Predicting self‐injurious behavior at age three among infant siblings of children with autism

Existing research suggests that self‐injurious behavior (SIB) is a relatively common interfering behavior that can occur across the lifespan of individuals with autism spectrum disorder (ASD). We previously reported that SIB or proto‐injurious SIB at 12 months was related to increased risk of SIB at 24 months among a preschool sample of children with a high familial likelihood for ASD (Dimian et al., 2017). In the present study, we extend these findings, examine SIB occurrence, and associated potential risk factors at 36 months. The present sample included 149 infants with an older sibling with ASD (65.8% male) who completed assessments at ages 12, 24, and 36 months. Descriptive analyses and binary logistic regression models were utilized. SIB was more prevalent among those children who received a diagnosis of ASD. Logistic regression indicated that presence of SIB, stereotypy, hyper‐ and hypo‐ sensory responsivity, and lower intellectual functioning at age 12 months significantly predicted the occurrence of SIB at 36 months. These findings have implications for understanding developmental processes culminating in persistent SIB and may inform prevention programming.

Existing research suggests that self-injurious behavior (SIB) is a relatively common behavior disorder among people with intellectual and developmental disabilities (I/DD) and is observed across the lifespan (Davies & Oliver, 2013;Taylor et al., 2011).Individuals with autism spectrum disorder (ASD) tend to engage in SIB at a higher rate in comparison to those with developmental delay (Soke et al., 2019), those at risk for I/DD (Schroeder et al., 2014) and Down syndrome (Richards et al., 2012).A recent meta-analysis estimated that the prevalence of SIB is 42% among individuals with ASD with and without co-occurring intellectual disability (Steenfeldt-Kristensen et al., 2020).SIB is considered a form of restricted and repetitive behavior and includes any rhythmic, self-directed behavior that has the potential to result in injury, such as self-directed hitting or biting, head banging, or skin picking (Tate & Baroff, 1966).While there are many cross-sectional studies of SIB, longitudinal studies of SIB are limited, particularly those using prospective designs during early childhood (e.g., Schroeder et al., 2014).SIB is notoriously demanding and costly to treat once it is established in a behavioral repertoire and can be highly resistant to treatment.In general, SIB can have negative impacts on health, as well as quality of life for those affected and their caregivers.For instance, children with ASD who engage in SIB are twice as likely to be hospitalized in a psychiatric unit than children with ASD that do not engage in SIB (Mandell, 2007).Additionally, SIB persistence (i.e., ch ronic SIB) estimates have been reported as high as 84% (Taylor et al., 2011) and 78% over time (Richards et al., 2016).It is therefore critical that early intervention and prevention programming address and target precursors of SIB for children at risk as early as possible.
The development and emergence of SIB has received considerably less empirical attention than the persistence and treatment of established SIB.The phenomenology of SIB is not unique to people with intellectual and developmental disabilities but is also observed within neurotypical individuals.Thelen (1979) identified early rhythmic behavior developed in synchrony with motor and vocal development among neurotypically developing infants.RRB and SIB are prevalent within the first years of life but tend to be less prominent among neurotypically developing populations (Baranek et al., 2007;Fodstad et al., 2012;MacLean et al., 2020;Wolff et al., 2014).Of the few comparative analyses of preschoolers to date, SIB rates tend to be twice as high among children experiencing developmental delays in comparison to children without delays (Fodstad et al., 2012;Hoch et al., 2016;MacLean et al., 2020).Although base rates differ even early on, children with or at high likelihood for ASD have a higher propensity to engage in chronic SIB (e.g., Rattaz et al., 2015).
