eetine
se mortality in the United 3 and 1985(I). A similar ortality was observed in ul metropolitan area during that time. ronary heart disease deat te in 30- to 74-year old women declined from 182.3/100, in 1963to . In similarlyaged men, the rate declined 97.0/100,ooO in from 582.51100, in 1963to 297.1/100,009in 19g5(2). The relative contributions to this decline are hypothesized and include changes in lifestyles, risk factors and medical care. During this time the use of coronary artery bypass graft From the Division of Epidemiology. School of Public Health, University of Minnesota, Minneapolis, Minnesota: t DukeUniversity Medical Center, Durham, North Carolina: *Bowman Gray School of Medicine, WinstonSalem, North Carolina. This study was supported by Grant ROI HL23727 from the Nctional Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland and by Grant HS-06503 from the Agency for Health Care Policy and Research, Rockville. Flaryland. Manuscript received August 2,1993: revised manuscript received January 27, 1994. accepted February IO, 1994. Address for COE-: Dr. Katherine M. Doliszny, Division of Epidemiology, School of i’ublic Health, University of Minnesota, 1300 South Second Street, Suite 300, Minneapolis, Minnesata 554~4-1015. 81994 b;r the American College of Cardiology
first iotroduced in 1968(3 roced~res in 1971(4)to 228, of this therapeutic app prolonginglife have prompted s mate the co~t~botion of coron gery to the decline in corona (4,6-9). The estimated contribution of coronary artery bypass graft surgery to the decline in coronary heart disease mortality ranges from 1%to 42%,with most reports clustering around 3% to 5%. These estimates, however, are based on speculations about both the real frequency of coronary artery bypass graft surgery and on survival differences between patients treated surgically and medically. T estimates were derived from randomized clinical trials enrolled a highly select group of patients, or and did not incorporate the simultaneous graphic and clinical characteristics affecting the survival of surgicallyor medically treated patient cal outcomes in smaller community r from those of academic centers that participate in randomized trials. Finally, they failed to account for the 0735-1097/94/$7.00
%
DOLlSZNYETAL. CABGANDCORONARYHEARTDISEASEMORTALlTYDECLiINE
dynamic changes in case selection, use of SWWY ad improved operating techniques over time. ‘the unique features of this study are the use of actual lisfrequmcy and survivaldata from residentsof the St. Paul metropolitanarea (n = 2 million)undergomgCOCOEW artery bypass grafl surgery between 1970and coronary heart disease mortality data from the populationduring this same period and a data-base edical survival that assumed coronary artery bypass surgery was not available. The difference between rates was then used to calculate the estimated contrisurgery to the deciin~ bution of coronary artery bserved in ~~innea in coronary heat disease &-St. Paul ~tw~en 197
JACC Vol. 24.N~. II July I :%%-I03
cohort was fQ~Iowed
ale population in the s~ven~coMnty larea. Prom 1971to 1979andfrom 1981 ion was used to estimate population heart disease mortality for sed on the basis d death Minnesota Department of codes 410-414, 427-429 and 1984,as ICD-9 codes 410-414,
and between 1979 and
nesota, were enumerated Addrtional data from the hospit& record were collected for a le of coronary artery bypass tients medical history and physical ex~rnlnat~on, chest roentgenogram, electrocardiogram cardiac catheterization and coronary artery bypass kgery. A stratified sampling technique was used to this 15% sample, which was proportional to the numberof surgeries performed in a given vear at a given hospital. up. All patients were followed up to determine vital status using the National Death Index for events occurring after 1979and MINNDEX, a Minnesotadex. for delrhs occurring system has been validated x and was found to agree on the survival status of 9g% of the people tested (10). Deaths were classified as either coronary heart disease related or ~~oron~ heart disease related. Codes used for underlying cause of death were identical to those for the population data. Patients were assumed to be alive on December 31, 13g5
previously by these investigators(I l-15). Important pr inneapolis-St. dais base and the D eases were included in Cox regression a using data from 2,864 consecutive, sym~to treated patients with angiographicallyproved coronary artery disease evaluated at Duke University between 1969and 1984.These patients underwent medical treatment for at least 3 months and did not have 1)congenital heart disease, 2) primary valve disease, 3) primary cardiomyopathy, 4) acute evolving myocardfal infarction, 5) postinfarction ventticular septal defects, 6) previous coronary revascularization, and 7) previous coronary angioplasty. Cox regression analyses were performed on a randomly selected two-thirds of the population and validated in the remaining one-third. The final model used in this analysis was based on 22 characteristicsfrom the history, physical examination, chest X-ray film, ECC and cardiac catheterization and included a time-of-catheterization variable. It should be noted t whereas the Duke data were c~l~ec~e~pros time of catheterization, the Minneapolis-St. Paul data were cotlected retrospectively. Every etIort was made, however, to abstract the Minneapoli~St. Paul patient informationand
catbeterizat~onfor the ent
ry artery bypass graft surgery sarn~~c. s co~t~ib~t~ by a given by~othet~c m a given year were summed. Therefore, the e mean of the indimber of the cohort.
