1. RESEARCH
Effect of calcium supplements on risk of myocardial
infarction and cardiovascular events: meta-analysis
Mark J Bolland, senior research fellow,1 Alison Avenell, clinical senior lecturer,2 John A Baron, professor,3
Andrew Grey, associate professor,1 Graeme S MacLennan, senior research fellow,2 Greg D Gamble, research
fellow,1 Ian R Reid, professor1
1
Department of Medicine, Faculty ABSTRACT INTRODUCTION
of Medical and Health Sciences, Objective To investigate whether calcium supplements
University of Auckland, Private Bag Osteoporosis is a major cause of morbidity and mor-
92 019, Auckland 1142, New increase the risk of cardiovascular events. tality in older people.1 Calcium supplements margin-
Zealand Design Patient level and trial level meta-analyses. ally reduce the risk of fracture,2 3 and most guidelines
2
Health Services Research Unit, Data sources Medline, Embase, and Cochrane Central recommend adequate calcium intake as an integral
University of Aberdeen
3
Register of Controlled Trials (1966-March 2010), part of the prevention or treatment of osteoporosis.4 5
Department of Medicine, and
Department of Community and reference lists of meta-analyses of calcium supplements, Consequently, calcium supplements are commonly
Family Medicine, Dartmouth and two clinical trial registries. Initial searches were used by people over the age of 50. Observational stu-
Medical School, NH, USA carried out in November 2007, with electronic database dies suggest that high calcium intake might protect
Correspondence to: I R Reid searches repeated in March 2010.
i.reid@auckland.ac.nz against vascular disease,6-8 and the findings are consis-
Study selection Eligible studies were randomised, tent with those of interventional studies of calcium sup-
Cite this as: BMJ 2010;341:c3691 placebo controlled trials of calcium supplements plements that show improvement in some vascular risk
doi:10.1136/bmj.c3691
(≥500 mg/day), with 100 or more participants of mean factors.9-11 In contrast, calcium supplements accelerate
age more than 40 years and study duration more than one vascular calcification and increase mortality in patients
year. The lead authors of eligible trials supplied data. with renal failure, in both dialysis and predialysis
Cardiovascular outcomes were obtained from self reports, populations.12-14 Furthermore, a five year randomised
hospital admissions, and death certificates. controlled trial of calcium supplements in healthy
Results 15 trials were eligible for inclusion, five with older women, in which cardiovascular events were
patient level data (8151 participants, median follow-up prespecified as secondary end points, recently
3.6 years, interquartile range 2.7-4.3 years) and 11 with
reported possible increases in rates of myocardial
trial level data (11 921 participants, mean duration
infarction and cardiovascular events in women allo-
4.0 years). In the five studies contributing patient level
cated to calcium.15 16 We carried out a meta-analysis
data, 143 people allocated to calcium had a myocardial
of cardiovascular events in randomised trials of cal-
infarction compared with 111 allocated to placebo
cium supplements.
(hazard ratio 1.31, 95% confidence interval 1.02 to 1.67,
P=0.035). Non-significant increases occurred in the
METHODS
incidence of stroke (1.20, 0.96 to 1.50, P=0.11), the
composite end point of myocardial infarction, stroke, or In November 2007 we searched Medline, Embase, and
sudden death (1.18, 1.00 to 1.39, P=0.057), and death the Cochrane Central Register of Controlled Trials for
(1.09, 0.96 to 1.23, P=0.18). The meta-analysis of trial randomised placebo controlled trials of calcium sup-
level data showed similar results: 296 people had a plements, using the terms “calcium”, “randomised
myocardial infarction (166 allocated to calcium, 130 to controlled trial”, and “placebo” as text words, and cor-
placebo), with an increased incidence of myocardial responding MeSH terms (full details are available from
infarction in those allocated to calcium (pooled relative the authors). We searched for studies in the reference
risk 1.27, 95% confidence interval 1.01 to 1.59, lists of meta-analyses published between 1990 and
P=0.038). 2007 of the effect of calcium supplements on bone den-
Conclusions Calcium supplements (without sity, fracture, colorectal neoplasia, and blood pressure,
coadministered vitamin D) are associated with an and in two clinical trial registries (ClinicalTrials.gov
increased risk of myocardial infarction. As calcium and Australian New Zealand Clinical Trials Registry).
supplements are widely used these modest increases in No language restrictions were applied. In March 2010
risk of cardiovascular disease might translate into a large we updated the searches of the electronic databases
burden of disease in the population. A reassessment of (Medline: January 1966-March 2010, Embase: Janu-
the role of calcium supplements in the management of ary 1980-March 2010, Central Register of Controlled
osteoporosis is warranted. Trials: first quarter 2010).
