Using Base Rate of Low Scores to Identify Progression from Amnestic Mild Cognitive Impairment to Alzheimer's Disease

OBJECTIVES: To investigate the implications of obtaining one or more low scores on a battery of cognitive tests on diagnosing mild cognitive impairment (MCI). DESIGN: Observational longitudinal study. SETTING: Alzheimer's Disease Neuroimaging Initiative. PARTICIPANTS: Normal controls (NC, n = 280) and participants with MCI (n = 415) according to Petersen criteria were reclassified using the Jak/Bondi criteria and number of impaired tests (NIT) criteria. MEASUREMENTS: Diagnostic statistics and hazard ratios of progression to Alzheimer's disease (AD) were compared according to diagnostic criteria. RESULTS: The NIT criteria were a better predictor of progression to AD than the Petersen or Jak/Bondi criteria, with optimal sensitivity, specificity, and positive and negative predictive value. CONCLUSION: Considering normal variability in cognitive test performance when diagnosing MCI may help identify individuals at greatest risk of progression to AD with greater certainty.