Use of neighborhood characteristics to improve prediction of psychosocial outcomes: a traumatic brain injury model systems investigation

Abstract

Objective: To determine whether use of neighborhood characteristics derived from U.S. Census Bureau information contributes to the prediction of outcomes after traumatic brain injury (TBI) beyond the variance accounted for by individual characteristics.

Design: Cross-sectional analysis of follow-up interviews conducted 1, 2, and 5 years postinjury.

Setting: Twelve Traumatic Brain Injury Model Systems (TBIMS) centers.

Participants: Patients 16 years of age and older with moderate or severe TBI enrolled in the TBIMS National Database (N=472).

Interventions: None.

Main outcome measures: Satisfaction With Life Scale and Participation Assessment with Recombined Tools-Objective.

Results: Individual characteristics alone accounted for 26% and 48% of variance in life satisfaction and participation, respectively; neighborhood characteristics alone accounted for 6% and 9% of variance, respectively. Models combining both types of characteristics included significant neighborhood and individual predictors for participation but not life satisfaction; however, for participation, prediction only improved beyond that found for individual characteristics alone by 1.2%.

Conclusions: The results did not support the hypotheses that characteristics of a person's neighborhood would increase outcome prediction beyond that which can be accomplished based on characteristics of the individual alone. Though neighborhood characteristics were statistically significant in prediction models, the improvement in percent of variance accounted for was negligible. Refinements in conceptualization and methodology are suggested for continued exploration of the contribution of neighborhood characteristics to TBI outcomes.