Neurocognitive disorders are under-diagnosed. A previously validated public domain cognitive screening test, Self-Administered Tasks Uncovering Risk of Neurodegeneration (SATURN), may address this need. However, there are no prior reports of its performance in its intended clinical settings. From a single dementia clinic, we retrospectively collected and analysed scores on SATURN, the Montreal Cognitive Assessment (MoCA), Mini-Mental State Examination (MMSE), Saint Louis University Mental Status (SLUMS), and the Functional Activities Questionnaire (FAQ), obtained over a two-year period. We found that SATURN was well-correlated with each of the other assessments (MoCA: r = 0.82, MMSE: r = 0.76, SLUMS: r = 0.81, and FAQ: r = -0.55). In receiver operating characteristic (ROC) analyses, SATURN was statistically non-inferior to those other tests, and indeed was usually numerically (but not statistically) superior. In short, SATURN compared favorably to legacy cognitive screening instruments in regular clinical practice.

We retrospectively analyzed electronic medical records of patients by a single cognitive neurologist between August 2020 and July 2022. Records from a total of 1033 patients were potentially eligible for this retrospective analysis. After exclusions, we analysed data from 113 patients tested with SATURN and 219 patients not tested with SATURN. When SATURN was used to screen for cognitive impairment, it was the index event. When SATURN was not used, the first instance of using one of the other tests (MMSE, MoCA, SLUMS) was the index event. Age, sex, race/ethnicity, level of education, and neurologist diagnosis from the index event were collected. We also collected any other test scores within 365 days of the index event, allowing linear regression comparison of various test scores (e.g., SATURN vs. MoCA). Where available, the diagnosis from an independent neuropsychologist is also documented, if within 365 days of the index event. Etiologic diagnosis is documented based on the best understanding of the patient's condition at the time of data extraction. All other information beyond July 2022 (or 365 days after the index event, whichever happened first) is censored. To appropriately de-identify records for Dryad, we binned ages and levels of education.