Previous reports indicate that children who exhibit SIB engage in multiple forms or topographies (Berkson et al., 2001;Kurtz et al., 2012;MacLean et al., 2020) that can change over time (e.g., Richman & Lindauer, 2005).Some studies indicate that head hitting tends to emerge first (e.g., Berkson et al., 2001) and is the most common topography among individuals with ASD (Steenfeldt-Kristensen et al., 2020).Moreover, SIB is commonly characterized as proto-injurious early on when the behavior does not cause tissue damage yet has the potential to do so (e.g., hand mouthing could deteriorate and break down the skin with sufficient exposure).There is no predetermined threshold at which caregivers tend to exhibit concern about SIB or consensus about when SIB becomes clinically significant and requires treatment.Because SIB is dynamic, prospective, longitudinal analyses of the forms, function, severity, and the contributing factors are necessary to identify who may develop chronic or clinically significant SIB.
Researchers, clinicians, and stakeholders have hypothesized as to the etiology and the function or reason why individuals engage in SIB.For example, Guess and Carr (1991) posited that stereotyped behavior and SIB emerge as repetitive behaviors that at first serve a homeostatic function and then become sensitive to socially mediated reinforcement contingencies.Similarly, applied behavior analytic research repeatedly demonstrates empirical evidence that SIB can be maintained by operant reinforcement contingencies and functions include attention, tangible, escape/avoidance, and automatic/sensory (e.g., Iwata et al., 1994).Some caregivers also report that SIB occurs due to anxiety (Russell et al., 2019).Other research with autistic youth and adults who self -report about their experiences with SIB indicate that it is a means of regulating negative emotions/depression and high energy states such as anger (Moseley et al., 2019).Alternatively, there is a body of research suggesting SIB is due to underlying biochemical processes (e.g., endogenous opioids; Huisman et al., 2018).How and why SIB developmentally unfolds, however, ultimately requires longitudinal investigations.
The biobehavioral or psychosocial factors associated with the emergence and persistence of SIB in ASD need to be characterized to identify who is likely to develop chronic versus transient SIB.Putative risk markers and predictors investigated for the emergence and persistence of SIB among individuals with ASD include symptom severity (Baghdadli et al., 2003;Rattaz et al, 2015), impulsivity/hyperactivity (Richards et al., 2012;Richards et al., 2016), intellectual functioning (Dimian et al., 2017;Richards et al., 2012), sensory features (Duerden et al., 2012), and stereotypy (Richman et al., 2013).Methodological variability regarding how SIB was measured, the age groups included, sampling methods, and the study designs utilized make it difficult to generalize and come to a consensus about risk factors for SIB.
Within the repetitive behavior domain, atypical sensory responses are a common feature of ASD (Baranek et al., 2006;Ben Sasson et al., 2007).The relation between SIB and sensory features has received limited attention; however, Duerden et al. (2012) reported that atypical sensory processing was significantly associated with SIB.Sensory features explained approximately 12% of the variance within their statistical model.Wolff et al. (2019) also reported that SIB and hyper-responsivity were highly correlated at 12 and 24 months among infants with a high likelihood for ASD.Further, early sensory behaviors have been associated with later adaptive behaviors among individuals with ASD as well (Williams et al., 2018).Taken together, evaluating the predictive value of early sensory responses could help identify who is at risk for chronic and persistent SIB over time and identify efficacious interventions.Similarly, the extent to which an ASD diagnosis contributes to the risk for SIB also requires further investigation.

