EHCB 4(1) - Cholesterol and coronary heart disease

FEBRUARY 1998 VOLUME 4 NUMBER 1 ISSN: 0965-0288
Effective
Health Care
Bulletin on
the effectiveness
Cholesterol and coronary
of health service
heart disease: screening
interventions for
decision makers
and treatment
NHS Centre for Reviews
s Despite the declining
and morbidity. Therapy
coronary heart disease
should be targeted at
and Dissemination,
(CHD) mortality rate, CHD
people who are at high risk
remains a major cause of
of coronary heart disease
University of York
premature death and
rather than be based upon
imposes high personal,
cholesterol levels. In
social and economic costs.
asymptomatic people, at
s
low risk of coronary heart
Blood cholesterol is an
disease, the costs of
important risk factor for
cholesterol lowering using
CHD but should be
statins are high relative to
considered in the context
the benefits and their use
of other risk factors such as
is contentious.
smoking, raised blood
pressure and physical
s Cholesterol lowering is one
inactivity.
of a number of methods of
s
reducing the risk of
Blood cholesterol alone is a
cardiovascular disease. The
relatively poor predictor of
cost-effectiveness of some
individual CHD risk. The
anti-hypertensives, aspirin
majority of CHD events
and beta-blockers is greater
occur in people with
than statins.
average or low blood
cholesterol levels.
s Greater priority should be
Consequently, cholesterol
given to the appropriate
screening is unlikely to
use of other drug
reduce mortality and can
treatments and non-
be misleading or even
pharmacological
harmful.
interventions in the
s
primary and secondary
Cholesterol lowering using
prevention of coronary
statins is effective at
heart disease.
reducing CHD mortality
The contents of this bulletin are likely to be valid for around one year, by which time significant new research evidence may have become available.

---

A. Coronary
taken into account when making
of identifying those at highest risk
projections of the population
and effective interventions once
heart disease
benefits flowing from
they have been identified.
interventions to reduce CHD such
A.1 The importance of CHD:
as cholesterol lowering. Declining
In contrast, a population approach
CHD is a major cause of morbidity
CHD mortality rates are only
focuses more on trying to reduce
and mortality in the UK,
partly explained by reductions in
levels of risk factors in the
accounting for just under one
established cardiovascular risk
population as a whole. The logic
quarter of all deaths in 1995: 27%
factors3, 4, 5 and it is probable that
here is that even though CHD risk
among men and 21% among
general social and economic
for any individual may be lowered
women.1 While many CHD deaths
improvement over time has
by only a small amount, the
occur among elderly people, CHD
contributed to this trend.6
population effect could be
accounts for 31% of male and 13%
However, it is noteworthy that
substantial because so many
of female deaths within the 45–64
these benefits have not been
people are affected. Furthermore, a
age group.
observed in the lowest socio-
large percentage of events occur in
economic groups.7
people who are at only average
CHD leads to obstruction of blood
risk and who would otherwise be
flow through the coronary arteries
Because of the importance of CHD,
missed by approaches targeted at
to the heart muscle, due to
considerable effort has been made
those at high risk. Public health
atherosclerosis (fibro-fatty
to identify the major risk factors
policy is based on a combination
deposits) and associated blood
associated with the disease and to
of population and targeted
clots. This can lead to sudden
modify them by drugs, lifestyle
approaches.
death, heart attack (myocardial
and environmental change in
infarction) which may be fatal,
order to prevent CHD occuring
One CHD risk factor is serum
angina or heart failure.
(primary prevention) or preventing
cholesterol. Much attention has
death or (further) coronary events
been focused on screening people
CHD imposes high social costs,
in people with established disease
to identify those with raised
including impaired quality of life
(secondary prevention). One
cholesterol levels and then trying
and reduced economic activity. A
approach to disease prevention
to lower these levels through diet
large share of NHS resources are
requires identification of people at
and/or medical treatment. This
also accounted for by CHD.2
high risk of CHD and the
topic was covered in a previous
However, CHD rates have been
subsequent application of
issue of Effective Health Care.8
declining in the UK for almost 20
interventions which will reduce
However, since then a new class of
years (Fig. 1) and this needs to be
their risk factors. Such targeted
cholesterol lowering drugs – the
strategies require efficient means
statins – has been developed and
evaluated.
Percent of 1980 rate
100
The expenditure on statin drugs
*
was over £20 million in 1993 and
*
*
90
*
Men
*
*
by 1997 had risen to over £113
*
Women
million (Fig. 2). Expenditure on
80
*
other lipid-lowering drugs is still
*
increasing but at a much slower
70
rate and in 1997 amounted to £21
*
*
million, resulting in a total of £134
60
million.
*
50
This bulletin considers whether
cholesterol screening is worthwhile
40
and examines the effectiveness and
cost-effectiveness of the statins and
30
a range of other interventions to
reduce CHD. It aims to provide a
20
summary of the research evidence
which can be used to establish
10
cost-effective policies for reducing
CHD.
0
1980
81
82
83
84
85
86
87
88
89
90
91
92
93
94
1995
A.2 Cholesterol and other risk
Year
factors: Cholesterol is a fatty
substance which is manufactured
Fig. 1 CHD death rate as a percentage of 1980 rate among men and women aged
in the body – particularly in the
55–64 years, England & Wales, 1980–1995
2 EFFECTIVE HEALTH CARE Cholesterol and CHD: screening and treatment
FEBRUARY 1998

