Bullying Victimization of Children with Mental, Emotional, and Developmental or Behavioral (MEDB) Disorders in the United States

Abstract

Though preventable, children with disabilities have a high risk of victimization, contributing to worsening health conditions. Hence, this study examined the exposure of school-age children with mental, emotional, developmental, or behavioral (MEDB) disorder to bully victimization. This study used the 2018 National Survey of Children’s Health (NSCH) data of 23,494 children ages 5–17 to estimate multilevel logistic regression with fixed and random effects. Children’s health conditions were treated as level one variables, while family poverty level and neighborhood characteristics such as vandalism and presence/absence of recreational centers were treated as level two variables. The paper presents the prevalence of bullying victimization among children with at least one disorder (MDBB = 39.5%), anxiety (20.6%), depression (10.8%), ADD/ADHD (18.3%), behavioral problems (14.9%), learning disability (11.9%), Tourette syndrome (0.5%), developmental delay (10.1%), Autism spectrum disorder (4.6%), speech disorder (10.7), and intellectual disability (1.6%), respectively. Bullying victimization was positively associated with anxiety (AOR = 1.995, 95% CI = 1.634–2.436), depression (AOR = 2.688, 95% CI = 2.031–3.557), developmental delay (AOR = 1.804, 95% CI = 1.422–2.288), but inversely associated with Autism spectrum disorder (AOR = 0.614, 95% CI = 0.399–0.946). Neighborhood disorganization and poverty were also associated with bullying victimization. The NSCH data suggests that children with disabilities in the US had a higher prevalence rate of victimization. Consequently, effective bullying prevention strategies that can protect and improve children’s quality of life with special needs should be prioritized.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40653-021-00368-8.

Introduction

Bullying is a form of childhood adverse experience, social violence, and an essential public health issue (Center for Disease Control and Prevention, 2019). It has received enormous attention from researchers as well as public and private domains. Broadly defined, bullying is a form of aggressive behavior in which someone intentionally and repeatedly causes another person injury or discomfort in physical contact, words, or more subtle actions (Center for Disease Control and Prevention, 2019). Worldwide, one in five children experience bullying, and school-going children are particularly at risk. Overwhelming evidence shows that victims of bullying were at greater risk for depression, anxiety, sleep difficulties, lower academic achievement, and school drop out (Arseneault et al., 2010; Blake et al., 2012; Lebrun-Harris et al., 2019; PACER’s National Bullying Prevention Center, 2020; Swearer et al., 2001; Zablotsky et al., 2013; Zinner et al., 2012).

According to Healthy People 2020 Initiatives, the target prevalence rate of bullying is 17.9% among US adolescents (Healthy People Initiatives, 2020). However, meeting this target requires continuous research to keep track of the change based on reliable national data. Hence, this study contributes to this national commitment toward reducing bullying victimization and perpetration, particularly among vulnerable children and adolescents in the US. Following this section is the review of the prevalence of bullying victimization and disability and their risk factors. The ‘Method’ section presents data and analytical techniques followed by presentation and interpretation of results. The paper concludes by discussing the results, the study limitations, and suggested policies based on this study’s findings.

Literature Review: The Prevalence of Bullying and Disability

In medical geography literature, bullying is being referred to as a psychosocial insult exerted by external forces (Meade, 2014). Various typologies of bullying victimization, which can be perpetrated by close relatives or outsiders, had been researched and compiled (Salmon et al., 2018). Traditional forms of bullying include physical abuse and harassment such as race-based, gender-based, sexual orientation-based, disability-based, and physical appearance-based harassment. In the last five decades, bullying has taken a different dimension due to social technology advancements leading to cyber-victimization and perpetration (Code, 1971). Hence, research in cyberbullying and cyber-victimization has burgeoned (Antoniadou et al., 2019; Camerini et al., 2020; Holfeld et al., 2019; Kokkinos et al., 2016; Kokkinos & Antoniadou, 2019; Larrañaga et al., 2016; Musharraf et al., 2019; Sam et al., 2019). Though cyberbullying /cyber-victimization is a common platform for bullying nowadays, most national surveys are yet to collect information regarding cyberbullying. Hence, the analysis of bullying in this study is limited to the traditional form of bullying victimization.

Despite the long-range discussion and research on bullying and bullying victimization among the general population and at different strata, scholars have paid little attention to how both individual-level and neighborhood-level factors affect the risk of victimization among children with disabilities and within the same statistical models. As a result, there is an underlying question of whether the nature of bullying victimization is determined by individual, situational, or contextual factors. Furthermore, the geography of bullying victimization and children with disabilities has been less researched in the US. This is particularly important because of the varying characteristics of a place, including policies, socio-cultural and economic contexts that shape places. It will be very erroneous to depend on single statistical values at the national level, especially toward meeting the target of reducing the prevalence of bullying to 17% in the US by the end of 2020 (Healthy People Initiatives, 2020).

Due to different nationally representative self-report surveys providing a varying perception of bullying prevalence in the US, inconsistencies in rates abound in reports (Lebrun-Harris et al., 2019). In the general student population, it ranged between 16% to 23%. However, bullying prevalence rates ranged from 19% in students with specific learning disabilities to 35.3% among students with behavioral and emotional disorders (PACER’s National Bullying Prevention Center, 2020). About a quarter (33.9%) of students with autism, 24.3% of students with intellectual disabilities, 20.8% of students with health impairments face high bullying victimization (Rose & Monda-Amaya, 2012). Accordingly, several strategies to reduce the prevalence had been recommended in the literature. The strategies include promoting family environments that support healthy development, provide quality education early in life, strengthen youth’s skills, connect youth to caring adults and activities, create protective community environments, intervene to lessen harm and prevent future risk (Center for Disease Control and Prevention, 2019; Rose & Monda-Amaya, 2012). Nevertheless, bullying perpetration and victimization have not been reduced in the United States (Healthy People Initiatives, 2020), particularly among children and adolescents with disabilities.

