Research in Developmental Disabilities 36 (2015) 419–427

Contents lists available at ScienceDirect

Research in Developmental Disabilities

Stereotyped behaviors predicting self-injurious behavior in individuals with intellectual disabilities Lucy Barnard-Brak a,*, Johannes Rojahn b, David M. Richman a, Steven R. Chesnut a, Tianlan Wei c a b c

Texas Tech University, Department of Educational Psychology and Leadership, PO Box 41071, Lubbock, TX 79409, United States George Mason University, Department of Psychology, 4400 University Drive, Fairfax, VA 22030, United States Mississippi State University, Department of Educational Psychology, PO Box 6334, Starkville, MS 39761, United States

A R T I C L E I N F O

A B S T R A C T

Article history: Received 23 April 2014 Received in revised form 5 August 2014 Accepted 12 August 2014 Available online 11 November 2014

We examined the relation between stereotyped behavior and self-injurious behavior (SIB) for 1871 individuals with intellectual disabilities who had a score of >0 on the Behavior Problem Inventory (BPI-01; Rojahn et al., 2001). We report three main findings: First, structural equation modeling techniques (SEM) revealed that the BPI-01stereotyped behavior subscale scores predicted BPI-01 SIB subscale scores. Second, when stereotyped behavior was modeled as a predictor of SIB, mixture-modeling techniques revealed two groups of individuals: one in which stereotyped behavior was a strong, statistically significant predictor of SIB (69% of the sample), and another one in which stereotyped behavior was not a predictor of SIB (31%). Finally, two specific stereotyped behavior topographies (i.e., body rocking and yelling) were identified that significantly predicted five different SIB topographies (i.e., self-biting, head hitting, body hitting, self-pinching, and hair pulling). Results are discussed in terms of future research needed to identify biobehavioral variables correlated with cases of SIB that can, and cannot, be predicted by the presence of stereotyped behavior. ß 2014 Elsevier Ltd. All rights reserved.

Keywords: Self-injurious behavior Stereotyped behavior Proto-SIB Intellectual disabilities Developmental disabilities Neurodevelopmental disorders Behavior Problems Inventory

Self-injurious behavior (SIB) refers to self-directed acts that either cause immediate tissue damage or they have the potential1 to do so if left untreated (Rojahn, Schroeder, & Hoch, 2008). The most common SIB topographies among individuals with intellectual disabilities (ID) are head banging, body hitting, self-biting, and self-scratching, but depending on the definition of SIB they also include less prevalent behaviors such as, hair pulling, gouging and air swallowing (aerophagia) (Rojahn et al., 2008). Individuals with ID and autism are especially vulnerable to develop SIB at some point in their life showing higher SIB prevalence rates than their typically developing counterparts (e.g., Baghdadli, Pascal, Grisi, & Aussiloux, 2003; Bodfish, Symons, Parker, & Lewis, 2000; Kozlowski & Matson, 2012; McClintock, Hall, & Oliver, 2003; Oliver, Murphy, & Corbett, 1987). Point-prevalence studies indicate that SIB occurs in approximately 10% to 12% of individuals with ID in general (Didden et al., 2012). Rojahn et al. (2008) estimated that the prevalence of SIB in individuals with mild ID may be 4%, in those with moderate ID 7%, in severe ID 16%, and 25% in individuals with profound ID. SIB prevalence tends to be higher in childhood and adolescence as compared to older age groups (Rojahn et al., 2012a). However, the SIB prevalence in a

* Corresponding author. E-mail address: [email protected] (L. Barnard-Brak). 1 This definition of SIB refers to firmly established, often chronic behavior (as opposed to emerging proto-injurious behavior) that would cause tissue damage without ongoing treatment. http://dx.doi.org/10.1016/j.ridd.2014.08.017 0891-4222/ß 2014 Elsevier Ltd. All rights reserved.