---

Net ingredient cost (£ millions)
and alcoholism, among others.9, 10, 11
140
Thus it is the pre-existing disease
which causes both the low
Other Drugs
Statins
120
cholesterol and raised mortality
and not the low cholesterol levels
100
themselves which produce the
elevated mortality rates.12
80
Differences in average levels of
60
blood cholesterol between
communities or populations are
40
largely determined by differences
in diet. Countries with high
20
dietary saturated fat intake and a
low ratio of polyunsaturated to
0
saturated fatty acids have high
1991
1992
1993
1994
1995
1996
1997
average cholesterol levels.13
Randomised controlled trials
Fig. 2 Prescribing trends for cholesterol lowering drugs in England, 1993–1997
(RCTs) in institutional settings
Source: Department of Health, Statistics Division 1E, Prescription Cost Analysis system
demonstrate that if components of
liver – and plays a vital role in the
calculations. The higher the ratio
the diets of individuals are
functioning of cell membranes.
the higher the CHD risk.
changed substantially then large
Cholesterol is found in several
changes in blood cholesterol levels
forms in the body and, when
The average level of blood
can be achieved.14
bound to proteins, forms
cholesterol within a population is
lipoproteins. Cholesterol and other
an important determinant of the
Although blood cholesterol is an
fatty blood components are often
CHD risk of the population. In
important risk factor, by itself it is a
referred to collectively as ‘blood
countries where the average
relatively poor predictor of who will
lipids’.
cholesterol levels of the population
go on to have a CHD event. Fig. 4
are low, CHD tends to be
shows the relationship between
Blood lipids can be divided into
uncommon. Prospective studies
blood cholesterol and CHD rates in
different fractions or components:
show that groups of individuals
British men; only 42% of those who
low density lipoprotein (LDL) and
with lower levels of cholesterol
will suffer an event over 15-years
high density lipoprotein (HDL)
run less risk of developing CHD.
have blood cholesterol over 6.5
cholesterol and triglycerides. High
The association between
mmol/l. This is further illustrated in
levels of LDL and low levels of
cholesterol level and future risk of
Fig. 5 which shows that the
HDL are associated with increased
CHD is graded and continuous:
distribution of blood cholesterol in
risk of CHD. The ratio of plasma
there is no threshold above which
British men aged 40–59 who
total (or LDL) cholesterol to HDL
CHD risk begins to increase (Fig. 3).
subsequently went on to suffer
cholesterol is often used in risk
from CHD and in those who did
There has been some concern that
not, overlap considerably.
Age-adjusted 6-year death rate per 1000 men
low levels of blood cholesterol
40
increase the risk of mortality from
Other major independent risk
Total mortality
causes other than CHD, including
factors (e.g. smoking, high blood
35
cancer, respiratory disease, liver
pressure, diabetes, physical
disease and accidental/violent
inactivity, and obesity) also exist
30
death. Fig. 3 shows a U-shaped
and should be considered in
25
curve in which men with the
defining individual risk of CHD.
lowest cholesterol levels have
Fig. 6 shows the importance of
20
higher rates of total mortality than
considering risk factors together.
CHD mortality
men with higher (but still well
Smokers with high blood pressure
15
below average) levels. Several
have three times the risk of dying
studies have now demonstrated
of CHD compared to non-smokers
10
that this phenomenon is mostly, or
with low blood pressure where
5
entirely, due to the fact that this
both have the same level of blood
group of people with low
cholesterol. Risk scoring systems
0
cholesterol levels includes a
developed from the British
3.6 4.1 4.7 5.2 5.7 6.2 6.7 7.3 7.8 8.3
disproportionate number whose
Regional Heart Study were no
Serum cholesterol (mmol/l)
cholesterol has been reduced by
more accurate in predicting who
illness – early cancer, respiratory
suffered from coronary heart
Fig. 3 Age-adjusted 6-year CHD and
disease, gastrointestinal disease
disease with blood cholesterol
total mortality per 1,000 men screened for
MRFIT according to serum cholesterol
FEBRUARY 1998
Cholesterol and CHD: screening and treatment EFFECTIVE HEALTH CARE 3

