Abstract
The most common reason parents seek medical assistance during the first 3 months of life is unsettled infant behaviour [1], usually involving some combination of irregularity of biological function (feeding/sleeping), a tendency to fussing, crying and negative mood, a tendency to withdraw from new situations and stimuli, and slow adaptability to change. Unsettled infants are variously referred to as ‘colicky’, ‘regulatory disordered’ and ‘temperamentally difficult’. Irrespective of terminology, longitudinal studies suggest that parents of ‘colicky’ infants still report more family distress 3 yearslater [2], [3], that ‘regulatory disordered’ infants may be more vulnerable to later sensory integrative difficulties and regulation of attention during cognitive tasks [4], [5], and that ‘temperamentally difficult’ infants are at increased risk of later behaviour problems [6].
Unsettled infant behaviour causes significant concern to parents and disruption to families [1]. This paper examines relations between unsettled infant behaviour and postnatal depression in mothers admitted to a parentcraft hospital, Tresillian Family Care Centres. Tresillian offers three tiers of assistance to parents: a parent helpline (24-hour telephone advice regarding any concerns parents have regarding their child's development or behaviour); an outreach and day-stay service and a residential care service where mothers and infants are admitted for 5 to 7 days. With the exception of the telephone helpline, all services require referral from a health professional (typically an early childhood health centre nurse or general practitioner). During 1999 therewere 87 289 babies born in New South Wales and therewere 55 000 calls to the parent helpline (primary tier), 2792 clients were seen by the outreach service, 2735 clients used the day-stay service (secondary tier) and 2189 mothers were admitted to one of the three residential care units in Sydney (tertiary tier). Seventy-three per cent of admissions were for unsettled infant behaviour, following unsuccessful community based or daystay intervention, so the residential care group may be regarded as mothers having significant ongoing difficulties managing their infants. The unit is staffed by professionals from child health nursing, social work, paediatric medicine and child and adolescent psychiatry. Referrals are accepted from across the State, but the majority are from the Sydney Metropolitan area.
Unsettled infant behaviour and postnatal depression
A high proportion of mothers admitted to residential care facilities in parentcraft hospitals have been reported to score above the cut-off point (12) for a likely major depressive disorder on the Edinburgh Postnatal Depression Scale (EPDS) [7], with figures rangingfrom 39% [8] to 86% [9]. Studies of community samples have also reported that infants of depressed mothers cry more than other infants [10]. However, it has been difficult to separate the chicken from the egg with respect to difficult infant behaviour and maternal depressivesymptoms [11]. Indeed the cause and effect debate may be futile given the close attunement between mothers and infants at this early stage and the dynamic ways in which they influence each other [12]. On the one hand there is evidence that prenatal depression and anxiety may predict new-born irritability and infant colic (e.g. [13]) and that insensitive maternal responses to infant signals as observed in depressed mothers may result in excessive infant crying [14]. However, other research has demonstrated that infant factors measured soon after birth such as poor motor functioning [15], neonatal irritability [15] and persistent infant crying [16] predict maternal depression. Irrespective of the direction of effect, distress in the parent–infant relationship and related mood disorders in the postnatal period have considerable public health significance given the growing body of research evidence demonstrating possible adverse effects on the social, emotional and cognitive development ofchildren [17–21].
Summary and aims of present study
Both unsettled infant behaviour and mood disorders in the postnatal period are risk factors for parenting problems, infant development and family adjustment, and may be particularly problematic incombination [12], [22]. Although clinicians can attest to the prevalence of these difficulties in mothers and infants admitted to residential facilities, there are few research data available in these settings. Furthermore, while it can be inferred that mothers who seek or are referred to residential care services differ from non-referred mothers with respect to both their own subjective level of distress and their infants' irritability [12], few studies have compared referred mothers and infants with parents who have not sought this level of professional help [1]. Such data have both theoretical and practical clinical implications. This paper has three major objectives: (i) to compare maternal mood state and infant temperament in a sample of mothers admitted to a parentcraft unit (residential care mothers) with a demographically matched comparison group; (ii) to explore differences within the residential care group according to diagnosis of maternal depression; and (iii) to explore relations among the measures.