Duke ‘Universitydata base accurately r survival of Minneapolis-St. Paul patients, rience of a sample of medically treated Minneapolis-St. Paul ts was determined. These patients underwent cardiac res but were not immediately referred to coross graft surgery. If coronary artery bypass after cathete~~atio~, it must e of the “surgical window,” defiled as the number of days after catheterj~at~o~within inneapolis-!St. Ml patients ~~de~we~~ coror.ary artery bypass graft surgery. The surgical window was determined for each hospital, during three different periods: 197Gto 1974,1975to 1979and 1980to 1984.
e total number of hypothetic medical deaths in 1974 would include deaths in patients catheterized in 1970, 1971, 1972, 1973 and 1974, res tively. The rates used to calculate the ann%alpercent contribution of coronary artery bypass graft slirgcry to the decline in coronary heart disease mortality and an outline of the computational method used are described in the ~~~e~d~x.
bypass graft All hospitals p ok-St. Paul surgery between participated in the enumeration of coronary artery graft surgery patients. Of the 13,995coronary artery bypass graft surgery procedures performed during this period, 9,548 (68.2%) were eligible for inclusion, 4,268 (30.5%) :a!ere
JACC Vol. 24. No. 1 July I :95-103
DOLISZNYETAL.
98
CABG AND CORONARY HEART DISEASE MORTALITY DECLINE
TOTAL
120 100
eo 60
WO
40 20 0
ary artery
. Paul raid
ineligible, and 179 (1.3%) primary reason for inelig
records were missi was residency out
rate increased from 19.4 MM. Although the actual rates over time were signific~tly n compared with men, the percent of rate ater in women. There was a steady increase surgery until 1980, in the rate of coronary artery bypass g after which the rate was level. of the coronary artery bypass patients increased over time from 55 years in 1970 to 61 years in III he percent of coronary artery su 265 years iuc by 111%. TRble I ilh tes the chart distribution of coronary ns in the neapolis-!&. 1 coronary artery bypass surgery sample. Over time, significantly fewer patients with one-vessel disease were referred to surgery, whereas
outcome within the 1st 30 d
Furthermore, surv and female patients. w tic medical survival estimates were calcula for 1 inneapo1is-S. Paul coronary artery by surgery patients for whom ext clinical and surgical data were collected who met t uke University mo inclusion criteria. A comparison of clinical characteristics between the sample of the 247 inneapok-St. Paul medically treated
1. AngiographicExtent of Coronary Artery Disease(270% Stenosis) in 1,408 Patients rgoingCoronaryArteryBypassGrd?Surgeryby 5-YearPeriods:The Minnesota HeartSurvey Period 1970-1974 (n=464)
No diseasedvessels One-vesseldiscrae Twcwesseldisease Three-vesseldisease Leftmaincoronaryartery?
No. of Pts
(%I
7 IO8 I60 I89 52
1.5 23.3 34.5 40.7 II.2
*Linear test of trendovertime. kSO%
1975-1979 (n=46W No. of Pts 112 94 166 I88 a4
stesoses.pts = patients.
(%I
No. of Pts
(W
p Value
2.6 20.4 36.1 40.9 18.3
6 79 I73 226 I00
1.2 16.3 35.7 46.7 20.7
NS 0.007* NS NS O.lwoI
dell versus Kaplan-Meier eapolis-St. Paul medically
treated patients.