---

increased by using good
20%
equipment and repeat analyses. A
National Initiative on Cholesterol
Accuracy, Methods and
Standardisation has been launched
which aims to improve the
15%
standardisation of cholesterol
measurement.
The increasing use of compact
measuring devices such as desk
10%
top analysers in GP surgeries and
their spread to high street chemist
and health food stores, is of
potential concern. They are less
5%
accurate,20, 21, 22 making it difficult to
distinguish confidently between
people with raised and normal
Percentage of CHD events occuring over 15 years
cholesterol levels,23 and are less
amenable to national initiatives for
0
quality assurance. Studies of the
<4.5
5.0–
6.0–
7.0–
8.0–
use of such analysers in general
Serum cholesterol (mmol/l) at baseline
practice suggest that quality
control is a major problem due to
Fig. 4 Blood cholesterol distribution and percentage of CHD events occuring at each
lack of time, poor technique, and
level over 15 years follow up Source: British Regional Heart Study
the use of outdated test strips.24,25
included than without,
not accurately reflect the true
highlighting the importance of
cholesterol level due to
Availability of analysers was
these other major risk factors.15
measurement error (bias and
associated with a three-fold
imprecision) and natural biological
increase in cholesterol estimation,
variation in cholesterol levels
although the value of this extra
within an individual. These
information was not assessed.26
B. Detecting
sources of error can result in
misclassification and lead to
Even when evaluated in optimal
raised
incorrect diagnosis and the
conditions the performance of
possibility of unnecessary
some machines has been
cholesterol
treatment.
inadequate,27 although more
recent disposable devices have
Cholesterol screening may be done
B.1 Measurement error:
achieved reasonable accuracy and
either by testing the entire adult
Measurement error can be the
precision.28 Home cholesterol
population or by making use of
result of bias (the degree to which
testing kits using such disposable
routine contacts in primary health
a reading systematically differs
devices, which have not been
care (opportunistic screening). Both
from a gold standard or reference
evaluated under the circumstances
these methods are essentially
value) or imprecision (where
for which they are marketed, are
untargeted, giving equal priority to
measurements are subject to
unlikely to perform well.29
both high and low CHD risk
random measurement error).
individuals. In a recent survey two-
B.2 Biological variability: In any
thirds of British general
There is considerable evidence
individual the blood cholesterol
practitioners said they offered
that different laboratory analysers
concentration is not constant over
some form of cholesterol testing.16
can give different readings for the
time. This random biological
In England, the Health Survey
same blood sample.18 For example,
variation is quite large and results
shows that 28% of people aged
a UK study found that laboratory
in considerable misclassification.
45–64 have had their blood
equipment systematically
Estimates of within-person
cholesterol measured in the last 3
overestimated cholesterol levels by
variation show a coefficient of
years.17
over 4% at the cut off of 7.8
variation for measurements made
mmol/l.19 This would result in a
one-year apart of 7% which is
The main screening test for blood
50% increase in the number of
large compared with the between-
cholesterol is the measurement of
people tested subsequently being
person coefficient of variation of
total blood cholesterol in blood
recommended for treatment. Bias
15%. In British men, the
samples obtained by either
can be reduced in laboratory
implication of this biological
venepuncture or finger prick.
equipment by regular calibration
variation is that 28% of men
Cholesterol measurements may
against a standard, and precision
classified as having a raised blood
4 EFFECTIVE HEALTH CARE Cholesterol and CHD: screening and treatment
FEBRUARY 1998

---

B.4 Other effects of screening:
Screening, either by mass
%
approaches or opportunistically, is
25
never entirely without the risk of
harm. Knowledge of the presence
of a risk factor may result in
20
CHD
people who previously felt well
Non CHD
behaving as if they were sick
(adopting a sick role) with adverse
15
consequences for the individual
and society. It is also possible that
knowledge of the absence of a risk
10
factor may result in adverse life-
Percentage of men
style choices.
5
It has been shown that classifying
people as ‘suffering with
hypertension’ is associated with
0
<3.5
4-
5-
6-
7-
8-
9-
10-
increased sickness absence and
Serum cholesterol (mmol/l) at baseline
adoption of the sick role – a
labelling phenomenon.35 The
negative effects of labelling people
Fig. 5 Distributions of blood cholesterol among British men who did and did not
develop coronary heart disease over 15 years follow up Source: British Regional Heart Study
as hypertensive have not been
found in all studies and it has been
cholesterol on a single testing will
the USA and Britain show that
suggested that professional
have a normal long-term blood
untargeted general population
support, individualised care and
cholesterol.30 In order to reduce
screening coupled with dietary
follow up, and attention to
misclassification several (at least
advice have little effect on
compliance-improving strategies,
two) measurements should be
cholesterol levels.31, 32, 33
may overcome adverse labelling
made separated by a few weeks,
effects.36 Only limited evidence is
and clinical decisions should be
An Australian study demonstrated
available to determine the
based upon the average of several
that the majority (61%) of people
potential influence of blood
readings rather than a single
who had their blood cholesterol
cholesterol screening on labelling
measurement.
tested by case-finding were
with case studies showing similar
unwilling to make dietary changes
effects to those seen in
B.3 Effect of screening on
to their fat intakes on the grounds
hypertension.37 However, a trial
cholesterol levels: Early
that their cholesterol levels were
and a before–after study failed to
enthusiasm in the United States
all right. This suggests that
demonstrate any adverse effects.38, 39
for a patient-centred approach –
screening may interfere with
the ‘know your number’ campaign
general public health strategies to
– resulted in many people being
reduce the dietary fat intake of the
screened and given dietary advice.
whole population.34
However, evidence from RCT’s in
(A) Non smokers
(B) Smokers
9.5
10.6
12
12
6.1
6.1
5.5
10
5.2
10
8.2
3.7
5.7
5.6
8
8
4.1
4.0
6
DBP
6
DBP
2.4
tality rate/1000 person-yrs
4
High
3.4
3.4
High
3.1
tality rate/1000 person-yrs
4
1.7
2.0
Average
Average
2
2
Low
Low
CHD Mor
0
CHD Mor
0
Low
Average
High
Low
Average
High
Serum cholesterol
Serum cholesterol
Fig. 6 CHD mortality among British men according to their levels of blood cholesterol, diastolic blood pressure and smoking behaviour
FEBRUARY 1998
Cholesterol and CHD: screening and treatment EFFECTIVE HEALTH CARE 5