Methods
Sample
Residential care group
Mothers were recruited from one of the Residential Care Units at Tresillian Family Care Centres in New South Wales. All mothers admitted to Tresillian are routinely screened using the EPDS. For the purposes of this research, all women with a singleton first-born child under the age of 4 months (range = 6–20 weeks, mean = 11 weeks) were screened as usual, but also given information and consent forms regarding the longitudinal study. A psychologist then visited the centre on day two of their admission to explain the research and ask mothers if they were willing to be contacted and visited at home when their infants were 4 months old. One hundred and twenty-eight mothers who agreed (80% of those approached) then completed consent forms.
Comparison group
A comparison group of 58 first-time mothers was recruited through an obstetrician whose practice had the same postcode as the residential care centre (response rate 70%). Comparison mothers were participating in a larger study of the transition to parenthood and were recruited during the third trimester of pregnancy, and subsequently visited at home when their infants were 4 months old. Mothers in the comparison cohort who were admitted to a residential parentcraft unit during the first 4 months postpartum were excluded (n = 4).
Procedure
Self-report questionnaires were mailed to mothers in both groups and they were visited at home when their infants wereapproximately 4 months old (infant age range: residential care group = 16–20 weeks; comparison group = 16–22 weeks). Except where otherwise specified, all questionnaire and interview data were collected from mothers in both groups on this visit.
Measures
Obstetric and perinatal history
Mothers were interviewed when their infants were 4 months old about their childbirth experiences, and information regarding infant birthweight, gestational age and admission to neonatal intensive care was collected.
Depression
Anxiety: Spielberger State-Trait Anxiety Inventory
The Spielberger State-Trait Anxiety Inventory (STAI) [26] is a valid, reliable, self-report scale with 20 items referring to current anxietyand 20 assessing general anxiety levels. Mothers in both groups completed this questionnaire when their infants were 4 months old. Women scoring above 40 on the STAI were defined as ‘highly anxious’ in keeping with an earlier study of Australian childbearing women in New South Wales [21].
Marital adjustment: the Dyadic Adjustment Scale
The Dyadic Adjustment Scale (DAS) [27], a widely used 32-item self-report questionnaire on marital satisfaction, was completed by both groups when their infants were 4 months old. The measure yields overall scores ranging from 0 to 151. The generally accepted cut-off point in non-clinical community samples is 100, with scores below 100 designated as indicating marital distress.
Infant temperament
The Short Temperament Scale for Infants (4–8 months of age) (STSI) [28] measures mothers' perceptions of their infant's behaviour and temperament. It is a 30-item questionnaire yielding five factor scores: (i) approach (the extent to which the child is likely to approach vs withdraw), (ii) rhythmicity (the regularity and rhythmicity of the baby's eating and sleeping behaviour), (iii) cooperation– manageability (the ease with which the baby accepts caretaking such as nappy changes or interruptions to feeds), (iv) activity-reactivity (the extent to which the baby squirms, wriggles or becomes excited by toys or changes of environment) and (v) irritability (the tendency to cry, and soothability). An overall ‘easy/difficult’ score is calculated comprising the sum of the approach, cooperation–manageability and irritability scores. There are normative data for Australian infants [28] and infants are classified as ‘difficult’ if they score more than one standard deviation above the community mean score on the difficulty rating. Given the potential for maternal ratings of infant temperament to be confounded by maternal depression [10], the study also included father ratings for infant temperament in both groups. Both mothers and fathers completed the STSI when their infantswere 4 months old.
Statistical methods
In a first stage of analysis groups were compared on demographic (parental age and education) and perinatal variables to establish the comparability of the two groups on these criteria, which have been demonstrated to be salient predictors of both parental adjustment and child outcomes [29]. Second, distributions of the core variables were examined in each group separately. Unless otherwise stated variables were normally distributed as tested by the Kolmogorov–Smirnov (K–S) statistic. In a third stage, group means were compared using paired t-tests. In order to further explore the within-group differences in the residential care group according to diagnosis of depression, data are presented both for the overall Tresillian group and also for diagnostic subgroups. Because of the relatively large number of comparisons, a conservative p-value of < 0.01 was considered significant.
