1-36
Observed Benefits of Antihypertensive Drug Treatment
Results of the Trials of Traditional Therapy
Controlled Clinical Trials of Drug Treatment
for Hypertension
A Review
Jeffrey A. Cutler, Stephen W. MacMahen, and Curt D. Furberg
Downloaded from http://hyper.ahajournals.org/ by guest on June 17, 2017
We reviewed the designs and major results of 17 large-scale, controlled, clinical trials that
reported the effects of drug treatment for hypertension on morbidity or mortality. Seven trials
conducted in study populations with more-severe hypertension (diastolic blood pressures 100120 mm Hg or higher), including the more-severe stratum of the Veterans Administration Trial,
showed large reductions in stroke, other "hypertensive" events, and, in one trial, total
mortality. Of 11 trials in populations with less-severe hypertension (diastolic blood pressures
predominantly below 105 mm Hg), including the less-severe stratum of the Veterans Administration Trial, nine met the criteria for pooling of results. Among the aggregate 43,000 patients
in the nine trials who were followed up for an average of 5.6 years, mean diastolic blood
pressure reduction was 5.8 mm Hg, and a significant 11% reduction in total mortality was
observed. This benefit was largely attributable to a 38% reduction in fatal strokes; nonfatal
strokes were similarly reduced. Coronary heart disease mortality was 8% lower in drug
treatment than in control groups, but this difference was not significant. A similar result was
observed for combined coronary mortality and nonfatal myocardial infarction. A possible
explanation for the inconclusive result regarding coronary end points was an adverse trend,
observed in several trials, in a subgroup with baseline resting electrocardiographic abnormalities. Because all the trials except the propranolol arm of the Medical Research Council trial
used drug regimens based on thiazide-like diuretic agents, and because there are now several
new drug classes proposed as initial therapy, additional large-scale clinical trials may need to be
considered. (Hypertension 1989;13(suppl I):I-36-I-44)
T
hree conditions are generally considered necessary to justify a systematic effort by the
health care system to detect and treat a risk
factor for common and serious diseases such as
cardiovascular disease: 1) the risk factor has a high
prevalence, 2) the risk factor has a strong association with the disease, and 3) treatment has been
shown to be efficacious as well as safe, tolerable,
and affordable by society. Other articles in this
From the Prevention and Demonstration Research Branch,
Division of Epidemiology and Clinical Applications, National
Heart, Lung, and Blood Institute, National Institutes of Health,
Bethesda, Maryland (J.A.C); the Medifcal Research CouncilImperial Cancer Research Fund Clinical' Trials Service Unit,
Nuffield Department bf-Clini£al Medicine, Radcliffe Infirmary,
Oxford, England (S .W. M.); and the Center for Prevention Research
and Biometry, Bowman Gray School of Medicine of Wake Forest
University, Winston-Salem, North Carolina (C.D.F.).
Address for correspondence: Jeffrey A. Cutler, MD, Room
604, Federal Building, National Institutes of Health, Bethesda,
MD 20892.
The authors are grateful to Dr. Robert M. Graham for acting as
guest editor in the evaluation of this manuscript.
supplement have summarized the evidence establishing the first two of these conditions for primary
hypertension. This article introduces and reviews
the principal evidence regarding the efficacy of
antihypertensive drug treatment in reducing morbidity and mortality, namely, the evidence from
large-scale, long-term, randomized, controlled, clinical trials. The approach is quasihistorical: the earlier trials that dealt predominantly with more-severe
hypertension are summarized first, and then the
more recent widely debated trials in less-severe
hypertension are discussed in greater detail. Except
for this crude stratification by level of blood pressure (BP), this review does not address important
further questions including whom to treat, what
drugs to use, or the balance of benefits and costs
(medical and financial). These topics have been
previously addressed1 and are considered further
in other contributions to this supplement. A discussion of the relation between baseline electrocardiographic abnormalities and treatment effects
is also included.
Cutler et al
Trials in More-Severe Hypertension (Diastolic
Blood Pressure 100-120+ mm Hg)
The beginnings of randomized, controlled, clinical trials in hypertension can be traced to a doubleblind placebo-controlled trial conducted by the Veterans Administration (VA) in hypertensive men of
all severity levels and first reported in I960.2 This
trial was not designed to detect the effects of
treatment on morbidity and mortality; rather, the
trial apparently served as a pilot study for the
landmark VA trial that separately reported the
results for severe and less-severe hypertension in
1967 and 1970, respectively.- Before the 1967 VA
report, two other trials in middle-aged subjects with
predominantly severe hypertension were completed
by Hamilton et al and by Wolfe and Lindeman;
another trial was reported several years afterward
by Barraclough et al.
Downloaded from http://hyper.ahajournals.org/ by guest on June 17, 2017
All these trials used a variety of drug classes,
including the thiazides, but only the VA trial had a
standardized initial regimen (a combination of hydrochlorothiazide, reserpine, and hydralazine). The
VA trial was also the only one of the four trials that
was double-blind as well as placebo controlled. The
Hamilton trial was potentially more subject to bias
than the others because alternate allocation rather
than randomization was used for assignment to drug
treatment or no treatment (without placebo).
