Variation in Hospital Length of Stay in Patients With
Acute Myocardial Infarction Undergoing Primary
Angioplasty and the Need to Change the Diagnostic-
Related Group System
Beth A. Bartholomew,
MD
, Kishore J. Harjai,
MD
, Cindy L. Grines,
MD
,
Judy A. Boura,
MS
, Lori L. Grines,
PhD
, Gregg W. Stone,
MD
, David A. Cox,
MD
,
Bruce R. Brodie,
MD
, and William W. O’Neill,
MD
Our findings show significant variation in length of
hospital stay among patients with acute myocardial
infarction treated with primary percutaneous coro-
nary intervention; length of stay can be predicted
from baseline clinical and angiographic characteris-
tics. Our investigation suggests the need for a multi-
tier diagnostic-related group system for patients with
acute myocardial infarction treated with percutane-
ous coronary intervention. ᮊ2003 by Excerpta
Medica, Inc.
(Am J Cardiol 2003;92:830–833)
D
iagnostic-related groups represent a classification
system to organize medical illness into cohesive
groups that have similar length of stay and hospital
resource utilization. In the current U.S. reimbursement
system, there is 1 diagnostic-related group category
for patients with acute myocardial infarction (AMI)
treated with percutaneous coronary intervention
(PCI). This is in contrast with many other diseases and
procedures that have multiple diagnostic-related
groups. The purpose of this investigation was to eval-
uate variations in length of hospital stay (LOHS) and
assess the need for multiple diagnostic-related groups
in patients with AMI undergoing PCIs.
•••
The patient population in the Primary Angioplasty
in Myocardial Infarction (PAMI) studies enrolled pa-
tients presenting with symptoms consistent with MI
within 12 hours of onset.
1–8
Patients were included if
they had ST-segment elevation of Ն1 mm in 2 con-
tiguous leads or a nondiagnostic electrocardiogram
(including left bundle branch block, paced rhythm)
with findings of high-grade coronary stenosis and
associated ventricular wall-motion abnormalities in
the catheterization laboratory. Exclusion criteria from
the PAMI studies were: cardiogenic shock (systolic
blood pressure Ͻ80 mm Hg for Ͼ30 minutes, not
responsive to fluids), previous thrombolytic therapy
for the index infarction, stroke in the preceding month,
child bearing potential (unless recent pregnancy test
results were negative), life expectancy Ͻ1 year, cur-
rent use of warfarin, renal failure, and known contra-
indications to heparin, aspirin, or ticlopidine. All stud-
ies were conducted in accordance with the Declaration
of Helsinki principles.
To best reflect current therapies of patients with
AMI treated with PCI and the separate diagnostic-
related group categories, we utilized patients enrolled
after 1994. Additionally, we excluded these patient
subsets: those who did not undergo PCI, treatment
with only thrombolytic therapy, coronary artery by-
pass grafting during hospital stay, death within 7 days
of their AMI, patients without an length-of-stay entry,
and those at a non-U.S. site.
Data collection for the detailed case report forms
was prospectively completed by clinical coordinators
at each site. The data were reviewed at the clinical
coordinating center of William Beaumont Hospital.
Each PAMI study submitted cine angiograms to an
independent core laboratory for analysis. The Throm-
bolysis In Myocardial Infarction (TIMI) trial classifi-
cation system for myocardial perfusion scoring was
used for evaluation.
9
Congestive heart failure was
classified by Killip class and considered present for
Killip classes II to IV.
10
Length of stay was defined as
the time from admission to discharge in days.
Statistical analysis was completed using commer-
cially available statistical analysis software (SAS 8.0;
SAS Institute, Cary, North Carolina). Patients were
divided based on the last digit of their enrollment
number. All patients with even numbers were placed
in the derivation cohort and all odd numbers in the
validation cohort. The derivation cohort was utilized
for the univariate and multivariate regression analysis
to derive the predictors of an extended hospitalization
Ͼ5 days.
The correlation between length of hospitalization
and study variables in the derivation group was tested
using Wilcoxon’s rank test for categoric variables and
Spearman’s coefficient for continuous variables. Vari-
ables that had a univariate association (p Ͻ0.10) were
used in the first step of the multivariate regression
analysis. The outcome of the model was extended
hospitalization Ͼ5 days. The least significant variable
was dropped at each step until only those variables
with a p value of Ͻ0.05 remained in the final model.
A multivariate score was developed to predict the
probability of an extended hospitalization Ͼ5 days.
From William Beaumont Hospital, Royal Oak, Michigan; Lenox Hill
Hospital, Lenox Hill, New York; Mid Carolina Cardiology, Charlotte,
North Carolina; and Lebauer Cardiovascular Research Foundation,
Greensboro, North Carolina. Dr. Grines’ address is: Cardiac Cath-
eterization Laboratories, William Beaumont Hospital, 3601 West 13
Mile Road, Royal Oak, Michigan 48073. E-mail: cgrines@
beaumont.edu. Manuscript received May 12, 2003; revised manu-
script received and accepted June 16, 2003.
830
©2003 by Excerpta Medica, Inc. All rights reserved. 0002-9149/03/$–see front matter
The American Journal of Cardiology Vol. 92 October 1, 2003 doi:10.1016/S0002-9149(03)00893-2