trends in outcome and costs of coronary intervention in the 1990s
TRANSCRIPT
Trends in Outcome and Costs of CoronaryIntervention in the 1990s
William S. Weintraub, MD, Elizabeth M. Mahoney, SCD, Ziyad M.B. Ghazzal, MD,Spencer B. King III, MD, Steven D. Culler, PhD, Douglas C. Morris, MD, and
John S. Douglas, Jr., MD
Our objective was to examine trends in outcome andcost of percutaneous coronary intervention (PCI) be-tween 1990 and 1999. PCI has become the most com-mon form of myocardial revascularization in recentyears, rivaling the more established coronary arterybypass surgery. There has been increasing interest inimproving outcome of PCI while also seeking to mini-mize cost. A total of 21,755 patients undergoing PCIwere evaluated. Clinical data were gathered from theEmory Cardiovascular Database and financial data fromthe UB92 formulation of the hospital bill. Charges werereduced to cost using departmental cost-to-charge ra-tios. Costs were inflated to 1999 dollars using medicalcare inflation rates. Mortality varied without a signifi-
cant trend from 0.63% to 0.44% (p 5 0.64). The Q-wavemyocardial infarction rate decreased from 0.68% to0.40% (p 5 0.0003). Emergent coronary surgery de-creased from 3.50% to 1.25% (p <0.0001). Mean hos-pital inflation-adjusted cost decreased from $10,478 to$8,367 (p <0.0001). Length of stay after the proceduredecreased from 2.8 to 1.8 days (p <0.0001). Outcomeof PCI continues to improve, with a decrease in coronarysurgery and Q-wave myocardial infarction but with nosignificant change in mortality. This was accomplishedwhile also decreasing costs and length of stay. Whetherthese favorable trends will continue remains to be seen.Q2001 by Excerpta Medica, Inc.
(Am J Cardiol 2001;88:497–503)
Percutaneous coronary intervention (PCI), includ-ing balloon angioplasty, intracoronary stents, and
other devices, has become the most common form ofmyocardial revascularization in recent years, rivalingthe more established coronary artery bypass grafting(CABG).1–3 Concurrent with increasing PCI has beenan escalating concern over cost and outcome. Anexamination of the outcome after PCI has been a focusof regional groups,4 professional societies,5,6 and stategovernments.7,8 Assessment of cost was a major com-ponent of 2 major trials in the United States compar-ing outcome of PCI with that of CABG.9,10 Coronarystenting has been shown to be useful in treating acuteclosure or threatened acute closure,11 as well as pre-venting restenosis.12,13 Although stents, as originallyused with coumadin, could not be shown to decreasecost,14 more recent data suggest declining cost withstenting.15 Other new therapies, especially the use ofglycoprotein IIb/IIIa blockers, have been shown todecrease cardiac events after PCI,16,17with little effecton cost.18,19Changes in in-hospital outcomes and costof PCI during the 1990s is the subject of this study.
METHODSPatient population: From January 1990 through De-
cember 1999, 21,755 patients underwent PCI atEmory University Hospitals. All patients undergoingPCI were included.
Definitions: Diabetes was based on a physician tell-ing a patient he/she has diabetes (nonstressed bloodsugar.140 mg/dl on at least 2 occasions, althoughthis cannot reasonably be verified) or the patient tak-ing or having had taken oral hypoglycemic agents,insulin, or diet therapy.Vessels diseasedwas the num-ber of vessels with$50% diameter luminal narrowingin the left anterior descending, left circumflex, or rightcoronary artery, or a major branch artery.Left maindiseasewas the presence of$50% diameter luminalnarrowing in the left main coronary artery.Coronaryanatomic featureswere lesion length and percent di-ameters stenosis measured by quantitative coronaryarteriography, whereas descriptive features includingdiffuse disease, calcium, an ulceration lesion, and abranch point lesion were judged by the operator.An-giographic successwas a decrease in postprocedurediameter stenosis to,50% and a decrease in percentstenosis by$20%.Myocardial infarctionwas new Qwaves after the procedure.Non–Q-wave myocardialinfarction was elevation in creatine phosphokinase to3 times the upper limit of normal.Variables defined bypatient historywere hypertension, severity of angina,and prior myocardial infarction. Angina was definedby the Canadian Cardiovascular Society Classifica-tion.20 Heart failure was defined by the New YorkHeart Association criteria.21
Percutaneous coronary intervention procedure anddata collection: All PCI procedures were performedusing standard techniques in use during the time of theprocedure. PCI procedures included balloon dilata-tions, stent implantation, and the use of other newdevices.
