Journal of lnternul Medicine 1997; 2 4 1 : 485-492
Diabetes as a risk factor for myocardial infarction:
population and gender perspectives
V. LUNDBERG", B. STEGMAYRb, K. ASPLUNDb, M. ELIASSON" & F. HUHTASAARF
From the Departments of Medicine at aKaJix Hospital,
Abstract. Lundberg V, Stegmayr B, Asplund K,
Eliasson M, Huhtasaari F (Kalix Hospital, Umel
University Hospital and Lulei-Boden Hospital,
Sweden). Diabetes as a risk factor for myocardial
infarction : population and gender perspectives.
J Intern Med 1997; 241: 485-92.
Objectives. To investigate diabetes as a risk factor for
acute myocardid infarction (AMI)from a population
perspective in a region with high cardiovascular
disease (CVD) risk.
Design. Population screenings for diabetes and a
population-based AMI register.
Setting. Northern Sweden MONICA area.
Subjects. Representative sample (Norrbotten and
Vasterbotten counties) of 2432 men and women
35-64 years was investigated 1990 and 1994. All
patients with AMI aged 35-64 years were included,
in total 3031 between 1989 and 1993.
Results. The prevalence of diabetes was 5% in men
and 4.4% in women. The relative risk (RR) in
Introduction
Diabetes is a well-known risk factor for the de-
velopment of coronary heart disease. The relative
risk (RR) seems to be higher in diabetic women than
in men [ 1 4 ] . In the Rancho Bernado study, a 40-79
years old cohort was followed prospectively for seven
years [3]. The relative risk of death attributed to
ischaemic heart disease was 2.5 for diabetic men and
3.4 for diabetic women compared to non-diabetic
control subjects after adjusting for smoking, hy-
pertension and hypercholesterolaemia. During the
follow-up in the Framingham study, diabetes doubled
the risk of fatal cardiovascular disease in men and
almost tripled the risk in women [l].
An increased mortality from acute myocardial
0 1997 Blackwell Science Ltd
University Hospital and 'Lul&-Boden Hospital, Sweden
diabetic men was 2.9; 95% coniidence interval (CI)
2.6-3.4, and in diabetic women, RR 5.0 : CI 3.9-6.3.
The risk for re-infarction was about twice as large in
patients with diabetes as in patients without diabetes.
In both sexes the overall 28 day case fatality (CF) was
significantly higher in diabetic compared to non-
diabeticsubjects. When compared to the non-diabetic
population, the overall mortality from AM1 in the
diabetic population was 4 times higher among men
and 7 times higher among women. The population
attributable risk (PAR), a crude estimate of all AMIs
ascribed to diabetes, was 11% in men and 17% in
women.
Conclusions. Diabetes increases the risk for AM1
attack rate, incidence, case-fatality, recurrence and
mortality and is an important contributor to all AMIs
in middle-aged people.
Keywords: acute myocardial infarction, case fatality,
diabetes mellitus, incidence, mortality, population
attributable risk.
infarction (AMI) in diabetic subjects may result from
a higher incidence of AME or a higher case fatality
(i.e. proportion of fatal events out of all AMIs in a
specified time period from onset). In a Finnish study,
case fatality in AMI patients before admission to
hospital was the same in diabetic as in non-diabetic
subjects, but 28 days case fatality was twice as high
in diabetic patients as in non-diabetic patients [l,51.
Myocardial infarctions were not found to be larger in
diabeticthan non-diabeticpatients but cardiac failure
was the main cause of death significantly more often
in diabetic than in nondiabetic patients [1, 51. Only
a few studies have focused on the gender differencein
the impact of diabetes on AMI [l,3, 51 and no one
has quantified the contribution of diabetes to the
burden of AM1 in the society.
48 5
486 V. LUNDBERG et al.
The aim of the present study was to investigate
diabetes as a risk factor for acute myocardial in-
farction (AMI)from a population perspective in the
ages 35-64 years in a region with high cardio-
vascular disease (CVD) risk [6]. Incidence and case
fatality rates have been measured and the relative
contribution of diabetes to all AMIs in the community
has been estimated.
Subjects and methods
The two northern-most counties in Sweden con-
