- Email: [email protected]
Comparison of risk and patterns of practice in patients older and younger than 70 years with acute myocardial infarction in a two-year period (1987–1989)
The
CORONARY ARTERY DISEASE
American Journal
OCTOBER 1, 1991, VOL. 68, NO. 9
of Cardiology
Comparison of Risk and Patterns of Practice in Patients Older and Younger than 70 Years with Acute Myocardial Infarction in a Two-Year Period (19874989) Terrence J. Montague, MD, Roland M. Ikuta, MD, Roger Y. Wang, BMSC, Kyung S. Bay, PhD, Koon K. Teo, MD, PhD, and Norman J. Davies, MD
To further evaluate contemporary risk and practice patterns in acute myocardial infarction (AMI), 402 consecutive patients with AMI between July 1,1966, and June 30,1989 were studied. The clinical Investigations, medical therapy and outcome of patients aged 170 years (n = 132; group 1) were compared with patients aged <70 years (n = 270; group 2). In group 1, 20% of patients had no typical cardiac pain ver&IS 6% in group 2 (p
from younger ranted.
patients.
Further
(Am J Cardiol
studies are war-
1991;6&643-647)
n a previous comparative analysis of mortality risk and patterns of medical practice in older versus younger patients with acute myocardial infarction (AMI), .we found the elderly to be at substantially higher risk, but, ironically, to receive significantly less aggressiveinvestigative and therapeutic attention than their younger counterparts with AMI.’ In that previous study of an AM1 patient cohort selectedfrom 2 major Canadian tertiary cardiac care institutions during 1987, age was found to be a risk factor for short-term mortality, independent of other conventional risk factors.’ However, we were concernedthat the results of the previous analysis might be affected by selection bias, because the study samples were selected retrospectively and, perhaps more importantly, nonconsecutively.l To confirm or refute the results of our previous study,’ we reinvestigatedmortality risk and patterns of practice in another consecutiveAM1 patient cohort.
I
METHODS Patients: The current sample population consisted
of 402 consecutive patients with a diagnosis at discharge or death of AM1 and abnormally elevated creatine kinase levels during the interval of July 1, 1988 to June 30, 1989 at the University of Alberta Hospitals. Two hundred four of the study patients were admitted to the Division of Cardiology via the Emergency Department of the University of Alberta Hospitals; another 32 patients were originally admitted to non-cardiology servicesof the Hospitals. The remaining I66 study patients were transferred from other hospitals.
PATTERNS OF PRACTICE IN ACUTE MYOCARDIAL INFARCTION
843
1
TABLE I Distribution of Demographic and Clinical Variables in 402 Consecutive Patients with Acute Myocardial Infarction in 1988-1989
Sex Men Women Previous AMI Cardiac pain on presentation ECG infarct type Q wave Non-Q wave Medications ASA Pre-AMI Post-AM1 Nitrates Pre-AMI Post-AM1 p blockers Pre-AMI Post-AM1 Calcium antagonists Pre-AMI Post-AM1 Antiarrhythmics Pre-AMI Post-AM1 Acute thrombolysis Investigations Exercise tolerance test Coronary angiography Ejection fraction calculation Mortality AMI = acute myocardlal NS = not significant.
infarction;
Group 1 (z7Oyrs; n = 132) (%I
Group 2 (<70yrs; n = 270) (%)
60 40 30 80
80 20 25 94
58 42
67 33
NS NS
15
TABLE II Univariate Comparison of Mortality Risk in 402 Consecutive Patients with Acute Myocardial Infarction in 1988-1989
p Value
Sex Men Women Previous AMI Cardiac pain on presentation ECG infarct type Q wave Medications ASA post-AM1 Nitrates post-AM1 f3 blockers post-AM1 Calcium antagonists post-AM1 Antiarrhythmics post-AM1 Acute thrombolysis
12 56
74
NS
23 76
15 86
NS
Abbreviations
18 24
14 39
NS
33 58
23 60
<0.05 NS
4
2
47 4
42 20
NS
22 20
54 51 47 8
19
27 ASA = aspirin;
THE AMERICAN
JOURNAL
NS
ECG = electrocardiographic;
OF CARDIOLOGY
VOLUME
68
Variable Absent (%)
11 22
-
p Value
21
11
12
32
15
12
NS
7 11 4
30 29
10 18
20 11 15
10.01 10.01
5
19
as in Table I.
