Prognostic significance of the electrocardiogram after Q wave myocardial

infarction. The Framingham Study

ND Wong, D Levy and WB Kannel
Circulation 1990, 81:780-789
doi: 10.1161/01.CIR.81.3.780
Circulation is published by the American Heart Association. 7272 Greenville Avenue, Dallas, TX
72514
Copyright 1990 American Heart Association. All rights reserved. Print ISSN: 0009-7322. Online
ISSN: 1524-4539

The online version of this article, along with updated information and services, is
located on the World Wide Web at:
http://circ.ahajournals.org/content/81/3/780

Subscriptions: Information about subscribing to Circulation is online at

http://circ.ahajournals.org//subscriptions/
Permissions: Permissions & Rights Desk, Lippincott Williams & Wilkins, a division of Wolters
Kluwer Health, 351 West Camden Street, Baltimore, MD 21202-2436. Phone: 410-528-4050.
Fax: 410-528-8550. E-mail:
journalpermissions@lww.com
Reprints: Information about reprints can be found online at
http://www.lww.com/reprints

Downloaded from http://circ.ahajournals.org/ by guest on November 25, 2011

780

Prognostic Significance of the

Electrocardiogram After Q Wave
Myocardial Infarction
The Framingham Study
Nathan D. Wong, PhD, Daniel Levy, MD, and William B. Kannel, MD
The prognostic value of abnormalities on the electrocardiogram (ECG) present 1 year after
initial myocardial infarction (MI) is examined in relation to reinfarction and coronary death
throughout 32 years (mean, 10.1 years) of follow-up in the Framingham Heart Study. Resting
12-lead ECGs were available in 251 survivors (190 men and 61 women) of clinically recognized
Q wave MI. The ECG reverted to normal in 31 (12.4%) cases and was abnormal but without
Q waves in 37 (14.7%). Q waves persisted without other significant abnormalities in 108 (43.0%)
and with other abnormalities in 75 (29.9%) cases. Electrocardiographic abnormalities at
follow-up were more common in women and in those persons whose initial MI was anterior as
compared with inferior. Nonspecific T wave, ST segment changes, and electrocardiographic left
ventricular hypertrophy on the ECG before or after MI were powerful predictors (p<0.01) of
coronary death. The relation of these residual post-MI electrocardiographic findings to
reinfarction and coronary death was assessed by Cox regression analysis. The follow-up
electrocardiographic status was unrelated to the risk of subsequent reinfarction. Subjects who
lost Q wave evidence of MI but whose ECG continued to show evidence of repolarization
abnormalities, left ventricular hypertrophy, or blocked intraventricular conduction were at a
3.5-fold increased risk (p<0.01) of coronary death as compared with those reverting to a
normal ECG. Persons with a persistent Q wave MI accompanied by these abnormalities were
at a 2.7-fold excess risk (p=0.01) of coronary death as compared with those with a normalized
ECG. These findings remained significant when considering age and standard coronary risk
factors. The presence of other electrocardiographic abnormalities without persistent Q waves
yields a worse prognosis than a Q wave persisting alone. The prognostic value of a follow-up
ECG with abnormalities other than a persistent Q wave MI also remained after considering the
effects of left ventricular hypertrophy and cardiac enlargement on x-ray, functional classification, and diuretic usage. Specific electrocardiographic abnormalities present before infarction,
however, were potent indicators of long-term prognosis and diminished the importance of the
follow-up ECG. Although survival after initial MI is improved only if the ECG reverts to
normal, information on electrocardiographic abnormalities before MI can be especially useful
in evaluating long-term risk. (Circulation 1990;81:780-789)

From the Division of Cardiology, Department of Medicine,

University of California, Irvine, California; and the Framingham
Heart Study, Framingham, and the Section of Preventive Medicine
and Epidemiology, Boston University School of Medicine, Boston,
Massachusetts.
Supported in part by grant HL-01243 from the National Heart,
Lung, and Blood Institute, Bethesda, Maryland.
Preliminary results of this study were presented at the 60th
Scientific Sessions of the American Heart Association, November
18, 1987.
Address for correspondence: Nathan D. Wong, PhD, Preventive
Cardiology Program, C240 Medical Sciences I, Department of
Medicine, University of California, Irvine, CA 92717.
Received August 15, 1988; revision accepted November 2,
1989.

T he usefulness of the resting post-myocardial

infarction electrocardiogram (ECG) as a tool
for assessing subsequent risk in patients with
myocardial infarction (MI) is uncertain. The risk of
death or complications is higher in those with generalized abnormalities' and, specifically, in individuals with
Q or QS wave abnormalities, left ventricular hypertrophy, ST segment elevation or depression, and conduction block disturbances.2-8 Data from population-based
studies with long-term follow-up are lacking, however.
Furthermore, few studies have examined the relation
between the persistence or regression of electrocardiographic abnormalities and prognosis after MI.

Downloaded from http://circ.ahajournals.org/ by guest on November 25, 2011

Wong et al The Electrocardiogram and Post-Myocardial Infarction Prognosis

The purpose of this investigation was to study the

association of electrocardiographic abnormalities
after MI with the long-term risk of reinfarction and
coronary death. Whether specific electrocardiographic
abnormalities such as ST segment and T wave
changes, left ventricular hypertrophy, and conduction
block disturbances present 1 year after MI can modify
the long-term prognostic outlook in those persons with
or without persistent Q wave evidence of MI was
evaluated. The Framingham Heart Study offered a
unique opportunity to examine these issues in a
population-based cohort of survivors of MI with biennial ECGs and follow-up of up to 32 years.
Methods

The general methodology of the Framingham

Study has been previously described.9 In short, it
involved the follow-up of 5,209 men and women,
initially (in 1948) aged 30-62 years, for the evaluation of risk factors for coronary heart disease. Biennial examinations included a detailed physical examination, chest x-ray, 12-lead resting ECG, and a
complete panel of blood tests.9

