METHODS

Study population

We followed up students from 17 of 18 special education schools in the Los Angeles Unified School District (LAUSD) from the 1994–98 academic years. One excluded school services homebound students with severe medical needs.

Definitions and variables

An injury case was defined as physical trauma to any disabled student sustained during a school sponsored activity. We identified and collected details of injuries (nature, cause, location, activity) from pupil accident reports completed by school staff.

We linked cases to disability, age, gender, and race/ethnicity data from the Information Technology Division. To classify disability, a team of experts used the following categories defined by the California Department of Education: aphasic, autistic, blind, deaf/blind, developmentally disabled, mentally retarded, established medical disability, hard of hearing, language/speech impaired, learning disabled, multidisabled, other health, orthopedic, partially sighted, seriously emotionally disturbed, and traumatic brain injured.¹⁵ Developmentally disabled are lowest functioning with severe, lifelong mental and/or physical disability. Multiple disabled students have more than one diagnosed disability.¹⁵

Disabilities were collapsed into five broader groups due to limited person-time and few injury cases observed in most individual disability types. These categories reflect similar physical and cognitive limitations: (1) developmentally disabled; (2) emotional/mental disabled (autistic, learning disabled, mentally retarded, and emotionally disturbed); (3) physically/sensory disabled (orthopedic, aphasic, blind, deaf/blind, hard of hearing, language/speech impaired, partially sighted); (4) multidisabled; and (5) other disabled (established medical disability, other health impaired, undetermined disability, and traumatically brain injured).

We used the primary cause in the chain of events to describe the external cause. For example, if a child first collided with a pole and then fell, the primary cause was a collision. Because of the limited cognition of many subjects, we did not ascertain intentionality. Assault such as bites, pushes, or kicks were labeled “injuries by other students”.

The population was identified and quantified using yearly enrollment rosters. We accrued student-years and corrected for absenteeism by multiplying by (1−average absentee rate) for each school. Absentee figures were provided by the LAUSD budget department. We assumed that enrollment was stable and students experienced the average absentee rate of their school of attendance. Incidence was calculated per 100 student-years.

Analysis

We calculated odds ratios using generalized estimating equations (GEE) models. Data were time stratified to adjust for time dependent exposures (age, school of enrollment). GEE regression accounted for dependence among repeated students nested within schools and thus controlled hierarchically for school of attendance.^16–18 We imposed an autoregressive correlation structure since subjects closer in time are more likely correlated. The deviance test was employed to assess model fit. No interactions were added because of power limitations.

RESULTS

Of 6769 students enrolled into the study, 60% (n = 4076) were boys and 40% (n = 2693) were girls (table 1). Students were predominately Hispanic (61%). Most students (45%) had emotional/mental disabilities (n = 3061), 19% had multiple disabilities (n = 1304), and 18% were developmentally disabled (n = 1229). Fewer than 12% had physical disabilities (n = 799), and 6% had other disabilities (n = 376).

We identified 933 injury events, of which 236 were excluded (121 during summer sessions, two from the excluded medical school, 78 reports of general pain or no “visible injury”, and 35 that did not link to school demographic data). A total of 697 events were analyzed.

Injury rate

The overall injury rate was 4.7/100 student-years (table 1). Rates were highest among students with multiple (6.4/100), other (6.2/100), and physical disabilities (5.3/100). Boys sustained about five injuries per 100 student-years, while girls suffered four injuries per 100 student-years. Black (6.1/100) and Hispanic students (4.6/100) had higher rates of injury than other race/ethnic groups. Rates generally increased with age.

Injury types and body locations

During the 697 injury events, students sustained 888 distinct physical insults to the body. A single region was affected during 547 events. The remaining 150 events resulted in multiple injuries either to the same or different body regions. Single and multiple injuries were collapsed across disability groups (table 2).

Most injuries were superficial bruises (28%), cuts (28%), and abrasions (18%) (table 2). Unspecified wounds and bleeding made up almost 7% of injuries. Another 8% were “other” injuries (poisoning, insect bites, burns, dislocations, fractures, strains/sprains, foreign objects).

Forty eight percent of insults occurred to the face (table 2). The upper extremity (19%) and head/neck (16%) regions were also common locations. Fewer injuries were to the lower extremity (10%). Injury types and body locations were similarly distributed across disability groups (results not shown).

External cause

About two thirds of injuries were either fall related (34%) or involved other students biting, pushing, tripping, striking, or kicking the injured student (31%) (table 3). Seventeen percent of cases collided into, were caught between, or cut by an object. About 12% of injuries involved various mechanisms, including seizures, self inflicted bites, headbangings, poisoning, contact with hot objects/fire, and foreign objects.

