Abstract
ABSTRACT
Objective
The Frontal Behavioural Inventory (FBI) was validated in behavioural‐variant frontotemporal dementia (bvFTD) with a cut‐off of 25. We tested whether this threshold is appropriate in amyotrophic lateral sclerosis (ALS), where behavioural change is milder and partly confounded by motor pathology.
Methods
We analysed 506 ALS patients with a valid FBI assessment. Behavioural impairment was defined using two complementary reference standards, to minimise dependence on any single behavioural instrument: a consensus rule (pathology on ≥ 2 of three independent instruments: Edinburgh Cognitive and Behavioural ALS Screen, Frontal Systems Behaviour Scale, Beaumont Behavioural Inventory) and Latent Class Analysis. ROC analysis identified the Youden‐optimal cut‐off; agreement was quantified by Cohen's
κ
; internal validity was assessed by stratified 5‐fold cross‐validation. Total and subscale scores were analysed separately.
Results
Among 345 complete‐case patients, 24.6% met criteria for behavioural impairment. The FBI total discriminated impairment well (AUC 0.855, 95% CI 0.806–0.898). The Youden‐optimal cut‐off was FBI ≥ 9 (sensitivity 0.87, specificity 0.74), whereas the legacy cut‐off of 25 showed sensitivity 0.22. Subscale recalibration yielded cut‐offs of FBI‐Apathy ≥ 7 (AUC 0.808) and FBI‐Disinhibition ≥ 4 (AUC 0.851).
Conclusions
In ALS, the bvFTD‐derived threshold of 25 misses most patients with behavioural involvement. FBI ≥ 9 is a promising candidate ALS‐specific screening threshold, with scores ≥ 15 indicating a higher probability of behavioural involvement; exploratory subscale thresholds of ≥ 7 for Apathy and ≥ 4 for Disinhibition are also proposed. These cut‐offs substantially improve sensitivity while maintaining acceptable specificity, but require prospective external validation before replacing current practice.