Вы находитесь на странице: 1из 11

CLINICAL RESEARCH STUDY

Combination Therapy Versus Monotherapy in Reducing Blood


Pressure: Meta-analysis on 11,000 Participants from 42 Trials
David S. Wald, MD, Malcolm Law, FRCP, Joan K. Morris, PhD, Jonathan P. Bestwick, MSc, Nicholas J. Wald, FRS
Wolfson Institute of Preventive Medicine at Barts and The London Queen Mary’s School of Medicine and Dentistry, Charterhouse
Square, London, United Kingdom.

ABSTRACT

OBJECTIVE: To quantify the incremental effect of combining blood pressure-lowering drugs from any 2
classes of thiazides, beta-blockers, angiotensin-converting enzyme inhibitors, and calcium channel block-
ers over 1 drug alone and to compare the effects of combining drugs with doubling dose.
METHODS: Meta-analysis of factorial trials in which participants were randomly allocated to 1 drug alone,
another drug alone, both drugs together, or a placebo.
RESULTS: We identified 42 trials (10,968 participants). With a thiazide used alone, the mean placebo-
subtracted reduction in systolic blood pressure was 7.3 mm Hg and 14.6 mm Hg combined with a drug
from another class. The corresponding reductions were 9.3 mm Hg and 18.9 mm Hg with a beta-blocker,
6.8 mm Hg and 13.9 mm Hg with an angiotensin-converting enzyme, and 8.4 mm Hg and 14.3 mm Hg
with a calcium channel blocker. The expected blood pressure reduction from 2 drugs together, assuming
an additive effect, closely predicted the observed blood pressure reductions. The ratios of the observed to
expected incremental blood pressure reductions from combining each class of drug with any other over that
from 1 drug were, respectively, for thiazides, beta-blockers, angiotensin-converting enzyme inhibitors, and
calcium channel blockers: 1.04 (95% confidence interval [CI], 0.88-1.20), 1.00 (95% CI, 0.76-1.24), 1.16
(95% CI, 0.93-1.39), and 0.89 (95% CI, 0.69-1.09); the overall average was 1.01 (95% CI, 0.90-1.12).
Comparison of our results with those of a published meta-analysis of different doses of the same drug
showed that doubling the dose of 1 drug had approximately one fifth of the equivalent incremental effect
(0.22 [95% CI, 0.19-0.25]).
CONCLUSION: Blood pressure reduction from combining drugs from these 4 classes can be predicted on
the basis of additive effects. The extra blood pressure reduction from combining drugs from 2 different
classes is approximately 5 times greater than doubling the dose of 1 drug.
© 2009 Elsevier Inc. All rights reserved. • The American Journal of Medicine (2009) 122, 290-300

KEYWORDS: Angiotensin-converting enzyme inhibitor; Beta-blocker; Blood pressure; Calcium channel blocker;
Combination blood pressure therapy; Randomized trial; Thiazide

Monotherapy is the standard initial treatment for reducing classes) when stepwise increases in the dose of 1 drug fail
blood pressure in most patients with hypertension, moving to achieve the desired decrease in blood pressure.1-4 A
to combination therapy (2 or more drugs from different meta-analysis published in 2003 showed that halving the
dose of most blood pressure-lowering drugs substantially
Funding: None. reduced the prevalence of adverse effects but reduced the
Conflict of Interest: Nicholas J. Wald and Malcolm Law hold patents blood pressure-lowering effect by only approximately
(EU1272220 and GB2361186) for a combination pill for the prevention of
cardiovascular disease (Polypill) and together with David Wald have in- 20%,5 supporting proposals for the use of low-dose drug
terests in its development. combinations as the first-line treatment for the control of
Authorship: All authors had access to the data and played a role in blood pressure.5-8
writing this manuscript. The effectiveness of this approach relies on there being
Requests for reprints should be addressed to David S. Wald, MD,
Wolfson Institute of Preventive Medicine, Charterhouse Square, London
additive effects between the different classes of drugs when
EC1M, 6BQ, United Kingdom. used together, such that the combined blood pressure-low-
E-mail address: d.s.wald@qmul.ac.uk ering effect of 2 together is the sum of each alone. Ran-

0002-9343/$ -see front matter © 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.amjmed.2008.09.038
Wald et al Combination Versus Monotherapy for Blood Pressure Reduction 291

