Background-The purpose of this study was to describe the public locati
ons of cardiac arrest and to estimate the annual incidence of cardiac
arrest per site to determine optimal placement of automatic external d
efibrillators (AEDs), This was a retrospective cohort study. Methods a
nd Results-Locations of cardiac arrest were abstracted from data colle
cted by emergency medical service programs in Seattle and King County,
Washington, from January 1, 1990, through December 31, 1994. Types of
commercial and civic establishments were tallied and grouped into 23
location categories consistent with Standard Industrial Codes, and the
number of sites within each location category was determined. With th
e addition of ''public outdoors'' and ''automobiles'' as categories, t
here were 25 location categories. During the study period, 7185 arrest
s occurred, 1130 (16%) of which were in public locations. An annual in
cidence of cardiac arrest per site was calculated. Ten location catego
ries with 172 sites were identified as having a higher incidence of ca
rdiac arrest (greater than or equal to.03 per year per site). Thirteen
location categories had a lower incidence of arrest (less than or equ
al to.01 per year per site). There were approximate to 71 000 sites wi
thin these categories. Conclusions-Placement of 276 AEDs in the 172 hi
gher-incidence sites would have provided treatment for 134 cardiac arr
est patients in a 5-year period, 60% of whom were in ventricular fibri
llation. We estimate between 8 and 32 lives could be saved in 5 years.
To cover the remaining 347 arrests occurring in public in a fi-year p
eriod, defibrillators would have to be placed in 71 000 sites, not inc
luding outdoors and automobiles. Placement of AEDs in public locations
can be guided by the site-specific incidence of arrest.