420

L. Barnard-Brak et al. / Research in Developmental Disabilities 36 (2015) 419–427

general ID population is difficult to determine because it depends so heavily on factors such as level of ID, age, sex, and etiology of ID and others (and complex interactions among them). Non-injurious stereotyped behaviors, which can be described as atypical, idiosyncratic, and repetitive behaviors such as body rocking, hand flapping, spinning, and hand clapping are also very common in the ID population (Berkson, Tupa, & Sherman, 2001; Bodfish et al., 1995; Murphy, Hall, Oliver, & Kissi-Debra, 1999; Rojahn, Matlock, & Tasse´, 2000). Stereotyped behavior is quite common in individuals with ID. Didden et al. (2012) estimated that perhaps as much as 50% of the population may demonstrate one or more stereotypic response topographies. Many studies have shown a high probability of a concurrence of SIB and stereotyped behavior. For instance, Rojahn et al. (2012a) reported a Spearman correlation between the SIB and stereotyped behavior frequency ratings of 0 on the Behavior Problem Inventory (BPI-01; Rojahn, Matson, Lott, Esbensen, & Smalls, 2001) in 1122 individuals; 744 had a positive SIB subscale score (i.e., a score of 1 or larger), and 955 had a positive Stereotyped Behavior subscale score. Given the high prevalence of both behaviors in that population, a correlation between SIB and stereotyped behavior may be simply a function of chance and does not necessarily imply a functional relationship. Guess and Carr (1991) provided persuasive conceptual and empirical evidence in support of their proposition that some cases of stereotyped behavior and SIB first emerge as developmentally delayed, repetitive motor behaviors that initially serve a homeostatic function in under- or overstimulating environments, and later may become sensitive to socially mediated contingencies such as access to caregiver attention. However, speculations that some cases of SIB may be causally (or functionally) related to stereotyped behavior have a long history in the empirical literature on chronic SIB (e.g., Baumeister & Forehand, 1973; Baumeister & Rollings, 1976). In one of the first studies on the early development of SIB, Murphy et al. (1999) asked teachers from several special education classes for children with severe ID or autism below the age of 11 years to identify those who were ‘‘beginning to show early SIB’’ (p. 149) according to a list of nine specific SIB topographies (which, interestingly, were taken from an earlier but quite similar version of the BPI-01 (Rojahn et al., 2001) that is the basis of the current study). Seventeen children comprised the SIB index group; the control group consisted of 14 children without any signs of emerging SIB and matched on age and ambulation. While the index children showed indeed more potential SIB, there was also a significant overlap between the groups with regard to percent duration of potential SIB, suggesting the difficulties inherent in reliably identifying initial or early onset of SIB. After that, behavior observations of the index group were repeated for 18 months in 3month intervals. Clear SIB escalations were found for some of those children, but not for all. Using regression analyses (except for earlier teacher concerns about behavior problems) none of several other independent variables, including early stereotyped behavior, were found to predict SIB. In a subsequent study Hall, Oliver, and Murphy (2001) focused on four of those children that had exhibited escalating SIB over the 18 months observation period. They found a significant association between early SIB and low levels of social contact from teachers. The authors hypothesized that this association may be a risk marker for the exacerbations of SIB. The term proto-injurious behavior, introduced by Berkson, Tupa, and Sherman (2001), refers to certain types of stereotypic behaviors or non-repetitive, discrete bouts of behavior that may lead to topographically similar forms of SIB. Proto-injurious behavior is viewed as a potential risk marker for emerging SIB (Furniss & Biswas, 2012; Langthorne & McGill, 2008; Petty, Allen, & Oliver, 2009; Richman & Lindauer, 2005; Symons, Sperry, Dropik, & Bodfish, 2005) under the assumption implied in the term that the intensity or frequency of that proto-injurious behavior may be shaped such that it ultimately causes tissue damage. It also implies that some kind of topographic similarity between the proto-injurious behavior and the subsequent SIB should exist. In addition to the Hall et al. (2001) study just mentioned several other studies have provided evidence of the proto-injurious behavior hypothesis for some topographies of SIB. Berkson et al. (2001) examined body rocking and SIB in 39 young children between the ages of 3 and 40 months with severe disabilities in a longitudinal study. Twenty-one of the participants exhibited proto-SIB and/or SIB, while 10 displayed proto-SIB alone. The authors concluded that some proto-SIBs seemed to lead to SIBs of similar topographies. Richman and Lindauer (2005) followed emerging patterns of SIB in very young children with moderate-to-profound developmental delays. They confirmed that the majority of the proto-SIB remained topographically unchanged as they morphed into SIB for some of the children that developed SIB during the course of the study. Similarly, Petty et al. (2009) studied the temporal relationship between stereotypic, proto-injurious, and self-injurious behaviors in six children with intellectual disabilities and also found that stereotyped behavior was a precursor to proto-SIB and that proto-SIB in turn was a precursor to SIB in all of the children assessed. However, not all studies provided empirical evidence in support of the proto-injurious stereotyped behavior hypothesis, at least if proposed as an exclusive description of how SIB may develop. Kurtz, Chin, Huete, and Cataldo (2012), for instance reported that most of the young children in their sample of 32 either showed stereotypic behaviors and SIB simultaneously or they exhibited SIB only. Therefore, they concluded that SIB in their sample of children did not emerge from stereotypic behavior. A certain limitation of this study was, however, that the data for documenting the emergence of stereotyped behavior and SIB were based exclusively on parental report. In other words, it is unclear at this point whether and under what conditions certain stereotyped behavior may be precursors of certain proto-injurious behaviors and whether and under what circumstance proto-injurious behavior can be a precursor to SIB. In the current study, we examined whether the construct of stereotyped behavior (represented by a subscale score across many different stereotyped behavior topographies) can predict the construct of SIB (represented by a subscale score across many different SIB topographies). Second, after a stereotyped behavior construct was indeed found to be predictive of SIB, we

