripartum mental illness [2], with observations that psychiatric illness after child bearing had many unique and distinguishing features, as well as many symptoms appearing in illnesses unrelated to child bearing. This view was at odds with those of Kraeplin [3], Bleuler [4] and emergent syndromal focused systems of classification of psychiatric disorder in the twentieth century, in which the puerperal psychoses were regarded as no different to their non-puerperal counterparts. The American Psychiatric Association's Diagnostic and Statistical Manual of Mental Disorders, First Edition (DSM-I) in 1952 [5] removed the term ‘postpartum’, and recommended that psychiatric disorders after child bearing be classified and named according to the predominant syndrome. Consequently, many clinicians adopted the precepts of the nonexistence of postpartum psychiatric illness as a distinct entity, with childbirth held to simply expose latent functional illnesses, including schizophrenia and schizoaffective disorder. This would not have been a climate conducive to research nor management specifically tailored to puerperal psychiatric disorder had it not been for the formation of mother-baby units in the UK, which sparked a sustained clinical and research interest in this area.
Definitions and classification
The puerperium or postpartum period is variably defined, ranging from weeks to months after childbirth. There is no consensus on the maximum time between childbirth and psychiatric presentation during which an illness should be categorized as puerperal. Tetlow [6] and Paffenbarger [7] regarded this period as six months, whereas Brockington [8] argued that the temporal limits of ‘puerperal psychosis’ (encompassing DSM-IV [9] schizophrenia and schizoaffective disorder, bipolar type) were not known, but thought an arbitrary limit of two weeks postpartum would minimize ‘contamination’ of a research series by other disorders. DSM-IV does not list postpartum psychotic disorders as a distinct category, but rather under psychotic disorder not otherwise specified, where the criteria for other psychotic disorders, such as DSM-IV schizophrenia or schizoaffective disorder, are not met. Both the ninth and tenth editions of the International Classification of Diseases (ICD) [10] pay minimal attention to postpartum psychiatric illness. On the other hand, mother-baby units in general utilize a 12-month definition for the postpartum period. This occurs for reasons of safety of admission, so that toddlers are not placed at risk by virtue of their mobility.
Understandably, varying definitions of puerperal length affect case detection of postpartum psychiatric disorder. Those with shorter times between illness onset and child bearing are possibly biased towards biological or physiological mechanisms of onset and relapse, which are discussed later. Shorter time period definitions potentially underestimate the number of cases of schizophrenia and schizoaffective disorder affected.
People suffering with a more or less equal mixture of schizophrenic and prominent affective symptoms are syndromally classified under schizoaffective disorder in both DSM-IV and ICD-10. Nonetheless, schizoaffective disorder probably represents a heterogenous group of disorders: some with schizophrenia with prominent affective symptoms; others with a mood disorder with prominent schizophrenic symptoms; and others with distinct clinical syndrome [11]. First presentation of these disorders differs from a relapse of a pre-existing disorder, in terms of symptomatology, disability encompassing disruptiveness of acute versus chronic or residual features of illness, and exposure to and experience of mental health services.
Epidemiology
The link between schizophrenia and puerperal relapse in terms of frequency and proximity to parturition is more tenuous than that between relapsed affective psychosis and the puerperium [12, 13].
The relative risk for a woman to be admitted with a postpartum psychotic illness in the first month after childbirth is about 22 times greater than in the two years preceding childbirth: this risk is 35 times more likely after the first baby [13]. The majority of severely psychiatrically ill postpartum women satisfy Research Diagnostic Criteria (RDC) [14] for bipolar, schizoaffective or major depressive disorder [15–19]. Dean and Kendell [17] also found that 80% of postpartum psychotic disorders were affective, with a high proportion of manic disorders. Estimates of the prevalence of psychosis after childbirth have usually been derived from studies of mothers admitted to psychiatric hospitals within a given period postpartum. Such estimates range from 1 to 2 per 1000 deliveries [20, 21]. However, studies of complete populations of child bearing women [22], rather than those only admitted to a psychiatric hospital, suggest a higher prevalence of psychosis of up to 4 per 1000 live births.