Each trial produced a large net reduction of
diastolic BP in the treatment group. However, with
the modest sample sizes (61-143 participants) and
relatively short follow-up periods (only Hamilton's
trial went beyond 18 months), there were neither
enough deaths nor coronary heart disease (CHD)
events, even in the aggregate, to evaluate the effects
of treatment on these end points. Specifically, there
were six deaths in treated groups and 10 in the
control groups (total from all four trials). Three of
the trials3-5-6 also reported "hypertensive" events
(strokes, congestive heart failure, dissecting aneurysm, or nephropathy). The results, which indicated substantial benefits (>50%), have been discussed in detail in a previous review.8
Three other long-term randomized trials in moresevere hypertension have been conducted in populations differing in major ways from those described
above. Carter's9 trial in 99 patients with a previous
stroke demonstrated a statistically significant 57%
reduction in total mortality, largely attributable to
fewer recurrent strokes. Treatment in this trial
consisted mainly of use of the thiazides and sodium
restriction. Sprackling et al10 reported the results of
a trial conducted in 123 very elderly patients (mean
age, 81 years) randomized to methyldopa or no
treatment. Mortality was very high (90%) in both
randomized groups, and survival was not significantly affected by treatment. Finally, Coope and
Warrender" recently reported the results of another
trial in the elderly, aged 60-79 years, with qualifying diastolic BP of 105-120 mm Hg or systolic BP
Review of Drug Treatment Trials
1-37
more than 170 mm Hg. Eight hundred eighty-four
subjects were randomized to active treatment (with
atenolol, bendrofluazide, or both drugs) or no treatment and were followed up for 4.4 years on average. There was a significant reduction in fatal and
nonfatal stroke incidence (20 vs. 39); the favorable
trends in CHD events and total mortality were not
significant.
In summary, seven controlled clinical trials that
were designed with various degrees of rigor, that
studied a variety of moderately-to-severely hypertensive populations, and that used various treatment regimens showed reductions in hypertensive
events, especially stroke. Although there was some
suggestion of benefit, there were too few deaths and
CHD events to reliably determine moderate treatment effects on these end points.
Trials in Less-Severe Hypertension
Beginning with the initial report on the lesssevere stratum of the VA trial,3 results have been
published from nine randomized trials that enrolled
subjects with entry diastolic BP as low as 85-95
mm Hg and that also met the following inclusion
criteria for this review: 1) the trial objective was to
test the effect of BP reduction on morbidity or
mortality (i.e., the trial was not limited to comparing active drugs); 2) the focus was on primary
prevention, so that most subjects enrolled were free
of clinical cardiovascular disease; 3) allocation to
treatment groups was by randomization of individual subjects; and 4) data were reported for one or
more of the following end points: total mortality,
stroke, and major CHD events (CHD death or
nonfatal myocardial infarction) by "intentionto-treat" analysis (i.e., with inclusion of events
occurring throughout follow-up in all subjects by
original randomization group). In the interest of
drawing unbiased conclusions and maximizing use
of available information from the trials collectively,
studies that met the above criteria were not excluded
or subgrouped by such features as demographic
characteristics of subjects, drug doses, or secondstep and subsequent drugs. These subset issues are
discussed in a previous report.8
Information from the nine trials about study
designs, drug regimens, adherence to assigned regimens, and BP results is summarized in Table 1.
Other than the VA trial, these trials include the US
Public Health Service (PHS) Hospitals Cooperative
study,12 the VA National Heart, Lung, and Blood
Institute (VA-NHLBI) Feasibility study,13 the mild
hypertension trial of the Oslo study,14 the Australian National Blood Pressure (ANBP) study,15 the
trial of the European Working Party on Hypertension in the Elderly (EWPHE),16 the British Medical
Research Council (MRC) trial of treatment of mild
hypertension,17 the Hypertension Detection and
Follow-up Program (HDFP),18-20 and the hypertensive stratum (entry diastolic BP 90-114 mm Hg or
on antihypertensive medication) of the Multiple
1-38
Supplement I
Hypertension
Vol 13, No 5, May 1989
TABLE 1. Randomized Drug Treatment Trials in Less-Severe Hypertension: Design Features and Protocol Implementation
Study population
Age
Number of
participants (years)
Length of
follow-up
(years)
380
Mean, 52
Entry DBP
(mm Hg)
90-114
389
1,012
785
3,427
840
21-55
21-50
40-49
30-69
Mean, 72
90-114
85-105
95-109
95-109
90-119
6.5-9.0
1.5
5.5
4.0
4.7
MRC, 1985
17,354
35-64
90-109
5.5
HDFP, 1979
10,940
30-69
90+
5.0
8,012
35-57
90-114
7.0
Study, year
reported
VA, 1970
PHS, 1977
VA-NHLBI, 1978
Oslo, 1980
ANBP, 1980
EWPHE, 1985
Downloaded from http://hyper.ahajournals.org/ by guest on June 17, 2017
MRFIT, 1982
(hypertensive
subjects)
3.8
DBP effect
(mm Hg)
Maximum %
crossover
Net
Step 1
drug(s) 1Baseline change Control Treatment
104
14
DB-placebo
HCTZ
-19
16
+RES
+HDRZ
33
DB-placebo
CTZ
99
-10
35
21
19
DB-placebo
93
-7
CTLD
Open-untreated HCTZ
97
1
-10
20
34
SB-placebo
CTZ
100
-6
37
SB-placebo
101
-10
37
36
HCTZ
+TMTR
SB-placebo
98
-6
44
40
BDFZ or
PROP
Open-referred CLTD
101
-5
22
58
care
Open-usual
CLTD or
96
-4
23
65
HCTZ
care
Designcontrols
DBP, diastolic blood pressure; VA, Veterans Administration; PHS, US Public Health Service Hospitals Cooperative Study;
VA-NHLBI, VA National Heart, Lung, and Blood Institute Feasibility Study; Oslo, Oslo study; ANBP, Australian National Blood
Pressure Study; EWPHE, European Working Party on Hypertension in the Elderly; MRC, British Medical Research Council trial;
HDFP, Hypertension Detection and Follow-up Program; MRFIT, Multiple Risk Factor Intervention Trial; DB, double-blind; SB, singleblind; HCTZ, hydrochlorothiazide; RES, reserpine; HDRZ, hydralazine; CTZ, chlorothiazide; CTLD, chlorthalidone; TMTR,
triamterene; BDFZ, bendrofluazide; and PROP, propranolol.
