- American Journal of Cardiology

Transcription

- American Journal of Cardiology
A Meta-Analysis of 94,492 Patients With Hypertension Treated With
Beta Blockers to Determine the Risk of New-Onset Diabetes
Mellitus
Sripal Bangalore, MD, MHAa,*, Sanobar Parkar, MD, MPHa, Ehud Grossman, MDb,
and Franz H. Messerli, MDa
Beta blockers used for the treatment of hypertension may be associated with increased risk
for new-onset diabetes mellitus (DM). A search of Medline, PubMed, and EMBASE was
conducted for randomized controlled trials of patients taking ␤ blockers as first-line
therapy for hypertension with data on new-onset DM and follow-up for >1 year. Twelve
studies evaluating 94,492 patients fulfilled the inclusion criteria. Beta-blocker therapy
resulted in a 22% increased risk for new-onset DM (relative risk 1.22, 95% confidence
interval [CI] 1.12 to 1.33) compared with nondiuretic antihypertensive agents. A higher
baseline fasting glucose level (odds ratio [OR] 1.01, 95% CI 1.00 to 1.02, p ⴝ 0.004) and
greater systolic (OR 1.05, 95% CI 1.05 to 1.08, p ⴝ 0.001) and diastolic (OR 1.06, 95% CI
1.01 to 1.10, p ⴝ 0.011) blood pressure differences between the 2 treatment modalities were
significant univariate predictors of new-onset DM. Multivariate meta-regression analysis
showed that a higher baseline body mass index (OR 1.17, 95% CI 1.01 to 1.33, p ⴝ 0.034)
was a significant predictor of new-onset DM. The risk for DM was greater with atenolol, in
the elderly, and in studies in which ␤ blockers were less efficacious antihypertensive agents
and increased exponentially with increased duration on ␤ blockers. For the secondary end
points, ␤ blockers resulted in a 15% increased risk for stroke, with no benefit for the end
point of death or myocardial infarction. In conclusion, ␤ blockers are associated with an
increased risk for new-onset DM, with no benefit for the end point of death or myocardial
infarction and with a 15% increased risk for stroke compared with other agents. This risk
was greater in patients with higher baseline body mass indexes and higher baseline fasting
glucose levels and in studies in which ␤ blockers were less efficacious antihypertensive
agents compared with other treatments. © 2007 Elsevier Inc. All rights reserved. (Am J
Cardiol 2007;100:1254 –1262)
The prevalence of hypertension is increasing in the United
States.1 At the same time, the prevalence of diabetes mellitus (DM) and prediabetes is also on the increase,2 and this
parallels the growing prevalence of obesity and sedentary
lifestyles. Given that DM is associated with accelerated
atherosclerosis, the primary prevention of the development
of DM is of foremost importance. Patients with hypertension and DM are especially at high cardiovascular risk,3,4
making the primary prevention of DM in patients with
hypertension very important.
In addition to weight loss, exercise, and diet, the avoidance of medications that cause weight gain or impair weight
loss and/or impair insulin sensitivity or glucose tolerance is
thus of paramount importance. Previous studies have shown
that ␤ blockers and diuretics, the most commonly used
treatments for hypertension, can cause metabolic derangements leading to increased risk for new-onset DM.5–11 Howa
Department of Medicine, Division of Cardiology, St. Luke’s-Roosevelt Hospital and Columbia University College of Physicians and Surgeons, New York, New York; and bChaim Sheba Medical Center and
Sackler School of Medicine, Tel-Hashomer, Israel. Manuscript received
April 25, 2007; revised manuscript received and accepted May 11, 2007.
*Corresponding author: Tel: 212-523-5678; fax: 212-957-3680.
E-mail address: [email protected] (S. Bangalore).
0002-9149/07/$ – see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.amjcard.2007.05.057
ever, other studies have shown conflicting results,12–14 with
some studies even showing reduced risk for DM with ␤
blockers.15 The limitations of previous studies has been
short follow-up periods,15 the derivation of data from observational studies,13 randomized controlled trials with a
lack of adequate statistical power to evaluate the risk for
new-onset DM,12 the inclusion of studies in which ␤ blockers were not the first-line therapy, and a dearth of randomized controlled trials with new-onset DM as a predefined
end point. The aim of this meta-analysis was thus to explore
the role of ␤ blockers in the risk for new-onset DM in
patients with hypertension.
Methods
Study selection: We conducted a search of studies on
Medline, PubMed, and EMBASE using the terms: “␤ adrenergic blockers,” “adrenergic ␤ antagonist,” “␤ blockers,”
and “hypertension.” We limited our search to studies in
human subjects published in English in peer-reviewed journals from 1966 to March 2007. We included only randomized controlled studies with randomized comparisons of
regimens based on ␤ blockers with those using other agents,
with follow-up of ⱖ1 year, and evaluating the occurrence of
new-onset DM as an end point.
www.AJConline.org
Systemic Hypertension/␤ Blockers and New-Onset Diabetes
1255
variate analysis, the following covariates were considered:
systolic and diastolic blood pressure at entry into and end of
study, systolic and diastolic blood pressure difference between the 2 treatment modalities, baseline body mass index
(BMI), baseline fasting glucose level, age, and follow-up
duration. The selection criterion for multivariate regression
analysis was based on univariate statistical significance
and/or clinical judgment. The estimated between-study variance (␶2) was calculated using an estimate based on restricted maximum likelihood and is a measure of the residual heterogeneity having adjusted for the covariates. A p
value ⬍0.05 was considered significant.
Figure 1. Study flow diagram. BB ⫽ ␤ blockers; HTN ⫽ hypertension;
RCT ⫽ randomized controlled trial.
Data extraction: The primary end point was new-onset
DM. We extracted the baseline characteristics, definition of
new-onset DM (if available), and the outcome of interest for
each of the studies. The summary statistics for the end point
of interest were evaluated only in the subset of patients
without DM at baseline in each of the studies. The secondary end point of interest was death (from all causes), myocardial infarction, and stroke.