Study Population
During the first 30 years of follow-up (through the
16th biennial examination), a total of 388 patients
with an initial Q wave MI were identified, with each
event assigned the date of hospitalization or diagnosis by a physician. These clinically recognized infarctions required symptoms that promoted medical evaluation followed by a diagnosis based on characteristic
electrocardiographic changes (loss of initial QRS
potentials indicated by the development of Q waves
of at least 0.04 seconds).3 Q wave infarctions that
were clinically unrecognized (n=225) and incidentally detected by routine ECGs were not included in
this investigation because of the uncertainty of their
dating. Additionally, MIs prevalent at the initial
Framingham examination (n=49) and diagnosed by
autopsy evidence only (n=34), or clinically recognized but without a diagnostic Q wave pattern ECG
(n=84) were not eligible for inclusion in this study.
Of the 388 initial recognized Q wave MIs, 97 did not
survive to attend a follow-up exam, leaving 291
eligible for baseline examination of risk factors. An
additional 40 (14%) survived but did not attend the
required follow-up exam for inclusion. Of the 97 who
did not survive, 88 died of coronary heart disease
(including three who died suddenly within 1 hour),
and nine died of noncoronary causes. This left 251
(190 men and 61 women) subjects who returned for a
follow-up examination, on average, 1 year after their
initial MI, at which time a resting 12-lead ECG and
risk factor measurements were obtained (Table 1).
Those who did not survive to be included in the study
were more likely to be female and to have an abnormal ECG before infarction. Of those who did survive
to attend the postinfarction examination, participants
and nonparticipants did not differ significantly by

781

TABLE 1. Ascertainment of Study Population

Study population
Original Framingham cohort
Total MI (exams 1-16, 30-year follow-up)
Exclusions
Prevalent MI at exam 1
MI by autopsy only
MI without diagnostic ECG
Unrecognized MI
Total eligible MI for study (recognized MI by ECG)
Cases not surviving to follow-up exam
Survivors not attending follow-up exam
Total eligible MI attending follow-up examination
Exclusions for reinfarction analysis
Reinfarction before follow-up exam
Coronary death not because of reinfarction

n
5,209
780
392
49
34
84
225
388
97
40
251
65
10
55

gender, age, or ECG general impression before the

infarction.
On examination, subjects were classified according
to their follow-up ECG, demonstrating either 1) no
definite abnormality, 2) definite abnormalities but
without a definite persistent Q wave MI, 3) a persistent Q wave MI without other specified abnormalities, or 4) a persistent Q wave MI with other specified
abnormalities. Specified abnormalities included the
definite presence of one or more of the following:
nonspecific ST segment or T wave changes, left
ventricular hypertrophy, atrioventricular block (first
or higher degree), or complete bundle branch block.
Risk factors including systolic blood pressure, relative weight (percentage of ideal weight), serum cholesterol, and history of diabetes were measured
according to standard criteria.10,11

Definition of End Points and Follow-up

All subjects were followed for the occurrence of
recurrent MI (fatal or nonfatal) or death from coronary heart disease through the 17th biennial examination, which included up to 32 years (mean, 10.1
years) of follow-up. Reinfarctions were used as
acceptable end points regardless of whether they
were clinically recognized; those that were unrecognized were assigned the midpoint of the date
between the examination first demonstrating electrocardiographic evidence of the new MI and the date of
the last attended examination where it was not
present. Deaths from coronary heart disease
included only those deaths from coronary arteriosclerosis, thus excluding about 10% of cardiovascular
deaths from other causes (e.g., aortic aneurysm and
congenital heart disease). Follow-up time was
defined as the number of days from the assigned date
of the initial MI to the date of reinfarction or death
from coronary heart disease.
Data Analysis
The x2 test was used as a means of comparing

electrocardiographic recovery patterns 1) between

Downloaded from http://circ.ahajournals.org/ by guest on November 25, 2011

782

Circulation Vol 81, No 3, March 1990

men and women and 2) by location of initial MI

(anterior vs. inferior), by examining the proportions
of those in each of the four follow-up electrocardiographic categories. The percentages of men and
women with other specified electrocardiographic
abnormalities (as previously described) were compared in a similar manner. The Cox proportional
hazards model1213 was used to demonstrate the association of specific electrocardiographic abnormalities
and electrocardiographic classification (as compared
with those with a normal ECG) on the follow-up
exam with the long-term hazard of reinfarction and
coronary mortality. Total sample sizes varied by the
number with each specified characteristic. These
analyses were performed, adjusting for age and sex
alone, and in multivariable analysis (for analyses
involving electrocardiographic classifications),
adjustments were made for the effects of age, sex,
and standard risk factors (i.e., systolic blood pressure, serum cholesterol, relative weight, and diabetes, depending on the end point) previously
identified14 to be of prognostic value in postinfarction participants in Framingham. Smoking status is
excluded from these analyses because of its failure
to demonstrate prognostic value among Framingham post-MI subjects.15 Further analyses with coronary mortality as an end point were done with
additional adjustment for previously identified15
clinical variables of prognostic importance in
Framingham, including New York Heart Association (NYHA) functional classification, x-ray left
ventricular hypertrophy, x-ray generalized cardiac
enlargement, and diuretic usage. NYHA functional
classification involved comparing those persons
having moderate or worse impairment (level 2 or
higher) with those persons having minimum or no
impairment (level 1). X-ray left ventricular hypertrophy required that the heart chamber be enlarged
by extension of the left apex of the heart towards
the left, with an elevation and squaring of the apex,
whereas generalized cardiac enlargement indicated
a rounding and expanding of the heart to both the
right and left.3
Where electrocardiographic classification was used
to predict reinfarction, an additional 65 subjects were
excluded. This included 1) 10 subjects who experienced reinfarction before attending the post-MI
baseline examination for measurement of risk factors, and 2) 55 subjects who died of coronary heart
disease but not identified as having a documented
recurrent MI. A total of 186 subjects with a follow-up
ECG were, therefore, eligible for analysis with reinfarction as an end point. Multivariable analyses were
based on the number of cases with complete riskfactor data available.
The association of each of four specific electrocardiographic abnormalities present in the ECG before
MI, that is, 1) ST segment abnormality, 2) T wave
abnormality, 3) left ventricular hypertrophy, and
4) atrioventricular block (first degree or higher) with
the long-term risk of coronary death adjusted for age