In light of the very similar data from our 1987 AM1 patient cohort (n = 207),’ data from that earlier study’ and the current 1988-1989 patient sample were combined. The enlarged total sample of 609 contem@rary patients with AM1 allowed not only more reliable assessmentof the relation of mortality and the other variables, but also comparison of the older (n = 233) and younger (n = 376) subgroups with enough mortality end points in each subgroup to make intergroup comparison meaningful.
Based on the results of our 1987 study,’ patients with AM1 in 1988 to 1989 were divided into 2 groups. Group 1 consistedof 132 patients 270 years of age, 79 men and 53 women, average age 78 years, range 70 to 97. Group 2 consistedof 270 patients <70 years of age, 217 men and 53 women, average age 56 years, range 22 to 69. The variables assessedin this study included preadmission demographic and clinical data (age, sex, history of previous AM1 and cardiac medication profile) and acute presentation (symptoms, Q wave/non-Qwave AM1 type, creatine kinase peak level) and in-hospital (risk assessmenttests, medication and mortality) clinical variables. All grouped data were expressedas mean f standard deviation. Inferential univariate comparison of continuous data between the 2 study groups was by 2tailed, unpaired t tests; nominal scalevariables were by chi-square tests. The overall assessmentof association between mortality, as the dependent variable, and all other clinical and biologic variables, as independent variables, was evaluated using stepwiselogistic regression analyses. 844
Variable Present (%I
RESULTS The distribution of clinical variables in the older group 1 and younger group 2 AM1 patients is compared in Table I. Men predominated in both groups, although less so in the elderly group (Table I). Among the traditional risk factors, history of previous AMI and Q-wave AM1 were equally distributed in the 2 study samples(Table I). In addition, peak creatine kinase levels were not different in group 1 (1,402 f 1,635 IU) compared with group 2 (1,533 f 1,344 IU) (p = not significant). The most important study variable, in-hospital mortality, was 27% among patients 270 years and 8% in patients <70 years (p
OCTOBER
1, 1991
very significantly lessthan the 20% use in patients <70 years (p
LOGISTIC 80
REGRESSION
(ALL
SUBJECTS,
nificantly lessoften in the older patient group (on average, 20%) than in the younger patients (on average, 51%) (Table I). The univariate comparisons of mortality risk in the presenceand absenceof selectedclinical variables among the 402 patients with AM1 during 1988-1989 are displayed in Table II. History of previous infarction and absenceof cardiac pain on presentation conferred excessrisk, acute-phase thrombolysis and the postinfarction use of acetylsalicylic acid, nitrates, /3 blockers and calcium antagonistswere all associatedwith significantly lower mortality risk (Table II). The multivariate analysesof the associationbetween mortality risk and the other clinical variables in the total sample of 609 patients with AM1 and the older (n = 233) and younger (n = 376) subgroups, respectively, are presentedin Figures 1 to 3. Overall, previous AMI, presentation without cardiac pain and age 270
1x609)
MULTIPLE
I
LOGISTIC
REGRESSION
(~70 YRS, ~376)
120 100
Relative
Risk
80
*
60 40
pqo.05
0
Relative (%I -40 -
*
risk
* pco.05
‘O 0 -20
**
-60 -
-40
-SO-
-60 -80 -100 ADYS PREY AMI
FIGURE 1. Relative mortality risk associated with sekcted clinical variables in 609 patknts with acute myocardiil infarctlon (AMI) 1987-1666. ADYS = antianhythmii me&cations after AMI; ASA = aspirin after AMI; BB = B Mockers after AMI; CCB = calcium antagonists after AMI; CP = cardii pain; NIT = nitrates after AMI; Prev MI = previous AMI. MULTIPLE
LOGISTIC
REGRESSION
NO CP
M
CCB
ASA
BB
NIT
FIGURE 3. Relative mortality risk associated with selectad clinical variables in 376 patients aged <70 years with acute myocardiil infarction (AMI) 1987-1999. Abbreviations as in Figures 1 and 2.
