Académique Documents
Professionnel Documents
Culture Documents
32 8
AT TA C H M E N T R E P R E S E N TAT I ON S IN A B NOR
NO R M A L I L L N
NEE S S B E H AV I OU R BY P R OX Y
T
Table
able 1 Factitious illness by proxy (FIP) behaviours (Hesse, 1999). This allows the rater to (7). Only 5 mothers had previous
in 67 mothers categorise the transcribed narratives as diagnoses of personality disorder, although
either (a) secure or insecure; or (b) insecure 14 had demonstrated behaviours that might
Type of behaviour n (%)1
dismissing, insecure preoccupied or cannot be associated with personality disorder,
classify. In addition, transcripts may also such as self-harm and/or taking overdoses.
Repeated presentation to GP 16 (23) be categorised as either resolved or unre-
or hospital solved for traumatic experiences of loss or
Psychiatric diagnosis at interview
Poisoning 12 (17) fear.
After transcription, each narrative was Based on the clinical assessment of mental
Fabricated apnoeic attacks 10 (15) state, only 12 mothers received any sort of
rated for both security and type of in-
Previous SIDS plus new concerns 9 (13) diagnosis at interview (depression, 8;
security. The rating could not be done with
about mother/child personality disorder, 4). None received a
complete masking, given the context, but
Other fabricated symptoms 7 (10)
the rater was not aware of the details of diagnosis of severe mental illness.
Factitious disorder in mother 5 (7) the allegations or established behaviours.
Somatising disorder in mother 5 (7) In addition to data collected from the Psychological assessment
Smothering 5 (7) AAI, data analysed included available Thirty-nine participants completed the
Non-accidental injury 5 (7) demographic information, results of the MCMI–III. One set of responses showed
Dilution of feeds 3 (4) MCMI–III personality assessment, and an invalid response pattern. Of the remain-
Neglect 2 (3) psychiatric diagnoses (based on clinical ing 38, only in 9 cases (23%) were the
Invalidism of child 2 (2) interview and examination of past records). scores indicative of definite personality dis-
Previous FIP behaviour 2 (3) order. Four out of those nine participants
Killed a previous child 1 (2) RESULTS with a score indicative of personality dis-
order had profiles indicating borderline
GP, general practitioner; SIDS, sudden infant death
syndrome. Eighty-five mothers completed the AAI. personality disorder. Seventeen women
1. Mothers may show more than one behaviour. Five refused to take part in the research showed abnormally elevated personality
and three women’s cases were not estab- traits, most commonly compulsive, depen-
parent’s own history of being cared for as a lished as factitious illness by proxy dent or histrionic traits. These women had
child is indicated as an essential part of behaviour. Ten interviews were lost in also had psychiatric care in the past, and
assessment in child protection cases transcription. This left 67 transcripts for included those who had previously received
(Department of Health, 2000). analysis. a diagnosis of personality disorder. Twelve
The interview used was the Adult The mean age of the mothers was 28 mothers showed no evidence of abnormal
Attachment Interview (AAI; George et al, al, years and the mean age of the index child responses.
1994), which is a semi-structured interview, was 2.3 years. Most mothers (38, 55%)
providing a detailed account of early care ex- were in a partnership or married. Most Attachment data
periences. It has excellent psychometric prop- children had siblings and no particular
erties when carried out and rated by trained Secure or insecure
birth order was more common than
professionals (Hesse, 1999). K.B. is trained another; 32 (53%) of the children were Fifty-seven (85%) of the transcripts were
in the use of the AAI and carried out the inter- male. rated as insecure. The different categories
views. These were taped and transcribed, and Of the 67 mothers, 31 (46%) had his- of insecurity are described in Table 2,
then sent for rating. K.B. was not involved in tories of childhood trauma and 19 (28%) together with data from previously pub-
the rating of the interviews. gave histories of adult trauma, usually rape lished studies of non-clinical samples, clini-
Transcripts were rated according to a or domestic violence; 30 mothers (45%) cal samples, abusive mothers and offender
manual. Raters undergo an extensive 2- had histories of childhood illness or injury groups.
week training and then complete a 30-case and 16 (24%) recalled a member of the The secure (F) category was under-
interrater reliability check. To qualify as a family suffering from significant illness dur- represented in our sample, which is not
rater for research purposes, raters must ing their childhood; 36 mothers (54%) had surprising given the prevalence of past
achieve at least 80% interrater reliability. experienced bereavement, either as adults psychiatric illness, personality abnormal-
G.A. is trained as a rater. or children. ities and childhood histories of divorce
Information from this interview was (7.3%) or maltreatment/neglect (51%).
