Psychiatric symptoms and problem behaviours in people with intellectual disabilities
Abstract
Background Previous studies have suggested different patterns of associations between psychiatric symptoms and problem behaviours in people with intellectual disabilities (ID). The aim of this study was to investigate which problem behaviours are associated with specific psychiatric symptoms and the relative strength of these specific associations.
Method A cross sectional survey using the Psychiatric Assessment Schedule for Adults with Developmental Disabilities Checklist and the Disability Assessment Schedule was carried out in a sample of 214 adults with ID.
Results Self-injurious and, to a lesser extent, aggressive problem behaviours were most associated with affective type symptoms. Screaming and destructive behaviours tended to be more associated with autism-related social impairment rather than conventional psychiatric symptoms.
Conclusions This study gives further evidence of associations between psychiatric symptoms and specific problem behaviours in people with ID. It may be particularly useful to consider the diagnosis of affective disorders if a person with ID shows self-injurious or aggressive behaviours.
Introduction
Problem behaviours have often been considered as possible ‘behavioural equivalents’ of psychiatric symptoms in people with intellectual disabilities (ID) (Clarke & Gomez 1999). Recognizing and treating any associated psychiatric disorders in individuals with ID may therefore potentially reduce the human suffering and care burden caused by problem behaviours. It is still the case, however, that the relationships between problem behaviours and psychiatric symptoms remain an under-researched area. Cross-sectional studies have been the most common approach in the studies to date (McBrien 2003). Until very recently, study samples have tended to be small and non-standard diagnostic criteria and rating instruments were used. Some of these studies have suggested that associations between psychiatric symptoms and problem behaviours do exist, but others have not.
In the last 5 years, there have been three large studies published which have all shown associations between psychiatric symptoms and problem behaviours in adults with ID. Moss et al. (2000) used the Psychiatric Assessment Schedule for Adults with Developmental Disabilities (PAS-ADD) Checklist to estimate prevalence of different psychiatric symptoms in 320 people with ID who were also rated for presence of problem behaviours. Psychiatric symptoms were increased significantly in those with problem behaviours. Diagnostic categories were then derived from the PAS-ADD Checklist scores, although these did not correspond with ICD-10 diagnostic criteria. The prevalence of ‘depression’ and ‘hypomania’ was highest in those participants with ‘more demanding’ behaviours. Holden and Gitleson (2003) also used the PAS-ADD Checklist in a sample of 165 adults with ID to show that problem behaviours were associated with an increased prevalence of psychiatric symptoms. This study found that ‘anxiety’ and ‘psychosis’ were the diagnoses most commonly associated with problem behaviours. Once again, non-standard diagnostic criteria were used. Rojahn et al. (2004) studied 180 adults with more severe ID who were assessed with the Behavior Problems Inventory (Rojahn et al. 2001) and the DASH-II (Matson 1995). Those with self-injurious, aggressive/destructive or stereotyped behaviours had generally higher levels of psychopathology.
The patterns of the relationships found in the foregoing three studies have thus been inconsistent. Such relationships that have been shown may also have occurred simply by chance. Thus, there is a need for a further large study using standardized instruments, not only to strengthen the evidence base for any associations, but also to explore and clarify their likely complex patterns. Also, studies such as those described above have generally tried to look at associations at the syndromal level, but a symptomatic approach to the classification of psychiatric symptoms and problem behaviours may be more relevant to adults with ID, especially at more severe levels (Deb et al. 2001).
Aims
The aims of this study were to answer the following two research questions: Which psychiatric symptoms (if any) are associated with specific problem behaviours? To what extent do the presence of certain psychiatric symptoms predict the simultaneous presence of specific problem behaviours?
Method
Study design
A cross-sectional survey was carried out to determine the simultaneous prevalence of psychiatric symptoms and problem behaviours in a large sample of people with ID.
Participants
Participants were recruited from a total sample of 739 adults (18–85 years) known to a local register for people with ID in Southeast London. Of this total sample, 359 replied to a letter inviting them to take part in a concurrent study. Of those who replied, 214 gave consent and/or assent and took part in subsequently arranged clinical assessments. Residential status of participants included a variety of settings such as care homes, family homes and independent and supported living. More than half (60.3%) were in residential care. Participants included both those with and without known problem behaviours. Ethics approval was obtained from the local research ethics committee to collect data for this specific purpose. Exclusion criteria were lack of consent or assent to take part from participants or carers, living too far away or repeatedly failing to attend for arranged assessments.
