Alimentary Pharmacology and Therapeutics

Systematic review with meta-analysis: the association between

the use of calcium channel blockers and gastrointestinal bleeding
Y. He*, E. W. Chan*, W. K. Leung, S. Anand* & I. C. K. Wong*

*Department of Pharmacology and

Pharmacy, Centre for Safe Medication
Practice and Research, The University
of Hong Kong, Hong Kong.

Department of Medicine, Li Ka Shing

Faculty of Medicine, The University of
Hong Kong, Queen Mary Hospital,
Hong Kong.

Correspondence to:
Prof. I. C. K. Wong, Department of
Pharmacology and Pharmacy, Centre
for Safe Medication Practice and
Research, The University of Hong
Kong, 2/F Laboratory Block, 21
Sassoon Road, Hong Kong.
E-mail: wongick@hku.hk

Publication data
Submitted 6 February 2015
First decision 17 February 2015
Resubmitted 19 March 2015
Resubmitted 1 April 2015
Accepted 1 April 2015
As part of AP&Ts peer-review process, a
technical check of this meta-analysis
was performed by Dr Y. Yuan. This
article was accepted for publication after
full peer-review.

SUMMARY
Background
Due to their potential anti-platelet effect, it is suggested that calcium channel blockers (CCBs) are associated with gastrointestinal bleeding (GIB).
However, results from previous studies are conicting.
Aim
To conduct a systematic review and meta-analysis of randomised controlled
trials (RCTs) and observational studies to clarify the association between
CCBs and GIB.
Methods
We conducted a systematic search of PubMed, EMBASE, Cochrane library
and Trial Register databases up to January 2015. Studies that evaluated
exposure to CCBs reporting GIB outcomes were included in the metaanalysis. The inverse variance method with random effects model was used
to calculate the pooled estimates.
Results
Seventeen studies (four RCTs, eleven casecontrol and two cohort studies)
were included in the meta-analysis. The summary risk ratio (RR) for GIB
was 1.17 (95% CI 1.011.36) for CCB users vs. non-users. Subgroup analysis showed that CCB use was associated with a moderately higher risk of
lower GIB (RR = 1.83, 95% CI 1.172.84) but not upper GIB. However,
data from four RCTs did not support association between CCBs and GIB
(RR = 0.93, 95% CI 0.821.05). Subgroup analyses further showed that the
increased risk of GIB among CCB users was only observed in studies that
failed to adjust for prior history of GIB (RR = 1.67, 95% CI 1.342.08) or
use of anti-ulcer drugs (RR = 1.40, 95% CI 1.191.65).
Conclusion
Our meta-analysis showed a marginal association between calcium channel
blocker use and the risk of gastrointestinal bleeding. This association is of
dubious clinical signicance, as the effects of different comparators or
adjustment for confounding factors render this association nonsignicant.
Aliment Pharmacol Ther

2015 John Wiley & Sons Ltd

doi:10.1111/apt.13211

Y. He et al.
INTRODUCTION
Calcium channel blockers (CCBs) inhibit calcium transport mainly by blocking the voltage-gated calcium channel in the cellular membrane of vascular and cardiac
cells, which leads to vasodilation and a reduction in
heart rate.1 The CCBs are a class of medication that
plays an important role in the clinical management of
cardiovascular and cerebrovascular diseases including
hypertension, angina pectoris, cardiac arrhythmias, vasospasm and ischaemic neurological decits. Apart from
the usual reported adverse reactions of CCBs,2, 3 there is
a possibility of an increased risk of gastrointestinal bleeding (GIB) due to their potential anti-platelet effect. This
is further supported by animal, in vivo and ex vivo studies that demonstrated the inhibition of platelet aggregability of CCBs in putative pro-haemorrhagic activity.49
The promotion of vasodilation by CCBs also opposes the
usual vasoconstriction response to bleeding.10, 11 Therefore, it is biologically plausible that CCBs potentially
increases the risk of GIB.
In the past two decades, several studies have been
conducted to investigate the efcacy and safety of
CCBs.1229 However, the ndings with respect to the
association between CCB use and GIB risk are conicting. Due to the inconclusive ndings, and to better
understand this question, we conducted this systematic
review with meta-analysis of existing randomised controlled trials (RCTs) and observational studies that investigated the association between CCB use and the risk of
developing GIB.
METHODS
This systematic review was conducted following guidance
provided by the Cochrane Handbook30 and is reported
in accordance with the Preferred Reporting Items for
Systematic reviews and Meta-Analyses (PRISMA)31 and
the Meta-analysis Of Observational Studies in Epidemiology (MOOSE).32
Data sources and search strategy
A systematic literature search on PubMed, EMBASE and
the Cochrane library was performed using key words,
MeSH and Emtree terms. The following search terms
were used; calcium antagonist, calcium channel blockers, names of specic calcium antagonists combined
with gastrointestinal haemorrhage, peptic ulcer disease
and other subtypes/synonyms for gastrointestinal bleeding. Trial registers; the Clinical trials government (http://
www.ClinicalTrials.gov), the World Health Organisation
International Clinical Trials Registry Platform (ICTRP)
2

(http://www.who.int/ictrp/en/) and the metaRegister of

Controlled Trials (http://www.controlled-trials.com/)
were also searched to identify potentially relevant studies.
All databases were searched till 12 January 2015. Duplicates were removed and titles, abstracts and article contents were screened to determine whether the inclusion/
exclusion criteria were met. The bibliographies of some
relevant review articles were also searched to identify
any additional studies.