L. Barnard-Brak et al. / Research in Developmental Disabilities 36 (2015) 419–427

421

examined whether we could detect subgroups of individuals with different patterns of stereotyped behavior-SIB prediction. Finally, we investigated whether specific stereotyped behavior topographies predicted any specific SIB topographies. 1. Method 1.1. Participants and data collection The data set consisted of a total 1871 cases. Data sets from five different regions of the world were collated (Within USA: Rojahn et al., 2010; Rojahn, Rowe, Kasdan, Moore, & van Ingen, 2011; Zaja, Moore, van Ingen, & Rojahn, 2011; Outside of USA: Hastings, Beck, & Hill, 2005; Mircea, Rojahn, & Esbensen, 2010; Owen et al., 2004). Participants were selected to be part of the sample studied if they had a total score greater than 0 on the Behavior Problem Inventory (Rojahn et al., 2001). Data were cross-sectional though collected in a period from 2004 to 2010. For the respondents, approximately 45% (n = 847) completed the instrument in English while 55% (n = 1024) completed the instrument in a language other than English (e.g., Spanish, Dutch, Hungarian, and Romanian). For the sample, approximately 58% (n = 1086) were male while approximately 42% (n = 697) were female. The ages ranged from the youngest of 2.1 years to the oldest of 93 years with a mean of 28.99 years (SD = 20.82). The participants’ levels of intellectual disability (ID) ranged from average to profound ID. Approximately 2.5% (n = 37) of the sample were reported as having an average level of intelligence (IQ scores ranging from 85 to 114). While 4.2% (n = 63) were reported as having borderline intelligence with IQ scores ranging from 70 to 84. Approximately 9.5% (n = 141) were reported as having a mild level of ID, 13.2% (n = 196) were reported as having a moderate level of ID, 27.6% (n = 410) were reported as having a severe level of ID, and 30.8% (n = 458) were reported as having a profound level of ID. Approximately 72% (n = 1342) were located in residential facilities while 28% (n = 529) were located in the community. While utilizing data from single informants not any one type of informant information was utilized, the sample consists of diversity of demographic characteristics of both sample and respondents. 2. Measures The long form of the Behavior Problem Inventory (BPI-01; Rojahn et al., 2001, 2012a, 2012b) is a 49-item measure consisting of three subscales: Self-injurious Behavior (k = 14), Stereotyped Behavior (k = 24), and Aggressive/Destructive Behavior (k = 11). For the current study, we focused on the Self-injurious Behavior and Stereotyped Behavior subscales. A score of 1 indicated the presence of SIB or stereotyped behavior. Zeros indicated no reported presence of SIB. Cronbach’s alpha (a) values indicated an acceptable internal consistency of scores for each subscale with an a = 0.74 for SIB and a = 0.92 for stereotyped behavior. As no measure can identify every possible problem behavior, each subscale contained an open response item of ‘Other’ that permitted the identification of problem behaviors not included in the BPI; however, these responses were not included in our analyses. Table 1 provides the descriptive statistics for stereotyped behavior and selfinjurious behavior according to gender and age. 3. Procedure For our first research question (Was stereotyped behavior a significant predictor of SIB?), structural equation modeling techniques were utilized to examine the relationship between stereotyped behavior as predicting SIB via MPlus (v. 7.0; Muthe´n & Muthe´n, 2012). The exogeneous variables of age and level of intellectual disability (ID) were statistically controlled in examining this relationship in view of the relationship of the rest of the endogenous variables. The latent

Table 1 Descriptive statistics by gender and age. Self-injurious behavior

Stereotyped behavior

M

SD

M

SD

By gender Male Female

3.71 2.86

5.07 4.47

12.10 9.33

15.13 14.08

By age 0–10 10.1–15 15.1–20 20.1–30 30.1–35 35.1–40 40.1–45 45.1–50 50.1–60 60.0+

5.97 4.27 4.01 3.18 1.84 1.94 2.30 2.09 1.96 1.36

6.02 5.60 5.43 4.30 3.09 2.89 3.61 3.61 2.99 2.64

15.45 14.75 15.38 12.97 7.90 6.47 7.21 7.07 6.42 3.27

15.90 16.19 18.54 16.91 12.90 8.37 9.85 11.90 9.55 7.04

422

L. Barnard-Brak et al. / Research in Developmental Disabilities 36 (2015) 419–427

variable of stereotyped behavior was estimated using all items of stereotyped behavior subscale of the BPI while the latent variable of SIB was estimated using all items of self-injurious behavior subscale of the BPI. Both age and level of ID were observed variables. Fig. 1 provides the conceptual model that was tested. Several goodness-of-fit indices were used as criteria for judging the results. A statistically significant chi-square (x2) value was the first index of goodness of fit considered. We then evaluated the root mean square error of approximation (RMSEA) as indicating good fit if below 0.05, mediocre fit if between 0.08 and 0.10, and poor fit if greater than 0.10 (Browne & Cudeck, 1993; MacCallum, Browne, & Sugawara, 1996). Comparative Fit Index (CFI) and the Tucker-Lewis Fit Index (TLI) also known as the Non-Normed Fit Index were both considered to be an index of a well-fitting model if the values were greater than 0.95 (Hu & Bentler, 1999). In testing whether stereotyped behavior may be considered a predictive indicator of self-injurious behavior, this analysis does not provide any evidence of a causal relationship between the two variables as that would require an experimental design to examine. We estimated the direct effect between stereotyped behavior and self-injurious behavior and did not hypothesize any mediating or moderating relationships but we did statistically control for age and level of ID. For the second research question (Was there any relationship between stereotyped behavior and any individual topographies of SIB?), we tested whether stereotyped behavior as a construct appeared to be more associated with certain topographies of SIB over others for the sample as a whole. In considering the statistical significance, we first calculated the probability of committing at least one Type I error with the following family-wise or experiment-wise error rate equation:

Family-wise error rate ¼ 1  0:95k In this equation, k equals the number of multiple comparisons. We planned 14 multiple comparisons given the exploratory nature of this research question, which resulted in a family-wise error rate of 0.51. This error rate may be interpreted that there would be a 51% chance of committing one or more Type I error without adjusting the a-level. Given this relatively substantial probability of committing one or more Type I errors, we conservatively adjusted the alpha level via a Bonferroni correction with a resulting new a-level of 0.003. For the third research question (When stereotyped behavior is considered a predictor of SIB, what groups (if any) emerged from mixture modeling techniques?), we utilized mixture-modeling techniques in particular to latent class analyses to determine whether relevant hypothesized groups could be examined via MPlus (v. 7.0). A combination of statistics was evaluated in determining model fit. In particular, values of BIC (Bayesian Information Criterion) and AIC (Akaike Information Criterion) were evaluated. Smaller values of BIC and AIC indicate better comparative fit (D’Unger, Land, McCall, & Nagin, 1998; Nagin, 2005). Differences in both information criteria of 10 or more suggest very strong evidence of model fit corresponding to the odds of 150:1 (Raftery, 1995, p. 115). The Lo–Mendell–Rubin likelihood ratio test (LMR–LRT) was examined, which evaluated the null hypothesis that one fewer classes than tested (e.g., Ho: c-1 has better model fit) (Lo, Mendel, & Rubin, 2001). 4. Results 4.1. Was stereotyped behavior a significant predictor of SIB? To test the predictive relationship between stereotyped behavior and SIB, a path model was established using structural equation modeling techniques where we statistically controlled for level of ID and chronological age. Model fit was evaluated via values for root mean square error of approximation (RMSEA), Tucker–Lewis index (TLI), Comparative fit index (CFI), and chi-square (x2) for statistical significance (Byrne, 2011). Based upon the pre-existing item to factor structure of the BPI-01, the structural model estimates the fourteen SIB topographies as indicators of a SIB construct variable. The SIB construct variable was regressed onto the stereotyped behavior construct variable. The stereotyped behavior construct variable was estimated according to the 24 stereotyped behavior items per the BPI-01. Fig. 1 contains a visual diagram of the structural model. Results suggest adequate overall model fit: x2(736) = 2377.78, p < 0.05, RMSEA = 0.039, CFI = 0.913, and TLI = 0.908. SIB positively regressed onto stereotyped behavior (b = 0.69, p < 0.001) indicating that SIB appears to be