BMJ | ONLINE FIRST | bmj.com page 1 of 9

2. RESEARCH
Study selection previously reported,18 20 or no cardiovascular data
We included studies if they were randomised, double were available.23 26 Thus, patient level data on cardio-
blind, placebo controlled trials; elemental calcium was vascular outcomes were available for 63% of partici-
administered at a dose of ≥500 mg/day; the partici- pants in the 15 eligible studies, complete trial level
pants’ mean age at baseline was more than 40 years; data for 85% of participants, and at least partially com-
100 or more participants were randomised; partici- plete trial level data for 93% of participants. Basic
pants of either sex were studied; and the trial duration demographic and other trial related data were either
was more than one year. supplied by the lead authors (or nominated deputies)
We excluded trials concerning calcium and vitamin or extracted from the original publication by an inves-
D given together with a placebo comparator (trials tigator (MB).
were only eligible if vitamin D was given to both inter-
vention and control groups, because vitamin D supple- Ascertainment of cardiovascular events
mentation has been associated with decreased We considered a myocardial infarction to have
mortality17); trials in which calcium was administered occurred when either of the terms “myocardial infarc-
in the form of dietary modification or a complex nutri- tion” or “heart attack”, or code 410 (international clas-
tional supplement; and trials in which most partici- sification of diseases, ninth revision), was used to
pants had a major systemic disease other than describe the event. A stroke was considered to have
osteoporosis. occurred when any of the terms “stroke”, “cerebral
infarction”, “intracerebral hemorrhage”, “subarach-
Search results noid hemorrhage”, or “cerebrovascular accident”, or
One investigator (MB) carried out the initial search and any of the ICD-9 codes 430, 431, 433, 434 were used
two investigators (AG and MB) independently to describe the event. We considered a sudden death to
reviewed all potentially relevant studies. Overall, 190 have occurred when the term “sudden death” or the
potentially relevant reports of studies were identified ICD-9 code 798 was used to describe the event.
from the initial searches, but only 15 studies were eli-
Five studies contributed patient level data on cardio-
gible for analysis.15 16 18-33 Thirteen studies compared
vascular events. For one study,21 22 self reports of unad-
calcium supplements with placebo,15 16 18-27 29 31-33 one
judicated events were supplied. Another study28
study had a 2×2 factorial design allowing comparison
supplied self reports of hospital admissions and cause
of calcium with placebo and calcium plus vitamin D
of death from death certificates. Each event was then
with vitamin D,28 and one study compared calcium
independently adjudicated by two physicians blinded
plus alendronate with alendronate.30
to the treatment allocation of the participants (MB,
AG), and any disagreements were resolved by consen-
Data description
sus. For another study,25 verified events from hospital
We invited the lead author of each eligible study to
discharge data were supplied along with causes of
provide patient level data on cardiovascular events
deaths from death certificates. The causes of death
that occurred during the study irrespective of whether
were again independently adjudicated. For two
the participant was still taking the trial drug. When
studies,15 33 data from self reports, hospital admissions,
such data were not available we requested summary
and death certificates were independently adjudicated
data at trial level. We obtained patient level data on
by a cardiologist or neurologist, as previously
cardiovascular outcomes for five studies, and partially
complete trial level data for six. No data were available described.16 For six studies contributing trial level
for four studies because the original records were no data, all data on cardiovascular events were supplied
longer available and cardiovascular events were not by the lead authors and were obtained from physician
reported cause of death in one study27 and a combina-
tion of self reports and hospital discharge data in five
studies.19 24 29-32
Reports of studies identified by initial search (n=11 363)
End points
Reports of potentially relevant studies
identified and screened for retrieval (n=190) The prespecified primary end points were time to first
myocardial infarction, time to first stroke, and time to
Reports of studies excluded (n=162):
Study size (n=111)
first event for the composite end point of myocardial
Duration ≤12 months (n=30) infarction, stroke, or sudden death. The secondary end
Study design (n=21) point was time to death (all cause mortality).