domized trials of factorial design are required to quantify the average reduction from the comparison drugs, and the 2
the effect of giving 2 drugs together, using 4 groups with 1 drugs together. For example, in all randomized trials that
drug alone, the other drug alone, both drugs together, and considered pairwise comparisons of thiazides and another
placebo. Such trials have been published on each of the 4 most class of drug, we calculated the mean placebo-adjusted
widely used classes of drugs (thiazides, beta-blockers, angio- reduction in blood pressure on thiazides alone, on the com-
tensin-converting enzyme [ACE] parison drugs alone, and on both
inhibitors, and calcium channel drugs together. We specified
blockers).9 We examine the evi- equivalent doses of different drugs
CLINICAL SIGNIFICANCE
dence for additive effects of all by identifying the usual mainte-
pairwise combinations on a class- ● Monotherapy is the standard initial nance dose of each drug as recom-
specific basis. For each of the 4 treatment for reducing blood pressure, mended in reference pharmaco-
classes of drug we sought to quan- with stepwise increases in dose if the poeias.5,52 We referred to this as
tify the incremental blood pres- the “standard dose” and expressed
desired decrease in blood pressure is
sure-lowering effect of using any the dose of each drug in each trial
1 class of drug in combination
not achieved. as a multiple of the standard dose.
with another class and to assess ● Combining drugs from different classes Meta-regression analyses of blood
the efficacy of combinations com- is approximately 5 times more effective pressure reduction on the stan-
pared with using 1 drug in double in lowering blood pressure than increas- dardized doses of the drugs were
dose. ing the dose of 1 drug. used to investigate possible
sources of heterogeneity. STATA
● Combination therapy is the preferred software was used (StataCorp,
MATERIALS AND METHODS initial strategy in the treatment of high College Station, Tex).
Randomized trials using a facto- blood pressure. For each class of drug the ob-
rial design were identified using a served blood pressure reduction
search of Medline, Cochrane Col- on the combination was compared
laboration, and EMBASE data- with the expected blood pressure
bases in English from 1966 to March 2008. We used generic reduction based on the effect of both drugs together being
and trade names of all drugs in the 4 classes, thiazide, additive. Because the blood pressure-lowering effect of a
beta-blocker, ACE inhibitor, and calcium channel blocker given dose of drug depends on pretreatment blood pressure,
taken from reference pharmacopoeias as key or text words the expected blood pressure reduction from 2 drugs was the
and combined them in pairs. The resulting citations were sum of each drug alone allowing for the smaller reduction
limited to those of Medline publication type “clinical trial” from an added drug because of the decreased blood pressure
or “randomized-controlled trial.” We excluded trials under from the initial drug.52 For example, if 2 drugs, A and B,
2 weeks duration, with no placebo group or with a nonran- lower systolic blood pressure by a mm Hg and b mm Hg,
domized order of treatment and placebo. These exclusions respectively, from a given pretreatment blood pressure (z
apart, we included all randomized placebo-controlled trials mm Hg) when used alone, the expected effect of both
comparing any drugs of 2 of the 4 main classes specified together is less than (a ⫹ b) mm Hg because drug B effectively
above. The initial search identified 1697 articles, which was operates from a pretreatment blood pressure that is a mm Hg
reduced to 778 after screening the title, to 92 after inspec-
lower than z mm Hg, as a result of drug A. A previous
tion of the abstracts, and to 4210-51 after examining the full
meta-analysis5 showed that the blood pressure-lowering effect
articles, including a hand search of citations in the reports of of a drug is approximately 1 mm Hg less for each 10 mm Hg
published trials and systematic reviews. We also undertook decrement in pretreatment blood pressure. So, the expected
a search of Food and Drug Administration and Industry blood pressure reduction due to A plus B, is (a ⫹ b ⫺ a ⫻ 0.1)
websites but identified no additional trials that met the mm Hg, taking A as the initial drug and (b ⫹ a ⫺ b ⫻ 0.1) mm
inclusion criteria. Data were abstracted independently by 2 Hg, taking B as the initial drug; the average is a ⫹ b ⫺
investigators, and any inconsistencies were resolved by dis- 0.1 ⫻ (a ⫹ b)/2 or 0.95 (a ⫹ b) mm Hg. The expected effect of
cussion and referral back to the original articles. 2 classes of drug taken together is therefore 95% of the sum of
the blood pressure reductions for each class of drug taken
Statistical Analysis alone.
We calculated the mean blood pressure reductions in each We also calculated the observed incremental blood pres-
trial as the reduction in the treated group minus that in the sure reduction from 2 classes of drug together relative to 1
placebo group or, in crossover trials, end-treatment minus drug alone and divided this by the expected incremental
end-placebo blood pressure (with its standard error) for each effect, to give a ratio of observed to expected incremental
drug taken separately and for both drugs taken together. For effects. For each trial, the observed incremental blood pres-
each of the 4 classes of drug in turn, we used a random sure reductions from 2 classes of drug together (A and B)
effects model to estimate the average placebo-adjusted compared with the reduction from 1 alone was calculated by
blood pressure reduction from the specified class of drug, subtracting the average blood pressure reduction for each
292 The American Journal of Medicine, Vol 122, No 3, March 2009