ify the co~tr~b~t~~~ of to the decline in coroase rnQrta~~ty, an adjusted estimate of the artery bypass gr t disease mortal
truncated at 12 years. Figure 5 shows a receiver-operating characteristic curve (20) for the use ting 3-year survival of the 1medically treated patients. T is O.g3,indicating very good discrimination. Table 3 presents the contribution of coronary artery bypass graft surgery to the decline in coronary mortality, expressed as a percent, for each yeai between 1990and B9M.In 1990,coronary artery bypass graft surgery did not contribute to the disease mortality rate; rather, the actual nu experienced by the surgicalcohort during tha cted number of medical deaths, resufting in a uegative percent difference. From 1991to 1984 the percent contribution increased from 0.2% to +6.6%. Because the Duke model tended to overestimate the mortality cf the medically treated Mnneapolis-St. Paul
Patients With Coronary Heart Disease: The Minnesota Heart Survev
SurvivalEstimates(%I Survival At I yr At 2 yr At 3 yr Aa 4 yr At 5 yr At 6 yr At 7 yr At 8 yr At 9 yr At 10 yr At 11 yr
At 12yr
--_
Observed
Predicted
89.9 85.7 80.3 74.7 72.6 69.3 66.6 63.8 61.1 5b.7 53.6
90.0 84.5 79.0 74.9 41.7 68.8 ti5.4 6X 60.2 56.9 54.8
-0.1 t1.3 t1.3 -0.2 to.7 to.5 t1.2
53.6
50.9
t2.7 __
*Usingthe Duke model.
Diiwence
to.6 +0.9 t1.8
t
100
DGLISZNY ET AL. CABG AND CORONARY HEART DISEASE MORTALITY DECLINE
tween 1970and 19 bypass graft surgery heart disease mo contribution bad
JACC Vol. 24, No. 1 July 199495-103
s, c~r~~a~yartery ence on CQr~~ar~
c~~r~cte~5tic curve for the Duke modelin predictingJ-yearsurvival of the sample ofMinneapolis-gt. Paul medically treated patients.
was cah~iated. The ~nt~u~~ num
ividingthe numbe (the mean annual difference icted medical surv Table 2), and this value was used in the computation of the adjusted estimate of the percent contribution of coronary artery bypass graft surgery to the decline in coronary heart disease mortality. This adjusted estimate is included in Table 3.
extent had not operative
3. Estimated Contributiun of Coronary Artery Bypass Graft Surgery to the Decline in Coronary Heart Disease Mortality Rates in the Minneapolis-St. Paul Area.Maleand Pem~ePatients
Combined:The MinnesotaHeartSurvey Estimated Percent Contribution of
Year 00 1310 1976 1972 I973 1974 1972 I976 1977 197i? IQ79
I 1 1 1983 1984
Nun&r of Lives Saved (dlu. - ds.1 -8.0 4.7 13 3 22.3 39.0 45.8
3.5
Rate of Lives Saved dnaa- dsn -N, -1.0 0.6
1.7 2.9 4.9 5.7 7.4 9.7 9.2 13.0
80.6 77.2 110.9 125.8
14.5
182.1
23.4
160.1 153.6 211.5
17.5
16.5 22.1
I6
Cwonary Artery Bypass Graft Surgery to Decline in Cowwry Hemt Disease MortaW -0.3 (-0.3)t 0.2 to.21 05 (0.5) 0.9 (0.8) 1.5 (1.4) 1.7 (1.7) 2.2 (2.2) 2.9 (2.9) 2.7 (2.7) 3.9 (3.8) 4.3 (4.3) 6.1 (6.0) 5.2 (5.2) 4.9 (4.8) 6.6 (6.5)
‘See Appemiix for computational method. ‘Adjusted vlue. dMl - d,, = medical deaths - surgical deaths; N, = number of 30- to 74.year old Minneapolis- St. Paul area residents in year X.
