To determine whether AIDS surveillance misses a substantial number of perso
ns who die with unreported AIDS, we conducted a cross-sectional survey of d
ecedents examined by the San Francisco (SF) Medical Examiner. Decedents who
received toxicology screening were tested for HIV antibody and examined fo
r evidence of AIDS. Names of decedents with positive or indeterminate HIV a
ntibody test results were cross-referenced against the SF AIDS registry to
identify previously reported AIDS cases. Medical records of unreported case
s were reviewed to determine whether AIDS had been diagnosed prior to death
. Of 1959 decedents tested, 176 (9%) were HIV positive; 105 (60%) were iden
tified as having AIDS by the Medical Examiner. Of the 105 AIDS cases, 101 (
96%) had been previously diagnosed; 98 (97%) had been previously reported.
Overall, diagnosis and reporting were 93% complete. HIV-infected decedents
were more likely than those uninfected to be men and < 45 years old, and le
ss likely to be Asian/Pacific Islander or Native American (p < .001). They
were more likely to have died of suicide (p < .05) or drug abuse/overdose (
p < .001). In SF, AIDS case reporting is highly complete. Current surveilla
nce activities, which identify cases from health care settings, are appropr
iate. To decrease deaths among HIV-infected persons, suicide prevention and
substance abuse treatment programs are needed.