drug alone 0.5 (a ⫹ b) mm Hg from the reduction due to the Figure 2 shows, for each of the 4 classes of drug com-
combination of 2 drugs. The expected incremental blood bined with any other, the ratio of observed to expected
pressure reduction from 2 classes of drug compared with 1 incremental systolic blood pressure-lowering effects of 2
alone was the expected blood pressure reduction due to the drugs compared with 1 drug alone. For each class of drug
combination (0.95 (a ⫹ b) mm Hg, as derived above) minus the effect of adding a second drug was close to that ex-
the average effect of each drug alone 0.5 (a ⫹ b) mm Hg. pected, that is, a ratio of 1.0. The estimates for each com-
The ratios of observed to expected incremental effects had bination were thiazide plus any other class 1.04 (0.88-1.20),
a log Gaussian distribution, so a weighted geometric mean beta-blocker plus any other class 1.0 (0.76-1.24), ACE in-
was calculated for each class of drug by weighting the ratios hibitor plus any other class 1.16 (0.93-1.39), and calcium
for each trial by the number of participants allocated to each channel blocker plus any other class 0.89 (0.69-1.09). The
treatment in each study. average across all classes was 1.01 (0.90-1.12). Using 1
We compared the ratio of observed to expected incre- drug in double dose achieved ratios, respectively, of 0.19
mental blood pressure reductions with the equivalent effects (0.08-0.30), 0.23 (0.12-0.34), 0.20 (0.14-0.26), and 0.37
of doubling the dose of each class of drug by using the (0.29-0.45) for thiazides, beta-blockers, ACE inhibitors, and
results of a previous meta-analysis that examined the blood calcium channel blockers, respectively, an average of 0.22
pressure-lowering effects of different classes of drugs at (0.19-0.25). In every instance, combination therapy was
more effective than increasing the dose of 1 drug, and this
fixed dose.5 For each of the 4 classes of drug, the observed
was statistically significant (P ⬍.05) for all comparisons.
incremental effect of doubling dose was calculated by sub-
The mean doses of the drugs in the trials were close to
tracting the blood pressure reduction at twice the standard
the standard (or usual maintenance) dose,52 ranging be-
dose from that at the standard dose.5 The expected effect of
tween 0.5 and 1.6 multiples of standard. There was evidence
doubling dose (eg, of drug A), assuming an additive effect,
of heterogeneity of blood pressure-lowering effects across
was double the blood pressure reduction from using the
the individual trials of thiazides (P ⬍.01), beta-blockers
drug at standard dose (a mm Hg), allowing for the effect of (P ⫽ .06), ACE inhibitors (P ⫽ .08), and calcium channel
the lower pretreatment blood pressure, as described above blockers (P ⬍.01), which was largely explained by the dif-
(ie, 0.95 ⫻ 2a mm Hg). The expected incremental effect was ferent doses of drugs used in the trials. A meta-regression
therefore 0.95 ⫻ 2a – a, or 0.9a mm Hg. For each of the 4 analysis of blood pressure reduction on dose (with all trials
classes of drug, the ratio of the observed to the expected of a given class of drug stratified according to the dose used)
incremental effect was calculated and compared with the showed that the heterogeneity was no longer present
incremental effects of combining each class of drug with (P ⬎.05 for all classes of drug).
any other class. Figure 3 shows, for each of the 4 classes of drug, a plot
of the observed placebo-subtracted blood pressure reduction
RESULTS for that class of drug combined with any other drug com-
pared with that expected, assuming an additive interaction
Table 1 shows details of the 42 randomized factorial trials
and allowing for the blood pressure reduction due to the
included, involving 101 comparisons between pairs of drugs
initial drug. Each circle represents a different 2-drug com-
(some trials compared 2 drugs in different doses) and
bination at the doses used. The area of the circle reflects the
10,698 participants (10,333 in parallel group design trials
statistical precision of the points plotted. For each class of
and 365 in crossover trials). All but 1 trial (conducted in
drug, the circles are on or close to the line of identity (where
general practice) recruited patients attending hospital out-
observed equals expected), revealing a consistent effect
patient hypertension clinics, generally without a history of
across all studies and across doses of drug within trials.
coronary heart disease, stroke, diabetes, or renal disease. In
the individual trials, the duration ranged between 4 and 12
weeks, mean age was between 46 and 71 years, and pre- DISCUSSION
treatment blood pressure was between 136 and 173 mm Hg The results from this meta-analysis show that for each of the
systolic and 84 and 110 mm Hg diastolic. 4 classes of blood pressure-lowering drug considered, the
Figure 1 shows the mean (and 95% confidence intervals) blood pressure reduction from each class of drug combined
placebo-subtracted systolic blood pressure reductions ob- with 1 from another class is approximately additive. The
served in the trials for each of the 4 drug classes alone, for additional effect of combining given doses of 2 classes of
the comparison drug alone (from any of the other 3 classes), drug is approximately 5 times more effective than doubling
and for 2 drugs together. The expected blood pressure re- the dose of 1 drug.
duction from both drugs together (assuming an additive The incremental effect of an additional drug was ex-
interaction) is shown by the upper dotted line in Figure 1 for pressed as the ratio of the observed to expected extra blood
each of the 4 drug classes. The observed and expected pressure reduction. The latter is the sum of the reductions
effects of both drugs together are close, showing that the from each drug allowing for the reduced effect of the added
average effect of combining each class of drug with a drug drug due to the lower blood pressure achieved by the ex-
from another class is approximately additive. isting drug. This is needed to take account of a drug at a
Wald et al
Table 1 Details of the 42 Trials Included in the Meta-analysis

Combination Versus Monotherapy for Blood Pressure Reduction


Mean Pretreatment
Blood Pressure
(mm Hg)
Combination of Drugs Year of Number of Treatment Duration Dose
First Author Publication Participants (Weeks) Mean Age Trial Design Drug (Multiple of Standard) Systolic Diastolic
Thiazide ⫹ beta-blocker
Bateman10 1979 15 4 54 Cross-over chlorthalidone 1
155 104
atenolol 2
11
Chalmers 1976 16 8 45 Cross-over HCTZ
2 163 106
pindolol
12
Chalmers 1976 20 8 44 Cross-over HCTZ 2
159 99
timolol 3
13
Chalmers 1982 16 8 53 Cross-over indapamide 1
164 94
pindolol 0.67
14
Chrysant 1992 256 4 55 Parallel group HCTZ 0.5
atenolol 0.5 148 97
atenolol 1
Durel15 1992 5 4 47 Cross-over chlorthalidone
2 147 93
atenolol
16
Erwteman 1984 50 4 46 Cross-over chlorthalidone 1
143 94
metoprolol 2
Frishman17 1994 512 4 53 Parallel group HCTZ 0.25
HCTZ 1
bisoprolol 0.25 148 97
bisoprolol 1
bisoprolol 4
La Courciere18 1994 240 12 52 Parallel group HCTZ 0.5
HCTZ 1
147 96
nebivolol 0.2
nebivolol 2
Zachariah19 1993 1059 4 47 Parallel group HCTZ
0.25
bisoprolol 149 95
bisoprolol 1
Thiazide ⫹ ACE inhibitor
Brown20 1990 40 4 58 Parallel group HCTZ** 1
170 99
perindopril 1