28 reports of 15 studies eligible for inclusion
Statistical analysis
In trials with patient level data, we analysed each end
Fig 1 | Flowchart of studies. Initial search was in November
point using a Cox proportional hazards model, with a
2007: 9358 reports were identified, 173 reports of potentially
relevant studies retrieved, 150 reports excluded, and 23
dummy coded variable representing each study in the
reports of 15 individual studies identified. Search was model, and we reported the hazard ratio, with 95%
updated in March 2010: a further 2005 reports were identified confidence interval, and number needed to treat. The
and 17 reports retrieved but no new studies identified assumption of proportional hazards was explored by
page 2 of 9 BMJ | ONLINE FIRST | bmj.com

3. RESEARCH
Table 1 | Characteristics of 15 studies eligible for inclusion in meta-analysis
No in calcium group/ Daily dose Trial duration Baseline mean
Studies No in control group and supplement type (years) Primary end point age (years) % female
Patient level data on cardiovascular outcomes:
Reid 199321 22 68/67 1 g lactogluconate-carbonate 4 Bone mineral density 58 100
Baron 199925 464/466 1.2 g carbonate 4 Colorectal adenoma 61 28
Grant 200528 2617/2675 1 g carbonate 4* Low trauma fracture 77 85
Reid 200615 16 732/739 1 g citrate 5 Clinical fracture 74 100
Reid 200833 216/107 0.6 g or 1.2 g citrate 2 Spine bone mineral density 56 0
Subtotal/average† 4097/4054 — 4.1 — 73 78
Trial level data on cardiovascular outcomes‡:
Dawson-Hughes 199019 238/123 0.5 g carbonate or citrate 2 Spine bone mineral density 58 100
Riggs 199824 119/117 1.6 g citrate 4 Bone mineral density 66 100
Bonithon-Kopp 200027 204/212 2 g lactogluconate-carbonate 3 Colorectal adenoma 59 37
Prince 200629 730/730 1.2 g carbonate 5 Osteoporotic fracture 75 100
Bonnick 200730 282/281 1 g carbonate 2 Spine bone mineral density 66 100
Lappe 200731 32 446/288 1.4 g citrate or 1.5 g carbonate 4 Fracture incidence 67 100
Subtotal/average† 2019/1751 — 3.8 — 68 93
Total/average† 6116/5805 4.0 72 83
No data on cardiovascular outcomes:
Smith 198918§ 84/85 1.5 g carbonate 4 Arm bone mineral density 51 100
Elders 199120 198/97 1g or 2g lactogluconate-carbonate 2 Spine bone mineral density NA 100
or citrate
Recker 199623 95/102 1.2 g carbonate 4 Vertebral fracture 74 100
Peacock 200026 126/135 0.75 g citrate 4 Hip bone mineral density 76 72
Subtotal/average† 503/419 — 4.0 — 69 88
NA=not available.
*All participants were followed for two years with a median follow-up of 45 months.
†Weighted by person years of follow-up.
‡Complete trial level data were available for two studies.29 31 Partially complete data were available for four studies19 24 27 30
(see table 2 for details).
§No original records remained but lead author recalled no heart attacks in either treatment group.
carrying out a test for proportionality of the interaction of the analyses. We used random effects models to
between variables included in the model and the loga- pool summary data at trial level. Publication bias was
rithm of time. For a small number of events (<10%) the assessed using Funnel plots and Egger’s regression
timing was unknown. We treated these events as if they model.34 Analyses were done using SAS version 9.1
had occurred on the last day of follow-up for that par- or Comprehensive Meta-analysis version 2 (Biostat,
ticipant. Possible confounding by covariates was Englewood, NJ). All tests were two tailed and we con-
assessed by repeating these models including prespeci- sidered P<0.05 as significant.