Studies have found that various mental health conditions, behavioral disorders, and psychiatric disorders are predictors of bullying (Christensen et al., 2012; Rose & Gage, 2017; Rose & Monda-Amaya, 2012). In a US study, Lebrun-Harris et al. (2019) found that children aged 6–11 with internalizing problems had a 47% prevalence rate of bullying victimization; children with speech or language disorder experienced a 61% prevalence in bullying victimization, and those with behavioral or conduct problem experienced 41% prevalence rate. On the other hand, the rate of bullying victimization among adolescents aged 12–17 with internalized problems, behavioral or conduct problems were 52% and 69%, respectively. In Brazil, da Silva et al. (2020) found that bully-victims were nine times more likely to be depressed. Furthermore, the study also showed that depression was significantly correlated with all forms of bullying and victimization (physical, verbal, relational) among girls. However, only verbal and relational victimization was significantly correlated with depression among boys. Among Lebanese children, younger age, lower socioeconomic status, lower family income, and anxiety or disruptive behavior disorder were identified as predictors of bullying victimization (Halabi et al., 2018). The odds of being a victim of bullying were 2.95 for children with any disorder, 1.90 for externalizing disorder, and 3.53 for children with emotional disorders.

In the literature, there is evidence of gender disparity among students. A higher percentage of male than female students were likely to be physically bullied, whereas a higher percentage of females than males were subjects of rumors (18% vs. 9%) and more likely to be excluded from activities on purpose (7% vs. 4%) (PACER’s National Bullying Prevention Center, 2020). Nevertheless, children with neuropsychiatric disorders and other chronic health conditions are at higher risks of peer victimization, ostracization, and social neglect (Arseneault et al., 2010; Bejerot et al., 2013; Bifulco et al., 2014; Blake et al., 2012; Christensen et al., 2012; Rose & Gage, 2017; Rose & Monda-Amaya, 2012; Swearer et al., 2001; Zablotsky et al., 2013; Zinner et al., 2012).

Furthermore, several studies have shown the impact of poverty, neighborhood social, economic, physical environments on bullying victimization/perpetration in the literature. Compared to the 2009 children poverty rate of 20% (Yoshikawa et al., 2012), 14.4% of all children under 18 in the US were living below the official poverty measure in 2019. According to Yoshikawa et al. (2012), poverty is an important risk factor for many mental, emotional, and behavioral disorders of children and youths. Essentially, poverty is inherently linked to other co-factors such as prior generational low school attainment and teen parenting. Prior generational school attendants and teen parenting increase adolescents’ chances of raising their children in poverty. Similarly, education, achievement, and family structure in one generation can be determinants of family income poverty and then children’s health and development in the next generation. These factors could also predispose children from poor households to bullying victimization. For example, in a meta-analysis, Tippett and Wolke (2014) found a significant association between family SES and bullying victimization (OR = 1.54; 95% CI = 1.36, 1.74).

Evidence of neighborhood contextual factors has also been reported for adverse life experiences such as bullying victimization and mental health. Using a multiple mediational analysis, Choi et al. (2021) investigated the link between neighborhood disadvantage, childhood adversity, bullying victimization, and adolescent depression. The study showed that bullying victimization was a risk factor for social and emotional problems and found a direct and indirect effect of the neighborhood on mental health illness (depression) among adolescents. Furthermore, the study’s findings suggest that both neighborhood structural disadvantage and collective efficacy have direct impacts on adverse childhood experiences, bullying victimization, social-emotional development, and indirect impacts on adolescents’ depressive symptoms (Choi et al., 2021). Using structural equation modeling, a study found being bullied in early adolescence was most strongly predicted by having fewer close friends, higher family poverty, and living in neighborhoods with higher levels of disorder. However, the social disadvantage of a young person’s school did not impact being bullied in adulthood (D’Urso et al., 2020). Despite the enormous research conducted on bullying victimization and risk factors for many of the mental, emotional and behavioral disorders of children and youths, there is a lack of study to show the geographic distribution of bullying and MEBD in the US. In addition, based on the extant literature and the mixed results of association of bullying victimization and MEDB, it is incredibly essential to consistently examine factors contributing to the dynamics of exposure and vulnerability of bullying victimization among children with mental, emotional, developmental, and behavioral disorders to keep up with the ever-changing social and physical environments.

The Current Study

The purpose of the present study is two-fold. First, it investigates the effects of individual-level (disabilities) and neighborhood-level factors on bullying victimization prevalence. Second, it examines these relationships across the 50 states and the District of Columbia (DC) and stratifies by age categories of children in the United States to assess whether the effects are general or not. To that end, both individual-level and neighborhood-level factors should be associated with traditional forms of bullying (physical, verbal, relational) that occur in and the off-school environment—coded as a binary variable. The central question answered in the study is whether bullying victimization is associated with the ten selected childhood disorders in the country to keep up with the existing literature. Based on extant literature, the study hypothesized that children’s health status (MEDB), social position, and physical environment would account for exposure to bullying victimization. Results from this study are essential for policies toward reducing violence toward children with Special Needs.

Methods

Data for this study were obtained from the 2018 National Survey of Children’s Health (NSCH). The project was supported by the Health Resources and Services Administration (HRSA) of the US Department of Health and Human Services (HHS) and the National Maternal and Child Health Data Resource Initiative (Child and Adolescent Health Measurement Initiative (CAHMI), 2020).