Study selection
Studies included in this meta-analysis were either RCTs
or observational studies that investigated the association
between CCB use and the risk of GIB. Studies were
included if they: (i) clearly dened exposure to CCBs
and other comparative exposure groups; (ii) clearly
dened the outcome of GIB; (iii) reported the GIB
patient number and total patient number of each comparative group for RCTs; (iv) reported relative risks (RR)
(cohort studies) or odds ratios (OR) (casecontrol studies) or provided data for calculation of RR/OR. Only
data from the most recent comprehensive reports were
included when multiple publications were from the same
study. There were no restrictions on study size or language. Conference proceedings were excluded as we were
unable to assess the quality of these studies. Studies were
excluded if there was insufcient data for determining
the RR/OR and the 95% condence intervals (CI); or if
it was not conducted using a classical observational study
design. However, these studies were still included in the
narrative review and discussed qualitatively.
Quality assessment
The methodological quality of the included RCTs was
assessed using the Cochrane Collaborations tool for
assessing the risk of bias.33 The study was regarded as
low risk of bias for selective outcome reporting, provided that the outcome of GIB was pre-specied and
reported in the study. The included cohort and case
control studies were assessed for methodological quality
using the NewcastleOttawa scale34 as recommended by
the Cochrane Collaboration. This scale provides two sets
of specic criteria for cohort and casecontrol studies, to
assess the study selection (four items), comparability
(two items) and ascertainment of exposure/outcome
(three items). The high quality items were scored with
a star. A study can be awarded a maximum of one star
for each item within the selection and exposure/outcome
categories, while a maximum of two stars can be given
for comparability category. The maximum score would
Aliment Pharmacol Ther
2015 John Wiley & Sons Ltd

Systematic review with meta-analysis: calcium channel blockers & gastrointestinal bleeding
be nine. A nal score 7 was considered as high quality.
The quality assessment of included studies was performed independently by two researchers (YH and SA).
Any discrepancies were addressed by re-evaluation to
reach consensus (Table S1).

Data abstraction
YH performed the initial searches, screened abstracts
and full texts for eligibility. Study characteristics and
measure of effect were extracted from the included studies which are summarised in Table S2 and S3. For each
study, the risk estimates that were adjusted for the largest number of confounding variables were extracted for
analysis. For studies that did not report an adjusted
result,25, 28 the unadjusted result was extracted for analysis. If the unadjusted estimates were not directly
reported, patient and event number were used to calculate the OR/RR using the calculator of Review Manager
5.3.35
Statistical methods
The main analysis focused on assessing the risk of GIB
among CCB users. RRs and ORs with their 95% CIs in
the observational studies were extracted and log-transformed for the meta-analysis. For the studies that only
reported ORs, the adjusted ORs were used to summarise
the pooled RR. Total number of persons in each group
(exposed vs. not exposed to CCBs) were extracted for
the analysis of RCTs. GIB data from RCTs were analysed
according to participants randomised treatment assignments regardless of their subsequent status (i.e. intention-to-treat analysis). The inverse variance method with
random effects model was used to calculate summary RR
and 95% CI. Heterogeneity was assessed using Cochrans
Q statistical test and P < 0.10 was considered signicant.
I2 statistic was also calculated to estimate the proportion
of total variation among studies, where a value of 25%,
50% and 75% was regarded as low, moderate and high
heterogeneous, retrospectively.36
Subgroup analyses stratied according to study characteristics were carried out to investigate the source of
heterogeneity, which included study design (cohort,
casecontrol and RCT), different comparison groups
(beta-blocker, non-CCB and non-use of any anti-hypertensive agents), study quality (high vs. low), and the
sources of GIB including upper GIB (UGIB), lower GIB
(LGIB), peptic ulcer bleeding (PUB) and all GIB. The
classication of the sources of GIB was based on the
description in the original studies. If the subtypes of
GIB by location were not specied, the outcome was
Aliment Pharmacol Ther
2015 John Wiley & Sons Ltd

considered as all GIB. UGIB was dened as bleeding

due to a peptic lesion (including ulcers, erosions and
acute mucosal lesions) in the upper GI tract. LGIB was
dened as bleeding from the lower GI tract, including
colonic diverticular bleeding and bleeding of rectum or
anus. PUB was dened as bleeding due to peptic ulcers.
Diagnosis of GIB was generally identied by International Classication of Diseases, Ninth Revision, Clinical
Modication (ICD-9-CM) codes, and conrmed by
endoscopy or validated by medical records/interview.
Studies were further stratied based on whether there
was adjustment of certain confounders including the use
of drugs that were associated with GIB (aspirin/antiplatelet agents, nonsteroidal anti-inammatory drugs
(NSAIDs), corticosteroids, anti-coagulants and anti-ulcer
drugs) and history of GIB. Kelly et al.17 and Suissa
et al.18 excluded subjects with a history of GIB and
therefore, were considered as having adjusted for the
confounder of a history of GIB. The four RCTs1214, 24
were considered as having adjusted for all mentioned
confounders due to the intrinsic characteristic of RCT
study design.
Publication bias was assessed qualitatively using a funnel plot and quantitatively using the Egger test as the
number of included studies exceeded 10, with P < 0.10
indicating signicant publication bias. Statistical analyses
were conducted using Review Manager 5.335 and StatsDirect.37 P-values (two-tailed) <0.05 was regarded as statistically signicant, except for heterogeneity and
publication bias tests.