Fig. 1. Path diagram of the relationship between stereotyped behavior and SIB. yPlease note: Only the first and last item of the BPI-01 subscales ‘‘Stereotyped Behavior’’ and ‘‘SIB’’ are listed in the figure to conserve space. The rest of the items are represented by the dotted vertical lines. Black arrows indicate statistically significant paths at p < 0.05. The gray arrow indicates a modeled but statistically non-significant path.

L. Barnard-Brak et al. / Research in Developmental Disabilities 36 (2015) 419–427

423

significantly predicted by stereotyped behavior. In statistically controlling for age, the relationship of age with SIB was statistically significant with a standardized path coefficient value of b = 0.09, p < 0.05 indicating that as individuals grew older, than there was an associated decrease in SIB. There was a statistically significant relationship between age and stereotyped behavior with a standardized path coefficient value of 0.35, p < 0.05. Thus, as age increased, the frequency of stereotyped behavior also decreased. In statistically controlling for level of ID, there was a statistically significant relationship between level of ID and SIB with a standardized path coefficient value of 0.09, p < 0.05. As level of ID became less functional, SIB increased. There was not a statistically significant relationship between level of ID and stereotyped behavior. Table 2 provides the structural path model estimates for each item as they load onto their respective constructs. 4.2. Was there any relationship between stereotyped behavior and any individual topographies of SIB? Before adjusting the alpha level, there were 17 comparisons that revealed statistically significant results. After adjusting the level of statistical significance to p < 0.003 according to the Bonferroni correction, seven comparisons remained statistically significant. These comparisons revealed that stereotyped behaviors yelling and body rocking were significantly associated with five different topographies of SIB: self-biting; head hitting; body hitting; self-pinching; and hair pulling. 4.3. When stereotyped behavior is considered a predictor of SIB, what groups (if any) emerged from mixture modeling techniques? When applying mixture-modeling techniques, results revealed that a two-class solution was the best fitting model for the data, considering stereotyped behavior as a predictor of SIB. Confirming the two-class solution, a combination of fit statistics Table 2 Summary information for each of the two classes by BPI-01 item. Constructs and variables

b

SE

Class 1

Class 2

M

SD

M

SD

Self-injurious behavior Self-Biting Hitting Head Hitting Body Self-Scratching Vomiting Self-Pinching Pica Inserting Objects Pulling Fingers Inserting Fingers Air Swallowing Hair Pulling Extreme Drinking Teeth Grinding

0.59 0.79 0.78 0.54 0.45 0.65 0.69 0.72 0.58 0.67 0.61 0.54 0.65 0.50

0.03 0.03 0.03 0.03 0.05 0.03 0.03 0.04 0.05 0.03 0.06 0.04 0.04 0.04

0.53 0.79 0.63 0.37 0.11 0.25 0.25 0.03 0.03 0.12 0.05 0.18 0.14 0.24

0.96 1.14 1.08 0.85 0.49 0.71 0.76 0.26 0.21 0.54 0.38 0.63 0.64 0.77

0.32 0.48 0.24 0.26 0.16 0.12 0.35 0.08 0.08 0.21 0.10 0.20 0.18 0.40

0.82 0.99 0.75 0.76 0.61 0.55 0.91 0.43 0.46 0.71 0.48 0.67 0.70 0.96

Stereotyped behavior Rocking back and forth Sniffing objects Spinning own body Waving or shaking arms Rolling head Whirling on spot Repetitive body movements Pacing Twirling things Repetitive hand movements Yelling and screaming Sniffing own body Bouncing around Spinning objects Bursts of running Complex hand/finger mov. Manipulating objects Sustained finger movements Rubbing self Gazing at hands or objects Bizarre body postures Clapping hands Grimacing Waving hands

0.62 0.76 0.77 0.81 0.68 0.82 0.76 0.67 0.87 0.80 0.64 0.75 0.81 0.88 0.82 0.83 0.77 0.84 0.53 0.74 0.70 0.76 0.55 0.86

0.03 0.03 0.03 0.02 0.03 0.02 0.02 0.03 0.02 0.02 0.03 0.03 0.02 0.02 0.02 0.02 0.02 0.02 0.04 0.03 0.03 0.03 0.03 0.02

1.55 0.40 0.41 1.17 0.61 0.29 1.33 0.69 0.38 1.41 0.89 0.21 0.25 0.28 0.46 0.55 0.83 0.53 0.56 0.76 0.50 0.51 0.60 0.74

1.59 1.01 1.04 1.51 1.26 0.86 1.55 1.30 1.03 1.56 1.28 0.76 0.84 0.87 1.07 1.20 1.35 1.19 1.15 1.32 1.15 1.07 1.18 1.31

0.63 0.28 0.34 0.37 0.29 0.27 0.27 0.40 0.29 0.26 0.75 0.19 0.39 0.26 0.55 0.17 0.38 0.15 0.25 0.33 0.19 0.24 0.47 0.21

1.23 0.86 0.93 0.99 0.88 0.82 0.88 1.03 0.87 0.87 1.23 0.73 0.99 0.84 1.13 0.71 0.99 0.66 0.81 0.93 0.72 0.79 1.06 0.78

Notes: Class 1, Stereotyped behavior predicting SIB; Class 2, Stereotyped behavior NOT predicting SIB.

