Skip to main content
Download PDF

Persistence and predictors of self-injurious behaviour in autism: a ten-year prospective cohort study

Molecular Autism volume 11, Article number: 8 (2020) Cite this article

Abstract

Background

Self-injurious behaviours, such as head banging, hair pulling, skin picking and scratching, are common in individuals with autism. Despite high prevalence rates, there is a paucity of longitudinal research to refine models of risk and mechanism and inform service planning. In this longitudinal study, we investigated self-injury in a cohort of individuals with autism over 10 years to identify behavioural and demographic characteristics associated with persistent self-injury.

Methods

Carers of 67 individuals with autism completed questionnaires relating to the presence of self-injury and relevant risk markers at T1 (mean [SD] age in years 13.4 [7.7]) and T3 (mean [SD] age in years 23.9 [7.7]) 10 years later. Forty-six of these also took part at T2 (3 years after initial participation). Analysis assessed demographic and behavioural risk markers for self-injury, as well as the predictive value of items assessed at T1and T2.

Results

Self-injury was persistent in 44% of individuals over the 10-year period, with behavioural characteristics of impulsivity (p < .001) and overactivity (p = .002), identified as risk markers for persistence. A predictive model of self-injury was derived from LASSO analysis, with baseline impulsivity, interest and pleasure, stereotyped behaviour, social communication and adaptive functioning predicting self-injury over 10 years.

Conclusions

In this unique longitudinal investigation into the persistence of self-injury in a non-clinical sample of individuals with autism over a 10 year period, we have identified a novel, robust and stable profile of behavioural characteristics associated with persistent self-injury. Findings support an early intervention strategy targeted towards individuals identified to be at a higher risk of developing self-injurious behaviour.

Background

Self-injurious behaviour (SIB), defined as a physical non-accidental act of producing injury to one’s body [1], encompasses behaviours such as head banging, hair pulling and skin picking [2]. In addition to the direct negative physical consequences of SIB, the presence of SIB increases the risk of family, educational and residential placement breakdowns [3], restrictive practices in primary care settings [4] and use of psychotropic medications [5]. Whilst SIB is detrimental to the individual and those around them, limited epidemiological data exist delineating the developmental trajectory of these behaviours. Given the significant financial burden for service providers [6] and noted lack of interaction with professionals to alleviate behaviours at the individual level [7], early intervention arguments to ameliorate the development of SIB are growing. It is imperative that mechanisms underpinning SIB are understood to optimise the value of such strategies.

Prevalence of SIB in individuals with autism is reported to be as high as 50% [8]; significantly higher than that for individuals with intellectual disability (12%) [9, 10]. Strikingly, the presence of characteristics associated with autism are associated with a higher prevalence of SIB [11] in multiple genetic syndromes indicating that both the presence of diagnosable autism and the presence of autism characteristics (such as stereotyped behaviour, insistence on sameness and repetitive use of language) elevate risk even in the highest risk groups [8, 11]. Prospective longitudinal cohort studies are required to explore the characteristics underpinning SIB, to ultimately reduce poor population outcomes for those with autism characteristics.

Current research demonstrates persistence of SIB across the life span [12], with one review suggesting SIB is both common and stable over time for individuals with autism [13]. Further studies also highlight persistence in adolescents and adults without autism and with broader developmental disabilities [14, 15]; however, further research is needed to extend this evidence. With research focussing on the development of SIB in clinical samples over short periods of time [9, 10, 16, 17], the naturalistic and longitudinal trajectory of SIB over extended periods of time remains unexplored. Cross-sectional data in people with intellectual disability contradicts the assumption of linear persistence, demonstrating a peak in SIB towards late adolescence before a fragmented decline with age [18]. Longitudinal research in autism is necessary to delineate purported age-related changes and describe the naturalistic developmental trajectory of SIB in a non-clinical sample.

Research provides evidence for demographic and behavioural risk markers associated with the presence of SIB [8, 17, 19, 20] that inform theoretical models. Historically, operant models explaining the maintenance of SIB have prevailed [21], yet such models do not consider the importance of individual characteristics, instead suggesting equal risk across individuals. Oliver and Richards proposed an extended model, integrating childhood characteristics which account for variability in both the presence of SIB and developmental trajectory [22]. Identification of demographic and behavioural markers as robust variables associated with the presence, severity and persistence of SIB in autism would lend further support to this model and implicate potential causal mechanisms driving poor clinical outcomes.

Overactivity and impulsivity have consistently been identified as behavioural characteristics associated with the presence of SIB [22]. Within multiple samples of individuals with autism, overactivity and impulsivity predict both the presence and severity of SIB [23,24,25], with emerging evidence suggesting that these characteristics also predict persistence [24]. Research further highlights this association amongst samples with genetic syndromes [21]. Importantly, overactivity and impulsivity are known behavioural markers for impairments in behavioural inhibition. Behavioural inhibition comprises both the capacity to inhibit prepotent responses to evoking stimuli and the capacity to inhibit a response once emitted [26,27,28]. Thus, the association between impulsivity/overactivity and self-injury alludes to a fundamental cognitive vulnerability which may act as a mechanism interacting with operant learning to drive both the presence and severity of SIB [22].

Concomitant evidence for this hypothesis is the presence of self-restraint. Self-restraint behaviours are those which restrict the movement of an individual’s body parts using clothing, objects or a person’s own body [29, 30]. Self-restraint is significantly more common in individuals with self-injury [31] and is described as serving the purpose of inhibiting severe SIB [29, 31]. The presence of these behaviours suggests that environmental and sensory contingencies alone are insufficient to constrain self-injury. Description of the putative association between SIB and self-restraint in a prospective longitudinal at-risk cohort, such as those with autism, will provide a useful context in which to evaluate the hypothesised model implicating impaired behavioural inhibition as a risk marker.

In summary, SIB leads to significant physical, financial and emotional cost for individuals and caregivers. A paucity of research has evaluated persistence of SIB in autism. Whilst current data support the cross-sectional associations of behavioural characteristics such as overactivity and impulsivity with SIB, there is little evaluation of these associations longitudinally. A prospective longitudinal cohort affords the opportunity to describe and evaluate the presence, persistence and predictive associations with SIB in autism. Time 1 data (T1) and subsequent 3-year follow up (T2) of this prospective cohort identified behavioural risk markers for persistent SIB within the current sample of individuals with autism [8, 24]. The present study (T3) extends the longitudinal study, investigating the persistence of SIB over 10 years. The following hypotheses are made:

  1. 1.