Results
Study selection: Of the 805 randomized controlled trials
of ␤ blockers in patients with hypertension, 12 studies
evaluating 94,492 patients fulfilled the inclusion criteria
(Figure 1, Tables 1 to 3). We excluded the results of the
Metoprolol Atherosclerosis in Hypertension (MAPHY)
trial19 because this was a subgroup analysis from the Heart
Attack Primary Prevention in Hypertension (HAPPHY)
trial.20 Similarly, we excluded the results of the Systolic
Hypertension in the Elderly (SHEP) study,21 the Antihypertensive Treatment and Lipid Profile in a North of Sweden
Efficacy Evaluation (ALPINE) study,22 and the Atherosclerosis Risk in Community (ARIC) study8 because ␤ blockers
were not the first-line agents in these studies. The definition
of new-onset DM used in various studies was not uniform
(Tables 2 and 4).
Statistical analyses: Statistical analysis was done using
standard software (Stata version 9.0; StataCorp LP, College
Station, Texas) using the METAN program.16 Given the
known deleterious metabolic effects of diuretics, analysis
was performed comparing (1) ␤ blockers with placebo,
(2) ␤ blockers with diuretics, and (3) ␤ blockers with nondiuretic antihypertensive agents (angiotensin-converting
enzyme [ACE] inhibitors, angiotensin receptor blockers
[ARBs], and calcium channel blockers [CCBs]). The pooled
effect for each grouping of trials was derived from the point
estimate for each separate trial weighted by the inverse of
the variance (1/SE2). Heterogeneity was assessed visually
using funnel plots, Q (chi-square) statistics, and or I2 statistics.17 If trials were homogenous (p ⬎0.05), a fixedeffects model was used to calculate pooled effect sizes.
Otherwise, the random-effects model of DerSimonian and
Laird18 was applied to calculate overall differences. Publication bias was estimated using the weighted regression test
of Egger. For this analysis, we defined the younger cohort as
studies in which the mean age of the population was ⬍60
years and the elderly cohort as studies in which the mean
age of the population was ⱖ60 years. A sensitivity analysis
was performed after excluding mixed trials of ␤ blockers
and diuretics in which patients could be randomized to
either agent.
Beta blockers versus placebo: CHARACTERISTICS OF THE
We identified 2 placebo-controlled trials enrolling
16,372 patients who were followed for a mean of 5.3 years
(Table 1).23,24 In essence, a total of 1,102 patients receiving
atenolol and 4,403 patients receiving propranolol (without
DM at baseline) were compared with those receiving placebo. Compared with placebo, ␤ blockers resulted in reductions in systolic (mean 11.2 mm Hg) and diastolic (mean 6
mm Hg) blood pressure (Table 2).
NEW-ONSET DM: Beta-blocker therapy (pooled) resulted in
a 33% increased risk for new-onset DM (relative risk [RR]
1.33, 95% confidence interval [CI] 1.00 to 1.76, p ⫽ 0.05)
on the basis of the fixed-effects model or a trend toward
44% increased risk for new-onset DM compared with placebo (RR 1.44, 95% CI 0.69 to 3.00, p ⫽ 0.33, heterogeneity chi-square 6.18, p ⫽ 0.013) on the basis of the random-effects model. This was more pronounced in the
elderly cohort (aged ⱖ60 years; RR 2.13, 95% CI 1.34 to
3.38).24 In the younger cohort (aged ⬍60 years), however,
propranolol-based therapy did not result in any significant
increase in new-onset DM compared with placebo (RR
1.01, 95% CI 0.70 to 1.45). However, these subgroup analyses were based on single studies alone (the Medical Research Council Trial of Treatment of Mild Hypertension
[MRC] and the Medical Research Council Trial of Treatment of Hypertension in Older Adults [MRC-O]).
Meta-regression analysis: Meta-regression analysis
was used to explore and explain diversity (heterogeneity)
among the results of different studies. Univariate and multivariate regression analyses were performed. For the uni-
Beta blockers versus thiazide diuretics: CHARACTERISTICS
We identified 5 randomized controlled trials
comparing ␤ blockers with thiazide diuretics, enrolling
17,860 patients who were followed for a mean of 1 to 10
TRIALS:
OF THE TRIALS:
1256
The American Journal of Cardiology (www.AJConline.org)
Table 1
Clinical trials comparing ␤ blockers with placebo or diuretics
Trial
BBs vs placebo
MRC23 (placebo arm)
MRC-O24 (placebo arm)
BBs vs diuretics
HAPPHY20
MRC23 (diuretic arm)
MRC-O24 (diuretic arm)
Berglund12
VACS (1982)25
Patient Characteristics
No. of Follow-up
Patients
(yrs)
Adults without MI/CVA within 3 mo, angina, or DM
Adults without MI/CVA within 3 mo, renal dysfunction, or DM
BB
Comparison
13,057
3,315
4.9
5.8
Propranolol
Atenolol
Placebo
Placebo
Men without MI, angina, CVA
6,569
3.8
BFZ/HCTZ
Adults without MI/CVA within 3 mo, angina, or DM
Adults without MI/CVA within 3 mo, renal dysfunction, or DM
Men with SBP ⬎170 or DBP ⬎105 mm Hg
Men with DBP 95–114 mm Hg
7,077
2,183
106
394
4.9
5.8
10
1
Atenolol or
metoprolol
Propranolol
Atenolol
Propranolol
Propranolol
BFZ
HCTZ ⫹ amiloride
BFZ
HCTZ
BB ⫽ ␤ blocker; BFZ ⫽ bendrofluazide; CVA ⫽ cerebrovascular accident; DBP ⫽ diastolic blood pressure; HCTZ ⫽ hydrochlorothiazide; MI ⫽
myocardial infarction; SBP ⫽ systolic blood pressure; VACS ⫽ Veterans Administration Cooperative Study Group on Antihypertensive Agents.