424

L. Barnard-Brak et al. / Research in Developmental Disabilities 36 (2015) 419–427

was evaluated. These statistics indicated that a two-class solution had the smallest BIC and AIC values that also satisfied the LMR-LRT for comparative statistical significance. The number of parameters was also evaluated as more parsimonious models tend to more accurately model data in the case of models that were equally well-fitting; however, this did not occur in the current study. The path estimates suggest that one class emerged in which stereotyped behavior was a strong, statistically significant predictor of SIB with a standardized path estimate of b = 0.82, p < 0.001. This class of individuals comprised approximately 69% (n = 1291) of the sample. Another class emerged in which stereotyped behavior was not a statistically significant predictor of SIB with a standardized path estimate of b = 0.18, ns. This class of individuals comprised approximately 31% (n = 580) of the sample. Table 3 provides a summary of the mixture modeling results. There was no statistically significant relationship between class membership and gender, ethnicity, level of ID, or aggressive/destructive behavior. There was a statistically significant relationship between class membership and age, t(1869) = 2.56, p < 0.001, d = 13. Individuals belonging to the class where stereotyped behavior significantly predicted SIB were significantly younger (M = 27.69, SD = 18.75) than individuals belonging to the class where stereotyped behavior did not significantly predict SIB (M = 30.44, SD = 21.79). As for SIB, individuals belonging to the class where stereotyped behavior significantly predicted SIB exhibited significantly more topographies of SIB (M = 2.33, SD = 2.25) versus individuals belonging to the class where stereotyped behavior did not significantly predict SIB (M = 0.64, SD = 1.20), t(1869) = 18.41, p < 0.001, d = 0.90. As for stereotyped behavior, individuals belonging to the class where it significantly predicted SIB exhibited significantly more stereotyped behavior (M = 5.17, SD = 5.32) versus individuals belonging to the class where stereotyped behavior did not significantly predict SIB (M = 2.29, SD = 3.51), t(1869) = 18.41, p < 0.001, d = 64. Table 1 summarizes stereotyped behavior and self-injurious behaviors for both classes. 5. Discussion The primary purpose of this study was to examine the viability of a model of stereotyped behavior predicting SIB among individuals with ID. Utilizing a large, international dataset, findings from this study indicate that stereotyped behavior is a significant predictor of SIB for some, but not all, individuals with ID. Specifically, stereotyped behavior was a strong, statistically significant predictor of SIB for approximately 69% (n = 1291) of the sample, but stereotyped behavior did not predict SIB in approximately 31% (n = 580) of the sample. However, broad participant characteristic variables such as gender, ethnicity, level of ID, or levels of aggressive/destructive behavior did not predict group membership for stereotyped behavior predicting SIB. The only participant variables that discriminated the individuals that stereotyped behavior predicted SIB were: (1) younger chronological age, (2) greater number of topographies of SIB, and (3) a greater number of topographies of stereotyped behavior. These findings suggest that early intervention for individuals with multiple topographies of SIB and stereotyped behavior may be one area for future research on attempting to prevent the development of additional topographies of SIB. However, additional research is needed to discriminate abnormal stereotypy from stereotypy that is part of the process for developing coordinated and more sophisticated motor skills in children (Thelen, 1979). Discriminating abnormal from typical stereotypy observed in infants and young children will likely require quantification of (1) chronicity, (2) intensity, and (3) resistance to change (i.e., blocking, redirecting motor responses to functional tasks), among other variables. This will be a critical discrimination to assure that early intervention efforts are allocated to individuals exhibiting abnormal stereotypy as opposed to transient stereotypy that is part of typically developing motor movements. If this can be done with high levels of accuracy, early intervention efforts can be modeled after the functional assessment approach discussed in Richman (2008). That is, behavior management strategies that have been shown to be effective with well established cases of stereotypy and SIB can be matched to the current function of these behaviors, while simultaneously focusing on skill acquisition (e.g., adaptive behavior such as social and play skills) and Functional Communication Training for broad classes of reinforcement (e.g., mands for positive and negative reinforcement). Another area of future research may be to examine for the presence of mediating or moderating variables in the relationship between stereotyped behavior and self-injurious behavior among those individuals where stereotyped behavior was a strong predictive indicator of self-injurious behavior. For instance, preliminary results did not indicate a relationship between aggressive and destructive behaviors (as measured by the BPI-01 as well) with group membership (e.g., whether stereotyped behavior predicted SIB). Future research should continue to investigate the possible role of a wide variety of participant characteristics serving as mediating or moderating variables for stereotypy predicting development of SIB.

