The mental health impact of stillbirth, neonatal death or SIDS: Prevalence and patterns of distress among mothers

Soc. Sci. Med. Vol. 43, No. 8, pp. 1273-1282. 1996

Pergamon S0277-.9536(96)00039-1

Copyright ,~ 1996 ElsevierScience Ltd Printed in Great Britain. All rights reserved 0277-9536/96 $15.00 + 0.00

THE MENTAL HEALTH IMPACT OF STILLBIRTH, NEONATAL DEATH OR SIDS: PREVALENCE A N D PATTERNS OF DISTRESS A M O N G MOTHERS F R A N C E S M. BOYLE, I J O H N C. M. J O H N ~Department of Social and Preventive Medicine, Anthropology and Sociology, The University of

VANCE," J A K E M. N A J M A N 3 and THEARLE 2 2Department of Child Health and SDepartment of Queensland, Brisbane, Queensland 4072, Australia

Abstract--Although stressful events have long been implicated in the onset of psychological disorder, available data suggest that the majority of individuals appear to escape serious impairment even following highly traumatic events. Related to this is the question of chronicity and whether those who do become impaired develop mental health problems of an ongoing nature. This paper documents the psychological adjustment of 194 women following a highly stressful event--the death of an infant due to stillbirth, neonatal death or SIDS. Anxiety and depression were measured on four occasions--at 2, 8, 15 and 30 months post-loss--using the Delusions Symptoms States Inventory (DSSI/sAD). For comparative purposes, the mental health of 203 mothers of a surviving infant was similarly assessed. The results demonstrate that bereaved mothers, as a group, manifest significantly higher rates of psychological distress than mothers of living infants for at least 30 months after their loss. Their impairment may be either acute or chronic in form. The majority of bereaved mothers appear not be develop serious mental health problems in response to the loss or experience psychological impairment that is usually self-limiting. For a smaller group of women, the death of a baby may herald serious and ongoing distress. Bereaved mothers who were not distressed soon (2 months) after the loss were unlikely to become so later, but those who were still distressed at 8 months were likely to remain so subsequently. Copyright © 1996 Elsevier Science Ltd Key words--stillbirth, neonatal death, SIDS, mental health, stressful life events

INTRODUCTION

BACKGROUND

Life events, because of their potential to invoke stress, have long been studied as possible precursors to the onset of mental health problems. A striking feature of this body of research is the seemingly contradictory nature of the accumulated findings. On the one hand, numerous studies, employing a variety of methodologies, have shown that persons exposed to stressful life events experience mental health problems at a rate higher than that observed in the general population [I-4]. At the same time, however, there is strong evidence that life events do not invariably cause psychological disturbance. Even following highly traumatic events, such as the death of a spouse, only a minorit~ of individuals appear to develop serious or ongoing psychological disorder [5-8]. There is a need to monitor both the short- and medium-term consequences of stressful events, and to determine the natural history of the impact of such events on mental health. This paper is concerned with the experience of one type of life event, namely infant death, and the extent to which such an event produces mental health problems in women over a 30-month follow-up period.

Several issues must be considered in any evaluation of the impact of life events on mental health. First, to demonstrate a causal relationship between life events and mental health problems, it must be established that the event(s) of interest preceded the onset of the disorder. In some cases the disorder may have existed prior t o - - o r even led to---the occurrence of the event(s) in question. Unless this sequence is ascertained, the direction of the association between life events and mental health problems remains uncertain. Second, some categories of life events have a greater propensity than others to produce mental health problems. Events perceived by individuals as undesirable, out of their control and outside the normal expected sequence of life tend to be followed by higher levels of psychological distress than other forms of life change or disruption [9]. Related to this is the issue o f individual differences with respect to the meaning and context of events. Some studies have included life events of an arguably trivial nature, resulting in a weaker association between measures of events and disorder than might otherwise have been found. Third, once psychological disorder is found to follow a particular life event, the severity and

1273

1274

Frances M. Boyle et al.

chronicity of the resulting mental health impairment need to be determined. It is of particular importance to distinguish short-term self-remitting mental health problems from those which endure and may warrant a substantial clinical response. Focusing on a discrete event, such as infant death, offers a valuable opportunity to assess the link between stressful life events and mental health. The death of a baby is likely to be largely independent of prior mental health, reducing the possibility of confounding between cause and outcome. The unambiguously stressful nature of infant death also offers a degree of standardisation of the amount of stress experienced by individuals. Even so, some diversity of response may be expected. For example, the loss of an unplanned and unwanted pregnancy may elicit a different response than the loss of a much-wanted pregnancy conceived after years of infertility. At a broader level, personal or interpersonal resources may protect individuals from the impact of such a loss [10]. It must also be noted that even the birth of a surviving infant may produce mental health problems in a proportion of mothers. Two large-scale studies suggest that 7 8% of women may experience clinically significant post-partum disturbance [1 I, 12]. Previous studies of the impact ~/ bereat,ement on maternal mental health

