Measurement implications on the association between self-reported concussion history and depression: An NFL-LONG study.

This study examined how associations of self-reported concussion history and depression vary based on different variable characterizations. Former NFL players (=1,697) completed a General Health Survey, indicating the number of concussions they sustained during their football career and whether a physician had diagnosed them with depression, and the PROMIS 4-item Depression scale. Self-reported concussion history was characterized as: a 3-category variable (0, 1-2, 3+) with (1) indicator variables and (2) as an ordinal variable; a 5-category variable (0, 1-2, 3-5, 6-9, 10+) with (3) indicator variables and (4) as an ordinal variable; and (5) the original interval scale (0, 1, 2,…, 10, 10+). Depression was characterized as: (1) physician diagnosis (yes/no); and the PROMIS 4-item Depression scale treated as: (2) the original -score variable and (3) using a cut-off of a -score >60 versus ≤60. Regression models with various combinations of the characterizations, while adjusting for age, race/ethnicity, education, pain interference, and stress level, estimated prevalence ratios and mean differences for binary and continuous outcomes, respectively. Concussion history-related effect estimates were compared across all models. Self-reported concussion history emerged as a significant predictor of each depression measure. With a higher number of concussions reported, be it via the categories or the interval scale, stronger associations between self-reported concussion history and depression were observed. The various approaches to characterize self-reported concussion history and depression provided evidence of significant associations between the two variables, with the degree of association varying based on characterization of each construct.