Financial data were obtained from the hospitalcomputerized billing systems. Charges were obtained
From the Division of Cardiology, School of Medicine, Health Policyand Management, School of Public Health, Emory University, Atlanta,Georgia. Manuscript received January 8, 2001; revised manuscriptreceived and accepted April 9, 2001.
Address for reprints: William S. Weintraub, MD, 1256 BriarcliffRoad, Suite 1 North, Atlanta, Georgia 30306. E-mail: [email protected].
497©2001 by Excerpta Medica, Inc. All rights reserved. 0002-9149/01/$–see front matterThe American Journal of Cardiology Vol. 88 September 1, 2001 PII S0002-9149(01)01726-X
from the UB92 (UB82 before 1993) formulation of thehospital bill.22 Costs were derived from the chargesusing both whole hospital and departmental cost-to-charge ratios. Costs presented below will be thosederived from yearly departmental cost-to-charge ratiosunless otherwise noted. All costs were inflated to 1999dollars using both the medical component of and thetotal consumer price index (CPI) inflation rate.23
Demographic, clinical, angiographic, and proce-dural data including complications were recorded pro-spectively on standardized forms and entered into acomputerized database. All fields are defined in a datadictionary.
Statistical analyses: Data are expressed as propor-tions or as mean6 SD. Trends in categorical variablesover time were analyzed using the Mantel-Haenszeltest for trend,24 and differences in continuous vari-ables by analysis of variance with a test of linear trend.Missing covariate data were imputed using an S-Plusprocedure that performs multiple imputation usingrelationships among all variables to predict the miss-ing observations.25 Multivariate correlates of discreteend points were analyzed with logistic regression.Potential nonlinear effects of each of the continuouspredictor variables were checked using restricted cu-bic splines. Correlates of hospital cost and length ofstay (LOS) were determined by linear regression. Theyear of the procedure was modeled as both a contin-uous variable as well as with indicator variables toallow for potential nonlinear effects. Hospital cost andLOS were modeled using their original scale as wellas their natural logarithms as dependent variables.Because little difference in R2 was noted using theoriginal scale or the log transform, models based onthe original scale are presented. Statistical modelingand testing were performed using S-Plus (Mathsoft,Inc. Seattle, Washington) and BMDP (SPSS Corpo-ration, Chicago, Illinois). The ability of the models todiscriminate among patients with respect to their out-comes was measured using the c-index, equivalent tothe area under the receiver-operating characteristic(ROC) curve. Bootstrap validation of the nonimputedmodels was performed using the “validate” function inthe “Design” library of S-Plus statistical functions.25
All models showed little difference in c-index or R2
when validated.
RESULTSThere were increases in mean age, percent of
women, and percent with hypertension, prior myocar-dial infarctions, or diabetes over the period studied(Table 1). Heart failure and the severity of anginavaried somewhat over the period, with little cleartrend. Heart failure remained uncommon, whereasmost patients had severe angina. There was also noclear trend in the proportion of patients with priorCABG. Mean ejection fraction decreased over theperiod, and there was fluctuation in the number ofvessels diseased. Most procedures were elective, witha small but increasing number of procedures in thesetting of acute ischemia (urgent procedures) or acutemyocardial infarction (emergent procedures). Inci-
TABL
E1
Clin
ical
Cha
ract
eris
tics
1990
(n5
2,38
0)19
91(n
52,
443)
1992
(n5
2,31
5)19
93(n
52,
010)
1994
(n5
1,94
5)19
95(n
52,
096)
1996
(n5
2,13
9)19
97(n
52,
014)
1998
(n5
2,13
6)19
99(n
52,
277)
pVa
lue
Age
(yrs
)60
611
616
1161
611
626
1162
611
626
1162
611
626
1263
612
626
12,
0.00
01W
omen
26%
26%
27%
27%
29%
29%
30%
29%
31%
30%
,0.
0001
Hyp
erte
nsio
n50
%52
%50
%55
%57
%56
%60
%64
%65
%65
%,
0.00
01Pr
ior
MI
46%
47%
47%
49%
51%
53%
51%
54%
40%
32%
,0.
0001
Dia
bete
s19
%21
%23
%21
%23
%24
%25
%25
%27
%29
%,
0.00
01H
eart
failu
re4.
4%5.
8%5.
9%6.
6%6.
2%9.
4%6.
3%6.
1%8.
6%8.
8%,
0.00
01C
lass
II–III
angi
na71
%77
%83
%85
%85
%76
%75
%78
%70
%68
%,
0.00
01Pr
ior
PCI
34%
34%
35%
38%
39%
38%
38%
38%
38%
36%
0.00
07Pr
ior
CA
BG22
%21
%24
%26
%26
%24
%22
%24
%24
%23
%0.
