|
Abstract
|
BACKGROUND: Studies of pedal cyclist injuries have largely focused on individual injury categories, but every region of the cyclist's body is exposed to potential trauma. Real-world injury patterns can be complex, and isolated injuries to one body part are uncommon among casualties requiring hospitalization. Latent class analysis (LCA) may identify important patterns in heterogeneous samples of qualitative data.
METHODS: Data were taken from the Trauma Quality Improvement Program of the American College of Surgeons for 2017. Inclusion criteria were age 18 years or less and an external cause of injury code for pedal cyclist. Injuries were characterized by Abbreviated Injury Scale codes. Injury categories and the total number of injuries served as covariates for LCA. A model was selected on the basis of the Akaike and Bayesian information criteria and the interpretability of the classes. Associations were analyzed between class membership and demographic factors, circumstantial factors, metrics of injury severity, and helmet wear. Within-class associations of helmet wear with injury severity were analyzed as well.
RESULTS: There were 6151 injured pediatric pedal cyclists in the study sample. The mortality rate was 0.5%. The rate of helmet wear was 18%. LCA yielded a model with 6 classes: 'polytrauma' (5.5%), 'brain' (9.0%), 'abdomen' (11.0%), 'upper limb' (20.9%), 'lower limb' (12.4%), and 'head' (41.2%). Class membership had highly significant univariate associations with all covariates except insurance payer. Helmet wear was most common in the 'abdomen' class and least common in the 'polytrauma' and 'brain' classes. Within classes, there was no association of helmet wear with severity of injury.
CONCLUSIONS: LCA identified 6 clear and distinct patterns of injury with varying demographic and circumstantial associations that may be relevant for prevention. The rate of helmet wear was low, but it varied among classes in accordance with mechanistic expectations. LCA may be an underutilized tool in trauma epidemiology.
Language: en |