stitute together one of the 38 collaborating centres in
the WHO MONICA (Multinational Monitoring of
Trends and Determinants in Cardiovascular Diseases)
Project. The main objective of this prospective project
is to assess how changes in the incidence of coronary
heart disease and cerebrovascular disease over a 10-
year-period relate to changes in cardiovascular risk
factors in the population. The Northern Sweden
MONICA Project covers a sparsely populated area of
154000 km2,with a total population of 510000
inhabitants.
Population risk factor surveys
The prevalence of diabetes (known diabetes and
diabetes diagnosed by an oral glucose tolerance test)
was estimated from two population surveys per-
formed in 1990 and 1994. The target population
(35-64 years) in the 1990 survey was 193 168 and
in 1994 it was 198473. The participants in the
surveys were selected by stratified randomization for
age ( 3 5 4 4 , 45-54, 55-64 years) and sex. In each
10-year stratum, 250 men and 250 women were
selected from continuously updated population regis-
ters. The selected persons were invited by letter to
participate in a health survey. Persons who did not
attend at the first invitation received a reminder
letter and were given a new appointment. Of the
1500 eligible and invited persons in 1990, 1236
participated, a response rate of 82.4%. The cor-
responding figures for the 1994 year survey were
1196 out of 1500, a response rate of 79.7%. A
questionnairethat included amongst other, questions
on smoking habits, previous cardiovascular diseases
and previously known diabetes mellitus was com-
pleted by the participants. Of all participants in the
two surveys, 65 persons (2.7%) stated that they had
diabetes. Among the remaining subjects a randomly
@ 1997 Blackwell Science Ltd Journal of Internal Medicine 241: 485-492
selected sub-sample of 1269 men and women under-
went a glucose tolerance test after an overnight fast.
Seventy-five grams of glucose was dissolved in
300 ml water and the solution was ingested withii
5 min. A venous blood sample was taken in sodium
fluoride tubes immediately before the glucose load
and after 2 hours. The samples were analysed by a
hexokinase method (Boehringer Mannheim Auto-
mated Analysis for BM/Hitachi System 717). Ac-
cording to WHO criteria, subjects were classiiied as
diabetics if the fasting plasma value exceeded
7.7 mmol L-’ or the 2-hour plasma glucose was
3 11.1 mmolL-l [7].
Myocardial infarction registration
Data in this study were collected from 1 April 1989
to 31 December 1993. AMI events in diabetic and
non-diabetic subjects were evenIy distributed over
the year. To get 5 full years, the numbers for 1989
have been multiplied by 1.33 throughout this paper.
During the study period, all AMIs in 35-64 year old
men and women were registered. The study popu-
lation for AM1 registration and for the risk factor
surveys was the same; 194 775 in 1991 (mid year).
All hospital discharge records, general practitioners’
reports and death certiicates with ICD codes
4 1 0 4 1 4 were screened for acute events and vali-
dated (methods and definitions presented elsewhere)
[8]. A M I was defined according to the WHO criteria
and AMI cases were registered in a standardized way
according to the WHO MONICA manual (WHO
MONICA Manual 1990).
Qualitycontrols have been carried out throughout
the study period. The same personnel have been
registering the AMI cases throughout the five years
of this study, ensuring consistency over time.
Data collection consisted of hospitalized patients’
admission records including information on medical
history, symptoms, ECG and cardiac enzymes, and in
some cases, general practitioners’reports. All patients
in hospitals were interviewed by specially trained
nurses concerning risk factors and medical history of
previous AMI.In fatal cases, information was also
obtained from death certificatesand necropsy reports.
In sudden death cases, a close relative Ned in a
questionnaire about risk factors with the same
questions as in the nurses’ interviews. Data about
type of diabetes were collected from hospital records
and general practitioners’ records.
 DIABETES A N D AM1 487