Schopf et al. [23] re-examined a group of 57 women, who had been hospitalized for puerperal psychiatric disorders from 1952 to 1977, and found women with RDC schizoaffective psychosis to be particularly susceptible to postpartum onset. The authors also found that puerperal relapses occurred at a significantly higher rate in patients who had experienced non-puerperal relapses. Recurrence rates for affective psychosis in the postpartum period are estimated at 30–50% [24].
Yarden et al. [25] found no difference between the illness course of 67 pairs of child bearing and non-child bearing women with schizophrenia over a five-year period. However, diagnoses were not made with the use of operational criteria, and there was some suggestion that the child bearing group had a worse outcome in the first postpartum year.
Davidson and Robertson's [26] follow-up study of 82 patients treated for mental illness in the puerperium (defined as up to 12 weeks postpartum) between 1946 and 1978, of whom 16% were diagnosed with schizophrenia, appeared to show a relatively strong link between schizophrenia and puerperal relapse. A small sample size and lack of operational criteria were significant weaknesses of the study. McNeil's study [12] of 88 subjects, in which 17 were diagnosed with schizophrenia by his own criteria and a further 14 using the broader RDC, noted a postpartum breakdown of 23.5% in those with schizophrenia compared with 46.7% in affective illness or cycloid psychosis. In this study, women with schizophrenia became unwell later in the postpartum than those with affective illness. Also, subjects with more restrictive definitions of schizophrenia had a lower relapse rate than those with more broadly defined illness (24% vs 36%). This study was limited by the absence of data regarding antenatal clinical condition and treatment. Davies et al. [27] studied a series of 180 admissions to a mother-baby unit and compared 16 subjects meeting Feighner's criteria for schizophrenia [28] and 21 meeting ICD-10 but not Feighner criteria. The authors found that 43% of the broad group experienced an acute illness episode after delivery compared with none of the narrow group, a contrast which wasn't attributable to differences in clinical state or treatment during pregnancy. This data suggested that childbirth could exert a differential effect on the course of illness in severe and more benign forms of schizophrenia, and that the more severe schizophrenic illnesses may not be influenced by the changes associated with childbirth, such as the fall in oestrogen levels. But, the study was retrospective in design and the prevalence of schizophrenia was dependent on the stringency of the diagnostic system used, with ICD-10 requiring a one-month duration of symptoms as compared to six months for Feighner's criteria. The latter also required the exclusion of cases that met the diagnostic criteria for mood disorder or substance abuse, and was biased towards identifying patients with a poorer prognosis by virtue of the criterion of six months of continuous illness. Hence, Feighner's reliable, albeit restrictive criteria, potentially excluded aetiologically similar cases. Additionally, the study population was not random by being drawn from a mother-baby unit and did not include postpartum women with schizophrenia who didn't have significant difficulties in terms of relapse or impaired mother-infant relationship, nor those who were too ill to be admitted to this type of psychiatric ward. Although there was a higher acute illness or relapse rate in the broad schizophrenic group, which potentially included women with schizoaffective disorder, the study was not designed to explore the relationship between schizophrenia and schizoaffective disorder. Terp et al. [29] compared the psychiatric hospital admission rates for women diagnosed with psychosis within 91 days postpartum with non-puerperal women in the Danish population through linkage of the Danish Medical Birth and Psychiatric Central Registers between 1973 and 1993. This study found an increased rate of readmission for women diagnosed with schizophrenia and women with previous psychiatric admission compared with those admitted with other ‘functional’ psychotic disorders. The majority of readmissions in the study were related to patient psychopathology and lack of social support.