Table 2
Blood pressure response: ␤ blockers versus placebo or diuretics
Trial
Age
(yrs)
Men
(%)
Study-end BP (mm Hg)
(BB comparison)*
New-onset DM (BB
comparison)
Definition of New-onset DM
MRC23 (placebo arm)
MRC-O24 (placebo arm)
HAPPHY20
52
70
52
52
42
100
⫺9.0/⫺5.0
⫺13.5/⫺7.0
0.0/⫺1.0
43 (1.0%)
36 (3.3%)
86 (2.6%)
84 (1.0%)
34 (1.5%)
75 (2.3%)
MRC23 (diuretic arm)
MRC-O24 (diuretic arm)
Berglund12
VACS25 (1982)
52
70
50
50
52
42
100
100
⫹4.0/⫹1.0
⫹1.0/⫺0.5
NA/⫹1.0
⫹9.2/⫹1.8
43 (1.0%)
36 (3.3%)
5 (9.4%)
5 (4%)
106 (2.5%)
43 (4.0%)
1 (1.8%)
11 (6%)
Impaired glucose tolerance
Impaired glucose tolerance
FPG ⬎6.8 mmol/L (122 mg/dl) and 2 positive
dipsticks for glucosuria
Impaired glucose tolerance
Impaired glucose tolerance
FPG ⬎6.7 mmol/L (121 mg/dl) and glucosuria
FPG ⬎150 mg/dl
* Negative numbers indicates that BBs are more efficacious than comparison group.
BP ⫽ blood pressure; FPG ⫽ fasting plasma glucose; NA ⫽ not available; other abbreviations as in Table 1.
Table 3
Clinical trials comparing ␤ blockers with nondiuretic antihypertensive agents
Trial
BBs vs CCBs
ASCOT27
INVEST11
NORDIL28
STOP229 (vs CCB)
AASK31 (vs CCB)
BBs vs ACE inhibitors/ARBs
CAPPP30
STOP229 (vs ACE inhibitors)
LIFE26
AASK31 (vs ACE inhibitors)
Patient Characteristics
No. of
Patients
Follow-up
(yrs)
Adults with ⱖ1 other CV risk factor
but no CAD
Adults with CAD
Adults with DBP ⬎100 mm Hg
Adults without MI or CVA within
previous 12 mo
African-Americans with hypertensive
renal disease
19,257
5.5
Atenolol
Amlodipine
22,576
10,881
4,409
2.7
4.5
5.0
Verapamil SR
Diltiazem
Felodipine/isradipine
658
3.0
Atenolol
Any diuretic/any BB
Atenolol/metoprolol/pindolol/
HCTZ/amiloride
Metoprolol
Adults without renal disorders
10,985
6.1
Captopril
4,418
5.0
9,222
4.8
Atenolol or
metoprolol/HCTZ/BFZ
Atenolol/metoprolol/pindolol/
HCTZ/amiloride
Atenolol
877
3.8
Metoprolol
Ramipril
Adults without MI or CVA within
previous 12 mo
Adults with LVH without MI or CVA
in previous 6 mo
African-Americans with hypertensive
renal disease
BB
Comparison
Amlodipine
Enalapril/lisinopril
Losartan
ASCOT ⫽ Anglo-Scandinavian Cardiac Outcomes Trial; CAD ⫽ coronary artery disease; CV ⫽ cardiovascular; INVEST ⫽ International VerapamilTrandolapril Study; LIFE ⫽ Losartan Intervention for Endpoint Reduction in Hypertension; LVH ⫽ left ventricular hypertrophy; NORDIL ⫽ Nordic
Diltiazem; SR ⫽ sustained release; STOP2 ⫽ Swedish Trail in Old Patients With Hypertension-2; other abbreviations as in Table 1.
years (Table 1).12,20,23–25 In essence, a total of 4,399 patients
receiving atenolol or metoprolol and 4,581 patients receiving propranolol (without baseline DM) were compared with
those receiving diuretics (bendrofluazide, hydrochlorothiazide, and amiloride). Only 1 study, HAPPHY, showed comparable blood pressure reduction with a ␤ blocker (atenolol)
Systemic Hypertension/␤ Blockers and New-Onset Diabetes
1257
Table 4
Blood pressure response: ␤ blockers versus nondiuretic antihypertensive agents
Trial
Age
(yrs)
Men
(%)
Study-end BP (mm Hg)
(BB comparison)*
ASCOT27
INVEST11
NORDIL28
CAPPP30
STOP229 (BB vs CCB)
STOP229 (BB vs ACE inhibitor)
AASK31 (BB vs CCB)
63
66
60
53
76
76
54
77
48
49
53
33
33
62
⫹2.7/⫹1.9
⬍1
⫺3.0/0.0
⫺1.0/⫺1.0
⫺1.0/⫹1.0
⫺1.0/0.0
⫹2.0/0.0
AASK31 (BB vs ACE inhibitor)
LIFE26
54
67
62
46
0.0/⫺1.0
⫹1.1/⫺0.2
New-onset DM
(BB comparison)
799 (11.3%)
665 (8.2%)
251 (4.9%)
380 (7.3%)
97 (4.9%)
97 (4.9%)
567 (8.0%)
569 (7.0%)
216 (4.3%)
337 (6.5%)
95 (4.8%)
93 (4.7%)
319 (8.0%)
241 (6.0%)
Definition of New-onset DM
Not specified
Not specified
Not specified
2 FPG ⬎6.7 mmol/L (121 mg/dl)
2 FPG ⬎6.7 mmol/L (121 mg/dl)
2 FPG ⬎6.7 mmol/L (121 mg/dl)
Clinical diagnosis of DM or FSG
level ⱖ126 mg/dl
As above
2 FPG ⬎6.7 mmol/L (121 mg/dl)
* Negative numbers indicate that BBs were more efficacious than comparison group.
FSG ⫽ fasting serum glucose; other abbreviations as in Tables 1 to 3.
Figure 2. Effect of ␤ blockers on the risk for new-onset DM when
compared to nondiuretic antihypertensive agents (CCBs, ACE inhibitors
[ACEi], and ARBs). **Variation in RR attributable to heterogeneity. ASCOT ⫽ Anglo-Scandinavian Cardiac Outcomes Trial; df ⫽ degree of
freedeom; INVEST ⫽ International Verapamil-Trandolapril Study; LIFE ⫽
Losartan Intervention for Endpoint Reduction in Hypertension; NORDIL ⫽
Nordic Diltiazem; STOP2 ⫽ Swedish Trial in Old Patients With Hypertension-2; other abbreviation as in Figure 1.
compared with diuretics (Table 2). In all other studies,
diuretics reduced blood pressure to a greater extent (mean
3.5/0.5 mm Hg higher) compared with ␤ blockers.