Information on previous AMI was available in all found in the population surveys. The 9 5 % confidence
but three cases. In 1 % (30/3064) (one man treated intervals (CI) were calculated according to the
in hospital and 24 men and 5 women with sudden binomial distribution for the number of events within
death), no information on diabetes was available. the age groups. The Mantel-Haenszel x2 statistics
These 33 cases were excluded from the study. In nine were used to test .differencesin proportions. In small
patients (all men), diabetes was diagnosed in con- samples, Fisher’s exact test was used. Confidence
nection to the hospital treatment for AMI;they were intervals for relative risks were calculated using the
included in the analyses. Epi-Info computer program [9]. Population attribu-
table risk (or aetiologic fraction) was calculated as
described by Annitage and Berry [lo].
Classification of diagnostic findings
The study was approved by the National Computer
Methods for classification and diagnosis are described Data Inspection Board and the Ethical Committee at
in detail elsewhere [8]. In addition to the enzymes the UmeH University.
recommended in the MONICA core study, namely
creatinine phosphokinase (CK), aspartate trans-
aminase (ASAT) and hydroxybutyric dehydrogenase
(LD),some optional cardiac enzymes were included Diabetes in the population
in the Northern Sweden MONICA Study (LD-1 and
CK-MB). In the two population risk factor surveys, 2.7%of the
AMI cases were classified into one of the categories participants had previously known diabetes mellitus
‘definite infarction’, ‘possible infarction or coronary and an additional 2.0% were found by an oral
death’. ‘ischaemic cardiac arrest with successful glucose tolerance test to have undiagnosed diabetes.
resuscitation not fulfilling criteria for definite or The prevalence of diabetes in the 35-64 year
possible myocardial infarction ’, ‘unclassi6able in- population was 5.0% in men and 4.4%in women.
farction ’ or ‘ not infarction ’. UnclassXiable events Table 1shows the age-specific diabetes prevalence in
were mostly fatal cases with a death certificate northern Sweden in men and women.
diagnosis of AM1 where no information on previous
history of AMI or of the clinical event was obtainable. Acute myocardial infarction
In this paper, only cases classified as ‘definite
During the 5 years 1989 to 1993, 2415 men and
infarction’ have been included in non-fatal events. In
616 women fulfilled the MONICA criteria for acute
fatal events, ‘possible infarction ’ and ‘unclassifiable
myocardial infarction (first ever or recurrent). Of all
infarction’ have also been included. As agreed upon
by the MONICA collaborators (WHO MONICA
AMI patients, 15% of the men and 20% of the
Manual 1990) ‘attack rate’ includes all events (first
Table 1 Previously known diabetic subjects and diabetic
ever or recurrent), whereas ‘incidence’refers to first subjects detected by an oral glucose tolerance test in individuals
ever AMI events only. Subjects who died within 2 8 participating in population surveys 1990 and 1994
days from the onset of AMI were recorded as fatal
Previously
cases. known Detected at
Diabetes was subtyped as IDDM (insulin dependent diabetes survey Total prevalence
diabetes mellitus) or NIDDM (non insulin dependent
diabetes mellitus) according to the WHO criteria [7]. n (%I n (%I (%) 95%CI
Men
35 4 4 51379 1.3 21184 1.1 2.4 (1.1-3.7)
Statistical analyses 45-54 91398 2.3 41212 1.9 4.2 (2.6-5.7)
5 5-64 211413 5.1 71216 3.2 8.3 (6.2-10.4)
The population sue increased slightly from 19 19 5 1 131612 2.1 5.0 (4.0-6.0)
35 - 6 4 3511190 2.9
in 1989 in the 35-64 year age range to 197 172 in Women
1993. Calculations of attack rate and incidence were 35 4 4 51407 1.2 01206 0 1.2 0.4-2.1)
based on the 1991 midyear population (194775). 45-54 101416 2.4 71236 3.0 5.4 (3.7-7.1)
5564 151417 3.6 61215 2.8 6.4 (4.48.4)
The total number of diabetic persons in the popu- 3 5-64 131657 2.0 4.4 (3.5-4.7)
3011240 2.4
lation was estimated from the prevalence of diabetes
8 1997 Blackwell Science Ltd Journal of Internal Medicine 241: 4 8 5 4 9 2
488 V. LUNDBERG et al.