Risk Factor Intervention Trial (MRFIT).21-22 MRFIT
is the only one of these trials with additional interventions by design (lipid-lowering diet and smoking
cessation), and its exclusion would not materially
affect the review findings.
The study populations of these trials ranged in
size from 380 in the VA trial to more than 17,000 in
the MRC trial. All trials included predominantly
young-to-middle-aged adults except for the EWPHE,
which enrolled men and women aged 60 years and
older. The other trials also admitted subjects of
either gender, with the exception of the VA, Oslo,
and MRFIT trials. Generally, entry diastolic BP had
to be in the 90-114-mm-Hg range. Some subjects of
the Oslo trial had isolated systolic hypertension;
some in the VA-NHLBI trial had borderline diastolic hypertension. Follow-up periods were generally about 4-7 years.
The three earliest trials had double-blind placebocontrolled designs. Three later studies, the ANBP,
EWPHE, and MRC trials, used a single-blind
approach with placebo controls. The other three
studies had open designs, with the Oslo trial providing monitoring and treatment as necessary for
control subjects through the research clinic, while
in the HDFP and MRFIT, control subjects were
referred at the outset to community sources of care
for decisions about management. All trials used a
thiazide or a thiazide-like diuretic agent (e.g., hydrochlorothiazide, chlorothiazide, chlorthalidone, or
bendrofluazide) as the mainstay of treatment except
for the propranolol arm of the MRC trial. (Compar-.
ison of the effects of diuretic agents and /3-blockers
is the subject of the article by Furberg and Cutler in
this supplement.23) With the use of these drugs, as
well as designated "step-up" drugs in most trials,
net changes in mean diastolic BP of 4-19 mm Hg
were achieved. The mean change in diastolic BP for
participants in all trials was 5.8 mm Hg. An additional feature of the implementation of these protocols, which is important for understanding the BP
changes and interpreting trial results, is shown in
the last two columns of Table 1. In all trials, a
substantial proportion of control subjects were
receiving active treatment or were of unknown
treatment status at the end of follow-up. This proportion ranged from 14% in the VA trial to 58% and
65% in the HDFP and MRFIT, respectively. The
high "drop-in" was inherent in the design of the
latter studies; the maintenance of substantial diastolic BP differences, nevertheless, was in part due
to the relatively low dropout rates from the active
treatment groups (22% and 23%). The proportion
classified as dropouts or of unknown treatment
status in the other seven trials ranged from 1%
(Oslo) to 40% (MRC).
Morbidity and mortality data are presented in
Figures 1, 2, and 3 for the nine trials individually
and collectively (as pooled results). The rationale
for combining results was to estimate any effect of
treatment as precisely as possible, as well as to
provide maximal statistical power to detect an effect
of modest size, as there may be for CHD (see
MacMahon et al24 for details). To avoid bias in the
estimates, it is important that 1) all relevant trials
are included and 2) in each trial, all morbid events
Cutler et al Review of Drug Treatment Trials
INTERVENTION BETTER
THAN CONTROL
CONTROL BETTER
THAN INTERVENTION
0 -20-40-60-80-100 %
+ 100+80 + 60 + 40 + 20
i
1
1
T"
1—
1
VA
{ 10/21 )
PHS
(
2/4 )
VA NHLBI
(
2/0 )
OSLO
( 10/9 )
ANBP
( 25/35 )
EWPHE
(135/1491
MRC
(248/253)
HDFP
(349/419)
MRFIT
(174/169)
ALL TRIALS
1-39
FIGURE 1. Bar graph showing estimates
(x) with approximate 95% confidence intervals (-) of the relative difference in total
mortality between intervention and control groups. The numbers in parentheses
are the numbers of events (intervention/
control). VA, Veterans Administration;
PHS, US Public Health Service Hospitals
Cooperative Study; VA-NHLBI, VA
National Heart, Lung, and Blood Institute
Feasibility Study; OSLO, Oslo study;
ANBP, Australian National Blood Pressure study; EWPHE, European Working
Party on Hypertension in the Elderly;
MRC, British Medical Research Council;
HDFP, Hypertension Detection and
Follow-up Program; and MRFIT, Multiple Risk Factor Intervention Trial.
