Electroencephalography and survival in patients with Alzheimer's disease

Abstract

Reply from the Authors: We thank Claus et al. for their interesting comments regarding our article,' in which we found that abnormal spectral EEG at baseline examination was associated with cognitive and functional decline in probable AD, but not with death. They performed a similar analysis in their cohort, and found that abnormal baseline EEG was associated with cognitive and functional decline, and also with death. They felt that the discrepancy between the studies was that they followed their patients for a longer period of time and that we used an EEG composite measure (parasagittal mean frequency).We agree with Claus et al. that age is a predictor of death in AD. However, longitudinal studies that compared early versus late-onset AD have shown that early onset AD is associated with faster time to death,? although this is not a universal Other studies have found that older age (>70 years) is associated with death in AD.4 The reasons for these discrepancies are not clear, and may be due to sample characteristics, size of the cohorts, and other unidentified variables. For example, AD patients with late onset (age >65 years) have more cerebrovascular risk factors than those with an early onset,7 which may contribute to shorter survival.Comorbid medicalheurologic conditions can affect survival. All our patients had probable AD; none of them had a history of strokes or transient ischemic attacks or "silent" infarcts detected on neuroimaging studies a t the time of EEG. Furthermore, because our longitudinal study was oriented to examine AD in its "purest" form, there were strict selection criteria in terms of cerebrovascular risk factors. Patients with poor control hypertension (e.g., inadequate medical control of the medication, inadequate patient compliance) or secondary forms of hypertension with poor therapeutic response were excluded, as were those with severe uncontrolled ischemic heart disease (e.g., signs and symptoms of congestive heart failure, cardiac dysrhythmia, history of myocardial infarction within the previous 6 months). None of these patients had diabetes mellitus. Thus, these age-related health factors were less likely to affect survival in our sample.Finally, we have repeated our analysis in the same cohort. As of December 31, 1997, 51/95 (54%) patients have died, and the follow-up time is longer (mean, 4.24 ? 2.4 years; range, 0.9 to 12.3 years). Using a Cox proportional analysis adjusted by age, sex, education, severity of the symptoms of dementia (Mini-Mental State Examination), and duration of follow-up, we did not find an association between separate spectral EEG bands (as recommended by Claus et al.) at baseline and time to death. In addition, we reanalyzed composite spectral measures, including the parasagittal mean frequency, theta-beta, and the combined delta + theta percentage, and did not find an association between these parameters and survival.Abnormal EEG in AD patients appears to be a strong predictor of cognitive and functional decline. The relation between abnormal EEG and deat...