fied covariates likely to be associated with cardio-
vascular outcomes (age, sex, smoking status, presence RESULTS
of diabetes, dyslipidaemia, and hypertension at base- Figure 1 shows the results of the literature search and
line, and history of coronary heart disease) where data table 1 the characteristics of the eligible studies. The
were available for more than 80% of participants. Pre- quality of the five studies contributing patient level
specified subgroup analyses for dietary calcium, age, data has been independently assessed in previous sys-
sex, vitamin D status (serum 25-hydroxyvitamin D tematic reviews. 3 35 36
≥50 nmol/l or <50 nmol/l), and supplement type All 11 eligible trials were double blind, randomised
were done using interaction terms between treatment studies. The method of randomisation was stated expli-
allocation and the factor of interest. citly in seven: one used a central randomisation service
We used Poisson regression models to assess the and six used computer generated random numbers.
relation between the total number of events and treat- Allocation concealment was explicitly described in
ment allocation. Because recurrent cardiovascular four studies. Ten studies gave details of participants
events in an individual are unlikely to be independent, who withdrew or were lost to follow-up. Compliance
we used Poisson regression with general estimating was reported in all 11 studies, but the definitions for
equations to account for the intra-individual depen- compliance differed between studies and were not
dence of events. always comparable. In general, studies reported com-
To assess statistical heterogeneity between summary pliance of more than 75% in participants who were tak-
data at trial level we used Cochran’s Q statistic ing tablets at study completion. Table 2 shows the
(P<0.10) and the I2 statistic (I2>50%). No significant baseline cardiovascular characteristics, dietary cal-
statistical heterogeneity existed between trials in any cium intake, and vitamin D status of participants.
BMJ | ONLINE FIRST | bmj.com page 3 of 9

4. RESEARCH
Table 2 | Baseline cardiovascular and other variables in trials with patient or trial level data available for cardiovascular outcomes. Values are means
(standard deviations) unless stated otherwise
Dietary calcium Vitamin D* Weight Current smoker Hypertension Diabetes Ischaemic heart Lipid disorder
Studies (mg/day) (nmol/l) (kg) (%) (%) (%) disease (%) (%)
Dawson-Hughes19 406 (84) NA NA NA NA NA NA NA
Reid21 22 750 (290) 93 (37) 65 (9) 10 9 0 2 1
Riggs24 710 (290) 80 (25) NA NA NA NA NA NA
Baron25 880 (440) 73 (27) 82 (15) 19 37 10 12 32
Bonithon-Kopp27 980 (380) NA NA NA NA NA NA NA
Grant28 820 (350) 45 (18)† 65 (13) 12 NA 8 NA NA
Reid15 16 860 (390) 54 (18) 67 (11) 3 29 3 8 8
Prince29 915 NA 69 (13) NA NA NA NA NA
Bonnick30 1240 (580) NA NA 0.4 NA NA NA NA
Lappe31 32 1070 72 (20) 77 (15) NA NA NA NA NA
Reid33 870 (450) 92 (33) 83 (12) 3 8 0.3 0.3 4
NA=not available.
*25-hydroxyvitamin D.
†25-hydroxyvitamin D measured in sample of 80 participants.
Cardiovascular events by treatment allocation are for the effect of calcium treatment on myocardial
shown in table 3. infarction were 1.18 (0.70 to 2.00) for <500 mg/day,
0.68 (0.39 to 1.18) for 500-699 mg/day, 2.28 (1.26 to
Patient level analysis 4.15) for 700-899 mg/day, 1.81 (0.97 to 3.41) for 900-
Table 4 shows the baseline characteristics of the treat- 1099 mg/day, and 1.41 (0.81 to 2.48) for ≥1100 mg/
ment groups in the five studies contributing patient day; test for linear trend when hazard ratios are
level data. The median (interquartile range) duration expressed relative to the <500 mg/d fifth, P=0.12.
of follow-up in both groups was 3.6 (2.7-4.3) years. Interactions between treatment allocation and age,
In total, 143 people allocated to calcium had a myo- sex, vitamin D status, or supplement type for myo-
cardial infarction during follow-up compared with 111 cardial infarction were not significant, nor were they
allocated to placebo. The risk of incident myocardial between treatment allocation and any of these vari-
infarction in those allocated to calcium increased by ables or dietary calcium intake for stroke, the compo-
31% (hazard ratio 1.31, 95% confidence interval 1.02 site end point, or death.
