Inaccuracy of self-reported migraine: insights from 601 paired assessments from a single-centre study in Australia

Abstract

Both clinical and research settings may rely on accurate self-diagnosis of migraine for medical triage, population-based surveys or epidemiological studies. Unfortunately, accessing neurologists for diagnosis may be cost-prohibitive, delayed due to workforce shortages or impractical given the significant burden of the condition on limited public health resources. We aimed to ensure concordance between self and specialist diagnosis to confirm reliability of this proxy diagnostic measure for clinical workflow and validate studies relying on self-reported migraine. Single-centre, prospective, cross-sectional study at a women’s-specific mental health clinic, comparing self-reported lifetime migraine diagnosis to that of consultant headache neurologist blinded to initial self-reported diagnosis. The primary endpoint was overall agreement between the self and specialist diagnosis using a binary yes and no classification. Secondary analyses were performed for inter-rater agreement using Cohen’s kappa and systematic disagreement using McNemar test, in addition to reporting sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV). A total of 601 female patients were included (mean age 40.0y, SD 11.8). The overall agreement between self-diagnosis and specialist was 69.6% (418/601). After correcting for chance, agreement between self-reported migraine and neurologist-diagnosed migraine was fair (κ = 0.424, 95% CI (0.354–0.493), p < 0.001). Using neurologist diagnosis as the reference standard, self-reported migraine showed low sensitivity (50.4%) but high specificity (95.3%). Positive self-report was highly predictive of neurologist-diagnosed migraine (PPV = 93.6%), whereas negative self-report was less reliable (NPV = 58.8%). McNemar test was significant ( p < 0.001) indicating asymmetry between the two diagnostic methods. Specifically, self-diagnosis under-recognised migraine relative to specialist assessment. These findings suggest that participants who self-identified migraine were usually confirmed by neurologist assessment. However, many participants who did not self-identify migraine nevertheless met diagnostic criteria. Reliance on self-reported migraine may significantly underestimate true disease burden in clinical and research settings.