NEW-ONSET
DM:
Compared with thiazide diuretics,
␤-blocker therapy resulted in a 26% decreased risk for
new-onset DM (RR 0.74, 95% CI 0.61 to 0.90, p ⫽ 0.002)
on the basis of the fixed-effect models or a trend toward
21% decreased risk for new-onset DM compared with placebo (RR 0.79, 95% CI 0.45 to 1.41, p ⫽ 0.43, heterogeneity chi-square ⫽ 23.18, p ⬍0.0001) on the basis of the
random-effects model. In the younger cohort (aged ⬍60
years), ␤-blocker therapy resulted in a trend toward an 18%
reduction in the risk for new-onset DM compared with diuretics (pooled RR 0.82, 95% CI 0.37 to 1.84, p ⫽ 0.63, heterogeneity chi-square ⫽ 22.97, p ⬍0.0001). In the only study in
Figure 3. Effect of different ␤-blocker types on the risk for new-onset DM
compared with other agents. *Versus placebo; †versus diuretics; †versus
ACE inhibitors; ‡versus CCBs. **Variation in RR attributable to heterogeneity. VACS ⫽ Veterans Administration Cooperative Study Group on
Antihypertensive Agents; other abbreviations as in Figures 1 and 2.
the elderly cohort in this group (MRC-O), ␤-blocker therapy
(atenolol) resulted in a trend toward an 18% reduction in the
risk for new-onset DM compared with diuretics.
Beta blockers versus nondiuretic antihypertensive
agents: CHARACTERISTICS OF THE TRIALS: We identified 7 randomized controlled trials comparing ␤ blockers with nondiuretic antihypertensive drugs (ACE inhibitors, ARBs, or
CCBs), enrolling 65,765 patients who were followed for a
mean of 2.7 to 6.1 years (Table 3).11,26 –31 In essence, a total
of 19,097 patients receiving atenolol, 405 patients receiving
metoprolol, and 14,247 patients receiving mixed therapy
with ␤ blockers and diuretics were compared with those
receiving ACE inhibitors or ARBs (lisinopril, enalapril,
ramipril, captopril, and losartan) and CCBs (amlodipine,
felodipine, diltiazem, sustained-release verapamil, and isradipine). The blood pressure–lowering efficacy of ␤ blockers
was comparable with that of with nondiuretic antihyperten-
1258
The American Journal of Cardiology (www.AJConline.org)
Table 5
Meta–regression analysis
Variable
Univariate analysis
SBP difference
DBP difference
Baseline SBP
Baseline DBP
Fasting blood glucose (baseline)
End-of-trial SBP
End-of-trial DBP
Baseline BMI
Multivariate analysis
SBP difference
DBP difference
Fasting blood glucose (baseline)
Baseline BMI
Regression Coefficient
95% CI
p Value
␶2
0
0.000034
0.0114
0.0113
0
0.0087
0.0094
0.0079
0
0.051
0.059
⫺0.003
⫺0.004
0.012
⫺0.008
⫺0.010
0.074
0.051
0.013
⫺0.012
⫺0.022
0.001
⫺0.018
⫺0.035
⫺0.017
to
to
to
to
to
to
to
to
0.081
0.105
0.003
0.014
0.021
0.002
0.014
0.165
0.001
0.011
0.320
0.656
0.004
0.140
0.415
0.110
⫺0.054
⫺0.021
0.026
0.169
⫺0.170
⫺0.102
⫺0.005
0.013
to
to
to
to
0.063
0.061
0.057
0.326
0.364
0.621
0.104
0.034
Abbreviations as in Table 1.
sive medications (mean reduction 0.5/⫺0.2 mm Hg; Table
4). However, this apparent efficacy was driven mainly by
the mixed trials with ␤ blockers and diuretics. If these
studies were excluded, the antihypertensive efficacy of ␤
blockers was less than that of nondiuretic antihypertensive
agents.
NEW-ONSET DM: Beta-blocker therapy resulted in 22%,
21%, and 19% increased risk for new-onset DM compared
with nondiuretic antihypertensive agents, CCBs, and ACE
inhibitors or ARBs, respectively (Figure 2).
In the studies in a elderly cohort (aged ⱖ60 years),
␤-blocker therapy resulted in a 22% increased risk for
new-onset DM (RR 1.22, 95% CI 1.10 to 1.35, p ⬍0.0001,
heterogeneity chi-square 12.06, p ⫽ 0.034). In the younger
cohort (aged ⬍60 years) (in the African American Study of
Kidney Disease and Hypertension [AASK] and the Captopril Prevention Project [CAPPP]), ␤-blocker therapy resulted in a 17% increased risk for new-onset DM compared
with other antihypertensive agents (pooled RR 1.17, 95% CI
1.03 to 1.33, p ⫽ 0.018, heterogeneity chi-square 3.20, p
0.074).
Beta-blocker type and new-onset DM: In studies in
which the ␤ blocker evaluated was atenolol,11,24,26,27 the risk
for new-onset DM was 30% greater than with other agents
(Figure 3). In the studies excluding diuretics,20,24 as expected, atenolol resulted in an even higher risk for newonset DM compared with other agents (pooled RR 1.35,
95% CI 1.17 to 1.56, p ⬍0.0001, heterogeneity chi-square ⫽
10.59, p ⫽ 0.014).
In the study in which the ␤ blocker evaluated was metoprolol,31 the risk for new-onset DM was 34% compared
with that with other agents (pooled RR 1.34, 95% CI 1.04 to
1.73, p ⫽ 0.026, heterogeneity chi-square 1.93, p 0.164)
using the fixed-effects model (Figure 3 shows the same
results with random-effects model).
In the studies in which the ␤ blocker evaluated was
propranolol,12,23,25 the risk for new-onset DM showed a
trend toward 23% lower risk compared with other agents
(Figure 3). However, the studies were heavily weighted by
comparison with diuretics (3 of the 4 studies in this group).
