Measuring clinically relevant change in apathy symptoms in ADMET and ADMET 2.

Abstract

Among participants with Alzheimer's disease (AD) we estimated the minimal clinically important difference (MCID) in apathy symptom severity on three scales. Retrospective anchor- and distribution-based analyses of change in apathy symptom scores. Apathy in Dementia Methylphenidate Trial (ADMET) and ADMET 2 randomized controlled trials conducted at three and ten clinics specialized in dementia care in United States and Canada, respectively. Two hundred and sixty participants (60 ADMET, 200 ADMET 2) with clinically significant apathy in Alzheimer's disease. The Clinical Global Impression of Change in Apathy scale was used as the anchor measure and the MCID on the Neuropsychiatric Inventory - Apathy (NPI-A), Dementia Apathy Interview and Rating (DAIR), and Apathy Evaluation Scale-Informant (AES-I) were estimated with linear mixed models across all study visits. The estimated thresholds were evaluated with performance metrics. Among the MCID was a decrease of four points (95% CI: -4.0 to -4.8) on the NPI-A, 0.56 points (95% CI: -0.47 to -0.65) on the DAIR, and three points on the AES-I (95% CI: -0.9 to -5.4). Distribution-based analyses were largely consistent with the anchor-based analyses. The MCID across the three measures showed ∼60% accuracy. Sensitivity analyses found that MMSE scores and apathy severity at baseline influenced the estimated MCID. MCIDs for apathy on three scales will help evaluate treatment efficacy at the individual level. However, the modest correspondence between MCID and clinical impression of change suggests the need to consider other scales.