MULTIPLE
(270 YRS, n=233)
LOGISTIC
REGRESSION
(ALL
SUBJECTS,
n=609)
50 40 ,
I
I
.
40 30 0
Relative (%)
.
20 -
Risk
Relative Risk (%I * pco.05
-20
. 10. .
0-
-40
-10 4
-60
.20 1 60 PREV AMI
NO CP
M
NIT
ASA
CCB
BB
.
70
80
90
Age (years) I
FlGURE 2. Relative mortalii risk associated with selected clinical variables in 233 patients aged 270 years with acute myocardial infarction (AMI) 1667-1666. M = male; other abbreviations as in Figure 1.
FIGURE 4. Relative mortality risk ot various ages among 606 patients with acute myocardial infarction 1997-1999. Those aged <70 years were associated with lower ridr, those aged >70 years were associated with progressively increased risk.
PATTERNS OF PRACTICE IN ACUTE MYOCARDIAL INFARCTION
845
years were independently associatedwith the greatest increased relative mortality risk. Post-AM1 use of p blockers was, in contrast, associatedwith least relative risk (Figure 1). Among the older patient subset, previous AMI and absenceof chest pain also had the highest relative risk and @blockers, the lowest (Figure 2). Interestingly, in the younger patient group, although previous AM1 and absenceof cardiac pain were once again associatedwith increased risk of death, the post-AM1 use of antiarrhythmic drugs was associatedwith the highest relative mortality risk (Figure 3). It must be cautioned that the relation of mortality risk and therapeutic agents in this practice pattern analysis should not be viewed as causal; this was not a clinical trial. For example, the apparently counterintuitive association of post-AM1 calcium antagonist use with decreasedrelative risk doesnot mean these agents causedthe lower risk. Rather, as in our previous analysis,l they were probably given to less ill AM1 patients with lesser myocardial injury and consequent lesser risk. When age was considered as a continuous seriesof discrete variables, relative to mortality risk, there was a progressiveincreasein risk with increasein age (Figure 4). Thus, age 270 years is a variable that confers excess risk independently of other more traditional risk factors and the risk appears to increase continuously with increasing age. DISCUSSION The results of the presentstudy confirmed the findings of our previous study.’ Specifically, patients aged >70 years have a markedly increasedrisk of in-hospital death, and the associationof age and increasedmortality risk is independent of other variables that have been traditionally considered part of AM1 pathophysiology. Furthermore, elderly patients with AM1 are investigated and treated differently from their younger counterparts. In particular, they have significantly fewer risk prediction tests and receive proven effective medical therapy significantly less often than younger patients. The findings of our 2 studies are very similar to those of a recent subgroup analysis of an AM1 data base collected between 1980 and 1990 in 4 Canadian and United States medical centers. This latter study revealed in-hospital mortality of 20% and a 12 month postdischargemortality of 18% for patients >75 years of age, with age independently related to increased postdischargemortality risk.2 The incidencesof exercise testing, cardiac catheterization and thrombolytic and @blocker therapy were very low in all patients with AM1 26.5 years of age and were significantly lower in patients >75 years, compared with patients between 65 and 75 years.2 846
THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 68
Currently, there are 2 reasonableinterpretations of the available clinical trial data regarding the likely outcome with thrombolytic, p blocker, acetylsalicylic acid and nitrate use in elderly patients with AMI. The first is the elderly are likely to derive positive benefits at least comparable to the generally younger patients with AM1 who comprisedthe bulk of the previous trial populations. Alternatively, the available data could be interpreted as showing no convincing evidence of treatment benefit to elderly patients with AMI. Whereas the former interpretation is more likely correct, the practice pattern analysesimply the latter interpretation may, in fact, be very prevalent in clinical practice. Another major implication of these contemporary practice analysesis that proven effective medical therapy for AM1 may not be used optimally even in younger patients. For example, only 20% of patients <70 years received thrombolytic therapy in our 1988-1989 AM1 cohort, and only 39% of the same group received p blockers (Table I), both of which have been shown, incontrovertibly, to decrease mortality risk in patients with AMI. Conversely, calcium antagonist therapy, proven ineffective in post-AM1 prophylaxis,3 was used in 60% of our 1988-1989 patients with AM1 (Table I). Our