Table 2 Total number of diabetic and nondiabetic AMI patients

in the 35-64 year age group, 1989-1993 in northern Sweden.
Per cent (%) of diabetic and nondiabetic patients in parentheses

Diabetic Non-diabetic
patients patients
G
8
n n
Total
n
4 In
m
(%I (%I &
M
8
Men m
35-44 18 (11.5) 139 (88.5) 157 s
45-54 76 (11.8) 567 (88.2) 643 $
55-64 269 (16.7) 1346 (83.3) 1615 2
Total 363 (15.0) 2052 (85.0) 2415 -C

Women ye 5
3 5-44
45-54
8
12
(21.6)
(9.6)
29
113
(78.4)
(90.4)
37
125
Ba s
m
m
55-64 105 (23.1) 349 (76.9) 454 $ c
0

:b
Total 125 (20.2) 491 (79.8) 616 % %
? 6

women had diabetes (Table 2). In men with first ever I

AMI, 13% had diabetes, in women 17%. Among Y G

8
diabetic men and women having their 6rst AMI 26 In
m
out of 266 subjects were classified as IDDM. 9
As shown in Table 3, attack rates, incidence and G
rates of recurrent AMIs increased markedly with age s
In
m
with one exception: in diabetic women, the rates
were consistently higher in the 3 5 4 4 year than in
the 45-54 year age group. As a result, and in
contrast to all other age and sex categories, there
was no gender difference in attack rate or incidence 0
0
of AMl in the 3 5 4 4 year old diabetic population. 0

The crude AMI attack rate (fist and recurrent 5

events) was higher in diabetic men compared with
non-diabetic men (RR 3.7 : 9 5 % CI 3 . 3 4 . 1 ) and in
diabetic women compared to non-diabetic women 9 G
(RR 6.1; 95% CI 5.0-7.5) (Table 3). Middle-aged 8
In
diabetic individuals had an incidence (hst ever event) m
of AMI that in men was 3.0 times higher and in
women 5.0 times higher than in non-diabetic sub-
jects. The rate of recurrent infarctions in diabetic men
was 5.5 times higher than in non-diabetic men, and
in diabetic women 9.3 t i e s higher than in non-
diabetic women.
In calculations of the risk for a recurrent AMI in
diabetic vs. non-diabetic patients, adjustment for
early case fatality (before 28 days) after the first AMI 3*
Y

was made in order to include only those at risk for a

recurrent event. The risk for a recurrence was
doubled in diabetic subjects not only in absolute -a
0

terms (data not shown) but also in relative terms. 2

V
Thus, the ratio between recurrent and Erst AM1 rates
was 1.11 in diabetic men vs. 0.54 in nondiabetic
0 1997 Blackwell Science Ltd Journal of Intern! Medicine 241 :485-492
 DIABETES AND AM1 489

men, and 1.02- in diabetic women vs. 0.47 in non-

diabetic women.
In the youngest age-group there was a marked
difference in the age at onset of diabetes between
men and women. Among the 3544-year-old
patients with a first Ah4I, the mean age at onset of
diabetes was 24.7 years (95% CI 12.2-37.1) in men
and 11.8 years (95% C15.5-18.0) in women. In
45-54-year-old patients, the mean age at onset of
diabetes was 42.3 years (95% CI 39.0-45.7) in men
and 45.8 years (95% (338.1-53.5) in women,
whereas in the oldest age-group it was 51.2 years
(95% (349.6-52.0) in men and 48.8 years (95%
CI 45.6-52.0) in women.
Table 4 shows the 28day case fatality (out of
hospital and in hospital deaths together) in all and
first AMIs in different age groups of men and women.
The overall 28day case fatality was Significantly
higher in diabetic patients with AMI compared with
non-diabetic patients (men P = 0.0002, women
P = 0.02 by Mantel Haenszel chi-square statistics).
The 28day case fatality after a first AMI was
significantly higher in diabetic men than in non-
diabetic men (31 vs. 23%; P = O . O 3 ) , but the
difference did not reach statistical significance in
women (36 vs. 25%; P = 0.09). The case fatality
after a first AML in men was higher in IDDM than in
NIDDM (41 vs. 27%), in women it was the opposite
(13 vs. 36%) but due to the low numbers of IDDM
patients, none of the differences reached statistical
significance. The case fatality in recurrent A M I s was
consistently higher than in first events and differed
little between diabetic and nondiabetic patients (data
not shown).
One year case fatality after a first Ah4I was
significantly higher in diabetic men compared to
nondiabetic men (3 7 vs. 26%; P = 0.0009), but not
in women (37 vs. 29%; P = 0.23).
In the 35-64 year age range, total AMI mortality
rates were 4.3 times higher in diabetic as in non-
diabetic men (604 vs. 140 per 100000 per year),
and 7.3 times higher (262 vs. 36 per 100000) in
diabetic versus non-diabetic women.

