Drug overdose (OD) is one of the most common single diagnoses admitted to m
edical intensive care units (MICUs). The diagnosis results in relatively li
ttle morbidity or mortality, suggesting a need to improve the methods utili
zed in deciding on MICU admission. Our objective in this study was to devel
op a quantitative system whereby the emergency room (ER) physician, with re
asonable safety and ease, would be able to discriminate between the need fo
r an ICU and a nonmonitored bed. We reviewed the charts of 216 consecutive
MICU admissions for intentional OD involving 199 patients between the years
1995 and 1998. Clinical histories, vital signs, laboratory data inclusive
of toxicologic analysis, and both APACHE II and Glasgow coma scores (GCS) w
ere assessed from the ER and on transfer to the MICU 4-6 hours later. These
scores, as well as individual components of the APACHE II score, were eval
uated for significance. Of 216 admissions reviewed, 75 (35%) had MICU-requi
ring morbidity: intubation 61%, pneumonia 20%, arrhythmia or EKG changes 20
%, and hypotension 3%. Mortality was 2.7%. The remaining admissions were fo
r ICU monitoring due to lethargy, irritability, laboratory abnormalities, o
r simply based on the diagnosis of drug overdose. Urine examinations for dr
ugs were positive in only 53%, with the most common agent identified being
benzodiazepines (39%). Age, Apache II score, and GCS were significantly dif
ferent between those patients who developed MICU-requiring morbidity and th
ose who did not, as well as when comparing the morbidity with the mortality
group. Receiver operator control (ROC) curves reveal that both the APACHE
and GCS are excellent and equal predictors of morbidity, with a GCS of less
than or equal to 12 having 88% sensitivity and 92% specificity in predicti
ng MICU-requiring morbidity. We conclude that ER evaluation of GCS can be u
sed to accurately assess and predict the need for MICU monitoring in drug o
verdose. OD patients with a persistent GCS of greater than 12 or who do not
demonstrate any hemodynamic, infectious, or electrocardiographic complicat
ions in the ER do not require MICU admission.