---

C. Cholesterol
diets or lived in institutions where
However, all these dietary trials
control over diet was much
were of relatively short duration
lowering
greater. However, despite the
and did not consider clinical
greater fall in blood cholesterol,
endpoints. Therefore there is no
interventions
the meta-analysis failed to find any
evidence that they lower CHD risk.
significant CHD mortality risk
Cholesterol levels can be lowered
reduction (RR = 0.94; 95% CI: 0.84
Drugs
by several types of interventions,
– 1.06).44
diet and drugs being the most
C.3 The statins: Over the last few
important.
The generally poor performance of
years a new class of more powerful
some lipid lowering diets may be
cholesterol lowering drugs – the
Diet
partly explained by the fact that
statins (HMG CoA reductase
they often substitute complex
inhibitors) – has become available
C.1 Low fat diets: Changes in
carbohydrates for total fat
which is able to reduce LDL
individual dietary intake of
resulting in a reduction in both
cholesterol levels by more than
saturated fats and cholesterol have
HDL as well as LDL cholesterol.43
20%.
been studied extensively (Table 1).
This reduces total cholesterol but
The effectiveness of low fat diets
leaves the more important
A total of 22 published RCTs of
depend critically on how
LDL/HDL ratio unaffected and so
cholesterol lowering in which
restrictive they are and the degree
does not reduce CHD risk.14 This
clinical outcomes were recorded
of adherence. In settings where
highlights the fact that the real
were identified and their results
patients’ diets are controlled by
aim should be to lower CHD risk
pooled to give an overall estimate
others such as in metabolic wards
rather than focusing on lowering
of treatment effect. Overall, these
where adherence to diets is likely
serum cholesterol levels per se,
trials show that statins reduce the
to be high, dietary changes can be
which is relatively ineffective.
risk of CHD mortality by around
expected to produce substantial
25% (see Table 2). The trials which
reductions in blood cholesterol,
C.2 Garlic, oats and soy protein:
contributed most to the pooled
though no clinical event data have
A systematic review of trials
estimates were the West of
been reported.14 However, studies
suggested that garlic may exert a
Scotland Coronary Prevention
in the general population have
cholesterol lowering effect with
Study (WOSCOPS),49 the
shown only small changes in
falls of 0.65 mmol/l (95% CI: 0.53
Scandanavian Simvastatin Survival
cholesterol.40 These studies suggest
– 0.76) or around 10%.45 However,
Study (4S),50 the Cholesterol and
that the extent of cholesterol
some of the trials are severely
Recurrent Events (CARE) trial,51
reduction which may be expected
flawed and, therefore, the evidence
and the recently reported Long-
from recommending lipid lowering
is not reliable. Systematic reviews
term Intervention with Pravastatin
diets is likely to be very small
of studies evaluating the effects of
in Ischaemic Disease (LIPID) trial.52
(1–5%), and the effect on clinical
consuming oats46 or psyllium-
events has been shown to be
enriched cereals47 show a small
C4. Statins compared to other
disappointing (OR = 0.96; 95% CI:
cholesterol lowering effect of
cholesterol lowering drugs: The
0.89 – 1.04).41
around 2–5% respectively. A meta-
efficacy (relative risk) of statins in
analysis of 38 trials of soy protein
primary and secondary prevention
The effects of dietary interventions
as a substitute for meat protein
is summarised for a range of
used alone following myocardial
also demonstrated a net fall in
endpoints in Table 3. For
infarction demonstrated a greater
cholesterol of 0.60 mmol/l (95%
comparative purposes similar
fall in blood cholesterol than the
CI: 0.35 - 0.85), which was greater
information for fibrates (clofibrate
other dietary trials,44 probably
in people with high baseline
and bezafibrate) is also given.
because the participants were
cholesterol levels.48
more motivated to follow strict
Older drugs (e.g. fibrates) are not
as effective as the newer statins in
Table 1
The effect of lipid lowering diets in reducing blood cholesterol levels
lowering blood cholesterol and in
reducing CHD event rates. The
Blood cholesterol reduction (percent)
overall efficacy of older cholesterol
lowering drugs is strongly related
Multiple risk factor intervention trials41
0.14mmol/l
(2%)
to the baseline level of coronary
heart disease risk. In high risk
Dietary interventions
populations (>3% annual CHD
(i) general population
death rate), treatment benefits
outweigh treatment risk, whereas
Brunner42
0.22mmol/l
(3%)
in lower risk populations there is
(ii) including high risk
no place for these older drugs
Ebrahim & Davey Smith44
0.65mmol/l
(9%)
which may do harm.53
6 EFFECTIVE HEALTH CARE Cholesterol and CHD: screening and treatment
FEBRUARY 1998