16Ej
ectio
nfra
ctio
n55
612
576
1155
611
546
1254
611
546
1153
611
536
1952
611
526
10,
0.00
011-
vess
eldi
seas
e50
%47
%46
%43
%41
%42
%42
%41
%40
%39
%,
0.00
012-
vess
eldi
seas
e30
%31
%29
%29
%34
%29
%32
%32
%30
%32
%0.
040
3-ve
ssel
dise
ase
21%
22%
25%
28%
25%
27%
26%
27%
29%
29%
,0.
0001
Elec
tive
94%
94%
95%
95%
95%
93%
94%
91%
90%
91%
,0.
0001
Uns
tabl
ean
gina
2.86
%3.
52%
2.81
%1.
79%
1.34
%2.
19%
1.82
%2.
73%
2.34
%3.
43%
,0.
0001
Acu
teM
I2.
69%
2.58
%2.
68%
3.23
%3.
65%
5.06
%4.
39%
6.26
%7.
41%
5.49
%0.
29C
ardi
ogen
icsh
ock
NA
NA
NA
NA
0%0.
68%
0.43
%0.
44%
0.26
%0.
25%
0.06
6
MI5
myo
card
iali
nfar
ctio
n;N
A5
nota
vaila
ble.
498 THE AMERICAN JOURNAL OF CARDIOLOGYT VOL. 88 SEPTEMBER 1, 2001
dences in which patients required intervention whilein cardiogenic shock were rare throughout.
About 1 in 4 patients had multisite procedures(Table 2). Lesion length and maximum percent diam-eter stenosis increased as did diffuseness of diseaseand thrombus at the site. For lesion calcium, ulceratedlesions, and branch point lesions, there was less of atrend. The use of stents increased dramatically andangiographic success also improved. There was somevariability in the rate of catheterization laboratorycomplications. Dissections and acute closure de-creased, while side branch closure did not change.Distal embolization, the need for transfusion, and tem-porary pacemaker decreased, while tamponade, loss ofpulse, and hypotension varied little.
Although there was a decrease in Q-wave myocar-dial infarctions, non–Q-wave myocardial infarctionsincreased (Table 3). Access site complications variedlittle. CABG and stroke decreased from 3.50% to1.25% and 0.38% to 0.18%, respectively, while deathvaried some, but without a significant trend. LOS afterthe procedure decreased by 1 day over the decadestudied. Although hospital charges have changed little(after accounting for inflation), hospital cost, derivedfrom charges using departmental cost-to-charge ratiosand inflated to 1999 dollars using the Medicare infla-tion rate, significantly declined over time, from$10,4786 7,936 to $8,3676 4,583. This represents a20% decrease overall, with decreases in cost mostyears but especially from 1994 to 1995. Cost also fellusing the total CPI inflation rate. With use of thewhole hospital cost-to-charge ratio, the cost fell from$9,2196 $7,609 to $7,9866 $4,302 over the period(p ,0.0001). The R2 correlating departmental to hos-pital cost-to-charge ratios was 0.98.
Several models were created to predict hospitalcosts. Although adjusted for other variables, the addi-tional cost attributed to any factor must be consideredan approximation because of the limited ability of themodels to account for the variability in cost. In thefirst model, clinical site and procedural variables wereincluded. The ability to account for variability in costwas low, with an R2 of 0.15. Nonetheless, the year ofthe procedure compared with the baseline cost in 1990was a powerful correlate of cost, with the decreaseranging from $173 in 1991 to $4,852 in 1999. Presen-tation with an acute coronary syndrome, use of a stent,multisite procedures, clinical factors, and several sitedescriptors also were associated with increases in cost.
In the second model, complications were added(Table 4), and the corrected R2 increased to 0.36.Coronary surgery as a complication of unsuccessfulangioplasty was the most powerful correlate, increas-ing cost by $17,650. The year of the procedure re-mained important even after accounting for the de-crease in coronary surgery over the period. In additionto the use of stents, multisite procedures, and in-creased urgency, the complications of the need fortransfusion, hypotension, stroke, death, and Q-wavemyocardial infarction also were associated with anincrease in cost. Other clinical variables had lesserimpacts. The year of the procedure remained an im-
portant correlate when the total CPI was used tocorrect for inflation. When LOS was added to themodel, the year remained a significant, although lesspowerful, variable, with the R2 rising to 0.65.
The ability to predict postprocedure LOS was mod-est (corrected R2 5 0.262), with coronary surgerybeingthe most significant correlate, adding 6.08 days toLOS and stroke adding 3.84 days. Between 1990 and1999, the postprocedure stay decreased by 1.55 days(p ,0.0001). Increased urgency, other complications,and other variables were also related to increasedLOS.
