Background In people with learning disability one of the most frequent reasons for psychiatric referral is challenging behaviour.

Aims To determine what proportion of people with challenging behaviour actually have psychiatric symptoms.

Method Using an instrument specifically designed for use by informants, a sample of 320 people with administratively defined learning disability, with and without challenging behaviour, was surveyed for the presence of psychiatric symptoms.

Results Increasing severity of challenging behaviour was associated with increased prevalence of psychiatric symptoms, depression showing the most marked association. Anxiety symptoms were associated with the presence of self-injurious behaviour.

Conclusions There is clearly the potential for reducing challenging behaviour by improved identification and treatment of coexisting psychiatric disorders. The possibility of modifying diagnostic criteria for depression in people with learning disability, by including aspects of challenging behaviour, merits attention.

A total of 16.7% of people with administratively defined learning disability in the UK have challenging behaviour (e.g. aggression or self-injury) (Qureshi & Alborz, 1992). This behaviour is costly to manage and frequently leads to significant burden of care, high health care and social costs and social exclusion (Felce et al, 2000). Recently, it has been suggested that some challenging behaviour may be caused or exacerbated by a coexisting psychiatric disorder (Emerson et al, 1999). Prompt identification and treatment of these psychiatric disorders could thus have a positive impact on the challenging behaviour, and hence on the burden of care, the quality of life of the individual and carers and the cost-effectiveness of service provision. At the present time, however, the mental health needs of this group of individuals are poorly understood. This paper presents the results of a study investigating the prevalence of psychiatric symptoms in people with learning disability and challenging behaviour in the North West of England.


Design considerations

Age, IQ and gender are factors known to be associated with the presence of challenging behaviour (e.g. Borthwick-Duffy, 1994):

  1. the overall prevalence of challenging behaviour increases with age during childhood, reaches a peak during the age range 15-34 years and then declines;

  2. the prevalence of aggression, property destruction, self-injurious behaviour and other forms of challenging behaviour is positively correlated with degree of intellectual impairment;

  3. males are more likely to be identified as showing challenging behaviour than females.

Because the incidence of challenging behaviour decreases markedly in later life, the confounding effect of age was minimised by restricting the investigation to people under 60 years and matching by age bands. Because IQ is also a powerful confounder, the samples were stratified by their level of speech, as an approximate measure of intellectual level. There was no significant difference between the proportion of males in the sample groups.

Derivation of the samples

In 1988, seven district health authorities in the area then covered by the North West Regional Health Authority participated in a total population survey of the extent and nature of challenging behaviour among people with learning disability (Qureshi & Alborz, 1992; Kiernan & Qureshi, 1993). The study included individuals with challenging behaviour from age 3 to 87 years. A further study was undertaken in 1995 (Emerson et al, 1997). This involved: the attempted follow-up of all people identified in 1988 as showing more severe challenging behaviour; a repeat of the total population screening in two of the seven districts; and the attempted follow-up of all people identified in 1988 as showing less severe challenging behaviour. The current investigation uses the 1995 data on those individuals who were between 18 and 60 years of age at the time of follow-up (n=234).

A control group (n=86) was drawn from a study of the health needs of 200 people with learning disability, commissioned by Tameside Learning Disability Services. The control sample was drawn from those individuals living in staffed accommodation, because there were few missing cases (in comparison with those living alone or with family members). Those individuals identified on the Wessex Scale (Kushlick et al, 1973) as having no challenging behaviour were randomly selected and age-matched to produce the same age banding as in the challenging behaviour study. Those in the latter study tended to be younger than the Tameside population, so the selection was achieved by randomly removing Tameside cases from the upper age bands to achieve the same proportions.

Information collected

Challenging behaviour

A full account of the extensive information collected in this sample can be found in the report by Emerson et al (1997). Sample members in the challenging behaviour studies were rated on four basic types of challenging behaviour: aggression; destruction of property; self-injury; and other unacceptable behaviour. Each of these types was rated on a four-point scale (serious, controlled, lesser, no problem). Overall severity of challenging behaviour was estimated using a compound dichotomous variable (less demanding/more demanding) derived from other information collected on the sample members. A client was coded as ‘more demanding’ if any of the following applied:

  1. the person showed any challenging behaviour at least once a day;

  2. the challenging behaviour usually prevented the person from taking part in programmes or activities that would otherwise be appropriate to his or her level of skills and competence;

  3. physical intervention by more than one member of staff was the usual level needed to control any of the person's challenging behaviour;

  4. the usual consequence of challenging behaviour was major injury to the self, to another person with learning disability or to staff; ‘major injury’ was defined as that requiring hospital treatment (e.g. broken bones, cuts or stab wounds).