PURPOSE OF THE PRESENT STUDY
We previously reported that lower intellectual functioning, SIB and stereotypy endorsement at 12 months predicted SIB at 24 months (Dimian et al., 2017).consistent with previous studies (Richards et al., 2016), lower communication or speech level (Baghdadli et al., 2008), and greater severity of ASD symptoms (Baghdadli et al., 2008) were more likely to engage in or persist with SIB.The purpose of the present study was to extend our previous findings of predictive risk markers to 36 months of age.We hypothesized that the potential risk factors we previously reported (Dimian et al., 2017) in addition to sensory features would predict SIB at 36 months among toddlers with a high likelihood for ASD.We further explored topographical changes in SIB over repeated measurement points during the first 3 years of life.Our specific research questions therefore include (1) Do potential risk factors reported at age 12 months in Dimian et al. (2017)  (e) significant perinatal adversity or exposure to neurotoxins, (f) contraindication for MRI, (g) predominant home language other than English, (h) adopted or half siblings, (i) first-degree relative with bipolar disorder, psychosis, or schizophrenia, and (j) twins.
Study procedures were approved by institutional review at each clinical site and written, informed consent was obtained.All participants were sibling infants at high familial risk for ASD (i.e., all participants had an older sibling with an existing diagnosis of ASD confirmed by the Autism Diagnostic Interview-Revised (ADI-R; Lord, Rutter, & Couteur, 1994) and the Social Communication Questionnaire (SCQ; Rutter et al., 2003)).Participants received an autism diagnostic assessment at 24 and 36 months of age.Clinical best-estimate diagnosis was made by expert clinicians using DSM-IV-TR criteria for autistic disorder or PDD-NOS.The Autism Diagnostic Observation Scale (ADOS; Lord et al., 2000;ADOS-2;Lord et al., 2012), ADI-R, and Mullen Scales of Early Learning (MSEL; Mullen, 1995) were used to inform if a diagnosis was given.Overall, this study included 149 children of whom 41 (27.5%) met diagnostic criteria for ASD at 24 or 36 months.
Table 1 displays demographic information for the sample.The study sample was 65.8% male and mean ages at 12, 24, and 36 -month assessment dates were 12.50 (SD = 0.58), 24.72 (SD = 1.45), and 40.02 months (SD = 6.51).Among the sample, approximately 80% were white and most families reported an income of over $100 k a year.Most mothers and fathers had a college degree.
Four groups (persistent SIB, transient SIB, ASD, and no ASD) were created to evaluate the difference in SIB endorsement and severity over 12, 24, and 36 months of age.Persistent cases included participants that received an endorsement of SIB at all three time points and transient cases were defined as endorsement of SIB across one or two time points.Those who received a diagnosis of ASD at 24 months and those who did not receive a diagnosis of ASD were also compared.

Measures
Repetitive Behavior Scale -Revised.(RBS-R; Bodfish et al., 2000).The RBS-R is a caregiver rated measure of RRB, which includes 43 observable behavioral topographic items across six separate subscales (stereotyped motor behavior, self-injurious behavior, compulsive behavior, ritualistic behavior, sameness behavior, and restricted behavior).The current study utilized the stereotyped and SIB subscales and calculated items endorsed and severity using frequency rating scores (0-behavior does not occur, 1-behavior occurs and is a mild problem, 2-behavior occurs and is a moderate problem, 4-behavior occurs and is a severe problem).Presence or absence of SIB was defined by endorsement on the RBS-R.
Vineland adaptive behavior scales-II.(Vineland-II; Sparrow, Balla, & Cicchetti, 2005).The Vineland-II is a standardized, semi-structured assessment of adaptive behavior.The Adaptive Behavior Composite score (ABC) was derived from the Vineland-II.Daily living scores were not examined for this study given limited utility and variability for these scores in toddlers.
Mullen scales of early learning.The MSEL is a standardized measure of motor and cognitive development for children from birth to age 68 months.It provides an Early Learning Composite (ELC) score that was utilized as a general developmental index.
Sensory experiences questionnaire, version 2.1.(SEQ; Baranek et al., 2006).The SEQ is 33 items Likert scaled caregiver questionnaire on patterns of sensory response that include Hyper-responsiveness, Hypo-responsiveness, and Sensory Seeking, occurring within Social or Nonsocial contexts.The items are grouped into the following five categories: auditory, tactile, visual, gustatory/olfactory, and vesitibular/proprioceptive.The present study primarily focused on response pattern scores for Hyperresponsiveness (overactivity/avoidance of a stimuli such as covering your ears to avoid auditory stimuli), Hyporesponsiveness (underactivity/a lack of or a delayed response to a stimuli), and Sensory Seeking (repeated or intense interest with specific sensory stimuli) to test specific a priori hypotheses.We did not include sensory features in our previous model (i.e., Dimian et al., 2017) but have added them in the current study based on the literature to data (Duerden et al., 2012;Wolff et al., 2019).
Autism diagnostic observation schedule.The ADOS is a semi-structured diagnostic assessment designed to examine core ASD symptoms and behavior.Participants were assessed at 36 months using either module 1 or 2. Symptom severity was characterized using standardized severity scores from the ADOS (Gotham et al., 2009).