and sex, was evaluated by Cox proportional hazards

regression. Because of the small number of subjects
with bundle branch block prior to MI, this was not
evaluated in this portion of the analyses. Each of
these features, one or more of which were present in
approximately 80% of those persons with an abnormal overall electrocardiographic impression were
also examined together by similar techniques in
multivariable analyses with post-MI electrocardiographic classification. This allowed for examining
whether the prognostic value of the post-MI ECG
could be explained by preexisting electrocardiographic abnormalities.
Analysis of variance was used to compare riskfactor differences (for serum cholesterol, systolic
blood pressure, and relative weight) among categories of post-MI electrocardiographic status, and the
Student's t test was used to evaluate such differences
according to the presence or absence of specified
electrocardiographic abnormalities (i.e., nonspecific
ST segment and T wave abnormalities, left ventricular hypertrophy, and atrioventricular block). The x2
test of proportions was used for these analyses for
diabetes, which was a binary variable.
From the regression coefficients obtained, relative
risks (RRs) and 95% confidence limits were calculated
to compare the risk of long-term reinfarction or coronary mortality in each abnormal electrocardiographic
classification group as compared with the reference
group consisting of those with no definite abnormalities. RRs relating the presence versus the absence of
specific electrocardiographic abnormalities with these
end points were calculated; ap value of 0.05 or less was
considered statistically significant.

Results
Of the 251 subjects in whom a follow-up ECG was
obtained, 101 subsequently died of coronary disease
and 70 were diagnosed with a recurrent infarction.
One third of coronary deaths occurred within 5 years
of initial infarction, and 62% occurred within 10
years. Persistence of pathological Q waves was evident in 183 (73.0%) subjects, a definite electrocardiographic abnormality without Q wave MI was
present in 37 (14.7%), and 31 (12.4%) showed a
normalized ECG. The presence of ST segment or T
wave abnormalities, left ventricular hypertrophy,
atrioventricular block, bundle branch block or any of
these in combination accompanied 75 of those with a
persistent Q wave MI. Electrocardiographic classification on follow-up examination was significantly
different between men and women, with women
more likely to have residual abnormalities other than
a diagnostic Q wave pattern (p<0.05) (Table 2).
Furthermore, electrocardiographic recovery patterns
were significantly different by location of initial
infarction, with twofold the proportion of those with
an inferior MI as compared with those with an
anterior MI demonstrating a follow-up ECG with no
definite abnormality (17% vs. 8%, respectively).
Compared with the proportion of those with no

Downloaded from http://circ.ahajournals.org/ by guest on November 25, 2011

Wong et al The Electrocardiogram and Post-Myocardial Infarction Prognosis

783

TABLE 2. Electrocardiogram Classification by Sex on Follow-up Examination

Men
(n) (%)
26 (13.7)

Classification

Women
(n) (%)
5 (8.2)
17t (27.9)
23 (37.7)
16 (26.2)
61 (100.0)

Total
(n) (%)
31(12.4)
37 (14.7)
108 (43.0)
75 (30.0)
251 (100.0)

No definite abnormality
20 (10.5)
Definite ECG abnormality without persistent Q wave MI
85 (44.7)
Persistent Q wave MI without other specified abnormalities*
59 (31.0)
Persistent Q wave MI with other specified abnormalities*
190 (100.0)
Total
tp<0.05, as compared with proportions of men and women with no definite abnormality.
*Indicates definite presence of one or more of following: ST segment changes, T wave changes, left ventricular
hypertrophy, atrioventricular block, or bundle branch block.

definite abnormality, a persistent Q wave MI with

other abnormalities was significantly more common
in those with an anterior infarct (p<0.05) (Table 3).
Similar proportions of those with anterior and inferior MIs, however, demonstrated a persistent Q wave
MI on follow-up ECG (78% and 70%, respectively).
Of other specified abnormalities examined on the
follow-up ECG, there were no significant differences
in prevalence among men and women (Table 4).
Data on the separate effects of each of several
specific electrocardiographic abnormalities present
on the follow-up ECG with the long-term risks for
reinfarction and coronary mortality are presented in
Table 5. The reference group in each instance
included only those cases with a normalized post-MI
ECG. ST segment and T wave abnormalities, as well
as left ventricular hypertrophy, were significantly
associated (p<0.01) with the long-term risk for coronary mortality with RRs of 3.49, 3.27, and 7.73,
respectively. Although a doubling of risk was noted
for other bivariate associations of electrocardio-

graphic abnormalities with reinfarction or coronary

death, none was statistically significant.
In Cox regression analyses adjusted for the effects
of age and sex (Table 6), neither an abnormal ECG
alone nor a persistent Q wave MI (with or without
other abnormalities) conferred a significant excess
risk of reinfarction. When predicting coronary death,
however, those with a definite electrocardiographic
abnormality, but without a persistent Q wave MI,
were at more than a threefold increased risk
(RR=3.46,p<0.01) as compared with those persons
with no definite ECG abnormality. Subjects with a
persistent Q wave MI with other abnormalities were
also at a significantly increased risk of coronary death
(RR=2.70, p=0.01). In those with a persistent
Q wave MI, but without other indicated abnormalities, no significant increase in risk of coronary death
was demonstrated. In multivariable analyses with
additional adjustment for the effects of systolic blood
pressure, serum cholesterol, and diabetes, the
increased risk in those persons with definite electro-

