بررسی شیوع خشونت و خودآزاری مرتبط با سن در افراد مبتلا به کم توانی ذهنی: بازنگری

The aim of this study was to analyse statistically published data regarding the age related prevalence of aggression and self-injury in persons with intellectual disability. Studies including prevalence data for aggression and/or self-injury broken down by age band were identified and relative risk analyses conducted to generate indices of age related change. Despite conflicting results, the analysis conducted on included studies considered to be the most methodologically robust indicated that the relative risk of self-injury, and to a lesser extent aggression, increased with age until mid-adulthood, with some indication of a curvilinear relationship for self-injury. These conclusions have implications for the understanding of the development of different forms of challenging behaviour and the importance of early intervention strategies.

Research indicates a range in prevalence for challenging behaviours, such as self-injury and aggression, of 10–20% of all people with intellectual disability (Emerson and Bromley, 1995, Emerson et al., 2001, Kiernan and Qureshi, 1993 and Lowe et al., 2007). Similar prevalence rates of between 10 and 15% for self-injury have been reported (Ando and Yoshimura, 1978, Ballinger, 1971, Borthwick-Duffy, 1994, Eyman and Call, 1977, Jacobson, 1982, Kebbon and Windahl, 1986, Oliver et al., 1987 and Saloviita, 2000). The range of prevalence rates for aggression appears to be slightly larger with estimates of between 2 and 20% (Cooper et al., 2009, Cooper, 1998, Crocker et al., 2006, Harris, 1993 and Sigafoos et al., 1994). There have been few statistical analyses of age differences in prevalence rates of challenging behaviour. Several methodologically robust studies report an increase in challenging behaviour with age. More specifically, a small number of studies indicate an increase in prevalence until the mid-30s followed by decline (e.g. Kiernan and Kiernan, 1994 and Oliver et al., 1987). Many researchers however, have failed to identify any association between age and challenging behaviour (e.g. Fraser et al., 1986 and Hillery and Mulcahy, 1997) whilst others report age related changes in the prevalence of challenging behaviour but are imprecise with regard to the age at which the prevalence begins to change (Collacott et al., 1998, Hemmings et al., 2006, Kiernan and Alborz, 1996 and Maisto et al., 1978). Disparity in general and age related prevalence might result from the divergent methodologies and samples employed. For example, the prevalence of challenging behaviour is likely to depend on the form of the behaviour and definition. The criterion for the presence of behaviour also has a significant effect, so that the age related prevalence of severe self-injury causing tissue damage will be different than that for milder self-injury. Additionally, the size and origin of the sample will influence results so that, for example, prevalence of challenging behaviour is likely to be higher in a sample of participants recruited from institutions where individuals are referred as a result of challenging behaviour (e.g. Emerson et al., 2001). Establishing the prevalence of challenging behaviour with age might inform models of the development of the behaviour. Whilst both biological and operant processes have been implicated in the development of challenging behaviour (Oliver, 1993), the ontogeny of specific forms of challenging behaviour are not well understood. Guess and Carr's (1991) stage model indicates that self-injury emerges from repetitive behaviour, thus this behaviour is proposed to have a specific developmental trajectory. However, very little research has been devoted to the development of other forms of challenging behaviour. From a clinical perspective, being aware of age related changes in the prevalence of challenging behaviour would not only enable services to plan effectively for the future needs of children with intellectual disabilities, but could potentially help services to target early intervention at different age bands before the prevalence of challenging behaviour begins to increase. There are therefore, advantages to further investigating the prevalence of challenging behaviour with age. The aim of this study was to review and analyse published data regarding the age related prevalence of aggression and self-injury in persons with intellectual disability. These specific forms of challenging behaviour were reviewed due to their clinical significance and, generally, well defined nature. To generate an accurate review, the inclusion criteria for all studies included the provision of prevalence of aggression and/or self-injury by age band data in addition to the number of participants in each age band so that these data could be analysed statistically within each study (statistical differences in data across studies were not analysed, although these data were compared using visual analysis). Whilst this inevitably limited the number of studies included, this also enabled a robust assessment of the consistency of the data across studies with different sample sizes and methodologies and thus extended the findings of previous studies. When interpreting the results, the focus was on papers with a more robust methodology, including a larger, more representative sample with use of standardised measures with established psychometric properties. Several frequently cited articles include data on the prevalence of aggression and/or self-injury by age band, but do not report the number of participants in each age band and thus could not be included within this review. Although these studies cannot be included in this review, the trends are worth noting. Oliver et al. (1987) and Borthwick-Duffy (1994) report the highest prevalence of self-injurious behaviour to be in the teenage years, whilst the highest prevalence rate reported by Rojahn (1986) was in those in their mid-20s. Conversely, Griffin et al. (1987) reported a decrease in the prevalence of self-injury in 14–22 year olds compared to younger individuals aged 4–14 years. With regard to aggression, Borthwick–Duffy reported a slight increase in prevalence after the age of 20, although this difference is not analysed statistically. Conclusions drawn from comparisons between the results of different studies should be tentative as statistical significance of age related change in prevalence within studies is not evaluated. Nevertheless, these results do provide an indication of the trends demonstrated by published results not meeting criteria for inclusion in this study and allude to the need for a review.