data do not allow evaluation of exactly why the patterns of practice were as observed. However, the very low overall utilization rates of thrombolytic and p blockade therapy make it unlikely that clinical contraindications were present in all the patients who did not receive these medications. Moreover, the data were analyzed in an intention-to-treat manner, when any useof the study medications during any part of the hospital coursewas recordedas the drug being present, irrespective of whether it was subsequentlystopped becauseof adverseeffects. Thus, the utilization rates representthe best casescenario. In contrast, it is possiblethat physician desire to avoid side effectsmay have contributed to the low administration. of somedrugs. Such perceptions may simply have been more prevalent in the management of elderly patients, i.e., bias may have occurred.’ If physician bias is a factor in contemporary AM1 practice patterns, it may be secondary to ineffective translation of viable results from important clinical trials to everyday clinical practice. If this supposition is correct, how can the situation be improved? Certainly, continuing medical education by unbiased and thoughtful personsand consensuspanels is important to disseminationof the correct interpretations of clinical trials. That these standard modes of altering the patterns of practice have been lessthan totally satisfactory is, perhaps, best supported by the persistently high utilization rates of calcium antagonists in AMI, despiteconsistentdata from repeatedtrials showing no improvement in survival for the treated groups.3 Imperfect translation of trial data is also indicated by
OCTOBER 1,
1991
the limited use of p blockers in AMI, despite available knowledge of their proven ability to reduce total AMI mortality, sudden death and reinfarction since 1981.415 The Canadian Cardiovascular Society’s consensus conference on postinfarction management has recently recommendedthat all patients with AMI, except possibly those at very low risk, receive 0 blockers in the absenceof any clinical contraindication.6 This consensus conference also suggested that benefit from 0 blockers was greatest among those at highest risk.6 The data from our patient outcome analysescertainly indicate that patients with AM1 aged 270 years have a markedly increased risk of death. If the consensusview is correct,6 elderly patients may, becauseof their very high risk status, benefit even more from effective medical therapy than the average patient with AMI. However, this concept remains untested. A novel approach that may lead to faster dissemina tion of important trial results, and a concurrently faster adaptation in clinical practice patterns, may be the recruitment of investigators for large, simple trials from widespreadgeographic areas and a wide variety of real practice situations. The involvement of primary care physicians as trial investigators would simultaneously provide a pool of community clinician-scientists and medical educators. Such physicians would be integrally
involved in the progress of trials to answer important clinical questions.Their enthusiasm and anticipation of the correct answer would be transmitted to their local colleaguesand likely result in a more efficient acceptance of trial outcome data than would be accomplished by a visiting educator or publication. This concept also remains untested. Further studies are indicated.
REFERENCES 1. Montague T, Wang R, Crowell R, Bay K, Marshall D, Tymchak W, Tea K, Davies N. Acute myocardial infarction: contemporary risk and managementin older versus younger patients. Canad J Cardiol 1990;6:241-246. 2. Smith SC, Gilpin E, Ahnve S, Dittrich H, Nicod P, Henning H, Ross J. Outlook after myocardial infarction in the very elderly compared with that in patients aged 65 to 75 years. J Am Co11Cardiol 1990;16:784-792. 3. Yusuf S, Wittes J, Friedman L. Overview of results of randomizedclinical trials in heart disease. Treatments following myocardial infarction. JAMA 1988;260:2088-2093. 4. Hjalmarson A, Elmfeldt D, Herlitz J, Holmberg S, Malen I, Nyberg G, Ryden
L, SwedbergK, Vedin A, WaagsteinP, WaldenstromA, WaldenstromJ, Wedel H, WilmhelmsenL, WilmhelmsenC. Effect on mortality of metoprolol in acute myocardial infarction. A double-blind randomisedtrial. Lancer 1981;2:823-827. 5. The Norwegian Multicenter Study Group. Timolol-induced reduction in mortality and reinfarction in patientssurviving acute myocardial infarction. N Engl J Med 1981;304:801-807. 6. Canadian Cardiovascular Society. Report of the consensusconferenceon the managementof the post myocardial infarction patient. Read before the 43rd Scientific Sessionsof the CanadianCardiovascularSociety,Halifax, Nova Scotia, 18 October 1990.
