Measures of Short-Term Test–Retest Reliability of Computerized Neurobehavioral Tests

Abstract

Neurobehavioral test batteries are often administered repeatedly to evaluate changes over time or effects of clinical interventions or neurotoxic exposures. Time intervals between test sessions range from very short (hours) to very long (decades). The aim of this study was to compare the impact of two brief time intervals on the test–retest reliability of frequently used neurobehavioral tests. Fifty healthy adults were recruited through newspaper advertisements in Portland, Oregon. Participants were divided into either a 6 h (same-day) or 1-week retest group. All participants completed a battery of tests from the computerized Behavioral Assessment and Research System (BARS). Reliability was assessed by Pearson product–moment correlation and by intraclass correlation coefficient (ICC). The test battery generally showed adequate reliability in the short-term (week) and very short-term (day) and stability in performance over repeated administration when examined by multiple measures. Intraclass correlation coefficient ranged from 0.35 to 0.85. The magnitude of variation of performance in the administered tests was equally distributed around zero (i.e. no difference). The findings suggest that neurobehavioral tests such as BARS may be a useful tool for the assessment of acute exposures and clinical status where short-term evaluation is required (e.g. in the same-day or within 1 week).