Objective - To investigate whether an association exists between midaz
olam use and serious cardiorespiratory events or death. Design - Retro
spective analysis of data in an inpatient record linkage database, col
lected between March 1986 and October 1987 from 14 hospitals in the Un
ited States. Patients - A cohort of 19,112 patients who received injec
table midazolam or diazepam on the same day that a medical procedure w
as performed. Main outcome measures - An attempt was made to identify
suspected serious cardiorespiratory adverse events and deaths that occ
urred within 24 h of study-drug administration. Death rates within 24
h following study-drug administration were compared between patients w
ho received injectable midazolam and those who received injectable dia
zepam. Results - Validation analysis of the information in the compute
rized record linkage database indicated that serious cardiac and respi
ratory adverse events that occurred within 24 h of study-drug administ
ration could not be reliably identified. Therefore, a comparison of th
e serious cardiorespiratory adverse event rates following administrati
on of midazolam versus diazepam could not be made. Death rates within
24 h of study-drug administration could be evaluated. For the cohort a
s a whole, the death rate was significantly lower among patients who r
eceived midazolam than among those who received diazepam (0.76% versus
1.93%, p<0.01). This difference remained statistically significant, a
fter adjusting for age, sex, comorbidity diagnosis, concomitant drug u
se, type of medical procedure, and hospital size and teaching capabili
ty. Conclusions - The results suggest that there is no increased risk
of death associated with midazolam administration when compared to dia
zepam administration for endoscopic, conscious sedative, and general a
nesthetic procedures in hospitals. (C) 1997 by John Wiley & Sons, Ltd.