Procedure for Participants Selection and Inclusion

To overcome the limitations of web-based surveys such as internet access and computer literacy, the 2018 National Survey of Children’s Health was administered online and by mail. Randomly selected addresses from households across the US were mailed instructions to access the survey online; some addresses also received a paper version of the screening questionnaire. After two reminder letters and postcard reminders to complete the survey by web, those households who had not accessed the online survey were mailed a paper screening questionnaire. To increase sampling efficiency, administrative data were used to determine addresses that were more likely to be households with children aged 0–17 years. The survey over-sampled children 0–5 years and children with special health care needs (CHSCN). This means that households with two or more children, CSHCN and children 0–5 years old, had a higher probability of being selected as compared to other children in the household (U. S. Census Bureau, 2019).

NSCH data were collected using a two-stage paper survey instrument and a single-stage web-based survey instrument from a sample of 176,000 households in the Census Master Address File allocated across the 50 states and DC. The sample was then stratified by the state of residence and a child-presence indicator that allowed the Census Bureau to oversample households that were more likely to have children using a screener (U. S. Census Bureau, 2019). The details of the sampling method have been published online (Child and Adolescent Health Measurement Initiative (CAHMI), 2020; U. S. Census Bureau, 2019). Briefly, the selection process took a step-by-step approach (see Fig. S1). At the first stage, participants were sent a mailed invitation to fill out the online survey. Some participants were also mailed a paper screening questionnaire, either with the initial invitation or after reminder letters.1 At the second stage, both paper and online respondents were initially asked if one or more children ages 0–17 living in the household and households that qualify are prompted to continue to the next stage by filling the initial Screener. Respondents filled out an initial Screener with the age and sex of all children in the household. Online respondents were prompted to submit Screener and move directly to the Topical Questionnaire and paper respondents mailed back their screener form. Next, one child from each household was randomly selected to be the subject of the main Topical Questionnaire, after which online respondents were automatically directed to, and paper respondents received (through the mail) an age-appropriate Topical Questionnaire. Depending on the age of the child, respondents complete one of the three surveys based on age groups: 0–5, 6–11, and 12–17. For more details of the survey source and accuracy statement, U. S Census Bureau provides useful documents online.2 To account for children with disabilities, children with Special Health Care Needs were oversampled (at 80%) in order to allow robust estimates of this critical population.

The overall response rate for the 2018 NSCH was 43.1% out of the 71,000 screener questionnaires that were completed between June 2018 and January 2019 of those 38,140 households eligible for follow-up. Of those eligible households, 30,530 completed a topical interview for children age 0–17 years (U. S. Census Bureau, 2019). Because this study focused on the school-age group (5–17 years), only 23,494 children were included and formed the study sample size. The sample size thus varies depending on households with children with disabilities or disorders. Data for this analysis was obtained with permission from the data resources center for child and adolescent health (https://www.childhealthdata.org/).

The NCHS Research Ethics Review Board approved all study procedures and modifications. To protect the confidentiality of individual respondents and children, responses for certain variables were collapsed or suppressed. The Census Bureau, Health Resources and Sevices Administration, and Department of Maternal and Child Health Bureau take extraordinary measures to ensure that survey subjects’ identities are protected. All direct identifiers and characteristics that might lead to identification have been omitted from the data files (Health Resources & Services Administration, 2017).

Measures and Operationalization of Variables

Dependent Variable: Bullying victimization is the dependent variable based on a question which asked: “During the past 12 months, how often was this child bullied, picked on, or excluded by other children, age 6-17 years?” Five response options were given based on the frequency of exposure: Never (in the past 12 months), 1–2 times (in the past 12 months), 1–2 times per month, 1–2 times per week, or almost every day. These options were recoded to ‘1’ if the child had been bullied in the last 12 months and to ‘0’ if not.

Multilevel variables: There were 11 main variables of interest in this study. Ten variables captured children screened for one or more mental, emotional, developmental, or behavioral disorders (MEDB), and one index variable represented households with children with at least one of the ten MEDB. The single MEDB variable was developed based on this question: “Does this child have current mental, emotional, developmental or behavioral (MEDB) conditions from a list of 27 conditions, or the CSHCN Screener item on ongoing emotional, developmental or behavioral problems?” Parents answered yes or no for all the 11 outcome variables, and the ‘no’ was used as the reference category for all of them.

Residential and neighborhood variables used include mold in the home, recreational center presence, poorly kept or rundown housing, and vandalism as part of level two variables. Rundown features are evidence of structural deterioration of property and signs of social inequity and environmental injustice. Hence, we hypothesized that children living under such conditions might be more exposed to bullying victimization. Conversely, the presence (reference: absence) of a recreational center/park is an indicator of neighborhood quality and has implications on health. Because literature suggests that more males than females have a higher risk of being bullied, the study stratified the model by gender and included age as a scale variable. The state of residence was also used as a level two identifier and family poverty level (FPL) as a level two predictor of the outcomes in the multivariate statistical models.

Statistical Analysis

All Analyses were conducted in SPSS v20. Data were cleaned, sorted, and examined for evidence of multicollinearity, which might bias the results of the analyses among variables. The Pearson Chi-Square test was used to determine the prevalence rate of bullying. To achieve the most parsimony models, only variables that showed significant associations with bullying victimization were entered in the multivariate logistic model using the stepwise method (Tabachnick & Fidell, 2007). Next, an age-stratified multilevel logistic regression was developed to examine the disparity in exposure to bullying victimization between young children (6–11) and transition adolescents (12–17). At first, constant model was estimated to determine the within-subject variation and the second model included level one (individual/household) variables. The third model included neighborhood variables (park/recreational center, rundown housing, vandalism). Lastly, the federal poverty level (FPL) was finally included as a level two variable. The NSCH generated survey weight (FWC) was applied to reflect the population of children and youth ages 0–17 in the US and not the family or household. Results were interpreted in odds ratios (ORs) along with their 95% confidence intervals (CI).