Characteristics of diabetic patients with AM1

Overall, there were only small differences between
men and women in mean age of onset of diabetes
(45.8 vs. 44.9 years). When the myocardial in-
farction occurred, 41% of men (149/363) were
0 1997 Blackwell Science Ltd Journal of Internal Medicine 241: 4 8 5 4 9 2
490 V. LUNDBERG et al.
being treated with insulin, 29%(106/363) with oral
hypoglycaemic agents, 27%(98/363) with diet only
and in 3% (10/363) the treatment was unknown.
The corresponding proportions for women were 5 5 %
(69/125), 20% (25/125), 20% (25/125) and 5%
(6/125) and there was no patient with unknown
treatment.
Population attributable risk
In the 3 5-64 year age range, population attributable
risk for diabetes in AMI was calculated to be 11 % in
men and 17% in women.
A sensitivity analysis was performed based on the
upper and lower limits of the confidence intervals of
the relative risks which, in turn,was calculated from
the prevalence of diabetes in the entire population. In
these sensitivity estimates, the population attribu-
table risk ranged from 7.6 to 14.2% in men and from
10.2 to 25.0% in women. The population attribu-
table risk calculated for men and women together
was 12 % in the 3 5-64 year age group.
Discussion
With nearly 500 AMI events in diabetic patients and
more than 2500 in non-diabetic subjects, this is the
first population-based study in which it has been
possible to estimate the incidence of AMI in a diabetic
compared to a non-diabetic population with reason-
able accuracy. Middle-aged diabetic patients were
found to have an incidence of AMI that, in men, was
three times higher and, in women, five times higher
than in non-diabetic subjects. Diabetic patients had
a higher 28 day case fatality than non-diabetic
patients. This agrees with previous observationsfrom
Sweden [ll], Finland [S] and a report from the
Minnesota Heart Survey, in which a 40%higher
mortality in diabetic AMI patients compared with
nondiabetic AMI patients was observed during a 6
year follow-up [121.
The total prevalence of diabetes was based on the
presence of previously known diabetes and newly
detected diabetes diagnosed by an oral glucose
tolerance test. Since diabetic subjects have regular
medical check-ups, they may be less likely to par-
ticipate in population screenings. This would lead to
an underestimation of the prevalence of diabetes in
the population. However, the non-response rate in
the present surveys was relatively lo& (19 %). Other
0 1997 Blackwell Science Ltd Journal oJInternal Medicine 241: 4 8 5 4 9 2
Swedish population studies have found about the
same prevalence of diabetes as we found in com-
parable age groups [ll,131.
Whereas the prevalence of diabetes in the popu-
lation was based on known diabetes and diabetes
detected by a glucose tolerance test, the presence of
diabetes in AMI patients was based only on diabetes
that was previously known or newly detected under
routine clinical conditions when the patient was
treated for an AMI. If a glucose tolerance test had
been performed in the AMI patients, the proportion
of diabetic AMI patients would probably have been
higher, resulting in even higher relative risk for AMI
in diabetic patients.
Confidence intervals for diabetic group in Table 3
are based on the assumption that the total number of
diabetics in the population is known. In our study
the total number of diabetic men and women has to
be estimated, and thereby the confidenceintervals for
diabetic group in Table 3 are not strictly valid.
In this paper, emphasis is on crude attack and
incidence rates of AMI and not on age-standardised
rates. Crude age-specific rates reflect more accurately
the impact of diabetes on AMI than age-standardised
rates do. When age-standardisation was applied, the
relative risks for AMI conferred by diabetes were
somewhat lower than the age-specific relative risks,
reflecting the age distribution of diabetes in the
reference population.
When the diabetic and non-diabetic populations
were compared, the AMI mortality was more than
four times greater in male and more than seven times
greater in female diabetic people. Three components
contributed to this: a higher incidence of firstever
AMIs, a higher incidence of recurrent A M I s and a
higher case fat&@ rate. The greater relative increase
in mortality from AMI in female compared with male
diabetic patients was because the relative increase in
both incidence (first AMI)and recurrent AMI rates
were higher in women. On the other hand, the
excess case fatality among diabetics was of the same
magnitude in men and women. The result of the
excess risk conferred by diabetes being even more
pronounced in women than in men is that diabetes
tends to even out some of the gender differences in
AMI [14, 151.
A particularly large impact of diabetes was noted
on the rate of recurrent AMIS. The ratio between the
diabetic and the non-diabetic populations was 5.5-
fold in men and 9.3-fold in women. Both diabetic