to 1.67, P=0.035. fig 2). During follow-up, 167 people Recurrent cardiovascular events tended to be more
allocated to calcium and 143 allocated to placebo had a common in people allocated to calcium. Comparing
stroke (1.20, 0.96 to 1.50, P=0.11), 293 people allo- people allocated to calcium with those allocated to pla-
cated to calcium and 254 allocated to placebo had cebo, 19 versus 13 had more than one myocardial
any of myocardial infarction, stroke, or sudden death infarction (P=0.38), 21 versus 13 had more than one
(1.18, 1.00 to 1.39, P=0.057), and 519 people allocated stroke (P=0.23), and 59 versus 32 had more than one
to calcium and 487 allocated to placebo died (1.09, of myocardial infarction, stroke, or sudden death
0.96 to 1.23, P=0.18). The number needed to treat (P=0.006). Poisson regression models with general esti-
(NNT) with calcium for five years to cause one incident mating equations were used to estimate the effect of
event was 69 for myocardial infarction, 100 for stroke, calcium on the total number of events, including inci-
61 for any of myocardial infarction, stroke, or sudden dent and recurrent events. Overall, 164 myocardial
death, and 77 for death. Adjusting for prespecified cov- infarctions occurred in people allocated to calcium
ariates related to cardiovascular outcomes with data compared with 125 in those allocated to placebo (rela-
available for more than 80% of participants (age, sex, tive risk 1.32, 95% confidence interval 1.02 to 1.71,
smoking status, and diabetes) did not change the results P=0.032). Stroke occurred in 190 people allocated to
of the primary analyses. calcium compared with 156 allocated to placebo (1.24,
Prespecified subgroup analyses showed a significant 0.99 to 1.56, P=0.07). In total, 361 myocardial infarc-
interaction between treatment allocation and dietary tions, strokes, or sudden deaths occurred in people
calcium intake for myocardial infarction. Calcium allocated to calcium compared with 287 in people allo-
treatment was associated with an increased risk of myo- cated to calcium (1.27, 1.07 to 1.51, P=0.006).
cardial infarction in people with dietary calcium intake
above the median of 805 mg/day (hazard ratio 1.85, Trial level analysis
95% confidence interval 1.28 to 2.67) but no increased Table 3 shows summary data on cardiovascular events
risk in those with dietary calcium intake below the at trial level. Eight studies15 16 21 22 25 28-33 were included
median (0.98, 0.69 to 1.38, P for interaction 0.01). in the main analysis. A further three trials19 24 27 had
When the cohort was divided by fifths of dietary cal- data only available for subgroups of participants.
cium intake rounded to the nearest 100 mg/day, the These three trials were included in a sensitivity analysis
respective hazard ratios (95% confidence intervals) that included data from all 11 trials. Publication bias
page 4 of 9 BMJ | ONLINE FIRST | bmj.com

5. RESEARCH
Table 3 | Number of people with cardiovascular events and deaths by treatment allocation
Calcium group Placebo group
No of Myocardial No of Myocardial
Studies participants infarction Stroke Composite* Death participants infarction Stroke Composite* Death
Dawson-Hughes19† 238 0 0 NA NA 123 0 1 NA NA
Reid21 22 68 0 2 2 0 67 0 1 1 0
Riggs24‡ 119 0 0 0 1 117 0 0 0 0
Baron25 464 20 15 31 25 466 17 11 28 22
Bonithon-Kopp27§ 204 0 1 1 8 212 0 0 1 9
Grant28¶ 1311 45 56 97 238 1332 39 48 86 217
Grant28** 1306 44 60 100 220 1343 34 58 89 218
Reid15 16 732 31 34 60 34 739 21 25 50 29
Prince29†† 730 21 38 56 29 730 17 40 56 38
Bonnick30‡‡ 282 0 1 NA 2 281 0 2 NA 1
Lappe31 32†† 446 2 5 8 NA 288 2 4 8 NA
Reid33 216 3 0 3 2 107 0 0 0 1
Total 6116 166 212 358 559 5805 130 190 319 535
NA=not available.
*Any of myocardial infarction, stroke, or sudden death. Seventeen events were classified as sudden deaths that occurred in eight people allocated to calcium and nine allocated to placebo.