used to inform the conclusion of the final Our sample therefore resembles clinical
report and to make therapeutic recommen- Previous psychiatric diagnosis
groups, rather than normal mothers. It is
dations. The mothers gave their consent to (based on case notes or GP
perhaps surprising to find any secure
the use of anonymised material from these records only)
mothers in this group.
interviews, either directly or through their Thirty-six mothers (52%) had histories of
legal representatives. psychiatric treatment, either in childhood
Patterns of insecurity
or adulthood, at the level of primary care
only; 6 had histories of childhood psychi- (D). This category was
Insecure ^ dismissing (D).
Analysis atric illness alone. The most common overrepresented in our sample. A similar
The AAI is rated using linguistic analysis of previous psychiatric diagnoses were soma- excess of D representations is described in
transcriptions of the taped interviews tising disorder (7) and eating disorders offender groups (Fonagy et al, al, 1997) and
329
A D S H E A D & B LU G L A S S
T
Table
able 2 Adult Attachment Interview categories
Group Secure (F), Enmeshed (E), Dismissing (D), Cannot classify (CC), Unresolved state of mind with
% % % % regard to trauma or loss (U), %
Non-clinical mothers2 58 18 24 0 19
Mothers of disturbed children2 14 19 41
Psychiatric patients2 22 64 14 82
Forensic patients (male)3 19 45 36 36
This study 18 (12/67) 12 (8/67) 46 (31/67) 27 (18/67) 60 (40/67)
1. Subjects can be assigned to both an unresolved state of mind and one other category.
2. Data from van IJzendoorn & Bakermans-Kranenburg, 1996.
3. Data from Fonagy et al,
al, 1997.
in parents who maltreat their children Main & Hesse (1990) suggested that years), the prevalence of bereavement ex-
(Crittenden et al,
al, 1991). Individuals with CC mothers are both frightened and periences seems high. For some, the fact
a dismissing approach to attachments dis- frightening as parents, which may then that they were bereaved as children may
miss distress at times of illness, fear or loss, cause disorganisation of attachment in their have been traumatic; for others, the
which may be associated with lack of children. Long-term follow-up studies of bereavement may have been traumatic
empathy for others’ distress (Fonagy et al, al, children with disorganised attachment have because it involved a sudden death. Nine
1997). There is a particular sub-class of found that they often become either com- mothers had previously lost a child to
dismissing attachment (Ds2), which is pulsive carers or controlling carers. Facti- sudden infant death syndrome. Two out
associated with fear of the loss of a child tious illness by proxy behaviour could be of the nine had two previous children who
through death, and could theoretically give considered both compulsive and controlling had died of this syndrome. Some had
rise to excessive medical help-seeking caregiving, and therefore as one manife- experienced multiple bereavements in their
behaviour. We did not find evidence of this station of highly disorganised attachment lives or been bereaved when pregnant.
subgroup in our sample. in mothers. The question raised by these data is
whether the experience of bereavement led
to a profound disorganisation of the
(E). This category is
Insecure ^ enmeshed (E). Lack ofresolution of psychological distress follow- attachment system for these mothers and
underrepresented in our sample. Individ- (U). The AAI asks individ-
ingtraumaorloss (U). contributed to the risk of this type of
uals with the E categorisation find it hard uals about experiences of trauma and loss maltreatment.
to see themselves as separate persons from in childhood. Responses are rated for lin-
their families of origin; the ‘enmeshment’ guistic evidence of resolution of normal
is with the attachment figures of childhood. distress or evidence of abnormal distress. DISCUSSION
Such individuals often have had over-close Ratings are made for both trauma (such
relationships with their own mothers or as childhood abuse, carefully defined in Normal and abnormal illness
had to act as carers for their own parents. the manual) and bereavement experiences. behaviour
The E category has also been associated There was an excess of the unresolved (U) Illness behaviour is a term that refers to the
with borderline personality disorder, which categorisation in our sample compared ways individuals perceive, experience and
has been described as common in factitious with both clinical and non-clinical groups, respond to illness. Illness behaviour is influ-
illness by proxy (Bools et al,
al, 1994). suggesting that these mothers may still be enced by individual, social and cultural ex-
struggling with unresolved fear or grief periences and constructions of illness, and
reactions. may be subject to special expectations.