Statistical analyses
Point-prevalence of SIB at each time point, cumulative incidence, and relative risk were calculated to assess the occurrence of SIB over time.Binary logistic regression models were fitted to specifically test T1 predictors of parent-endorsed SIB (binary variable including SIB or no SIB) at T3.We utilized odds ratios and X 2 omnibus tests of model coefficients to evaluate each predictor variable tested and Nagelkerke R 2 to assess model fit.Variance inflation factors (VIF) were computed to assess multicollinearity between the predictors variables and was within the acceptable range.Group-level differences in terms of SIB form at each time point were based on SIB status across time and are descriptively presented.Missing data were excluded from all analyses and only participants with scores on each of the measures at each time point were included in the statistical analyses.All analyses were conducted using SPSS 26.

Prevalence, incidence, and relative risk estimates
Self-injurious behavior point prevalence estimates based on an any parent endorsement of SIB on the RBS-R for the whole sample were 38.9% at 12 months, 34.9% at 24 months, and 22.1% at 36 months.SIB cumulative incidence (i.e., new cases at 36 months) was 2% over 12 months (at T3).ASD diagnosis prevalence (diagnosis at 24 or 36 months) was 27.5%.Prevalence of SIB at 36 months among children diagnosed with ASD was 48.8% compared to 12.1% among those who did not.The relative risk of engaging in SIB at 36 months was 4.05 times higher among children who were diagnosed with ASD compared to children with no diagnosis (95% CI = 2.23,7.37).

Persistence of SIB
Self-injurious behavior was reported at all three time points for 6% of the sample and at least one or two time points for 50% of the sample (i.e., transient SIB).SIB at both 24 and 36 months was observed in 14% of the participants.Caregivers did not endorse any SIB items at any time point for 42% of the sample.Of the participants with no SIB reported by caregivers at any time point, 15% received a diagnosis of ASD.Additionally, among the participants that were characterized as having transient SIB, 33% received a diagnosis of ASD.In contrast, 56% of the participants with persistent SIB received a diagnosis at age 3 years.

SIB topography over time
The most common type of SIB in both persistent and transient SIB groups was "hits self against surface" (Figure 1).At 36 months, "rubs/scratches and skin picking" increased in endorsement and severity for the persistent group and are displayed in Figure 1.There were higher rates of SIB overall, for the ASD group (Figure 2) compared to those without a diagnosis.Compared to 12 months, approximately 65% of the ASD group endorsed a new form of SIB at 24 or 36 months in comparison to 29% of the non-ASD group.At 24 and 36 months, the most common new forms of SIB endorsed by caregivers across both groups was "hits self with body part" (ASD = 24%; Non-ASD = 8%) and "inserts finger or object" (ASD = 22%; Non-ASD = 10%).Severity scores were greater at 12 months among the non-ASD group but decreased by 36 months.In contrast, rates and severity scores on average increased by 36 months among the ASD group with "hits self against surface" remaining high.
Figure 3 displays the number of emergent topographies (i.e., topographies that were not endorsed at 12 months but were at follow up) at 24 and 36 months for both the participants with a diagnosis of ASD and those without.The radar graphs show disparate patterns at each time point with the participants with ASD tending to develop "hits self against surface or object" but not  "hits self with object" and "skin picking" at 24 months.In contrast, participants without a diagnosis of ASD at 24 months endorsed "inserts finger or object" (e.g., eye poking/ear poking) as the most common emergent topography and "pulls" and "rub or scratches self" as the least.At 36 months, among the participants with ASD, "hits self with body part" was the primary topography that emerged and for participants without ASD, "skin picking" was the most common.Both "hits self with an object" and "biting" were least common for those with ASD and "hits self against surface or object" and "hits self with object" were the least common emergent topographies for participants without a diagnosis.Overall, the absolute number of topographies that newly emerged at 24 and 36 months was greater for those who received a diagnosis of ASD.