In the only nondiuretic study in this group (the placebo arm
Figure 4. RR for new-onset DM (NODM) as a function of systolic blood
pressure (SBP) difference between ␤ blockers and nondiuretic antihypertensive agents. Other abbreviations as in Figures 2 and 3.
of MRC), there was no reduction in the risk for new-onset
DM with propranolol compared with placebo.
Sensitivity analysis: Sensitivity analysis was performed
to evaluate for the effects of mixed studies of ␤ blockers and
diuretics.28 –30 In the studies using mixed ␤ blockers and
diuretics, the ␤-blocker and diuretic arms resulted in an 11%
increased risk for new-onset DM compared with other
agents (CCBs and ACE inhibitors; pooled RR 1.11, 95% CI
1.01 to 1.22, p ⫽ 0.038, heterogeneity chi-square ⫽ 0.69,
p ⫽ 0.875). However, in the studies excluding mixed ␤
blockers and diuretics, ␤ blockers resulted in an even
greater risk (30%) for new-onset DM compared with nondiuretic agents (pooled RR 1.30, 95% CI 1.22 to 1.39, p
⬍0.0001, heterogeneity chi-square 14.54, p ⫽ 0.024).
Meta-regression analysis: Meta-regression analysis
was performed to evaluate for the heterogeneity observed in
the previous analyses. In the studies comparing ␤ blockers
with nondiuretic antihypertensive agents, univariate meta-
Systemic Hypertension/␤ Blockers and New-Onset Diabetes
Figure 5. RR for new-onset DM as a function of baseline fasting blood
glucose levels. Abbreviations as in Figures 2 to 4.
1259
Figure 7. RR for new-onset DM as a function of length of follow-up.
Abbreviations as in Figures 2 to 4.
these variables. Multivariate meta-regression analysis, using
the systolic and diastolic blood pressure differences between the 2 treatment modalities, fasting blood glucose at
baseline, and baseline BMI (forced in the model), explained
most of the between-study variance in the group (␶2 reduced
from 0.00080 to 0). There was an exponential increase in
RR for new-onset DM with higher baseline BMI (OR 1.17,
95% CI 1.01 to 1.33, p ⫽ 0.034; Figure 6). Finally, for the
entire cohort, the RR for new-onset DM increased with
increasing length of follow-up on ␤ blockers (Figure 7).
Secondary analyses: Pooled analysis showed that ␤blocker therapy resulted in a trend toward 4% increased risk
for death (pooled RR 1.04, 95% CI 1.00 to 1.09, p ⫽ 0.056,
heterogeneity chi-square 10.63, p ⫽ 0.560) and 15% increased risk for stroke (pooled RR 1.15, 95% CI 1.01 to
1.30, p ⫽ 0.029, heterogeneity chi-square 27.8, p ⫽ 0.001),
with no benefit for the end point of myocardial infarction
(pooled RR 1.02, 95% CI 0.92 to 1.12, p ⫽ 0.769, heterogeneity chi-square 19.30, p ⫽ 0.023).
Figure 6. RR for new-onset DM as a function of baseline BMI. Abbreviations as in Figures 2 to 4.
Discussion
regression analysis showed that systolic (odds ratio [OR]
1.05, 95% CI 1.05 to 1.08, p ⫽ 0.001) and diastolic (OR
1.06, 95% CI 1.01 to 1.10, p ⫽ 0.011) blood pressure
differences between the 2 treatment modalities at end of
study and baseline fasting blood glucose levels (OR 1.01,
95% CI 1.00 to 1.02, p ⫽ 0.004) were predictive of the risk
for new-onset DM (Table 5). The RR for new-onset DM
followed an exponential relation with the systolic blood
pressure difference between the 2 treatment modalities at
study end (Figure 4), with an increase in RR with increasing
systolic blood pressure difference between the 2 arms. A
similar exponential relation was observed between baseline
fasting glucose levels and RR for new-onset DM, with
increased risk with increase in baseline fasting glucose
levels (Figure 5). The heterogeneity in the analysis as shown
by between-study variance was substantially explained by
The results of the present meta-analysis show that ␤-blocker
therapy for hypertension is associated with increased risk
for new-onset DM compared with nondiuretic antihypertensive drugs and also compared with placebo. This excess risk
was greater in the cohort aged ⱖ60 years, in the cohort with
higher baseline fasting glucose levels and higher baseline
BMIs, and when the systolic blood pressure difference between the 2 treatment groups at study end was greater. The
results of this analysis let us calculate that the treatment of
1,000 patients with ␤ blockers for 4.4 years will result in 14
excess cases of DM, 3 excess deaths, and 4.7 excess strokes.
Given that 65 million Americans have hypertension, this
would account for 910,000 cases of DM, 195,000 deaths,
and 305,500 excess strokes. This is hardly an acceptable
risk/benefit ratio.
The relation between insulin resistance and primary hypertension is complex. Previous studies have shown that
1260
The American Journal of Cardiology (www.AJConline.org)
Figure 8. Beta blockers for primary hypertension: pathogenesis for increased risk. BP ⫽ blood pressure; SM ⫽ smooth muscle; other abbreviation as in Figure 1.
hypertension per se worsens insulin resistance32 and that
insulin resistance and hyperinsulinemia can cause or predispose patients to become hypertensive.33,34 In this study,
although blood pressure was reduced significantly with ␤
blockers, there was a trend toward greater risk for new-onset
DM compared with placebo, implying that the risk is in part
related to the medication itself. Similarly, in the comparison
against diuretics, although diuretics resulted in a greater
blood pressure reduction, there was a trend toward greater
risk for DM in the diuretic group compared with ␤ blockers,
emphasizing that reduction in blood pressure is necessary
but not sufficient to reduce the risk for new-onset DM, and
the type of antihypertensive agent used may play an important role as well.
