Main findings

We found high rates of chronic somatoform disorders in these women, with enduring somatoform disorders in 57% and fabricated symptoms in 64% (11 individuals had both chronic somatisation and factitious illness). In many participants the use of healthcare services was chaotic, with frequent visits to different accident and emergency departments, frequent changes of GP (often instigated by the GP) and lack of continuity of care. Discussion with the individual's GP seldom yielded any useful information because the GP's knowledge of the individual was often fragmented and brief. This association between persistent somatisation and factitious illness is known to occur in a subset of very high users of medical and surgical services. ^{Reference Fink10} We were able to establish a positive diagnosis of factitious disorder in our participants because we had access to so much clinical data from different sources.

Of the medically unexplained symptoms, pseudoneurological complaints (faints and pseudoseizures) and gastroenterological symptoms (abdominal pain, nausea) were the most common. Nine people (32%) reported pseudoseizures and, significantly, epilepsy and anoxic episodes had been diagnosed in seven of their children. Of the nine women who had been diagnosed as asthmatic, four also reported pseudoseizures. The association between pseudoseizures and reported asthma has been noted by others, ^{Reference de Wet, Mellers, Gardner and Toone11} and it has been suggested that anxiety, hyperventilation and dissociative elaboration might account for the observed association. Both asthma and anxiety/hyperventilation may be important risk factors for the development of pseudoseizures, but the reported asthma itself may be psychogenic in origin, as seems likely in the majority of our participants. The finding of pseudocyesis or false pregnancy in a fifth of mothers is of interest; to our knowledge it has not been reported before in these women.

Over half our participants had histories of self-harm, which often began in adolescence and in some continued into adult life. Only three had either current or lifetime histories of substance misuse, and ten had forensic histories; most often this involved shoplifting, theft or arson. Two women received custodial sentences after the assessment. Although we did not formally measure personality disturbance or disorder in our participants, a diagnosis of personality disorder had been established in the psychiatric notes in about three-quarters of those individuals who had received psychiatric treatment. Often coexisting with the personality difficulty was a defensive or sometimes frankly hostile attitude towards Social Services, especially among those women who had reported negative experiences while in care.

Methodological issues and limitations

These women represent a consecutive case series referred to one unit over a period of 15 years and so it is possible that they were not representative of individuals with fabricated or induced illness. All but three had been referred from the family courts for psychiatric assessment after episodes of alleged fabricated or induced illness, accompanied by a request for an opinion on the suitability of reunification of mother and child. Our sample is therefore biased towards those women who are detected, suspected and referred for detailed assessment, rather than a description of all women who carry out fabricated or induced illness behaviour. These limitations aside, however, we were able, because of the stringent requirements of the court, to obtain detailed medical and social histories for all of our participants except one. This included all relevant primary care data, which were invaluable in establishing the history and duration of each person's physical and psychological symptoms. The similarities between our data and the only other major published series ^{Reference Bools, Neale and Meadow6} lead us to believe that our sample is not atypical or unrepresentative. Furthermore, just over half the children in our series were victims of fabricated epilepsy or anoxic episodes, which is similar to the 42% reported in a well-known North American review. ^{Reference Rosenberg12} Seizures have also been reported to be the most common presentation of fabricated illness in children. ^{Reference Barber and Davis13} It remains possible that there is a wider group of people who carry out fabricated or induced illness behaviours, perhaps a milder version, who do not necessarily respond with hostility and lies when detected.

Early childhood experiences and intergenerational factors

The most striking abnormality in the childhood experiences of these women was the high rate of early family disruption and loss. Only 2 of the 28 had not experienced loss of a parent (through death, separation/divorce or removal of a parent to prison) before the age of 11 years, and over a third had spent time in foster care during their formative years. High rates of physical or sexual abuse have been reported by others, ^{Reference Gray and Bentovim14} and in another study insecure attachment as well as high rates of unresolved trauma and loss reactions was identified in these mothers. ^{Reference Adshead and Bluglass15}

The disruptive and turbulent childhoods experienced by the majority of these women is reflected in the very high referral rates to child and adolescent psychiatry services. Over half of the sample was assessed in these services, and three-quarters went on to have contact with adult psychiatric services. It is also worth noting that studies of children and adolescents with factitious disorder referred to child and adolescent consultation liaison services have revealed strikingly similar biographical and clinical characteristics to our sample. ^{Reference Ehrlich, Pfeiffer, Salbach, Lenz and Lehmkuhl16}

Recent studies have found support for the hypothesis that the children of parents with somatoform disorders are at increased risk of having high rates of contact with medical services as well as developing abnormal health beliefs. ^{Reference Craig, Cox and Klein17,Reference Marshall, Jones, Ramchandani, Stein and Bass18} Our findings confirm this intergenerational component, and also suggest an important association between individuals who fabricate symptoms first in themselves and later in their children. ^{Reference Feldman, Rosenquist and Bond19} For this reason our findings emphasise the important implication of establishing a diagnosis of factitious disorder in women of childbearing age. Because of the increased subsequent risk of fabricated illness occurring in a child, any pregnant woman who has received this diagnosis should undergo a comprehensive multidisciplinary assessment to assess the risk of harm to the unborn child.