RESULTS
Search results
A total of 1508 records were retrieved from the literature search. Titles and abstracts were screened and full
texts of 55 articles were further reviewed. Eighteen
studies met the criteria for this systematic review
(Figure 1). Three publications were from the same
clinical trial,12, 38, 39 and hence only the most recent
comprehensive report with a validated outcome was
included.12
Characteristics and quality of included studies
Four RCTs were conducted in Europe, North America
and collaboratively in multiple centres (Europe, North
America and Mid-East), respectively. Of the four RCTs,
three studies13, 14, 24 compared the use of CCB vs. placebo, whereas Phillips et al.12 compared CCB vs. angiotensin-converting enzyme inhibitor (ACEi) or diuretics.
3

Y. He et al.
1012 records identified through
database searching (PubMed +
Embase)

496 records identified through

Cochrane library, clinical trial
registries and other sources

Screening
450 duplicates excluded, 1058 records remained

1058 records screened

1003 records excluded: basic

science/animal studies, casereports, not related to GIB,
reviews

Eligibility
55 full-text articles assessed for
eligibility

37 records excluded:
conference abstracts, not
reporting GIB results

Included
18 studies included in the systematic review (17 included in meta-analysis)

Figure 1 | PRISMA ow chart summarising study identication and selection.

The observational studies were conducted in Asia, North

America and Europe (Table S2). Desboeuf et al.23
adopted a very different study design using French spontaneous reporting data, i.e. case/non case study and
therefore was discussed separately in the narrative
review. Of the remaining 13 observational studies, ten
casecontrols and one cohort study compared CCB with
non-use; the remaining two specically compared CCB
with beta-blockers.
The overall risk of bias of the four RCTs assessed
in accordance to Cochrane Collaborations tool was
low in all subcategories of criteria. As shown in Table
S1 and S2, the quality scores of the observational studies ranged from 5 to 8. Nine casecontrol and two
cohort studies were considered high quality (NewcastleOttawa score 7). Most of the studies adjusted or
matched for demographical variables, and several also
accounted for other potential confounders, including
the use of drugs that are associated with GIB (NSAIDs, aspirin, corticosteroid and anti-coagulants), antiulcer agents or other anti-hypertensive drugs, as well
as comorbidities (cardiovascular and cerebrovascular
diseases, GIB history). For the RCTs, exposure ascertainment was obtained through drug dispensing and
medical records. The observational studies relied on
medical diagnostic codes, hospital medical records or
structured interviews to ascertain GIB outcome as well
as drug exposure.
4

Risk of gastrointestinal bleeding

RCTs. Four RCTs were included in the meta-analysis.
The pooled estimated RR was 0.93 (95% CI 0.821.05),
showing no signicant difference of GIB risk between
CCB users and non-users (Figure 2). Phillips et al. conducted a post hoc analysis of the ALLHAT trial12 and
reported that the GIB risk among CCB users was lower
compared with ACEi users but similar to diuretics. Combining both ACEi and diuretics as the non-CCB comparison group, the estimated RR of GIB among CCB users
compared with non-users was 0.88 (95% CI 0.761.02).
Observational studies. Eleven casecontrol and two
cohort studies were included in the primary meta-analysis (Figure 2). In summary, the pooled estimated RRs of
casecontrol and cohort studies were 1.20 (95% CI 1.00
1.45) and 1.44 (95% CI 0.862.40) respectively. Pahor
et al.22 reported the rst prospective cohort study, where
CCB was associated with an increase in risk of GIB compared with beta-blockers among elderly patients in the
USA (RR = 1.86, 95% CI 1.232.82). Interestingly, three
casecontrol studies25, 26, 28 found that there was a trend
of a higher risk of colonic diverticular bleeding (a common LGIB) among CCB users compared with non-users.
Furthermore, a casecontrol study investigating GIB of
different sources by Kaplan et al.16 also supported the
association for all GIB (RR = 2.03, 95% CI 1.313.14)
and LGIB (RR = 2.56, 95% CI 1.086.05), whereas the
Aliment Pharmacol Ther
2015 John Wiley & Sons Ltd

Systematic review with meta-analysis: calcium channel blockers & gastrointestinal bleeding
Risk Ratio
Study or Subgroup
log [Risk Ratio]
SE Weight
IV, Random, 95% CI
2.1.1 RCT
Cellis 2001
0.207 0.5035
2.0%
1.23 [0.46, 3.30]
Petruk 1988
0.6293 1.216
0.4%
0.53 [0.05, 5.78]
Phillips 2013
0.1316 0.0751 11.8%
0.88 [0.76, 1.02]
0.0862 0.1277
Poole-Wilson 2006
9.7%
1.09 [0.85, 1.40]
Subtotal (95% CI)
23.9%
0.93 [0.82, 1.05]
Heterogeneity: Tau2 = 0.00; Chi2 = 2.68; df = 3 (P = 0.44); I2 = 0%
Test for overall effect: Z = 1.13 (P = 0.26)

Risk Ratio
IV, Random, 95% CI

2.1.2 Case-Control
0.6366 0.4538
2.3%
1.89 [0.78, 4.60]
Jansen 2009
0.708 0.2225
6.3%
2.03 [1.31, 3.14]
Kalpan 2000
0.1823 0.1468
8.9%
1.20 [0.90, 1.60]
Kelly 1999
0.1054 0.1468
8.9%
0.90 [0.67, 1.20]
Lanas 2003
0.3567 0.275
4.9%
0.70 [0.41, 1.20]
Nagata 2015
0.6259 0.4377
2.5%
1.87 [0.79, 4.41]
Pillotto 1996
0.3365 0.1282
9.7%
1.40 [1.09, 1.80]
Rodriguez 1998
0 0.5271
1.8%
1.00 [0.36, 2.81]
Suh 2012
0.0583 0.1528
8.7%
1.06 [0.79, 1.43]
Suissa 1998
0.0408 0.2209
6.3%
0.96 [0.62, 1.48]
Weil 2000
0.6419 0.4047
2.8%
1.90 [0.86, 4.20]
Yamada 2008
63.3%
1.20 [1.00, 1.45]
Subtotal (95% CI)
Heterogeneity: Tau2 = 0.04; Chi2 =19.78; df = 10 (P = 0.03); I2 = 49%
Test for overall effect: Z = 1.94 (P = 0.05)
2.1.3 Cohort
0.6206 0.2123
6.6%
Pahor 1996
1.86 [1.23, 2.82]
0.0953 0.2221
6.3%
Smalley 1998
1.10 [0.71, 1.70]
12.9%
1.44 [0.86, 2.40]
Subtotal (95% CI)
Heterogeneity: Tau2 = 0.09; Chi2 = 2.92; df = 1 (P = 0.09); I2 = 66%
Test for overall effect: Z = 1.38 (P = 0.17)
100.0%
1.17 [1.01, 1.36]
Total (95% CI)
2
2
Heterogeneity: Tau = 0.04; Chi = 36.73; df = 16 (P = 0.002); I2 = 56%
0.1
Test for overall effect: Z = 2.07 (P = 0.04)
Test for subgroup differences: Chi2 = 6.78; df = 2 (P = 0.03); I2 = 70.5%