(CC). This cat-
Insecure ^ cannot classify (CC). Loss or trauma experiences were For example, in Western cultures, those
egorisation indicates an attachment common in our sample. Out of 40 mothers, who are ill are expected to desire to get
narrative that is highly disorganised. There 22 (55%) described experiences of either well, to comply with treatment and to seek
are different ways for an attachment physical or sexual abuse or neglect in child- help (or elicit care) appropriately (Parsons,
narrative to be rated as CC: the entire hood; 19 described adult trauma such as 1964). Although not specified, it may be
narrative may be grossly incoherent or the domestic violence or rape; 25 (62.5%) assumed that proxy illness behaviour has
narrative may indicate that the individual had experienced bereavement, either in the same expectations.
exhibits a mixture of both D and E states childhood or adulthood. These data are Factitious illness by proxy behaviour
of mind with regard to attachment (Hesse, similar to those from the Great Ormond involves three forms of abnormal illness
1999). Participants in the CC category Street cohort (Gray & Bentovim, 1996). behaviour: false accounts of symptoms,
seemed unable to find any words to Traumatic bereavement is the most fabricated symptoms and induction of
describe past relationships (although they common type of trauma that people are symptoms (Bools, 1996). The last group is
were perfectly articulate in other areas). likely to experience (Breslau, 1998), and is most dangerous and most likely to come
They were also unable to access any defined as sudden, unexpected or violent to the attention of healthcare professionals.
memories of their childhoods, and seemed bereavement. Given the comparatively The few available cohort studies describe
to find the task alien and uninteresting. young mean age of our participants (28 cases of more serious and life-threatening
330
AT TA C H M E N T R E P R E S E N TAT I ON S IN A B NOR
NO R M A L I L L N
NEE S S B E H AV I OU R BY P R OX Y
behaviours, resulting in referral to (and hostility to the child. The AAI categorisa- ill as children and requiring either regular
detection in) specialist centres (Bools et al,
al, tion can indicate only a general state of visits to the GP or hospital. A further 12
1994; Gray & Bentovim, 1996; 1996; Southall mind in relation to attachment relation- had experiences of a family member suffer-
et al,
al, 1997). The exact prevalence of this ships; it does not offer details of those ing illness when they were children. These
behaviour is unknown; the more severe relationships. This would come from more experiences would have put them at risk
forms are rare (0.5 per 100 000 children; detailed analysis of the narratives, on which of somatising disorder in adulthood.
McClure et al, al, 1996) and are associated we are currently engaged. Why a mother The key question is how the experience
with appreciable mortality, similar to other would choose one type of behaviour over of illness or poor care in childhood affects
types of child abuse (Sheridan, 2003). another is likely to be influenced by many the capacity to care for others in adulthood.
Meadow’s original account stated that factors, including unconscious symbolism Mothers who have a secure attachment to
the mother’s intention was to obtain atten- and ease of access/opportunity. Our data their own parents are generally better able
tion to her own needs, and this is still his support the general premise that there is to provide good enough care for their chil-
position (Meadow, 2002). Other theorists no direct causal link between any single dren (van IJzendoorn, 1995). Mothers who
have argued that it is a manifestation of a psychiatric diagnosis and complex behav- provide poor care (in terms of neglect or
perverse relationship between mothers and iours that are deemed criminal or other- maltreatment) are more likely to have
doctors, or it is an extended form of self- wise dangerous by the courts. insecure models of attachment than
harm (Schreier, 2002). We suggest, based mothers who provide good enough care
on our data, that the mother’s own care- (Crittenden et al, al, 1991). Somatising
giving and care-eliciting behavioural sys- Illness behaviour and childhood mothers have children with more health
tems have become disorganised, perhaps care and attachment problems and with higher medical consulta-
as a consequence of unresolved grief after Factitious illness by proxy might be better tion rates than those of well mothers or
bereavement or as one aspect of disordered understood using the concept of ‘abnormal organically ill mothers (Craig et al,
al, 2002).