TABLE 3. Number and Percentage of Cases by Electrocardiogram Classification on Follow-up Examination by

Location of Initial Myocardial Infarction

Anterior

Inferior

(n) (%)

(n) (%)

22 (16.9)
9 (8.0)
No definite abnormality
17 (13.1)
16 (14.3)
Definite ECG abnormality without persistent Q wave MI
61 (46.9)
46 (41.1)
Persistent Q wave MI without other specified abnormalities*
41 (36.6)
Persistent Q wave MI with other specified abnormalities*
30t (23.1)
112 (100.0)
130 (100.0)
Total
n=242. Location of initial myocardial infarction identified to be anterior and inferior in six cases, and was not
available in three cases.
*Presentation on follow-up ECG with one or more of following: ST segment or T wave changes, left ventricular
hypertrophy, atrioventricular block, or bundle branch block.
tp<0.05, as compared with proportions of anterior and inferior MIs with no definite abnormality on the followup ECG.
TABLE 4. Percentages of Specified Electrocardiogram Abnormalities on Follow-up Examination
Women
Men
Abnormality
18.0
18.4
Nonspecific ST segment changes (%)
31.2
20.5
Nonspecific T wave changes (%)
9.8
5.8
Left ventricular hypertrophy (%)
18.0
11.6
Atrioventricular block (%)
4.9
6.8
Bundle branch block (%)

Differences in proportions between males and females were not statistically significant.
Downloaded from http://circ.ahajournals.org/ by guest on November 25, 2011

Total
18.3
23.1
6.8
13.2
6.4

784

Circulation Vol 81, No 3, March 1990

TABLE 5. Specified Electrocardiogram Abnormalities on Followup ECG and Long-term Risk of Reinfarction and Coronary Death
After Recognized Q Wave Myocardial Infarction

Abnormality
Nonspecific ST segment
changes
Nonspecific T wave
changes
Left ventricular

hypertrophy
Atrioventricular block
Bundle branch block

Reinfarction
RR*
1.63
(0.56,7.86)
1.61
(0.55,4.77)
1.91

Coronary death
RR*
3.49t
(1.43,8.53)
3.27t
(1.31,8.18)
7.73t

(0.20,17.75)

(1.92,31.07)

1.92
(0.58,633)
1.30
(0.39,4.39)

2.37
(0.75,7.47)
1.76
(0.57,5.46)

*RR, Adjusted relative risks are estimates from Cox proportional hazards model adjusted for age and sex; comparisons are to
reference group with normalized follow-up ECG. Numbers in
parentheses indicate 95% confidence limits of relative risk.

p<0.01

cardiographic abnormalities with or without a persistent Q wave MI remained highly significant and of
similar magnitude as in the analyses adjusted for age
and sex (Table 7).
Further adjustment for clinical variables including
NYHA functional class (level 2 or higher vs. 1),
generalized cardiac enlargement and left ventricular
hypertrophy by x-ray, and diuretic usage, demonstrated in this sample and previously15 to be important prognostic factors, weakened slightly the effect
of a persistent Q wave MI in the presence of other
abnormalities. An abnormal ECG with a normalized
Q wave pattern, however, continued to remain an
important prognostic indicator of subsequent coronary death (Table 8).
The 1) prevalences of preexisting electrocardiographic abnormalities on the ECG before MI and
2) associations of these abnormalities with the risk of
TABLE 6. Follow-up ECG Abnormality Status and Long-term
Risk of Reinfarction and Coronary Death After Recognized
Q Wave Myocardial Infarction
ECG Abnormality status
Definite ECG abnormality,
without definite Q wave MI
Persistent Q wave MI without
other specified
abnormalities
Persistent Q wave MI
with other specified
abnormalities

Reinfarction
RR*
1.13
(0.42,3.01)
1.22
(0.56,2.63)

Coronary death
RR*
3.46t
(1.45,8.26)
1.78
(0.83,3.83)

1.32
(0.57,3.02)

2.70t
(1.25,5.84)

*RR, Adjusted relative risks are estimates from Cox proportional hazards model adjusted for age and sex; comparisons are to
reference group with follow-up ECG showing no definite abnormality. Numbers in parentheses indicate 95% confidence limits of
relative risk.
tp<0.05.
tp<0.01.
Presentation on follow-up ECG with one or more of following:
ST segment or T wave changes, left ventricular hypertrophy,
atrioventricular block, or bundle branch block.

TABLE 7. Follow-up ECG Abnormality Status and Long-term

Risk for Reinfarction and Coronary Death After Recognized
Q Wave Myocardial Infarction
Reinfarction Coronary death
RR*
RR*
ECG Abnormality status
3.68t
Definite ECG abnormality,
0.58
(1.41,9.56)
(0.16,2.02)
without definite Q wave MI
1.41
1.93
Persistent Q wave MI,
(0.85,4.41)
(0.61,3.27)
without other specified

abnormalities
1.39
2.92t
Persistent Q wave MI,
(1.27,6.72)
with other specified
(0.56,3.46)
abnormalities
*RR, Risk factor adjusted relative risks are estimates from Cox
proportional hazards model; reinfarction analysis (n=139)
adjusted for age, sex, systolic blood pressure, relative weight, and
serum cholesterol; coronary death analysis (n=199) adjusted for
age, sex, systolic blood pressure, serum cholesterol, and diabetes.
Reference group for all estimates are cases with no definite ECG
abnormality on follow-up examination. Numbers in parentheses
indicate 95% confidence limits of relative risk.

tp<0.05.
tp<0.01.
Presentation on follow-up ECG with one or more of following:
ST segment or T wave abnormality, left ventricular hypertrophy,
atrioventricular block, or bundle branch block.

coronary death post-MI are presented in Table 9.