293
294
Table 1 Continued

Mean Pretreatment
Blood Pressure
(mm Hg)
Combination of Drugs Year of Number of Treatment Duration Dose
First Author Publication Participants (Weeks) Mean Age Trial Design Drug (Multiple of Standard) Systolic Diastolic
Canter21 1994 460 8 53 Parallel group HCTZ 0.25
HCTZ 0.5
HCTZ 1
159 102
quinapril 0.125
quinapril 0.5
quinapril 2
Chalmers22 1986 21 4 61 Cross-over HCTZ 2
173 92
enalapril 2
Chrysant23 1994 505 8 53 Parallel group HCTZ 0.5
HCTZ 148 98
1
lisinopril

The American Journal of Medicine, Vol 122, No 3, March 2009


Chrysant24 1996 334 6 53 Parallel group HCTZ
1 159 98
benazepril
25
Fernandez 1994 67 8 53 Parallel group HCTZ 0.5
144 93
fosinopril 2
26
Kayanikis 1987 211 8 54 Parallel group HCTZ 1
161 93
captopril 1
27
Pool 1997 550 8 52 Parallel group HCTZ 0.2
HCTZ 0.5
HCTZ 1.5
147 96
fosinopril 0.25
fosinopril 1
fosinopril 4
Pordy28 1994 1162 4 54 Parallel group HCTZ 0.5
HCTZ 1
cilazapril 0.2 144 95
cilazapril 2
cilazapril 4
Wald et al
Table 1 Continued

Mean Pretreatment
Blood Pressure
(mm Hg)

Combination Versus Monotherapy for Blood Pressure Reduction


Combination of Drugs Year of Number of Treatment Duration Dose
First Author Publication Participants (Weeks) Mean Age Trial Design Drug (Multiple of Standard) Systolic Diastolic
Scholze29 1993 534 6 48 Parallel group HCTZ 0.5
HCTZ
1
ramipril 159 104
ramipril 2
ramipril 4
Thiazide ⫹ calcium-channel blocker
Burris30 1990 297 6 52 Parallel group HCTZ 0.5
HCTZ 1
HCTZ 2
diltiazem 0.5 152 99
diltiazem 0.75
diltiazem 1
diltiazem 1.5
Pool31 1993 298 6 53 Parallel group HCTZ
1 150 95
diltiazem
32
Salvetti 1991 66 4 55 Cross-over chlorthalidone
1 158 97
nifedipine
33
Weir 1992 298 4 54 Parallel group HCTZ 0.5
HCTZ 1
diltiazem 0.5 149 95
diltiazem 0.75
diltiazem 1
Wing34 1997 19 4 71 Cross-over HCTZ
1 163 85
4 71 lacidipine
Beta-blocker ⫹ ACE inhibitor
Wald35 2008 47 4 62 Cross-over atenolol
0.5 136 84
lisinopril
36
Wing 1988 16 4 59 Cross-over atenolol
1 171 97
enalapril
Beta-blocker ⫹ calcium-channel blocker
Clement37 1987 17 3 55 Cross-over atenolol 2
152 94
nifedipine 1
Dargie38 1986 14 4 52 Cross-over propranolol
1.5 157 107
verapamil
39
Lyons 1994 12 2 52 Cross-over atenolol 2
154 102
lacidipine 1

295
296
Table 1 Continued

Mean Pretreatment
Blood Pressure
(mm Hg)
Combination of Drugs Year of Number of Treatment Duration Dose
First Author Publication Participants (Weeks) Mean Age Trial Design Drug (Multiple of Standard) Systolic Diastolic
40
Maclean 1990 136 6 52 Parallel group atenolol
1 173 110
nitrendipi
41
McInnes 1985 14 4 51 Cross-over propranolol
1.5 158 107
verapamil
42
Tonkin 1990 17 4 61 Cross-over atenolol
1 181 102
diltiazem
Calcium-channel blocker ⫹ ACE
inhibitor
Chan43 1997 156 12 71 Parallel group lisinopril 1
diltiazem 0.5 167 104
diltiazem 1
Cushman44 1998 891 12 56 Parallel group enalapril
0.5
diltiazem 156 98
diltiazem 0.75
Frishman45 1995 401 8 54 Parallel group benazepril 0.5

The American Journal of Medicine, Vol 122, No 3, March 2009


157 101
amlodipine 0.5
46
Kuschnir 1996 30 8 56 Parallel group benazepril
1 164 103
amlodipine
47
Levine 1995 186 4 56 Parallel group enalapril 1
verapamil 0.5 153 100
verapamil 1
Messerli48 1998 631 6 54 Parallel group trandolopril 4
153 100
verapamil 1
49
Scholze 1998 456 6 55 Parallel group trandolopril 0.5
trandolopril 1
trandolopril 2 163 104
verapamil 0.5
verapamil 0.75
Scholze50 1999 507 6 50 Parallel group ramipril 1
ramipril 2
ramipril 4 155 99
felodipine 1
felodipine 2
Veratran51 1997 272 8 51 Parallel group trandolopril 1
149 97
verapamil 0.75
*ACE-I ⫽ ACE inhibitor; BB ⫽ beta blocker; CCB ⫽ calcium-channel blocker; **HCTZ ⫽ hydrochlorothiazid.
Wald et al Combination Versus Monotherapy for Blood Pressure Reduction 297