Table 3 Mixture modeling two-class solution. Classes

Parameters

BIC

AIC

LMR-LRT

1 2 3 4 5 6

77 119 161 203 245 No convergence

32,894.94 32,833.13 32,922.46 33,083.35 33,157.90

32,494.85 32,214.82 32,085.92 32,028.58 31,884.89

– Reject, p < 0.05 Do not reject Do not reject Do not reject

L. Barnard-Brak et al. / Research in Developmental Disabilities 36 (2015) 419–427

425

With regards to specific topographies of stereotyped behavior that predicted topographies of SIB, yelling and body rocking were significantly associated with the following topographies of SIB: self-biting; head hitting; body hitting; selfpinching; and hair pulling. This finding is difficult to interpret in terms of the current literature on proto-injurious behavior, given that there does not appear to be any topographical association between stereotyped behaviors that predicted SIB, especially with regard to yelling. That is, previous research (Guess & Carr, 1991; Richman & Lindauer, 2005) has suggested that some forms of SIB may evolve from topographically similar forms of stereotyped behavior such as hand flapping in the head and shoulder region evolving into head hitting. However, body-rocking as a potential risk marker for later development of gross motor SIB may be easier to explain if we allow for an extension of the proto-injurious behavior model of SIB. One could argue that restricting proto-injurious behavior to topographic similarities with the eventual self-injurious behavior is unnecessarily limiting. Instead, the similarity between proto- and actually damaging behavior outcome could perhaps also be based on other behavioral dimensions, such as dynamic similarities. For instance, (repetitive) body-rocking as in this data set may have gross-motor, rhythmic-repetitive, dynamic characteristics that are similar to gross-motor, rhythmic-repetitive head hitting and body hitting. That is, body rocking could be viewed as a proto-injurious behavior with a certain likelihood for evolving into head or body hitting/ banging on the basis of their dynamic similarities. There are several behavioral principles that could account for such a process of evolution, including behavior shaping and stimulus fading contingencies, or a growing tolerance to the stimulatory consequences of body-rocking that require head banging as response with the same dynamic but increased intensity that produces a ‘‘dose increase’’ of those consequences produced by engaging in the behavior. Future research will need to replicate and extend the current study to determine if yelling and body rocking are indeed high risk stereotyped behavior for later development of topographically dissimilar SIB. The primary implication of the current study is how results inform our conceptualization of the relation between stereotyped behavior and SIB among individuals with intellectual and developmental disabilities. Results from the current study indicate that stereotyped behavior is a significant predictor of SIB in 69% of our sample of 1871 cases of individuals with ID. Our results suggest that stereotyped behavior is a potential risk marker for SIB for a subset of individuals with ID. If these findings are replicated, future research should examine whether early intervention for chronic stereotyped behavior could prevent some forms of later development of SIB. The younger the participants, and participants with relatively greater number of topographies of SIB and stereotyped behavior, showed the highest level of stereotyped behavior predicting SIB. These findings are also important for consideration as we move from a reactive model of treatment of SIB to a preventative model of SIB. Our findings suggest that intervening earlier with regards to the participants’ chronological age and intensive intervention for individuals with multiple topographies of stereotyped behavior or SIB may prevent the development of additional topographies of SIB. Additional research is needed to replicate these findings and to identify the number of topographies of SIB or stereotyped behavior that may increase the probability of stereotyped behavior predicting later development of SIB. Several limitations emerged as part of the current study. First, mixture-modeling techniques may be prone to arriving at local solutions that are data-driven rather than conceptual or theoretical findings (Hipp & Bauer, 2006). Local solutions are typically only valid to the sample of data from which they are derived rather than being generalizable across different samples. To address this issue of possibly arriving at a local solution in the current study, results were cross-validated with a random split-half of the sample, which revealed a two-class solution similar to that found from the whole sample. Second, our measure of SIB was not all inclusive of all possible topographies of SIB. Confirmatory factor analyses although do indicate the psychometric sufficiency of the SIB subscale (and stereotyped behavior as well). Third, individuals with ID belonging to stereotyped behavior predicting SIB group remain quite nebulous in terms of determining a statistically significant predictor of SIB from the data available. For instance, as approximately half of the sample that belonged to the group where stereotyped behavior did not predict SIB, these individuals may have had some coexisting or comorbid psychiatric disorder. The presence of some intervening or confounding additional disorder may account for the lack of a relationship between stereotyped behavior and SIB. The data set did not provide this information as to the specific type of comorbid psychiatric diagnosis precluding statistical analysis. A review of the literature does appear to indicate the presence of mental health disorders as co-occurring with intellectual and developmental disorders at a higher rate than the general population (e.g., for intellectual disability and autism spectrum disorders, Bradley, Summers, Wood, & Bryson, 2004; Hill & Furniss, 2006). Thus, a comorbid mental health disorder may account for some proportion of the SIB not being predicted by stereotyped behavior among individuals with intellectual and developmental disabilities (see Matson & Turygin, 2012). For example, Richman et al. (2013) showed that high level of impulsivity, as measured by the Aberrant Behavior Checklist, was one variable (in addition to levels of ID and stereotypic behavior) that predicted SIB in a sample of 617 individuals with autism. Additional information such as the percent of cases in nonambulatory care settings or level of communication ability (e.g., nonverbal status) was not provided precluding analysis of these variables as well. Finally, results of our second research question examining whether stereotyped behavior may serve as a predictive indicator of any individual topographies of SIB may be considered an exploratory analysis of the data rather than based upon a theoretical or conceptual model in contrast to our first research question, which maintained a clear hypothesis. Future research should utilize results of the current study to aid in the development of hypotheses between stereotyped behavior and individual topographies of SIB. As a result, future research should consider the role of coexisting and comorbid disorders among individuals with intellectual and developmental disorders in relation to stereotyped behavior as a possible risk marker for SIB. Future research should also consider whether defining and measuring SIB as repetitive or discrete bouts of SIB for individuals with intellectual and developmental disabilities is helpful to investigators studying stereotyped behavior as a potential risk