Predicting SIB at 3 years of age
Two logistic regression models were calculated to evaluate which psychosocial variables at 12 months (T1) predicted SIB occurrence at 36 months (T3).Table 2 displays the results of the models conducted.The first model replicated Dimian et al. (2017) and included sex, Mullen early learning composite score, SIB endorsed, stereotyped behavior endorsed, and the Vineland adaptive behavior composite score (ABC) at 12 months predicting SIB occurrence at 36 months.The model was statistically significant at an alpha level of 0.05 (χ 2 = 29.67,df = 5, p < 0.001, R 2 pseudo = 0.29).Mullen early learning composite score and stereotyped behavior scores were significantly associated with SIB at 36 months.The odds of SIB at 36 months was two-fold for children with stereotyped behavior at 12 months (OR = 2.24, 95% CI [1.54,3.25])and odds of SIB increased by 4% for each point decrease in Mullen composite score (OR = 0.96, 95% CI [0.92,1.00]).Model A's positive predictive value (PPV) was 84.2%.

DISCUSSION
We found that SIB in 36 month old children at high likelihood for ASD was more prevalent among children who received a diagnosis of ASD; four times higher than children without ASD.Logistic regression results were mixed, however, the best fitting model with the highest positive predictive value showed that presence of SIB, stereotypy, hyper-and hypo-responsivity, and lower intellectual functioning at age 12 months significantly predicted the occurrence of SIB at 36 months.Descriptive data indicated increased SIB severity at time 3 relative to earlier time points.This suggests that even though fewer children are engaging in SIB by age 3 years, the behavior becomes more severe for those who continue to engage in it.

SIB topographies
The results also indicate that "hitting self against surface" may be one of the first forms of SIB to emerge and may increase in severity over time.The data indicate that skin picking is also a topography that becomes more problematic over time.Overall, for the persistent cases (i.e., SIB at each time point), "hits self against surface" was endorsed the most and resulted in the greatest severity scores, particularly at 24 months."Rubs or scratches" and "skin picking" showed an accelerating trend with increased endorsement and severity at 36 months, which was a differentiated the persistent group from the others.The transient cases also indicated, "hits self against surfaces" as more severe, peaking at 12 months and then a decreasing over time.Additionally, the ASD group overall had higher endorsement of all topographies and greater severity scores over time.Again, "hits self against surface" was endorsed the most for both diagnostic groups but was more severe by 36 months for those with an ASD diagnosis relative to those without.For the ASD group, "hits self against surface" was the emergent topography endorsed most often at 24 months, whereas "inserts finger or object" was endorsed the most for the group without a diagnosis.At age 36 months, there were fewer new topographies reported within the sample overall.Age trends related to SIB topography provide valuable information about the emergence of SIB.The extent to which young children persist with a specific form may elucidate which topographies may be more salient early intervention targets.Because we know that SIB and proto-injurious self -directed behaviors are prevalent in the early years of life among both neurotypical and neurodiverse children (e.g., Roane et al., 2007;Wolff et al., 2016), tracking changes in form, frequency, and severity may facilitate the creation of a taxonomy and nosology of SIB.The present topography results are somewhat consistent with findings that SIB resulting in tissue damage can emerge before 2 years, and that head hitting or banging is the earliest topography observed in young children with intellectual and developmental disabilities (Baghdadli et al., 2008;Berkson et al., 2001;Hall et al., 2001;Richman & Lindauer, 2005).These topographies have also been found to be transient and children may stop or conversely may begin to engage in multiple new topographies (Kurtz et al., 2012).Among the ASD group, there was a higher absolute number of emergent or new topographies and "hits self with body part" increased in prevalence and severity at 36 months, which is the primary topography reported in the literature early on (Steenfeldt-Kristensen et al., 2020).

