Data from a probability sample of casualty patients treated at a count
y hospital emergency room (ER) during a 1 year period (N = 1124) are c
ompared to data from coroner reports of all fatalities arising from un
natural causes during the same time period in the same county (N = 304
). The two samples are compared on: demographic characteristics, cause
s of casualty (fall, laceration/puncture wound, motor vehicle, fire, i
ngestion, other cause), place of injury, and alcohol and drug use prio
r to the event. Alcohol and drug use data were obtained by breathalyze
r and self-reports in the ER sample and by toxicology screening of blo
od upon autopsy in the coroner sample. The coroner sample was signific
antly more likely to be male, younger and white compared to the ER sam
ple. A significantly larger proportion of the coroner sample was posit
ive for alcohol (43%) compared to those breathalyzed within 6 hr of in
jury who reported no drinking after the event (11%) and to those who r
eported drinking within the 6 hr prior to the event (28%) in the ER sa
mple. Among those who were alcohol positive no difference was found be
tween the coroner sample and the ER sample for the proportion of those
who were also drug positive (24% in each). Cases in the coroner sampl
e were no more likely to involve violence (17%) than those in the ER s
ample (20%). Violence-related fatalities were more likely to involve a
lcohol (47%) than non-fatal injuries (19%), but were no more likely to
involve drug use in combination with alcohol (39% vs. 31%, respective
ly).