Figure 1 Mean placebo-subtracted systolic blood pressure reduction from a meta-analysis of


42 randomized factorial trials of thiazides, beta-blockers, ACE inhibitors, or calcium channel
blockers using each class of drug separately, any 1 of the other 3 classes alone, and in
combination with the specified drug class (95% confidence interval). The dashed line represents
the expected blood pressure reduction from the combination assuming an additive effect,
allowing for the smaller reduction from 1 drug given the lower pretreatment blood pressure
because of the other. BP ⫽ blood pressure; ACE ⫽ angiotensin-converting enzyme.

given dose having a smaller blood pressure-lowering effect additive effect), and 1.5 indicates a supra-additive (or syn-
in a person with a lower blood pressure than in a person ergistic) effect (equivalent to 50% greater than additive).
with a higher blood pressure.52 An incremental effect of 1.0 Overall, our result of 1.01 (the average of the summary
thus indicates that the effect is exactly additive, 0.5 indi- estimates from each class of drug) is close to the effect
cates a subadditive effect (equivalent to 50% of the extra being exactly additive.

Figure 2 Ratio of observed to expected incremental blood pressure-lowering effects of


adding a drug or doubling the dose according to the class of drug. The expected incremental
effect is the incremental blood pressure reduction of the added (or doubled drug), assuming
an additive effect and allowing for the smaller reduction from 1 drug (or dose of 1 drug)
given the lower pretreatment blood pressure because of the other. ACE ⫽ angiotensin-
converting enzyme.
298 The American Journal of Medicine, Vol 122, No 3, March 2009

systolic or ⱖ90 mm Hg diastolic) and in patients with


specific indications (eg, diabetes or a myocardial infarc-
tion). Although the value of routinely starting treatment
with combination therapy, particularly with low doses, has
been proposed,5-8 this has not been widely accepted. No
guideline recommends combination rather than mono-
therapy as a matter of routine in all patients. The substantial
advantage of this approach, over increasing dose, is clear
from the results presented here, based on many studies,
across different doses and pretreatment blood pressure lev-
els. The results leave little doubt over the advantages of
adopting low-dose combination blood pressure-lowering
treatment as routine initial therapy for all, instead of a
monotherapy and stepped-care approach.
A single blood pressure-lowering drug at standard dose
reduces diastolic blood pressure by approximately 5 mm Hg,5
equivalent to approximately a 25% reduction in risk of cor-
onary heart disease events (relative risk 0.75) and approxi-
mately a 35% reduction in stroke (relative risk 0.65), at age
65 years, from a meta-analysis of 61 cohort studies sup-
ported by a meta-analysis of 147 randomized trials.57,58 Our
Figure 3 Comparison of observed placebo-subtracted reduc- results indicate that doubling the dose of a single drug
tion in systolic blood pressure of 2 drugs together against the would increase the blood pressure reduction from approxi-
expected effect of 2 drugs together according to the class of drug
mately 5 to 6 mm Hg, which would reduce coronary heart
in factorial trials of 2 blood pressure-lowering drugs (the expected
effect assumes a full additive effect and allows for the smaller
disease events by 29% (because 0.756/5 ⫽ 0.71), an addi-
reduction from 1 drug given the lower pretreatment blood pressure tional 4 percentage points, and reduce stroke by 40%
because of the other). Each circle represents a different drug and (0.656/5 ⫽ 0.60), an additional 5 percentage points. Com-
dose combination within a trial. There are more circles than trials bining 2 drugs from different classes would increase the
because several trials examine different doses of different drugs. blood pressure reduction from approximately 5 to 9 mm Hg,
The area of the circle reflects the statistical precision of the points which would reduce coronary heart disease events by 40%
plotted. The diagonal lines are the lines of identity where observed (0.759/5 ⫽ 0.56), an additional 15 percentage points, and
equals expected. ACE ⫽ angiotensin-converting enzyme. reduce stroke by 54% (0.659/5 ⫽ 0.46), an additional 19
percentage points. This means that for every 1 incremental
coronary heart disease event or stroke prevented by dou-
This analysis combined data from trials of all possible bling the dose of a single drug, 4 events would be prevented
pairwise combinations of the 4 most widely used classes of by using combination therapy.
blood pressure-lowering drugs, allowing the additive effects of Low-dose therapy has the advantage of reducing adverse
each class of drug to be quantified on a class-specific basis. effects that, with the exception of ACE inhibitors and an-
Angiotensin-II receptor blockers were not included in the giotensin receptor blockers, are strongly dose related; for 2
meta-analysis because being a newer class of drug, there classes (thiazides and calcium channel blockers), for exam-
are few factorial trials and they do not encompass all the ple, adverse effects are 80% lower at half-standard than
pairwise comparisons. Nonetheless, 3 published trials showed standard dose.5 The prevalence of adverse effects from
additive effects in combination with thiazides53-55 and 1 pub- combining 2 drugs at half-standard dose would therefore,
lished trial showed additive effects in combination with cal- for most combinations, be lower than with 1 drug at stan-
cium channel blockers.56 dard dose. Using more than 2 drugs in combination also
“Monotherapy” and “stepped-care” is the usual initial would increase efficacy; 3 drugs at half-standard dose (com-
approach to treating blood pressure in most patients with pared with 2 at standard dose) would, for example, reduce
hypertension, in which a trial of treatment is started in each diastolic blood pressure by approximately another 2 mm Hg
patient, increasing the dose of 1 drug before adding others (from 9 to 11 mm Hg) with expected reductions in the risk
if specified blood pressure “targets” are not reached. The of coronary heart disease and stroke of 46% and 63%,
British BHS 2004 and NICE/BHS 2006 clinical practice respectively. The use of combination low-dose therapy
guidelines recommend this as the initial approach in all therefore has greater efficacy and less toxicity than using a
patients.1,2 The American Joint National Committee VII3 higher dose of a single drug.
and the European Society of Cardiology/European Society There may be concerns that such an approach may lower
of Hypertension 20074 guidelines also advocate this as the blood pressure below the so-called blood pressure “targets”
general approach, but recommend using 2 drugs initially if often regarded as optimal.1,2 The evidence, however, is
a person’s blood pressure is particularly high (ⱖ160 mm Hg against the view that there is some target blood pressure
Wald et al Combination Versus Monotherapy for Blood Pressure Reduction 299

