Longitudinal changes in cognitive ability following traumatic brain injury: a systematic review and meta-analysis

The longitudinal impact of traumatic brain injury on cognitive functioning has been widely studied, though findings remain inconsistent. This may be due to variations in injury severity between participants, and the timing of assessments between studies. This meta-analysis aims to address these limitations by providing an overview of longitudinal cognitive changes in moderate-severe traumatic brain injury cohorts. This meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocols and was registered with PROSPERO (CRD42024562211). PubMed, EMBASE via Ovid and APAPsychInfo via EBSCOHost were last searched in August 2025 and were screened based on pre-established eligibility criteria, regardless of publication date, including longitudinal cognitive assessment in moderate-severe traumatic brain injury cohorts. Studies without follow-up assessments in at least 3 months after the baseline assessment were excluded. Separate random effects meta-analyses were conducted for each assessed cognitive domain (memory, processing speed, verbal ability, visuospatial ability, executive functioning), due to the expected variability in the change trajectories between them. Where appropriate, meta-regression analyses using a random effects maximum likelihood estimator examined the impact of assessment timing on cognitive change. Risk of bias was assessed using the Newcastle Ottawa Scale and Critical Appraisal Skills Programme. Of 9481 original articles, 21 met the eligibility criteria and included a total of 972 participants. Significant improvements were found in memory, processing speed, verbal and visuospatial ability, with effect sizes ranging from 0.27 to 0.38. However, longitudinal assessment of executive functioning did not indicate significant changes (P = 0.712). Meta-regression revealed that shorter durations between injury and baseline assessment predicted greater improvements in processing speed (β = -0.01, SE = 0.00, 95%CI [-0.009, -0.005], z = -6.18, P < 0.001) and memory (β = -0.005, SE = 0.00, 95%CI [-0.007, -0.003], z = -6.31, P < 0.001). Findings suggest consistent, subtle improvements across most cognitive domains, with potential non-linear trajectories, especially for memory and processing speed. However, it is acknowledged that statistically significant improvements have limited clinical significance. Insignificant changes in executive functioning are discussed in relation to potential differing profiles of dysfunction following traumatic brain injury.