PAlTERNS OF PRACTICE IN ACUTE MYOCARDIAL INFARCTION
847
CORONARY ARTERY DISEASE
American Journal
OCTOBER 1, 1991, VOL. 68, NO. 9
of Cardiology
Comparison of Risk and Patterns of Practice in Patients Older and Younger than 70 Years with Acute Myocardial Infarction in a Two-Year Period (19874989) Terrence J. Montague, MD, Roland M. Ikuta, MD, Roger Y. Wang, BMSC, Kyung S. Bay, PhD, Koon K. Teo, MD, PhD, and Norman J. Davies, MD
To further evaluate contemporary risk and practice patterns in acute myocardial infarction (AMI), 402 consecutive patients with AMI between July 1,1966, and June 30,1989 were studied. The clinical Investigations, medical therapy and outcome of patients aged 170 years (n = 132; group 1) were compared with patients aged <70 years (n = 270; group 2). In group 1, 20% of patients had no typical cardiac pain ver&IS 6% in group 2 (p
from younger ranted.
patients.
Further
(Am J Cardiol
studies are war-
1991;6&643-647)
n a previous comparative analysis of mortality risk and patterns of medical practice in older versus younger patients with acute myocardial infarction (AMI), .we found the elderly to be at substantially higher risk, but, ironically, to receive significantly less aggressiveinvestigative and therapeutic attention than their younger counterparts with AMI.’ In that previous study of an AM1 patient cohort selectedfrom 2 major Canadian tertiary cardiac care institutions during 1987, age was found to be a risk factor for short-term mortality, independent of other conventional risk factors.’ However, we were concernedthat the results of the previous analysis might be affected by selection bias, because the study samples were selected retrospectively and, perhaps more importantly, nonconsecutively.l To confirm or refute the results of our previous study,’ we reinvestigatedmortality risk and patterns of practice in another consecutiveAM1 patient cohort.
I
METHODS Patients: The current sample population consisted
of 402 consecutive patients with a diagnosis at discharge or death of AM1 and abnormally elevated creatine kinase levels during the interval of July 1, 1988 to June 30, 1989 at the University of Alberta Hospitals. Two hundred four of the study patients were admitted to the Division of Cardiology via the Emergency Department of the University of Alberta Hospitals; another 32 patients were originally admitted to non-cardiology servicesof the Hospitals. The remaining I66 study patients were transferred from other hospitals.
PATTERNS OF PRACTICE IN ACUTE MYOCARDIAL INFARCTION
843
1
TABLE I Distribution of Demographic and Clinical Variables in 402 Consecutive Patients with Acute Myocardial Infarction in 1988-1989
Sex Men Women Previous AMI Cardiac pain on presentation ECG infarct type Q wave Non-Q wave Medications ASA Pre-AMI Post-AM1 Nitrates Pre-AMI Post-AM1 p blockers Pre-AMI Post-AM1 Calcium antagonists Pre-AMI Post-AM1 Antiarrhythmics Pre-AMI Post-AM1 Acute thrombolysis Investigations Exercise tolerance test Coronary angiography Ejection fraction calculation Mortality AMI = acute myocardlal NS = not significant.
infarction;
Group 1 (z7Oyrs; n = 132) (%I
Group 2 (<70yrs; n = 270) (%)
60 40 30 80
80 20 25 94
58 42
67 33
NS NS
15
TABLE II Univariate Comparison of Mortality Risk in 402 Consecutive Patients with Acute Myocardial Infarction in 1988-1989
p Value
Sex Men Women Previous AMI Cardiac pain on presentation ECG infarct type Q wave Medications ASA post-AM1 Nitrates post-AM1 f3 blockers post-AM1 Calcium antagonists post-AM1 Antiarrhythmics post-AM1 Acute thrombolysis
12 56
74
NS
23 76
15 86
NS
Abbreviations
18 24
14 39
NS
33 58
23 60
<0.05 NS
4
2
47 4
42 20
NS
22 20
54 51 47 8
19
27 ASA = aspirin;
THE AMERICAN
JOURNAL
NS
ECG = electrocardiographic;
OF CARDIOLOGY
VOLUME
68
Variable Absent (%)
11 22
-
p Value
21
11
12
32
15
12
NS
7 11 4
30 29
10 18
20 11 15
10.01 10.01
5
19
as in Table I.