†Unpublished trial level data provided by author. Data on stroke available only for participants who withdrew from study.
‡Unpublished trial level data provided by author. Data available only for participants who withdrew from study.
§Unpublished trial level data provided by author. Data on cause of death only.
¶Calcium versus placebo arms in Randomised Evaluation of Calcium or Vitamin D (RECORD) study.
**Calcium and vitamin D versus placebo plus vitamin D arms in Randomised Evaluation of Calcium or Vitamin D (RECORD) study.
††Unpublished trial level data provided by author.
‡‡Unpublished trial level data provided by Boyd Scott.
was not evident on inspection of Funnel plots or in not apply to coadministered calcium and vitamin D
Egger’s regression model in any analysis (P>0.40 for supplements. None of the trials had cardiovascular out-
all analyses). Figure 3 shows the results of the main comes as the primary end points, and data on cardio-
analysis. Allocation to calcium supplements was asso- vascular events were not gathered in a standardised
ciated with an increased risk of myocardial infarction manner. In only two of the trials were the data adjudi-
(relative risk 1.27, 95% confidence interval 1.01 to cated by blinded trial investigators. However, unless
1.59, P=0.038) but not stroke, the composite end there was differential misclassification or misreporting
point, or death. Including data from the three addi- of cardiovascular events in people treated with cal-
tional studies in the sensitivity analysis did not signifi- cium, this is unlikely to alter the results, because the
cantly change the results for any end point. data came from blinded, placebo controlled trials.
Incomplete or no data on cardiovascular outcomes
DISCUSSION were available for seven trials in our analysis, compris-
ing about 15% of the total number of participants.
In this pooled analysis of around 12 000 participants
However, the small size of these trials and the consis-
from 11 randomised controlled trials, calcium supple-
tency of the findings in the other eight larger trials sug-
ments were associated with about a 30% increase in the
gest the missing data are unlikely to have substantially
incidence of myocardial infarction and smaller, non-
changed the results.
significant, increases in the risk of stroke and mortality.
When recurrent events in 10-17% of participants were
Comparison with other studies
included in analyses, the results were similar, although
The current findings are consistent with trials of
the relative risks tended to be slightly larger. The find-
patients with renal failure, in which calcium supple-
ings were consistent across trials, with an increased ments were associated with an increase in mortality.13
relative risk of myocardial infarction with calcium Few comparable data are available from observational
observed in six of the seven trials in which at least studies of calcium supplements. One study reported a
one event occurred, although no individual trial 24% increase in coronary heart disease in Finnish post-
reported a statistically significant effect. The risk of menopausal women using calcium supplements (with
myocardial infarction with calcium tended to be or without vitamin D) compared with non-users.37
greater in those with dietary calcium intake above the Non-fatal myocardial infarction in US men using cal-
median but was independent of age, sex, and type of cium supplements compared with non-users did not
supplement. increase significantly, although the relative risk for
each fifth of supplement intake ranged between 1.02
Limitations of the review and 1.07.38
Our study has some limitations. We excluded studies The relations between dietary calcium intake and
that compared coadministered calcium and vitamin D cardiovascular events have also been examined. The
supplements with placebo. The results therefore may inverse relation between calcium intake and
BMJ | ONLINE FIRST | bmj.com page 5 of 9

6. RESEARCH
Myocardial infarction Stroke
Cumulative incidence (%)
6 8
Hazard ratio 1.31 (95% CI 1.02 to 1.67), P=0.035 Hazard ratio 1.20 (95% CI 0.96 to 1.50), P=0.11
5
Calcium 6
4
Placebo
3 4
2
2
1
0 0
No at risk
Calcium 4097 3870 3539 2670 1294 373 4097 3865 3541 2659 1294 373
Placebo 4054 3865 3588 2728 1320 388 4054 3859 3589 2730 1312 386
Composite of myocardial infarction, stroke, or sudden death Death
Cumulative incidence (%)
12 18
Hazard ratio 1.18 (95% CI 1.00 to 1.39), P=0.057 Hazard ratio 1.09 (95% CI 0.96 to 1.23), P=0.18
10 15
8 12
6 9
4 6
2 3
0 0
0 1 2 3 4 5 0 1 2 3 4 5
Years Years
No at risk
Calcium 4097 3848 3517 2635 1271 360 4097 3889 3580 2699 1322 389
Placebo 4054 3848 3566 2692 1292 376 4054 3875 3618 2767 1340 399
Fig 2 | Cumulative incidence of myocardial infarction, stroke, composite of myocardial infarction, stroke, or sudden death, and
death by treatment allocation in five studies that contributed patient level data
standardised mortality ratios for ischaemic heart dis- studies of US men,38 39 or of Dutch civil servants.40
ease in the United Kingdom was strong.6 In two US Thus, in contrast with the observational and inter-
prospective observational studies, women in the high- ventional studies of calcium supplements, these obser-
est fourth of calcium intake had 30-40% lower cardio- vational studies do not show increased cardiovascular
vascular mortality than those in the lowest fourth,7 and risks with higher dietary calcium intake. These differ-
those in the highest fifth of calcium intake had a 30-40% ences suggest that cardiovascular risks from high cal-
lower risk of ischaemic stroke than those in the lowest cium intake might be restricted to use of calcium
fifth.8 No relation between calcium intake and ischae- supplements.