The death of a child is, undisputedly, a highly traumatic life event. The death of a very young child, as in the case of stillbirth (SB), neonatal death (NND) or Sudden Infant Death Syndrome (SIDS), has the potential to invoke severe emotional distress in parents [13 17]. From a community perspective, however, little is known about the impact of SB, NND or SIDS on the mental health of parents. While researchers have highlighted the range and depth of emotion that bereaved parents may experience, rarely has this form of loss been studied in a systematic way. In short, the findings of most published studies are difficult to interpret due to methodological weaknesses including: small, unrepresentative samples; retrospective data collection; the use of idiosyncratic or subjective outcome criteria; short follow-up periods; and the failure to include an appropriate non-bereaved comparison group. Large scale longitudinal studies of widowhood provide clear indications that major bereavement can seriously affect mental health [7, 8, 18, 19]. Taken together, these studies suggest that up to one-third of the bereaved may manifest serious psychological disturbance within 2 months of the death. One to two years later, up to one-fifth can be expected to report such disturbance. Notably, this mainly seems to reflect ongoing distress, in that depression soon after the loss appears to predict later depression [7]. In one study, 26 of 28 widows who were distressed 2 years post-loss had also been distressed at the initial l-month assessment [8].

There is reason to believe that the psychological impact of perinatal or infant death is no less substantial than that following widowhood. The death of a baby represents the loss of a unique bond of physiological, psychological and social significance. For mothers, especially, it is argued that a physiological and emotional relationship with their child begins during pregnancy as the unborn infant's physical presence becomes increasingly apparent. With the impending birth, the parents, in line with their own and society's expectations, begin to define themselves by their parental role [20]. Bereaved parents may also experience a variety of intangible losses, including the loss of strongly held future expectations. Reference has been made to child death as a violation of the very order of life [20-22] and studies of life events are consistent in showing that events which disrupt the normal expected sequence of life pose a particular threat to mental health [9]. There may also be a sense of guilt or anger at failing to keep the child alive. In developed nations, public health and medical advances have resulted in a dramatic decline in perinatal mortality and a corresponding expectation that babies "do not die". Trends towards declining birthrates and later childbearing may result in each pregnancy having greater significance for the parents and, when unsuccessful, a greater impact than when both birth rates and infant mortality rates were higher. The fact that perinatal or infant death is a relatively rare event may well serve to heighten the impact of such a loss for bereaved parents, who frequently feel isolated and unsupported in their loss. Available data suggest that between 13% and 33% of bereaved mothers manifest serious psychological distress within 2 months of the death of a baby [13, 16, 23, 24]. At 12-14 months, these prevalence estimates range from 13% to 28% [24-26]. However, sample bias (including that due to selective attrition in the case of multiple assessments) and subjective outcome criteria mean that these estimates are, at best, speculative. Only one published study has considered the issue of changes in mental health status over time. Following ectopic pregnancy, miscarriage, SB or NND, 23% of 138 mothers scored significantly higher at 2 years than at 2 months post-loss on the authors' Perinatal Grief Scale [27]. While this suggested that a sizeable minority of women may develop symptoms some time after their loss, whether score increases reflected the onset of disorder or more minor fluctuations of mood as might be found in a non-bereaved sample is unclear. This paper reports findings from a longitudinal study of a large community sample of mothers recently bereaved by SB, NND or SIDS. It evaluates the short- and medium-term impact of such an event on the mental health of mothers. It also seeks to provide a better understanding of the course of

Mental health impact of infant death distress that may be associated with this form of loss by examining patterns of change in mental health through time. METHOD

Participants and procedure The study draws on data from the Family and Child Health Study (FACHS), a longitudinal study of families* bereaved by SB, N N D and SIDS in south-east Queensland, Australia between 1985 and 1988. This predominantly urban region ha,; a population of about 1.6 million of mainly white European origin with minorities of people fi'om indigenous and non-English speaking backgrounds. Seven hospitals (including the state's two major obstetric hospitals) and Queensland Health (the state health department) assisted in the recruitment of families to the study and the protocol for initiating contact with families met their ethical requirements. Upwards of 85% of affected families were approached to join the study. The remaining deaths occurred at a provincial hospital which declined to participate in the study, at small private hospitals scattered throughout the study area or, for unknown reasons, were not referred to the researchers. For ethical reasons, no further follow-up of these families was possible. However, it seems unlikely that, as a group, they would differ markedly from the families who participated. In most cases of SB and N N D , contact details for families were sent to a designated member of the research team who made initial contact with the family, described the study and invited tlheir participation. In a small number of SB and N N D cases, and in all cases of SIDS, contact was first made with the family by a hospital or health department social worker. If the family agreed to be contacted, then details were sent to the researchers. In all cases first contact from the research team occurred at about 4-6 weeks post-loss. Contact with control families who had a surviving infant of the same age, sex and hospital insurance as the infant who died was initiated in the same way. When a bereaved family agreed to participate, the first of three eligible control families was contacted. The remaining control families were approached in turn if a family refused to participate or could not be contacted. Matching for hospital insurance, a crude measure of socioeconomic status, sought to increase the comparability of the bereaved and control groups with respect to social circumstances. In 1988, slightly less than :half the Australian population had any form of private health insurance. The majority of those insured were from high- and middle-income groups [28]. Bereaved families (and their controls) were interviewed in their homes by a psychologist or social worker at about 2, 8, 15 and 30 months following the *The term family also includes single parents.