personality. Personality disorder is a com- care-eliciting behaviours’. Mechanic (1978) The fact that these mothers involve
mon diagnosis in perpetrators of child mal- described a number of different variables, their children in abnormal illness behaviour
treatment generally (Famularo et al,al, 1992; both social and individual, associated with suggests that their attachment to the chil-
Dinwiddie & Bucholz, 1993), and particu- ‘normal’ care-eliciting behaviour. Hender- dren, and related capacity to care for them,
larly in those with factitious illness by son (1974) first suggested that adverse has become disorganised. The argument
proxy (Bools et al,al, 1994). childhood experiences of being cared for that their experience of illness in childhood
We did not find a high prevalence of might explain ‘abnormal’
‘abnormal’ care-eliciting be- affects their proxy care-eliciting behaviours
personality disorder in our group. This haviour in adulthood, such as malingering is consistent with the work of Craig et al
may be because we were not able to use a or fabricating accounts of symptoms, and (2002) on somatising mothers and research
structured assessment measure of personal- self-harming behaviours. Bools found a his- by Hill et al (2004), who found that a
ity disorder in the whole group, and were tory of factitious disorder and somatising mother’s experience of maternal care
able to use the MCMI–III in only half the disorder to be more common in this group; affected how she related to professionals
sample. Our data suggest that abusive or hence the suggestion that the mothers them- involved in the care of her child’s health.
risky behaviour is not synonymous with a selves should be diagnosed as having facti- It is reasonable to assume that factitious
diagnosis of personality disorder, and may tious illness by proxy (Bools et al,
al, 1994, illness by proxy behaviours represent soma-
be compatible with comparatively good Bools, 1996). In our series, a subgroup of tising disorder extended to the care of a
mental health. Also, the attachment data the mothers did have apparent histories of child. This may be true for a minority but
may reflect an aspect of psychological func- somatising disorder, insofar as they either is unlikely to be a sufficient explanation;
tioning that is not measured by personality had been given that diagnosis in the past rather, somatising disorder in mothers
disorder assessment tools. Alternatively, if or their medical records strongly indicated may be a risk factor in some cases, if only
personality disorder is considered a disor- such a disorder (repeated presentation for because it indicates both personality
der of interpersonal functioning, then per- unexplained medical symptoms in them- pathology and heightened anxiety in
haps it is possible for the disorder to be selves, as opposed to their children). relation to physical illness.
localised to one set of relationships – Somatising disorders are arguably an-
perhaps those characterised by disparities other form of abnormal illness behaviour,
of vulnerability or those that have an in which individuals present to GPs and The mothers’ attachment styles
attachment function. One might then think hospital medical services with medically In terms of their attachment histories, the
of these mothers as having a personality unexplained symptoms. There is evidence study group differs from non-clinical sam-
disorder confined to caring relationships; that the experience of childhood illness ples of normal mothers. They resemble clin-
just as some men are only abusive or cruel and lack of parental care is associated with ical samples, with an excess of unresolved
in the context of marital relationships. somatising disorders in adulthood (Craig et states of mind and CC transcripts. This sup-
As expected, we did not find any con- al,
al, 1993; Hotopf et al,
al, 1999; Craig et al,
al, ports the thesis that these mothers are in
nection between any one type of AAI classi- 2002; Crane & Martin, 2002). One expla- highly disorganised states of mind when
fication and factitious illness by proxy nation for the abnormal caregiving behav- they think about caring and being cared
behaviour. The illness behaviour is likely iour in our study is that caregiving for in their own childhoods, and confirms
to be a ‘final common pathway’ of a gener- attitudes were disorganised by an experi- that an attachment paradigm is reasonable
ally unstable relationship between mother ence of illness in childhood. Out of 67 for hypothesis generation and testing for
and child, and evidence of the mother’s mothers, 37 (55%) had experience of being this type of child maltreatment.
3 31
A D S H E A D & B LU G L A S S
The excess of insecure attachment attachments to others and reduce parenting It is also problematic to make inferences
representations is in accordance with the capacity as much as other types of psycho- from mothers referred for forensic assess-
Henderson hypothesis (Henderson, 1974) logical dysfunction. Unresolved distress in ment. Clearly this leads to sampling bias in
that insecurity of attachment influences relation to trauma would certainly affect that only identified cases will be included,
abnormal care-eliciting behaviour in adults, ability to care for a child, and has signifi- which are likely to overrepresent cases of
and with the now abundant evidence that cant therapeutic implications. A subgroup perceived high risk to children. In terms of
early childhood experience of illness influ- of these women might benefit from inter- psychological mechanisms, it would be use-
ences medical help-seeking behaviour, both ventions for post-traumatic stress sympto- ful to have a comparison group of mothers
for adults themselves and when they seek matology, such as cognitive–behavioural who present with less extreme or harmful
care on behalf of others (Craig et al, al, therapy, exposure therapy or antidepres- abnormal care-eliciting by proxy, although
2002). Our data are also consistent with sant medication. Bereavement during it is hard to see how these might be identi-
evidence that attachment style influences pregnancy or in the first postnatal year fied. Sampling bias could also be introduced
general illness behaviour in both adults might also significantly affect the mother’s in the choice of expert assessor, which is
and children (Feeney & Ryan, 1994; Cie- attachment to her own child, and this made partly by lawyers, but partly by family
chanowski et al, al, 2001; Wilkinson, 2003). could be asked about routinely, so court judges. However, judges’ decisions are
The resemblance of our sample to a that therapeutic interventions could be affected by factors independent of the asses-
clinical group is striking, given that only a offered. sor, such as pressure on court time and avail-
minority of participants received psychi- ability of other witnesses, and it is hard to
atric diagnoses or (for those who completed see how this would affect the results.