men and diabetic women surviving their first in-
farction have about twice as large risk for re-
infarction as their non-diabetic counterparts. This
would imply that secondary prevention measures
after a first AM1 are applied less rigorously in diabetic
than in non-diabetic patients, or they are less effective
in diabetic subjects [16].
The age distribution of diabetic women with AMI
differed from that in non-diabetic women and in all
male categories. Thus, the attack rate, the incidence
and the rate of recurrent AMIs were all higher in the
3 5 4 4 year than in the 45-54 year age group. In the
3 5 4 4 year diabetic population there were no gender
differences in the risk of AMI. Most of the women
who suffered an AMI at age 3 5 4 4 had an early
onset of diabetes and a long duration of the disease.
Mean age at onset of diabetes for the youngest age
group encountering their fist AMI was 11.8 years in
women, 24.7 years in men. This implies that women
with diabetes have greater risks to encounter AMI in
younger ages compared to men. In the older age
groups there were no big gender differences in the
age at onset of diabetes for first AMIs.
The lower attack and incidence rates in women in
the age group, 45-54 years, may be due to the fact
that a large proportion of non-insulin dependent
diabetic patients with relatively short diabetes dur-
ation have been added. As diabetic women pass into
the next decade of life (55-64 years), there is a very
dramatic increase in the incidence of AMI. It may be
speculated that the interaction of diabetes and
postmenopausal loss of oestrogens confers particu-
larly high risk for AMI [15, 171.
The overall population attributable risk of AMI
associated with diabetes was calculated to be about
12% for the ages covered by the present study.
Attributable risk is a measure that is based on
relative risk and proportion of the population ex-
posed. It should be regarded as an estimate of the
proportion of all AMI events that would be avoided if
diabetes did not exist in the community. This is a
crude measure that assumes a direct causal re-
lationship between diabetes and AMI,and it does not
take into account interactions with other risk factors.
Also, if the prevalence of diabetes in the general
population is underestimated, the relative risk of AMI
in diabetic vs. non-diabetic subjects would be in-
creased and the population attributable risk over-
estimated. However, in sensitivity analyses, the
population attributable risk was only modestly re-
0 1997 Blackwell Science Ltd Journal ojInterna2 Medicine 241: 4 8 5 4 9 2
DIABETES AND AM1 491
duced if the upper 9 5 % confidence limit of relative
risk of diabetes in the general population was taken
into consideration. In absolute numbers, around 40
AMI events per 100000 inhabitants would be
averted annually in the age group 35-64 years, if
diabetes did not exist in the population. It should be
noted that, in absolute numbers, diabetes is likely to
cause considerably more AMIs in higher age groups,
not covered by the present study.
Our results emphasise the importance of diabetes
as a risk factor for AMI from a population perspective
and that preventive strategies to reduce this impact
are essential. It is possible to reduce the risk of AMI
by improved metabolic control of diabetic patients
with IDDM [16, 18, 191. There is also room for other
strategies to prevent AMI in diabetic patients by the
use of interventions with documented effects, such as
the use of antiplatelet drugs [20], smoking cessation,
physical exercise, improved control of hypertension
and detection and treatment of dyslipidaemia
[16,21,22]. In view of the high rate of recurrent
AMIs observed in the present study, the effects of
various secondary prevention measures after AMI
should be explored in more detail in diabetic subjects
[16, 191.
Acknowledgements
The study was supported by grants from Norrbotten
and Vasterbotten County Councils, The Swedish
Medical Research Council (grant 27X-07192 to KA),
the Swedish Public Health Institute, the Joint Com-
mittee of the Northern Sweden Health Care Region,
and Swedish Tobacco Company, and the Heart and
Chest Foundation.
References
1 Kannel WB,McGee DL. Diabetes and cardiovascular disease.