1275

loss to determine patterns of adjustment over time. In total, 51 i mothers were enrolled in the study. Neither initial participation--SB (63%); N N D (68%); SIDS (53%); control ( 6 5 % ) - - n o r retention rates for the four interviews--SB (79%); N N D (74%); SIDS (69%); control (81%)--differed significantly by category. The following results are restricted to the cohort of mothers who completed all four interviews (SB = 78; N N D = 80; SIDS = 36; control = 203) to ensure that any changes over time were not attributable to changes in sample composition. Analyses were conducted to determine whether characteristics other than category of loss were associated with selection into the study or later attrition. These are detailed elsewhere [29] but, briefly, responders more often had a male infant, private health insurance, were older, married, employed and contacted by telephone. Characteristics associated with attrition are shown in Table 1. Mothers who were not married and who did not have private health insurance were significantly more likely to leave the study. A trend towards higher rates of attrition for mothers with other children was at the border of statistical significance. Notably, however, mothers lost to follow-up did not differ significantly from those who remained in the study in terms of initial anxiety or depression scores [29]. Bereaved (combined SB, N N D and SIDS) and

Table I. Characteristicsof mothers lost to follow-upbetweenthe 2and 30-monthinterviews Lost to follow-up at 2 months 30 months Z-" Characteristic (Np (%) (P) Age (years) < 20 (24) 29.2 20-24 (102) 28.4 25-29 (213) 19.8 30-34 (120) 19.2 35 + (46) 26.I NS" Marital status Married (427) 20.6 Unmarried (84) 30.9 0.04 Health insurance Public (328) 25.6 Private (182) 16.5 0.02 Religion "Catholic (152) 18.4 Anglican (129) 27.I Other religion (166) 20.5 No religion (59) NS Usual church attendance At least monthly 047) 22.4 Less than monthly (359) 22.3 NS Employment Employed outside home 028) 24.2 Housekeeper (318) 21.1 Other (61) 24.6 NS Other livingchildren None (209) 18.2 One or more (302) 25.2 0.06 ~Totals vary slightlydue to missingdata. hNS = not significant.

1276

Frances M. Boyle et al.

control groups were comparable across a range o f variables including: family income, education, religious affiliation a n d frequency of church attendance [29]. However, bereaved mothers were more likely to be younger t h a n 25 years ( P < 0.05). They also tended more often to be u n m a r r i e d and to have other children, but these differences reached only borderline statistical significance ( P = 0.06). In line with epidemiological findings [30, 31], SIDS mothers were younger ( P < 0.01) a n d more likely to have other living children ( P < 0.05) t h a n SB or N N D mothers. N o differences were observed across the three control groups a n d they have been aggregated into a single comparison group.

Measurement of mental health Mental health was assessed using the anxiety and depression scales of the Delusions S y m p t o m s States Inventory (DSSI/sAD) [32], a self-report checklist consisting of two seven-item scales measuring symptoms of anxiety and depression. The main criterion for disturbance is the existence of functional limitations which disrupt normal life. Typical items are: I have worried about every little thing; I have been so worked up that I couldn't sit still (anxiety); 1 have gone to bed not caring if I never woke up; The future seems hopeless (depression). The D S S I / s A D has been validated in a variety of settings, including adult Australian populations [33] and is appropriate for use as a c o m m u n i t y screening instrument. Somatic symptoms do not feature heavily in the DSSI/sAD, an i m p o r t a n t consideration given the possible overlap between some such s y m p t o m s and normal physiological changes associated with pregnancy, childbirth and lactation. Anxiety and depression are k n o w n to feature prominently as part of normal bereavement, including the death o f an infant [21]. W o m e n were asked to report on their feelings in the preceding m o n t h by stating how frequently they had experienced each symptom on a scale ranging from " n e v e r " to "all the time". Those who reported five or more of the seven symptoms at least "'some of the time" were classified as anxious or depressive. W h e n applied to control mothers, this cutoff yielded prevalence rates of 5-6.5% (anxiety) and 2 - 3 . 5 % (depression) at the four interviews. This is a conservative criterion of mental disorder c o m p a r e d to other c o m m u n i t y studies which have sometimes suggested that over 20% of the population manifest symptoms of disorder at any chosen point in time. F o r example, a British survey using the Present State Examination (PSE) found that 14.9% o f women in the c o m m u n i t y had a psychiatric disorder [34]. The choice of the above cutoff reflected a key aim o f the study: to identify women who manifested serious mental health problems following their loss. Since anxiety and depression are normal bereavement

Table 2. Rates of anxiety at each of the four interviewswith relative risks and 95% confidence intervals (CI) for bereaved compared with control mothers Anxious (%) Bereaved Control Relative Time (N + 194) (N = 203) risk" 95% CI 2 months 33.5 6.5(a)h 5.15 2.94-9.04 8 months 18.3(a) 5.5(b) 3.35 1.75-6.40 15 months 15.3(c) 5.1(d) 3.01 1.51-6.00 30 months 13.8(c) 5.9 2.33 1.21-4.48 "The control group is the referencegroup with a relative risk of 1.00. bData incomplete for (a) three women; (b) two women; (c) five women; (d) seven women.

responses, a high p r o p o r t i o n of mothers would be expected to manifest some degree o f distress. Arguably, however, it is those w o m e n whose distress is o f sufficient severity to w a r r a n t clinical attention who are of particular concern. As no clear distinction exists between " n o r m a l grief" and psychological disorder [7, 22], a conservative outcome criterion was considered appropriate. RESULTS