The various ECG abnormalities are not exclusive
classifications; more than one abnormality can be
present in a given individual. Although 236 subjects
had an ECG before MI, not all subjects had information on specific electrocardiographic abnormalities available. Approximately one fourth (25.8%) of
subjects who subsequently had an infarction had an
abnormal ECG before MI, with nonspecific ST
segment and T wave abnormalities being the most
prevalent of the specified abnormalities. In Table 9,
the association of post-MI electrocardiographic status adjusted for the effect of preexisting electrocardiographic abnormalities is also presented. The electrocardiographic abnormalities before MI that are
significantly associated with coronary death include
nonspecific ST segment changes (RR=4.67,p <0.001),
nonspecific T wave abnormalities (RR=2.76,
p<O.OO1), and left ventricular hypertrophy (RR=4.29,
p<0.01). ST segment and T wave changes and left
ventricular hypertrophy each remained significant
prognostic factors even after adjusting for post-MI
electrocardiographic classification. An abnormal
follow-up electrocardiographic examination was no
longer prognostic once ST segment or T wave abnormalities or atrioventricular block on the electrocardiographic tracing before MI were considered. When
simultaneous adjustment is made for each of the
preexisting electrocardiographic abnormalities (ST
segment and T wave abnormalities, left ventricular
hypertrophy, and atrioventricular block), the risk conferred by an abnormal follow-up ECG (RR=1.221.82, depending on category) is no longer statistically
significant. An abnormal follow-up ECG, however,
particularly in the absence of a Q wave MI (RR=3.6,
p<0.01), retained predictive value after adjustment

Downloaded from http://circ.ahajournals.org/ by guest on November 25, 2011

Wong et al The Electrocardiogram and Post-Myocardial Infarction Prognosis

TABLE 8. Follow-up ECG Abnormality Status and Long-term
Risk for Coronary Death: Multivariable Analysis With Standard
and Clinical Risk Factors
Relative risk
Variable
1.35
Age (10-year increment)
Sex (female: male)
0.15t
1.54i:
Systolic blood pressure (25 mm Hg increment)
1.37:
Serum cholesterol (50 mg/dl increment)
2.804
Diabetes (yes: no)
1.74t
Functional classification (level 2+ :1)
1.88
X-ray heart enlargement (definite: other)
X-ray left ventricular hypertrophy (definite: other)
2.63t
3.824
Diuretic usage (current:none)
1.93
Diuretic usage (previous:none)
ECG abnormality status
Definite ECG abnormality, without definite
4.264
Q wave MI: normal
Persistent Q wave MI, without other specified
1.90
abnormalities: normal
Persistent Q wave MI, with other specified
2.48
abnormalities: normal
Values are for 164 cases under analysis.
tp<0.05; tp<0.Ol in multivariable analysis.

for an ECG before MI that demonstrated left ventricular hypertrophy.

Table 10 shows the association of serum cholesterol,
systolic blood pressure, relative weight, and diabetes
assessed at the follow-up examination with post-MI
electrocardiographic classification. Both age and systolic blood pressure differed significantly (p<0.05 and
p<0.01, respectively) across categories of electrocardiographic classification, with those subjects having an
abnormal ECG on follow-up, but without a persistent
Q wave MI, being older and having a higher systolic
blood pressure than those whose ECGs returned to
normal. Table 11 shows the mean levels of these risk
factors with the presence versus absence of specified
electrocardiographic abnormalities. Those subjects
with left ventricular hypertrophy were significantly
(p<0.01) older and had higher systolic blood pressure
levels, and those subjects with atrioventricular block
were also significantly (p<0.01) older and had a
higher (p<0.05) prevalence of diabetes than those
without atrioventricular block.

Discussion
The findings from this population-based study
demonstrate the long-term (mean follow-up of 10
years with available follow-up of up to 32 years)
prognostic implications of a routine ECG obtained
after recovery from an initial MI. The presence of an
abnormal ECG on follow-up after infarction is sufficient to predispose the surviving MI patient to a
significantly greater risk of coronary death. We demonstrate such individuals to be at more than threefold the risk of subsequent death from coronary
disease as compared with those with a normalized
post-MI ECG, a risk that remained important even
after considering standard coronary risk factors.

785

The Q wave MI is documented by a panel of

Framingham physicians from the ECG obtained at
the time of hospitalization. Its persistence or disappearance was based on the findings of the follow-up
ECG performed at a subsequent biennial examination. We showed a Q wave MI to persist in approximately 73% of cases in whom a follow-up ECG was
available. Other previous reports have demonstrated
varying rates of regression of Q waves. Throughout
a 1-2-year follow-up, reports have demonstrated
Q waves to persist in approximately 85% of patients,16,17 although one report demonstrated complete regression of Q waves to occur in 21% of
patients during 3 months.18 Others report disappearance of Q-QS abnormalities to occur in nearly half of
patients in the first year, with the majority of regression occurring within 2 months.19 In those subjects
with a follow-up ECG performed at least 2 years after
the infarction, partial or total regression has been
demonstrated to occur in 60% of patients.20
We demonstrated that 12.4% of persons with a
Q wave MI showed a normalized ECG on follow-up.
This contrasts with other reports showing that only in
5.6% of such patients do the ECGs return to normal
(although reported rates of Q wave regression were
similar).21 We report that if a reference category of
4'no definite ECG abnormality" is used, an increased
risk is seen in those subjects with a persistent Q wave
MI, which is statistically significant when accompanied by other electrocardiographic abnormalities.
This latter group comprised a full 30% of cases
returning for a follow-up examination. Data from the
Health Insurance Plan of New York showed the risk
of death during 4.5 years to be significantly less in
those persons with a history of MI or angina pectoris,
whose ECG returned to normal within about 6
months, as compared with those whose ECGs did not
return to normal.22 Another 5-year follow-up study
demonstrated that those persons who have Q wave
changes or other significant ECG changes persisting
on follow-up ECG, 1-2 years after the primary
infarct, have a significantly higher mortality than
those with a normalized ECG.23 Furthermore, the
Coronary Drug Project2 has reported Q-QS abnormalities on the follow-up ECG to be significantly
associated with subsequent coronary mortality, even
in multivariable analyses with other clinical and
standard risk factors. Reinfarction has also been
shown to be more common in those persons with
progression of Q or QS wave changes during the first
year after MI24 and in those with a persistent U wave
at the time of hospital discharge.25 Other studies,
however, have reported the persistence of diagnostic
Q waves to be unrelated to subsequent survival20,26'27
or recurrent MI.21
It is uncertain how long other specified electrocardiographic abnormalities examined in this investigation were present. By comparing the follow-up postMI ECG with the ECG obtained before MI, we were
able to demonstrate that 28% of those persons with
an abnormal ECG (persistent Q wave with or without