Downloaded from http://hyper.ahajournals.org/ by guest on June 17, 2017
INTERVENTION BETTER
THAN CONTROL
CONTROL BETTER
THAN INTERVENTION
100 + 80 + 60 + 40+20
r
0-20-40-60-80-100%
FATAL STROKE
-K-
1/7
PHS
RELATIVE DIFFERENCE NOT ESTIMATABLE
0/0
VA-NHLBI
RELATIVE DIFFERENCE NOT ESTIMATABLE
VA
i
OSLO
-
0/2
-
3/6
( 21/31 )
( 18/27 )
-X-
ANBP
-X-H-
EWPHE
MRC
FIGURE 2. Bar graph showing estimates
(X) with approximate 95% confidence inter( 29/52 ) vals (-) of the relative difference in fatal
( 11/9 ) and nonfatal stroke between intervention
and control groups. The numbers in parentheses are the numbers of events (intervention/control). See the legend to Figure 1
for explanation of abbreviations.
-X-
HDFP
MRFIT
0/0
-•
ALL TRIALS
NON FATAL STROKE
(
4/13 )
(
0/2 )
(
0/0 )
OSLO
(
0/3 )
ANBP
( 10/16 )
-X-
VA
PHS
-
VA-NHLBI
*
-
-X-
•
RELATIVE DIFFERENCE NOT ESTIMATABLE
"
-¥r
MRC
HDFP
MRFIT
-X-+
ALL TRIALS
-*-
( 42/82 )
I 73/106)
( 27/23 )
1-40
Supplement I Hypertension
CONTROL BETTER
THAN INTERVENTION
100+80+60+40+20
T"
T
r T" T
FATAL CHD
Vol 13, No 5, May 1989
INTERVENTION BETTER
THAN CONTROL
0 -20-40-60-80-100 %
~i
:
VA
1
1
1
( 6/11 )
PHS
-K-
( 2/4 )
VA-NHLBI
• • — X>100%-
( 2/0 )
OSLO
• * — X>ino%-
( 6/2 )
ANBP
(
EWPHE
( 29/47 )
-X-
(106/97 )
-K-
MRC
5/11")
HDFP
1131/148)
MR FIT
( 80/79 )
Downloaded from http://hyper.ahajournals.org/ by guest on June 17, 2017
ALL TRIALS
NONFATAL Ml
VA
I 5/2
•-X>100%-
PHS
"*
-*-
VA-NHLBI
•*•
-K-
OSLO
-*•
ANBP
•*-
MRC
HDFP(ECG)
MRFIT
FIGURE 3. Bar graph showing estimates
(x) with approximate 95% confidence
intervals (-) of the relative difference in
fatal coronary heart disease (CHD) and
nonfatal myocardial infarction (MI)
between intervention and control groups.
The numbers in parentheses are the numbers of events (intervention/control). (Note
that the data from the EWPHE trial are
reported only for total cardiac mortality.)
See the legend to Figure 1 for explanation
of abbreviations.
( 7/6
( 6/5
! 8/8
-*•
( 28/22
-K-
-X-
(116/137)
( 40/41 )
1174/197)
ALL TRIALS
have been reported. With regard to the first point,
we have endeavored to include all published trials
meeting the review criteria stated above; we have
not sought data from unpublished trials. With regard
to the second point, all trials provided data on
cause-specific mortality and, with one exception,14
on nonfatal CHD and stroke by intention to treat.
The statistical method for combining results is that
of Yusuf et al.25 This procedure, which ensures that
treatment and control subjects are compared only
within each trial and not between trials, provides a
"typical" odds ratio as the weighted mean estimate
of effect, a Z statistic and corresponding p value,
and a 95% confidence interval (CI). Similar statistics are provided for each individual trial; in Figures
1, 2, and 3, the odds ratios are presented as percent
differences in risk of end points.
Total mortality was lower in the intervention than
in the control groups in six of the nine trials (Figure
1). In one (HDFP), the difference was statistically
significant (i.e., the 95% CI does not overlap zero).
The finding for this trial population, which was
drawn by probability sampling from residential areas,
may be the most generalizable to the US population
with due consideration of the BP reduction achieved
in clinical practice. Also, the estimated relative
benefit for the HDFP trial falls within the confidence limits of all other trials. Three trials (VA,
ANBP, and HDFP) showed distinctive favorable
point estimates of a 17-50% reduction, while four
trials (Oslo, EWPHE, MRC, and MRFIT) showed
small differences (8% benefit to 4% harm). The two
remaining trials reported very few deaths. The
pooled results demonstrated a statistically significant 11% reduction in total mortality (95% CI:
-19%, -2%).