0.2

0.5

Favours CCB

10

Favours non-CCB

Figure 2 | Summarised estimates of primary analysis of GIB risk among users of CCB vs. non-CCB.

association was found to be nonsignicant for UGIB

(RR = 1.54, 95% CI 0.922.59) and PUB (RR = 1.17,
95% CI 0.622.21).

All studies. Based on the meta-analysis of all studies (4

RCTs, 11 casecontrol and 2 cohort studies) (Figure 2),
the use of CCB was marginally associated with an
increased risk of GIB, with an RR of 1.17 (95% CI 1.01
1.36). There was, however, considerable heterogeneity
observed across studies (Cochrans Q-test, P = 0.002;
I2 = 56%).
Narrative review. Desboeuf et al.23 used the case/non
case methodology to investigate the association between
CCB use and the risk of GIB by analysing adverse drug
reaction (ADR) reports in the French pharmacovigilance
system database. Cases were dened as GIB reports
Aliment Pharmacol Ther
2015 John Wiley & Sons Ltd

recorded in the spontaneous system, while non cases

were all other ADRs recorded. Exposure was considered
as the presence in an ADR report of CCB. The association between CCB use and GIB risk was not supported
(OR = 1.2, 95% CI 0.91.6).

Subgroup analysis
By comparison groups. The comparison groups of the
studies included beta-blockers, non-CCB (use of drugs
other than CCB, including other anti-hypertensive
agents) and non-use of any anti-hypertensive agents. The
pooled result of Kaplan et al.16 and Pahor et al.22
showed that compared with beta-blockers, CCB use was
signicantly associated with an increased risk of GIB
(RR = 1.94, 95% CI 1.442.62). Comparison to non-use
of any anti-hypertensive agents yielded a marginally signicant association (RR = 1.17, 95% CI 1.021.35).
5

Y. He et al.
However, the association between CCB use and GIB was
not signicant when compared with non-CCB
(RR = 1.02, 95% CI 0.851.21; Table 1 and Figure S1).

By sources of GIB. Of all included studies, Kaplan

et al.16 reported the outcomes of UGIB, LGIB, PUB and
all GIB, respectively, and the corresponding RRs were
extracted for subgroup analyses. Of the remaining 16
studies, there were four, three, three and six studies that
reported UGIB, LGIB, PUB and all GIB as outcomes
respectively. Results showed that CCB use was associated
with a signicantly increased risk of LGIB with an RR of
1.83 (95% CI 1.172.84). There was no signicant difference in the risks of UGIB (RR = 1.23, 95% CI 0.99
1.53), PUB (RR = 0.97, 95% CI 0.761.24) or all GIB
(RR = 1.23, 95% CI 0.941.61) between CCB and nonusers (Table 1, Figure S2).

By confounders adjustment. Subgroup analyses were also

carried out to examine the impact of adjustment on confounders, particularly on the history of GIB and use of
other drugs that would alter the risk of bleeding. An
increased risk of GIB were observed (RR = 1.67, 95% CI
1.342.08) in studies that failed to adjust for history of
GIB, whereas adjustment of history of GIB renders this
association not signicant (Table 1, Figure S3). Interestingly, stratied analysis of studies that failed to adjust
for the use of anti-ulcer drugs revealed an increased risk
of GIB (RR = 1.40, 95% CI 1.191.65, Table 1, Figure S4)
but not in those studies that adjusted for this confounder. Among subgroups that have adjusted or not
adjusted for the use of NSAIDs, aspirin/anti-platelet
agents or corticosteroids/anti-coagulants, no signicant
association between CCB use and GIB risk was shown
(Table 1, Figures S5S7).

Table 1 | Subgroup analyses of the GIB risk among users of CCB vs. non-CCB
Group variable
All studies
Comparison group*

Sources of GIB

Study quality
Adjustment for confounders
History of GIB**
Anti-ulcer drug
NSAID
Aspirin/anti-platelet agents
Corticosteroids/anti-coagulants

Subgroups

No. of studies

Adjusted RR (95% CI)

Heterogeneity between groups

Beta-blockers
Non-use
Non-CCB
Upper GIB
Lower GIB
PUB
All GIB
High
Low

17
2
6
9
5
4
4
7
15
2

1.17
1.94
1.17
1.02
1.23
1.83
0.97
1.23
1.17
1.21

(1.011.36)
(1.442.62)
(1.021.35)
(0.851.21)
(0.991.53)
(1.172.84)
(0.761.24)
(0.941.61)
(0.991.38)
(0.761.95)

P = 0.001, I2 = 84.9%

Yes
No
Yes
No
Yes
No
Yes
No
Yes
No

11
6
7
10
14
3
9
8
8
9

1.03
1.67
0.93
1.40
1.17
1.32
1.14
1.22
1.21
1.15

(0.911.16)
(1.342.08)
(0.841.03)
(1.191.65)
(0.991.38)
(0.742.34)
(0.921.41)
(0.971.53)
(0.951.53)
(0.941.40)