When causes were examined by disability, the emotionally/mentally disabled suffered more injuries by other students (38%) than from falls (23%). In contrast, the developmentally disabled, multidisabled, and other disabled were injured more frequently from falls (40%–44%) than by other students (21%–34%). For the physically disabled, almost an equal number of students were injured from these two causes.

School activity

Only 60% of cases had complete information on school activity (table 3). Almost 27% were sports related, with twice as many injuries during organized play (physical education/interscholastic sports) (n = 125) than unorganized play (recess) (n = 62). Sixty six injuries (10%) were sustained during class time. A remaining 22% occurred during other activities: feeding, bus transportation or off-campus activity, passing period, and toileting/diaper/dress change.

Differential patterns were seen in specific disability groups. About a third of injuries to emotional/mental, physical, and multiple disabilities occurred during school related play (table 3). Students with developmental disabilities were injured more often during class time (11%); only 8% of their injuries were sports related.

School location

More than 20% of all injuries occurred in the playground or on athletic fields (table 3). The next most common location was the classroom (13%). Twenty five percent of injuries occurred in “other” locations including the auditorium/gymnasium/cafeteria, bathroom/locker room, bus/bus loading area/lot, and hallway. Forty percent of cases had missing data on both external cause and school location.

Students with emotional/mental, physical, multiple, and other disabilities were injured most often in the playground/athletic field. In contrast, those with developmental disabilities suffered more injuries in the classroom (23%) than in the playground/athletic field (9%).

Measures of association

Our final model included disability, gender, age, school, and year of enrollment. Including race/ethnicity did not improve fit (χ² = 3.29, p = 0.35). Compared with developmentally disabled students, the odds of injury was increased among the multiple disabled (odds ratio (OR) 1.7, 95% confidence interval (CI) 1.3 to 2.3), other disabled (OR 1.5, 95% CI 1.0 to 2.3), and the physically disabled (OR 1.4, 95% CI 1.0 to 1.9) (table 4). Males had a slightly increased odds compared with females (OR 1.2, 95% CI 1.0 to 1.5). Associations between age and injury were weak and imprecise. Odds ratios from the GEE model were similar to estimates from conventional logistic models.

DISCUSSION

This is the first study to describe school related injury in a large population of over 6000 disabled students from multiple schools. Our estimated rate of 4.7/100 is within the range reported in the general school population (1.67 to 7.45/100).¹⁹

Of all disabled groups, the multidisabled appear to be at greatest risk. Their combination of impairments that can occur in dyads, triunes, or greater forms could profoundly interfere with activities of daily living.²⁰ For example, a deaf mentally retarded child cannot process both auditory messages and visual hazards.

Those with physical disabilities also have an increased potential for injury. Other studies report the blind to be at high risk of pedestrian injuries and those with sensorimotor deficits at risk of burns.^21,22 Similarly at school, students with orthopedic handicaps may have difficulty using their limbs for balance to protect against falls. Visually impaired youth often cannot avoid collision hazards even with assistive devices. However, comparatively higher cognitive function may lead them to explore environments similarly to non-disabled children, providing more potential exposure to injury hazards.

Students with developmental disabilities have the weakest injury association. These students have substantial limitations in at least three major life activities (self care, language, learning, mobility, self direction, independent living, and economic sufficiency) and require special lifelong, coordinated care.²⁰ The developmentally disabled have limited physical movement and less contact with their classmates and hence participate in fewer school activities. Decreased exposure probably results in a lower injury risk.

Mental disabilities affect expectations of a child’s behavioral, social, and functional capacity.²³ Prior studies have reported high injury risk among those with behavioral problems.^13,14 Yet, in our study, students with mental/emotional disabilities were at low risk of injury. Perhaps, despite cognitive limitations, these children may function at higher motor and sensory levels compared with their disabled peers and better handle physical hazards. Thus, physical aptitude, rather than mental processing, may most influence injury risk. Limited support of this hypothesis was found in a study of physical education among mentally retarded students. Students with low motor ability suffered substantially more injuries than students with average motor skill.²⁴ Additional research is needed to further test how motor and mental functioning affect injury risk.

Study limitations

No national or international database on school related injuries exists.²⁵ Like many existing school based studies, we are limited to data provided by single school districts maintained primarily to protect against litigation. Unfortunately, reporting biases may compromise the reliability and validity of school data.²⁶

We compared injury rates across schools and found some differences which could reflect true rate differences or a reporting bias. Some variability is expected due to different definitions and reporting. For example, at one school, the nurse files all reports; in another school an administrator handles this responsibility. Without details about school specific procedures, we make no conclusions about the degree of potential bias. At best, we hierarchically adjusted for unexplained variability by including schools as a variable in our GEE models.