Predicting SIB
The logistic regression results indicated that presence of stereotypy, lower intellectual functioning, and hyper and hypo-responsiveness at age 12 months significantly predicted SIB at 36 months.Higher SIB at 12 months was also statistically significant, however, odds ratios indicated that children with SIB at age 12 months were less likely to engage in SIB at 36 months.This particular finding was surprising given that we previously found higher SIB at 12 months predicted SIB status at age 24 months (Dimian et al., 2017).A decelerating trend in SIB overall for most of the children may have contributed to this disparate finding.but it could also be that proto-injurious forms of SIB at 12 months (i.e., less severe, non-tissue damaging SIB) are not as predictive of more severe or persistent SIB at 3 years.We adopted a liberal approach to our definition of SIB in this analysis (i.e., any endorsement of SIB) given that the phenomena are poorly understood in very young children.A more stringent definition, perhaps based on direct observation versus parent report, as well as accounting for SIB trajectories over longer periods (e.g., into school-age) may provide valuable information about who goes on to develop clinically significant SIB that requires treatment.
SIB endorsement and severity at 36 months was significantly associated with each sensory response evaluated.Hypo-responsivity yields the strongest association and was a slightly stronger individual predictor of SIB at 36 months when entered in the model with hyper and sensory seeking response patterns.Hypo-responsiveness has been found to be associated with being less likely to be able to adapt to an environment (i.e., it takes them longer) among children with ASD (e.g., Brock et al., 2012).This pattern of sensory responsiveness may affect how a child interacts with their environment, and may be indicative of more rigid or ritualistic behavioral patterns.SIB could be more likely to occur for these children in particular when transitions or new settings/people are introduced.Likewise, hyper-responsiveness involves over-reactivity or over focus on stimuli in the environment that may contribute to behavior-environment interaction patterns early on.In general, these data point to potential intervention targets early in life that may entail teaching adaptive/functional skills as well as coping skills that address inflexible behavior that may be protective and increase resilience against SIB engagement.
Stereotypy was the strongest predictor of SIB at 36 months and is consistent with other empirical work and hypotheses related to the emergence of SIB.Previous research provides some evidence among disparate disability samples that stereotyped behavior is highly correlated with SIB and could be a precursor to the development of SIB for some individuals.Barnard-Brak et al. (2015) found that stereotyped behavior predicted SIB for subgroup of children and adults with intellectual disabilities in a large cross sectional sample.Younger age and having multiple topographies of SIB and stereotypy characterized the group where stereotypy significantly predicted SIB.Similar to the current study, Rojahn and colleagues (2016) conducted a longitudinal study with 160 infants and toddlers at risk for developmental delay and evaluated the relationship between stereotypy and SIB over across three time points in a year.The best fitting model showed that earlier stereotyped behaviors predicted later SIB.Similarly, Richman et al. (2013) found that impulsivity and severity of stereotyped behavior were the strongest predictors of SIB in a sample of individuals with ASD 3 to 35 years old.Overall, our results do suggest that stereotypy that is maintained into late infancy and early toddlerhood may be a developmental precursor to SIB.