P = 0.09, I2 = 53.9%

P = 0.89, I2 = 0%
P = 0.0002, I2 = 92.8%
P < 0.0001, I2 = 94.0%
P = 0.70, I2 = 0%
P = 0.66, I2 = 0%
P = 0.75, I2 = 0%

CI, condence interval; CCB, calcium channel blockers; GIB, gastrointestinal bleeding; NSAID, nonsteroidal anti-inammatory drug;
PUB, peptic ulcer bleeding; RR, risk ratio.
* Non-use indicates patients in this group did not receive any other anti-hypertensive drugs; Non-CCB indicates patients in this
group might receive other drugs including anti-hypertensive agents.
Kaplan et al. reported stratied outcomes of upper GIB, lower GIB and PUB.
A total score 7 of NewcastleOttawa scale was considered as high quality for observational studies; The four RCTs were of
low risk of bias and considered as with high quality.
The four RCTs were considered as having adjusted for all mentioned confounders.
** Kelly et al. & Suissa et al. excluded subjects with GIB history and were considered as with adjustment of GIB history.
6

Aliment Pharmacol Ther

2015 John Wiley & Sons Ltd

Systematic review with meta-analysis: calcium channel blockers & gastrointestinal bleeding
By study quality. Subgroup analysis by study quality
showed no signicant difference between pooled estimates of high-quality and low-quality studies (Table 1,
Figure S8). Excluding the two studies with quality scores
<7 render the overall association nonsignicant
(RR = 1.17, 95% CI 0.991.38).
Publication bias
The funnel plot basically shows symmetrical distribution
of the 17 included studies, with P value of 0.11 for
Eggers test, indicating that there was no evidence of signicant publication bias (Figure S9).
DISCUSSION
To our knowledge, this is the rst meta-analysis of all
the RCTs and observational studies investigating the
association of CCB use and the risk of GIB. We undertook a rigorous systematic review and meta-analysis with
data extraction and statistical analysis by independent
reviewers. From our results, CCB use showed a weak
association with an increased GIB risk compared with
non-CCB use, with an RR of 1.17 (95% CI 1.011.36).
From our meta-analysis, there was a signicant heterogeneity between studies. Three studies (two case
controls16, 20 and one cohort22) reported signicant association between CCB use and GIB risk, whereas the
remaining studies showed no association. Of the three
studies, Garcia Rodriguez et al.20 found that current use
of CCBs were associated with GIB (RR = 1.7, 95% CI
1.32.1). However, the association was less obvious after
further adjustment for the use of aspirin and anti-platelet
drugs, various markers of comorbidity including number
of previous hospitalisations, use of beta-agonists, insulin,
oral hypoglycaemic drugs and cardiovascular drugs
(RR = 1.4, 95% CI 1.11.8).20 In addition, the study
claimed that past CCB users of more than 1 year prior
to the episode of GIB already presented a slightly
increased risk of GIB (RR = 1.5, 95% CI 1.31.8); hence,
an acute effect of CCB use is not supported. The authors
did not exclude the possibility of uncontrolled confounding or selection bias that may affect the accuracy of
results.
Moreover, the two studies using beta-blockers as the
comparator group16, 22 may explain the major reason for
heterogeneity between studies. Subgroup analysis of these
two studies showed that users of CCB were 1.9-fold
(95% CI 1.442.62) more likely to develop GIB compared with users of beta-blockers, while this association
was nonsignicant in subgroups with unspecied comparators. Since, beta-blockers can reduce cardiac output
Aliment Pharmacol Ther
2015 John Wiley & Sons Ltd

by blocking b1 receptors and induce splanchnic vasoconstriction by blocking b2 receptors,40 this may be a biologically plausible mechanism for the protective effects of
GIB. Indeed, beta-blockers have been shown to prevent
variceal GIB.41, 42 Findings from two studies included in
our meta-analysis may imply that beta-blockers also
could prevent nonvariceal GIB. Suissa et al.18 investigated the association of the use of different anti-hypertensive drugs including ACE inhibitors, CCBs, betablockers and diuretics compared with non-use. They
found that beta-blockers demonstrated a lower risk of
hospitalisation for GIB, with an RR of 0.66 (0.440.98),
while the association of CCB and GIB was nonsignicant, with an RR of 1.06 (0.791.43). Garcia Rodriguez
et al.20 also reported that beta-blockers were associated
with the lowest GIB risk (RR = 1.0, 95% CI 0.71.4) of
all anti-hypertensive drugs compared with non-users.
Interestingly, subgroup analyses by different sources of
GIB showed that CCB use was associated with an
increased risk of LGIB, and there was no signicant
association with UGIB or PUB. Our nding is consistent
with a recent meta-analysis by Kvasnovsky et al.43 which
showed that CCB use was signicantly associated with
an increased risk of colonic diverticular bleeding
(OR = 2.50, 95% CI 1.444.35). Besides, Pilotto et al.21
also commented that the positive association between
CCB use and GIB risk found by Pahor et al.22 may be
due to the inclusion of bleeding outcomes related to cancers or diverticula of the intestine, rather than peptic
ulcers or gastritis. These imply that CCB use may be
more associated with LGIB than UGIB/PUB. However, it
is difcult to draw rm conclusion due to limited number of published studies and potential effects of different
confounding factors. Further investigation is warranted.
In addition, it is worth noting that adjustment for
GIB-associated drugs (aspirin/anti-platelet agents and
anti-ulcer drugs) attenuated the association between
CCB use and the risk of GIB. The positive nding of an
increased risk of GIB in CCB users were only found in
studies that failed to adjust for the use of anti-ulcer
drugs or history of GIB. As the use of anti-ulcer drugs
or whether having a history of GIB would alter the risk
of developing GIB, the positive association may be due
to inadequate adjustment of certain important confounding factors in observational studies.
A case/non case study by Desboeuf et al.23 did not
support the association between CCB use and GIB. One
of the disadvantages of this case/non case study was due
to the inherent limitations of spontaneous reporting system by physicians, including underreporting as well as
7