---

Table 2
Summary of major trials of statins
Trial
CHD
Patient group
Treatment
Follow
Sex
Number
Base-line
Total/CHD
death
up (yr.)
(mean age)
T/C
CHO
mortality
rate*
odds ratio (95% CI)
WOSCOPS
3.8
No CHD evidence
Pravastatin 40mg
4.9
Men only
3302 vs 3293
7.03
0.78 (0.60–1.00)
(1995)49
cho: 6.5+mmol/l
vs placebo
(55)
0.67 (0.45–0.99)
4S trial
15.7
Post MI or angina,
Simvastatin 20–
5.4
Men 81%
2221 vs 2223
6.74
0.69 (0.56–0.84)
(1994)50
cho: 5.5–
40mg vs placebo
(60)
8.0mmol/l
0.75 (0.64–0.88)
CARE
11.5
Post MI 3–20 mths
Pravastatin 40mg
5.0
Men 86%
2081 vs 2078
5.40
0.91 (0.74–1.12)
(1996)51
cho: <6.2mmol/l
vs placebo
(59)
0.80 (0.61–1.05)
LIPID
Pravastatin 40mg
Men 83%
4512 vs 4502
5.60
0.76 (0.67–0.86)
13.8
Post MI/ unstable
6.0
(1997)52
angina 3– 36 mths,
vs placebo
(31–75)
cho:
0.75 (0.64–0.88)
4.0–7.0mmol/l
* Control group CHD mortality rate per 1000 patient years; CHO = cholesterol level T = Treatment group C = control group
Pravastatin and simvastatin appear
limited data.55 A meta-analysis for
among younger participants.
to be equally effective in reducing
this bulletin of the recently
Pooling of these sub-group
CHD event rates. However, less
published data on women from
analyses from the major statin
data from large scale trials are
the 4S,56 LIPID study (preliminary
trials (CARE, 4S, WOSCOPS, pooled
currently available for fluvastatin,
data),52 CARE study51 and pooled
pravastatin trials) carried out for
atorvastatin and cerivastatin and
data from several pravastatin
this bulletin, demonstrates a
consequently their clinical efficacy
trials57 shows that if both fatal and
relative risk of combined fatal and
is not yet proven, although they
non-fatal coronary heart disease
non-fatal CHD events of 0.70 (95%
lower LDL cholesterol to an extent
events are considered, women
CI: 0.62 – 0.79) for older people.
similar to or greater than other
have an on-treatment relative risk
People in their late 70s and 80s,
statins.
of 0.77 (95% CI: 0.64 – 0.92), which
while obviously at increased
is similar to men (no significant
absolute risk of coronary heart
Further important trial results are
interaction effect for gender P =
disease, have not been studied in
awaited which appear likely to
0.46). A report of an increased risk
the recent statin trials. Treating
extend the range of indications for
of breast cancer among treated
people in this age group with
use of statins. The next trial to
women in the CARE study was not
statins must, therefore, remain a
report will be the Air Force/Texas
confirmed in the 4S or the LIPID
matter of clinical judgement until
Coronary Atherosclerosis
studies. The pooled results from
the Anti-hypertensive, Lipid
Prevention Study of lovastatin in
the three studies shows no
Lowering after Heart Attack Trial
6605 people (15% women) with no
association with breast cancer (RR
(ALLHAT), which is examining the
evidence of coronary heart disease
= 1.0; 95% CI: 0.44 – 2.24).
efficacy of statin treatment in older
and with average blood cholesterol
people, reports in 2002.
levels. This trial was stopped early
C.6 Statins and older people:
after finding a 36% reduction in a
Statin treatment in older people is
combined fatal and non-fatal CHD
as effective as in middle-aged
endpoint.54
adults. The sub-group analyses of
D. Non-
those aged 55+ and 65+ years
C.5 Statins in women: The
within individual trials have
cholesterol
efficacy of statin treatment among
reported risk reductions at least as
women is less certain due to
good as, if not better than, those
lowering
Table 3
The relative efficacy of treatment with cholesterol lowering drugs+
alternatives
Primary prevention
Secondary prevention
Secondary prevention
Cholesterol lowering is only part
with statins
with statins
with fibrates
of the repertoire of possible
effective interventions to reduce
Total mortality
0.77 (0.60–0.99)
0.79 (0.73–0.86)
0.97 (0.90–1.05)
CHD risk and not necessarily the
most important. CHD risk can also
CHD mortality
0.68 (0.46–1.00)
0.74 (0.66–0.83)
0.93 (0.85–1.01)
be significantly reduced by
Non-fatal MI
0.68 (0.56–0.84)
0.70 (0.61–0.80)
0.57 (0.28–1.11)
changes in lifestyle (e.g. smoking
Net cholesterol
cessation, exercise and the use of
lowering
20%
21%
9.5%
non-cholesterol lowering diets)
+ Figures are pooled relative risk estimates (95% confidence intervals).
and drug treatments (e.g. to lower
FEBRUARY 1998
Cholesterol and CHD: screening and treatment EFFECTIVE HEALTH CARE 7