Downloaded from http://circ.ahajournals.org/ by guest on November 25, 2011

786

Circulation Vol 81, No 3, March 1990

TABLE 9. Specified ECG Abnormalities on ECG Before Myocardial Infarction and Long-term Risk of Coronary Death
Adjusted for age, sex,
Adjusted for age
and ECG status
and sex (RR)
ECG abnormality
(95% CL)
(95%
CL)*
(percentage with abnormality)
1
1
3.891
Nonspecific ST segment changes (13.4%)
4.67t
(1.92,7.84)
(2.39,9.11)
2.33
ECG1
1.00
ECG2 1
1.63
ECG3T
2.301
2.761
T
wave
abnormalities
Nonspecific
(17.2%)
(1.34,3.94)
(1.65,4.61)
2.34
ECG1
1.40
ECG2
2.11
ECG3
4.291
3.08t
Left ventricular hypertrophy (3.8%)
(1.18,8.07)
(1.67,11.06)
ECG1
3.61t
ECG2
1.68
2.761
ECG3
Atrioventricular block (first degree or
1.28
1.73
higher) (8.1 %)

(0.36,4.49)

(0.52,5.83)
ECG1
ECG2
ECG3
All specified abnormalities together
Nonspecific ST segment
abnormalities
Nonspecific T wave abnormalities
Left ventricular hypertrophy
Atrioventricular block

ECG1
ECG2
ECG3

2.10
1.03
2.28

3.02t
1.66
4.271
1.16
1.82
1.22
1.64

Values are relative risk (RR) estimates.

*95% confidence limits of relative risk.

tp<0.05.
tp<0.Ol.
ECG1, Comparison of cases at follow-up ECG with definite ECG abnormality but without definite Q wave MI as
compared with those with normalized ECG.
1 ECG2, Comparison of cases at follow-up ECG with persistent Q wave MI but without other abnormalities (ST
segment or T wave abnormality, left ventricular hypertrophy, or atrioventricular block) as compared with those with
normalized ECG.
ECG3, Comparison of cases at follow-up ECG with persistent Q wave MI with other abnormalities as compared
with those with normalized ECG.

other abnormality) after MI had abnormalities other

than Q waves on examination done an average of 1
year before infarction. Furthermore, of those persons
with electrocardiographic abnormalities other than
Q wave MI at follow-up, 41% had such abnormalities
present before MI. In this study, the prognostic
importance of an abnormal ECG before MI was also
examined. The presence of nonspecific ST segment
changes, T wave abnormalities, or left ventricular
hypertrophy on the electrocardiographic tracing
before infarction were grave prognostic indicators,
alone, and also when considering post-MI electrocardiographic classification. In fact, ST segment and
T wave abnormalities before MI were such powerful

risk factors that, when adjusted for, diminished the

value of the follow-up ECG. In the Coronary Drug
Project,2 ST segment depression was the most important prognostic indicator after MI, and ventricular
conduction defects remained significantly related to
coronary death, even in multivariable analyses.
Although strong associations are apparent when
predicting coronary death, this investigation failed to
show a significantly increased risk of reinfarction
from follow-up electrocardiographic status. It is possible that preexisting abnormalities made it more
difficult to diagnose new infarctions, especially
because all cases, on entry into the study, had
diagnostic Q waves. This might be, in particular, the

Downloaded from http://circ.ahajournals.org/ by guest on November 25, 2011

Wong et al The Electrocardiogram and Post-Myocardial Infarction Prognosis

787

TABLE 10. Risk Factor Mean Levels by Postinfarction ECG Classification

Postinfarction ECG classification*

0
1
2
3
p
Age (yr)
59.2
64.3
59.0
61.1
<0.05t
Serum cholesterol (mg/dl)
249.7
262.1
246.1
242.4
Systolic blood pressure (mm Hg)
135.7
147.7
131.1
139.3
<0.Olt
Relative weight (% of ideal)
120.4
118.1
120.3
118.8
Diabetes (%)
6.5
18.9
8.3
16.0
...
*Categories of ECG classification defined as follows: 0, reference group, no definite ECG abnormality; 1, definite
ECG abnormality without definite Q wave MI; 2, persistent Q wave MI without other specified (one or more of
following: ST segment or T wave abnormality, left ventricular hypertrophy, atrioventricular block, or bundle branch
block) abnormalities; 3, persistent Q wave MI with other specified abnormalities.
tp value across categories of ECG classification.

case for those reinfarctions that were clinically

unrecognized, or identified only by the observation
of new diagnostic changes on the follow-up ECG
(i.e., new location of reinfarction as compared with
initial MI). If so, reinfarctions in those persons with
persistent Q wave MI would have been underestimated (i.e., when occurring in a location adjacent to
TABLE 11. Risk Factor Mean Levels by Presence of Specified ECG
Abnormalities Postinfarction