Results
Comparability of groups on demographic, obstetric history and perinatal variables
While it is acknowledged that this is not a strict case–control design, the two groups were very similar with respect to demographic, obstetric history and perinatal variables. Both groups were somewhat older and better educated than the general population of mothers with firstborn children [30], with no significant difference between the groups on either variable. The mean age for the residential care motherswas 31 years (SD = 4.2), and for the comparison group 32 years (SD = 2.3). A high proportion of mothers in both groups had completed tertiary education: 50% and 55% for residential care and comparison groups, respectively. Nor were there any significant differences between the two groups on infant gender (male: residential care = 49%, comparison group = 47%) rate of caesarean birth (28% in both groups), or number of babies born earlier than 37 weeks gestation (residential care = 9% comparison group = 7%). Only 1% of infants had a birthweight lower than 2500 g and there was a non-significant trend for more residential care infants to have spent time in Neonatal Intensive Care (14% compared with 7% in the comparison group; Pearson χ2 = 4.12, p = 0.11). As the groups did not differ on any of these variables, no covariates were used in subsequent analyses.
Depression
Thirty-six per cent of residential care mothers compared with 6% of comparison mothers scored above 12 on the EPDS (χ2 = 21.7, df = 1, n = 186, p = 0.000). When group means were compared, the residential care mothers scored significantly higher on the EPDS (t = −9.7, df = 186, p = 0.000) but interpretation of these results is somewhat limited by the fact that the residential care mothers completed the EPDS when their baby was younger (mean age = 11 weeks vs 18 weeks for the comparison group). Only the residential care mothers were interviewed with the CIDI (when their infants were four months old). Sixty-two per cent (n = 79) met DSM-IV criteria for a major depressive episode since childbirth and 25% (n = 19) of the mothers meeting these criteria had been prescribed antidepressant medication. Thirteen per cent (n = 16) of the residential care mothers met DSM-IV research criteria for minor depression and 24% (n = 33) reported that they had not been depressed. There was a high agreement between the EPDS coded dichotomously (≥ 12 or < 12) and diagnosis of an episode of depression (κ = 0.44, p = 0.000). The sensitivity, specificity, positive and negative predictive value of the EPDS (cut-off = 12.5) for a diagnosis of major depression were 52%, 94%, 93% and 54%, respectively. With a cut-off of 9.5 these values were, respectively, 75%, 63%, 80% and 55%.
Anxiety
Thirty-five per cent of the residential care mothers were classified as highly anxious compared with 9% of the comparison mothers (χ2 = 13.4, df = 1, n = 186, p = 0.000). When group means were compared, the residential care mothers scored significantly higher than the comparison mothers on both trait anxiety (t = −3.4, df = 186, p = 0.001) and state anxiety (t = −5.7, df = 186, p = 0.000). (Mothers in both groups completed the anxiety state and trait questionnaires when their infants were 4 months old.)
Table 1 shows mean EPDS, A-State and A-Trait scores for the comparison and the residential care mothers. In this and subsequent Tables, the means are presented for the whole residential care group as well as for subgroups according to DSM-IV diagnosis.
Comparison group and residential care means for the Edinburgh Postnatal Depression Scale (EPDS), A–State and A-Trait∗
Marital satisfaction
Fourteen per cent of residential care mothers compared with 10% of comparison mothers reported total scores on the DAS in the distressed range. This difference was not significant, nor were there significant differences when group means were compared.
Infant temperament
With reference to the normative data, 20% of residential care mothers rated their infants as ‘difficult’ (i.e. < 1 SD above the mean community score), compared with 5% of the comparison mothers when their infants were 4 months old, (χ2 = 6.5, df = 1, n = 184, p = 0.01). Comparisons of mean ‘difficulty’ scores showed that while there were no differences within the residential care group between depressed and non-depressed mothers, the residential care mothers' ratings were significantly higher than those of the comparison mothers (t = 4.26, df = 184, p = 0.000). When individual temperament factors were considered, again there were no differences within the residential care group according to diagnosis of depression, but the residential care and control groups differed significantly on irritability (t = 4.6, df = 184, p = 0.000), and cooperation–manageability (t = 4.0, df = 184, p = 0.000), but not on activity/reactivity, rhythmicity or approach. Table 2 shows mean temperament factor scores for the comparison and the residential care mothers. The Table also includes normative values for Australian infants aged between 4 and 8 months.