The results for the incidence of fatal and nonfatal
stroke from each trial are shown in Figure 2 except
that the EWPHE trial contributed data on fatal
strokes only for this intention-to-treat analysis. There
were fewer strokes in the group randomized to
active treatment in all trials, except for one (VA-
Cutler et al
Downloaded from http://hyper.ahajournals.org/ by guest on June 17, 2017
NHLBI) that reported no such events and the
MRFIT, in which there was essentially no difference in stroke rates. In three trials, the VA, MRC
(nonfatal stroke only), and HDFP, the differences
were significant. The pooled estimates from all
trials are a 38% reduction in stroke mortality (95%
CI: - 2 1 % , -55%) and a 36% (±10%) reduction in
combined fatal plus nonfatal stroke (data not shown
in the figure).
The data for CHD events are presented in Figure
3. Judgments were required regarding inclusion of
data from two trials. In the HDFP, nonfatal myocardial infarctions were ascertained by three methods:
1) new pathological Q waves on electrocardiograms
recorded at the 2nd and 5th years of follow-up, 2)
Rose questionnaire at the same intervals, and 3) an
annual query to participants about diagnosed heart
attacks.20 As the last two measures are wholly
subjective, we were concerned about bias in this
unblinded trial. There was some evidence of overascertainment of nonfatal myocardial infarction
(false- positives) when these were included, because
the ratio of nonfatal myocardial infarction to fatal
CHD was twice as high in the referred-care group
compared with control groups in other trials. Therefore, we have included only those myocardial infarctions as ascertained by serial electrocardiogram
from the HDFP trial, with the recognition that the
rates would be substantially underestimated in both
groups but without serious bias (although the proportion with no follow-up electrocardiogram was
6% higher in the referred-care than in the steppedcare group).
The other issue concerns the reporting of cardiac
rather than coronary mortality from the EWPHE
trial. It is reasonable to assume that the reduction in
cardiac mortality in this trial was at least partly due
to a reduction in mortality from congestive heart
failure. However, because most of the deaths were
probably from CHD, we included the EWPHE
mortality data in our CHD analysis; however, we
excluded nonfatal myocardial infarctions from the
EWPHE, as these events were not reported by
intention to treat.
The results for CHD death and for nonfatal
myocardial infarction from individual trials are similarly distributed on both sides of the zerodifference line, although somewhat more trials show
favorable trends for mortality than for nonfatal
myocardial infarction (Figure 3). For CHD mortality alone, the pooled estimate is a reduction of 8%
(95% CI: - 2 1 % , +5%); only the EWPHE trial
showed a significant benefit (see the qualification
noted above). For fatal and nonfatal CHD combined, the pooled treatment effect is estimated as
- 9 % , a favorable trend, but the 95% CI ranges from
-19% to +1%.
Two other drug trials in mild-to-moderate hypertension were not included in the pooled analyses
because they did not meet one of the review criteria. The Hypertension-Stroke Cooperative Study26
enrolled only those subjects with a recent clinical
Review of Drug Treatment Trials
1-41
history of cerebrovascular disease, so this trial was
not focused on primary prevention. In this trial,
differences between drug treatment and control
groups for stroke, CHD, and total mortality were
not significant, although they were in a favorable
direction and are quite consistent with our pooled
estimates of effect. In Morgan et al's 27 trial, men
were allocated to four treatment arms on randomly
chosen weeks rather than individually. End-point
events, primarily CHD deaths, were few, with more
deaths in the combined-drug treatment groups (thiazide or propranolol) than in control groups. This
finding was attributed to a significant increase in the
thiazide-treated group; mortality was not different
between propranolol-treated and control groups.
Hypotheses Regarding Effects of Treatment
on Coronary Heart Disease
The inconclusive results of the trials with regard
to treatment effects on CHD have stimulated much
debate and analysis. The article by MacMahon et
al24 considers in some detail the question of whether
the observed effects of treatment are consistent
with those that might reasonably be expected with
the achieved reduction in BP. It has been suggested
that reliable detection of plausible treatment effects
on CHD would have required somewhat larger
numbers of patients than were present and randomized, collectively, in the trials. However, the effects
of treatment in the trials might actually be less than
those expected on the basis of BP reduction alone,
as the thiazide diuretic agents are known to produce
several metabolic changes that could offset some
benefits of BP reduction. Diuretic drugs appear to
increase blood cholesterol levels by about 3-5% 2829 ;
this increase is primarily in low density lipoprotein
cholesterol. During these trials, such an effect on
blood cholesterol might be expected to increase the
risk of CHD by about 6-10%.3° Clearly, this effect
alone could substantially offset the potential benefits of BP reduction on CHD, which might have
been as great as a 15-20% reduction in risk.