    SIB will be persistent between T1 and T3 for the majority of individuals

  2. 2.

    Higher levels overactivity and impulsivity at T3 will be positively associated with the following:

    1. (a)

      The presence of self-injurious behaviour at T3

    2. (b)

      The presence of self-restraint behaviours at T3

  3. 3.

    Higher levels overactivity and impulsivity at T2 will predict longitudinally the presence of self-injurious behaviour at T3

  4. 4.

    Higher levels overactivity and impulsivity at T1 will predict longitudinally the following:

    1. (a)

      The presence of self-injurious behaviour at T3

    2. (b)

      The presence of self-restraint behaviours at T3

Method

Participants

At Time 1 (T1) participants were recruited through the National Autistic Society [8]. All participants who consented to future contact were invited to take part in the present study (N = 241), independent of participation at the Time 2 (T2) follow- up [24]. Seventy-two participants completed the study (return rate: T2 = 35.78%, T3 = 29.58%). Participants were excluded if (a) they were under the age of four at T1, (b) they did not have a diagnosis of autism confirmed by a relevant professionalFootnote 1, (c) they had a diagnosis of a genetic syndrome and (d) they scored above the autism threshold on the Social Communication Questionnaire on fewer than two of the three time points. Five participants were therefore excluded, leaving a final sample of 67.

Procedure

Information packs containing an invitation letter and link to the online survey were sent to all participants. Using unique identifiers, participants completed the relevant consent forms, before being directed through each measure and submitting responses. Paper versions of questionnaires were available upon request. All participants were sent individual feedback reports upon completion of data analysis, detailing responses from participation in T1, T2 and T3 studies. Ethical approval for this study was obtained from the ethical review committee at the University of Birmingham.

Measures

The following questionnaires, suitable for carer report in individuals with intellectual disabilities, were included:

A demographic questionnaire detailing person characteristics, age, mobility and verbal ability was used. Inclusion allowed for the assessment of potential associations demographic characteristics may have in subsequent self-injury analysis. A service receipt sub-section was also included, detailing clinical services accessed over the 10-year period, and carer’s evaluation of their utility.

The Wessex was used to assess self-help adaptive functioning [32]. The questionnaire has been shown to be successful in measuring ability amongst those with an intellectual disability and has good inter-rater reliability at the subscale and item levels [33]. Inclusion allowed for exploration of how individual adaptive functioning had developed since T1.

The Activity Questionnaire (TAQ) assessed impulsivity and overactivity [34]. It consists of three subscales, and cut offs are established to account for unusually high scores [34]. The measure has been shown to have good inter-rater reliability (mean .56), test-retest reliability (mean .75) with assessments of internal consistency showing all subscales correlate to a moderate degree [35] Impulsivity was associated with persistent self-injury at T2 analysis [24], with current analysis therefore exploring the development of this association.

The Social Communication Questionnaire (SCQ) was used to assess behaviours associated with autism within the sample [36]. The measure demonstrates good concurrent validity (ADOS [37]; ADI-R [38]), and internal consistency (α = .90 for the total scale). It is a non-diagnostic screening tool and was used to exclude participants at T1. The measure has a recommended cut-off score of 15 [36], although it is argued this benchmark should not be rigid and can vary based upon individual characteristics and severity of symptoms [39]. Thus, as all participants had a clinical diagnosis of autism, participants were only excluded from T3 analysis if they scored below this cut off on more than two data collection points. Given the longitudinal nature of the study, the lifetime SCQ was used to collect data at T2 and T3, measuring individual change over time.

The Repetitive Behaviour Questionnaire (RBQ) was used to rate frequency of repetitive behaviour and severity on a Likert scale [40]. Repetitive behaviours are considered to be a risk marker for self-injury [41, 42]. It was therefore considered a relevant measure to include, exploring how such behaviours develop with age. The measure has been shown to have good reliability in a sample of individuals with heterogeneous causes of intellectual disability [35]. Concurrent, content and face validity has also been evidenced and shown to be robust [35].

The Challenging Behaviour Questionnaire (CBQ) evaluated self-injury, aggression, destruction of property and stereotyped behaviour within the past month [43]. The questionnaire allows for topographies and severity of SIB to be described. Analysis of psychometric properties has found good inter-rater reliability [43].

In addition to measures assessed at T1 [8], The Self-Restraint Questionnaire was included at T3 [30]. Self-restraint behaviours are described to serve the purpose of inhibiting severe SIB [23]. The measure describes seven topographies of self-restraint, with a checklist to indicate any behaviour present. The measure has been shown to be reliable with fair inter-rater agreement across all items, and good reliability on three of the subscales [35]. Validity has also been evidenced through a series of direct observations (89.6% across observation and scores) [35].

Data analysis

Normality of data was assessed using Kolmogorov-Smirnov tests. Due to the dataset significantly deviating from normal distributions (p < .05), non-parametric analyses were employed. Mann-Whitney U tests were conducted to assess demographic differences between those who took part in T3 study and those who declined to take part, to evaluate how representative the T3 sample was of the original T1 sample. Chi-Square and relative risk statistics were conducted to assess service use between those presenting with self-injury at T3, and those who did not. Chi-Square and Mann-Whitney U analyses were also used to explore demographic and behavioural differences between those who showed SIB at T3 and those who did not. McNemar analyses were conducted to assess persistence and topographies of self-injury from T2 to T3 and T1 to T3. Kruskal-Wallis analyses were used to evaluate putative risk markers between T2 and T3, whereby participants were split into absent (self-injury absent at both T2 and T3), transient (self-injury absent at either T2 or T3) and persistent (self-injury present at both T2 and T3) groups. This analysis was also repeated for data collected at T1 to T3, data was again split into three groups: absent (self-injury absent at both T1 and T3), transient (self-injury absent at either T1 or T3) and persistent (self-injury present at both T1 and T3) groups. Pairwise post hoc Mann-Whitney U analyses corrected for multiple comparisons were used to assess significant differences between the categorical groups. Kruskal-Wallis analyses were also used to explore putative risk markers associated with self-restraint at T3. In order to summarise data collected at each of the three time points and clearly depict behavioural characteristics that cross-sectionally and longitudinally predicted SIB, standardised effect sizes were calculated. Data from T1 [8] and T3 [24] were reassessed, and Z scores were extracted, with standardised effect sizes then calculated. Finally, to explore the predictive value of items assessed at T1, least absolute shrinkage and selection operator (LASSO) analysis was conducted, with the outcome variables being the presence of self-injurious or self-restraint behaviour at T3. LASSO analysis was chosen as evaluation of variance inflation factors indicated high levels of multicollinearity within the predictor variables, violating assumptions of traditional regression analysis [44]. As LASSO analysis is a penalised form of regression, poorer parameters are reduced where there is multicollinearity, minimising over-prediction in smaller samples [45]. Analysis utilised R software for statistical computing (version 3.5), operating the ‘glmnet’ package [46].

