Predicting risk of suicidality is generally difficult. We studied whet
her it is possible to predict postdischarge suicide attempts of nonpsy
chotic inpatients (n = 42) on the basis of the Hamilton Depression Rat
ing Scale (HDRS) scored during the inpatient period. Twelve patients (
29%) attempted suicide during a 3-year follow-up period. Total HDRS sc
ores during the inpatient period did not predict suicide attempts. The
re were no differences in changes of total HDRS score during the first
half of the inpatient period, but an unchanged total HDRS score durin
g the second half of the inpatient period was associated with postdisc
harge suicide attempts. Predicting lack of suicidality succeeded well
(83%) on the basis of HDRS item three on discharge. Two-thirds (63%) o
f those who attempted suicide during follow-up were assessed as suicid
al on discharge. A structured clinical assessment like HDRS item three
is reliable in predicting lack of suicidality and fairly good also in
predicting suicidality.