Nonspecific ST segment changes

Age (yr)
Systolic blood pressure (mm Hg)
Serum cholesterol (mg/dl)
Relative weight (% of ideal)
Diabetes (%)
Nonspecific T wave changes
Age (yr)
Systolic blood pressure (mm Hg)
Serum cholesterol (mg/dl)
Relative weight (% of ideal)
Diabetes (%)
Left ventricular hypertrophy
Age (yr)
Systolic blood pressure (mm Hg)
Serum cholesterol (mg/dl)
Relative weight (% of ideal)
Diabetes (%)
Atrioventricular block
Age (yr)
Systolic blood pressure (mm Hg)
Serum cholesterol (mg/dl)
Relative weight (% of ideal)
Diabetes (%)
Bundle branch block
Age (yr)
Systolic blood pressure (mm Hg)
Serum cholesterol (mg/dl)
Relative weight (% of ideal)
Diabetes (%)

Absent

Present

61.2
136.2
245.8
120.1
11.7

59.3
138.3
257.5
117.4
13.0

60.7
135.5
245.2
119.3
11.9

61.4
140.2
257.1
120.5
12.0

67.6*

60.4
134.5
249.8
119.9
10.7

223.4
114.9
29.4

66.7
143.6
245.5
117.2
24.2

135.6
248.7
120.0
10.1t

60.7
136.6
248.1
119.6
11.9

63.4
135.6
248.2
119.1
12.5

164.2t

60.Ot

tp<0.05 or tp<0.01 as compared with mean or prevalence of

risk factor in those without specified ECG abnormality.

the initial Q wave MI), partially explaining the

attenuated findings.
Our discovery of women having a markedly different post-MI electrocardiographic profile than men,
with a substantially greater proportion of women
demonstrating electrocardiographic abnormalities in
the absence of a persistent Q wave MI (28% vs. 11%
in men), to our knowledge, has not been demonstrated elsewhere. These abnormalities other than
Q wave MI, if more common in women than in men,
can contribute to their greater early mortality after
MI, as demonstrated by other investigators.2829 In
relation to location of initial MI, our discovery that
those persons with an anterior MI were less likely to
have a normalized ECG after MI might help explain
reports in the literature demonstrating that such individuals have a poorer prognosis.30-32
It is clear that other clinical variables and standard
risk factors have an important role in post-MI risk
stratification. Some investigators have questioned
whether the follow-up ECG yields additional prognostic usefulness if other clinical information such as
persistent heart failure or heart enlargement on x-ray
are available.33 We demonstrated the continuing
grave prognosis of an abnormal ECG without a
persistent Q wave MI even after noting other relevant clinical and standard risk factors. Although
systolic blood pressure was strongly associated with
post-MI electrocardiographic status (apparently
because of its strong relation to left ventricular
hypertrophy), both remained independent predictors
of coronary heart disease death in multivariable
analyses.
The fact that individuals must survive 1 year, on
average, to attend the follow-up exam precludes the
ability to detect the prognostic importance of those
electrocardiographic abnormalities that are associated with the greatest early mortality. A larger study
has demonstrated the presence of bundle branch
block and premature ventricular beats are important
predictors of 5-year all-cause and cardiovascular
mortality.2 In our investigation, however, the ability
to detect the separate impact of these factors was
limited because of the few survivors presenting with
these abnormalities on their follow-up ECG. Only for
nonspecific ST segment and T wave abnormalities

Downloaded from http://circ.ahajournals.org/ by guest on November 25, 2011

788

Circulation Vol 81, No 3, March 1990

were the age- and sex-adjusted RRs statistically

significant for predicting coronary death. The findings of this study are, therefore, most applicable to
relatively healthy cohorts of survivors of MI (e.g.,
those who have survived an average of 1 year).
Our data underscore the prognostic importance of
the follow-up ECG, after an initial infarction, as a
means for risk stratification, independent of the
effects conferred by the standard risk factors. This is
important, considering that prognosis is poorer for
those with high blood cholesterol, hypertension, and
diabetes.2,14,15,22 The follow-up ECG after MI, in
which the Q wave has disappeared but with other
abnormalities present, is demonstrated in this report
to be a grave prognostic indicator. This influence
remains strong even after the effect on prognosis of
an abnormal ECG or left ventricular hypertrophy
detected before infarction is considered, although
certain abnormalities existing before MI, such as ST
segment and T wave abnormalities are strongly associated with post-MI prognosis and, when taken into
consideration, attenuate markedly the predictive
value of the follow-up ECG. Nevertheless, our data
emphasize the continuing importance of the routine
follow-up ECG, years after recovery from MI. Additional information on electrocardiographic abnormalities present before infarction can also be useful
in helping to predict long-term prognosis. Abnormalities detected on the follow-up ECG, even if not
accompanied by symptoms, may warrant more
aggressive diagnostic or therapeutic measures.

References
1. Brush JE, Brand DA, Acampora D, Chalmer B, Wackers FJ:
Use of the initial electrocardiogram to predict in-hospital
complications of acute myocardial infarction. N Engl J Med
1985;312:1137-1 141
2. Schlant RC, Forman S, Stamler J, Canner PL: The natural
history of coronary heart disease: Prognostic factors after
recovery from myocardial infarction in 2789 men: The 5-year
findings of the Coronary Drug Project. Circulation 1982;
66:401-414
3. Sorlie P: Cardiovascular diseases and death following myocardial infarction and angina pectoris: Framingham Study, 20year follow-up, in Kannel WB, Gordon T (eds): The Framingham Study. Department of Health, Education, and Welfare
publication no. (NIH) 77-1247. Washington, DC, US Government Printing Office, 1977, pp 1-153
4. Schatzkin A, Cupples LA, Heeren T, Morelock S, Kannel WB:
Sudden death in the Framingham Heart Study: Differences in
incidence and risk factors by sex and coronary disease status.
Am J Epidemiol 1984;120:888- 899
5. Siltanen P, Pohjola-Sintonen S, Haapakoski J, Makijarvi M,
Pajari R: The mortality predictive power of discharge electrocardiogram after first acute myocardial infarction. Am Heart J
1985;109:1231-1237
6. Velema JP, Lubsen J, Pool J, Hugenholz PG: Can cardiac
death be predicted from an ambulatory 24-hour ECG?
J Chronic Dis 1985;38:233-239
7. Kreger BE, Kannel WB, Cupples LA: Electrocardiographic
precursors of sudden unexpected death: The Framingham