---

blood pressure, beta-blockers after
MI, which also had no effect on
use of drugs which may have had
a myocardial infarction (MI), and
cholesterol levels (RR = 0.24 95%
adverse effects and generally, the
aspirin). A recent Effective Health
CI: 0.1–0.8).61, 62 The most
variable quality of the
Care bulletin on Stable Angina
prominent change in the
programmes.
(1997, vol 3 no.5) reviewed the use
intervention group was an increase
of invasive treatments such as
in consumption of alpha-linolenic
Evidence from trials of post-MI
coronary artery bypass grafting
acid from rapeseed margarine
rehabilitation are also relevant as
(CABG) and angioplasty (PTCA).
(used as the participants found it
many of these included smoking
difficult to consume high intakes
cessation together with increases
D.1 Smoking cessation: Smoking
of olive oil).
in physical activity. Trials that
cessation advice given in primary
attempted to modify several risk
care settings has a small but
The striking findings of the trials
factors, including smoking, and
important effect on long-term
of oily fish and Mediterranean diet
not just increase physical activity,
behaviour. Pooled estimates from
certainly require replication, and if
showed reductions in CHD
188 trials show that around 2%
substantiated, these diets would
mortality (RR = 0.63; 95% CI: 0.51
(95% CI: 1 – 3%) of those given
have an important role in reducing
– 0.80) and total mortality (RR =
personal advice during one routine
mortality following myocardial
0.77; 95% CI:.0.64 – 0.94).44 The
consultation stopped smoking and
infarction. The effect of these
absolute levels of CHD mortality
did not relapse up to 1-year later.58
interventions in people at lower
in these trials were of the order of
The use of nicotine gum increases
risk of CHD is not known.
4% per year in the control group,
the quit rates to about 4% (95%
giving a number needed to treat of
CI: 2 - 6%). This will lead to
D.3 Exercise: Lack of physical
about 13 people for 5 years to
approximately a 1 – 2% overall
activity has been shown to be a
avoid one CHD death. A future
reduction in mortality and
strong independent risk factor for
Effective Health Care bulletin will
morbidity. The effect is much
death from CHD.63 It is estimated
provide a more comprehensive
larger in those who quit, but only
that a sedentary lifestyle doubles
summary of the research evidence
a small percentage quit with simple
the risk of CHD mortality (95% CI:
about cardiac rehabilitation.
advice.
1.6 – 2.2). However, there are no
reliable trials examining the
D.5 Aspirin: In primary
Advice to stop smoking is much
impact on survival of interventions
prevention aspirin does not reduce
more effective among those people
solely aimed at promoting exercise
all-cause mortality significantly.67
who have suffered a myocardial
and there is considerable debate
However, the participants in both
infarction, with up to 36%
about the level or intensity of
of the large primary prevention
stopping.59 This results in over a
exercise which confers
trials were physicians – a group at
30% reduction in the mortality
cardiovascular benefit.64 A recent
very low risk of CHD. Aspirin
risk. The next issue of Effectiveness
review found that a proportion of
appears to reduce mortality among
Matters published by the NHS
patients did respond positively to
people who have not yet had a
Centre for Reviews and
exercise advice given in a primary
heart attack but who are at high
Dissemination will summarise
care setting.65
risk of such an event (e.g. unstable
reviews of the effectiveness of
angina, stable angina and
interventions to promote smoking
A computer simulation based on
peripheral vascular disease).68
cessation. A future Effective Health
the epidemiological evidence of
Care bulletin will examine the
the association between exercise
D.6 Lowering raised blood
effectiveness of ways to prevent
and CHD mortality has estimated
pressure and beta-blockers post
the uptake of smoking in young
that if the proportion of the
MI: Systematic reviews of RCTs
people.
population undertaking moderate
show that for people with high
activity were increased by 25%,
blood pressure, anti-hypertensive
D.2 Oily fish and Mediterranean
the number of life years gained
medication reduces the risk of
diets: Increased intake of oily fish
would be similar to a 2% reduction
CHD and all-cause mortality.69
has been shown to reduce
in the proportion of smokers.66
cardiovascular mortality after
Epidemiological studies among
heart attack without reducing
D.4 Multiple interventions: Trials
survivors of cardiovascular disease
cholesterol levels (RR = 0.65 95%
of multiple risk factor
show that the relationship between
CI: 0.5–0.9). In the DART trial60
interventions for primary
blood pressure and both total and
22% of participants did not like
prevention in workplace settings
CHD mortality follows a U-shaped
oily fish and consequently were
and primary care show very small
relationship. This has caused
given maxepa supplements.
and non-significant effects on
concern that treating high blood
CHD mortality (RR = 0.96; 95% CI:
pressure following MI may cause,
Significant reductions in CHD
0.89 – 1.04).41 This is probably due
rather than prevent, mortality.
were also found in a trial of
to poor adherence to non-
However, new evidence shows that
Mediterranean diet in people after
pharmacological interventions, the
the poor prognosis in those with
8 EFFECTIVE HEALTH CARE Cholesterol and CHD: screening and treatment
FEBRUARY 1998