Results

The median age was 12 years (M = 11.65, SD =3.775, 95% CI 12.07–12.16). Table 1 presents the prevalence of bullying victimization and all the ten MEDB conditions by age group. The prevalence of bullying victimization was 64.4% among younger children and 44% among adolescents. Based on the composite index of mental, emotional, behavioral disorders (MEDB), adolescents (41.7%) have a higher rate more than younger children (30.3%). More notably, adolescents with anxiety and depression reported a higher prevalence rate of bullying victimization than younger children. The prevalence was twice as high in adolescents with Tourette syndrome than younger ones. Table 1 also shows that bullying victimization was highest among children with ADD/ADHD and anxiety but lowest among children screened for intellectual disability and Tourette syndrome.

Table 1.

The Prevalence Rate of Bullying Victimization Among Children with Disabilities in the United States in 2018

	Age-Groups
Variables	6–11 years [n (%)]	12–17 years [n (%)]	Total, n (%)
MEDBa	1699 (30.3)	2280 (41.7)	3979 (35.9)
Tourette Syndrome	18 (0.30)	36 (0.7)	54 (0.5)
Anxiety	811 (14.5)	1461 (26.8)	2272 (20.6)
Depression	226 (4)	963 (17.7)	1189 (10.8)
Behavior Problems	841 (15.1)	808 (14.8)	1649 (14.9)
Developmental Delay	550 (9.8)	562(10.3)	1112 (10.10)
Intellectual Disability	69 (1.2)	113 (2.1)	182 (1.6)
Speech Disorder	657 (11.7)	525 (9.6)	1182(10.7)
Learning Disability	542 (9.7)	773 (14.2)	1315 (11.9)
Autism ASD	211 (3.8)	298 (5.5)	509 (4.6)
ADD/ADHD	841 (15.2)	1163 (21.5)	2004 (18.3)
Bullying victimization b	5607 (64.4)	5473 (44)	11,080 (51)

^a Children experience one of the 10 MEDB conditions

^b Prevalence of bullying victimization by age group

The prevalence of MEDB (χ2 = 90.24, p < 0.001) and bullying victimization (χ2 = 268.79, p < 0.001) significantly vary across the 50 US states and the District of Columbia, respectively. Figure 1 shows the spatial distribution for both variables—i.e., bullying victimization and MEDB. Seven states, including Idaho, Utah, North Dakota, Wisconsin, Oklahoma, Arkansas, and North Carolina, were classified in the highest quantile (2.3%–2.7%). For MEDB prevalence rates, six states were classified in the highest quantile, and they include Louisiana, Mississippi, Arkansas, Kentucky and West Virginia, and Vermont— in the sunbelt region. Geographic information system detects significant cluster of MEDB in the southern states (Getis-Ord G = 0.020, Z = 2.261, p = 0.023). Given the z-score of 2.26, there is a less than 5% chance that this high-clustered pattern could result from a random chance. The prevalence of bullying victimization among children screened for at least one of the ten MEDB conditions was 35.9%, and Fig. 2 shows the hotspot of the prevalence rates prominient among the 48 contigous states except California, Washington, Oregon and Arizona with statistical confidence ranging between 90% in mostly northeastern states (also in Nevada and New Mexico) and 99% in midwestern and southern states. Hawaii falls in the cold region which indicates the prevalence of MEDB in children and adolescent was signifcantly lower. There was no significant hot or cold spot detected for bullying victimization.

Fig. 1.

Geographic distribution of bullying victimization and MEDB— mental, emotion, developmental, or behavior in the US Source: National Survey of Children Health (NSCH)

Fig. 2.

Hotspot of mental, emotional, developmental, or behavior in children in 2018

Association with MEDB Conditions: Results from the Multilevel Logistic Regression

Table 2 presents the systematic modeling of the association between bullying victimization and different childhood disorders and disabilities, mostly in the expected direction. Model 1 only assessed the association between bullying victimization and all the ten individual MEDB conditions and the MEDB-indicator. In models 1 and 2, children and adolescents screened for a single MEDB indicator, anxiety, depression, behavioral problem, a developmental problem, speech disorder, and ADD/ADHD were more likely to be bullied (Table 2). In model 2, bullying victimization was positively associated with neighborhoods characterized by vandalism (AOR = 2.168, 95% CI = 1.522–3.09). Lastly, age serves as a protective factor from bullying because it was inversely associated with bullying victimization (β = −0.13, OR = 0.87, 95% CI 0.86–0.88).

Table 2.

