Neuropsychological Function in Migraine Headaches: An Expanded Comprehensive Multidomain Meta-Analysis

Abstract

A sizable literature has studied neuropsychologic function in persons with migraine (PwM), but despite this, few quantitative syntheses exist. These focused on circumscribed areas of the literature. In this study, we conducted an expanded comprehensive meta-analysis comparing performance on clinical measures of neuropsychological function both within and across domains, between samples of PwM and healthy controls (HCs). For this Meta-analyses Of Observational Studies in Epidemiology-compliant meta-analysis, a unified search strategy was applied to OneSearch (a comprehensive collection of electronic databases) to identify peer-reviewed original research published across all years up until August 1, 2023. Using random-effects modeling, we examined aggregated effect sizes (Hedges' g), between-study heterogeneity (Cochran Q and I2), moderating variables (meta-regression and subgroup analyses), and publication bias (Egger regression intercept and Duval and Tweedie Trim-and-Fill procedure). Study bias was also coded using the NIH Study Quality Assessment Tools. Omnibus meta-analysis from the 58 studies included (PwM n = 5,452, HC n = 16,647; 612 effect sizes extracted) indicated lower overall cognitive performance in PwM vs HCs (g = -0.37; 95% CI -0.47 to -0.28; p < 0.001), and high between-study heterogeneity (Q = 311.25, I2 = 81.69). Significant domain-specific negative effects were observed in global cognition (g = -0.46, p < 0.001), executive function (g = -0.45, p < 0.001), processing speed (g = -0.42, p < 0.001), visuospatial/construction (g = -0.39, p = 0.006), simple/complex attention (g = -0.38, p < 0.001), learning/memory (g = -0.25, p < 0.001), and language (g = -0.24, p < 0.001). Orientation (p = 0.146), motor (p = 0.102), and intelligence (p = 0.899) were not significant. Moderator analyses indicated that age (particularly younger HCs), samples drawn from health care facility settings (e.g., tertiary headache centers) vs community-based populations, and higher attack duration were associated with larger (negative) effects and accounted for a significant proportion of between-study heterogeneity in effects. Notably, PwM without aura yielded stronger (negative) effects (omnibus g = -0.37) vs those with aura (omnibus g = -0.10), though aura status did not account for heterogeneity observed between studies. Relative to HCs, PwM demonstrate worse neurocognition, as detected by neuropsychological tests, especially on cognitive screeners and tests within executive functioning and processing speed domains. Effects were generally small to moderate in magnitude and evident only in clinic (vs community) samples. Aura was not meaningfully associated with neurocognitive impairment.