In light of the very similar data from our 1987 AM1 patient cohort (n = 207),’ data from that earlier study’ and the current 1988-1989 patient sample were combined. The enlarged total sample of 609 contem@rary patients with AM1 allowed not only more reliable assessmentof the relation of mortality and the other variables, but also comparison of the older (n = 233) and younger (n = 376) subgroups with enough mortality end points in each subgroup to make intergroup comparison meaningful.
Based on the results of our 1987 study,’ patients with AM1 in 1988 to 1989 were divided into 2 groups. Group 1 consistedof 132 patients 270 years of age, 79 men and 53 women, average age 78 years, range 70 to 97. Group 2 consistedof 270 patients <70 years of age, 217 men and 53 women, average age 56 years, range 22 to 69. The variables assessedin this study included preadmission demographic and clinical data (age, sex, history of previous AM1 and cardiac medication profile) and acute presentation (symptoms, Q wave/non-Qwave AM1 type, creatine kinase peak level) and in-hospital (risk assessmenttests, medication and mortality) clinical variables. All grouped data were expressedas mean f standard deviation. Inferential univariate comparison of continuous data between the 2 study groups was by 2tailed, unpaired t tests; nominal scalevariables were by chi-square tests. The overall assessmentof association between mortality, as the dependent variable, and all other clinical and biologic variables, as independent variables, was evaluated using stepwiselogistic regression analyses. 844
Variable Present (%I
RESULTS The distribution of clinical variables in the older group 1 and younger group 2 AM1 patients is compared in Table I. Men predominated in both groups, although less so in the elderly group (Table I). Among the traditional risk factors, history of previous AMI and Q-wave AM1 were equally distributed in the 2 study samples(Table I). In addition, peak creatine kinase levels were not different in group 1 (1,402 f 1,635 IU) compared with group 2 (1,533 f 1,344 IU) (p = not significant). The most important study variable, in-hospital mortality, was 27% among patients 270 years and 8% in patients <70 years (p
OCTOBER
1, 1991
very significantly lessthan the 20% use in patients <70 years (p
LOGISTIC 80
REGRESSION
(ALL
SUBJECTS,
nificantly lessoften in the older patient group (on average, 20%) than in the younger patients (on average, 51%) (Table I). The univariate comparisons of mortality risk in the presenceand absenceof selectedclinical variables among the 402 patients with AM1 during 1988-1989 are displayed in Table II. History of previous infarction and absenceof cardiac pain on presentation conferred excessrisk, acute-phase thrombolysis and the postinfarction use of acetylsalicylic acid, nitrates, /3 blockers and calcium antagonistswere all associatedwith significantly lower mortality risk (Table II). The multivariate analysesof the associationbetween mortality risk and the other clinical variables in the total sample of 609 patients with AM1 and the older (n = 233) and younger (n = 376) subgroups, respectively, are presentedin Figures 1 to 3. Overall, previous AMI, presentation without cardiac pain and age 270
1x609)
MULTIPLE
I
LOGISTIC
REGRESSION
(~70 YRS, ~376)
120 100
Relative
Risk
80
*
60 40
pqo.05
0
Relative (%I -40 -
*
risk
* pco.05
‘O 0 -20
**
-60 -
-40
-SO-
-60 -80 -100 ADYS PREY AMI
FIGURE 1. Relative mortality risk associated with sekcted clinical variables in 609 patknts with acute myocardiil infarctlon (AMI) 1987-1666. ADYS = antianhythmii me&cations after AMI; ASA = aspirin after AMI; BB = B Mockers after AMI; CCB = calcium antagonists after AMI; CP = cardii pain; NIT = nitrates after AMI; Prev MI = previous AMI. MULTIPLE
LOGISTIC
REGRESSION
NO CP
M
CCB
ASA
BB
NIT
FIGURE 3. Relative mortality risk associated with selectad clinical variables in 376 patients aged <70 years with acute myocardiil infarction (AMI) 1987-1999. Abbreviations as in Figures 1 and 2.
MULTIPLE
(270 YRS, n=233)
LOGISTIC
REGRESSION
(ALL
SUBJECTS,
n=609)
50 40 ,
I
I
.