Thiazide diuretics worsen glycemic control by impairing
insulin secretion and by increasing insulin resistance.35 Beta
blockers have been shown to inhibit pancreatic insulin secretion (via ␤-2 receptors), worsen insulin resistance, cause
weight gain,36 diminish peripheral blood flow, and lead to
increased glycogenolysis (by unopposed ␣-2 action), all of
which are implicated in adverse glycemic control.9,37 This is
not a class effect, and ␤ blockers with intrinsic sympathomimetic effects, ␤-1 selective blockers with ␤-2 agonist
properties, and newer noncardioselective ␤ blockers with
vasodilating properties (such as carvedilol) have minimal
effects on glycemic control.38 In this meta-analysis, atenolol
resulted in a 30% increased risk for DM, metoprolol in a
32% increased risk (trend), and propranolol in a 23% decreased risk (trend). However, most of the studies (3 of 4) of
propranolol compared it with diuretics, which were shown
to be more diabetogenic in this meta-analysis, so any meaningful conclusion on the beneficial effects of propranolol
over atenolol or metoprolol cannot be made.
In contrast, CCBs and ACE inhibitors or ARBs resulted
in 21% and 23% reductions, respectively, in the risk for
new-onset DM compared with ␤ blockers. CCBs cause
vasodilation and improve peripheral blood flow, which may
increase or improve insulin sensitivity.39 ACE inhibitors, in
contrast, improve insulin sensitivity by improving skeletal
and peripheral blood flow, promoting adipocyte differentiation, and inhibiting the ␣-2 receptor–mediated impairment
of insulin secretion and glucose uptake.40,41 Previous studies
have questioned whether the excess risk for DM is secondary to a beneficial effect of other drugs on insulin sensitivity
rather than a deleterious effect of ␤ blockers. However,
given a higher risk for DM in patients taking ␤ blockers,
even compared with placebo, the adverse metabolic profile
of ␤ blockers cannot be ignored. In the recently concluded
Diabetes Reduction Assessment With Ramipril and Rosiglitazone Medication (DREAM) trial,42 the use of ramipril (an
ACE inhibitor) in patients with impaired fasting glucose or
with impaired glucose tolerance did not reduce the incidence of DM, further attesting to the fact that new-onset
DM with ␤ blockers may be due to the adverse effect of ␤
blockers rather than beneficial effects of other medications.
In patients with insulin resistance, hyperinsulinemia can
cause or exacerbate hypertension, possibly by renal sodium
retention.43 In addition, insulin resistance causes vascular
smooth muscle proliferation, impairs endothelium-dependent vasodilation, increases peripheral vascular resistance,
and augments vasoconstriction in response to norepinephrine and angiotensin II (Figure 8), all of which can lead to
or potentially exacerbate hypertension.44 Further evidence
for the role of insulin resistance in hypertension comes from
the fact that treating insulin resistance with insulin sensitizers (metformin, thiazolidinediones), increased physical activity, and weight loss reduces blood pressure.44 – 46 It therefore appears that hypertension (untreated or treated with ␤
blockers or thiazide diuretics) leads to an increase in insulin
resistance, and insulin resistance at least in part worsens
hypertension control, leading to a vicious cycle.
In this meta-analysis, we showed that in comparison with
other antihypertensive agents, the antihypertensive efficacy
of ␤ blockers was inferior. This is consistent with a previous
analysis showing similar results.6 Conceivably, this could
be in part related to insulin resistance and new-onset DM. In
this analysis, however, diuretics resulted in an increased risk
for new-onset DM compared with ␤ blockers, but their
blood pressure–lowering efficacy was superior to that of ␤
blockers. Diuretics result in natriuresis, potentially reversing some of these effects on renal sodium retention in
insulin-resistant patients, possibly explaining this paradox.
Beta blockers cause a number of metabolic derangements (Figure 8), which can potentially explain some of
these results. However, whether new-onset DM caused by
these medications has an adverse prognostic effect has been
debated. Some studies have shown that new-onset DM
associated with antihypertensive medications enhances the
risk for cardiovascular events. After a median of 6 years, the
risk for cardiovascular events with new-onset DM was
found to be similar to the risk in patients with established
DM and hypertension at baseline.7 Alderman et al,5 in a
follow-up of 6,886 hypertensive patients, found a significantly higher incidence of cardiovascular events in patients
with in-treatment blood glucose levels of ⱖ139.5 mg/dl. In
the 18-year follow-up of the Multiple Risk Factor Intervention Trial (MRFIT), patients who had developed DM during
treatment had greater mortality rates than those without
DM.47 Similarly, in a population-based cohort study of
1,860 men followed for 17.4 years, increased blood glucose
during treatment for hypertension (mainly by thiazide di-
Systemic Hypertension/␤ Blockers and New-Onset Diabetes
uretics and ␤ blockers) was an independent risk factor for
myocardial infarction.7
However, in a long-term study of SHEP, patients with
DM at baseline had worse prognoses (adjusted hazard ratio
1.66, 95% CI 1.41 to 1.95), patients with new-onset DM
while taking placebo had worse prognoses (adjusted hazard
ratio 1.56, 95% CI 1.12 to 2.18), but patients with new-onset
DM while taking chlorthalidone had no excess cardiovascular morbidity or mortality (adjusted hazard ratio 1.04,
95% CI 0.74 to 1.46) after follow up of 14.3 years,21 leading
some investigators to conclude that new-onset DM caused
by these medications might not be harmful. However, before we jump to any conclusion, these data must be analyzed
carefully and critically. There were no data on the blood
pressure difference between the groups at the end of 14.3
years of follow-up. It is possible that patients with newonset DM on active therapy had substantially lower blood
pressures that offset some of the adverse effects of DM. It
is also unknown if these patients were managed differently
for their DM. Given the limitations of this analysis and the
data from previous studies showing detrimental effects of
new-onset DM, the atherogenic potential of DM, whether
primary or secondary (to medications), cannot and should
not be ignored.
As in other meta-analyses, given the lack of data in each
trial, we did not adjust our analyses for the doses of medications used, add-on therapies used, body weight changes,
or compliance with assigned therapies. The definition of
new-onset DM and the diagnostic criteria used in the studies
were heterogenous. New-onset DM was a predefined end
point in only a few studies and was a post hoc analysis in the
others and suffers from the limitations of such analysis.
Given the limitation of meta-analyses, a causal relation
between new-onset DM and cardiovascular outcomes cannot be established.