Aetiological factors relevant to postpartum relapse
1. Hormonal
The role of oestrogen in prophylaxis against schizophrenia in women has been explored by Kulkarni et al. [30, 31], based on the following observations: international consensus about the later age of onset of schizophrenia in women [32, 33]; better response to neuroleptics in women [34]; more treatment-resistant negative symptoms in men [35]; the vulnerability of women to psychotic episodes during menopause, the postpartum period and at low oestrogen phases of the menstrual cycle [35]; case reports, including that by Dennerstein et al. [36], of improvement of psychotic symptoms in women after synthetic oestrogen and progesterone; as well as Seeman and Lang's [37] hypothesis that oestrogen may protect against early onset of severe schizophrenia in women, including through an antidopaminergic effect [38–40].
Kulkarni et al. [30] performed an open clinical trial of 11 women with acute psychotic symptoms, as scored on the Brief Psychiatric Rating Scale [41], Scale for the Assessment of Positive Symptoms [42] and the Scale for the Assessment of Negative Symptoms [43]. These women had oestradiol added to their neuroleptic treatment for 8 weeks and were compared to seven women with similar symptom severity receiving neuroleptic treatment alone. The group receiving oestradiol had a more rapid improvement in their psychotic symptoms, with both groups reaching similar levels of recovery by the eighth week. Kulkarni et al. [31] subsequently conducted a double-blind placebo controlled study of standardized antipsychotic medication and placebo or transdermal oestradiol. This 28-day study had three groups of 12 actively ill women of child-bearing age suffering with schizophrenia, with psychopathology ratings utilizing the Positive and Negative Syndrome Scale [44] and weekly hormone levels. Preliminary analyses indicated that the group receiving 100 micrograms of oestradiol made greater symptomatic improvement. The findings of Kulkarni et al. [30, 31] might be relevant in the prevention of relapse of schizophrenic illnesses in the postpartum period, during which time oestrogen levels fall in comparison to those in pregnancy and remain low if a mother is breast-feeding. But, hormonal factors may have a limited role to play with respect to symptoms and longitudinal course in women with schizophrenic illnesses when interpreted in the context of previously discussed epidemiological studies exploring the temporal association of relapses with parturition itself. Further research is required to evaluate whether there is a role for oestrogen as a prophylactic treatment for postpartum schizophrenia and schizoaffective disorder.
2. Compliance with antipsychotic medication
In studies of ‘typical’ antipsychotic medication involving samples of patients of both sexes, approximately 40% of patients with schizophrenia cease these within one year [45] and about 75% within two years [46]. About 25% of patients with schizophrenia who are prescribed depot antipsychotic medication stop attending follow-up and no longer have their injections within one year of starting [45]. As many as 75% of patients with schizophrenia, who are non-compliant with their prescribed antipsychotic medication, experience a significant exacerbation of their symptoms over the course of one year in comparison to 25% who comply [47]. This may be particularly relevant to the postpartum relapse of schizophrenic illnesses, because antipsychotic medication may have been ceased (by the patient or their doctor) because of concerns regarding teratogenicity during pregnancy or effects on the infant during breastfeeding.
3. Life events
Life stress has been implicated in the exacerbation of symptoms in almost all psychiatric disorders, including schizophrenia [48]. Epidemiological studies that have looked at postpartum psychiatric illness, such as those of Kendell et al. [15] and Hafner et al. [49], are limited in their capacity to tease out factors that might have had an influence in terms of postpartum illness in a life event as complex as childbirth. Confounding psychosocial factors in such studies make it difficult to separate the independent contribution of variables, or their interacting effects, on postpartum psychiatric disorders.
The prospective study of Marks et al. [50] attempted to address these methodological issues by comparing 43 women with a history of psychosis or severe depression with a group of 45 pregnant control subjects without previous psychiatric disorder, using Paykel's life events interview schedule. After delivery, 51% of those with a history of previous disorder utilizing Research Diagnostic Criteria developed postpartum psychiatric illness, with 28% and 23% being categorized as psychotic and non-psychotic, respectively. The study found that all of the psychotic relapses were in women with a history of schizoaffective or bipolar disorder, and that only the non-psychotic relapses were associated with an increased likelihood of a severe life event in the 12 months preceding illness onset. Although the study suggested that life events could be significant precursors to postpartum illness, particularly non-psychotic disorders and with less importance in the aetiology of more severe psychotic postpartum illness, a major flaw with the study was the exclusion of women with schizophrenia. Additionally, the study did not examine the cumulative effects of life events, other variables such as chronic social, marital and other relationship difficulties, such as being single, and their interactions with relapse rates in the study. On the other hand, a 1-year prospective, longitudinal study of relapse in 30 subjects with schizophrenia found a significant increase in life events in the month preceding relapse, compared to an analogous non-relapse month for the same patients or for the average events per month for non-relapsing subjects [51]. Hence, life events may be relevant in postpartum schizophrenia relapse, but their overall importance in a myriad of other factors, including those biological, is yet to be established.
