Research Letter: Interview Versus Self-Administration of Retrospective Brain Injury Identification

(1) Determine concordance between retrospective identification of a prior history of brain injury exposure when the method of administration is either interview or self-administered; and (2) Determine the test/retest reliability of self-administered identification.

American Population Panel.

In total, 391 (54% female; 70% white, non-Hispanic) self-selected from 54 000 panel participants aged 18 years or older living in the United States.

Descriptive measures of concordance and test/retest reliability. The concordance sample was randomized for order of administration.

Ohio State University Traumatic Brain Injury Identification Method (OSU TBI-ID) with additional questions on anoxic/hypoxic events and prior diagnosis of other acquired brain injuries.

Substantial concordance was found for summary indices comparing OSU TBI-ID self-administered to interview results. High test/retest reliability was observed for most indices derived from the self-administered OSU TBI-ID. For both concordance and test/retest reliability, perfect agreement ranged from 85% to 90% for both the presence or absence of a history of any TBI, the presence or absence of a TBI that resulted in loss of consciousness, and the occurrence of repeated head impacts. Some indices of hypoxic/anoxic events could not be tested due to low frequency, but perfect agreement exceeded 90% for ever losing consciousness due to a drug overdose, or ever due to being choked or strangled. There was also very high perfect agreement for ever having been diagnosed with another acquired brain injury (91% for concordance; 89% for reliability).

These findings suggest that the psychometric properties of the self-administered OSU TBI-ID are comparable to those in the interview version. High concordance and test/retest reliability were also observed for identification of other sources of acquired brain injury. Findings for items used to capture certain anoxic/hypoxic events were also supported though low endorsement rates limited some analyses.