The ability to predict hospital death (Table 5) wasexcellent (ROC area 0.876). Increasing urgency wasthe most significant correlate. Other clinical correlateswere older age, female gender, heart failure, first PCI,severity of site narrowing, site thrombus, reducedejection fraction, and multivessel disease. The year ofthe procedure was not a significant correlate. Theability to predict postprocedural CABG was moremodest, with an ROC area of 0.695. The year of theprocedure was a significant correlate, with CABGdecreasing with an odds ratio of 0.90/year (95% con-fidence interval 0.86 to 0.93, p,0.0001) over theperiod of the study. Postprocedure CABG increasedsignificantly (all p ,0.01), with greater urgency ofpresentation (odds 1.78 with unstable angina, 3.16with acute myocardial infarction), lesion calcium(odds 1.93), total occlusions (odds 1.62), and mul-tivessel disease (odds 1.39 with 2-vessel disease, 1.92with 3-vessel disease), and decreased significantly (allp ,0.01) with advanced age (odds 0.88/10-year in-crease), diabetes (odds 0.74), and prior CABG (odds0.40). The ability to predict who would develop Q-wave myocardial infarction was similar to the abilityto predict CABG, with an ROC area of 0.736. Theprobability of developing a Q-wave myocardial in-farction decreased with an odds ratio of 0.90/year(95% confidence interval 0.83 to 0.97) over the pe-riod. The probability of developing a Q-wave myo-cardial infarction increased (all p,0.05) with greaterurgency (odds 1.92 for unstable angina, 3.69 for acutemyocardial infarction) at presentation, multivesseldisease (odds 1.35 for 2-vessel disease, 1.82 for 3-ves-sel disease), increasing ejection fraction (odds 1.09 fora 10% increase), longer length (odds 1.01/mm),thrombus (odds 2.03). Whether Q-wave myocardialinfarction occurring in patients with greater urgencyrepresents events that were destined to happen orcomplications associated with the procedure cannot bedetermined with certainty from these data.
DISCUSSIONThe outcome of PCI improved between 1990 and
1999, whereas LOS and cost decreased dramatically.When costs were adjusted for inflation using the totalCPI rather than the medical component, the decreasein cost was still noted, although it was less dramatic.These favorable trends occurred at a time when pa-tients became somewhat sicker and in an environmentin which attempts to control costs became a majorfocus. LOS continued to decrease as stent usage in-
CORONARY ARTERY DISEASE/OUTCOME OF CORONARY INTERVENTION 499
TABL
E2
Site
and
Proc
edur
e
1990
(n5
2,38
0)
1991
(n5
2,44
3)
1992
(n5
2,31
5)
1993
(n5
2,01
0)
1994
(n5
1,94
5)
1995
(n5
2,09
6)
1996
(n5
2,13
9)
1997
(n5
2,01
4)
1998
(n5
2,13
6)
1999
(n5
2,27
7)p
Valu
e
Mul
tisite
32%
27%
27%
23%
27%
24%
24%
25%
27%
28%
0.00
32
Vein
graf
t8.
5%8.
5%10
.0%
10.6
%11
.0%
9.5%
8.7%
11.8
%11
.1%
9.9%
0.00
05
Max
lesi
onle
ngth
(mm
)7.
16
6.9
6.8
64
7.6
66.
48.
56
7.2
8.8
66.
110
.86
7.4
11.6
614
.411
.96
8.3
12.9
68.
714
.56
10.5
,0.
0001
Max
%ste
nosi
s79
612
806
1082
611
836
1084
69
846
1184
611
836
1283
611
846
10,
0.00
01
Tota
locc
lusi
on10
%9%
10%
8%10
%10
%11
%12
%12
%11
%,
0.00
01
Diff
use
dise
ase
30%
27%
23%
42%
41%
35%
31%
39%
37%
36%
,0.
0001
Lesi
onca
lciu
m12
%10
%7%
15%
15%
17%
17%
22%
17%
13%
,0.
0001
Thro
mbu
s14
%10
%10
%11
%14
%17
%17
%18
%16
%19
%,
0.00
01
Ulc
er15
%11
%10
%24
%18
%15
%18
%22
%15
%15
%,
0.00
01
Bran
chpo
int
35%
31%
22%
19%
22%
28%
36%
36%
29%
31%
0.00
03
Sten
t1.
9%6%
4.5%
3%6.
4%10
.4%
31.7
%51
.9%
65.7
%79
.4%
,0.
0001
Ang
iosu
cces
s92
%93
%94
%93
%93
%94
%96
%95
%95
%96
%,
0.00
01
Any
in-la
bco
mpl
icat
ion
11.8
%10
.6%
12.2
%10
%11
.1%
10.8
%12
.1%
10%
8.5%
7.4%
,0.