Level of learning disability

There were no formal measures of intellectual ability collected on the sample members. However, a strong indication could be inferred from sample members' level of speech. In both the challenging behaviour study and the control group this was measured on a three-point scale (no language, words and phrases, full sentences).

Psychiatric symptoms

The Psychiatric Assessment Schedule for Adults with a Developmental Disability (PAS-ADD) Checklist (Moss et al, 1998) is a screening instrument designed to identify mental health problems in people with learning disability. In both the challenging behaviour and the Tameside study, a PAS-ADD Checklist was completed for each sample member by an individual (staff member or family member) judged to be in the best position to rate the symptoms. The PAS-ADD Checklist has a four-point rating scale, designed to be the best compromise between the loss of information resulting from a binary scale and the unreliability resulting from too many points. However, because the psychometric properties of the four-point scale were not known, a conservative analysis was performed, items being dichotomised into symptom present/absent.


Sample details

Age banding of the total sample of 320 people with administratively defined learning disability was as follows: 18-30 years, n=129 (40%); 31-45 years, n=125 (39%); 46+ years, n=66 (21%). Numbers in each level of challenging behaviour were: no problem, n=86 (26.9%); less demanding, n=148 (46.2%); more demanding, n=86 (26.9%). There were 64% males (n=205) and 36% females (n=115). There was no significant difference in the proportion of males to females in the Tameside and challenging behaviour studies.

Speech results indicated that there was an expected strong association between level of challenging behaviour and level of learning disability (see Table 1). Most of the 234 individuals in the two challenging behaviour groups showed aggressive behaviour of various kinds, including self-injurious behaviour. However, 57% of the people in the study (n=133) did not manifest any self-injurious behaviour at all. As expected, those who did manifest self-injury (n=101) included a significantly larger proportion of individuals with no speech (see Table 2).

View this table:
Table 1

Relation between level of speech and severity of challenging behaviour

View this table:
Table 2

Distribution of individuals in the two challenging behaviour groups by level of language and presence/absence of self-injurious behaviour

Group comparisons of psychiatric symptomatology

Increasing severity of challenging behaviour was strongly associated with the mean number of psychiatric symptoms (range 0-16) scored on the PAS-ADD Checklist: no problem, n=1.4; less demanding, n=2.0; more demanding, n=4.3; p < 0.0001 (Kruskal-Wallis test). Post hoc comparisons using the Mann-Whitney test indicated that the difference between the group without challenging behaviour and the group with less demanding challenging behaviour was not significant. The difference between the groups with less and more demanding challenging behaviour was highly significant (P < 0.0001).

Table 3 gives details of the individual symptoms in relation to each level of severity of challenging behaviour. It can be seen that for 23 of the 26 items the prevalence was highest in the group whose challenging behaviour was more demanding.

View this table:
Table 3

Prevalence of Psychiatric Assessment Schedule for Adults with a Developmental Disability (PAS-ADD) symptoms in the three samples

Four diagnostic categories were then derived from the PAS-ADD scores: anxiety, depression, hypomania and psychosis. In relation to anxiety, depression and psychosis it was possible to identify items on the schedule whose diagnostic significance was very clear-cut. With regard to hypomania, the core symptom in the PAS-ADD Checklist is elevated mood. However, the diagnostic significance of this symptom was not considered to be sufficiently clearcut, so for the present purpose individuals with probable hypomania were defined as those with elevated mood in conjunction with reduced need for sleep (either broken sleep or early waking). Core symptoms were thus as follows: anxiety — phobic anxiety, non-situational anxiety; depression — depressed mood, suicidal intent/actions; hypomania — elevated mood plus early waking or broken sleep; psychosis — hallucinations, delusions.

Table 4 shows the prevalence of individuals who either had at least one core symptom in the anxiety, depression or psychosis symptom groups or met the criteria for hypomania. These results show the prevalence of psychiatric disorders to be high in the whole sample and very high in the group with more demanding challenging behaviour. All four categories showed an increasing prevalence with severity of challenging behaviour, although this did not reach significance in the case of anxiety and psychosis. Depression showed a very marked prevalence that was differential across the three groups. For the whole study, the overall prevalence of psychiatric disorders is in accord with other published studies (Campbell & Malone, 1991).

