Background: Sedative-hypnotic drugs not only increase sedation, but also im
pair memory as serum concentration increases. These drugs also produce prof
ound changes in the auditory event-related potential (ERP). The ability of
various ERP components to predict changes in sedation and memory produced b
y various drugs was tested.
Methods: Sixty-five healthy volunteers randomly received intravenous placeb
o, midazolam, propofol, thiopental, fentanyl with ondansetron, or ondansetr
on alone at five different stable target concentrations (three increasing,
two decreasing) rising a computer-controlled infusion pump to produce varyi
ng degrees of sedation without loss of consciousness. ERPs were recorded wh
ile volunteer participants detected a deviant auditory stimulus and made a
button-press response to a target tone (standard oddball paradigm, 80:20 ra
tio, to elicit a P3 response). At each target concentration, volunteers lea
rned a list of 16 words. The predictive probabilities (P-k) of various ERP
components were deter-mined for word recognition at the end of the day (mem
ory) and log reaction time to the deviant stimulus (sedation).
Results: The N2 latency of the ERP consistently predicted log reaction time
in all groups (P-k +/- SE from 0.58 +/- 0.04 to 0.71 +/- 0.04). The N2P3 a
mplitude of the ERP was the best predictor of memory performance for midazo
lam (P-k, 0.63 +/- 0.04), propofol (P-k, 0.62 +/- 0.05), and thiopental (P-
k, 0.66 +/- 0.04). There was a differential ability to predict memory perfo
rmance from sedation for midazolam and propofol.
Conclusions: Midazolam and propofol affect memory differentially from their
sedative effects, and these are indexed by specific components of the audi
tory ERP. These components of the ERP are associated with specific, but not
necessarily unique, neuroanatomic structures. Thus, these drugs act by add
itional mechanisms beyond general central nervous system depression to prod
uce the effects of sedation and memory impairment.