level within the range of values in Western populations 12. Chalmers J, Horvath J, Tiller D, Bune A. Effects of timolol and
below which further blood pressure reduction has no further hydrochlorothiazide on blood-pressure and plasma renin activity. Lan-
cet. 1976;2:328-331.
effect in preventing cardiovascular disease. Epidemiologic 13. Chalmers JP, Wing LMH, Grygiel JJ, et al. Effects of once daily
studies show a continuous proportionate reduction in risk of indapamide and pindolol on blood pressure, plasma aldosterone con-
heart disease and stroke with decreasing blood pressure, centration and plasma renin activity in a general practice setting. Eur
without threshold.57 Over time, patients entered into trials J Clin Pharmacol. 1982;22:191-196.
of blood pressure reduction have been selected with lower 14. Chrysant SG, Chappel C, Farnham J, et al. Antihypertensive and
metabolic effects of single and combined atenolol regimens. J Clin
and lower blood pressures and the trials have shown no Pharmacol. 1992;32:61-65.
attenuation of the relative reduction in disease events,59-63 15. Durel LA, Hayashi PJ, Weidler DJ, Schneiderman N. Effectiveness of
as expected from the epidemiologic studies.5 Setting antihypertensive medications in office and ambulatory settings: a pla-
blood pressure targets needlessly limits the potential for cebo-controlled comparison of atenolol, metoprolol, chlorthalidone,
preventing heart attacks and strokes through blood pres- verapamil, and an atenolol-chlorthalidone combination. J Clin Phar-
macol. 1992;32:564-570.
sure reduction.
16. Erwteman TM, Nagelkerke N, Lubsen J, et al. b Blockade, diuretics,
and salt restriction for the management of mild hypertension: a ran-
domised double blind trial. BMJ. 1984;289:406-409.
CONCLUSIONS 17. Frishman WH, Burris JF, Mroczek WJ, et al. First-line therapy option
Combining blood pressure-lowering drugs from different with low-dose bisoprolol fumarate and low-dose hydrochlorothiazide
classes is approximately 5 times more effective than dou- in patients with stage I and stage II systemic hypertension. J Clin
bling the dose of 1 drug. It follows that to maximize efficacy Pharmacol. 1995;35:182-188.
combination therapy, preferably using low doses to mini- 18. Lacourcière Y, Arnott W. Placebo-controlled comparison of the effects
of nebivolol and low-dose hydrochlorothiazide as monotherapies and
mize side effects, is substantially better than monotherapy in combination on blood pressure and lipid profile in hypertensive
and should be considered as routine initial therapy. patients. J Hum Hypertens. 1994;8:283-288.
19. Zachariah PK, Messerli FH, Mroczek W. Low-dose bisoprolol/hydro-
chlorothiazide: an option in first-line, antihypertensive treatment. Clin
Ther. 1993;15:779-787.
20. Brown CL, Backhouse CI, Grippat JC, Santoni JP. The effect of
ACKNOWLEDGMENTS perindopril and hydrochlorothiazide alone and in combination on
We thank Andrew Archbold, Jan Mackie, and Mark Caul- blood pressure and on the renin-angiotensin system in hypertensive
subjects. Eur J Clin Pharmacol. 1990;39:327-332.
field for comments on drafts of this article. 21. Canter D, Frank GJ, Knapp LE, et al. Quinapril and hydrochlorothi-
azide combination for control of hypertension: assessment by factorial
References design. J Hum Hypertens. 1994;8:155-162.
1. Williams B, Poulter NR, Brown MJ, et al. Guidelines for the management 22. Chalmers JP, Morris MJ, Wing LMH, et al. Effects of enalapril and
of hypertension: report of the fourth working party of the British Hyper- hydrochlorothiazide on blood pressure, renin-angiotensin system, and