Rates of mental health problems At all four interviews, anxiety a n d depression rates were significantly higher for bereaved mothers t h a n mothers of surviving infants (Tables 2 and 3). N o t surprisingly, anxiety (33.5%) and depression (19.1%) rates were highest for bereaved mothers 2 m o n t h s after their loss, at which time they were, on average, at least five times more likely to be distressed (95% CI for anxiety = 2.94-9.04; depression = 2.4911.93). These rates fell most markedly between the initial and 8 m o n t h interviews, when 18% o f mothers were anxious and 9 % were depressed. C o r r e s p o n d ingly, at 8 m o n t h s the likelihood of being distressed had declined, with bereaved mothers a b o u t three times more likely to be anxious or depressed than the control mothers (95% CI for anxiety = 1.75-6.40; depression = 1.28-7.78). Notably, the fall-off in the prevalence of distress was minimal after 8 months. Despite gradual declines at 15 and 30 months, bereaved mothers remained more likely t h a n controls to display high levels of both anxiety (14%) a n d depression (7%) more than 2 years after their loss. At 15 months, bereaved Table 3. Rates of depression at each of the four interviews with relative risks and 95% confidence intervals (CI) for bereaved compared with control mothers Depressed (%) Bereaved Control Relative Time (N + 194) (N = 203) risk" 95% CI 2 months 19.1 3.5(a)" 5.45 2.49-11.93 8 months 9.4(a) 3.0(b) 3.16 1.28-7.78 15 months 8.5(c) 2.0(d) 4.15 1.41-12.18 30 months 7.4(c) 2.5 3.01 1.1(~8.19 ~'Thecontrol group is the referencegroup with a relative risk of 1.00. hData incomplete for (a) three women: (b) two women: (c) five women: (d) seven women.

Mental health impact of infant death m o t h e r s were three times more likely t h a n control,; to be anxious (95% CI = 1.51-6.00), a n d this risk remained more t h a n doubled at the final interview (95% CI = 1.21-4.48). With respect to depression, bereaved m o t h e r s remained three to four times more likely to be affected t h a n control mothers at the 15 (95% Ci = 1.41-12.18) a n d 30 m o n t h (95% C! = 1.10-8.19) interviews. Separate multivariate logistic regression analyses were u n d e r t a k e n to determine the extent to which factors other t h a n the d e a t h o f a baby m a y have accounted for these differences. Included in the models were m a t e r n a l age ( < 2 5 vs 2 5 + years); marital status (married vs u n m a r r i e d ) a n d other children (yes vs no). As noted above, these variables tended to differentiate the bereaved a n d control groups. A t 30 m o n t h s , the strength o f association between category (bereaved vs control) a n d anxiety, while reduced, remained statistically significant ( P = 0.02 for the adjusted model, c o m p a r e d with P < 0.01 for the unadjusted model presented in Table 2) when the effects o f age, marital status a n d other children were taken into account. With respect to depression at 30 months, a reduction in the strength o f association when these b a c k g r o u n d variables were added to the model produced, a difference which reached only the b o r d e r o f significance ( P = 0 . 0 7 for the adjusted model, c o m p a r e d with P < 0.05 for the unadjusted m~zdel presented in Table 3). At 15 m o n t h s , however, the association between category a n d depression remained statistically significant ( P = 0 . 0 1 for the adjusted model, c o m p a r e d with P < 0.01 for the unadjusted model presented in Table 3). While the a p p a r e n t decline in statistical significance o f some o f the above associations is likely to reflect the effect o f possible c o n f o u n d i n g variables, the disaggregation o f the data following the introduction o f these variables also contributes to this decline.

Type of loss and mental health M o t h e r s in the SIDS group manifested the highest rates of anxiety a n d depression and, at all four interviews, were at greater risk for distress t h a n control m o t h e r s (Tables 4 and 5). SB and N N D m o t h e r s displayed similar a n d lower rates of distress at each interview a n d b o t h groups were indistinguishable from control m o t h e r s by 15 m o n t h s with respect to depression, a n d by 30 m o n t h s with respect to anxiety. The initially higher rates o f anxiety a n d depression for SIDS m o t h e r s appeared to fall less sharply over time. Despite the apparently more prevalent a n d e n d u r i n g distress following SIDS, beyond the initial (2 m o n t h ) loss period, no more than 1 in 4 SIDS m o t h e r s met the study's criteria for mental health problems. At 30 m o n t h s following the loss a b o u t 1 in 5 m o t h e r s were anxious a n d a b o u t 1 in 6 depressed. The m o t h e r s o f SIDS infants were younger a n d more likely to have other living children than m o t h e r s in the other groups, but separate

1277

Table 4. Rates of anxiety according to type of death at each of the four interviewswith relative risks and 95% confidenceintervals (CI) Time Anxious (%) Relativerisk 95% CI 2 months SIDS NND SB Control 8 months SIDS NND SB Control 15 months SIDS NND SB Control 30 months SIDS NND SB Control

(36) (80) (78) (200)

55.5 28.7 28.2 6.5

8.55 4.42 4.34 1.00

4.68.-15.60 2.36--8.29 2.30 8.18

(36) (78) (77) (201)

25.0 19.2 14.3 5.5

4.57 3.51 2.61 1.00

2.04-10.23 1.69--7.31 1.18 5.77 --

(35) (80) (74) (196)

25.7 12.5 13.5 5.1

5.04 2.45 2.65 1.00

2.21--11.51 1.06--5.66 1.15--6.10

(36) (80) (73) (203)

22.2 11.2 12.3 5.9

3.76 1.90 2.09 1.00

1.65--8.55 0.83--4.34 0.92-4.74 --

analyses showed that adjustment for these variables did not materially alter the above results. Overall, the data d e m o n s t r a t e a consistent pattern of heightened risk for anxiety a n d depression associated with the d e a t h of a baby. In some cases, the associated 9 5 % Cls are wide, due mainly to small n u m b e r s especially in the SIDS group. Yet, the point (or " b e s t " ) estimates of R R tend to be large a n d reflect statistically significant differences, particularly at the earlier interviews.