The Framingham Study. JAMA 1979: 241: 2035-8.
Garcia MJ, McNamara PM. Gordon T, Kannel WB. Morbidity
and mortality in diabetics in the Pramingham population.
Diabetes 1974: 23: 105-11.
Barrett-Connor E. Wingard DL. Sex differential in ischemic
heart disease mortality in diabetics : a prospective population-
based study. Am J Epidemiol 1983: 118: 489-96.
Scheidt-Nave C. BarrettConnor E, Wingard D. Resting elec-
trocardiographic abnormalities suggestive of asymptomatic
ischemic heart disease associatedwith non-insulindependent
diabetes rnellitus in a defined population. Circulation 1990:
81 : 899-906.
492 V. LUNDBERG et al.
5 Lehto S. Pyorkila K, Mettinen H, Ronnemaa T, Palomiiki P,
TuomilehtoJ. Myocardial infarct size and mortality in patients
with non-insulindependent diabetes mellitus. J Intern Med
1994; 236: 291-7.
6 Tunstall-Pedoe H. WHO MONICA Project Principal Inves-
tigators. The World Health Organization MONICA Project:
Monitoring of Trends and Determinants in Cardiovascular
Diseases. Clin Epidernioll988; 41: 105-14.
7 WHO Study Group. Diabetes Mellitus. World Health Organ-
ization Tech Report Series 1985; 727: 10-13.
8 WHO MONICA Project (prepared by TunstaU PH, Kuulasmaa
K, Amouyel P. Arveiler D. Rajakangas AM, Pajak A).
Myocardial infarction and coronary deaths in the World
Health Organization MONICA Project. Registration pro-
cedures, event rates, and case-fatality rates in 38 populations
from21 muntrIes in four continents. Circulation 1994; 90:
583-612.
9 Dean K. Self-care components of lifestyles : The importance of
gender, attitudes and the social situation. Soc Sci Med 1989;
29: 137-52.
10 Armitage P. Berry G. Statistical Methods in Medical Research.
2nd edn. Oxford: Blackwell Scientific Publications. 1987.
11 Lundman B. Utvtkdering av vikdprogmm fdr diabetes i
Medelpad. In: Prirndwcirdens utvecklingsenbet i Medelpad.
Sundsvall, 1993.
12 Spratka JM,Folsom AR, Burke GL, McGovem PG. Hahn LP.
Trends in prevalence of diabetes mellitus in patients with
myocardial infarction and effect of diabetes on survival. The
Minnesota Heart Survey. Diabetes Care 1991; 14: 5 3 7 4 3 .
13 Andersson DKG, Sviirdsudd K, Tibblin G. Prevalence and
incidence of diabetes in a Swedish community 1972-87.
Diabetic Med 1991; 8: 428-34.
14 Kleinman JC. Donahue RP, Harris MI, Piucane IT,Madans
JH, Brock DB. Mortality among diabeticsin a national sample.
Am J Epiderniol 1988; 128: 389-401.
Q 1997 Blackwell Science Ltd Journal oflnternal Medicine 241: 485-492
15 Legato M.Cardiovascular disease in women: what's Merent?
What's new? What's unresolved? Ann NY Acad Sci 1994;
736: 147-57.
16 Kjekshus J, Pedersen TR. Reducing the risk of coronary
events: evidence from the Scandinavian Sivastatin Survival
study (4s). Am J Curdid 1995; 76: 6468C.
17 Beard CM,Kottke T& Annegers JF,Ballard DJ. The Rochester
Coronary Heart, Disease Project: eflect of cigarette smoking,
hypertension, diabetes, and steroidalestrogen use on coronary
heart disease among 40 to 59-year-old women 1960 through
1982. Mayo Clin Proc 1989; 64: 1471-80.
18 The Diabetes Control and Complications Trial Research Group
(1993). The dect of intensive treatment of diabetes on the
development and progression of long-term complications in
insulindependent diabetes mellitus. N EnglJ Med 1993 ; 329:
977-86.
19 Malmberg K, Ryden L, Efendic S. Herlilz J, Nicol P.
Waldenstrijm A et d. Randomized trial of insulin-glum
infusion followed by subcutaneous insulin treatment in
diabetic patients with acute myocardial infarction (DIGAMI
study): effects on mortality at 1 year. J Am toll cardioll995;
26: 57-65.
20 Yudkin J. Which diabetic patients should be taking aspirin?
BMJ 1995; 311: 641-2.
2 1 ShepherdJ, Cobbe SM,Ford I, Isles CG,Lorimer AR, Madarlane
PW, McKillop JH,Packard CJ. Prevention of coronary heart
disease with prevastatin in men with hypercholesterolemia.
N E q I J M e d 1995; 333: 1301-7.
22 Gould AL, Rossouw JE,Santanello NC, Heyse JF, Furberg CD.
Cholesterol reduction yields clinical beneEt. A new look at old
data. Circulntion 1995; 91: 2274-82.
Received 21 October 1996; accepted 30 December 1996.
Correspondence:V i v a Lundberg, MONICA-Project, Kalix Hospital,
$952 82 Kalix. Sweden (fax: +46-923-154 96).