The Veterans Administ~tioo Cooperative Study, European Coronary Surgery Study and the Coronary Artery Surgery Study identified patient sub~oups that be~efitted from coronary artery bypass graft surgery, among which were patients with left main coronary artery disease (26)and threevessel disease combined with moderate left ventricular dysfunction (24,251. Patients in the European Coronary Surgery Study, with two- or three-vessel disease and apparently normal left ventricular function, benefitted from coronary artery bypass graft surgery, p~icu~ar~y if the left anterior descending coronary artery was involved (27). It would foHourthat if more patients with these lesious were referred to coronary artery bypass graft surgery would bP:improved, i~fl~enc~~g overall coronary ease mortality rates. As shown in Table 1, the proportion of patients with three-vessel disease and left main artery disease undergoing coronary artery bypass graft surgery increased over time, a~t~o~gbonly the change in the latter group reached statistical significance(p < 0.0001). In the early IPgOs,patients with one- and two-vessel disease also had the option of undergoing percutaneous transluminal
JAW Vol. 24, No. I My 1994:95-m
ale. %t was essential
imates forthis group w th rate was assu
bution of coronary artery e in coronary heart disease European Coronary Surgery Study and the Coronary Artery Surgery Study) spanned the to 1984).Furthermore, ran represented a highly sele patients referred for coronary Although use of the Duke registry data representAa compromise, these data have been used extensively in models to y heart disease mortality (13-15,28) and ke data were collected prospectively, they lete than those obtainable using a
these studies were s able for the surgical
and survival data were not avail-
cohorts from 1972to 1981.The effect ot the latter approach was to underestimate the number of deaths contributed by early surgical cohorts and to overestimate the number of g ic the later cohorts. Coronary artery bypass estly to the overall decline m coronary flerent computational
arison of the observed survival dieted estimates based on the Duke model shows good agreemt-nt. Nevertheless, the survival estimates predicted by the Duke made1were more frequently slightlylower than
coronary artery bypass graft surgery to the decline in corotween nary heart disease mortality, while small, inc ity) to 1970and 1984from a negative effect (increas a positive one. This increasing contribution may have been
102
JAW
WLWNY ET AL. CABG AND CORONARY HEART DISEASE MORTALITY DECLINE
influencedby the increaseduse of coronaryarterybypass sraft surgery, improvedoperative mortality,changes in severityor referralpatterns.It is questionablewhetherthe contributionof coronaryarterybypass sraft surgeryto the decline ic coronaryheart disease mortalitycontinued to because of increasethroughthe late 198%and early s, such as competitionfrom other therapeuticapp ctomy and thrombolytics,whose use this period. The interactionof andits effecton survivahtl will have n futurederivationsof the cQnt~~utio~of core9 if any, to the dockingin
Using these rates, artery bypass graft su as mortality was cQ~pM~e~
cline in corona
Vol. 24, No. I July I :95-m
broad disease
ne in ischemic heart &case mo~~ity. Ann
A mm=
where d,9W = number of coronary heart disease deaths in ~p~lalion of 30. to 74year-old Minneapolis-!%. I area residents in 1970, and N,910 = number of 3@ to 74-year-of area residents in 1970. 2. Actual coronary hart disease ~Q~t~~it~ em ( Faul Area resipopulation) for 30- to 74-year old inneapolis& dents in year X (Ax), where X = 1970,1971. . . . , I enttnt and survival of patients with baronary artery disease-&e 14. Fryor DB. Harrell FE Jr. Rat&in JS. et al. The changing survival benefits
where d,
= number of coronary heart disease deaths in population ear old Minneapolis-St. Paul nrea residents in year X. pattitioned into the number of coronary heart disease ng patients who had undergone coronary artery any time between 1970 aud year X ids&) and the heart disease deaths in the remaining population (da&. Nx = number of 30. to frlyear old Minneapolis-!%. Paul area
residents in yearX. rate @2r I 1. Paul are navailable (E,) in da X Id EX=-=Nx
dMx + dll, Nx
x Ias.
Equation 3 is identical to equation 2, with the exception of the dsx term, which is replaced by d,,. The latter term is the number of deaths expected if, in lieu of coronary artery bypass graft surgery, the gtwp identified in rate 2 was treated medically.
of coronary revascuhuir.ation over time. Circulation 1987:76 Suppl V:Vlf21. Ii FE Jr. Lee KL. et al. The evotution of medical and surgical therapy for coronary artery disease: a Kyear perspective. JAMA I989:261:2077-86. 16. Cox DR. Regression models and Me tables. i Royal Stat Sot 1972;43: 187-220. 17. Bnslow NE. Covariaoce analysis of censored survival data biometrics. Biometrics 1974;3(0:89-99. 18. Harrell FE Jr. The PHGLM procedure in SAS institute Inc. SUGI Suoolemental Libmrv User’s Guide. Version 5 Edition. Carv. .~~, . North Ca&na 1986:437-661 19. Pierpont GL, KNW Practical problems in assessingrisk for core g. J Thorac Cardiovasc Surgery 1985;89:673-82. 20. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (RGC) curve. Diagn Radio1 1982;143:29-36. 21. Frye RL. Fisher L, Schaff Randomized trials in coronary artery bypass surgery. EtoB Cardiovasc Dis B?87%1-22. 22. Gersh BJ, Califf RM, Loop FD, Akins CW, Pryor DB, Takara TC. Coronary bypass surgery in chronic stable angina. Circulation 1989,79 Suppl1:1-146-59.
Study Qf with simi
arrell FE Jr, Lee K clinicd trials of comary
artery bypass surgery.