Multilevel Logistic Regression for Bullying and Childhood Disorders

Model Term	Model 1: Unadjusted				Model 1 + Neighborhood, Age				Full: + Poverty variable
	B	OR	95% CI		B	AOR	95% CI		B	AOR	95% CI
			LB	UB			LB	UB			LB	UB
Intercept	−0.298***	0.742	0.682	0.809	1.219***	3.383	2.778	4.121	0.983***	2.671	2.096	3.404
Tourette syndrome	0.645	1.906	0.83	4.38	0.813	2.255	0.75	6.783	0.853	2.346	0.821	6.703
Anxiety	0.641***	1.899	1.598	2.256	0.713***	2.04	1.651	2.522	0.691***	1.995	1.634	2.436
Depression	0.58***	1.787	1.348	2.368	0.954***	2.596	1.985	3.393	0.989***	2.688	2.031	3.557
Behavioral problem	0.661***	1.936	1.524	2.461	0.565***	1.759	1.397	2.215	0.59***	1.804	1.422	2.288
Developmental problem	0.392**	1.48	1.185	1.849	0.277**	1.32	1.073	1.623	0.287*	1.332	1.071	1.656
Intellectual disability	−0.142	0.868	0.557	1.352	−0.127	0.881	0.557	1.393	−0.097	0.908	0.565	1.459
Speech disorder	0.373**	1.453	1.145	1.843	0.234	1.264	0.996	1.604	0.195	1.215	0.951	1.552
Learning disability	−0.198	0.821	0.66	1.02	−0.1	0.905	0.742	1.104	−0.054	0.948	0.778	1.154
Autism ASD	−0.54*	0.583	0.378	0.9	−0.508	0.602	0.38	0.954	−0.488*	0.614	0.399	0.946
ADD/ADHD	0.098	1.103	0.783	1.553	0.288	1.334	0.932	1.908	0.276	1.318	0.906	1.918
MEDB10	0.496***	1.643	1.367	1.975	0.447***	1.564	1.314	1.861	0.448***	1.565	1.311	1.868
Child’s age					−0.138***	0.871	0.857	0.886	−0.138***	0.871	0.856	0.886
Mother’s age					−0.004*	0.996	0.993	1	−0.005c’	0.995	0.992	0.999
Recreational center					0.024	1.025	0.888	1.183	−0.001	0.999	0.875	1.141
Poorly kept/Rundown					−0.106	0.899	0.735	1.101	−0.069	0.933	0.773	1.127
Vandalism					0.774***	2.168	1.522	3.09	0.83***	2.294	1.589	3.312
Litter/garbage					0.034	0.967	0.848	1.102	0.027	1.027	0.906	1.164
Female (male ref.)					0.237***	1.267	1.126	1.425	0.23***	1.258	1.119	1.415
FPL:0%–99% ref.
FPL:100–199%									0.107	1.112	0.794	1.559
FPL:299%–399%									0.353**	1.424	1.166	1.738
FPL: 400%									0.452***	1.572	1.333	1.853

*** p < 0.001; ** p < 0.01; * p < 0.05

Note:

Model 1: Composed of only the disorder variables

Model 2: Variables in model 1 + child’s age + mother’s age + neighborhood variables

Model 3: Model 2 + Federal poverty level (FPL)

In the full model, the association between bullying victimization and the seven childhood conditions in the model remained consistently significant except for developmental delay after adjusting for other variables. Female sex was positively and significantly associated with bullying victimization (OR = 1.29, 95% CI 1.22–1.37) after adjusting for all other variables. However, when the model was flipped between males and females (female gender was used as the reference group), males had a negative and significant association with bullying victimization (data not presented). Bullying victimization was also significantly associated with each category of the family poverty level (FPL) compared to the lowest reference category (0%–99%). Lastly, the random effect estimation shows a significant variation in bullying victimization for different models across the 50 contiguous states and DC (see Table 3). Table 4 presents the random effect parameters. The table indicates that bullying victimization, as a social problem, varies across the United States.

Table 3.