426

L. Barnard-Brak et al. / Research in Developmental Disabilities 36 (2015) 419–427

marker for SIB. Future research needs to determine whether repetitive or discrete bouts of SIB are more likely to occur among individuals with intellectual and developmental disabilities with high and low levels of stereotyped behavior and presence or absence co-morbid psychiatric diagnoses. In conclusion, the current study provides an empirical examination of stereotyped behavior predicting SIB among individuals with intellectual and developmental disabilities. The current study provides evidence to suggest stereotyped behavior did predict SIB for 69% of the current large sample. The critical next steps in this line of research are to replicate the relation between stereotyped behavior and SIB, identify variables that discriminate abnormal stereotypy from stereotypy that is part of typical motor development, and to document the effects of early intervention strategies on preventing the development of additional topographies of stereotypy and SIB. References Baghdadli, A., Pascal, C., Grisi, S., & Aussilloux, C. (2003). Risk factors for self-injurious behaviours among 222 young children with autistic disorders. Journal of Intellectual Disability Research, 47, 622–627. Baumeister, A. A., & Forehand, R. (1973). Stereotyped acts. International Review of Research in Mental Retardation, 6, 55. Baumeister, A. A., & Rollings, J. P. (1976). Self-injurious behavior. In N. Ellis (Ed.), International review of research in mental retardation (8th). New York: Academic Press. Berkson, G., Tupa, M., & Sherman, L. (2001). Early development of stereotyped and self-injurious behaviors: I. Incidence. American Journal on Mental Retardation, 106, 539–547. Bodfish, J. W., Crawford, T. W., Powell, S. B., Parker, D. E., Golden, R. N., & Lewis, M. H. (1995). Compulsion in adults with mental retardation: Prevalence, phenomenology, and comorbidity with stereotypy and self-injury. American Journal on Mental Retardation, 100, 183–192. Bodfish, J. W., Symons, F. J., Parker, D. E., & Lewis, M. H. (2000). Varieties of repetitive behavior in autism: Comparisons to mental retardation. Journal of Autism and Developmental Disorders, 30, 237–243. Bradley, E. A., Summers, J. A., Wood, H. L., & Bryson, S. E. (2004). Comparing rates of psychiatric and behavior disorders in adolescents and young adults with severe intellectual disability with and without autism. Journal of Autism and Developmental Disorders, 34(2), 151–161. http://dx.doi.org/10.1023/ B:JADD.0000022606.97580.19 Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Bollen & J. S. Long (Eds.), Testing structural equation models (pp. 136–162). Newbury Park, CA: Sage. Byrne, B. (2011). Structural equation modeling with Mplus: Basic concepts, applications, and programming. New York: Routledge. Didden, R., Sturmey, P., Sigafoos, J., Lang, R., O’Reilly, M., & Lancioni, G. E. (2012). Nature, prevalence and characteristics of challenging behavior. In J. L. Matson (Ed.), Functional assessment for challenging behaviors. Autism and child psychopathology series (pp. 25–44). New York, NY, USA: Springer Science+Business Media http://dx.doi.org/10.1007/978-1-4614-3037-7_3 ix, 243pp. D’Unger, A., Land, K., McCall, P., & Nagin, D. (1998). How many latent classes of delinquent/criminal careers?. Results from mixed Poisson regression analyses of the London, Philadelphia, and Racine cohort studies. American Journal of Sociology, 103, 1593–1630. Furniss, F., & Biswas, A. B. (2012). Recent research on aetiology, development and phenomenology of self-injurious behaviour in people with intellectual disabilities: A systematic review and implications for treatment. Journal of Intellectual Disability Research, 56(5), 453–475. Guess, D., & Carr, E. (1991). Emergence and maintenance of stereotypy and self-injury. American Journal on Mental Retardation, 96, 299–319. Hall, S., Oliver, C., & Murphy, G. (2001). Early development of self-injurious behavior: An empirical study. American Journal on Mental Retardation, 106(2), 189–199. Hastings, R. P., Beck, A., Daley, D., & Hill, C. (2005). Symptoms of ADHD and their correlates in children with intellectual disabilities. Research in Developmental Disabilities, 26(5), 456–468. Hill, J., & Furniss, F. (2006). Patterns of emotional and behavioural disturbance associated with autistic traits in young people with severe intellectual disabilities and challenging behaviours. Research in Developmental Disabilities, 27(5), 517–528. http://dx.doi.org/10.1016/j.ridd.2005.07.001 Hipp, J. R., & Bauer, D. J. (2006). Local solutions in the estimation of growth mixture models. Psychological Methods, 11(1), 36–53. Hu, L-T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. Kozlowski, A. M., & Matson, J. L. (2012). An examination of challenging behaviors in autistic disorder versus pervasive developmental disorder not otherwise specified: Significant differences and gender effects. Research in Autism Spectrum Disorders, 6(1), 319–325. Kurtz, P. F., Chin, M. D., Huete, J. M., & Cataldo, M. F. (2012). Identification of emerging self-injurious behavior in young children: A preliminary study. Journal of Mental Health Research in Intellectual Disabilities, 5(3–4), 260–285. Langthorne, P., & McGill, P. (2008). Functional analysis of the early development of self-injurious behavior: Incorporating gene–environment interactions. American Journal of Mental Retardation, 113(5), 403–417. Lo, Y., Mendell, N., & Rubin, D. B. (2001). Testing the number of components in a normal mixture. Biometrika, 88, 767–778. MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power analysis and determination of sample size for covariance structure modeling. Psychological Methods, 1, 130–149. Matson, J. L., & Turygin, N. C. (2012). How do researchers define self-injurious behavior? Research in Developmental Disabilities, 33(4), 1021–1026. McClintock, K., Hall, S., & Oliver, C. (2003). Risk markers associated with challenging behaviours in people with intellectual disabilities: A meta-analytic study. Journal of Intellectual Disability Research, 47(6), 405–416. http://dx.doi.org/10.1046/j.1365-2788.2003.00517.x Mircea, C. E., Rojahn, J., & Esbensen, A. J. (2010). Psychometric evaluation of Romanian translations of the Behavior Problems Inventory-01 and the Nisonger Child Behavior Rating Form. Journal of Mental Health Research in Intellectual Disabilities, 3, 51–65. Murphy, G., Hall, S., Oliver, C., & Kissi Debra, R. (1999). Identification of early self-injurious behaviour in young children with intellectual disability. Journal of Intellectual Disability Research, 43(3), 149–163. Muthe´n, L. K., & Muthe´n, B. O. (2012). Mplus user’s guide (7th ed.). Los Angeles, CA: Muthe´n & Muthe´n. Nagin, D. S. (2005). Group-based modeling of development. Cambridge, MA: Harvard University Press. Oliver, C., Murphy, G. H., & Corbett, J. A. (1987). Self-injurious behaviour in people with mental handicap: A total population study. Journal of Intellectual Disability Research, 31(2), 147–162. http://dx.doi.org/10.1111/j.1365-2788.1987.tb01351.x Owen, D. M., Hastings, R. P., Noone, S. J., Chinn, J., Harman, K., Roberts, J., et al. (2004). Life events as correlates of problem behavior and mental health in a residential population of adults with developmental disabilities. Research in Developmental Disabilities, 25, 309–320. Petty, J., Allen, D., & Oliver, C. (2009). Relationship among challenging, repetitive, and communicative behaviors in children with severe intellectual disabilities. American Journal on Intellectual and Developmental Disabilities, 114(5), 356–368. Raftery, A. E. (1995). Bayesian model selection in social research. Sociological methodology, 25, 111–164. Richman, D. (2008). Early intervention and prevention of self-injurious behavior exhibited by young children with severe developmental delays. Journal of Intellectual Disability Research, 5, 3–17. Richman, D., & Lindauer, S. (2005). Longitudinal functional analysis of stereotypic, proto-injurious, and self-injurious behavior in young children with developmental delays. American Journal on Mental Retardation, 110, 439–450. Richman, D. M., Barnard-Brak, L., Bosch, A., Thompson, S., Grubb, L., & Abby, L. (2013). Predictors of self-injurious behaviour exhibited by individuals with autism spectrum disorder. Journal of Intellectual Disability Research, 57, 429–439. http://dx.doi.org/10.1111/j.1365-2788.2012.01628.x