Underreporting may still be problematic, although school policy requires reports for “all accidents however slight”. Time constraints or other reasons may have prevented the filing of reports. We argue, however, that underreporting is probably less problematic in the disabled than the able bodied population. Because of their fragile health condition, special needs students are routinely screened by full time nurses. We suspect then that injuries at special schools are reported at a higher level and more consistently than in other school settings.

Despite this favorable situation, existing reports often lacked details on location and activity. Data may be missing because reports were not completed or no adults witnessed the event. For example, a non-verbal cognitively impaired student may present with a physical injury but cannot verbalize how or where the injury occurred. Or, injuries may have occurred in locations with more diffused adult supervision, such as during recess or in the playground. We found no significant missing data patterns across disability and demographics. Nonetheless, a concerted effort is needed for more thorough documentation of injuries at schools. Together schools and the research community should collaborate to build systematic injury data collection systems.

Another potential limitation of our study is our categorization of disability. Although groupings show some similarity in physical, mental and cognitive impairment, categories are admittedly heterogeneous with a wide range of conditions. For instance, the physically disabled could have cerebral palsy, skeletal disorders, muscular dystrophy, or visual impairment.²³ Aggregating these children into one category may have pulled estimates towards the null value. A similar effect could have occurred within other disability categories. Without additional subjects and cases, separate examination of each disability was not feasible. We thus consulted with a special education specialist for expert advice. While categorization of disabilities is difficult in light of each child’s unique condition, our groupings based on these impairments can still help determine appropriate activities and classroom placing (personal communication, Special Education Administrator, 2003). Our findings are thus a general picture of the injury experience within broad types of similar limiting characteristics and serve as a platform for more in-depth investigation.

Migration and lost time from illness could also lead to biased rates. Children with disabilities typically remain at one school for several years, reducing the likelihood of migration (personal communication, Special Education Administrator, 1998). On the other hand, while average absenteeism was accounted, individual student absenteeism was unknown.

IMPLICATIONS FOR PREVENTION

In 2001, the Centers for Disease Control and Prevention (CDC) published recommendations for preventing unintentional injuries and violence in schools. CDC highlighted several areas including the social climate, physical environment, education, school health services, crisis response, and community integration.²⁸ Some of these recommendations are currently being put into action throughout the United States. In New York City, Safety Makes Sense trains teachers, school personnel, and students in unintentional injury prevention.²⁹ In Arizona, PeaceBuilders fosters prosocial behavior through exercises among parents, students, teachers, administrators, and teachers to prevent violence in elementary schools.³⁰ In California, the Department of Education published a resource guide listing several violence prevention programs delivered by classroom instruction. To our knowledge only PATHS in California, focused on violence prevention, has modules for children in special education (www.californiahealthykids.org, accessed 2003).

Many existing prevention programs may or may not be specified to special needs children. We take a first step by providing baseline scientific information about patterns of injury needed for assessing safety needs and structuring interventions. For example, a substantial number of bruises and cuts were reported to the face and head. To prevent these types of injuries, children with physical or multiple disabilities could wear protective headgear, like helmets, during sports or recreational play. Classrooms could also be designed with padding for edges and hard surfaces. Children with developmental disabilities, especially those prone to headbanging, may also benefit from helmet use in the classroom, where more of their injuries seem to occur. On the other hand, student-on-student injuries were most common among the emotionally/mentally disabled. School based conflict resolution measures are available and can be applied if tailored to the cognitive levels of students with mental disabilities.⁷ These are examples of what could be done in the special education environment.

But, before implementation, preventive measures should first incorporate a child’s functional impairment and then be evaluated for appropriateness and effectiveness.⁷ For example, a safety education program for special needs children should include assessment, developing teaching and learning objectives, identifying the capability to learn, preparation, and teaching.³¹ If thoughtfully developed, such programs can significantly impact the care and management of injuries at school.

Public Health Law 99–457 stipulates that schools provide specialized health services to children with disabilities.³² School nurses focus primarily on tracheostomy management, ventilator monitoring, tube feeding, temperature control, catherization, and seizure management.^23,33 The management of injuries is an area for further development but will require collective efforts among nurses, teachers, aides, administrators, and parents.

By creating a safer environment for all children, schools can better meet their educational goals. One goal is to teach children with disabilities life skills needed for social and physical integration into the community. Special education curriculum goes beyond academics and includes methods for student transitioning into society. Information about injury patterns can help this process. By teaching children with disabilities how to navigate their school environment safely, they may be better equipped to manage physical challenges in the outside world.