Impact of schizophrenic illnesses on infants and the mother–infant dyad
Women with schizophrenia are more likely to have multiple partners, no current partner, higher rates of unplanned pregnancy and be victims of violence in pregnancy [52].
There is evidence of mother–infant interactional disturbance in a dyad with a mother suffering with schizophrenia, manifestations of this (i.e. reduced eye contact, lack of stimulation, difficulty or inability to pick up cues and discordance between mother and infant) have lasting cognitive, emotional, behavioural and social consequences for infant development [53–57]. The quality of the mother–infant interaction may be significantly influenced by the ill mother's symptoms, which may lead to a lack of synchronous and contingent behaviours. These are considered to be vital for attachment development [58]. The psychotic symptoms of schizophrenia may affect the mother–infant relationship through a number of mechanisms [59]: through involvement of the infant in delusions, hallucinations or passivity experiences by rendering the mother unavailable to her infant when her symptoms demand preferential attention; through behavioural disorganization; and through abnormal expressions of emotion, such as blunted or perplexed affect. The infant may be incorporated into its mother's delusions, including being regarded as a source of persecution or as a potential victim of persecution, with potential risks of infanticide, neglect and abuse [60]. The distortion of ego boundaries, as postulated to occur in passivity experiences, may hinder the mother's ability to distinguish herself from her infant. Hence, positive symptoms of schizophrenic illnesses disrupt the mother–infant relationship through the presence of unwanted or excessive behaviours. On the other hand, negative symptoms such as affective flattening, avolition and apathy, anhedonia and disorganization may influence the mother–infant dyad through the absence of behaviours which promote a secure attachment [58]. There is no literature which supports the assumption that the resolution of maternal symptoms of mental illness results in normalization of mother–infant interactions. Snellen et al. [59] in their study of a group of 15 schizophrenic mother–infant dyads admitted to an Australian mother-baby unit found that mothers with florid positive symptoms and prominent negative symptoms were at particular risk of displaying disturbed interactions with their infants. These authors also found that the adverse contribution of the mothers’ negative symptoms was not evident until after the positive symptoms had resolved. However, the sample size was small and the study population was heterogenous in terms of symptoms, prominence of deficit features, illness severity and chronicity. They found that more than half of the mothers were single parents and that the majority had not planned their pregnancies. The majority of the mothers retained custody of their infants, which contrasted significantly to the study by Kumar et al. [21].
Kumar et al. [21] found that 9 out of 15 mothers (aged 19–35) with a diagnosis of schizophrenia, taken from a study sample of 100 consecutive admissions to a mother-baby unit, were separated from their infants at discharge or discharged under formal supervision from child protective services. Seventy-eight per cent of the mothers with schizophrenia were single. The authors found that the mothers with schizophrenia had interactional difficulties with their infants and long admissions (a mean of 10.7 weeks, with a minimum of three and a maximum of 25 weeks). The lack of blindness and a control group (e.g. mothers with infants and a diagnosis of schizophrenia managed in the community) as well as the selection bias of mothers and infants assessed to be suitable for a mother-baby unit admission in this study, limited the predictive validity of interactional difficulties being single or suffering with schizophrenia had for determining likelihood of separation or protective services involvement.
