Background. The broad picture of intensive care unit (ICU) outcomes an
d expenditures cannot be discerned from previous studies that were con
ducted at single hospitals and focused on narrow subsets of patients,
Methods. This study provides a comprehensive national profile of ICU u
sed by Medicare patients with cancer. The data source was the Medicare
Provider Analysis and Review file for fiscal year 1990, representing
100% of all hospital admissions that occurred within 723 ICD-9-CM code
s and organized into 11 code groups. Using screening criteria, admissi
ons were categorized as surgical(both major and minor procedures) or n
onsurgical (no procedures) and with and without involvement of the ICU
. The categories were compared using the following outcome variables:
total hospital charges, ICU charges, ancillary charges, average length
of stay, and in-hospital mortality. Results. This study population ac
counted for nearly 800,000 admissions, of which 143,458 (18.1%) involv
ed the use of the ICU. Actual ICU charges represented 4.9% of the $9.3
billion in total hospital charges. Intensive care unit use is associa
ted positively with service intensity, and 73% of all the admissions i
nvolving the ICU were for major procedures. Only 2% involved no proced
ures. Admissions involving use of the ICU generate higher charges and
longer lengths of stay than non-ICU admissions, although the differenc
es decrease with declining treatment intensity and resource use. In-ho
spital mortality rates, for those cases that used the ICU, were 9.8% f
or major procedures, 21.2% for minor procedures, and 37.6% for cases i
nvolving no procedures. Conclusions. Contrary to the conclusions drawn
from previous research, these findings suggest that patients who rece
ive less intense service and use fewer hospital resources are more lik
ely to die in the hospital than those who receive more care, with or w
ithout a stay in the ICU during the hospitalization. A global view of
ICU use does not support the conclusion that a disproportionate share
of special care resources is expended on futile care of the terminally
ill or excessive monitoring of low risk patients, although these prob
lems undoubtedly exist. Analysis of comprehensive national data regard
ing the use of intensive care provides a perspective that challenges s
ome of the conclusions based on more limited studies that were conduct
ed in single hospitals and focused on nonsurvivors or subsets of patie
nts narrowly defined in other ways.