Mental health changes over time We now address the question of whether distress rates at follow-up interviews reflect ongoing psychological problems or the later onset of symptoms, by d o c u m e n t i n g changes in mental health status (Figs 1 and 2). At each o f the three follow-ups, w o m e n were categorised according to whether they: (i) remained Table 5. Rates of depression according to type of death at each of the four interviews with relative risks and 95% confidence intervals (Cl) Time Depressed (%) Relativerisk 95% CI 2 months SIDS (36) 27.8 7.94 3.23-19.49 NND (80) 15.0 4.29 1.75-10.49 SB (78) 19.2 5.49 2.33 12.96 Control (200) 3.5 1.00 -8 months SIDS (36) 19.4 6.51 2.32-18.27 NND (78) 9.0 3.01 1.04-8.67 SB (77) 5.2 1.74 0.50-6.00 Control (201) 3.0 1.00 -15 months SIDS (35) 20.0 9.80 3.03-31.72 NND (80) 6.2 3.06 0.84-11.11 SB (74) 5.4 2.65 0.68-10.32 Control (196) 2.0 1.00 -30 months SIDS (36) 16.7 6.77 2.18-21.00 NND (80) 3.7 1.52 0.37-6.22 SB (73) 6.8 2.78 0.83-9.33 Control (203) 2.5 1.00

1278

Frances M. Boyle et al.

Anxiety Control mothers (N -- 191)

Bereaved mothers (N = 182)

I

At 2 months

I W 65.4%

34.6%

93.2%

6.8% Anxious

Anxious

At 8 months

/

\

\

15.4% 3.8% 19.2%

2.1%3.1%

61.6%

'-v-'

4.7%

90.1%

5.2% Anxious

19.2% Anxious

At 15 months

I

\

\

12.6% 2.7% 25.8%

58.8%

I

\

\

2.6% 1.6% 7.3%

15.3% Anxious

88.5%

4.2% Anxious

At 30 months

/ \ \ 12.1% 2.2% 29.1%

56.6%

14.3% Anxious

/\ ~ 2.1%3.1% 9.4%

85.3%

5.2% Anxious Remain anxious

~]

No longer anxious

Become anxious

[--q

Never anxious

Fig. I. Patterns of anxiety over time for bereaved and control mothers.

distressed; (ii) had become distressed; (iii) were not distressed but had been previously; or (iv) were not distressed and had not been previously. A t the third and fourth interviews, the " r e m a i n e d distressed" category includes women currently distressed and who had been distressed on at least one pevious occasion. The " n o longer distressed" women, a l t h o u g h not currently distressed,

*A small number of respondents with missing data at any one of the four assessments were excluded and, for this reason, figures differ slightly from those above. There was no reason to believe that the exclusion of this small number of missing cases, equally distributed across the bereaved and control groups, would appreciably alter the observed patterns.

had been so on at least one previous occasion. It follows that the " b e c o m e distressed" category includes only those w o m e n distressed for the first time at the current interview a n d the " n o t distressed" category only those w h o have also not been distressed at any previous interview.* Figures 1 a n d 2 reveal several i m p o r t a n t features. First, as s h o w n above, a m a r k e d reduction occurred in the prevalence o f anxiety and depression between 2 and 8 months. However, this fall-off was not sustained beyond 8 months, with prevalence rate reductions of only 5 % (anxiety) a n d 2 % (depression) during the remainder o f the study. Thus, the bereaved mothers, as a group, remained significantly more distressed t h a n the control mothers even at the 30 m o n t h follow-up interview.

Mental health impact of infant death Second, despite a clearly elevated risk for mental health problems, most bereaved mothers did not manifest psychological disorder. The majority adapted to their loss without evidence of serious mental health problems in the form of anxiety (57%) or depression (72%). Third, it is apparent that the higher rates of distress among bereaved mothers reflected ongoing mental health problems in a core group of women. Thus, at 30 months, 14.3% of the bereaved mothers were anxious, a figure made up of 12.1% who were anxious on one or more previous occasions (usu~.lly including the 2 m o n t h interview) and only 2.2% x~ho had become anxious since the last interview. Similarly, 7.6% were depressed at 30 months. This was made up of 4.9% who had been depressed

1279

previously and 2.7% who had become depressed since the last interview. Relatively few bereaved mothers became anxious or depressed beyond the initial 2 month interview and, indeed, the small proportion of the bereaved sample who became distressed at each successive interview (2-4%) differed little from that of the control group (1-3%). These rates of onset are unlikely to be any higher than those observed in a sample drawn from the population at large. Excluding mothers who were anxious at 2 months, a total of about 8% of bereaved and control mothers became anxious on a subsequent occasion; 8.8% of bereaved and 4.2% of controls developed depression (P = 0.07). It appears then that the death of an infant may precipitate ongoing or chronic distress in a small