Furthermore, treatment with diuretic agents is
known to increase ventricular ectopic activity,31
perhaps due to effects on serum levels of potassium
and magnesium, and the possibility that this effect
may subsequently produce an excess of sudden
CHD deaths has been raised. On the basis of data
from the MRFIT, it has been suggested that such an
effect of treatment might be greater in those who
already have minor abnormalities on the resting
electrocardiogram. In that study, treatment with
diuretic agents in the special-intervention group
was associated with a significantly greater risk of
CHD death in those with an abnormal baseline
electrocardiogram than in those with a normal electrocardiogram (Table 2). Such nonrandomized comparisons are, of course, potentially subject to bias.
However, a similar interaction was also suggested in
the randomized comparisons of the specialintervention group and the usual-care control group
1-42
Supplement I
Hypertension
Vol 13, No 5, May 1989
TABLE 2. Coronary Heart Disease Mortality for MRFIT SpecialIntervention Participants*
Treatment status
Prescribed diuretic
agent
Not prescribed diuretic
agent
Relative risk
(unadjusted)
Relative risk
(adjusted for other
risk factors)
Electrocardiogram
Normal
Abnormal
2.3
5.9
2.0
1.5
1.13
3.99
0.95t
3.34t
Downloaded from http://hyper.ahajournals.org/ by guest on June 17, 2017
*Based on diuretic treatment status at annual visits and the
presence or absence of baseline resting electrocardiographic
abnormalities.
Rates are per 1,000 person-years. Other risk factors are age
(years), baseline diastolic blood pressure (mm Hg), baseline and
annual cigarette smoking (yes/no) and serum cholesterol (mg/dl),
adjusted by Cox proportional-hazards model.32
t/J<0.01 for the difference between these relative risks.
MRFIT, Multiple Risk Factor Intervention Trial.
stratified by electrocardiographicfindings(Table 3).32
Similar comparisons have also been made for a
"MRFIT-like" subgroup of the HDFP. Although the
results from the HDFP subgroup are not so strongly
suggestive of a baseline electrocardiogram-treatment interaction, they are not inconsistent with it.33
A curious finding in both trials was that electrocardiographic abnormalities in this subset of mild hypertensive subjects were not predictors of CHD mortality in control (usual-care and referred-care)
participants, but this factor should not affect the
interpretation of the between-group findings. However, in both studies, the number of CHD deaths
recorded was small, and consequently, they do not
provide conclusive evidence about differential treatment effects on CHD deaths in groups defined by the
presence or absence of resting electrocardiographic
abnormalities at baseline.
Two other trials have reported results on this
subgroup issue. In the Oslo trial, which showed an
overall adverse trend with treatment for a combination of fatal and nonfatal CHD events, results in
those subjects with baseline resting electrocardiographic abnormalities were apparently less favorable
TABLE 3.
Conclusions
This review summarizes the strong evidence
from randomized clinical trials that drug treatment
for mild-to-moderate hypertension confers important health benefits. In the aggregate, the data are
based on approximately 43,000 subjects who were
studied for an average of 5.6 years. The intervention regimens were based on thiazide or related
diuretic drugs or on propranolol, a nonselective
j8-blocker, administered in a fashion that produced
mean diastolic BP levels about 6 mm Hg below
that of randomized control subjects. Attributable
to this difference was an estimated 11% reduction
in total mortality, much of it due to the 38%
reduction in death from stroke. Nonfatal stroke
was similarly affected. Undoubtedly, part of the
benefit is also due to reduction of other hypertensive events such as heart failure that is so convincingly shown by the treatment trials in severe
hypertension.
The results for CHD suggest that treatment is
more likely than not to result in some benefit. The
evidence of benefit is, however, less than conclusive. Even when the trials are considered in combination, there is inadequate statistical power to
detect relatively modest treatment effects. Nevertheless, there is some evidence that the potentially
beneficial effects of BP reduction for CHD may be
offset to some degree by the adverse effects of
diuretic drugs on serum cholesterol and a possible
adverse effect of the thiazides in a subgroup of
patients with resting electrocardiographic abnor-
Mortality From Coronary Heart Disease for Subsets of MRFIT and HDFP Cohorts
Electrocardiogram
status
Normal
Abnormal
than for the study group as a whole.34 The MRC trial
has not yet published outcome data by treatment
group according to the presence or absence of any
resting electrocardiographic abnormalities at baseline. However, that study has published an analysis
relating morbid events to the most recent electrocardiographic recording, an analysis which tends to
confound the classification by electrocardiographic
status with any effects of the assigned drugs. For
only two classes of the Minnesota code, 1.1-1.2
(abnormal Q-QS) and 8.1 (ventricular ectopic arrhythmia), was there any evidence of higher CHD mortality in the bendrofluazide group than with placebo.35
MRFIT 1(hypertensive subjects)
Special intervention
Usual care
% Difference
20.7
15.8
-24t
(44)
(58)
17.7
29.2
+65t
(36)
(21)
HDFP (' 'MRFIT-like" subgroup)
Stepped care
Referred care
% Difference
10.0
-36*
15.6
(16)
(25)
21.4
+22*
17.5
(20)
(18)
*Based on randomized groups and the presence or absence of resting electrocardiographic abnormalities at baseline.
Rates are per 1,000 during 7 years (MRFIT) or 5 years (HDFP). Numbers of deaths are shown in parentheses.