mic heart disease or stroke was observed in prospective A body of evidence related to the current work
comes from studies comparing coadministered cal-
cium and vitamin D supplements with placebo,
Table 4 | Baseline characteristics of participants in five studies included in patient level which were excluded from our meta-analysis.
analysis by treatment allocation. Values are percentages unless stated otherwise
Recently, the Women’s Health Initiative reported
Characteristics Calcium group Placebo group that calcium and vitamin D had no effect on the risk
Median (interquartile range) age (years) 74.5 (70-79) 74.6 (71-79) of coronary heart disease or stroke.41 The findings of
Women 76.5 79.2 that study might differ from ours for several reasons.
White ethnicity 97.2 97.7 The Women’s Health Initiative used low dose vitamin
Mean (SD) weight (kg) 68.4 (14.1) 67.9 (13.7) D supplements, and vitamin D deficiency has been
Mean (SD) dietary calcium (mg/day) 837 (377) 831 (370) associated with increased risk of cardiovascular
Mean (SD) 25-hydroxyvitamin D (nmol/l)* 66.1 (28.9) 64.3 (27.5) disease42 and vitamin D supplementation with
Current smoker 11.0 10.2 decreased mortality.17 Also, the participants differed
Hypertension† 28.0 28.4 from those in our meta-analysis: on average they
Ischaemic heart disease† 8.1 7.8 were younger (mean age 62 v 75 years), heavier
Lipid disorder† 14.8 15.4 (mean weight 76 v 68 kg; 34% in Women’s Health
Diabetes 7.0 6.7 Initiative v 10% of women in our meta-analysis
Proportion of women was significantly higher in placebo group because one study that only involved men had a weighed >80 kg), had higher calcium intakes (mean
2:1 ratio of allocation to calcium or placebo.33 No other differences existed between groups. Medical conditions 1150 v 830 mg/day), and a higher proportion were
at baseline were self reported by participants.
*Data available from four studies for 1445 participants in calcium groups and 1355 in placebo groups.