View this table:
Table 4

Percentages of individuals meeting defined diagnostic criteria, in relation to severity of challenging behaviour

Self-injurious behaviour

Within the 234 people in the challenging behaviour groups, the prevalence of PAS-ADD symptoms in people with and without self-injurious behaviour was investigated. In terms of the mean total number of psychiatric symptoms identified by the PAS-ADD Checklist, there was no significant difference. However, four individual symptoms showed significant differences in prevalence (Table 5).

View this table:
Table 5

The Psychiatric Assessment Schedule for Adults with a Developmental Disability (PAS—ADD) symptoms showing significantly different prevalence in people with self-injurious behaviour (n=101) compared with those who have challenging behaviour but no self-injurious behaviour (n=133)

The sympton ‘odd gestures and mannerisms’ probably features because people with self-injurious behaviour include a higher proportion of individuals with profound learning disability, many of whom demonstrate stereotyped behaviour. However, two of the symptoms are indicative of the fact that anxiety disorders are a possible factor in self-injurious behaviour (‘jumpy’ and ‘phobic anxiety’).


Significance of the association

The results provide some evidence for a statistical association between challenging behaviour and psychiatric disorders. The overall prevalence of psychiatric disorders, as measured by the four diagnostic categories, showed people with more demanding challenging behaviour to have over twice the prevalence compared with those who had no challenging behaviour. Depression was four times as prevalent in those whose challenging behaviour was more demanding than in people showing no challenging behaviour; hypomania was three times as prevalent. The strong association with depression is particularly important because this condition often remains undetected, both in the general population (Goldberg & Huxley, 1980) and in people with learning disability (Patel et al, 1993). It is thus probable that there are many individuals with learning disability and challenging behaviour who also have unrecognised psychiatric problems. This highlights the importance of introducing improved methods for identification of psychiatric problems in people with learning disability — methods such as the PAS-ADD Checklist (Moss et al, 1998) and the Mini PAS-ADD (Prosser et al, 1998).

Challenging behaviour as a possible diagnostic criterion for depression

Some authors (e.g. Meins, 1995) have suggested that challenging behaviour can sometimes be an atypical expression of depression and should be incorporated into diagnostic criteria modified specifically for this population. Although the results from the current study suggest that this assertion merits attention, it may be difficult to implement in practice. The determinants of challenging behaviour are likely to be highly complex — a combination of factors relating to history of learned behaviour and biological, environmental, social and psychological factors. Challenging behaviour may exacerbate a coexisting psychiatric disorder, whereas psychiatric disorders may express themselves partly in terms of a challenging behaviour. Given this complexity, the use of data on challenging behaviour to make psychiatric diagnoses would pose major questions of validity.

Self-injurious behaviour

Among people with self-injurious behaviour, anxiety disorders were identified as being more prevalent than among people without such behaviour. It is not clear whether this finding relates specifically to the presence of self-injurious behaviour or whether it is because this group contains more individuals with profound learning disability. It has been noted elsewhere (King et al, 1994) that anxiety disorder is one of the most frequent diagnoses made in people with this level of disability.

Reliance on untrained raters

A potential limitation of the current study was that the information on psychiatric disorders was provided exclusively by non-psychiatrists. Although the PAS-ADD Checklist has been validated for use by unqualified observers (Moss et al, 1998), accurate quantification of the statistical associations between challenging behaviour and psychiatric disorders would need a further study using comprehensive multidisciplinary assessment, including expert psychiatric opinion.

Clinical Implications and Limitations


  • Psychiatric disorders, particularly depression and hypomania, are significantly related to the presence of challenging behaviour.

  • There is a clear potential for reducing challenging behaviour through the identification and treatment of unrecognised psychiatric problems.

  • Improved methods for identifying psychiatric disorders in the community should be adopted as a routine part of service provision to people with learning disabilities.


  • Data on psychiatric symptoms were collected by support staff, rather than psychiatrists with expertise in learning disability.

  • Learning disability was administratively defined, rather than from an epidemiological perspective.

  • The intellectual level of the subjects could be inferred only approximately from their level of speech.

  • Received July 7, 2000.
  • Revision received May 30, 2000.
  • Accepted June 9, 2000.


View Abstract