Importance of pathological lying

Examination of social work, medical and psychiatric notes revealed that over half of the participants exhibited pathological lying (sometimes referred to as pseudologia fantastica ^{Reference King and Ford20} ) and in some this dated from a very early age; 6 years was the earliest recorded. These lies were often compulsive, habitual and sometimes self-aggrandising, and occurred throughout adolescence in many. In some they emerged at times of life stress and the episode of fabricated or induced illness was often accompanied by a re-emergence of lying and other deceitful activity such as repeated hoax telephone calls to police about ‘harassment’, which occurred in two women, financial fraud in one and nomadic wandering in another.

Enquiry into the motivations for feigning symptoms revealed links with these early abusive or key formative experiences in only about a third of the 28 women. For example, some patients ‘made up’ symptoms in order to avoid beatings or to prevent contact visits with abusive parents and others reported that they felt more cared for in hospital. In these cases developmental links could be detected between early disruptive and abusive experiences and later patterns of deceptive illness behaviour. The role of pathological lying in this behaviour is mentioned above.

We have already mentioned the possible links between early adverse experiences and subsequent illness behaviour: in over half our sample pathological lying was a key component of the psychopathology. It was often commented on in social work notes or case conferences. When confronted about the lying, individuals often responded by denial or disavowal of their contribution to the deception. As expected, all but 2 of the 17 participants exhibiting pathological lying also reported factitious physical symptoms in their own right.

Implications for clinical practice

Factitious disorder in adults and fabricated or induced illness (in children) can co-occur, so the detection of one should trigger a search for the other. Psychiatrists whose clinical work brings them into contact with parents with chronic somatoform and factitious illnesses should be alert to the impact of these disorders on the individuals' children. Our retrospective cohort does not allow us to identify which women with somatisation disorder and/or factitious disorder are at risk of carrying out this form of abuse. It may be that some women with a somatising disorder do not present a risk to their child, perhaps those at the milder end of the spectrum and those who respond to early identification without lying, hostility or deceit. Our sample necessarily describes a group who were detected, suspected and then referred, often from the family courts. It is probable therefore that our sample describes a severe end of a broader spectrum, although longitudinal studies of child-rearing women with a somatising disorder would be required to answer this question. At this point we are unable to speculate as to what proportion of women with a somatising disorder would go on to present a risk to their child. However, some ‘risk factors’ can be identified from our data, and identification of these should alert clinicians to those women who may be at increased risk of inducing abnormal illness behaviour in their children. These risk factors are shown in Appendix 2.

Our findings also suggest that no clinician should attempt to carry out an assessment of a woman suspected of this form of child abuse without access to extensive previous medical (and other) records. Neither should any opinion be expressed by the examining doctor about the mother's competence to look after a child based on a single assessment in an out-patient clinic. Our findings also imply that great care be taken in validating symptom reports by these women, especially those where evidence of pathological lying is suspected or has been established. In practice this means that the clinician needs to corroborate clinical data using a variety of different sources, both medical and non-medical, and a degree of healthy scepticism is required throughout the management of such cases.

Our sample of women had experienced high rates of early adverse experiences, including neglect and abuse as children and high rates of foster care. In adult life, somatoform disorders and factitious disorders also coexisted in nearly two-thirds of the women, with evidence of self-harm in about a half. The most frequently reported physical complaints were non-epileptic seizures and ‘psychogenic’ asthma, as well as gastrointestinal symptoms. Significantly, there were often similarities between the ‘functional’ symptoms reported by the mothers and the pseudoneurological symptoms fabricated or induced in their children. Such individuals are notoriously difficult to manage, but it is clear that the challenge of pregnancy and childbirth provide additional stressors for these women that invariably lead to ambivalent or hostile impulses towards their children. ^{Reference Adshead and Bluglass21}

Three-fifths of the mothers exhibited pathological lying from early adolescence and often this continued into adult life. When these mothers became distressed their lying invariably eventually involved the children, who then became the victims of fabricated and induced illness. Pathological lying is a rare but important symptom, and can occur in a variety of different clinical settings. ^{Reference Dike, Baranoski and Griffith9} Our findings suggest that detection of pathological lying in adolescent or young women should alert medical and social work staff to its potential significance when these women become pregnant. In some cases the pathological lying became established in early life to avoid some feared consequence or to elicit a desired response from others. It is important to stress, however, that the motivation for the induced illness in children was unclear in two-thirds of individuals. Reunification of the child with the mother can be attempted, but only in very carefully selected individuals. ^{Reference Berg and Jones22}