Results

Demographic characteristics of the sample

In order to ensure those who participated at T3 were representative of the original T1 sample, comparisons were made between those who took part at T3 and those who declined on a range of demographic and behavioural characteristics from T1. The data in Table 1 reveal that those who took part at T3 did not significantly differ from those who declined to take part on any of the demographic measures collected at T1. However, differences were obtained for some measures of behavioural characteristics. Individuals who took part at T3 showed significantly lower levels of activity, impulsivity, compulsive behaviour and restricted preferences. They also displayed higher levels of repetitive behaviour. The final sample was not significantly different regarding levels of self-injury and was therefore deemed representative of the T1 sample for the purposes of the present study. Demographic characteristics of parents and caregivers that participated at T3 are also presented (Table 2) detailing self-reported levels of education and household income.

Table 1 Demographic and behavioural characteristics of those who participated at T3 and those who declined to take part T3
Full size table
Table 2 Educational and financial characteristics of parents and caregivers of those who participated at T3 (67)
Full size table

Persistence of self-injury

In order to assess hypothesis 1 and 2, the persistence and stability of SIB was explored. Groups were first created based upon the presence of self-injury at T2 and T3: Absent, Remission, Incidence and Persistent. McNemar analysis was employed to explore significant differences between groups. Percentages of participants showing self-injury and individual topographies of self-injury were calculated for each of these groups (Table 3). Analysis showed no significant change in self-injury between these time points.

Table 3 Percentage (N) of participants showing remission, incidence, persistent or absent self-injurious behaviour between T2 and T3
Full size table

This analysis was repeated for data collected at T1–T3 (Table 4). The data in Table 3 show significant reductions in the presence of self-injury (p = .031), and the specific topography of self-biting (p = .039) from T1 to T3. Self-injury remitted in 56% of individuals displaying SIB at T1 but was persistent in 44% of individuals over 10 years. There were no other significant differences within individual topographies of self-injury.

Table 4 Percentage (Na) of participants showing remission, incidence, persistent or absent self-injurious behaviour between T1 and T3
Full size table

In order to explore any mediating effect of service use upon persistence of SIB at T3, Chi-squared analysis with accompanying relative risks were calculated (see Table 5). Results show that there were significant differences between the four groups (persistent, absent, remitted and incident SIB) regarding access to paediatricians (χ2 (2) = 12.765, p = .002). Post hoc analysis showed both the persistent and transient group accessed paediatricians more than the absent group (p < .001). There were no other significant differences regarding service providers, and relative risk analysis comparing absent and persistent group revealed no significant differences.

Table 5 Number and percentage of individuals with autism spectrum disorder accessing services and Chi-squared analysis
Full size table

In summary, the analyses support the null hypothesis as results show a significant reduction in self-injury within the sample over the longitudinal period.

Demographic and behavioural variables associated with the presence of self-injury and self-restraint

In order to assess hypothesis 4, analysis explored T3 demographic and behavioural variables associated with the presence of self-injury and self-restraint at T3. This analysis allows for insight into the presence of risk marker associated with behaviours cross-sectionally. Participants were grouped based upon the presence or absence of self-injury or self-restraint behaviours at T3. Chi-square, Fisher’s exact test’s and Mann-Whitney U analyses were conducted to compare scores between those with present versus absent self-injury at T3 (Table 6) and self-restraint at T3 (Table 7).

Table 6 T3 demographic and behavioural characteristics for participants with and without self-injury at T3
Full size table
Table 7 T3 Demographic and behavioural characteristics for participants with and without self-restraint at T3
Full size table

Results in Table 6 show no significant differences between the presence of self-injury at T3 and demographic measures collected at T3. Total activity scores (U = 187, Z = − 3.259, p < .001, rFootnote 2= 0.4) and subscales of overactivity (U = 176, Z = − 3.418, p < .001, r = 0.4) and impulsivity (U = 187, Z = − 3.264, p < .001, r = 0.4) were significantly higher for the self-injury group. Significant differences were also found on total repetitive behaviour scores (U = 228, Z = − 2.657, p = .008, r = 0.3), and insistence on sameness subscale (U = 224, Z = − 2.734, p = .006, r = 0.3), with the self-injury group scoring higher. Total autism characteristics group mean scores (U = 244.4, Z = − 2.422, p = .015, r = 0.3) were also significantly higher for the group showing self-injury at T3.

Results in Table 7 show no significant differences between those showing self-restraint at T3 and demographic measures collected at T3. Mood total score (U = 364, Z = − 2.371, p = .018, r = 0.3) and subscales of mood (U = 348, Z = − 2.591, p = .010, r = 0.3) and interest and pleasure (U = 384, Z = − 2.120, p = .034, r = 0.3) were significantly lower amongst those that showed self-restraint. Total activity scores (U = 282, Z = − 3.412, p < .001, r = 0.4) and subscales of overactivity (U = 293, Z = − 3.278, p < .001, r = 0.4) and impulsivity (U = 304, Z = − 3.139, p = .002, r = 0.4) were significantly higher within the self-restraint group. Repetitive behaviour total scores (U = 389, Z = − 2.052, p = .040, r = 0.3) and insistence on sameness (U = 396, Z = − 1.988, p = .047, r = 0.2), were also significantly higher amongst those who showed self-restraint. Finally, autism characteristics total score (U = 359, Z = − 2.439, p = .015, r = 0.3) was also significantly higher for individuals displaying self-restraint.

In summary, analyses support hypothesis 4, with behavioural measures of overactivity and impulsivity, alongside other behavioural characteristics, being significantly associated with both the presence of self-injury and self-restraint at T3.

