Increased usefulness of medical data for public health surveillance of child maltreatment

Date Published
Source

Schnitzer, P.G., Slusher, P.L., Kruse, R.L., & Tarleton, M.M. (2011). Identification of ICD codes suggestive of child maltreatment. Child Abuse & Neglect, 35(1), 3-17.

Reviewed by
Andreas Jud
Summary

Hospitals and medical professionals all over the world use International Classification of Diseases (ICD) codes to identify signs, symptoms, social circumstances, and external causes of injury or disease. Although specific ICD codes for child maltreatment exist, medical professionals fail to identify child maltreatment-related health conditions for numerous reasons. Focusing on hospital discharges and emergency department (ED) visits, the authors intended to increase the usefulness of these medical data for public health surveillance by developing a set of criteria to represent maltreatment, but not coded as such.

The study population included all children less than age 10 with a record in Missouri’s statewide hospital discharge and ED visit database in the year 2000. Fatalities and children already identified with an ICD code for child maltreatment were excluded. Three steps were conducted to retrieve the intended set of maltreatment-related ICD codes. First, an initial list of ICD codes representing illnesses or injuries suspicious for child maltreatment was compiled. Over 28,000 visits meeting these codes were identified. A second step reduced the study population to an eligible sample of 2,826 visits. If the code was assigned to 50 or fewer visits all the visits assigned that code were included; for codes which were assigned to more than 50 visits, a simple random sample of 50 visits was selected for review. In the final sample medical records were reviewed to categorize visits as probable, possible or unlikely maltreatment according to an operationalized classification system. A code was identified as maltreatment-related if more than 66% of medical records were classified as probable or possible maltreatment.

Methodological notes

The study displays a high degree of methodological rigor in its design. However, some problematic issues are due to the ICD and its use. First, ICD codes are mainly used for billing purposes in the U.S. Therefore, they are potentially biased. Second, there are no ICD codes for highly maltreatment-suggestive injuries such as cigarette marks. A further limitation note relates to generalizability. For reasons of frequency, car accidents were excluded although some might be related to maltreatment (e.g., supervisory neglect). Psychiatric hospitals were excluded as admissions to these institutions most likely reflect chronic or long term consequences of maltreatment rather than incidents of maltreatment, which was the focus of the surveillance program.