3.1. The prevalence of aggression by age To investigate the prevalence of aggression by age, the prevalence of aggression by age band data as described by the twelve studies identified were examined. These results are shown in Table 5. Table 5. Prevalence of aggression (%/n) by age bands (years) for the eleven studies identified meeting criteria. Each cell contains the age band (italicised), as well as the % prevalence and number of participants in parentheses. Studies were divided into categories: those using child and adult samples, adult only samples and child samples only and Table 6 shows the relative risks of aggression across age bands for studies using child and adult as well as adult only samples. Table 6. Relative risk for the prevalence of aggression for each older comparison age group as compared to the index group for each total population study. Child and adult samples above and adult samples only below the bold line. Each cell denotes a five year age band and cells are merged to signify multiple age bands. The index group is the first cell on the left of each row. Bold = p < .01. As demonstrated in Table 6, the relative risk of aggression increased significantly with age in two of the studies using a child and adult sample. Jacobson's (1982) study indicated a significant increase in the relative risk (RR) of aggression in adults aged 22 years or over (RR = 1.33, CI = 1.19, 1.48) compared to individuals aged between 0 and 21 years. Rojahn, Borthwick-Duffy, and Jacobson's (1993) study also indicated that compared to participants aged between 0 and 10 year olds, those aged 11–20 (RR = 1.64, CI = 1.53, 1.76) and 21–45 (RR = 1.99, CI = 1.88, 2.11) years are at significantly greater relative risk of aggression. Tyrer et al.’s (2006) study suggests a general decrease in the relative risk of aggression with increasing age after the age of 20, with the majority of adults age bands at significantly less relative risk (RR range = .26–.66) than individuals aged less than 19 years. The exception to this was adults aged between 30 and 39 years (relative risk = .71, CI = .49, 1.02). The results of Crocker et al. (2006) also indicated a significant decrease in the relative risk of aggression in 50–59 year olds (RR = .56, CI = .41, .78) as compared to 18–29 year olds, although there were no significant differences in the relative risk of aggression for other age bands as compared to the index group. In contrast to the results of Rojahn et al. (1993) and Jacobson (1982), both Eyman and Call (1977) and Harris (1993) failed to detect any significant changes in the risk of aggression with age. Eyman and Call's results might have been influenced by the very large older age band used masking any significant trends within this group, although using far more narrow age bands, Harris also failed to identify any significant difference with age. In order to clarify this result, relative risk analyses were conducted using age bands from Harris’ data made similar to those used by Rojahn et al. This analysis demonstrated that according to Harris’ (1993) results, the relative risk of aggression did not differ significantly with age so that participants aged between 10 and 19 years (RR = 1.68, CI = .52, 5.47) and 20 and 44 years (RR = 1.86, CI = .61, 5.7) were at no greater relative risk of aggression than participants aged between 5 and 9 years, in contrast to Rojahn et al.’s (1993) results which indicated an increased relative risk of aggression in 11–20 (RR = 1.64) and 21–45 (RR = 1.99) year olds. Whilst the results based on Harris’ modified age bands were not significant, they were similar to the relative risks produced from Rojahn et al.’s results. It might be hypothesised that the significantly reduced relative risk of aggression with age in two of the total population studies employing adult only samples was the result of the older index group utilised, so that the increase in the relative risk of aggression has already occurred in the index group and thus shows no significant difference to the older age groups. To test this, the age bands utilised in Harris’ (1993) study were altered to match those utilised by Crocker et al. (2006). Using an index group of participants aged 20–29 years to compare to Crocker et al.’s findings, relative risk analysis indicated that participants aged between 30 and 60 years of more were at no greater relative risk of aggression than participants aged between 20 and 29 years. These results are similar to those of Crocker et al. except for the significantly reduced relative risk of aggression identified in 50–59 year olds in Crocker et al.’s study, although again, whilst not reaching significance, the results gained from Harris’ modified age bands were similar. Since these modified age bands were the same as those employed by Smith, Branford, Collacott, Cooper, and McGrother (1996), the results produced were also compared to those for this study. This comparison indicated similarities between the data, so that there were no significant differences in the relative risk of aggression with age, although the relative risk figures were quite different. Finally, in order to examine how the use of the older index groups had affected the results of Deb, Thomas, and Bright (2001), the age bands employed by Harris (1993) were further modified in order to compare to the results of these two studies. Employing an index group of participants aged between 15 and 29 years, relative risk analysis indicated that participants aged between 30 and 64 years were at no significantly different relative risk for aggression to the younger index group, results similar to Deb et al., although the relative risk figures were quite different. To summarise, the results of two studies of aggression employing child and adult samples indicated an increase in relative risk ranging from approximately 1.3–2.0 with age from childhood and teenage years to adulthood. Two total population studies employing adult only samples indicated a decrease in the relative risk of aggression within adulthood. Further relative risk analysis based on the results of Harris’ (1993) modified age bands, illustrated similar relative risks to these studies (although the results were not significant). Modifying Harris’ age bands to fit those of Smith et al. (1996) and Deb et al. (2001) also indicated no significant differences in the relative risk of aggression with age. In combination these analyses potentially indicate a curvilinear relationship between age and the prevalence of aggression with a significant increase between childhood and adulthood followed by a decline in prevalence in adulthood. The relative risks across age bands for each study with a child only sample (Ando and Yoshimura, 1978 and Tavormina et al., 1976) were calculated and indicate that the relative risk of aggression does not significantly increase in later childhood. Illustrating this, the results of Ando and Yoshimura showed that 11–14 year olds were at no greater relative risk of aggression than 6–9 year olds (RR = .13, CI = .01, 2.02), whilst Tavormina et al. also did not identify a significantly greater relative risk of aggression in 8–12 (RR = .44, CI = .07, 2.86) and 12–17 year olds (RR = .94, CI = .21, 4.16) as compared to an index group of children aged 4–6 years.