Longitudinal risk markers for the presence of self-injury and self-restraint behaviours

In order to assess hypothesis 3 and evaluate putative risk markers in those with self-injury compared to those without, participants were categorised into three groups: absent (self-injury absent at both T2 and T3; N = 11), transient (self-injury absent at either T2 or T3; N = 6), and persistent (self-injury present at both T2 and T3; N = 29). T2 behavioural characteristics were assessed across the three groups (for brevity, these data are presented in the appropriate column in Table 8). Kruskal-Wallis analyses identified significant differences between groups on measures of impulsivity (χ2 (2) = 9.705, p = .008) and overactivity (χ2 (2) = 9.764, p = .005). Differences were also found for insistence on sameness (χ2 (2) = 6.994, p = .030), restricted repetitive and stereotyped behaviours (χ2 (2) = 7.102, p = .0.29) and reciprocal social interaction (χ2 (2) = 7.185, p = .028). Pairwise post hoc analysis corrected for multiple comparisons revealed significant differences between scores in the absent and persistent self-injury groups for all behavioural variables.

Table 8 Effect sizes for cross-sectional and longitudinal behavioural risk markers of self-injury over ten years
Full size table

This analysis was repeated in order to assess hypothesis 4 and evaluate putative risk markers between those with self-injury and those without. Participants were categorised into three groups: absent (self-injury absent at both T1 and T3; N = 37 mean [SD] age in years = 13 [10], % male = 81), transient (self-injury absent at either T1 or T3; N = 18 mean [SD] age in years = 11 [6], % male = 83) and persistent (self-injury present at both T1 and T3; N = 11 mean [SD] age in years = 10 [6], % male = 73)Footnote 3. T1 behavioural characteristics were assessed across the three groups (see Fig. 1 for the median, maximum and minimum scores and significant differences between groups). Kruskal-Wallis analyses identified significant differences between groups on measures of overactivity (χ2 (2) = 16.067, p < .001) and impulsivity (χ2 (2) = 20.418, p < .001). Pairwise post hoc analysis corrected for multiple comparisons revealed significant differences between scores in the absent and persistent self-injury groups, with the persistent group scoring significantly higher on measures of overactivity (U = 76, p = .002, r = 0.5) and impulsivity (U = 45.5, p < .001, r = 0.6).

Fig. 1
figure 1

T1 MIPQ, RBQ, TAQ, and SCQ total and subscale scores for absent, transient and persistent groups

Full size image

In order to evaluate putative risk markers associated with the presence of self-restraint behaviours at T3, measures of T1 behavioural characteristics were assessed (see Fig. 2 for the median, maximum and minimum scores and significant differences between groups). Mann-Whitney U analyses identified significant differences between groups on measures of compulsive behaviours (U = 368, Z = − 1.993, p = .046, r = 0.2) overactivity (U = 363, Z = − 2.387, p = .017, r = 0.3) and impulsivity (U = 333, Z = − 2.762, p = .006, r = 0.3), with those showing self-restraint behaviours at T3 scoring higher on the T1 measures. No other significant differences were found on other any other measures.

Fig. 2
figure 2

T1 MIPQ, RBQ, TAQ, and SCQ total and subscale scores for individuals with and without self-restraint behaviours at T3

Full size image

In summary, analysis suggests a profile of behavioural characteristics that are associated with the presence of self-injury. Impulsivity and overactivity scores obtained at T1 significantly predict the presence of self-injury and self-restraint behaviours 10 years later, at T3, supporting hypothesis 4.

Cross-sectional and longitudinal analysis summary

In order to compare both cross-sectional and longitudinal risk markers for self-injury over the 10-year data set, summary analyses are presented in Table 8. Table 8 presents effect sizes (R interpreted with Cohens D) of significant differences between present and absent SIB groups (cross-sectionally) and persistent and absent SIB groups (longitudinally). Data are drawn from the previously published studies [8, 24] and the analyses conducted in the present study. Analysis was also conducted for the demographic variables presented at each of the three time points; however, as none of these significantly predicted differences longitudinally they were not included in the final table. The results in Table 8 demonstrate that impulsivity and overactivity are the only behavioural variables that predict self-injury both cross-sectionally and longitudinally.

Predictive model of risk markers for longitudinally predicting the presence of self-injury and self-restraint behaviours

Finally, in order to further assess hypothesis 4 and evaluate the utility of scores obtained at T1 to predict self-injury severity and self-restraint at T3, least absolute shrinkage and selection operator (LASSO) analysis was employed. Behavioural variables collected at T1 were entered into the LASSO analysis, to control for potential multicollinearity. Outcome variables were set as T3 self-injury and T3 self-restraint in turn. As T3 self-injury severity scores were not normally distributed, responses were converted into factor variables (two levels: self-injury, no self-injury). Figures 3 and 4 present variables responding to the weight of the penalty increasing for each model. Cross-validation utilising the binomial deviance as a function of log lambda was then applied (Figs. 5 and 6). Shrinkage penalty parameters for Lambda (λ) were determined through tenfold cross validation [49]. All variables with zero coefficients were removed from each of the final models.

Fig. 3
figure 3

Solution path plotting Self-Injury variable coefficients against the L1 norm

Full size image
Fig. 4
figure 4

Solution path plotting self-restraint variable coefficients against the L1 norm

Full size image
Fig. 5
figure 5

Cross-validation plot for Self-Injury predictors, estimating optimal Lambda minimum and maximum estimates using the deviance metric

Full size image
Fig. 6
figure 6

Cross-validation plot for self-restraint predictors, estimating optimal Lambda minimum and maximum estimates using the deviance metric

Full size image

Impulsivity, interest and pleasure, stereotyped behaviour and ability at T1, as well as the presence of self-restraint at T3, were included in the final model predicting the presence of self-injury at T3. Overactivity scores from T1 were the only remaining variable predicting self-restraint at T3. Levels of predictive error as shown within cross-validation plots suggest models should be interpreted with caution, although variables remaining in the final self-injury model are supported by prior analyses.

In summary, the analysis presents two models of prediction for the presence of self-injury and self-restraint. Remaining variables presented in the model predicting self-injury support hypotheses 4 (a). The model of prediction for self-restraint fails to support hypothesis 4 (b), as variables hypothesised were converted into zero coefficients and not found to be predictive of self-restraint behaviours.

Discussion

This study details a unique longitudinal investigation into the persistence of self-injury and self-restraint within a sample of individuals with autism over a 10-year period. The use of robust measures at each point of data collection strengthens the validity and reliability of findings. Stringent exclusion criteria and evaluation of demographic variability between those who participated and those who did not ensures that the current sample is representative of the wider non-clinical sample, further contributing to the internal validity of conclusions. The use of novel, conservative longitudinal data analysis approaches allows for unique predictive models to be proposed. Results present a robust argument for the presence of behavioural risk markers such as impulsivity and overactivity successfully predicting self-injury in autism over a 10-year period.

