Predicting institutionalization and mortality across the spectrum of Alzheimer’s disease

Abstract

AbstractBackgroundPatients and care partners express a desire for accurate prognostic information. Previous studies predicting institutionalization and mortality focused on the dementia stage. Yet, Alzheimer’s disease (AD) is characterized by a long pre‐dementia stage. We aimed to predict institutionalization and mortality in patients across the cognitive spectrum of AD.MethodWe included n=1418 non‐demented patients with Subjective Cognitive Decline (SCD) or Mild Cognitive Impairment (MCI) (age 63±7, 38%F, Mini Mental State Examination (MMSE) 27±2) and n=1179 patients with AD dementia (age 65±7, 53%F, MMSE 20±5) from the Amsterdam Dementia Cohort. The dates of institutionalization (admission to a nursing home; follow‐up 3.4±2.7 years) and mortality (follow‐up 4.5±2.7 years) were derived from Statistics Netherlands. We constructed models, stratified by dementia status, to predict institutionalization and mortality using Cox regression. We evaluated the following determinants: MMSE, Neuropsychiatric Inventory (NPI), Charlson Comorbidity Index (CCI) and APOE e4 status, MRI (medial temporal lobe atrophy (MTA), global cortical atrophy (GCA), white matter hyperintensities (WMH)) and CSF biomarker levels (Aβ42 and p‐tau). Determinants were selected using backward selection (p<0.10). All models included age and sex. In addition, we evaluated the discriminative performance of models without MRI and CSF biomarkers. Discriminative performance of the models was assessed with Harrell’s C statistics.ResultIn non‐demented patients, n=123 (9%) patients died and n=74 (5%) patients were institutionalized. In AD dementia, n=413 (35%) patients died and n=453 (38%) patients were institutionalized. Table 1 shows the optimal prediction models for institutionalization and mortality. Discriminative performance was better in non‐demented patients than in those in the dementia stage: institutionalization (Harrell’s C 0.81 vs 0.68) and mortality (0.78 vs 0.66). Although model fit was better for the full model, discriminative performance was comparable for the model without MRI for institutionalization and without CSF for mortality in non‐demented patients (Table 1).ConclusionWe constructed prediction models to predict institutionalization and mortality in patients across the spectrum of AD. Highest discriminative performance was found in models for non‐demented patients with SCD or MCI. CSF biomarkers contributed most to predict institutionalization and MRI biomarkers to predict mortality.