Empirical Best Linear Unbiased Prediction of Bullying Victimization ^a

US States	Model 1				Model 2				Model 3
	Prediction	Sig.	95% CI		Prediction	Sig.	95% CI		Prediction	Sig.	95% CI
			Lower	Upper			Lower	Upper			Lower	Upper
Alabama	−0.14	0.000	−0.214	−0.065	−0.139	0.001	−0.222	−0.056	−0.099	0.024	−0.185	−0.013
Alaska	0.152	0.000	0.079	0.226	0.12	0.004	0.038	0.202	0.105	0.010	0.025	0.185
Arizona	−0.092	0.014	−0.165	−0.019	−0.131	0.001	−0.207	−0.055	−0.122	0.002	−0.198	−0.046
Arkansas	−0.014	0.706	−0.088	0.06	0.027	0.499	−0.052	0.106	0.058	0.175	−0.026	0.141
California	−0.068	0.078	−0.144	0.008	−0.098	0.027	−0.184	−0.011	−0.082	0.058	−0.166	0.003
Colorado	0.195	0.000	0.122	0.268	0.205	0.000	0.128	0.282	0.212	0.000	0.134	0.29
Connecticut	−0.032	0.390	−0.105	0.041	0.01	0.813	−0.07	0.089	−0.035	0.396	−0.117	0.046
Delaware	−0.292	0.000	−0.365	−0.22	−0.395	0.000	−0.471	−0.32	−0.382	0.000	−0.458	−0.305
DC	−0.464	0.000	−0.542	−0.387	−0.608	0.000	−0.722	−0.495	−0.598	0.000	−0.72	−0.477
Florida	−0.33	0.000	−0.405	−0.256	−0.301	0.000	−0.378	−0.225	−0.262	0.000	−0.34	−0.184
Georgia	−0.079	0.038	−0.154	−0.004	−0.098	0.012	−0.174	−0.022	−0.067	0.091	−0.145	0.011
Hawaii	−0.322	0.000	−0.4	−0.243	−0.353	0.000	−0.444	−0.262	−0.396	0.000	−0.49	−0.302
Idaho	0.562	0.000	0.487	0.637	0.662	0.000	0.571	0.752	0.664	0.000	0.563	0.765
Illinois	−0.336	0.000	−0.409	−0.262	−0.241	0.000	−0.324	−0.158	−0.239	0.000	−0.322	−0.157
Indiana	−0.192	0.000	−0.265	−0.119	−0.125	0.001	−0.201	−0.05	−0.111	0.004	−0.187	−0.035
Iowa	0.184	0.000	0.109	0.258	0.2	0.000	0.121	0.279	0.198	0.000	0.119	0.278
Kansas	−0.03	0.427	−0.103	0.044	−0.104	0.005	−0.177	−0.031	−0.116	0.002	−0.189	−0.043
Kentucky	0.006	0.872	−0.066	0.078	−0.002	0.958	−0.079	0.075	0.019	0.647	−0.061	0.098
Louisiana	−0.148	0.000	−0.224	−0.072	−0.157	0.000	−0.236	−0.078	−0.118	0.005	−0.201	−0.035
Maine	0.096	0.012	0.021	0.17	0.067	0.089	−0.01	0.144	0.028	0.466	−0.048	0.104
Maryland	−0.226	0.000	−0.299	−0.153	−0.227	0.000	−0.307	−0.148	−0.254	0.000	−0.333	−0.176
Massachusetts	−0.152	0.000	−0.223	−0.08	−0.091	0.022	−0.169	−0.013	−0.141	0.001	−0.223	−0.059
Michigan	0.005	0.891	−0.067	0.077	0.051	0.195	−0.026	0.128	0.056	0.150	−0.02	0.133
Minnesota	0.177	0.000	0.104	0.249	0.195	0.000	0.117	0.272	0.149	0.000	0.07	0.227
Mississippi	−0.06	0.114	−0.135	0.015	−0.073	0.095	−0.159	0.013	−0.007	0.877	−0.102	0.087
Missouri	0.03	0.414	−0.043	0.103	0.045	0.240	−0.03	0.121	0.05	0.193	−0.025	0.126
Montana	0.331	0.000	0.258	0.404	0.312	0.000	0.233	0.391	0.31	0.000	0.226	0.393
Nebraska	0.136	0.000	0.061	0.21	0.164	0.000	0.085	0.244	0.166	0.000	0.086	0.247
Nevada	−0.347	0.000	−0.42	−0.273	−0.398	0.000	−0.48	−0.316	−0.383	0.000	−0.463	−0.303
New Hampshire	−0.065	0.079	−0.138	0.008	0.019	0.640	−0.059	0.097	−0.027	0.499	−0.105	0.051
New Jersey	−0.432	0.000	−0.505	−0.359	−0.393	0.000	−0.47	−0.316	−0.412	0.000	−0.49	−0.333
New Mexico	0.331	0.000	0.257	0.406	0.195	0.000	0.118	0.272	0.225	0.000	0.148	0.303
New York	−0.284	0.000	−0.357	−0.21	−0.238	0.000	−0.326	−0.15	−0.25	0.000	−0.337	−0.162
North Carolina	−0.017	0.646	−0.091	0.056	0.003	0.931	−0.072	0.079	0.015	0.698	−0.062	0.093
North Dakota	0.241	0.000	0.168	0.315	0.167	0.000	0.092	0.241	0.142	0.000	0.068	0.216
Ohio	−0.166	0.000	−0.239	−0.094	−0.194	0.000	−0.27	−0.118	−0.192	0.000	−0.268	−0.117
Oklahoma	0.102	0.006	0.03	0.175	0.157	0.000	0.075	0.24	0.185	0.000	0.101	0.27
Oregon	0.291	0.000	0.217	0.365	0.27	0.000	0.193	0.348	0.263	0.000	0.185	0.34
Pennsylvania	0.056	0.146	−0.02	0.132	0.044	0.271	−0.035	0.123	0.037	0.358	−0.042	0.116
Rhode Island	−0.221	0.000	−0.296	−0.146	−0.267	0.000	−0.352	−0.182	−0.304	0.000	−0.395	−0.214
South Carolina	−0.177	0.000	−0.251	−0.103	−0.143	0.000	−0.221	−0.066	−0.121	0.003	−0.199	−0.042
South Dakota	0.435	0.000	0.359	0.511	0.454	0.000	0.376	0.532	0.463	0.000	0.386	0.539
Tennessee	−0.105	0.004	−0.177	−0.033	−0.13	0.001	−0.205	−0.055	−0.102	0.011	−0.18	−0.023
Texas	−0.296	0.000	−0.37	−0.223	−0.256	0.000	−0.333	−0.178	−0.238	0.000	−0.319	−0.157
Utah	0.661	0.000	0.586	0.736	0.709	0.000	0.627	0.792	0.709	0.000	0.619	0.799
Vermont	−0.097	0.011	−0.171	−0.022	−0.072	0.080	−0.152	0.009	−0.116	0.003	−0.192	−0.039
Virginia	0.102	0.009	0.025	0.179	0.103	0.011	0.024	0.183	0.082	0.044	0.002	0.161
Washington	0.223	0.000	0.151	0.295	0.173	0.000	0.098	0.248	0.144	0.000	0.07	0.217
West Virginia	−0.189	0.000	−0.264	−0.114	−0.201	0.000	−0.28	−0.121	−0.17	0.000	−0.252	−0.088
Wisconsin	0.367	0.000	0.294	0.441	0.389	0.000	0.311	0.466	0.387	0.000	0.311	0.463
Wyoming	0.689	0.000	0.612	0.767	0.694	0.000	0.612	0.776	0.679	0.000	0.597	0.761

^aIntercept Only Models for the 51 States in the US

Note:

Model 1: Composed of only the disorder variables

Model 2: Variables in model 1 + child’s age + mother’s age + neighborhood variables

Model 3: Model 2 + Federal poverty level (FPL)

Table 4.