L. Barnard-Brak et al. / Research in Developmental Disabilities 36 (2015) 419–427

427

Rojahn, J., Matlock, S. T., & Tasse´, M. J. (2000). The stereotyped behavior scale: Psychometric properties and norms. Research in Developmental Disabilities, 21(6), 437–454. Rojahn, J., Matson, J. L., Lott, D., Esbensen, A. J., & Smalls, Y. (2001). The Behavior Problems Inventory: An instrument for the assessment of self-injury, stereotyped behavior and aggression/destruction in individuals with developmental disabilities,. Journal of Autism and Developmental Disorders, 31, 577–588. Rojahn, J., Rowe, E. W., Macken, J., Gray, A., Delitta, D., Booth, A., et al. (2010). Psychometric evaluation of the Behavior Problems Inventory-01 and the Nisonger Child Behavior Rating Form with children and adolescents. Journal of Mental Health Research in Intellectual Disabilities, 3, 28–50. Rojahn, J., Rowe, E. W., Kasdan, S., Moore, L., & van Ingen, D. J. (2011). Psychometric properties of the Aberrant Behavior Checklist, the Anxiety, Depression and Mood Scale, the Assessment of Dual Diagnosis and the Social Performance Survey Schedule in adults with intellectual disabilities. Research in Developmental Disabilities, 32, 2309–2320. Rojahn, J., Rowe, E. W., Sharber, A. C., Hastings, R. P., Matson, J. L., Didden, R., et al. (2012a). The Behavior Problems Inventory-Short Form (BPI-S) for Individuals with Intellectual Disabilities I: Development and provisional clinical reference data. Journal of Intellectual Disability Research, 56, 527–545. Rojahn, J., Rowe, E. W., Sharber, A. C., Hastings, R. P., Matson, J. L., Didden, R., et al. (2012b). The Behavior Problems Inventory-Short Form (BPI-S) for Individuals with Intellectual Disabilities II: Reliability and validity. Journal of Intellectual Disability Research, 56, 546–565. Rojahn, J., Schroeder, S. R., & Hoch, T. A. (2008). Self-injurious behavior in intellectual disabilities. Amsterdam: Elsevier. Symons, F. J., Sperry, L. A., Dropik, P. L., & Bodfish, J. W. (2005). The early development of stereotypy and self-injury: A review of research methods. Journal of Intellectual Disability Research, 49(2), 144–158. Thelen, E. (1979). Rhythmical stereotypies in normal human infants. Animal Behaviour, 27, 699–715. Zaja, R. H., Moore, L., van Ingen, D. J., & Rojahn, J. (2011). Psychometric comparison of the functional assessment instruments QABF, FACT, and FAST for selfinjurious, stereotypic and aggressive/destructive behavior. Journal of Applied Research in Intellectual Disabilities, 24, 18–28.