The Frontotemporal Dementia versus Primary Psychiatric Disorder (FTD versus PPD) Checklist: A Bedside Clinical Tool to Identify Behavioral Variant FTD in Patients with Late-Onset Behavioral Changes: A Bedside Clinical Tool to Identify Behavioral Variant FTD in Patients with Late-Onset Behavioral Changes

Background: Differentiating early behavioral variant frontotemporal dementia (bvFTD) and primary psychiatric disorders (PPD) is complex and biomarkers have limited accuracy, leading to inaccurate diagnoses. Objectives: Develop a simple bedside clinical tool to differentiate bvFTD from PPD. Methods: A checklist of clinical features differentiating bvFTD from PPD was developed based on literature and clinical experience. The checklist was filled prospectively for 29 consecutive patients (Montreal Neurological Hospital) with lateonset (≥ age 40) behavioral changes suggestive of bvFTD. The checklist was subsequently retrospectively completed on the baseline visit (N = 137) of the Late-Onset Frontal Lobe study (Amsterdam). In both cohorts, patients were followed 2 years to establish a final best clinical diagnosis, categorizing patients into Probable FTD (N = 46), Possible FTD (N = 8), Other Cognitive Disorder (N = 36), Other Neurological Disorder (N = 10), or PPD (N = 66). Results: All items distinguished the two groups except "duration more than 5 years", which was removed to create a final 17-item version. Mean checklist scores were significantly different across all groups (Oneway ANOVA F(4,161) = 27.462, p < 0.001). The PPD group had lower scores than all other dementia categories, with the largest difference between Probable FTD (X = 12.04) and PPD (X = 7.48). A score ≥ 11 was found to be strongly indicative of bvFTD (specificity 93.9%, sensitivity 71.1%, PPV 89.2%). Scores ≤8 were strongly indicative of a PPD (specificity 91.3%, sensitivity 77.3%, PPV 92.7%). Patient with scores of 9.10 are considered indeterminate. Conclusions: Although further prospective validation is required, the "FTD vs PPD Checklist" could provide a simple tool to improve diagnostic accuracy, particularly in non-specialized settings.