The results show that self-injury was persistent in 44% of individuals over 10 years, with rates of self-injury significantly decreasing from T1. Findings support cross-sectional and longitudinal literature presenting an age-related decline in the persistence of self-injury [18, 19]. Significant reductions in self-injury suggest a divergent trajectory in autism compared to those with ID, where higher persistence rates are reported, 84% over an 18-year period [18]. Current findings must also be viewed independently of research involving clinical populations, where self-injury may also be driven by elevated levels of co-morbid mood, anxiety and behavioural disorders [50]. Age-related decline in behavioural measures of autism symptomology, stereotyped behaviours and repetitive behaviours are reported for individuals with autism post adolescence [16, 20, 51]. Current findings may therefore represent a global age-related decrease in clinical behaviours for some individuals with autism. Whilst overall persistence of SIB decreased over time, it is also important to note that SIB was persistent for a significant minority (44%) of individuals with autism. Self-injury beyond the age of 20 is suggested to be a chronic behaviour requiring professional intervention [18]. Thus, these data provide support for arguments advocating early intervention to prevent the behaviour from occurring and subsequently persisting over time.

Results show significant differences between absent and persistent groups regarding access to paediatricians, with no other significant differences regarding access to other professionals. Findings are consistent with literature highlighting a considerable paucity of service use amongst individuals with intellectual disabilities [7, 52]. Individuals who engage in self-injury are considered to present a greater need for professional input to reduce such behaviours [53], however current findings suggest this need is not met, despite the persistent presence of clinically significant SIB for 10 years. It could be argued for those without self-injury, services offer a protective role in preventing the development of the behaviour. Participants were initially recruited through a parent support group, with those participating in the current study potentially representing a subsample more willing or able to interact with professionals, inflating service use data estimates. Nevertheless, even with the consideration of inflation of data within those who do not present with self-injury, the potential un-met needs for individuals with self-injury is concerning. The lack of reported access to professional services to address self-injury is proposed be a key factor in its subsequent persistence [7]; it is therefore imperative future research and policy providers investigate this issue further to encourage proactive and persistent interventions from professionals for those with self-injury.

Cross-sectional analysis of T3 characteristics associated with self-injury and self-restraint revealed significant differences in the behavioural profile for individuals presenting with these behaviours. Higher scores on measures of overactivity, impulsivity and repetitive behaviours were associated with both self-injury and self-restraint, consistent with data in other studies [23, 54]. These results support the hypothesis that impaired behavioural inhibition may drive SIB in those with autism [55]. Autism phenomenology scores were also significantly higher in individuals presenting with self-injury at T3, supporting research associating severity of autism symptomology with severe SIB [19]. The use of a standardised screening tool to score autism symptomology allows robust conclusions to be drawn and supports the clinical implications of conclusions. Findings enhance understanding of the behavioural profile associated with individuals presenting with self-injury, but also how this is differentiated for individuals without the behaviour.

Individuals who presented with self-restraint behaviour at T3 also showed significantly lower mood, interest and pleasure scores and significantly higher impulsivity scores. Self-restraint behaviours are described to serve the purpose of inhibiting severe SIB [29, 31]. Results present an emerging behavioural profile of individuals who show self-restraint. Individuals appear to be more impulsive and experience more frequent and severe self-injury. It is well-documented that painful health conditions are more common in individuals with autism, elevated for those presenting with self-injury [56]. It could be argued that lower mood occurs as a result of pain associated with the complex behavioural profile for individuals presenting with self-restraint [57]. The identification of self-restraint behaviours within the current study was limited to behavioural presence, with no record of duration or severity for individual topographies and how this may relate to mood. However, literature supporting the association of pain with elements of the presented behavioural profile suggests lower mood linked to pain is a plausible explanation [58].

Investigation of T1 behavioural markers associated with the presence of self-injury and self-restraint at T3 revealed that overactive and impulsive behaviours continue to predict self-injury and self-restraint longitudinally, as found at T2 analysis [24]. The identification of stable and reliable behavioural markers of SIB considerably enhances current understanding of mechanisms underpinning the persistence of self-injury and its age-related developmental trajectory. Furthermore, results highlight the potential positive clinical impact of identifying individuals at greater risk of developing severe self-injury. Utilising behavioural characteristics that have been identified to reliably longitudinally predict the presence of negative behaviours would allow clinical services to orient to preventative rather than solely reactive interventions [23]. The use of validated behavioural assessments at each of the time point in the present study significantly enhances the internal validity of conclusions made. Future research should attempt to corroborate findings through the employment of behavioural focussed intervention strategies, whereby intervention techniques are tailored to individual risk to ensure maximum value for both individuals and service providers.

Results present two explorative models for demographic and behavioural variables that longitudinally predict the presence of self-injury and self-restraint behaviours in turn. T1 behavioural measures that remain in the final model as having predictive value for the presence of behaviours provide support for arguments of individual characteristics influencing the developmental trajectory of self-injury and self-restraint [22]. These analyses show that impulsivity, interest and pleasure, stereotyped behaviour, social communication and adaptive functioning predict the persistence of SIB over 10 years. The novel use of regularisation techniques (LASSO analysis) represents an emerging shift within the behavioural sciences towards adopting methods of machine learning. Such analysis has the capability of producing more robust and accurate predictions when compared to traditional techniques that often overfit data and lead to inflations of error [59]. It must be noted predictive error in current models is potentially inflated by smaller sample sizes and incomplete data sets. Yet the ability of such models to identify individuals at risk of developing severe negative behaviours is not limited as these approaches are more robust than traditional regression techniques. There is benefit to be gained through the use of such novel techniques within the behavioural sciences field, expanding capabilities of analysis.

In summary, findings reveal self-injury was persistent for 44% of individuals that presented with the behaviour 10 years ago, with a robust and stable profile of behavioural characteristics associated with self-injury and self-restraint presented.

Limitations

Small sample size may limit the population parameters drawn from statistical analysis in the current study. This may be amplified by the high attrition rates from initial T1 data collection and result in inflation of scores of measures such as service access. However, recent arguments suggest there is utility in smaller samples, offering the ability to investigate theoretical relationships at the individual participant level [60]. It must be taken into consideration when comparing the current sample with similar research that it is currently the largest longitudinal dataset utilising a non-clinical sample to explore SIB in individuals with autism and thus has significant value within its size. The current sample’s mean age from T1 to T3 stretches across early childhood to adulthood, offering significant value in its findings. Whilst future longitudinal investigations should attempt to potentially re-engage with individuals that declined the invitation to take part, the smaller sample within the current study has considerable clinical and scientific value.