Depression Control mothers (N = 191)

Bereaved mothers (N = 182) At 2 months

I

I U

3.1% Depressed

80.8%

19.2% Depressed At 8 months

/ \ 6.6% 3.3%12.6%

1.0% 1.6% 2.1%

77.5%

95.8%

2.6% Depressed

9.9% Depressed I

96.9%

At 15 months

/ ~ 6.0% 2.7% 16.5%

/ \ ~ 0.5% 1.6% 4.2%

74.7%

93.7%

2.1% Depressed

8.7% Depressed At 30 months

I

IH / ~ \ 4.9% 2.7% 20.3%

72.1%

1.0% 1.0% 5.2%

92.7%

2.0% Depressed

7.6% Depressed [~

Remain depressed

~]

No longer depressed

m

Become depressed

['-"1 Never depressed

Fig. 2. Patterns of depression over time for bereaved and control mothers.

1280

Frances M. Boyle et al.

group of women. These women are also likely to have a number of shared characteristics which differentiate them from women with more adaptive outcomes. Included among these, and detailed in a forthcoming monograph, are: inadequate social support, marital dissatisfaction, and other children in the family. The data provide little evidence for "delayed" mental health problems. If a bereaved mother is not experiencing high levels of distress at about 2 months post-loss, the likelihood of her developing mental health problems at a later time appears to be no greater than that of a mother of a surviving child, at least within the first 30 months of the death. However, if a mother remains anxious or depressed 8 months after the loss, her distress appears likely to continue for some time. DISCUSSION This study has demonstrated a clear impact of the death of a baby on women's mental health. Using comparative data from mothers of surviving infants, bereaved mothers were found to be at increased risk for psychological distress at 2, 8, 15 and 30 months after their loss. We estimated that 30 months after the death, some 14% of bereaved mothers displayed high levels of anxiety and about 7% high levels of depression. These represented 2- and 3-fold risks, respectively. An important issue is whether the observed higher rates of distress among bereaved mothers can be attributed to the death of a baby or whether they may reflect differences which existed prior to the loss. To exclude this second possibility would require pre-loss measures of mental health to assess the comparability of the bereaved and control groups. Given that mothers were recruited following their loss, this was not possible in the present study. We have, however, been able to demonstrate that the two groups were comparable across a number of background variables. We have also demonstrated that differences in terms of age, marital status and other children do not fully account for the higher rates of distress, especially anxiety, among bereaved mothers even at 30 months post-loss. A second issue involves sample representativeness and whether the results may have differed had all the eligible mothers been included. Characteristics associated with initial non-response and later loss to follow-up tended to be similar and included younger age, public health insurance, being unmarried and social support perceived as inadequate. Although these tend to be suggestive of social disadvantage and, consequently, poorer mental health, we have shown elsewhere [29] that mothers who left the study were not, on average, more psychologically impaired soon after the loss. While this cannot be established directly for those who did not participate at all, and it may be that they represent a more severely impaired group, we have argued that the effect of

sample loss on prevalence estimates and bereavedcontrol comparisons is likely to be modest. We conclude that the death of a baby has the potential to produce mental health problems of both an acute and chronic nature, yet the majority of bereaved mothers appear to escape serious psychological impairment. Mental health problems, if they occur, usually manifest soon after the loss and tend to be remitting in most cases. Nonetheless, for a small group of women, the death of an infant may herald ongoing mental health impairment. Like the present study, a number of previous studies of both conjugal and perinatal bereavement have focused on serious mental health problems as opposed to what may be considered the normal sadness and distress of grief. Reported rates of distress have ranged from 13% to 35% at 2 months and 11% to 28% at 1-2 years [7, 8, 13, 18, 19, 24, 26]. Our estimates of 33% (anxiety) and 19% (depression) at 2 months, 15% (anxiety) and 9% (depression) at 15 months, and 14% (anxiety) and 7% (depression) at 30 months are broadly comparable. Among mothers who were distressed soon after the loss, improvements in mental health most often occurred within about 8 months. From this point on, the higher rates of both anxiety and depression among bereaved mothers were largely attributable to an excess of women who experienced chronic distress. The relatively low onset rates (2%-4%) for anxiety and depression differed little from those observed in the control group (1%-3%). It is unlikely that these rates would be any higher than would be expected over time in a sample drawn from the population at large. This further highlights the immediacy of the impact of infant death on women's mental health. Anxiety was the more ubiquitous form of distress experienced by mothers following the birth, and particularly the death, of an infant. This may, in part, be attributable to the measure itself. Unpublished data from a study of 8556 pregnant women [35], which employed the same measure, also reveal higher rates of anxiety relative to depression. Nonetheless, the present pattern of results highlights the prominence of anxiety in the spectrum of responses following the death of a baby. Reactions to the death of a baby have been likened to post-traumatic stress disorder, essential clinical features of which include high levels of fear and anxiety accompanied by intrusive thoughts and the repeated re-living of an extremely traumatic experience [15]. Following the death of a baby, mothers may also experience anxiety and a sense of vulnerability in relation to the future well-being of both surviving and subsequent children. Losing an infant to SIDS rather than SB or NND appeared to increase the likelihood of both anxiety and depression. Dyregrov and Matthiesen [15], who reported a similar finding, suggested that this may reflect the particularly traumatic nature of S1DS. Not only did SIDS always occur without warning and in the home, but also it was inexplicable. Of course, in