---

Table 4
Relative treatment effects (ie. vascular deaths) and a number needed to treat for five years to avoid one event for alternative
treatments for the prevention of coronary heart disease at a range of baseline levels of CHD risk.
Number of people needed to treat for 5 years to avoid one event
Relative risk
by their annual percent risk of CHD
Treatment
(95% CI)
0.1%
0.5%
1.5%
3.0%
4.0%
6.0%
Primary prevention
Smoking advice58
0.99 (0.98–1.0)
20,000
4000
1333
666
500
333
Nicotine replacement58
0.98 (0.98–0.99)
10,000
2000
667
333
250
166
Aspirin72
0.98(0.78–1.18)*
10,000
2000
667
333
250
166
Anti-hypertensive drugs67, 73, 74
<60 yrs
0.79 (0.71–0.87)
950
190
63
31
24
16
60+yrs
0.75 (0.64–0.88)
800
160
53
26
20
13
Statins
0.68 (0.46–1.00)
625
125
41
21
16
10
Secondary prevention
Aspirin72
0.82 (0.76–0.88)
–
222
74
37
28
18
Beta-blockers75
0.78 (0.71–0.87)
–
181
61
30
23
15
Statins
0.74 (0.66–0.83)
–
154
51
26
19
13
Smoking advice59
0.68 (0.57–0.79)
–
125
42
21
16
10
Oily Fish60
0.65 (0.5–0.9)
–
114
38
19
14
9
Mediterranean diet61, 62
0.24 (0.1–0.8)
–
52
17
9
7
4
– Risk level too high for primary prevention or too low for secondary prevention
* Not a statistically significant treatment effect
NNTs in bold are those that equate to range of CHD event rates occuring in randomised controlled trials or meta-analysis
low blood pressure is due to
of some of these interventions in
considerable variation.
damaged heart muscle and not the
terms of numbers needed to be
Interventions generally considered
low blood pressure.70 This
treated (NNTs) for 5 years to avoid a
to be worthwhile (aspirin for
evidence, taken in conjunction
vascular death. A range of different
secondary prevention and anti-
with the trials of beta-blockers
baseline risks is used to compare
hypertensive treatment in older
conducted in post-myocardial
NNTs which correspond to the
people) have NNTs rather greater
infarction patients suggest that
differences that might be expected
than those for statins. The 5-year
treatment is beneficial.71, 76
in primary and secondary care
NNTs for smoking cessation advice
settings among men and women.
are very high but are not strictly
D.7 Numbers needed to treat:
comparable with drug NNTs as
Table 4 presents some summary
The NNT for 5 years of different
treatment is very cheap, is only
information on the potential effects
drug treatment options, shows
given once and the CHD events
Table 5
Costs per year of life gained (£PLYG) for a range of different interventions+
Drug interventions
£PLYG, gross (95% CI)
£PLYG, net
Statins
- Simvastatin 27mg/day (1.37/day)
£8,240 (£6,220, 11,280)
£7,240
Anti-hypertensives (bendrofluazide 2.5mg, 0.1p/day)*
- middle-aged
£ 70 (£ 40–130)
£ 580 (saved)
- elderly@
£ 45 (£ 30–180)
£ 870 (saved)
Anti-hypertensives (combined regimen of bendrofluazide, atenolol, enalapril, 53p/day)*
- middle-aged
£1,510 (£940–3,050)
£ 860
Aspirin (300mg/day,# 0.5p/day)
*£ 50 (£ 30–320)
£ 407(saved)
Aspirin (150mg) + dipyridamole (400mg, 24p/day)*
£2,800 (£1,500–17,080)
£2,340
Beta-blockers (atenolol 50mg, 3.8p/day)*
£ 230 (£170–410)
£ 130
Dietary interventions*
Fish diet, advice only (£41/yr)
£ 560 (£330–2,220)
£ 610
Fish diet + 20mg maxepa/week (£57/yr)
£ 780 (£460–3110)
£ 830
Mediterranean diet (£52/yr)
£ 290 (£200–1,980)
£ 180
+ Figures are £ (1998) per life year gained with discounting of costs and benefits at 6% for patients with an absolute baseline risk of CHD events of 3%
per year.
Net costs take into account projected savings from reduced admissions and treatment for clinical events avoided.
* No data on revascularisation procedures avoided by treatment, hence potential savings are underestimated.
@ CHD event rate for elderly people was derived from trials and was equivalent to 4.5% per year.
# aspirin dose used in post myocardial infarction trials was 1.2gm/day but current practice would favour a lower dose.
FEBRUARY 1998
Cholesterol and CHD: screening and treatment EFFECTIVE HEALTH CARE 9