Age-Stratified Sensitivity Analysis: Association Between Bullying Victimization and Childhood Disorders ^a

Model Term	Children: 5–11 Years					Adolescent: 12–17 Years
	B	AOR	95% CI		B	AOR	95% CI
			Lower	Upper			Lower	Upper
Intercept	−0.023	0.977	0.897	1.064	−0.999***	0.368	0.337	0.402
Tourette syndrome	−0.268***	0.765	0.746	0.785	1.264***	3.541	3.469	3.615
Anxiety	0.835***	2.304	2.293	2.314	0.541***	1.719	1.712	1.725
Depression	0.94***	2.56	2.534	2.586	0.969***	2.635	2.624	2.645
Behavioral problem	0.654***	1.923	1.915	1.931	0.472***	1.604	1.597	1.61
Developmental disability	0.205***	1.228	1.222	1.234	0.475***	1.608	1.6	1.617
Intellectual disability	−0.039***	0.961	0.952	0.971	−0.304***	0.738	0.731	0.744
Speech disability	0.18***	1.198	1.193	1.202	0.291***	1.338	1.332	1.344
Learning disability	−0.313***	0.731	0.728	0.735	0.106***	1.112	1.108	1.117
Autism ASD	−0.684***	0.504	0.502	0.507	−0.186***	0.831	0.825	0.836
ADD/ADHD	0.393***	1.481	1.475	1.487	0.021***	1.021	1.018	1.025
MEDB10 (composite)	0.392***	1.481	1.475	1.486	0.569***	1.766	1.76	1.772

*** p < 0.001; AOR adjusted odd ratios

^a Controls for mother’s age, neighborhood characteristics, and FPL

Age-Stratified Model

Bullying victimization was associated with all the 11 health outcomes in the age-stratified model (Table 4). There was a different dimension of association observed between the two age groups for children with Tourette syndrome and learning disability. The dimension of associations was consistent for children with anxiety, depression, behavioral problem, developmental disability, speech disability, autism ASD, and ADD/ADHD (Table 4). Overall, the composite variable of disability (MEBD) also shows a positive association with bullying victimization, but the odds ratio was significantly higher among adolescents (AOR = 1.766, 95% CI = 1.760–1.772) than children (AOR = 1.481, 95% CI = 1.475–1.487). All models controlled for neighborhood factors and mother’s demography.

Discussion and Conclusions

Meeting the target proposed by Healthy People 2020 Initiatives demands commitment at different levels, which could be achieved by multidisciplinary research. This study has presented the prevalence of bullying victimization among school-age children and adolescents based on the 2018 National Survey of Children’s Health data of 23,494 children ages 5–17 years in the 50 states and DC of the US. It also investigated the association between bullying victimization and childhood disorders at the individual level and neighborhood level. It should be noted that the overall prevalence rate of bullying victimization found in this study was 51%, higher than the rates (22.7%) reported in a 2016 national survey in the US by Lebrun-Harris et al. (2019), and 19.74% reported from trend analysis of the national Youth Risk Behavior Survey between 2011 and 2017 (Li et al., 2020). The NSCH data showed that the prevalence was statistically higher among the younger age group (5–11) than the older age group (12–17). The younger age group in this study corresponds to students in Elementary through Middle schools who are at higher risk of peer victimization (Son et al., 2012, 2014).

Findings on the association between bullying and disabilities among children with disabilities are consistent with other studies (Rose & Gage, 2017). Here, in this current study, we found a high prevalence rate of bullying among children screened for at least one of the ten mental health conditions—MEDB. Compared to the general children population, children with disabilities are of particular concern regarding bullying victimization. Similar to the findings from this study, Rose and Espelage (2012) reported that 35.3% of students with behavioral and emotional disorders, 33.9% of students with autism, 24.3% of students with intellectual disabilities, 20.8% of students with health impairments, and 19% of students with specific learning disabilities face high levels of bullying victimization. However, the rates for the individual disorder were lower than previously reported rates among students with disabilities. In line with the rate reported by Rose and Gage (2017) in the US and elsewhere (Halabi et al., 2018), students with internalizing problems and emotional disorders had the highest rate of bullying victimization. However, the prevalence of victimization of students with intellectual disability (ID = 1.6%) was considerably lower than the 62.2% earlier reported in the US by Christensen et al. (2012). The interpretation of the discrepancies in the prevalence of victimization among children with disabilities has different angles. Firstly, one could argue that children in this group are gaining more confidence to overcome peer victimization. Second, due to different nationally representative self-report surveys have presented a varying perception of the prevalence of bullying in the US (Lebrun-Harris et al., 2019), and this holds for victimization among children with disabilities. Lastly, because this study is based on parent reports of children screened for ID, parents may be unaware of their child being bullied, and because of the child’s intellectual disability, they may not be able to communicate this to their parents. Additionally, if bullying is occurring via text, social media, and other cyber platforms, parents may be unaware as less obvious than ‘traditional bullying.’

The multivariate analysis results showed that children with behavioral and internalizing problems have greater odds of being victimized. Consistent with the findings from longitudinal data analysis (Rose & Gage, 2017), the odds of bullying victimization was particularly highest among children with behavioral problems and lowered among those with attention deficit disorder. In the age-stratified model, the odds of being a victim of bullying for younger children and adolescents with Autism/ASD were significantly lower, which deflects from what was previously reported (Rose & Gage, 2017). The plausible explanation is that these groups may not report to their parents when bullied due to age or disabilities.