40 30 0
Relative (%)
.
20 -
Risk
Relative Risk (%I * pco.05
-20
. 10. .
0-
-40
-10 4
-60
.20 1 60 PREV AMI
NO CP
M
NIT
ASA
CCB
BB
.
70
80
90
Age (years) I
FlGURE 2. Relative mortalii risk associated with selected clinical variables in 233 patients aged 270 years with acute myocardial infarction (AMI) 1667-1666. M = male; other abbreviations as in Figure 1.
FIGURE 4. Relative mortality risk ot various ages among 606 patients with acute myocardial infarction 1997-1999. Those aged <70 years were associated with lower ridr, those aged >70 years were associated with progressively increased risk.
PATTERNS OF PRACTICE IN ACUTE MYOCARDIAL INFARCTION
845
years were independently associatedwith the greatest increased relative mortality risk. Post-AM1 use of p blockers was, in contrast, associatedwith least relative risk (Figure 1). Among the older patient subset, previous AMI and absenceof chest pain also had the highest relative risk and @blockers, the lowest (Figure 2). Interestingly, in the younger patient group, although previous AM1 and absenceof cardiac pain were once again associatedwith increased risk of death, the post-AM1 use of antiarrhythmic drugs was associatedwith the highest relative mortality risk (Figure 3). It must be cautioned that the relation of mortality risk and therapeutic agents in this practice pattern analysis should not be viewed as causal; this was not a clinical trial. For example, the apparently counterintuitive association of post-AM1 calcium antagonist use with decreasedrelative risk doesnot mean these agents causedthe lower risk. Rather, as in our previous analysis,l they were probably given to less ill AM1 patients with lesser myocardial injury and consequent lesser risk. When age was considered as a continuous seriesof discrete variables, relative to mortality risk, there was a progressiveincreasein risk with increasein age (Figure 4). Thus, age 270 years is a variable that confers excess risk independently of other more traditional risk factors and the risk appears to increase continuously with increasing age. DISCUSSION The results of the presentstudy confirmed the findings of our previous study.’ Specifically, patients aged >70 years have a markedly increasedrisk of in-hospital death, and the associationof age and increasedmortality risk is independent of other variables that have been traditionally considered part of AM1 pathophysiology. Furthermore, elderly patients with AM1 are investigated and treated differently from their younger counterparts. In particular, they have significantly fewer risk prediction tests and receive proven effective medical therapy significantly less often than younger patients. The findings of our 2 studies are very similar to those of a recent subgroup analysis of an AM1 data base collected between 1980 and 1990 in 4 Canadian and United States medical centers. This latter study revealed in-hospital mortality of 20% and a 12 month postdischargemortality of 18% for patients >75 years of age, with age independently related to increased postdischargemortality risk.2 The incidencesof exercise testing, cardiac catheterization and thrombolytic and @blocker therapy were very low in all patients with AM1 26.5 years of age and were significantly lower in patients >75 years, compared with patients between 65 and 75 years.2 846
THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 68
Currently, there are 2 reasonableinterpretations of the available clinical trial data regarding the likely outcome with thrombolytic, p blocker, acetylsalicylic acid and nitrate use in elderly patients with AMI. The first is the elderly are likely to derive positive benefits at least comparable to the generally younger patients with AM1 who comprisedthe bulk of the previous trial populations. Alternatively, the available data could be interpreted as showing no convincing evidence of treatment benefit to elderly patients with AMI. Whereas the former interpretation is more likely correct, the practice pattern analysesimply the latter interpretation may, in fact, be very prevalent in clinical practice. Another major implication of these contemporary practice analysesis that proven effective medical therapy for AM1 may not be used optimally even in younger patients. For example, only 20% of patients <70 years received thrombolytic therapy in our 1988-1989 AM1 cohort, and only 39% of the same group received p blockers (Table I), both of which have been shown, incontrovertibly, to decrease mortality risk in patients with AMI. Conversely, calcium antagonist therapy, proven ineffective in post-AM1 prophylaxis,3 was used in 60% of our 1988-1989 patients with AM1 (Table I). Our data do not allow evaluation of exactly why the patterns of practice