Reliability of Concussion Signs and Symptoms Reporting Among Former Professional American-Style Football Players

Retrospective evaluations of repeated head injury are needed to better understand associations between head injury exposure and later-life deleterious outcomes. However, there is limited assessment of whether head injury recall assessments produce consistent measures over time, and no assessment of whether the reporting is related to current health status. The concussion signs and symptoms scale (CSS; developed for the Football Players Health Study at Harvard University) was designed to measure cumulative head injury exposure history by asking about the frequency of 10 CSS during active football play. Responses are summed with a total CSS range of 0-130. Former professional American-style football players completed the CSS at two timepoints. A subset of participants also reported on current health (subjective cognitive symptoms [Quality of Life in Neurological Disorders], depression [Patient Health Questionnaire], anxiety [Generalized Anxiety Disorder], pain [Patient-Reported Outcome Measurement Information System (PROMIS) Global], and overall health [PROMIS Global]) at each timepoint. To examine reporting consistency and recall bias, we calculated the Spearman correlation between measures assessed an average of 74.5 (standard deviation [SD] = 41.2) months apart and estimated associations between change in demographic, football-related, and current health factors and change in CSS (ΔCSS) over time using multivariable linear regression. Across the 335 participants, the mean (SD) CSS score at times 1 and 2 were 30.2 (25.5) and 29.1 (25.2), respectively, with an average change in CSS (ΔCSS) of -1.1 (SD = 19.8). There was no significant association between ΔCSS and years since play, months between timepoints, or age at time 1 (0.49 < p < 0.84). Eighty-one (24.2%) participants completed identical questions on current health factors at times 1 and 2. In separate multivariable models, there was no association between changes in pain, cognitive symptoms, health, depression, and anxiety reporting and ΔCSS (0.17 < p < 0.92). On average, the CSS score as a measure of retrospective concussion exposure did not change meaningfully over an average of 75 months, and changes in current health status were not significantly associated with ΔCSS. Results suggest that the CSS scale is stable over time and appears robust against changes in health status. The CSS should be considered for other retrospective studies of brain-injured populations to measure prior cumulative concussion history.