Secondly, the choice of authors to utilise traditional significance statistic (p < .05) could be considered a limitation of analysis. However, as the nature of the research is largely exploratory, the use of a more modest estimate of significance alongside considerations of effect size was deemed sufficient in data interpretations. Where multiple comparisons have been made, stringent corrections have been put in place through the use of Bonferroni [61].

Another limitation considered by the research team is the bias seen within the socioeconomic descriptives of the sample remaining at the present time point. It is not uncommon within autism research for samples to be disproportionately representative of individuals that are highly educated and of higher socioeconomic status; however, it is something to be considered when interpreting findings proposed within the current paper.

Finally, the age suitability of measures used within the current study must also be considered. Although the SCQ is an appropriate screening measure for autism and for individuals with intellectual disabilities, questions were not adapted within the current investigation to represent the ageing sample. Literature suggests the potential benefits of modifying questions and subsequent cut-off scores to reflect samples [62]. Future research should therefore attempt to adapt questions to ensure accuracy of responses whilst maintaining the specificity of the measure.

Conclusions

A robust and stable profile of behavioural characteristics associated with self-injury and self-restraint is presented, with their role as putative risk markers further reinforced. The ability of measurable behaviours such as overactivity and impulsivity to successfully predict individuals at greater risk of poorer outcomes over a 10-year period has significant implications for clinical interventions. Explorative models further emphasise the predictive power these behaviours have, identifying their role as mechanisms that underpin negative behaviours. Early intervention attempts should therefore target individuals considered to be at greater risk of developing severe negative behaviours and prevent them from entering into individual’s behavioural repertoire.

Availability of data and materials

The datasets generated and/or analysed during the current study are not publicly available. Due to the sensitive nature of the research and ethical concerns surrounding the publication of sensitive personal data, no participants were asked for consent to their data being shared.

Notes

  1. Relevant professionals were considered to be any of the following: paediatrician, clinical psychologist, clinical geneticist, GP, psychiatrist, educational psychologist or significant other.

  2. Effect size r was employed as an alternative to standard difference statistics when assumptions are violated [47, 48]. Effect sizes presented are interpreted with Cohen’s d

  3. Total sample 66 due to missing data of one participant

Abbreviations

CBQ:

Challenging Behaviour Questionnaire

LASSO:

Least absolute shrinkage and selection operator

MIPQ:

Mood, Interest and Pleasure Questionnaire

RBQ:

Repetitive Behaviour Questionnaire

SCQ:

Social Communication Questionnaire

SIB:

Self-Injurious Behaviour

TAQ:

The Activity Questionnaire

References

  1. Tate BG, Baroff GS. Aversive control of self-injurious behavior in a psychotic boy. Behav Res Ther. 1966;4(4):281–7.

    Article  CAS  PubMed  Google Scholar 

  2. Matson JL, Lo Vullo SV. A review of behavioral treatments for self-injurious behaviors of persons with autism spectrum disorders. Behavior Modification. 2008;32(1):61–76.

    Article  PubMed  Google Scholar 

  3. Pritchard A, Roy A. Reversing the export of people with learning disabilities and complex health needs. Br J Learn Disabil. 2006;34(2):88–93.

    Article  Google Scholar 

  4. Heyvaert M, Saenen L, Maes B, Onghena P. Systematic review of restraint interventions for challenging behaviour among persons with intellectual disabilities: focus on experiences. Journal of Applied Research in Intellectual Disabilities. 2015;28(2):61–80.

    Article  PubMed  Google Scholar 

  5. Tsiouris JA, Kim SY, Brown WT, Pettinger J, Cohen IL. Prevalence of psychotropic drug use in adults with intellectual disability: positive and negative findings from a large scale study. J Autism Dev Disord. 2013;43(3):719–31.

    Article  Google Scholar 

  6. Barrett B, Mosweu I, CRG J, Charman T, Baird G, Simonoff E, et al. Comparing service use and costs among adolescents with autism spectrum disorders, special needs and typical development. Autism. 2015;19(5):562–9.

    Article  PubMed  Google Scholar 

  7. Awan N, Oliver C, Wilde L, Eden K, Waite J. Service receipt of adults with rare genetic syndromes who engage in challenging behaviour. J Intellect Dev Disabil. 2018;43(3):308–16.

    Article  Google Scholar 

  8. Richards C, Oliver C, Nelson L, Moss J. Self-injurious behaviour in individuals with autism spectrum disorder and intellectual disability. J Intellect Disabil Res. 2012;56(5):476–89.

    Article  CAS  PubMed  Google Scholar 

  9. Cooper SA, Smiley E, Allan LM, Jackson A, Finlayson J, Mantry D, et al. Adults with intellectual disabilities: prevalence, incidence and remission of self-injurious behaviour, and related factors. J Intellect Disabil Res. 2009;53(3):200–16.

    Article  PubMed  Google Scholar 

  10. Oliver C, Murphy GH, Corbett JA. Self-injurious behaviour in people with mental handicap: a total population study. J Intellect Disabil Res. 1987;31(2):147–62.

    Article  Google Scholar 

  11. Eden KE, De Vries PJ, Moss J, Richards C, Oliver C. Self-injury and aggression in tuberous sclerosis complex: cross syndrome comparison and associated risk markers. J Neurodev Disord. 2014;6(1):10.

  12. Dimian AF, Botteron KN, Dager SR, Elison JT, Estes AM, Pruett JR, et al. Potential risk factors for the development of self-injurious behavior among infants at risk for autism spectrum disorder. J Autism Dev Disord. 2017;47(5):1403–15.

    Article  PubMed  PubMed Central  Google Scholar 

  13. Totsika V, Hastings RP. Persistent challenging behaviour in people with an intellectual disability. Current Opinion in Psychiatry. 2009;22(5):437–41.

    Article  PubMed  Google Scholar 

  14. Nøttestad JA, Linaker OM. Self-injurious behaviour before and after deinstitutionalization. J Intellect Disabil Res. 2001;45(2):121–9.

    Article  CAS  PubMed  Google Scholar 

  15. Kieman C, Alborz A. Persistence and change in challenging and problem behaviours of young adults with intellectual disability living in the family home. J Appl Res Intellect Disabil. 1996;9(3):181–93.

    Article  Google Scholar 

  16. Shattuck PT, Seltzer MM, Greenberg JS, Orsmond GI, Bolt D, Kring S, et al. Change in autism symptoms and maladaptive behaviors in adolescents and adults with an autism spectrum disorder. J Autism Dev Disord. 2007;37(9):1735–47.