were as observed. However, the very low overall utilization rates of thrombolytic and p blockade therapy make it unlikely that clinical contraindications were present in all the patients who did not receive these medications. Moreover, the data were analyzed in an intention-to-treat manner, when any useof the study medications during any part of the hospital coursewas recordedas the drug being present, irrespective of whether it was subsequentlystopped becauseof adverseeffects. Thus, the utilization rates representthe best casescenario. In contrast, it is possiblethat physician desire to avoid side effectsmay have contributed to the low administration. of somedrugs. Such perceptions may simply have been more prevalent in the management of elderly patients, i.e., bias may have occurred.’ If physician bias is a factor in contemporary AM1 practice patterns, it may be secondary to ineffective translation of viable results from important clinical trials to everyday clinical practice. If this supposition is correct, how can the situation be improved? Certainly, continuing medical education by unbiased and thoughtful personsand consensuspanels is important to disseminationof the correct interpretations of clinical trials. That these standard modes of altering the patterns of practice have been lessthan totally satisfactory is, perhaps, best supported by the persistently high utilization rates of calcium antagonists in AMI, despiteconsistentdata from repeatedtrials showing no improvement in survival for the treated groups.3 Imperfect translation of trial data is also indicated by
OCTOBER 1,
1991
the limited use of p blockers in AMI, despite available knowledge of their proven ability to reduce total AMI mortality, sudden death and reinfarction since 1981.415 The Canadian Cardiovascular Society’s consensus conference on postinfarction management has recently recommendedthat all patients with AMI, except possibly those at very low risk, receive 0 blockers in the absenceof any clinical contraindication.6 This consensus conference also suggested that benefit from 0 blockers was greatest among those at highest risk.6 The data from our patient outcome analysescertainly indicate that patients with AM1 aged 270 years have a markedly increased risk of death. If the consensusview is correct,6 elderly patients may, becauseof their very high risk status, benefit even more from effective medical therapy than the average patient with AMI. However, this concept remains untested. A novel approach that may lead to faster dissemina tion of important trial results, and a concurrently faster adaptation in clinical practice patterns, may be the recruitment of investigators for large, simple trials from widespreadgeographic areas and a wide variety of real practice situations. The involvement of primary care physicians as trial investigators would simultaneously provide a pool of community clinician-scientists and medical educators. Such physicians would be integrally
involved in the progress of trials to answer important clinical questions.Their enthusiasm and anticipation of the correct answer would be transmitted to their local colleaguesand likely result in a more efficient acceptance of trial outcome data than would be accomplished by a visiting educator or publication. This concept also remains untested. Further studies are indicated.
REFERENCES 1. Montague T, Wang R, Crowell R, Bay K, Marshall D, Tymchak W, Tea K, Davies N. Acute myocardial infarction: contemporary risk and managementin older versus younger patients. Canad J Cardiol 1990;6:241-246. 2. Smith SC, Gilpin E, Ahnve S, Dittrich H, Nicod P, Henning H, Ross J. Outlook after myocardial infarction in the very elderly compared with that in patients aged 65 to 75 years. J Am Co11Cardiol 1990;16:784-792. 3. Yusuf S, Wittes J, Friedman L. Overview of results of randomizedclinical trials in heart disease. Treatments following myocardial infarction. JAMA 1988;260:2088-2093. 4. Hjalmarson A, Elmfeldt D, Herlitz J, Holmberg S, Malen I, Nyberg G, Ryden
L, SwedbergK, Vedin A, WaagsteinP, WaldenstromA, WaldenstromJ, Wedel H, WilmhelmsenL, WilmhelmsenC. Effect on mortality of metoprolol in acute myocardial infarction. A double-blind randomisedtrial. Lancer 1981;2:823-827. 5. The Norwegian Multicenter Study Group. Timolol-induced reduction in mortality and reinfarction in patientssurviving acute myocardial infarction. N Engl J Med 1981;304:801-807. 6. Canadian Cardiovascular Society. Report of the consensusconferenceon the managementof the post myocardial infarction patient. Read before the 43rd Scientific Sessionsof the CanadianCardiovascularSociety,Halifax, Nova Scotia, 18 October 1990.
PAlTERNS OF PRACTICE IN ACUTE MYOCARDIAL INFARCTION
847