0001
Dis
sect
ion
6.72
%5.
94%
7.26
%5.
57%
6.99
%6.
2%7.
67%
5.46
%5.
29%
5.09
%0.
015
Acu
tecl
osur
e3.
24%
2.91
%2.
16%
2.04
%2.
26%
1.81
%1.
5%1.
29%
1.22
%1.
36%
,0.
0001
Side
bran
ch1.
39%
1.76
%1.
64%
0.9%
1.34
%1.
38%
1.5%
1.44
%1.
08%
1.10
%0.
12
Clo
sedi
stale
mbo
lizat
ion
0.92
%0.
65%
1.94
%2.
54%
1.13
%1.
57%
1.59
%1.
24%
0.33
%0.
22%
0.00
11
Tran
sfus
ion
0.8%
0.57
%0.
6%0.
3%0.
31%
0.57
%0.
37%
0.25
%0%
0%,
0.00
01
Tem
pora
rypa
cem
aker
0.46
%0.
37%
0.6%
0.4%
0.21
%0.
72%
0.89
%0.
65%
0.19
%0.
09%
0.31
Tam
pona
de0%
0.08
%0%
0%0%
0%0.
05%
0%0.
14%
0.04
%0.
19
Loss
dista
lpul
se0.
17%
0.16
%0.
17%
0%0.
31%
0.14
%0.
09%
0.2%
0.8%
0.09
%0.
034
Hyp
oten
sion
0.29
%0.
49%
0.6%
0.35
%0.
26%
0.52
%0.
61%
0.89
%0.
42%
0.13
%0.
93
Ang
io5
angi
ogra
phic
;Max
5m
axim
al.
500 THE AMERICAN JOURNAL OF CARDIOLOGYT VOL. 88 SEPTEMBER 1, 2001
creased, reflecting the use of platelet blockers insteadof full anticoagulation and shorter stay with stents inrecent years. The overall decrease in cost over timereflects the net effect of several factors: there havebeen significant efforts to control costs as well asdecreases in the rate of complications that contributeto higher cost. However, over the same period of time,patients have been presenting with clinical character-istics associated with higher cost of treatment. Thereremain factors associated with declining costs that areunaccounted for, and which suggest decreased inten-sity of care. Importantly, this patient population hasbeen reasonably stable, and if there had been a muchlarger increase in PCI for acute myocardial infarction,the favorable trend for costs might have been obscuredfurther.
Since PCI was introduced in the late 1970s, theprocedure has grown in popularity, and there havebeen many technical improvements. Most recentlythere has been explosive growth in coronary stent-ing.11,12 Randomized trials have helped to define therelative merits of PCI versus medical therapy26 and ofPCI versus CABG.27,28 Concurrently, there has beenincreasing concern over the cost of medical care andwhether society would be able to afford to providehigh-quality medical care in the future as the popula-tion ages. The response has been the development ofthe managed care industry, in which various newstructures have been created to try to control cost.Providers have responded to this environment by de-veloping care maps and other strategies to try tocontrol the cost of services. Most of these strategieshave been used at Emory University Hospitals, includ-ing care maps (instituted in the mid-1990s), moreaggressive seeking of discounts on prices, and scalingback intensity of care when this was considered safe.A frequently voiced concern is that as services areoffered at lower prices, the result will be a steadydegradation in quality.29 The results presented herewould suggest that so far this has not happened.
This study is limited by being conducted at 1institution and may have uncertain generalizability.Not all technical details are available, including gly-coprotein IIb/IIIa use. Only hospital costs were ana-lyzed, because professional costs were not availablefor most patients. The study may be limited further byits use of hospital costs as a proxy for societal costs.Hospital costs were derived in this study using depart-mental cost-to-charge ratios. These ratios are devel-oped based on cost guidelines of the American Hos-pital Association and are reported to the Health CareFinancing Agency in a yearly hospital cost report,which is required of hospitals for Medicare reimburse-ment. Whether costs derived from charges in thismanner reflect true underlying costs of the procedurecannot be further resolved in a meaningful way. Infact, in a mixed product environment, there are alwayscross subsidizations that cannot be fully accountedfor. The correct inflation rate correction, whether thetotal CPI or the medical component, is uncertain.Medical care costs have increased more rapidly, andthus the results using the total CPI show less decrease
TABL
E3
Hos
pita
lOut
com
ean
dC
osts
1990
(n5
2,38
0)19
91(n
52,
443)
1992
(n5
2,31
5)19
93(n
52,
010)
1994
(n5
1,94
5)19
95(n
52,
096)
1996
(n5
2,13
9)19
97(n
52,
014)
1998
(n5
2,13
6)19
99(n
52,
277)
pVa
lue
Non
–Q-w
ave
MI
3.87
%2.