Data collection
Assessors consisted of three psychiatrists and one other trained rater. The same assessor recorded all the following data simultaneously for each of the participants:
Demographics
Age (under 35 or 35 and over), gender and severity of ID (mild/moderate or severe/profound) were also recorded.
Psychiatric symptoms
The participants of the study were screened for 28 psychiatric symptoms using the PAS-ADD Checklist (Moss et al. 1998). The PAS-ADD Checklist has been independently investigated for reliability (Sturmey et al. 2005). The data on psychiatric symptoms were collapsed into categories of ‘present’ or ‘absent’.
Level of intellectual disabilities
Clinical levels of ID, ‘mild’, ‘moderate’, ‘severe’ and ‘profound’ following ICD-10 (WHO 1992) diagnostic criteria were determined from the IQ scores when available in the clinical notes and/or from clinical assessment. The data were collapsed into binary categories of ‘mild/moderate’ and ‘severe/profound’.
Social impairment
Participants were assessed for presence or absence of autism-related ‘social impairment’ with the ‘quality of social interaction’ item from the Disability Assessment Schedule (DAS) (Holmes et al. 1982). The DAS has been investigated for reliability (Holmes et al. 1982; Ardoin et al. 1991) and has been widely used in research in people with ID. The data on pattern of autism related social impairment were collapsed to categories of ‘present’ (scores from 0 to 4) or ‘absent’ (scores 5 or 6) following the DAS guidelines for the scores on this item (Holmes et al. 1982).
Problem behaviours
The presence or absence of 13 problem behaviours for each individual was recorded using the ‘Behaviour problems’ section of the DAS. Data recorded were collapsed into binary categories of ‘severe problem’ (scores 0–5) and ‘lesser/no problem’ (score 6–8) corresponding to the problem behaviour being ‘present’ or ‘absent’.
Analysis of results
Data were analysed using spss (Version 11.0). Cross-tabulations with chi-squared tests (with correction using Fisher's exact test if appropriate) were performed with age, gender, level of ID, each psychiatric symptom and autism-related social impairment and each individual problem behaviour. Statistical advice was sought regarding the level of significance set for associations on bivariate analysis. As this was primarily an exploratory study, the Bonferroni correction was considered unnecessary. Therefore, those independent variables significantly associated (P < 0.05) in each bivariate analysis with each of the seven most prevalent problem behaviours were then entered into seven separate corresponding logistic regression models. This allowed odds ratios to be determined, that is, to determine which psychiatric symptoms and/or social impairment most strongly independently predicted the presence of each of the specific problem behaviours by comparison with a ‘reference’ participant (female, over 35, mild or moderate ID, no psychiatric symptoms or social impairment).
Results
Data were collected on 214 study participants. The study group was almost equally divided by gender (49.5% male, 50.5% female). The participants were predominantly older with most aged more than 35 years (72%). The majority (64%) of the sample had mild or moderate ID compared to more severe ID (36%).
Seven of the 13 problem behaviours were shown by at least 10% of the sample, this giving sufficient numbers for further statistical analysis. These were ‘difficult or objectionable personal habits’ (16.8%), ‘screaming’ (16.4%), ‘destructive’ (15%), ‘tantrums’ (14%), ‘aggressive’ (12.6%), ‘pesters/seeks attention’ (12.6%) and ‘self-injurious’ (10.3%) behaviours.
Psychotic symptoms, such as ‘strange experiences’ (3.3%) and ‘strange beliefs’ (1.1%) were of low prevalence. Affective symptoms such as ‘irritable mood’ (19.2%), feeling ‘sad or down’ (14.5%) and loss of enjoyment or interests’ (10.7%) were more common. Autism-related social impairment, as measured by the DAS, was more common than the psychiatric symptoms measured by the PAS-ADD Checklist; it was shown by 45% of the sample.