the MCMI–III) showed abnormal person- Strengths and weaknesses It is possible that because all our partici-
ality traits. It is possible that standardised of the study pants were involved in legal proceedings of
diagnostic interviews would have revealed There are no other studies to our knowl- enormous personal significance, this made
more psychiatric pathology. However, only edge that have examined attachment repre- them more defensive in their interactions,
a minority had a previous psychiatric sentations in individuals with abnormal and affected the coherence of the narrative
history, and these were usually those with illness behaviour by proxy. The main discourse. This could only be assessed using
clinical elevations on the MCMI–III. We strength of the study is the large size of a comparison group; for example, if the AAI
suggest that these data show the complexity the sample. The nature of the behaviour were offered in the context of treatment for
of the relationship between individual psy- (being both deceptive and illegal) has made these women, this might produce different
chopathology and interpersonal behaviour it difficult to identify large samples of narratives. It is highly unlikely that any
within family groups. Forensic studies on mothers for study. The Stafford group mother who had been detected in factitious
offenders within the family make it clear (Southall et al,
al, 1997) and the Bools illness by proxy behaviour would not face
that family violence is compatible with group (Bools et al,
al, 1994) both published legal proceedings; using samples in which
apparent psychiatric and psychological clinical data on such mothers, but their abnormal illness behaviour was merely sus-
health. This suggests that better research sample sizes (for interview) were 33 and pected would be practically difficult and
tools are needed to understand abusive 17, respectively. ethically problematic. It should also be
and exploitative relationships between Our attachment-based interviews pro- noted that the preponderance of the dismiss-
intimates. vide a wealth of data, not just about attach- ing category was found in other studies of
These data might also influence the advice ment representations, but also about the attachment in forensic groups.
we give to courts about mothers’ parenting mothers’ experiences of their children, their Our main finding is that mothers with
abilities. Attachment history and representa- fears, wishes and disappointments. The factitious illness by proxy are more like a
tion seem to be more relevant to the capacity study’s main weakness is the lack of a clinical group than not, and therefore are
to care for children than psychopathology per comparison group, allowing only limited likely to have unmet treatment needs,
se.
se. This is consistent with previous research inferences from our data. Insecurity of which may be relevant to future risk. There
which suggests that mental illness does not attachment is likely to be a risk factor for is a dearth of facilities for psychological
always reduce the capacity to parent. child maltreatment, prima facie,
facie, but it is and psychiatric treatment of maltreating
only one among many and is unlikely to parents, unless families are likely to be
be either predictive or specific. The type reconciled (Jones, 1987). Given the lack of
Unresolved distress of insecurity may be relevant, as may the resources, this discrimination is not un-
and bereavement nature of the experiences that give rise to reasonable. However, all of the women in
We found high levels of unresolved distress that insecurity. This can only be explored our study were of reproductive age, and in
in this group. The U categorisation is asso- with a comparison group and a prospective two cases there had been previous estab-
ciated with mental illness and dysfunction study. It is not clear which comparison lished factitious illness by proxy behaviour,
in some studies (Hesse, 1999). Although group would be most effective. Eminson which suggests that the behaviour might
the prevalence of childhood abuse or (2000) argued that mothers with psychi- recur. We would argue for the development
neglect was similar in our sample to that atric illness and normal children are the of psychological treatment services for
in previous studies, the key issue psycho- most informative comparison group, all maltreating parents, to include
logically is the degree to which the indivi- whereas Meadow (2000) suggested psychological therapy, psychoeducational
dual has come to terms with that mothers of children with established interventions and medication where appro-
experience mentally. If there is unresolved physical illness. We agree with Meadow priate, and an active approach to risk
distress, then this can disorganise and are recruiting for such a study now. management. From our own clinical
332
AT TA C H M E N T R E P R E S E N TAT I ON S IN A B NOR
NO R M A L I L L N
NEE S S B E H AV I OU R BY P R OX Y
333