69%
3.03
%3.
34%
3.41
%3.
09%
2.78
%3.
79%
5.03
%4.
82%
0.01
8
Q-w
ave
MI
0.68
%0.
9%0.
91%
0.9%
0.73
%1.
15%
0.38
%0.
2%0.
28%
0.40
%0.
0003
Qor
Non
-QM
I3.
87%
2.69
%3.
03%
3.34
%3.
41%
3.09
%2.
78%
3.79
%5.
03%
4.82
%0.
16
Vasc
ular
com
plic
atio
n
0.13
%0.
12%
0.22
%0.
15%
0.42
%0.
33%
0.23
%0.
2%0.
19%
0.27
%0.
28
Cor
onar
ysu
rger
y3.
50%
2.66
%3.
26%
2.36
%3.
12%
2.24
%1.
41%
1.54
%0.
99%
1.25
%,
0.00
01
Stro
ke0.
38%
0.16
%0.
52%
0.15
%0.
47%
0.19
%0.
05%
0.1%
0.09
%0.
18%
0.00
81
Dea
thin
hosp
ital
0.63
%0.
53%
0.69
%0.
8%0.
62%
0.52
%0.
42%
0.89
%1.
03%
0.44
%0.
64
LOS
(d)
2.8
64.
02.
86
3.11
3.0
63.
872.
66
3.5
2.7
63.
62.
66
3.3
1.9
62.
252.
06
3.0
1.9
62.
91.
86
3.0
,0.
0001
Tota
lLO
S(d
)4.
16
4.8
3.8
63.
84.
06
4.4
3.6
64.
03.
76
4.1
3.6
63.
62.
76
2.7
2.7
63.
52.
66
3.2
2.4
63.
3,
0.00
01
Hos
pita
lcha
rge
$15,
115
6$1
2,71
1$1
4,76
26
$12,
581
$16,
275
6$2
4,98
3$1
6,28
56
$15,
203
$15,
946
6$1
1,17
2$1
5,34
46
$9,7
88$1
5,31
46
$8,2
29$1
6,03
86
$9,7
75$1
6,13
16
$8,8
92$1
6,92
96
$9,2
68,
0.00
01
Hos
pita
lcos
t,
med
ical
CPI
$10,
478
6$7
,936
$10,
389
6$8
,333
$10,
047
6$8
,531
$9,
962
6$9
,095
$10,
315
6$6
,826
$7,
959
6$0
33$
8,08
76
$4,6
55$
7,95
66
$5,0
24$
7,86
16
$4,4
22$
8,36
76
$4,5
830.
0001
Hos
pita
lcos
t,
tota
lCPI
$8,
802
6$6
,663
$9,
130
6$7
,323
$9,
150
6$7
,766
$9,
303
6$8
,490
$9,
844
6$6
,514
$7,
692
6$4
,864
$7,
789
6$4
,480
$7,
742
6$4
,889
$7,
785
6$4
,380
$8,
367
6$4
,583
,0.
0001
Abb
revi
atio
nas
inTa
ble
1.
CORONARY ARTERY DISEASE/OUTCOME OF CORONARY INTERVENTION 501
in cost. The argument for the total CPI is that it reflectscompetition for all goods and services in society,whereas the argument for the medical component ofthe CPI reflects competition with other health caregoods and services. All this being said, the trendstoward improved results at lower cost and thus lowerresource use remain clear. However, the subsidizationprobably did not vary tremendously over the period,because results using whole hospital and departmentalcost-to-charge ratios were similar.
Whether the favorable trends noted in this studycan be maintained is unknown. After a certain point,presumably, further efforts to increase efficiency mayresult in less favorable results. Costs could perhaps bereduced further if complications were further reduced.
This cannot be easily accomplishedby patient selection, because the abil-ity to predict complications is mod-est (ROC curve area 0.695 forCABG, 0.736 for Q-wave myocar-dial infarction). However, techno-logic advances may reduce compli-cations further.
It is a societal demand and will-ingness to pay for PCI that will limitattempts to decrease cost. Given theevidence supporting the effective-ness of PCI, there is currently toomuch demand for the service to re-alistically expect a decrease in thisservice by a large amount in the nearterm. To the extent that Emory dataare reflective of national trends, agood faith effort to respond to soci-etal forces is being made. If expen-sive services are to continue to beavailable at an appropriate level, so-ciety will need to continue to be will-ing to pay for them.
Acknowledgment: The investiga-tors thank Lesley Wood for her careful editorial re-view.
