Injury Types that Predict Subsequent Child Maltreatment Diagnosis: Differences by age, race, and gender

• In Event: Poster Session 06
  In Poster Session: PS 06 Topic - Health, Growth, Injury

Abstract

Background: There is ambiguity regarding specific types of injuries that may indicate abuse and whether there is bias in diagnosis based on age, race/ethnicity, and gender. To address these gaps the current study tested injury categories of bruising, burns, lacerations, fractures, head injuries, and falls as predictors of a subsequent diagnosis of maltreatment by age, race/ethnicity, and gender.

Methods: Data were from the electronic health records of a large integrated healthcare system. The maltreatment cases were selected from members born between 1/1/2009 and 12/31/2019 who had at least one ICD code indicating child maltreatment. A control participant was selected randomly from candidates eligible for each maltreatment case based on age, gender, race/ethnicity and MediCal status. This resulted in 4576 participants per group (total N=9152). Injuries were captured using ICD codes and grouped into the following categories: bruising, fractures, lacerations, head injury, burns, falls, and unspecified injury. Chi-square was used to test the frequencies of each category of injury between the case versus control group by age group (birth to <1yr; 1 to 2.99yrs; 3 to 4.99yrs ; 5 to 10 yrs), race/ethnicity (white, black, Hispanic, Asian, multiple/other/unknown), and gender (male, female). Logistic regression was used to examine the seven injury categories as predictors of a subsequent child maltreatment diagnosis to determine the relative strength of each when all were included in the same model.

Results: Chi-square analyses for age group showed that all injury categories occurred more frequently in the case than control group for the birth-<1yr and 1-2.99yr age groups. None of the injury types were significant risks for maltreatment in the 3-4.99yr or 5-10yr age groups. Due to these differences by age we examined race/ethnicity and gender by age as well (combining the youngest groups into one birth-2.99yr old group). For boys in the birth-2.99yr group, all injury categories were more frequent in the case than control, whereas for girls all categories were more frequent for cases except burn. There were no significant differences in risk for the older age groups by gender. For Hispanic children in the birth-2.99yr old group, all the injury categories were more frequent for the case than control, whereas for black youth only fracture and laceration were more frequent, and for white children only fracture and head injury were more frequent. Lastly, in the final model with all injury types included together (see Table 1), fracture was the strongest predictor in the birth-2.99yr old age group (OR=4.88, 95% CI=2.77, 8.60), followed by head injury (OR=3.51, 95% CI=2.51, 4.91). No injury category was a significant risk for maltreatment in the older age groups (Table 2).

Conclusions: Injury type may predict child maltreatment in young children under 3, but not for older children. However, the utility of different injury categories appears to differ by gender and race/ethnicity. Mandated reporters should consider these potential biases when evaluating suspicions injuries. These results add to empirically informed recommendations regarding the types of injuries that may be indicative of abuse and predict future abuse.

Authors

• Sonya Negriff, Kaiser Permanente Southern California
  Presenting Author

• Mercie DiGangi, Kaiser Permanente Southern California
  Non-Presenting Author