Development and initial psychometric evaluation of the Brain Fog Questionnaire: a brief measure of subjective cognitive symptoms.

BACKGROUND

Brain fog is a heterogeneous experience across clinical conditions. We developed the five-item Brain Fog Questionnaire (BFQ) as a brief patient-reported measure of subjective cognitive symptoms and conducted an initial psychometric evaluation.

METHODS

In this cross-sectional study, 534 adults were recruited through an open online survey in Taiwan. The BFQ assesses five subjective cognitive symptoms on a four-point frequency scale. Internal consistency and dimensionality were examined using reliability analyses and exploratory factor analysis, with bootstrap estimates of factor-loading stability. K-means clustering and latent class analysis (LCA) characterized symptom patterns; agreement was assessed using the adjusted Rand index (ARI). Internally derived candidate thresholds and correspondence with a global subjective brain fog item were explored.

RESULTS

Internal consistency was good (Cronbach's α = 0.82; ordinal α = 0.87; ωt = 0.90). A dominant general factor was supported (loadings = 0.66-0.94; ωh = 0.85). Although the two-factor model had better numerical fit, its second factor was effectively a single-item factor defined by word-finding difficulty and showed unstable bootstrap estimates. K-means and LCA identified graded differences in symptom severity rather than distinct profiles. Agreement was modest for two-group solutions (ARI = 0.34) and higher for three-group solutions (ARI = 0.69). Candidate thresholds varied by internal classification method, with scores of ≥5-6 identified across analyses. For BFQ ≥6, correspondence with the global item was partial (sensitivity 54%, specificity 77%, positive predictive value 89%, negative predictive value 33%).

CONCLUSIONS

The BFQ showed preliminary internal psychometric properties as a measure of subjective cognitive symptoms. Validation in well-characterized clinical populations using independent cognitive, clinical, and functional measures is required.