Limitations
The current study has several limitations.The analysis primarily included parent-report measures; utilizing direct observations of repetitive behavior, stereotypy, sensory responsivity, and SIB would be valuable in substantiating the findings of this study.Another limitation is that we aggregated the SIB outcome and used a dichotomous variable in the analysis.Any endorsement of SIB on the RBS-R was utilized to categorize and assess outcomes in this study, which in effect considers all variety of SIB to be equivalent.The SIB data reported in this study indicated that form and severity across the sample were variable, and therefore we may be missing subtypes or more precise classifications based on form or frequency of SIB.Additionally, the RBS-R was used and was created based on adults with intellectual and developmental disabilities.More developmentally appropriate measures (e.g., the Repetitive Behavior Scale for Early Childhood; Wolff et al., 2016) may provide more accurate information on self-injurious (or proto-self-injurious) and repetitive behaviors.Also, we have no measure of function or reason why SIB is occurring.Finally, our study is limited in terms of sample demographics, as a more diverse sample of children is necessary to increase the external validity of our findings to the population of young children in the United States.

Future research
It is imperative that this line of inquiry continue with additional research so that we can better understand the pathogenesis of SIB and predict who may be at risk for developing chronic SIB.Our hope is that if we can identify who is at the highest risk for clinically significant SIB, we can intervene earlier and possibly prevent its emergence.The results of this study suggest some potential targets for further study, and provide a more comprehensive model of SIB risk that incorporates sensory responsiveness.Four decades of research on SIB within developmental disabilities suggest, however, that there may be multiple developmental pathways leading to SIB, especially within a heterogeneous disorder like autism.There are several competing hypotheses that include both biological and behavioral factors within the extant literature (e.g., McClintock et al., 2003).Additionally, the field of applied behavior analysis has provided rigorous evidence for operant maintenance of SIB, which has led to effective but resource intensive behavioral treatment.At issue is that not all SIB is maintained by operant/ social reinforcement (i.e., automatically maintained SIB), and what causes the emergence of SIB may substantively differ from what causes it to persist and maintain over time.The models created to date, including those in the present study, account for only some of the variance in later SIB, suggesting that a more precise or tailored risk model (and ultimately a risk algorithm) incorporating both biological and behavioral characteristics may be needed.Ultimately, we may need to examine the mechanisms that contribute to the emergence of SIB in terms of equifinality (Cicchetti & Rogosch, 1996) and generate multiple models that link to specific malleable intervention targets.

ACKNOWLEGMENTS
This work was supported by grants from the National Institutes of Health under awards R01MH116961, P30HD03110, and R01HD05574; Autism Speaks, and the Simons Foundation.

CONFLICT OF INTEREST STATEMENT
We have no conflicts of interests to disclose.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are openly available in NDA at.https://nda.nih.gov/edit_collection.html?id=19.

ETHICS STATEMENT
All study procedures were approved by the respective institutional review boards at each clinical data collection site and carried out in accordance with the Declaration of Helsinki, with informed written consent obtained from parents or legal guardians.

F
I G U R E 1 SIB topography endorsement by time point for persistent cases (SIB at all three time points; n = 9).The lower panel is SIB topography endorsement by time point for transient cases (SIB at one or two time points; n = 75).The size of the bubble indicates average severity endorsed (0 = not a problem; 3 = severe problem.)

F
I G U R E 2 SIB topography endorsement by time point for those who received a diagnosis of ASD at 24 or 36 months (n = 41).The lower panel is SIB topography endorsement by time point for those who did not receive a diagnosis of ASD at 24 or 36 months (n = 108).The size of the bubble indicates average severity endorsed (0 = not a problem; 3 = severe problem.) Demographic characteristics.
T A B L E 1Note: MSEL = Mullen scales of early learning; Standard scores are reported for the MSEL and Vineland (M = 100) and mean and standard deviations are presented.
At 12 months, SIB endorsement and severity scores were significantly associated with hyperresponsivity (r = 0.31, ).Odds of SIB increased by 5% for each point decrease in Mullen composite score (OR = 0.95, 95% CI Unadjusted odds ratios with 95% confidence intervals between psychosocial characteristics at 12 months and SIB at 36 months.