Table 1 shows the comparison of 23 independent binary variables (psychiatric symptoms or social impairment absent/present) that showed at least one significant association with the dependent binary variables (no or lesser problem/severe problem) of the seven most prevalent problem behaviours in the sample. Some psychiatric symptoms showed no associations with any of the most frequent problem behaviours. These were: ‘reduced concentration’, ‘increased appetite’, ‘easily startled’, ‘sudden fear’, ‘more forgetful/confused’, ‘suspicious’, ‘strange experiences’, ‘strange beliefs’, and ‘startled’.
| Independent variable | Problem behaviours | ||||||
|---|---|---|---|---|---|---|---|
| Aggressive | Self-injurious | Screaming | Destructive | Difficult habits | Tantrums | Pesters/seeks attention | |
| Age < 35 | * | * | |||||
| Severe ID | * | * | |||||
| Low energy | ** | ** | ** | ** | |||
| Anhedonia | ** | ||||||
| Sad or down | ** | ||||||
| Fearful/Panicky | ** | ||||||
| Repetitive actions | * | ||||||
| Too high or too happy | * | * | * | * | |||
| Suicidal | * | ** | |||||
| Loss of appetite | * | ||||||
| Weight change | * | ** | |||||
| Loss of confidence | ** | * | |||||
| Avoiding social contact | * | * | |||||
| Worthlessness | * | ||||||
| Delayed sleep | * | * | ** | ||||
| Early waking | ** | * | ** | * | |||
| Restlessness | * | ||||||
| Irritable mood | ** | ** | ** | * | ** | ** | |
| Loss of self-care | * | ||||||
| Odd language | * | ** | * | ** | * | ||
| Social impairment | ** | ** | * | ||||
- Chi-squared tests *P < 0.05; **P < 0.01.
- ID, intellectual disabilities.
A few symptoms were associated with several of the most frequent problem behaviours. The symptom of ‘irritable mood’ was associated with six of the seven problem behaviours. ‘Odd or repetitive language’ was associated with five of the seven problem behaviours. ‘Social impairment’ was associated with ‘screaming’, ‘destructive behaviours’ and ‘difficult personal habits’. Affective symptoms were associated with ‘self-injurious behaviours’ and ‘aggressive’ behaviours.
Each of the separate stepwise binary logistic regression models for the seven most frequent problem behaviours were significant. Table 2 shows the Chi-squared, P-values, −2 log likelihoods and the R-squared values for the different models.
| Problem behaviours | Chi-squared | Significance (P-value) | −2 Log likelihood | R-squared (Cox and Snell) |
|---|---|---|---|---|
| Aggressive | 19.3 | 0.00 | 134.0 | 0.09 |
| Self-injurious | 23.4 | 0.00 | 113.6 | 0.105 |
| Screaming | 20.9 | 0.00 | 165.4 | 0.094 |
| Destructive | 26.1 | 0.00 | 154.5 | 0.115 |
| Difficult personal habits | 6.93 | 0.04 | 187.0 | 0.032 |
| Tantrums | 19.5 | 0.00 | 142.5 | 0.098 |
| Pesters or seeks attention | 22.5 | 0.00 | 139.2 | 0.101 |
Table 3 shows the independent variables (psychiatric symptoms and autism-related social impairment) that predicted the presence of the seven most frequent problem behaviours. It shows the predicted odds ratios and 95% confidence intervals of a participant having specific problem behaviours present if a specific psychiatric symptom or ‘social impairment’ was also present.
| Problem behaviour | Predictor variable | Odds ratios | 95% CI | P-value |
|---|---|---|---|---|
| Aggressive | Early waking | 4.04 | 1.08–15.1 | 0.04 |
| Loss of energy | 3.72 | 1.21–11.4 | 0.02 | |
| Irritable mood | 3.0 | 1.16–7.8 | 0.02 | |
| Self-injurious | Irritable mood | 5.52 | 1.99–15.3 | 0.00 |
| Suicidal ideas | 5.19 | 1.22–22.1 | 0.03 | |
| Age < 35 | 2.81 | 1.02–7.73 | 0.04 | |
| Screaming | Odd language | 3.47 | 1.15–20.5 | 0.03 |
| Social impairment | 3.1 | 1.36–7.1 | 0.01 | |
| Destructive | Loss of energy | 4.36 | 1.43–13.3 | 0.01 |
| Social impairment | 4.09 | 1.7–9.82 | 0.00 | |
| Delayed sleep | 3.28 | 1.1–9.76 | 0.03 | |
| Difficult personal habits | Severe ID | 2.66 | 1.28–5.5 | 0.01 |
| Tantrums | Odd language | 5.25 | 1.6–17.3 | 0.01 |
| Irritable mood | 3.44 | 1.37–8.64 | 0.01 | |
| Pesters or seeks attention | Age < 35 | 2.52 | 1.03–6.15 | 0.04 |
| Irritable mood | 3.64 | 1.44–9.19 | 0.01 | |
| Panicky | 7.67 | 1.99–29.6 | 0.00 |
- CI, confidence interval; ID, intellectual disabilities.