tension Society, 2004 – BHSIV. J Hum Hypertens. 2004;18:139-185. atrial natriuretic factor in essential hypertension: a double blind fac-
2. CG18 Hypertension (persistently high blood pressure) in adults— torial cross-over study. Aust NZ J Med. 1986;16:475-480.
NICE guideline. Available at: www.nice.org.uk. 23. Chrysant SG. Antihypertensive effectiveness of low-dose lisinopril-
3. Chobanian AV, Bakris GL, Black HR, et al. The Seventh Report of the hydrochlorothiazide combination. Arch Intern Med. 1994;154:737-
Joint National Committee on Prevention, Detection, Evaluation, and 743.
Treatment of High Blood Pressure. The JNC 7 Report. JAMA. 2003; 24. Chrysant SG, Fagan T, Glazer R, Kriegman A. Effects of benazepril
289:2560-2573. and hydrochlorothiazide, given alone and in low- and high-dose com-
4. The Taskforce for the management of Arterial Hypertension of the binations, on blood pressure in patients with hypertension. Arch Fam
European Society of Hypertension (ESH) and of the European Society Med. 1996;5:17-24.
of Cardiology (ESC). Guidelines for the Management of Arterial 25. Fernández M, Madero R, González D, et al. Combined versus single
Hypertension. J Hypertens. 2007;25:1105-1187. effect of fosinopril and hydrochlorothiazide in hypertensive patients.
5. Law MR, Wald NJ, Morris JK, Jordan RE. Value of low dose com- Hypertension. 1994;23(Suppl I):I-207-210.
bination treatment with blood pressure lowering drugs: analysis of 354 26. Kayanakis JG, Baulac L. Comparative study of once-daily adminis-
randomised trials. BMJ. 2003;326:1427-1434. tration of captopril 50 mg, hydrochlorothiazide 25 mg and their com-
6. Frishman WH, Bryzinski BS, Coulson LR, et al. A multifactorial trial bination in mild to moderate hypertension. Br J Clin Pharmacol.
design to assess combination therapy in hypertension. Arch Intern 1987;23(Suppl 1):89S-92S.
Med. 1994;154:1461-1468. 27. Pool J, Cushman WC, Saini RK, et al. Use of the factorial design and
7. Moser M, Prisant LM. Low-dose combination therapy in hypertension. quadratic response surface models to evaluate the fosinopril and hy-
Am Fam Physician. 1997;56:1275-1279. drochlorothiazide combination therapy in hypertension. Am J Hyper-
8. Rosenthal T, Gavras I. Fixed-dose combinations as first-line treatment tens. 1997;10:117-123.
for hypertension. Prog Cardiovasc Dis. 2006;48:416-425. 28. Pordy RC. Cilazapril plus hydrochlorothiazide: improved efficacy
9. Fretheim A, Oxman AD. International variation in prescribing antihy- without reduced safety in mild to moderate hypertension. Cardiology.
pertensive drugs: its extent and possible explanations. BMC Health 1994;85:311-322.
Serv Res. 2005;5:21. 29. Scholze J, Breitstadt A, Cairns V, et al. Ramipril and hydrochlorothi-
10. Bateman DN, Dean CR, Mucklow JC, et al. Atenolol and chlorthali- azide combination therapy in hypertension: a clinical trial of factorial
done in combination for hypertension. Br J Clin Pharmacol. 1979;7: design. J Hypertens. 1993;11:217-221.
357-363. 30. Burris JF, Weir MR, Oparil S, et al. An assessment of diltiazem and
11. Chalmers JP, Korner PI, Tiller DJ, et al. Double-blind factorial trial of hydrochlorothiazide in hypertension. JAMA. 1990;263:1507-1512.
prindolol and hydrochlorothiazide in hypertension. Med J Aust. 1976; 31. Pool PE, Applegate WB, Woehler T, et al. A randomized, controlled
1:650-653. trial comparing diltiazem, hydrochlorothiazide, and their combination
300 The American Journal of Medicine, Vol 122, No 3, March 2009