1. Weintraub WS, Jones EL, King SB III, Craver J, Douglas JS Jr, Guyton R,Liberman H, Morris D. Changing use of coronary angioplasty and coronarybypass surgery in the treatment of chronic coronary artery disease: the Emoryexperience.Am J Cardiol1990;65:183–188.2. Detailed diagnosis and procedures for patients discharged from short-stayhospitals: United States, 1987.Vital Health Stat1989;100:181.3. Detailed diagnosis and procedures: National Hospital Discharge Survey, 1993.Vital Health Stat13 1994;122:121.4. Malenka DJ. Indications, practice, and procedural outcomes of percutaneoustransluminal coronary angioplasty in Northern New England in the early 1990s.Am J Cardiol1996:78:260–265.5. Grassman ED, Johnson SA, Krone RJ. Predictors of success and majorcomplications for primary percutaneous transluminal coronary angioplasty inacute myocardial infarction: an analysis of the 1990 to 1994 society for cardiacangiography and interventions registries.J Am Coll Cardiol1997:30:201–208.6. Weintraub WS, McKay CR, Riner RN, Ellis SG, Frommer PL, CarmichaelDB, Hammermeister KE, Effros MN, Bost JE, Bodycombe DP, for the AmericanCollege of Cardiology Database Committee. The American College of Cardiol-ogy National Database: progress and challenges.J Am Coll Cardiol 1997;29:459–465.7. Hannan EL, Arani DT, Johnson LW, Kemp HG Jr, Lukacik G. Percutaneoustransluminal coronary angioplasty in New York State: risk factors and outcomes.JAMA 1992;268:3092–3097.8. Ritchie JL, Phillips KA, Luft HS. Coronary angioplasty: statewide experiencein California.Circulation 1993;88:2735–2743.9. Weintraub WS, Mauldin PD, Becker ER, Kosinski AS, King SB III. Acomparison of the costs of and quality of life after coronary angioplasty andcoronary surgery: results from the Emory Angioplasty vs Surgery Trial (EAST).Circulation 1995;92:2831–2840.10. Hlatky MA, Rogers WJ, Johnstone I, Boothroyd D, Brooks MM, Pitt B,Reeder G, Ryan T, Smith H, Whitlow P, Wiens R, Mark DB. Medical care costsand quality of life after randomization to coronary angioplasty or coronary bypasssurgery.N Engl J Med1997;336:92–99.11. George BS, Voorhees WD III, Roubin GS, Fearnot NE, Pinkerton CA,Raizner AE, King SB, Holmes DR, Topol EJ, Kereiakes DJ, Hartzler GO.Multicenter investigation of coronary stenting to treat acute or threatened closureafter percutaneous transluminal coronary angioplasty: clinical and angiographicoutcomes.J Am Coll Cardiol1993;22:135–143.12. Fischman DL, Leon MB, Baim DS, Schatz RA. Savage MP, Penn I, Detre K,Veltri L, Ricci D, Nobuyoshi M, et al, for the Stent Restenosis Study Investiga-tors. A randomized comparison of coronary-stent placement and balloon angio-plasty in the treatment of coronary artery disease.N Engl J Med1994;331:496–501.
TABLE 4 Change in Cost, Corrected for Baseline and Procedural Variables andComplications
AdditionalCost ($) 95% CI p Value
Medical CPI (R2 5 0.36)1991 46 2269 to 360 0.77661992 2433 2754 to 2112 0.00811993 2502 2838 to 2167 0.00331994 2482 2827 to 2137 0.00621995 22,438 22,833 to 22,044 ,0.00011996 22,463 22,868 to 22,058 ,0.00011997 23,390 23,799 to 22,981 ,0.00011998 24,022 24,412 to 23,631 ,0.00011999 24,076 24,478 to 23,675 ,0.0001
Total CPI (R2 5 0.36)1991 449 160 to 738 0.00231992 342 48 to 637 0.02281993 505 197 to 812 0.00131994 740 423 to 1056 0.00001995 21,059 21,421 to 2697 0.00001996 21,113 21,485 to 2742 0.00001997 21,908 22,284 to 21,533 0.00001998 22,368 22,727 to 22,010 0.00001999 22,309 22,677 to 21,940 0.0000
CI 5 confidence interval.
TABLE 5 Correlates of Hospital Death (ROC curve area0.876)
VariableOddsRatio 95% CI p Value
Age (per 10 yrs) 1.61 1.36–1.90 ,0.0001Women 2.47 1.73–3.52 ,0.0001Heart failure 3.44 2.29–5.18 ,0.0001Urgent due to unstable
angina3.19 1.64–6.22 0.0007
Emergent due to acuteMI
5.89 3.70–9.35 ,0.0001
Prior PCI 0.53 0.35–0.81 0.0031Percent diameter stenosis
(10% increase)1.36 1.13–1.64 0.0010
Thrombus at site 1.62 1.09–2.40 0.0162Ejection fraction (10%
increase)0.81 0.70–0.94 0.0044
2-vessel disease 1.65 0.97–2.81 0.06583-vessel disease 2.54 1.57–4.08 0.0001
Abbreviations as in Tables 1 and 4.