The presence of autism in some individuals may have increased the risk of them manifesting certain problem behaviours. For example, the odds of a participant showing ‘screaming’ was increased by them also showing ‘odd or repetitive language’ (×3.47) and ‘social impairment’ (×3.1). The presence of autistic features may thus have increased the risk of this specific problem behaviour more than any symptom of mental illness.
Other results suggested that mental illness, in particular affective disorders, might have increased the risk of prevalence of certain problem behaviours in the study sample. For example, the odds of a participant showing ‘aggression’ were increased with the presence of three affective symptoms: ‘early waking’ (×4.04), ‘loss of energy’ (×3.72) and ‘irritable mood’ (×3.0). Similarly, the odds of a person showing ‘self-injurious behaviours’ were strongly increased by the affective symptoms of ‘irritable mood’ (×5.52) and ‘suicidal ideas’ (×5.19), and also simply by a person being of age less than 35 years (×2.81).
Other results showed a mixed picture. For example, the odds of a person showing ‘destructive behaviours’ was increased not only by the presence of ‘loss of energy’ (×4.36) and ‘delayed sleep’ (×3.28), but also by ‘social impairment’ (×4.09). The odds of a person showing ‘difficult personal habits’ were only increased significantly when that person had more severe ID (×2.66) and not by psychiatric symptoms or social impairment. The odds of a person showing the problem behaviour of ‘pesters or seeks attention’ was increased by ‘young age’ (×2.52) and ‘irritable mood’ (×3.64), but most increased by the symptom of ‘panicky or fearful’ (×7.67).
Discussion
Although it has often been assumed that psychiatric symptoms are associated with behavioural problems in people with ID, there has been no great evidence base for this assumption. This study reports data from a large sample of people with ID using standardized rating instruments, such as the PAS-ADD Checklist (Moss et al. 1998), which has allowed better comparison between this and the few existing studies. The data thus add further to the evidence base regarding the associations between psychiatric symptoms and problem behaviours. It supports the potential link between self-injurious and aggressive behaviours with affective disorders (Reiss & Rojahn 1994). The close association of irritable mood with problem behaviours has also been often noted (Meins 1995). These data support the long-standing idea that irritable mood may be a more frequent sign or symptom of distress in a person with ID compared to sadness or euphoria (Reiss & Rojahn 1994). Self-injurious and aggressive behaviours were most associated with affective disorder type symptoms. These results are of interest as depression has been the mental disorder most commonly suggested to be associated with problem behaviours (McBrien 2003).
Other findings from this study challenge previous findings. Age under 35 years, gender and severity of ID have been linked in previous studies to presence of problem behaviours (Borthwick-Duffy 1994). Aggressive behaviours were not associated with age, gender or severity of ID in this study. Self-injurious behaviours were associated with younger age but somewhat surprisingly not with gender or severity of ID. Younger age was only otherwise associated with the problem behaviour of pestering or seeking attention. The gender of participants was not significantly associated with any of the seven problem behaviours investigated. Having more severe ID was only associated with two problem behaviours: difficult or objectionable personal habits and tantrums.
Moss et al. (2000) acknowledged that one limitation of their own study was that it involved the rating of psychiatric symptoms by untrained carers. Therefore, a strength of this study is that it involved trained assessors. A further strength is that data were collected on autistic features because certain phenomena that might indicate psychiatric symptoms, such as social withdrawal and flat affect, may also be seen in people with autism who have no comorbid mental illness (Perry et al. 2001). An additional strength of this study is also that it took a symptomatic rather than the syndromal approach used in many previous studies.
Implications
This study gives further evidence that some problem behaviours are underpinned by psychiatric disorders in a proportion of individuals with ID. Understanding these associations may help to improve the formulation, accurate diagnosis and management of psychiatric disorders in this population. For example, these data suggest it is important to consider the possible diagnosis of an affective disorder if a person with ID shows self-injurious and/or aggressive behaviours.
This study also suggests that some problem behaviours, such as screaming, may be more likely when an individual has autism-related social impairment rather than psychiatric symptoms. It would be useful to better clarify the relative importance of autistic features and psychiatric symptoms in determining the expression of problem behaviours in people with ID. This may have important management implications given growing evidence that medication can reduce the severity and frequency of problem behaviours in adults with ID and autism who do not have any other clearly diagnosable psychiatric disorder (McDougle et al. 1998).