Comparison group, residential care group and normative mean factor scores∗ for the Short Temperament Scale for Infants (STSI)†
Validity of maternal rating
Given the potential for maternal ratings of infant temperament to be confounded by maternal depression [10], mother and father ratings for the infant difficulty score were compared for both groups. Even though the fathers overall were not depressed (only 7% scored in the clinical range on a self-report measure of depression), they showed a high level of agreement with maternal temperament ratings (residential care group: r = 0.73, p = 0.000; comparison group: r = 0.49, p = 0.000).
Within-group relations among measures
Finally, relations among maternal mood state, marital adjustment and infant temperament measures were examined for each group separately (See Table 3).
Correlations between maternal mood state measures, marital adjustment and maternal ratings of infant temperament for residential care and comparison mothers∗
Table 3 shows moderate to strong correlations between maternal self-ratings of anxiety and EPDS scores and moderate negative correlations between depression scores and marital satisfaction in both groups. Infant difficultness was moderately correlated with EPDS scores in both groups. The strongest correlations were between maternal state and trait anxiety scores and infant difficultness in the residential care group.
Discussion
The study has clearly demonstrated that first-time mothers admitted to family care centres report higher a high incidence of postnatal depression, and that they report higher state and trait anxiety and more difficult infant temperament than a demographically comparable group of mothers. The groups did not differ on marital satisfaction with relatively high levels of satisfaction reported by both groups. These findings will now be discussed with reference to their immediate clinical implications and to possible directions for future research.
Postnatal depression
Direct comparisons between the groups on depression are somewhat limited by a possible underestimation of the prevalence of depression in the comparison group due to the fact that these mothers were not assessed using the CIDI. Nevertheless, the thoroughness of the assessment of depression in the residential care group, the strong relationship between EPDS scores and diagnosis of depression and the size of the group difference (6% vs 36%) is consistent with other reports [8], [9] in suggesting that mothers whose infants are admitted to residential care facilities have a much higher rate of depressive symptomatology or more general psychological distress.
One needs to consider the implications of thelabel ‘postnatal depression’ and the mode of diagnosis, which have been the subject of considerable recentdebate [11], [31]. ‘Postnatal depression’ is a term appearing frequently in the lay literature and readily used by many women themselves. It seems generally helpful to both clinicians and women in identifying that a problem exists so that appropriate supportive interventions may be instigated. However, in a setting where 62% of mothers meet DSM-IV criteria for a major depressive episode, possible pejorative associations with ‘mental illness’ would not be helpful or appropriate, and might even in some cases act as a barrier to women reporting their distress and seeking help. For some of the women identified in this study the episode may be relatively brief or depressive symptoms may be directly related to sleep deprivation. For other women the depression may be severe, with melancholic or psychotic features, or long-lasting. Our prospective design will allow further clarification of these diagnostic complexities. We are also currently investigating relations between maternal reports of depression and night sleep patterns in this group.
The study findings also raise some interesting issues regarding the use of the EPDS. Despite the wellestablished validity of the EPDS there is ongoing discussion (particularly in the UK) regarding the use of the instrument as a screening tool [32]. In the current sample the high specificity (94%) and positive predictive value (93%) confirm that the cut-off of 12.5 accurately identifies true cases of postnatal depression at that point in time.
However, the study findings support the use of a structured diagnostic interview in addition to the EPDS in order to detect period prevalence. The relatively low sensitivity in the present study compared with other validation studies (see [32] for a recent review) may be explained by the fact that while the EPDS assessed symptoms in the week immediately preceding admission to Tresillian (when the babies were, on average, 11 weeks old), the diagnostic interview was conducted when the babies were 4 months old and probed for episodes of depression in any 2 week period since the baby had been born. At the time of admission a small number of mothers were already taking antidepressant medication, and/or were receiving other interventions for postnatal depression. In addition, many mothers commented on the fluctuating course of their depression. As a screening tool for current depression in parentcraft settings there can be little concern that the EPDS (with a cut-off of 12.5) is incorrectly identifying depressive cases. Probing for all episodes of depression since birth may be important in longitudinal evaluations of the impact of postnatal depression on child development.