    Article  PubMed  PubMed Central  Google Scholar 

  17. Baghdadli A, Picot MC, Pry R, Michelon C, Burzstejn C, Lazartigues A, et al. What factors are related to a negative outcome of self-injurious behaviour during childhood in pervasive developmental disorders? J Appl Res Intellect Disabil. 2008;21(2):142–9.

    Article  Google Scholar 

  18. Taylor L, Oliver C, Murphy G. The chronicity of self-injurious behaviour: a long-term follow-up of a total population study. J Appl Res Intellect Disabil. 2011;24(2):105–17.

    Article  Google Scholar 

  19. Rattaz C, Michelon C, Baghdadli A. Symptom severity as a risk factor for self-injurious behaviours in adolescents with autism spectrum disorders. J Intellect Disabil Res. 2015;59(8):730–41.

    Article  CAS  PubMed  Google Scholar 

  20. Esbensen AJ, Seltzer MM, KSL L, Bodfish JW. Age-related differences in restricted repetitive behaviors in autism spectrum disorders. J Autism Dev Disord. 2009;39(1):57–66.

    Article  PubMed  PubMed Central  Google Scholar 

  21. Arron K, Oliver C, Moss J, Berg K, Burbidge C. The prevalence and phenomenology of self-injurious and aggressive behaviour in genetic syndromes. J Intellect Disabil Res. 2011;55(2):109–20.

    Article  PubMed  Google Scholar 

  22. Oliver C, Richards C. Practitioner review: self-injurious behaviour in children with developmental delay. J Child Psychol Psychiatry Allied Discip. 2015;56(10):1042–54.

    Article  PubMed  Google Scholar 

  23. Richards C, Davies L, Oliver C. Predictors of self-injurious behavior and self-restraint in autism spectrum disorder: towards a hypothesis of impaired behavioral control. J Autism Dev Disord. 2017;47(3):701–13.

    Article  PubMed  Google Scholar 

  24. Richards C, Moss J, Nelson L, Oliver C. Persistence of self-injurious behaviour in autism spectrum disorder over 3 years: a prospective cohort study of risk markers. J Neurodev Disord. 2016;8(1):21.

  25. Richman DM, Barnard-Brak L, Bosch A, Thompson S, Grubb L, Abby L. Predictors of self-injurious behaviour exhibited by individuals with autism spectrum disorder. J Intellect Disabil Res. 2013;57(5):429–39.

    Article  PubMed  Google Scholar 

  26. Sonuga-Barke EJS. Psychological heterogeneity in AD/HD - a dual pathway model of behaviour and cognition. In: Behavioural Brain Research. 2002;130(1-2):29–36.

    Article  PubMed  Google Scholar 

  27. Turner M. Towards an executive dysfunction account of repetitive behaviour in autism. In: Autism As an Executive Disorder; 1997.

    Google Scholar 

  28. Nigg JT. Neuropsychologic theory and findings in attention-deficit/hyperactivity disorder: the state of the field and salient challenges for the coming decade. Biological Psychiatry. 2005;57(11):1424–35.

    Article  PubMed  Google Scholar 

  29. King BH. Self-injury by people with mental retardation: a compulsive behavior hypothesis. Am J Ment Retard. 1993;98(1):93–112.

  30. Powell SB, Bodfish JW, Parker D, Crawford TW, Lewis MH. Self-restraint and self-injury: occurrence and motivational significance. Am J Ment Retard. 1996;101(1):41–8.

  31. Fisher WW, Iwata BA. On the function of self-restraint and its relationship to self-injury. J Appl Behav Anal. 1996;29(1):93–8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  32. Kushllck A, Blunden R, Cox G. A method of rating behaviour characteristics for use in large scale surveys of mental handicap. Psychol Med. 1973;3(4):466–78.

  33. Palmer J, Jenkins J. The ‘Wessex’behaviour rating system for mentally handicapped people: reliability study. Br J Ment Subnorm. 1982;28(55):88–96.

    Article  Google Scholar 

  34. Burbidge C, Oliver C, Moss J, Arron K, Berg K, Furniss F, et al. The association between repetitive behaviours, impulsivity and hyperactivity in people with intellectual disability. J Intellect Disabil Res. 2010;54(12):1078–92.

    Article  CAS  PubMed  Google Scholar 

  35. Oliver, C., Royston, R., Crawford, H., Moss, J., Waite, J., Arron, K., Burbidge, C., Ellis, K., Nelson, L., Ross, E., Russell, H. & Welham A. Informant assessments of behaviour and affect for people with intellectual disability [Internet]. 2019. Available from: https://secure.viewer.zmags.com/publication/9964d37c#/9964d37c/1

    Google Scholar 

  36. Berument SK, Rutter M, Lord C, Pickles A, Bailey A. Autism screening questionnaire: Diagnostic validity. Br J Psychiatry. 1999;175(5):444–51.

    Article  CAS  Google Scholar 

  37. Lord C, Risi S, Lambrecht L, Cook EH, Leventhal BL, Dilavore PC, et al. The Autism Diagnostic Observation Schedule-Generic: a standard measure of social and communication deficits associated with the spectrum of autism. J Autism Dev Disord. 2000;30(3):205–23.

  38. Lord C, Rutter M, Le Couteur A. Autism Diagnostic Interview-Revised: a revised ver. of diagnostic interview for caregivers of individuals with possible PDDs. J Autism Dev Disord. 1994;24(5):659–85.

    Article  CAS  PubMed  Google Scholar 

  39. Corsello C, Hus V, Pickles A, Risi S, Cook EH, Leventhal BL, et al. Between a ROC and a hard place: decision making and making decisions about using the SCQ. J Child Psychol Psychiatry Allied Discip. 2007;48(9):932–40.

    Article  PubMed  Google Scholar 

  40. Moss J, Oliver C, Arron K, Burbidge C, Berg K. The prevalence and phenomenology of repetitive behavior in genetic syndromes. [Article]. J Autism Dev Disord. 2009;39(4):572–88.

    Article  PubMed  Google Scholar 

  41. Oliver C, Petty J, Ruddick L, Bacarese-Hamilton M. The association between repetitive, self-injurious and aggressive behavior in children with severe intellectual disability. J Autism Dev Disord. 2012;42(6):910–9.

    Article  Google Scholar 

  42. Davies LE, Oliver C. Self-injury, aggression and destruction in children with severe intellectual disability: incidence, persistence and novel, predictive behavioural risk markers. Res Dev Disabil. 2016;49:291–301.