in the therapy of essential hypertension. Pharmacotherapy. 1993;13: 48. Messerli F, Frishman WH, Elliott WJ. Effects of verapamil and tran-
487-493. dolapril in the treatment of hypertension. Am J Hypertens. 1998;11:
32. Salvetti A, Magagna A, Innocenti P, et al. The combination of 322-327.
chlorthalidone with nifedipine does not exert an additive antihyper- 49. Scholze J, Zilles P, Compagnone D. Verapamil SR and trandolapril
tensive effect in essential hypertensives: a crossover multicenter study. combination therapy in hypertension—a clinical trial of factorial de-
J Cardiovasc Pharmacol. 1991;17:332-335. sign. Br J Clin Pharmacol. 1998;45:491-495.
33. Weir MR, Weber MA, Punzi HA, et al. A dose escalation trial com- 50. Scholze J, Bauer B, Massaro J. Antihypertensive profiles with ascend-
paring the combination of diltiazem SR and hydrochlorothiazide with ing dose combinations of ramipril and felodipine ER. Clin Exp Hy-
the monotherapies in patients with essential hypertension. J Hum pertens. 1999;21:1447-1462.
Hypertens. 1992;6:133-138. 51. Veratran Study Group. Effects of verapamil SR, trandolapril, and their
34. Wing LMH, Arnolda LF, Harvey PJ, et al. Lacidipine, hydrochlorothi- fixed combination on 24-h blood pressure. Am J Hypertens. 1997;10:
azide and their combination in systolic hypertension in the elderly. 492-499.
J Hypertens. 1997;15:1503-1510. 52. Law M, Wald N, Morris J. Lowering blood pressure to prevent myo-
35. Wald DS, Law M, Mills S, et al. A 16-week, randomized, double-blind cardial infarction and strokes: a new preventive strategy. Health Tech-
placebo-controlled cross-over trial to quantify the combined effect of nol Assess. 2003;7:31.
an angiotensin-converting enzyme inhibitor and a ␤-blocker on blood 53. Kochar M, Guthrie R, Triscari J, et al. Matrix study of irbesartan with
pressure reduction. Clin Ther. 2008;30:2030-2039. hydrochlorothiazide in mild-to-moderate hypertension. Am J Hyper-
36. Wing LMH, Chalmers JP, West MJ, et al. Enalapril and atenolol in tens. 1999;12:797-805.
essential hypertension: attenuation of hypertensive effects in combi- 54. Benz JR, Black HR, Graff A, et al. Valsartan and hydrochlorothiazide
nation. Clin Exp Hypertens. 1988;10:119-133. in patients with essential hypertension. A multiple dose, double-blind,
37. Clement DL, De Pue NY, Packet L. Effect of calcium antagonists on placebo controlled trial comparing combination therapy with mono-
ambulatory blood pressure and its variations. J Cardiovasc Pharmacol. therapy. J Hum Hypertens. 1998;12:861-866.
1987;10(Suppl 10):S117-S119. 55. Mackay JH, Arcuri KE, Goldberg AI, et al. Losartan and low-dose
hydrochlorothiazide in patients with essential hypertension. Arch In-
38. Dargie H, Cleland J, Findlay I, et al. Combination of verapamil and
tern Med. 1996;156:278-285.
beta-blockers in systemic hypertension. Am J Cardiol. 1986;57:80D-
56. Chrysant SG, Melino M, Karki S, et al. The Combination of Olme-
82D.
sartan Medoxomil and Amlodipine Besylate in Controlling High
39. Lyons D, Fowler G, Webster J, et al. An assessment of lacidipine and
Blood Pressure: COACH, a randomized, double-blind, placebo-con-
atenolol in mild to moderate hypertension. Br J Clin Pharmacol.
trolled, 8-week factorial efficacy and safety study. Clin Ther. 2008;
1994;37:45-51.
30;587-604.
40. Maclean D, Mitchell ET, Lewis R, et al. Comparison of once daily
57. Prospective Studies Collaboration. Age-specific relevance of usual
atenolol, nitrendipine and their combination in mild to moderate es-
blood pressure to vascular mortality: a meta-analysis of individual data
sential hypertension. Br J Clin Pharmacol. 1990;29:455-463.
from one million adults in 61 prospective studies. Lancet. 2002;360:
41. McInnes GT, Findlay IN, Murray G, et al. Cardiovascular responses to
1903-1913.
verapamil and propranolol in hypertensive patients. J Hypertens. 1985; 58. Law M, Morris JK, Wald NJ. Use of blood pressure lowering drugs in
3(Suppl 3)219-221. the prevention of cardiovascular disease. BMJ. 2009. In press.
42. Tonkin AL, Wing LMH, Russell AE, et al. Diltiazem and atenolol in 59. Hansson L, Zanchetti A, Carruthers SG, et al., for the HOT Study
essential hypertension: additivity of effects on blood pressure and Group. Effects of intensive blood-pressure lowering and low-dose
cardiac conduction with combination therapy. J Hypertens. 1990;8: aspirin in patients with hypertension: principal results of the Hyper-
1015-1018. tension Optimal Treatment (HOT) randomised trial. Lancet. 1998;351:
43. Chan P, Lin C, Tomlinson B, et al. Additive effects of diltiazem and 1755-1762.
lisinopril in treatment of elderly patients with mild-to-moderate hy- 60. Lui L, Zhang Y, Liu G, et al., for the FEVER Study Group. The
pertension. Am J Hypertens. 1997;10:743-749. felodipine event reduction (FEVER) Study: a randomized long-term
44. Cushman WC, Cohen JD, Jones RP, et al. Comparison of the fixed placebo-controlled trial in Chinese hypertensive patients. J Hypertens.
combination of enalapril/diltiazem ER and their monotherapies in 2005;23:2157-2172.
stage 1 to 3 essential hypertension. Am J Hypertens. 1998;11:23-30. 61. Neaton JD, Grimm RH, Prineas RJ, et al., for the Treatment of Mild
45. Frishman WH, Ram CVS, McMahon FG, et al. Comparison of amlo- Hypertension Study Research Group (TOMHS). Treatment of Mild
dipine and benazepril monotherapy to amlodipine plus benazepril in Hypertension Study. JAMA. 1998;270:713-724.
patients with systemic hypertension: a randomized, double-blind, pla- 62. Schrier RW, Estacio RO, Esler A, Mehler P. Effects of aggressive
cebo-controlled, parallel-group study. J Clin Pharmacol. 1995;35: blood pressure control in normotensive type 2 diabetic patients on
1060-1066. albuminuria, retinopathy and strokes (the second Appropriate Blood
46. Kuschnir E, Acuña E, Sevilla D, et al. Treatment of patients with Pressure Control in Diabetes (ABCD) trial). Kidney Int. 2002;61:1086-
essential hypertension: amlodipine 5 mg/benazepril 10 mg compared 1097.
with amlodipine 5 mg, benazepril 20 mg, and placebo. Clin Ther. 63. Asselbergs FW, Diercks GFH, Hillege HL, et al., for the Prevention of
1996;18:1213-1224. Renal and Vascular Endstage Disease Intervention Trial (PREVENDIT)
47. Levine JH, Ferdinand KC, Cargo P, et al. Additive effects of verapamil Investigators. Effects of fosinopril and pravastatin on cardiovascular
and enalapril in the treatment of mild to moderate hypertension. Am J events in subjects with microalbuminuria. Circulation. 2004;110:2809-
Hypertens. 1995;8:494-499. 2816.