---

prevented are counted over a
among the simvastatin treated
cost-effectiveness estimates for a
lifetime rather than the 5-year
group were 32% lower than the
range of interventions. This
period. Nonetheless, they provide
placebo group,79 and that almost
demonstrates that the role of
some indication of the relative
90% of the drug costs were off-set
cholesterol lowering drugs must
effects of different types of
by savings in hospital admissions.80
be considered alongside other
intervention. A better guide to
However, because the rates of
appropriate options. It can be seen
policy however is provided by
revascularisation in the UK are
that several other interventions are
looking at the relative cost-
lower than in Scandinavia (where
more cost-effective than using
effectiveness of these options.
the trial was carried out), the
statins.
savings are unlikely to be as great.
However, more effective treatment
Smoking cessation interventions
of people at high risk of CHD
have also been shown to be highly
E. Cost-
events may reduce pressure for
cost-effective. The costs per life
increasing the rates of
saved are low and have been
effectiveness
revascularisation.
estimated to be about £500 per life
year gained.83 The additional cost
E.1 The model: The cost-
E.2 The importance of the level of
per life year gained of brief
effectiveness estimates of various
CHD risk: The baseline level of
counselling or the use of nicotine
interventions based on a life table
CHD risk has a major impact on
substitutes (e.g. gum), over and
model developed by the University
the absolute effect or impact of
above brief advice, is
of Sheffield are shown in Table 5.
interventions and should
approximately £2,500 if costs to
More details on the methods used
therefore, be taken into account
smokers as well as the NHS are
are given in the Appendix. The
when deciding who should receive
taken into account.
costs per life year gained in
which treatment.81 This is
primary and secondary prevention
illustrated in Table 4 and Fig. 7
If more people at increased CHD
with statins are very similar to
which show how the NNT and
risk were appropriately treated
previous estimates based on the
cost of achieving an extra year of
with aspirin and anti-hypertensive
WOSCOPS trial77 and by the 4S
life increase as people with lower
drugs, helped to stop smoking and
investigators,78 suggesting that the
initial CHD risk are treated. A
change their diet, then a large
methods used in the Sheffield
recent economic evaluation of
number (possibly over half) would
model are robust.
lipid lowering in primary care in
have their CHD risk sufficiently
patients with moderately raised
reduced to make statin treatment
The final column in Table 5 shows
risk doubted whether drug
unnecessary or relatively cost-
the net cost per life year gained
treatment as primary prevention is
ineffective.84
which takes into account potential
cost effective.82
savings due to avoiding CHD
E.4 Which statin: The net cost per
events and associated costs of
E.3 Alternatives to cholesterol
life year gained with statins of
treatment and hospitalisation. For
lowering: A major advantage of
around £7,000 (for patients with an
example, analyses of the 4S trial
the analyses presented in Table 5
annual CHD event risk of about
data showed that hospital costs
is that they provide comparable
3%), though quite high, compares
favourably with several other
interventions currently provided by
£ (thousands) per life year gained
the NHS, including those in the
20
management of coronary heart
disease. If a patient is still at
sufficiently high risk after using
other, more cost-effective options,
15
the use of a statin may be
appropriate. In such cases one of
the issues to be considered is
which one to use.
10
Fig. 8 shows the cholesterol
lowering effects of a range of
5
statins, both as the total
percentage reduction, and the
annual average drug costs per
percentage reduction for each
0
drug, in LDL cholesterol.
1.0%
1.5%
3.0%
4.5%
10.0%
CHD event rate
Fig. 7 Cost per life year gained with statins by initial CHD risk
10 EFFECTIVE HEALTH CARE Cholesterol and CHD: screening and treatment
FEBRUARY 1998

---