Mental health impact of infant death the case of SIDS, the older age of the child may be another contributing factor in that parents have had the opportunity to form a closer bond with the child. Overall, patterns of adjustment were highly consistent with those observed following widowhood [8] and confirmed in a comprehensive review of studies of bereavement [10]. These patterns were: (i) no evidence of high distress levels during the various study periods following the loss; (it) an initially high level of distress that gradually dissipated; and (iii) enduring high distress levels. This is not to suggest that the death of an infant to SB, N N D or SIDS does not produce intense and enduring sorrow, sadness and other emotions in mothers. However, in the general population, severe psychological disturbance appears to be relatively uncommon following s~ach events. Theoretically, our results are consistent with the seemingly paradoxical findings from life events studies. While exposure to stressful events is associated with mental health problems, it has become increasingly clear that most individuals do not become psychologically disturbed even after extremely traumatic events [5, 6]. Rather, the impact of stressful life events appears to be governed to a large degree by the personal and interpersonal resources available to those affected. Without information on pre-loss mental health, the extent to which some women developed problems in response to their loss or a worsening of pre-morbid symptoms cannot be established. However, it is tempting to speculate that this may distinguish acute from chronic forms of distress. Lasker and Toedter [27] mount a similar argument, suggesting that chronic grief is not simply an extension of acute (or more loss-specific) grief but a qualitatively different, more pervasive form of despair. Although we found little evidence of delayed mental health problems, it is possible that in sc,me cases it may be many years before high levels of distress are experienced in relation to a major loss [36, 37]. Certain future events or developmental stages may act as catalysts for the onset of a delayed response. The present findings also need to be considered in the context of depression as a recurrent disorder [5, 38]. Monroe and Depue, for example, maintain that 40-50% of those affected are likely to experience one or more recurrences. Importantly, they also suggest that life events may be particularly influential for early episodes of, and a subsequent susceptibility to, depression. Until populationbased longitudinal data covering a lengthy time-span become available, such possibilities remain speculative. The identification of mothers at risk for serious and enduring mental health problems is an important priority for clinicians and health policy planners, who are typically faced with limited resources for assisting families. The present study suggests that this broad aim could be quite easily met by using a simple

1281

clinical assessment at about 2 months post-loss, a time which usually coincides with a post-mortem conference or post-natal medical examination. Women who are not distressed at this time are unlikely to become so during the next 30 months. However, those who are distressed would constitute a high-risk group and possibly warrant more specialised assistance. In Australia presently and at the time of this study, services for bereaved families could best be described as occurring on an ad hoc basis. While some bereaved families may receive or initiate formal suport through health services or community support organisations, there is no comprehensive approach. Our data indicate that routine assessment, appropriate referral systems and opportunities for supportive counselling should be intrinsic components of care following the death of a baby. Acknowledgements--The financial support provided by the National Health and Medical Research Council, Australian Rotary Health Research Fund, and in memory of Thomas Edward Hayes is gratefully acknowledged, as is the involvement of all families who participated in the study. We thank Gail Williams for statistical advice and two anonymous reviewers for their helpful comments. Material used in this paper is derived from a thesis supported by an Australian Postgraduate Research Award and for which the first author was awarded a PhD by The University of Queensland in 1993. REFERENCES

I. Billings A. C. and Moos R. H. Stressful life events and symptoms: a longitudinal model. Hlth Psvehol. 1(2). 99, 1982. 2. Myers J. K., Lindenthal J. J., Pepper M. P. and Ostrander D. R. Life events and mental status: a longitudinal study. J. Hlth Soc. Behat'. 13(4), 398, 1972. 3. Paykel E. S., Myers J. K., Dienelt M. N., Klerman G. L., Lindenthal J. J. and Pepper M. P. Life events and depression: a controlled study. Arch. Gen. Psychiat. 21, 753, 1969. 760. 4. Pianta R. C. and Egeland B. Relation between depressive symptoms and stressful life events in a sample of disadvantaged mothers. J. Consult. Clin. Psychol. 62(6), 1229, 1994. 5. Monroe S. M. and Depue R. A. Life stress and depression. In Psychosocial Aspects of Depression (Edited by Becker J. and Kleinman A.). Lawrence Erlbaum Associates, Hillsdale, N J, 1991. 6. Brown G. W. and Harris T. Soeiul Origins 0[" Depression: A Study q[' Psychiatric Disorder in Women. Free Press, New York, 1978. 7. Zisook S., Shuchter S. R., Sledge P. A., Paulus M. P. and Judd L. L. The spectrum of depressive phenomena after spousal bereavement. J. Clin. Psyehiat. 55(4) (Suppl.), 29, 1994. 8. Vachon M. L. S., Rogers J., Lyall W. A., Lancee W. J., Sheldon A. R. and Freeman S. J. J. Predictors and correlates of adaptation to conjugal bereavement. Am. J. Psychiat. 139(8), 998, 1982. 9. Mirowsky J. and Ross C. E. Social Causes o[" Psychological Distress. Aldine de Gruyter, New York, 1989. 10. Wortman C. B. and Silver R. C. The myths of coping with loss. J. Consult. Clin. Psvchol. 57(3), 349, 1989. 11. Cooper P. J., Campbell E. A., Day A., Kennerley H.