Anxiety
The finding that the residential care mothers appear to be a particularly anxious group of women confirms the clinical impression of staff and highlights the importance of considering anxiety symptoms in this group of women. Barnett et al. [8] and others morerecently [31], [32] have suggested that the label ‘postnatal depression’ may place some unhelpful constraints on the way people think about the phenomenon of postnatal distress, by limiting the extent to which health professionals address the anxiety symptoms which co-occur with the depressive symptoms. For example, there are simple, non-stigmatizing measures for management of anxiety, such as cognitive and behavioural strategies, timemanagement and appropriate self-care, that could usefully be applied in these settings.
Infant temperament
The fact that 20% of residential care mothers, compared with 5% of comparison mothers, perceived their infant to be temperamentally difficult at 4 months may have implications for later behaviour problems in the child. Parental perceptions that an infant is more difficult than average has been shown to be a strong predictor of later child behaviour problems in longitudinal Australian research [22]. Currently, relations among validated diaries of infant crying and sleep patterns, reports of infant temperament (both nurse and mother) and maternal depression are also being investigated in this sample.
Maternal mood, infant temperament and infant behavioural development
In this study it was maternal anxiety symptoms which were most strongly correlated with maternal ratings of difficult infant temperament and behaviour lending further support to the need for a greater focus on anxiety symptoms both with respect to the interventions offered to mothers and the potential impact on infant development. Indeed, despite considerable research interest in recent years on the impact of postnatal depression on infant outcomes, only a few studies [33], [34], have considered the impact of maternal anxiety on infant development and behaviour.
Clinical implications
What are the practical implications of these findings regarding the prevalence of maternal depression and anxiety in child and family health settings? Given the close relationship between infant sleep and settling problems and maternal depressive symptoms [11] and the fact that family care units may be more acceptable to families than designated mental health facilities [8] residential care units in parentcraft hospitals may well be an appropriate setting for the provision of an integrated, non-stigmatizing approach to the management of both infant behaviour problems and maternal mood disorders. However, the findings also highlight the need for staff support and education in managing mental health problems. Recently national guidelines for clinical approaches to early intervention in perinatal mental health for primary care practitioners have becomeavailable [35]. The availability of adequate specialized mental health backup is also important both in providing consultation– liaison support for the front line, and in managing the small subset of these women who need to be referred for more specialized mental healthtreatment [36].
Study limitations and conclusions
It is acknowledged that the comparison group reported in this study is not an ‘ideal’ control group, and recruitment and procedural differences need to be taken into account in interpreting results. Nonetheless, in the absence of normative data for Australian child-bearing women on depression and anxiety and bearing in mind the relatively high maternal age and education level for the residential care group, we believe the inclusion of data from this demographically comparable group puts our findings regarding the high incidence of mood state problems in mothers admitted to residential care facilities into clearer perspective.
In conclusion, the study findings are consistent with clinical observations and earlier researchreports [8], [9], [16] in demonstrating a close association between unsettled infant behaviour and postnatal mood disorders. Although the findings do not clarify the direction of causality, they highlight the need to consider the psychological wellbeing of the mother despite the fact that the baby is the ‘presenting patient’. Given that both maternal mood disorders and difficult infant temperament are recognized as risk factors for subsequent child development and family adjustment, it is important to develop a clearer understanding of the longer term outcomes for both mothers and infants after they leave parentcraft services. Two aspects in particular need further clarification.
First, the course of postnatal depression during the first postnatal year needs to be explored to allow identification of those women whose difficulties are more persistent or severe. Similarly, it is important to explore the extent to which unsettled infant behaviour predicts ongoing temperamental difficulty and toddler behaviour problems.
Currently this group of mothers and infants are participating in a longitudinal follow up in order to allow a clearer understanding of the contribution of both maternal and infant problems to longer term child outcomes and to ascertain which mothers and infants are most at risk of ongoing problems and could therefore benefit from further or earlier intervention.
Footnotes
Acknowledgements
This research was supported by a Kinsman Fellowship Award from the Royal Australian and New Zealand College of Psychiatrists and Tresillian Family Care Centres.