Percent difference=([Special intervention-usual care]/usual care)x 100 or ([Stepped care-referred care]/referred care)x 100.
tp=0.08 for the difference between the corresponding relative risks.
tp=0.15 for the difference between the corresponding relative risks.
MRFIT, Multiple Risk Factor Intervention Trial and HDFP, Hypertension Detection and Follow-up Program.
Cutler et al Review of Drug Treatment Trials
Downloaded from http://hyper.ahajournals.org/ by guest on June 17, 2017
malities. On this basis, a case may be made for
alternate drug and nondrug treatments. There is
also a need to consider the feasibility of conducting trials to test whether such alternate treatment
strategies can reduce the incidence of CHD.
As impressive as these results are for stroke and
total mortality, they are not generally regarded as a
mandate to treat pharmacologically all individuals
who could have qualified for these trials. There is
the important consideration of absolute benefit of
treatment, which cannot be easily estimated from
trials due to subject selection and to the considerable use of active therapy in control groups. Rather,
information is required from more representative
populations about the risks of cardiovascular disease at different levels of blood pressure (for details,
see Browner and Hulley, this supplement).36 The
application of estimates of likely treatment effects
to these morbidity or mortality rates is likely to
provide the best estimate of the absolute effects of
treatment. Decisions about treatment must also, of
course, take into account the costs of treatment and
the side effects of antihypertensive drugs. It may be
concluded that treatment for hypertension is warranted, principally if the likely benefits of the
expected reduction in risks of vascular disease
warrant the costs of treatment and outweigh any
adverse effects of treatment on the quality of life.
Acknowledgment
The assistance of Colleen Brown in typing the
manuscript is hereby gratefully acknowledged.
References
1. MacMahon SW, Cutler JA, Furberg CD, Payne GH: The
effects of drug treatment for hypertension on morbidity and
mortality from cardiovascular disease: A review of randomized controlled trials. Prog Cardiovasc Dis 1986;29:99-118
2. Veterans Administration Cooperative Study on Antihypertensive Agents: A double-blind control study of antihypertensive agents. I. Comparative effectiveness of reserpine,
reserpine and hydralazide, and three ganglionic blocking
agents, chlorisondamine, mecamylamine, and pentolinium
tartrate. Arch Intern Med 1960;106:81-96
3. Veterans Administration Cooperative Study Group on Antihypertensive Agents: Effects of treatment on morbidity in
hypertension: Results in patients with diastolic blood pressures averaging 115 through 129 mm Hg. JAMA 1967;
202:1028-1034
4. Veterans Administration Cooperative Study Group on Antihypertensive Agents: Effects of treatment on morbidity in
hypertension. II. Results of patients with diastolic blood
pressure averaging 90 through 114 mm Hg. JAMA 1970;
213:1143-1151
5. Hamilton M, Thomson EN, Wisniewski TKM: The role of
blood pressure control in preventing complications of hypertension. Lancet 1964;l:235-238
6. Wolfe FW, Lindeman RD: Effects of treatment in hypertension: Results of a controlled study. J Chronic Dis 1966;
19:227-240
7. Barraclough M, Bainton D, Cochrane AL, Joy MD, Mac
Gregor GA, Foley TH, Lee MR, Rosendorff C, Holland
WW, Cranston WI, Rea JN, Greene J, Kilpatrick GS,
Wedell JM: Control of moderately raised blood pressure:
Report of a co-operative randomised controlled trial. Br Med
J 1973;3:434-436
1-43
8. Cutler JA, Furberg CD: Drug treatment trials in hypertension: A review. Prev Med 1985;14:499-518
9. Carter AB: Hypotensive therapy in stroke survivors. Lancet
1970; 1:485-489
10. Sprackling ME, Mitchell JRA, Short AH, Watt G: Blood
pressure reduction in the elderly: A randomised controlled
trial of methyldopa. Br Med J 1981;283:1151-1153
11. Coope J, Warrender TS: Randomised trial of treatment of
hypertension in elderly patients in primary care. Br Med J
1986;293:1145-1151
12. Smith WM, for the US Public Health Service Hospitals
Cooperative Study Group: Treatment of mild hypertension:
Results of a ten-year intervention trial. Circ Res 1977;
40(suppl I):I-98-I-105
13. Perry HM, Goldman AI, Lavin MA, Schnaper HW, Fitz
AE, Frohlich ED, Steele B, Rickman HG: Evaluation of
drug treatment in mild hypertension: VA-NHLBI feasibility
trial. Ann NY Acad Sci 1978,304:267-292
14. Helgeland A: Treatment of mild hypertension: A five year
controlled drug trial. Am J Med 1980;69:725-732
15. Report by the Management Committee: The Australian
therapeutic trial in mild hypertension. Lancet 1980;
1:1261-1267
16. Amery A, Birkenhager W, Brixko P, Bulpitt C, Clement D,
Deruyterre M, DeSchaepdryver A, Dollery C, Fagard R,
Forette F, Forte J, Hamdy R, Henry JF, Joossens JV,
Leonetti G, Lund-Johansen P, O'Malley K, Petrie J, Strasset T, Tuomilehto J, Williams B: Mortality and morbidity