1282

12.

13. 14.

15. 16. 17.

18.

19.

20.

21. 22. 23.

24. 25.

Frances M. Boyle et al.

and Bond A. Non-psychotic psychiatric disorder after childbirth: a prospective study of prevalence, incidence, course and nature. Br. J. Psychiat. 152, 799, 1988. Gotlib 1. H., Whiffen V. E., Wallace P. M. and Mount J. H. Prospective investigation of postpartum depression: factors involved in onset and recovery. J. Abnormal Psychol. 100(2), 122, 1991. Clarke M. and Williams A. J. Depression in women after perinatal death. Lancet 916, 1979. DeFrain J. Learning about grief from normal families: SIDS, stillbirth, and miscarriage. J. Marital Fam. Ther. 17(3), 215, 1991. Dyregrov A. and Matthiesen S. B. Stillbirth, neonatal death and Sudden Infant Death (SIDS): parental reactions. Scand. J. Psychol. 28, 104, 1987. Tudehope D. I., Iredell J., Rodgers D. and Gunn A. Neonatal death: grieving families. Med. J. Aust. 144, 290, 1986. Vance J. C., Foster W. J., Najman J. M., Embelton G., Thearle M. J. and Hodgen F. M. Early parental responses to sudden infant death, stillbirth or neonatal death. Med. J. Aust. 155, 292, 1991. Bornstein P. E., Clayton P. J., Halikas J. A., Maurice W. L. and Robins E. The depression of widowhood after thirteen months. Br. J. Psychiat. 122, 561, 1973. Stroebe W., Stroebe M. S. and Domittner G. Individual and situational differences in recovery from bereavement: a risk group identified. J. Soc. Iss. 44(3), 143, 1988. Rando T. A. The unique issues and impact of the death of a child. In Parental Loss of a ChiM (Edited by T. Rando). Research Press, Champaign, IL, 1986. Dyregrov A. Parental reactions to the loss of an infant child: a review. Scand. J. Psychol. 31, 266, 1990. Raphael B. The Anatomy of Bereavement. Basic Books, New York, 1983. Clyman R. I., Green C., Rowe J., Mikkelsen C. and Ataide L. Issues concerning parents after the death of their newborn. Crit. Care Med. 8(4), 215, 1980. Jensen J. S. and Zahourek R. Depression in mothers who have lost a newborn. Rocky Mtn Med. J. 69(11), 61, 1972. LaRoche C., Lalinec-Michaud M., Engelsmann F.. Fuller N., Copp M., McQuade-Soldatos L. and Azima

26.

27.

28.

29.

30. 31. 32.

33. 34. 35.

36.

37. 38.

R. Grief reactions to perinatal death--a follow-up study. Can. J. Psychiat. 29, 14, 1984. Forrest G. C., Standish E. and Baum J. D. Support after perinatal death: a study of support and counselling after perinatal bereavement. Br. Med. J. 285, 1475, 1982. Lasker J. N. and Toedter L. J. Acute versus chronic grief: the case of pregnancy loss. Am. J. Orthopsychiat. 61(4), 510, 1991. Australian Bureau of Statistics. Health Insurance Survey Australia, June 1988. Cat. no. 4335.0. Australian Bureau of Statistics, Canberra, 1989. Boyle F. M., Najman J. M., Vance J. C. and Thearle M. J. Estimating nonparticipation bias in a longitudinal study of bereavement. Aust. New Zealand J. Pub. Hlth, in press. Golding J., Limerick S. and Macfarlane A. Sudden Infant Death: Patterns, Puzzles and Problems. Open Books, Somerset, 1985. Resseguie L. Influence of age, birth order and reproductive compensation on stillbirth ratios. J. Biosoc. Sci. 5, 443, 1973. Bedford A., Foulds G. A. and Sheffield B. F. A new personal disturbance scale (DSSI/sAD). Br. J. Soc. Clin. Psychol. 15, 387, 1976. Henderson S., Byrne D. G. and Duncan-Jones P. Neurosis and the Social Environment. Academic Press, Sydney, 1981. Bebbington P., Hurry J., Tennant C., Sturt E. and Wing J. K. Epidemiology of mental disorders in Camberwell. Psychological Med. I1, 561, 1981. Keeping J. D., Najman J. M., Morrison J., Western J. S., Andersen M. J. and Williams G. M. A prospective longitudinal study of social, psychological and obstetric factors in pregnancy: response rates and demographic characteristics of the 8556 respondents. Br. J. Obstet. Gynaecol. 96, 289, 1989. Bowlby J. The making and breaking of affectional bonds: 1: Aetiology and psychopathology in the light of attachment theory (the Fiftieth Maudsley Lecture). Br. J. Psvchiat. 130, 201, 1977. Worden J. W. Grief Counselling and Grief Therapy: A Handbook .['or the Mental Health Practitioner, 2nd edition. Tavistock/Routledge, London, 1991. Coyne J. C. and Downey G. Social factors and psychopathology: stress, social support, and coping processes. Annu. Rev. Psychol. 42, 401, 1991.