Y. He et al.
notoriety bias. Moreover, it was not feasible to verify
matching between cases and comparators. Nevertheless,
the authors used NSAIDs and salicylates as positive controls to test the robustness of this method and results
were reasonable.
Several studies investigating the association of CCB
use and GIB risk were not included in the meta-analysis
for the following reasons. Zuccala et al.44 reported that
treatment of CCBs was associated with a reduction in
haemoglobin level during patients hospital stay, rather
than actual gastrointestinal bleeding. A randomised double-blind parallel group study by Smith et al.4 also found
that CCBs may have minor indirect anti-platelet function, as there was signicant decrease in beta-thromboglombulin level. However, Islim et al.45 reported that
untreated hypertensive patients had signicantly higher
levels of beta-thromboglombulin than normotensive controls. This suggested that the anti-platelet function of
CCBs in Smith et al.s study is of dubious clinical importance, since it may be due to an effect of lowered blood
pressure rather than a direct drug effect. Nevertheless,
the International Verapamil SR-Trandolapril Study
(INVEST) trial46 claimed that there was no signicant
association of CCB use and GIB risk compared with use
of beta-blockers, however, no detailed data on GIB was
presented. In addition, there are a few studies published
in abstract form that suggest a potential association
between CCB use and bleeding. Kuipers et al.47 conducted a systematic literature analysis investigating the
risk factors for UGIB in patients using low-dose acetylsalicylic acid, and found that concomitant CCB use were
associated with an increased risk of UGIB (one study),
with an OR of 2.53 (95% CI 1.255.14). A US community-based casecontrol study by Jalil et al.48 investigated
the risk factors for diverticular bleeding (the most common cause of LGIB) and found that CCB was a signicant risk factor (P = 0.009) of LGIB.

Strengths and limitations. We included all published

studies to date with available GIB outcome data in this
meta-analysis. The four RCTs included were of high
quality and low risk of bias. Included observational
studies were of an adequate quality of study design, as
indicated by the NewcastleOttawa Scale quality assessment. This review has several limitations. First, this
meta-analysis compiles the results from the available
studies and attempts to draw a conclusion. The
included studies did not adjust for the same confounders; hence, no direct comparisons can be made. We
have performed subgroup analysis by confounder
8

adjustments and some heterogeneity can be partially

overcome. The variability between studies was unavoidable and the study conclusions should be evaluated
alongside the reported heterogeneity. Nevertheless, we
conducted subgroup analyses to examine the impact of
the heterogeneity between studies, including stratication based on different study quality, comparison
groups and sources of GIB. Second, several casecontrol
studies reported RRs instead of ORs, and the exact statistical method was not clearly described. We were not
able to exclude the inuence on results by combining
RRs and ORs in the meta-analysis.

CONCLUSIONS
In conclusion, our meta-analysis suggests a marginal
association between CCB use and the risk of GIB, which
is of dubious clinical signicance, as evidence from the
RCTs did not support such an association and the effects
of different comparators or adjustment for confounding
factors render this association nonsignicant. Based on
the evidence from this meta-analysis, further studies are
needed to investigate the potential protective effect of
beta-blockers on GIB, as well as the association between
CCB use and the risk of LGIB.
AUTHORSHIP
Guarantor of the article: YH and ICKW.
Author contributions: YH and ICKW had the original
idea for this study and contributed to the development
of the idea and the study design. YH and SA independently conducted a systematic review and reviewed the
literature for relevance. YH, EWC and SA undertook the
primary analysis. YH, EWC, WKL and ICKW contributed to the interpretation of the analysis. YH and EWC
wrote the rst draft of the paper. EWC, WKL, SA and
ICKW critically reviewed the results and the manuscript.
EWC and ICKW provided oversight over all aspects of
this project. All authors had full access to all of the data
in the study and take responsibility for the integrity of
the data and the accuracy of data analysis.
All authors approved the nal version of the manuscript.
ACKNOWLEDGEMENTS
We thank Mr. Kirin QL Liu and Ms. Niya Li for their
support in preliminary literature screening and study
characteristics summary, respectively. We also thank
Mrs. Lisa Wong for proofreading and editing the manuscript.
Declaration of personal and funding interests: None.
Aliment Pharmacol Ther
2015 John Wiley & Sons Ltd

Systematic review with meta-analysis: calcium channel blockers & gastrointestinal bleeding
SUPPORTING INFORMATION
Additional Supporting Information may be found in the
online version of this article:
Figure S1. Summarised estimates of subgroup analysis
by different comparison groups.
Figure S2. Summarised estimates of subgroup analysis
by different sources of GIB.
Figure S3. Summarised estimates of subgroup analysis
by adjustment for history of GIB.
Figure S4. Summarised estimates of subgroup analysis
by adjustment for use of anti-ulcer drugs.
Figure S5. Summarised estimates of subgroup analysis
by adjustment for NSAID use.
Figure S6. Summarised estimates of subgroup analysis

by adjustment for use of aspirin/anti-platelet agents.

Figure S7 Summarised estimates of subgroup analysis
by adjustment for use of corticosteroids/anti-coagulants.
Figure S8. Summarised estimates of subgroup analysis
by study quality.
Figure S9. Funnel plot of included studies.
Table S1. NewcastleOttawa scale for assessment of
quality of included observational studies.
Table S2. Characteristics of included studies assessing
the risk of gastrointestinal bleeding (GIB) with calcium
channel blocker (CCB) use.
Table S3. Summary of results of included studies.