Significant rates of mental illness, including schizophrenia, have been found in families who are abusive to their children [61, 62]. Taylor et al. [61] found 42% of the mothers they reported as having mistreated their children suffered from schizophrenia, depression or personality disorder, while Falcov [62] reported that 25% of perpetrators of child abuse were mentally ill. The Victorian Department of Human Services [63] reported that 35% of families reported for abuse had a history of mental illness or substance abuse or dependence in one or both parents. The predictive value of schizophrenia in mothers of infants as a risk factor for child abuse was not able to be ascertained in these cross-sectional reports of a highly selected group consisting of one or both sex parents. The importance of associated factors, including low socioeconomic class, history of abuse and family violence, being single, separated or with a step-parent, were emphasized in Falcov's and the Victorian Department of Human Services’ reports. Although such factors are not specific to postpartum mothers with schizophrenic illnesses, child abuse perpetrated by mentally ill parents often occurs in a troubled social context.
Conclusions
Postpartum mothers with schizophrenia are more likely to have had an unplanned pregnancy, been victims of antenatal violence, or to have multiple partners or no current partner. The number of women with these disorders and dependent children is likely to increase in the setting of deinstitutionalization and availability of sole parent social security benefits. Early mother–infant interactional disturbance in dyads with maternal schizophrenia has lasting cognitive and emotional consequences for infant development. Mental illness, including schizophrenia, is overrepresented in families who are abusive to their children. Accordingly, the capacity of mothers with schizophrenic illnesses to care for their infants is of ongoing concern to mental health, social and child protective services. Mother-baby units are being asked to assess parenting skills in recently delivered mothers suffering from schizophrenia and schizoaffective disorder. Yet, the factors that help these mothers cope, as well as those that contribute to illness relapse, are not well understood.
The identification of coping as well as predictive factors for postpartum schizophrenic relapse have been sparse to date. Moreover, the research evaluating the impact of childbirth and child-rearing on mothers with schizophrenia and schizoaffective disorder has been hindered by a number of historical, classificatory and methodological constraints. The latter have included: a paucity of prospective studies with antenatal, rather than initial, postpartum recruitment; variable definitions of puerperal length; significant changes in psychiatric classification and the breadth of operational criteria; the heterogeneity of postpartum psychotic disorders, with the majority being mood or schizoaffective disorder rather than schizophrenia; selection biases inherent in studies involving only inpatients or mother-baby unit admissions; difficulties in life events research, including its retrospective nature and confounding, illness factors contributing to reported difficulties; and the specificity of childbirth as a unique or discrete life event.
Although potential risk factors for women suffering with schizophrenic illnesses most likely to have delayed postpartum courses include a concurrent affective component to the illness, a history of postpartum relapse, the absence of a partner and higher concurrent stress, further research is required. A clearer understanding of the role of hormones may pave the way for prevention, but until then research needs to look at the recruitment of antenatal women and follow-up through parturition and the puerperium, which has not occurred to date. If clinicians are to better understand which women suffering with schizophrenic illnesses and their dependent infants are most at risk, it is vital to study the impact of childbirth and child-rearing on mothers with initial antenatal recruitment of women with a history of a schizophrenic illness, including community mental health clinics and obstetric hospitals. Postpartum women subsequently admitted to mother-baby and acute adult psychiatry units groups need to be compared with those who relapse and are able to be managed in the community as well as those who do not relapse.
Further study is also necessary to determine whether women with postpartum schizophrenic relapses are likely to have slower recoveries than those women with similar diagnoses but without young children, in order to tailor services more specific to the needs of postpartum women suffering with these illnesses and their infants. Similarly, further exploration of whether a supportive partner is protective against poorer outcome is important in optimizing support available to these mothers. Such directions in research may facilitate the development of education programmes and secondary and tertiary consultation to the relevant mental, primary care, allied health and social services. Hopefully these combined strategies will contribute to the early recognition of difficulties that these mothers and their infants face, more accessible and effective intervention against maternal psychosocial disability and infant developmental delay, and possibly the reduction of child abuse and neglect. However, early intervention must be ethically balanced between the autonomy of postpartum women with schizophrenic illnesses and the duty of care to their children and families, as well as the paternalism and potential stigma of labelling a mother as being at high risk.