It would be also useful to explore the findings that gender and severity of ID were not largely associated with behavioural disturbance. These findings challenge those from both older studies and long-held orthodox views. Further research to clarify these specific findings would be therefore of interest.
Limitations
This study has several limitations. It must be noted that the study sample, although consisting of all adult ages and levels of ID, did consist predominantly of those from an older age group and with mild and moderate ID. Some of the PAS-ADD Checklist and DAS items are difficult to rate accurately (if at all) in people without verbal communication. This is particularly true of psychotic symptoms, which are predominantly language-based (King et al. 1994). There were low levels of psychotic symptoms present relative to affective symptoms in this sample, which thus reduces the power of this study to detect potential associations of behavioural disturbance and psychosis in people with ID. The DASH-II is the only well-validated psychopathology-rating instrument designed for severe and profound ID group, but is not designed for use in people with more mild ID. There is also the problem that psychiatric disorders may manifest with exacerbation in frequency and severity of background levels of problem behaviours and/or symptoms. The DAS, PAS-ADD Checklist and indeed other existing rating instruments arguably do not capture these nuances of change in behaviour and/or mental state. Data collected using the DAS and PAS-ADD Checklist were also collapsed into dichotomous categories to allow ease of analysis although this did mean loss of some of its complexity. Consideration of symptoms and problem behaviours that are ‘present’ or absent′ does, however, reflect the considerations of diagnosis and clinical practice. Difficulties also remain in that problem behaviours are not clearly defined (Lowry 1997). Deficiencies in rating instruments remain ongoing methodological problems in this field. However, the PAS-ADD Checklist and the DAS were used for these are well-recognized instruments for this population and are among the best currently available.
Social impairment may also be found in a range of psychiatric disorders, such as personality disorders and schizophrenia as well as in autism. This makes the postulated associations of specific problem behaviours with autism in this study tentative at this stage pending further research. Finally, it is most important to acknowledge that association is not causation. The logistic regression models did not account for the multitude of bio-psycho-social factors that are likely to be operating in each case of problem behaviour occurring in an adult with ID (McClintock et al. 2003). Therefore, such models allowed the possible relative importance of psychiatric symptoms to problem behaviours to be estimated but without proving that they are the causative factors. Could the results have been significantly altered by other variables, for example, whether or not the participants were on medication? If symptoms and behaviours are closely and even causally related, then it could be argued that other variables should not unduly change any associations between these two specific variables. It might be considered in the specific case of medication that associations between symptoms and behaviours should hold true whether in an untreated or treated individual. There is though a possibility that medication (and other factors) could have had a differential effect on psychiatric symptoms and problem behaviours in certain individuals leading to a change in the relationships seen.
Conclusions
Problem behaviours may sometimes be associated with psychiatric symptoms in some adults with ID. It is unlikely, however, that the majority of problem behaviours are underpinned by psychiatric disorders. Furthermore, there are not likely to be any simple relationships between individual problem behaviours and specific psychiatric disorders. Tsiouris (2003) has thus argued that problem behaviours should not be considered core features of a psychiatric disorder in adults with ID. However, they may sometimes be atypical features of psychiatric disorders in this patient group. Studies such as this one can therefore be used to explore the validity of modified diagnostic criteria such as the DC-LD (Royal College of Psychiatrists 2001) in adults with ID compared to standard diagnostic criteria such as the ICD-10 (WHO 1992).
A stronger argument for accepting associated problem behaviours as ‘behavioural equivalents’ of psychiatric symptoms could be made if they were shown to fluctuate together in a related manner over time. This argument would be made even stronger, and be highly clinically significant, if it could be shown that treatment of conventional psychiatric symptoms also were shown to improve any associated problem behaviours. Two case-study reports involving only four patients (Lowry & Sovner 1992; Sovner et al. 1993) showed that as psychiatric symptoms were reduced over time, problem behaviours reduced in a closely related manner. It is somewhat surprising then that larger prospective studies have not been reported. Randomized controlled trials are eventually needed to assess treatment of coexisting problem behaviours and psychiatric symptoms, but these must be carefully based on existing evidence in order to increase validity. Further prospective (descriptive and intervention) studies are thus first needed to clarify further the likely complex relationships between psychiatric symptoms and problem behaviours.
Acknowledgements
Dr David Boniface, Department of Statistics, University College London.