502 THE AMERICAN JOURNAL OF CARDIOLOGYT VOL. 88 SEPTEMBER 1, 2001
13. Serruys PW, de Jaegere P, Kiemeneij F, Macaya C, Rutsch W, HeyndrickxG, Emanuelsson H, Marco J, Legrand V, Materne P, et al. A comparison ofballoon-expandable-stent implantation with balloon angioplasty in patients withcoronary artery disease.N Engl J Med1994;331:489–495.14. Cohen DJ, Krumholz HM, Sukin CA, Ho KKL, Siegrist RB, Cleman M,Heuser RR, Brinker JA, Moses JW, Savage MP, Detre K, Leon MB, Baim DS.In-hospital and one-year economic outcomes after coronary stenting or balloonangioplasty: results from a randomized clinical trial.Circulation 1995;92:2480–2487.15. Blanco J, Weintraub WS, Shen J, Chronos H, McAdams D, Chronos N,Ghazzal ZMB, King SB III. Cost impact of high pressure inflations after therapywith aspirin and ticlopidine after coronary stenting (abstr).J Am Coll Cardiol1998;31(suppl):17A.16. The EPIC Investigators. Use of a monoclonal antibody directed against theplatelet glycoprotein IIb/IIIa receptor in high-risk coronary angioplasty.N EnglJ Med1994;330:956–961.17. The RESTORE Investigators. Effects of platelet glycoprotein IIb/IIIa block-ade with tirofiban on adverse cardiac events in patients with unstable angina oracute myocardial infarction undergoing coronary angioplasty.Circulation 1997;96:1445–1453.18. Mark DB, Talley JD, Topol EJ, Bowman L, Lai Choi L, Anderson KM, JollisJG, Cleman MW, Lee KL, Aversano T, Untereker WJ, Davidson-Ray L, CaliffRM. Economic assessment of platelet glycoprotein IIb/IIIa inhibition for preven-tion of ischemic complications of high-risk coronary angioplasty. EPIC Investi-gators.Circulation 1996;94:629–635.19. Weintraub WS, Culler S, Boccuzzi SJ, Cook JR, Kosinski J, for the Restore TrialStudy Group. Economic impact of GP IIB/IIIA blockade after high-risk angioplasty:results from the RESTORE Trial.J Am Coll Cardiol1999;34:1061–1066.
20. Campeau L. Grading of angina pectoris [letter].Circulation 1975;54:522–523.21. The Criteria Committee of the New York Heart Association: Nomenclatureand criteria for diagnosis of diseases of the heart and great vessels. 8th ed. Boston:Little, Brown, 1979.22. Weintraub WS, Mauldin PD, Becker ER, Kosinski AS, King SB III. Acomparison of the costs and quality of life after coronary angioplasty or coronarysurgery for multivessel coronary artery disease: results from the Emory Angio-plasty versus Surgery Trial (EAST).Circulation 1995;92:2831–2840.23. http. //www.economagic.com/blscu.htm.24. Mantel N. Chi-square tests with one degree of freedom: extensions of theMantel-Haenszel procedure.J Am Stat Assoc1963;58:690–700.25. Harrell FE. Design. S functions for biostatistical/epidemiological modeling,testing, estimation, validation, graphics, prediction, and typesetting. Available at:[email protected]. Parisi AF, Folland ED, Hartigan P, on behalf of the Veterans Affairs ACMEInvestigators. Comparison of angioplasty with medical therapy in the treatment ofsingle-vessel coronary artery disease.N Engl J Med1992;326:10–16.27. King SB III, Lembo NJ, Weintraub WS, Kosinski AS, Barnhart HX, KutnerMH, Alazraki NP, Guyton RA, Zhao XQ. A randomized trial comparing coronaryangioplasty with coronary bypass surgery: the Emory Angioplasty versus SurgeryTrial. N Engl J Med1994;331:1044–1050.28. The Bypass Angioplasty Revascularization Investigation (BARI) investiga-tors. Comparison of coronary bypass surgery with angioplasty in patients withmultivessel disease.N Engl J Med1996;335:217–225.29. Campion FX, Rosenblatt MS. Quality assurance and medical outcomes in theera of cost containment.Surg Clin North Am1996;76:139–159.
CORONARY ARTERY DISEASE/OUTCOME OF CORONARY INTERVENTION 503