REFERENCES
1. Kaplan NM. Calcium entry blockers in
the treatment of hypertension. Current
status and future prospects. JAMA
1989; 262: 81723.
2. Parmley WW. Efcacy and safety of
calcium channel blockers in
hypertensive patients with concomitant
left ventricular dysfunction. Clin
Cardiol 1992; 15: 23542.
3. Russell RP. Side effects of calcium
channel blockers. Hypertension 1988; 11
(3 Pt 2): Ii424.
4. Smith A, McPherson J, Taylor M,
Mason A, Carney S, Gillies A. Prohaemorrhagic effects of calcium
antagonists: a comparison of isradipine
and atenolol on ex vivo platelet
function in hypertensive subjects.
J Hum Hypertens 1997; 11: 7838.
5. Fitscha P, Virgolini I, Rauscha F,
Sinzinger H. Effects of isradipine on
platelet function in hypertension at rest
and during exercise. Am J Hypertens
1991; 4(2 Pt 2): 178s80s.
6. Ding YA, Han CL, Chou TC, Lai WY,
Shiao MF. Effect of isradipine on
platelet aggregation and oxygen free
radicals in hypertension. J Hypertens
Suppl 1991; 9: S3701.
7. Wallen NH, Held C, Rehnqvist N,
Hjemdahl P. Platelet aggregability
in vivo is attenuated by verapamil but
not by metoprolol in patients with
stable angina pectoris. Am J Cardiol
1995; 75: 16.
8. Blache D, Ojeda C. Comparative
inhibitory effects of dihydropyridines
on platelet aggregation, calcium uptake
and cyclic AMP concentration.
Pharmacology 1992; 45: 2509.
9. Ring ME, Corrigan JJ Jr, Fenster PE.
Antiplatelet effects of oral diltiazem,
Aliment Pharmacol Ther
2015 John Wiley & Sons Ltd

10.

11.

12.

13.

14.

15.

16.

17.

propranolol, and their combination. Br

J Clin Pharmacol 1987; 24: 61520.
Dollery CT. Clinical pharmacology of
calcium antagonists. Am J Hypertens
1991; 4: 88s95s.
Sica DA. Pharmacotherapy review:
calcium channel blockers. J Clin
Hypertens (Greenwich) 2006; 8: 536.
Phillips W, Piller LB, Williamson JD,
et al. Risk of hospitalized
gastrointestinal bleeding in persons
randomized to diuretic, ACE-inhibitor,
or calcium-channel blocker in
ALLHAT. J Clin Hypertens (Greenwich)
2013; 15: 82532.
Poole-Wilson PA, Kirwan BA, Voko Z,
de Brouwer S, van Dalen FJ, Lubsen J.
Safety of nifedipine GITS in stable
angina: the ACTION trial. Cardiovasc
Drugs Ther 2006; 20: 4554.
Celis H, Thijs L, Staessen JA, et al.
Interaction between nonsteroidal antiinammatory drug intake and calciumchannel blocker-based antihypertensive
treatment in the Syst-Eur trial. J Hum
Hypertens 2001; 15: 6138.
Lanas A, Serrano P, Bajador E, Fuentes
J, Sainz R. Risk of upper
gastrointestinal bleeding associated with
non-aspirin cardiovascular drugs,
analgesics and nonsteroidal antiinammatory drugs. Eur J Gastroenterol
Hepatol 2003; 15: 1738.
Kaplan RC, Heckbert SR, Koepsell TD,
Rosendaal FR, Psaty BM. Use of
calcium channel blockers and risk of
hospitalized gastrointestinal tract
bleeding. Arch Intern Med 2000; 160:
184955.
Kelly JP, Laszlo A, Kaufman DW,
Sundstrom A, Shapiro S. Major upper
gastrointestinal bleeding and the use of

18.

19.

20.

21.

22.

23.

24.

25.

calcium channel blockers. Lancet 1999;

353: 559.
Suissa S, Bourgault C, Barkun A,
Sheehy O, Ernst P. Antihypertensive
drugs and the risk of gastrointestinal
bleeding. Am J Med 1998; 105: 2305.
Smalley WE, Ray WA, Daugherty JR,
Grifn MR. No association between
calcium channel blocker use and
conrmed bleeding peptic ulcer disease.
Am J Epidemiol 1998; 148: 3504.
Garcia Rodriguez LA, Cattaruzzi C,
Troncon MG, Agostinis L. Risk of
hospitalization for upper
gastrointestinal tract bleeding associated
with ketorolac, other nonsteroidal antiinammatory drugs, calcium
antagonists, and other antihypertensive
drugs. Arch Intern Med 1998; 158:
339.
Pilotto A, Leandro G, Franceschi M, Di
Mario F, Valerio G. Antagonism to
calcium antagonists. Lancet 1996; 347:
17612.
Pahor M, Guralnik JM, Furberg CD,
Carbonin P, Havlik R. Risk of
gastrointestinal haemorrhage with
calcium antagonists in hypertensive
persons over 67 years old. Lancet 1996;
347: 10615.
Desboeuf K, Lapeyre-Mestre M,
Montastruc JL. Risk of gastrointestinal
haemorrhage with calcium antagonists.
Br J Clin Pharmacol 1998; 46: 889.
Petruk KC, West M, Mohr G, et al.
Nimodipine treatment in poor-grade
aneurysm patients. Results of a
multicenter double-blind placebocontrolled trial. J Neurosurg 1988; 68:
50517.
Yamada A, Sugimoto T, Kondo S, et al.
Assessment of the risk factors for
9

Y. He et al.

26.

27.

28.

29.

30.

31.

32.

33.