    Article  CAS  PubMed  Google Scholar 

  43. Hyman P, Oliver C, Hall S. Self-injurious behavior, self-restraint, and compulsive behaviors in Cornelia de Lange syndrome. Am J Ment Retard AJMR. 2002;107(2):146–54.

    Article  PubMed  Google Scholar 

  44. Thompson CG, Kim RS, Aloe AM, Becker BJ. Extracting the variance in flation factor and other multicollinearity diagnostics from typical regression results. Basic Appl Soc Psych. 2017;39(2):81–90.

    Article  Google Scholar 

  45. Dubocq M, Haddy N, Schwartz B, Rubino C, Dayet F, de Vathaire F, et al. Exploring the performance of methods to deal multicollinearity: simulation and real data in radiation epidemiology area. Int J Stat Med Res. 2018;7(2):33–44.

    Article  Google Scholar 

  46. Friedman J, Hastie T, Tibshirani R. Regularization paths for generalized linear models via coordinate descent. J Stat Softw. 2010;33(1):1.

  47. Rosenthal R. Parametric measures of effect size. In: The handbook of research synthesis; 1994;621:231–44.

  48. Fritz CO, Morris PE, Richler JJ. Effect size estimates: current use, calculations, and interpretation. J Exp Psychol Gen. 2012;141(1):2.

    Article  Google Scholar 

  49. Hastie T, Tibshirani R, Friedman J. Unsupervised learning. In: The elements of statistical learning: Springer; 2009. p. 485–585.

  50. Nixon MK, Cloutier PF, Aggarwal S. Affect regulation and addictive aspects of repetitive self-injury in hospitalized adolescents. J Am Acad Child Adolesc Psychiatry. 2002;41(11):1333–41.

    Article  Google Scholar 

  51. Murphy GH, Beadle-Brown J, Wing L, Gould J, Shah A, Holmes N. Chronicity of challenging behaviours in people with severe intellectual disabilities and/or autism: a total population sample. J Autism Dev Disord. 2005;35(4):405–18.

    Article  PubMed  Google Scholar 

  52. Ruddick L, Davies L, Bacarese-Hamilton M, Oliver C. Self-injurious, aggressive and destructive behaviour in children with severe intellectual disability: prevalence, service need and service receipt in the UK. Res Dev Disabil. 2015;45:307–15.

    Article  PubMed  Google Scholar 

  53. UK NCC for MH. Challenging behaviour and learning disabilities: prevention and interventions for people with learning disabilities whose behaviour challenges. 2015.

    Google Scholar 

  54. South M, Ozonoff S, McMahon WM. Repetitive behavior profiles in Asperger syndrome and high-functioning autism. J Autism Dev Disord. 2005;35(2):145–58.

    Article  PubMed  Google Scholar 

  55. Schmitt LM, White SP, Cook EH, Sweeney JA, Mosconi MW. Cognitive mechanisms of inhibitory control deficits in autism spectrum disorder. J Child Psychol Psychiatry Allied Discip. 2018;59(5):586–95.

    Article  PubMed  Google Scholar 

  56. Christensen TJ, Ringdahl JE, Bosch JJ, Falcomata TS, Luke JR, Andelman MS. Constipation associated with self-injurious and aggressive behavior exhibited by a child diagnosed with autism. Educ Treat Child [Internet]. 2008;32(1):89–103. Available from: http://0-muse-jhu-edu.brum.beds.ac.uk/content/crossref/journals/education_and_treatment_of_children/v032/32.1.christensen.html.

    Article  Google Scholar 

  57. Davies LE, Oliver C. The purported association between depression, aggression, and self-injury in people with intellectual disability: a critical review of the literature. American Journal on Intellectual and Developmental Disabilities. 2014;119(5):452–71.

    Article  PubMed  Google Scholar 

  58. Courtemanche AB, Black WR, Reese RM. The relationship between pain, self-injury, and other problem behaviors in young children with autism and other developmental disabilities. Am J Intellect Dev Disabil. 2016;121(3):194–203.

    Article  PubMed  Google Scholar 

  59. Yarkoni T, Westfall J. Choosing prediction over explanation in psychology: lessons from machine learning. Perspect Psychol Sci. 2017;12(6):1100–22.

    Article  PubMed  PubMed Central  Google Scholar 

  60. Smith PL, Little DR. Small is beautiful: In defense of the small-N design. Psychon Bull Rev. 2018;25(6):2083–101.

    Article  Google Scholar 

  61. Abdi H. The Bonferonni and Šidák corrections for multiple comparisons. In: Encyclopedia of measurement and statistics; 2007.

    Google Scholar 

  62. Barnard-Brak L, Brewer A, Chesnut S, Richman D, Schaeffer AM. The sensitivity and specificity of the social communication questionnaire for autism spectrum with respect to age. Autism Res. 2016;9(8):838–45.

    Article  PubMed  Google Scholar 

Download references

Acknowledgements

We are extremely grateful to all the parents and carers that gave their time to contribute and support this research.

Funding

Research Autism and Cerebra provided partial funding for this project.

Author information

Authors and Affiliations

  1. School of Psychology, University of Birmingham, Birmingham, B15 2TT, UK

    Catherine Laverty & Caroline Richards

  2. Cerebra Centre for Neurodevelopmental Disorders, School of Psychology, University of Birmingham, Birmingham, B15 2TT, UK

    Catherine Laverty, Chris Oliver, Lisa Nelson & Caroline Richards

  3. School of Psychology, University of Surrey, Guildford, Surrey, GU2 7XH, UK

    Jo Moss

Authors
  1. Catherine Laverty
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Chris Oliver
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Jo Moss
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Lisa Nelson
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Caroline Richards
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

CL collected and analysed the data and drafted the manuscript. CO contributed to the design of the study and revised the manuscript. JM and LN were involved in the data collection. CR contributed to the design of the study, data collection and analysis and revised the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Catherine Laverty.

Ethics declarations

Ethics approval and consent to participate

Ethical approval for this study was obtained from the ethical review committee at Coventry University.

Consent for publication

N/A

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Laverty, C., Oliver, C., Moss, J. et al. Persistence and predictors of self-injurious behaviour in autism: a ten-year prospective cohort study. Molecular Autism 11, 8 (2020). https://0-doi-org.brum.beds.ac.uk/10.1186/s13229-019-0307-z

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://0-doi-org.brum.beds.ac.uk/10.1186/s13229-019-0307-z

Keywords

Molecular Autism

ISSN: 2040-2392