Objective:
Although headache (HA) is a common sequela of traumatic brain injury (TBI), early predictors of chronic HA after moderate to severe TBI are not well established, and the relationship chronic HA has with psychosocial functioning is understudied. Thus, we sought to (1) determine demographic and injury predictors of chronic HA 1 or more years after moderate to severe TBI and (2) examine associations between chronic HA and psychosocial outcomes.
Setting:
Community.
Participants:
Participants in the TBI Model System (TBIMS) with moderate to severe TBI who consented for additional chronic pain questionnaires at the time of TBIMS follow-up.
Design:
Multisite, observational cohort study using LASSO (least absolute shrinkage and selection operator) regression for prediction modeling and independent t tests for psychosocial associations.
Main Outcome Measure(s):
Chronic HA after TBI at year 1 or 2 postinjury and more remotely (5 or more years).
Results:
The LASSO model for chronic HA at 1 to 2 years achieved acceptable predictability (cross-validated area under the curve [AUC] = 0.70). At 5 or more years, predictability was nearly acceptable (cross-validated AUC = 0.68), but much more complex, with more than twice as many variables contributing. Injury characteristics had stronger predictive value at postinjury years 1 to 2 versus 5 or more years, especially sustained intracranial pressure elevation (odds ratio [OR] = 3.8) and skull fragments on head computed tomography (CT) (OR = 2.5). Additional TBI(s) was a risk factor at both time frames, as were multiple socioeconomic characteristics, including lower education level, younger age, female gender, and Black race. Lower education level was a particularly strong predictor at 5 or more years (OR up to 3.5). Emotional and participation outcomes were broadly poorer among persons with chronic HA after moderate to severe TBI.
Conclusions:
Among people with moderate to severe TBI, chronic HA is associated with significant psychosocial burden. The identified risk factors will enable targeted clinical screening and monitoring strategies to enhance clinical care pathways that could lead to better outcomes. They may also be useful as stratification or covariates in future clinical trial research on treatments.