Study. Circulation 1987;75(suppl II):II-22-II-24

8. Lembo NJ, Starling MR, Dell'Italia LJ, Crawford MH, Chaudhuri TK, O'Rourke RA: Clinical and prognostic importance of
persistent precordial (V1-V4) electrocardiographic ST segment depression in patients with inferior transmural myocardial infarction. Circulation 1986;74:56-63

9. Dawber TR, Meadors GF, Moore FE: Epidemiologic

approaches to heart disease: The Framingham Study. Am J
Public Health 1951;41:279-286
10. Metropolitan Life Insurance Company Statistical Bulletin. Vol
40, 1959
11. Gordon T, Shurtleff D: Means at each examination and
interexamination variation of specified characteristics, in Kannel WB, Gordon T (eds): The Framingham Study. Department
of Health, Education, and Welfare publication no. (NIH)
74-478. Washington, DC, US Government Printing Office,
1973
12. Cox DR: Regression models and life tables (with discussion).
J Roy Stat Soc (B) 1972;34:187-220
13. Harrell FE: The PHGLM Procedure, in SUGI Supplemental
User's Guide. Cary, NC, SAS Institute, Inc, 1983
14. Wong ND, Cupples LA, Ostfeld AM, Kannel WB: Risk
factors for long-term prognosis after myocardial infarction:
The Framingham Study. Am J Epidemiol 1989;130:469-680
15. Wong ND: Clinical determinants for long-term risk of reinfarction and coronary mortality following initial myocardial
infarction: The Framingham Study (doctoral dissertation).
Yale University, New Haven, Conn, 1987
16. Kabfleisch JM, Shadaksharappa KS, Conrad LL, Sarkar NK:
Disappearance of the Q-deflection following myocardial
infarction. Circulation 1968;76:193-198
17. Anderssen N, Skjaeggestad 0: The electrocardiogram in
patients with previous myocardial infarction. Acta Med Scand
1964;176:123-126
18. Yusuf S, Estrada-Yamamoto M, Reyes CP, Herlitz J, Hjalmarson A: Factors of importance for QRS complex recovery
after acute myocardial infarction. Acta Med Scand 1982;
211:157-162
19. Pyorala K, Kentala E: Disappearance of Minnesota code
Q-QS patterns in the first year after myocardial infarction.
Ann Clin Res 1974;6:137-141
20. Marcus EB, Yano K, MacLean CJ: Regression of Q-waves
following acute myocardial infarction. Am J Epidemiol 1989;
129:105-111
21. Cox CJB: Return to normal of the electrocardiogram after
myocardial infarction. Lancet 1967;1:1194-1197
22. Weinblatt E, Shapiro S, Frank CW: Prognosis of women with
newly diagnosed coronary heart disease-A comparison with
course of disease among men. Am J Public Health 1973;
63:577-593
23. Leren P, Hjermann I, Bjornson J: Prognostic significance of
ECG-changes in survivors of myocardial infarction. Acta Med
Scand 1970;188:523-526
24. Rissanen V, Raunio H, Halinen MO, Rehnberg S: Changes
occurring in the postinfarction ECG in relation to age, sex,
and previous myocardial infarction. Am Heart J 1986;
111:286-292
25. Kishida HK, Hata N, Nejima J, Suzuki T, Otsu F, Saito T,
Kusama Y: Prognosis of patients with persistent negative
U wave following myocardial infarction. Jpn Heart J 1987;
28:15-25
26. Kaplan BM, Berkson DM: Serial electrocardiograms after
myocardial infarction. Ann Int Med 1964;60:430-435
27. Wasserman AG, Bren GB, Ross AM, Richardson DW,
Hutchinson RG, Rios JC: Prognostic implications of diagnostic Q-waves after myocardial infarction. Circulation 1982;
65:1451-1455
28. Kannel WB, Sorlie P, McNamara PM: Prognosis after initial
myocardial infarction: The Framingham Study. Am J Cardiol
1979;44:53-59
29. Puletti M, Sunseri L, Curione M, Erba S, Borgia C: Acute
myocardial infarction: Sex-related differences in prognosis.
Am Heart J 1984;108:63-66
30. Davis HT, DeCamilla J, Bayer LW, Moss AJ: Survivorship
patterns in the posthospital phase of myocardial infarction.
Circulation 1979;60:1252-1258
31. Luria MH, Debanne SM, Osman MI: Long-term follow-up
after recovery from acute myocardial infarction: Two and
five-year prognostic indices. Am J Med 1979;67:7-14

Downloaded from http://circ.ahajournals.org/ by guest on November 25, 2011

Wong et al The Electrocardiogram and Post-Myocardial Infarction Prognosis

32. Waters DD, Bosch X, Bouchard A, Moise A, Roy D, Pelletier
G, Theroux P: Comparison of clinical variables and variables
derived from a limited predischarge exercise test as predictors
of early and late mortality after myocardial infarction. J Am
Coll Cardiol 1985;5:1-8
33. Fioretti P, Tijssen JGP, Azar AJ, Lazzeroni E, Brower RW,
Katen JH, Lubsen J, Hugenholtz PG: Prognostic value of

789

predischarge 12 lead electrocardiogram after myocardial

infarction compared with other routine clinical variables.
Br Heart J 1987;57:306-312

KEY WORDS * cardiovascular diseases * electrocardiography a

myocardial infarction * coronary disease

Downloaded from http://circ.ahajournals.org/ by guest on November 25, 2011