Neighborhood and family socioeconomic factors were found to be associated with higher odds of bullying victimization. Children and adolescents living in neighborhoods characterized by a high rate of vandalism and dilapidated built environment were less likely to be victimized. This was contrary to expectation; it was expected that neighborhood disordered would serve as a risk factor, but instead, it was a protective factor to bullying victimization. It thus appears that children with a high level of neighborhood disorder are more likely to be bullied, probably because of a strong social network. According to Holt et al. (2014), the “presence of social disorganization and disorder can increase the likelihood that an individual will have low levels of self-control” (p.349). Low levels of self-control may prompt children in disorganized neighborhoods to quickly fight off bullies. The disorganized neighborhood, therefore, serves as an armor of protection from bully perpetrators. Although few studies have examined the influence of neighborhood effect on exposure to bullying and bullying victimization, this niche of research has been minimally investigated and, therefore, should be researched further.

The association of SES with victimization has been established (Piotrowska et al., 2015; Tippett & Wolke, 2014). The odds of victimization were more significant in children with FPL higher than 200%. From a medical geography perspective, this evidence suggests the potential effect of social class and the influence of poverty syndrome as a risk factor of social insults such as bullying victimization (Tippett & Wolke, 2014). Hence, this study adds to the cumulative evidence of low socioeconomic position and bullying victimization among children and adolescents. In line with the literature (Lebrun-Harris et al., 2019), this current study equally found that female children had a higher risk of bullying victimization. Additional investigation of the association may suggest that male children were less likely to experience bullying indicating the influence of masculinity in bullying exposure and prevention. However, a meta-analysis did not find a significant difference between male and female gender exposure to bullying victimization for both traditional and cyberbullying (Jadambaa et al., 2019).

Despite the presence of antibullying laws and policies across states in the US, the prevalence of bullying victimization persists (Li et al., 2020). Compared to other states with a significantly low rate of bullying of children identified with mental and behavioral disorders, a high prevalence of bullying victimization was observed in Idaho and Utah in the west region, North Dakota, Wisconsin, and Iowa in the mid-west, Oklahoma, Arkansas, and North Carolina in the south. In 2018, a study also reported similar bullying problems in the US and ranked Louisiana, Arkansas, Missouri, and Idaho between 1 and 4 out of 47 states (McCann, 2018).

Regarding the geographic distribution of mental health and disorder, this study showed that children across the 48 contigous states are significantly more likely to be screened with a mental, emotional, developmental, or behavioral disorder. One important factor that has been used to explain the disparity of the pattern of a mental, emotional, developmental, or behavioral problem in the US include poverty (National Academies of Sciences & Medicine, 2019; Yoshikawa et al., 2012). Coincidentally, the states with the highest prevalence and hotspot of MEDB index are also states with the highest poverty and inequality rates in the US. Recently, the National Academies of Sciences, Engineering, and Medicine called for a national agenda to effectively mitigate the risks for MEDB health outcomes in the US.

Limitation and Strength

Despite the evidence presented in this study, there are issues related to data, methodology, and measurement that worth noting while interpreting the results for preventive interventions. First, this study used one-point data from a cross-sectional design with inherent issues with the inability to account for the historical and progression effect of a phenomenon, particularly in health-related research. Hence, this study’s cross-sectional nature does not permit making causal inferences between bullying victimization and the several mental health/behavioral disorder conditions examined in this study. Therefore, the findings are limited to the mere association between variables. Second, data on bullying and bullying victimization was self-reported by parents of children in the selected household, and this might have led to underestimating the prevalence reported in this study. Children experiencing bullying may not always report to their parents.

Additionally, due to full knowledge of the spectrums of disability and confidentiality issues, parents may misreport or decline to report a child’s disability. All these factors should be borne in mind when interpreting a cross-sectional report like this study. Third, this study acknowledges other avenues for bullying victimization, such as cyberbullying/victimization; however, the NSCH survey only focused on the traditional form of bullying. It is hoped that the agency responsible for this project will, in the future, consider collecting information on cyberbullying. Lastly, the operationalization of bullying and wording in NSCH has changed markedly. The survey question and timeframe changed in 2018 to include the frequency of occurrence during the past 12 months (how often) rather than the degree of accuracy of the statement (how true). Response options were revised to give five frequency options: never (in the past 12 months), 1–2 times (in the past 12 months), 1–2 times per month, 1–2 times per week, or almost every day. Lastly, the combination of various responses to form bullying victimization may also be problematic.

Despite these limitations, this study not only contributes to the existing knowledge of children and adolescents with different mental and behavioral disorders exposure to bullying victimization; it provides a medical geography perspective of bullying as a psychosocial insult. It also provides updates on the national prevalence of children with neuropsychiatric disorders and other chronic health conditions and bullying victimization at the state and age levels.

This study’s intention toward bullying prevention is consistent with several others (Center for Disease Control and Prevention, 2019; Lebrun-Harris et al., 2019; Rose & Monda-Amaya, 2012), particularly children with disabilities across the United States. Sequel to this study’s findings, effective bully prevention strategies that can protect and improve the quality of life for children with disabilities would reduce the current prevalence. Based on the current findings, it is recommended that such programs begin at a younger age and progress with the increasing age of students with internalized problems, behavioral or conduct problems. Besides, improving the living and neighborhood environment of victims of bullying could reduce the exposure of children to bullying through relocation to a new and better living environment. Existing and new upstream preventive programs should be strictly enforced and specifically targeted at the 48 contigous states, where the rate of children with disabilities is significantly clustered.

Data Availability

All data and supporting documentation are publicly available from Data Resources Center for Child & Adolescent Health. https://www.childhealthdata.org/dataset.

Declarations

Conflict of Interest

The author declares no conflict of interest.

Financial Disclosure

The author of this paper reported no financial support.