1. Wolf-Maier K, Cooper RS, Banegas JR, Giampaoli S, Hense HW,
Joffres M, Kastarinen M, Poulter N, Primatesta P, Rodriguez-Artalejo
F, et al. Hypertension prevalence and blood pressure levels in 6
European countries, Canada, and the United States. JAMA 2003;289:
2363–2369.
2. National Institute of Diabetes and Digestive and Kidney Diseases.
National Diabetes Statistics Fact Sheet: General Information and National Estimates on Diabetes in the United States, 2005. Bethesda,
Maryland: U.S. Department of Health and Human Services, National
Institute of Health, 2006.
3. Gerber LM, Madhavan S, Alderman MH. Coincident hypertension.
Deleterious effects on patients with hyperglycemia. N Y State J Med
1983;83:693– 696.
4. Pell S, D’Alonzo CA. Some aspects of hypertension in diabetes mellitus. JAMA 1967;202:104 –110.
5. Alderman MH, Cohen H, Madhavan S. Diabetes and cardiovascular
events in hypertensive patients. Hypertension 1999;33:1130 –1134.
6. Davey Smith G, Bracha Y, Svendsen KH, Neaton JD, Haffner SM,
Kuller LH. Incidence of type 2 diabetes in the randomized multiple
risk factor intervention trial. Ann Intern Med 2005;142:313–322.
7. Dunder K, Lind L, Zethelius B, Berglund L, Lithell H. Increase in
blood glucose concentration during antihypertensive treatment as a
predictor of myocardial infarction: population based cohort study.
BMJ 2003;326:681.
8. Gress TW, Nieto FJ, Shahar E, Wofford MR, Brancati FL. Hypertension and antihypertensive therapy as risk factors for type 2 diabetes
mellitus. Atherosclerosis Risk in Communities Study. N Engl J Med
2000;342:905–912.
1261
9. Lithell H, Pollare T, Vessby B. Metabolic effects of pindolol and
propranolol in a double-blind cross-over study in hypertensive patients. Blood Press 1992;1:92–101.
10. Lithell HO. Effect of antihypertensive drugs on insulin, glucose, and
lipid metabolism. Diabetes Care 1991;14:203–209.
11. Pepine CJ, Handberg EM, Cooper-DeHoff RM, Marks RG, Kowey P,
Messerli FH, Mancia G, Cangiano JL, Garcia-Barreto D, Keltai M, et
al. A calcium antagonist vs a non-calcium antagonist hypertension
treatment strategy for patients with coronary artery disease. The International Verapamil-Trandolapril Study (INVEST): a randomized
controlled trial. JAMA 2003;290:2805–2816.
12. Berglund G, Andersson O, Widgren B. low-dose antihypertensive
treatment with a thiazide diuretic is not diabetogenic. A 10-year
controlled trial with bendroflumethiazide. Acta Med Scand 1986;220:
419 – 424.
13. Gurwitz JH, Bohn RL, Glynn RJ, Monane M, Mogun H, Avorn J.
Antihypertensive drug therapy and the initiation of treatment for diabetes mellitus. Ann Intern Med 1993;118:273–278.
14. Padwal R, Laupacis A. Antihypertensive therapy and incidence of type
2 diabetes: a systematic review. Diabetes Care 2004;27:247–255.
15. Padwal R, Mamdani M, Alter DA, Hux JE, Rothwell DM, Tu K,
Laupacis A. Antihypertensive therapy and incidence of type 2 diabetes
in an elderly cohort. Diabetes Care 2004;27:2458 –2463.
16. Bradburn MJ DJ, Altman D. Sbe24: metan—an alternative metaanalysis command. Stata Tech Bull Reprints 1998;8:86 –100.
17. Galbraith RF. A note on graphical presentation of estimated odds
ratios from several clinical trials. Stat Med 1988;7:889 – 894.
18. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin
Trials 1986;7:177–188.
19. Wikstrand J, Warnold I, Tuomilehto J, Olsson G, Barber HJ, Eliasson
K, Elmfeldt D, Jastrup B, Karatzas NB, Leer J, et al. Metoprolol versus
thiazide diuretics in hypertension. Morbidity results from the MAPHY
Study. Hypertension 1991;17:579 –588.
20. Wilhelmsen L, Berglund G, Elmfeldt D, Fitzsimons T, Holzgreve H,
Hosie J, Hornkvist PE, Pennert K, Tuomilehto J, Wedel H. Betablockers versus diuretics in hypertensive men: main results from the
HAPPHY trial. J Hypertens 1987;5:561–572.
21. Kostis JB, Wilson AC, Freudenberger RS, Cosgrove NM, Pressel SL,
Davis BR. Long-term effect of diuretic-based therapy on fatal outcomes in subjects with isolated systolic hypertension with and without
diabetes. Am J Cardiol 2005;95:29 –35.
22. Lindholm LH, Persson M, Alaupovic P, Carlberg B, Svensson A,
Samuelsson O. Metabolic outcome during 1 year in newly detected
hypertensives: results of the Antihypertensive Treatment and Lipid
Profile in a North of Sweden Efficacy Evaluation (ALPINE study).
J Hypertens 2003;21:1563–1574.
23. Medical Research Council Working Party. MRC Trial of Treatment of
Mild Hypertension: principal results. BMJ (Clin Res Ed) 1985;291:
97–104.
24. MRC Working Party. Medical Research Council Trial of Treatment of
Hypertension in Older Adults: principal results. BMJ 1992;304:405–
412.
25. Veterans Administration Cooperative Study Group on Antihypertensive Agents. Comparison of propranolol and hydrochlorothiazide for
the initial treatment of hypertension. II. Results of long-term therapy.
JAMA 1982;248:2004 –2011.
26. Dahlof B, Devereux RB, Kjeldsen SE, Julius S, Beevers G, de Faire U,
Fyhrquist F, Ibsen H, Kristiannson K, Lederballe-Pederson O, et al.
Cardiovascuar morbidity and mortality in the Losartan Intervention for
Endpoint reduction in hypertension study (LIFE): a randomised trial of
atenolol. Lancet 2002;359:995–1003.
