THE TEST-RETEST RELIABILITY OF A STANDARDIZED NEUROCOGNITIVE AND NEUROPHYSIOLOGICAL TEST BATTERY: “NEUROMARKER”

Abstract

NeuroMarker combines EEG and ERP measures with neurocognitive tests in a fully computerized and standardized testing system. It is designed for use across the lifespan and has a large normative database of over 1,000 subjects. This study was a preliminary evaluation of “NeuroMarker” in subjects spanning four decades. Twenty-one healthy subjects (12–57 years) were tested at baseline and four weeks later. From the “Neuromarker” battery, the authors analyzed EEG data (eyes open and closed) and ERPs elicited during auditory oddball (N100, P200, N200, P300) and working memory (P150, P300) tasks. Concomitant neuropsychological data, acquired using a touch-screen system, comprised measures of sensori-motor, attention, verbal, executive, and memory function. Test-retest data were examined using analyses of variance and correlational procedures (corrected for multiple comparisons), with parallel analyses of age. EEG data did not differ across sessions, and showed high test-retest reliability (.71–.95), particularly for theta and delta (>.85). ERP components also showed sound reliability, particularly for sites where components are maximal: fronto-central N100 (.76–.77), centro-parietal P300 (.78–.81) to oddball targets, N100 and P200 (.74–.86) to oddball non-targets, and P150 amplitude and latency (.84–.93) to working memory stimuli. Neuropsychological tests showed a similarly sound level of consistency (on average, .70), with the most consistent tests tapping simple motor function, estimated intelligence, switching of attention (Part 2), verbal interference response time and memory intrusions (.71–.89). Age and sex did not have a differential impact on reliability for EEG, ERP, or neuropsychology measures. These findings provide preliminary evidence that the “NeuroMarker” battery is reliable for test-retest assessments. The results suggest that the standardized approach has utility for providing sensitive clinical and treatment evaluations across age groups.

• Keywords
• EEG
• ERP
• reliability
• neuropsychology
• oddball test-retest

Notes

¹When age and sex were not included as covariates, the following session effects were observed: Theta power: Session effect (f = 16.62, p = .001); Alpha power: Session by midline interaction (f = 3.87, df = 2.19, p = .039), at the uncorrected level. This pattern of results suggests that any significant effects involving session are due to the contribution of demographic (age, sex) factors rather than to practice effects per se.

²When age and sex were not controlled for, both N200 latency (f = 4.90, df = 1.18, p = .042) and P300 latency (f = 4.84, df = 1.17, p = .042) for targets were slightly longer at re-test. For nontargets, N100 latency was also slightly longer at re-test (by about 5 ms, f = 4.92, df = 1.18, p = .04). Together, these data suggest a slight latency shift of the whole waveform at re-test, a shift that interacts with demographic data.

³When age and sex were not controlled, only one test-retest main effect was observed at the corrected alpha level: Switching of Attention, Part 2 (f = 9.47, df = 20, p = .006). At the uncorrected alpha level, the following test-retest effects were observed: Spot the Real Word accuracy (f = 7.18, p = .014), but ot the IQ estimate; Word generation score (FAS) (f = 5.69, p = .027); Total accuracy for Memory Recall (f = 10.27, p = .004).