results from the European Working Party on high blood
pressure in the elderly trial. Lancet 1985;1:1349-1354
17. Medical Research Council Working Party: MRC trial of
treatment of mild hypertension: Principal results. Br Med J
1985,291:97-104
18. Hypertension Detection and Follow-up Program Cooperative Group: Five-year findings of the Hypertension Detection and Follow-up Program. I. Reduction in mortality of
persons with high blood pressure, including mild hypertension. JAMA 1979;242:2562-2571
19. Hypertension Detection and Follow-up Program Cooperative Group: Five-year findings of the Hypertension Detection and Follow-up Program. III. Reduction in stroke incidence among persons with high blood pressure. JAMA 1982;
247:633-638
20. Hypertension Detection and Follow-up Program Cooperative Group: Five-year findings of the Hypertension Detection and Follow-up Program. V. Effects of stepped care
treatment on the incidence of myocardial infarction and
angina pectoris. Hypertension 1984;6(suppl I):I-198—1-206
21. Multiple Risk Factor Intervention Trial Research Group:
Multiple Risk Factor Intervention Trial: Risk factor changes
and mortality results. JAMA 1982;248:1465-1477
22. Multiple Risk Factor Intervention Trial Research Group:
Coronary heart disease death, nonfatal acute myocardial
infarction and other clinical outcomes in the Multiple Risk
Factor Intervention Trial. Am J Cardiol 1986;58:1-13
23. Furberg CD, Cutler JA: Diuretic agents versus /3-blockers:
Comparison of effects on mortality, stroke, and coronary
heart disease. Hypertension 1989;13(suppl I):I-57—1-61
24. MacMahon S, Cutler JA, Stamler J: Antihypertensive drug
treatment: Potential, expected, and observed effects on
stroke and on coronary heart disease. Hypertension 1989;
13(suppl I):I-45-I-50
25. Yusuf S, Peto R, Lewis J, Collins R, Sleight P: /3-Blockade
during and after myocardial infarction: An overview of the
randomized trials. Prog Cardiovasc Dis 1985;27:335-371
26. Hypertension-Stroke Cooperative Study Group: Effect of
antihypertensive treatment on stroke recurrence. JAMA
1974;229:409-417
27. Morgan TO, Adams WR, Hodgson M, Gibberd RW: Failure
of therapy to improve prognosis in elderly males with
hypertension. Med J Aust 1980;2:27-31
28. Medical Research Council Working Party on Mild to Moderate Hypertension: Adverse reactions to bendrofluazide
1-44
29.
30.
31.
32.
33.
Supplement I Hypertension
Vol 13, No 5, May 1989
and propranolol for the treatment of mild hypertension.
Lancet 1981 ;2:539-543
Weidmann P, Gerber A, Mardasini R: Effects of antihypertensive therapy on serum lipids. Hypertension 1983;
5(suppl I1I):III-12O-III-131
Peto R, Yusuf S, Collins R: Cholesterol-lowering trial results
in their epidemiological context (abstract). Circulation 1985;
72(suppl III):III-451
Holland OB, Nixon J, Kuhuert L: Diuretic-induced ventricular ectopic activity. Am J Med 1981;70:762-768
Multiple Risk Factor Intervention Trial Research Group:
Baseline rest electrocardiographic abnormalities, antihypertensive treatment, and mortality in the Multiple Risk Factor
Intervention Trial. Am J Cardiol 1985;55:1-15
Hypertension Detection and Follow-up Program Cooperative Research Group: The effect of antihypertensive drug
treatment on mortality in the presence of resting electrocardiographic abnormalities at baseline: The HDFP experience.
Circulation 1984;70:996-1003
34. Holme I, Helgeland A, Hjermann I: Treatment of mild
hypertension with diuretics: The importance of ECG abnormalities in the Oslo study and in MRFIT. JAMA 1984;
251:1298-1299
35. Miall WE, Greenberg G: Mild Hypertension: Is There Pressure to Treat? New York, Cambridge University Press, 1987
36. Browner WS, Hulley SB: Effect of risk status on treatment
criteria: Implications of hypertension trials. Hypertension
1989;13(suppl I):I-51-1-56
KEY WORDS • antihypertensive agents • clinical trials
prevention • coronary heart disease • stroke
Downloaded from http://hyper.ahajournals.org/ by guest on June 17, 2017
Controlled clinical trials of drug treatment for hypertension. A review.
J A Cutler, S W MacMahon and C D Furberg
Hypertension. 1989;13:I36
doi: 10.1161/01.HYP.13.5_Suppl.I36
Downloaded from http://hyper.ahajournals.org/ by guest on June 17, 2017
Hypertension is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 1989 American Heart Association, Inc. All rights reserved.
Print ISSN: 0194-911X. Online ISSN: 1524-4563
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://hyper.ahajournals.org/content/13/5_Suppl/I36
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in
Hypertension can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial
Office. Once the online version of the published article for which permission is being requested is located, click
Request Permissions in the middle column of the Web page under Services. Further information about this
process is available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Hypertension is online at:
http://hyper.ahajournals.org//subscriptions/