10

colonic diverticular hemorrhage. Dis

Colon Rectum 2008; 51: 11620.
Suh S, Seo PJ, Park H, et al. The risk
factors for colonic diverticular bleeding.
Korean J Gastroenterol 2012; 60: 349
54.
Nagata N, Niikura R, Sekine K, et al.
Risk of peptic ulcer bleeding associated
with Helicobacter pylori infection,
nonsteroidal anti-inammatory drugs,
low-dose aspirin, and antihypertensive
drugs: a case-control study. J
Gastroenterol Hepatol 2015; 30: 2928.
Jansen A, Harenberg S, Grenda U,
Elsing C. Risk factors for colonic
diverticular bleeding: a Westernized
community based hospital study.
World J Gastroenterol 2009; 15:
45761.
Weil J, Langman MJ, Wainwright P,
et al. Peptic ulcer bleeding: accessory
risk factors and interactions with nonsteroidal anti-inammatory drugs. Gut
2000; 46: 2731.
Higgins JPT, Green S (eds). Cochrane
Handbook for Systematic Reviews of
Interventions Version 5.1.0 [updated
March 2011]. The Cochrane
Collaboration, 2011. Available at:
www.cochrane-handbook.org (accessed
15 June 2014).
Moher D, Liberati A, Tetzlaff J, Altman
DG. Preferred reporting items for
systematic reviews and meta-analyses:
the PRISMA statement. BMJ 2009; 339:
b2535.
Stroup DF, Berlin JA, Morton SC, et al.
Meta-analysis of observational studies
in epidemiology: a proposal for
reporting. Meta-analysis Of
Observational Studies in Epidemiology
(MOOSE) group. JAMA 2000; 283:
200812.
Higgins JPT, Altman DG, Sterne JAC
(eds). Chapter 8: assessing risk of bias
in included studies. In: Higgins JPT,
Green S (eds). Cochrane Handbook for
Systematic Reviews of Interventions
Version 5.1.0 (updated March 2011).
The Cochrane Collaboration, 2011.

34.

35.

36.

37.

38.

39.

40.

41.

Available at: www.cochranehandbook.org (accessed 8 August 2014).

Wells GA SB, OConnell D, Peterson J,
et al. The NewcastleOttawa Scale
(NOS) for assessing the quality of
nonrandomised studies in metaanalyses. Available at: http://
www.ohri.ca/programs/
clinical_epidemiology/oxford.asp
(accessed 8 August 2014).
Review Manager (RevMan) [Computer
program], Version 5.3. Copenhagen:
The Nordic Cochrane Centre, The
Cochrane Collaboration, 2012.
Higgins JP, Thompson SG, Deeks JJ,
Altman DG. Measuring inconsistency
in meta-analyses. BMJ 2003; 327:
55760.
StatsDirect Ltd. StatsDirect Statistical
Software. England: StatsDirect Ltd,
2013. http://www.statsdirect.com.
The ALLHAT Ofcers and
Coordinators for the ALLHAT
Collaborative Research Group. Major
outcomes in high-risk hypertensive
patients randomized to angiotensinconverting enzyme inhibitor or calcium
channel blocker vs diuretic: the
Antihypertensive and Lipid-Lowering
Treatment to Prevent Heart Attack
Trial (ALLHAT). JAMA 2002; 288:
298197.
Leenen FH, Nwachuku CE, Black HR,
et al. Clinical events in high-risk
hypertensive patients randomly
assigned to calcium channel blocker
versus angiotensin-converting enzyme
inhibitor in the antihypertensive and
lipid-lowering treatment to prevent
heart attack trial. Hypertension 2006;
48: 37484.
Lopez-Mendez E, Uribe M. Beta
blockers in portal hypertension. Are
they really a good option? Ann Hepatol
2006; 5: 8691.
Poynard T, Cales P, Pasta L, et al.
Beta-adrenergic-antagonist drugs in the
prevention of gastrointestinal bleeding
in patients with cirrhosis and
esophageal varices. An analysis of data

42.

43.

44.

45.

46.

47.

48.

and prognostic factors in 589 patients

from four randomized clinical trials.
Franco-Italian Multicenter Study
Group. N Engl J Med 1991; 324: 1532
8.
Bernard B, Lebrec D, Mathurin P,
Opolon P, Poynard T. Betaadrenergic antagonists in the
prevention of gastrointestinal
rebleeding in patients with cirrhosis:
a meta-analysis. Hepatology 1997; 25:
6370.
Kvasnovsky CL, Papagrigoriadis S,
Bjarnason I. Increased diverticular
complications with nonsteriodal antiinammatory drugs and other
medications: a systematic review and
meta-analysis. Colorectal Dis 2014; 16:
O18996.
Zuccala G, Pedone C, Cocchi A, et al.
Use of calcium antagonists and
hemoglobin loss in hospitalized elderly
patients: a cohort study. Gruppo
Italiano di Farmacoepidemiologia
nellAnziano (GIFA) investigators.
Clin Pharmacol Ther 2000; 67:
31422.
Islim IF, Beevers DG, Bareford D. The
effect of antihypertensive drugs on
in vivo platelet activity in essential
hypertension. J Hypertens 1992; 10:
37983.
Cooper-DeHoff RM, Handberg EM,
Mancia G, et al. INVEST
revisited: review of ndings
from the International Verapamil SRTrandolapril Study. Expert Rev
Cardiovasc Ther 2009; 7:
132940.
Kuipers EJ, Hill C. Risk factors for
upper gastrointestinal bleeding in
patients using low-dose acetylsalicylic
acid: a systematic literature
analysis. Gastroenterology 2011; 140:
S174.
Jalil AA, Cronin R, Comianos M, Mann
M. Risk factors for diverticular bleeding
and re-bleeding: a United States
community-based hospital study. Am J
Gastroenterol 2013; 108: S177.

Aliment Pharmacol Ther

2015 John Wiley & Sons Ltd