27. Dahlof B, Sever PS, Poulter NR, Wedel H, Beevers DG, Caulfield M,
Collins R, Kjeldsen SE, Kristinsson A, McInnes GT, et al. Prevention
of cardiovascular events with an antihypertensive regimen of amlodipine adding perindopril as required versus atenolol adding bendroflumethiazide as required, in the Anglo-Scandinavian Cardiac Outcomes Trial-Blood Pressure Lowering Arm (ASCOT-BPLA): a
multicentre randomised controlled trial. Lancet 2005;366:895–906.
28. Hansson L, Hedner T, Lund-Johansen P, Kjeldsen SE, Lindholm LH,
Syvertsen JO, Lanke J, de Faire U, Dahlof B, Karlberg BE. Randomised trial of effects of calcium antagonists compared with diuretics and
beta-blockers on cardiovascular morbidity and mortality in hypertension: the Nordic Diltiazem (NORDIL) study. Lancet 2000;356:359 –
365.
1262
The American Journal of Cardiology (www.AJConline.org)
29. Hansson L, Lindholm LH, Ekbom T, Dahlof B, Lanke J, Schersten B,
Wester PO, Hedner T, de Faire U. Randomised trial of old and new
antihypertensive drugs in elderly patients: cardiovascular mortality
and morbidity the Swedish Trial in Old Patients With Hypertension-2
study. Lancet 1999;354:1751–1756.
30. Hansson L, Lindholm LH, Niskanen L, Lanke J, Hedner T, Niklason
A, Luomanmaki K, Dahlof B, de Faire U, Morlin C, et al. Effect of
angiotensin-converting-enzyme inhibition compared with conventional therapy on cardiovascular morbidity and mortality in hypertension: the Captopril Prevention Project (CAPPP) randomised trial. Lancet 1999;353:611– 616.
31. Thornley-Brown D, Wang X, Wright JT Jr, Randall OS, Miller ER,
Lash JP, Gassman J, Contreras G, Appel LJ, Agodoa LY, Cheek D.
Differing effects of antihypertensive drugs on the incidence of diabetes
mellitus among patients with hypertensive kidney disease. Arch Intern
Med 2006;166:797– 805.
32. Ferrannini E, Buzzigoli G, Bonadonna R, Giorico MA, Oleggini M,
Graziadei L, Pedrinelli R, Brandi L, Bevilacqua S. Insulin resistance in
essential hypertension. N Engl J Med 1987;317:350 –357.
33. Lissner L, Bengtsson C, Lapidus L, Kristjansson K, Wedel H. Fasting
insulin in relation to subsequent blood pressure changes and hypertension in women. Hypertension 1992;20:797– 801.
34. Taittonen L, Uhari M, Nuutinen M, Turtinen J, Pokka T, Akerblom
HK. Insulin and blood pressure among healthy children. Cardiovascular risk in young Finns. Am J Hypertens 1996;9:194 –199.
35. Fajans SS, Floyd JC Jr, Knopf RF, Rull J, Guntsche EM, Conn JW.
Benzothiadiazine suppression of insulin release from normal and abnormal islet tissue in man. J Clin Invest 1966;45:481– 492.
36. Sharma AM, Pischon T, Hardt S, Kunz I, Luft FC. Hypothesis:
beta-adrenergic receptor blockers and weight gain: a systematic analysis. Hypertension 2001;37:250 –254.
37. Pollare T, Lithell H, Selinus I, Berne C. Sensitivity to insulin during
treatment with atenolol and metoprolol: a randomised, double blind
study of effects on carbohydrate and lipoprotein metabolism in hypertensive patients. BMJ 1989;298:1152–1157.
38. Bakris GL, Fonseca V, Katholi RE, McGill JB, Messerli FH, Phillips
RA, Raskin P, Wright JT Jr, Oakes R, Lukas MA, et al. Metabolic
effects of carvedilol vs metoprolol in patients with type 2 diabetes
mellitus and hypertension: a randomized controlled trial. JAMA 2004;
292:2227–2236.
39. Pitre M, Gaudreault N, Santure M, Nadeau A, Bachelard H. Isradipine
and insulin sensitivity in hypertensive rats. Am J Physiol 1999;276:
E1038 –E1048.
40. Deibert DC, DeFronzo RA. Epinephrine-induced insulin resistance in
man. J Clin Invest 1980;65:717–721.
41. Lithell H. Metabolic effects of antihypertensive drugs interacting
with the sympathetic nervous system. Eur Heart J 1992;13(suppl):
53–57.
42. DREAM Trial Investigators, Bosch J, Yusuf S, Gerstein HC, Pogue J,
Sheridan P, Dagenais G, Diaz R, Avezum A, Lanas F, et al. Effect of
ramipril on the incidence of diabetes. N Engl J Med 2006;355:1551–
1562.
43. Zavaroni I, Coruzzi P, Bonini L, Mossini GL, Musiari L, Gasparini P,
Fantuzzi M, Reaven GM. Association between salt sensitivity and
insulin concentrations in patients with hypertension. Am J Hypertens
1995;8:855– 858.
44. Su HY, Sheu WH, Chin HM, Jeng CY, Chen YD, Reaven GM. Effect
of weight loss on blood pressure and insulin resistance in normotensive
and hypertensive obese individuals. Am J Hypertens 1995;8:1067–
1071.
45. Giugliano D, De Rosa N, Di Maro G, Marfella R, Acampora R,
Buoninconti R, D’Onofrio F. Metformin improves glucose, lipid metabolism, and reduces blood pressure in hypertensive, obese women.
Diabetes Care 1993;16:1387–1390.
46. Ogihara T, Rakugi H, Ikegami H, Mikami H, Masuo K. Enhancement
of insulin sensitivity by troglitazone lowers blood pressure in diabetic
hypertensives. Am J Hypertens 1995;8:316 –320.
47. Eberly LE, Cohen JD, Prineas R, Yang L. Impact of incident diabetes
and incident nonfatal cardiovascular disease on 18-year mortality: the
multiple risk factor intervention trial experience. Diabetes Care 2003;
26:848 – 854.