using hormone replacement therapy (52% v <3% in
†Data available from four studies for 1480 participants in calcium groups and 1379 in placebo groups. our meta-analysis). Overall, 54% were taking non-
page 6 of 9 BMJ | ONLINE FIRST | bmj.com

7. RESEARCH
Study Relative risk of myocardial Weight Study Relative risk of Weight
infarction (95% CI) (%) stroke (95% CI) (%)
Baron 1999 13 Reid 1993 1
Grant 2005 29 Baron 1999 6
Grant 2005 Vit D 26 Grant 2005 26
Prince 2006 13 Grant 2005 Vit D 30
Reid 2006 17 Prince 2006 20
Lappe 2007 1 Reid 2006 14
Reid 2008 1 Bonnick 2007 1
Total Lappe 2007 2
P=0.038 1.27 (1.01 to 1.59)
Test for heterogeneity: Total
P=0.25 1.12 (0.92 to 1.36)
I2=0% P=0.96 Test for heterogeneity:
I2=0% P=0.93
Study Relative risk of myocardial infarction, Weight Study Relative risk of Weight
stroke, or sudden death (95% CI) (%) death (95% CI) (%)
Reid 1993 0.4 Baron 1999 4
Baron 1999 9 Grant 2005 43
Grant 2005 27 Grant 2005 Vit D 42
Grant 2005 Vit D 28 Prince 2006 5
Prince 2006 17 Reid 2006 5
Reid 2006 16 Bonnick 2007 0.2
Lappe 2007 2 Reid 2008 0.2
Reid 2008 0.2 Total
Total Test for heterogeneity: P=0.26 1.07 (0.95 to 1.19)
Test for heterogeneity: P=0.13 1.12 (0.97 to 1.30) I2=0% P=0.84 0.5 0.8 1 1.2 1.5 2 3
I2=0% P=0.91 0.5 0.8 1 1.2 1.5 2 3 Favours Favours
calcium placebo
Favours Favours
calcium placebo
Fig 3 | Random effects models of effect of calcium supplementation on cardiovascular events and death. Full data were
available from these eight trials, but some trials do not appear in the figures because no events occurred during the trial: no
myocardial infarctions occurred in the study by Reid 199321 22 or Bonnick 2007,30 no strokes occurred in the study by Reid
2008,33 and no deaths occurred in the study by Reid 1993.21 22 Data on composite end point were not available for the study
by Bonnick 200730 or Lappe 2007.31 32 Grant 2005 is a Randomised Evaluation of Calcium or Vitamin D (RECORD) study calcium
versus placebo arms, and Grant 2005 vitamin D is a RECORD study calcium plus vitamin D versus vitamin D plus placebo arms
protocol calcium supplements at baseline (30% were with an increased incidence of myocardial infarction in
taking ≥500 mg/day), increasing to 69% at the final large observational studies.48-50 Primary hyperpara-
visit,43 44 compared with 1.2% taking non-protocol thyroidism, a condition in which serum calcium levels
calcium supplements at baseline in our meta-analysis. are raised, has also been associated with an increased
In the subgroups of women in the Women’s Health risk of cardiovascular events and death.51 52 Ingestion
Initiative who most closely approximated the partici- of equivalent doses of calcium from dairy products has
pants in our analyses (age 70-79 years, body mass a much smaller effect than calcium supplements on
index <30 kg/m2, total calcium intake <800 mg/day), serum calcium levels,53 which might account for the
the confidence intervals of the hazard ratios for coron- absence of a detrimental vascular effect of dietary cal-
ary heart disease with calcium and vitamin D included cium intake in the observational studies reviewed. Vas-
the hazard ratio for myocardial infarction we observed. cular calcification is an established risk factor for
It would be valuable to reanalyse the results of the cardiovascular disease,54 and the process of vascular
Women’s Health Initiative to assess the effects of cal- calcification is similar to osteogenesis.55 Because cal-
cium and vitamin D in non-obese women and in cium supplements increase bone density it is possible
women not taking non-protocol calcium supplements. that they may also increase vascular calcification and
Interestingly, the only study in our analysis that thereby cardiovascular events. In patients with renal
reported a relative risk of less than 1.0 for myocardial failure (both dialysis and predialysis populations) cal-
infarction with calcium also had high non-protocol use cium supplements accelerate vascular calcification and
of calcium supplements.31 32 45 increase mortality.12-14 Our graphical data are consis-
The current analyses do not deal with the mechan- tent with the possibility that an increased risk of myo-
isms by which calcium supplements might increase the cardial infarction with calcium supplements emerges
risk of myocardial infarction, but we have reviewed quickly, pointing to mechanisms such as increased coa-
this elsewhere.46 Calcium supplements acutely gulability or altered vascular flow, perhaps mediated
increase serum calcium levels to a modest degree.47 directly through the calcium sensing receptor or indir-
Serum calcium levels have been positively associated ectly through alterations in calcitropic hormones.
BMJ | ONLINE FIRST | bmj.com page 7 of 9

8. RESEARCH
AA is funded by a career scientist award of the chief scientist office of the
WHAT IS ALREADY KNOWN ON THIS TOPIC Scottish government health directorates. The Health Services Research
Unit is funded by the chief scientist office of the Scottish government
Calcium supplements are commonly taken by older people for skeletal health
health directorates. The sponsors of the study had no role in design and
A randomised placebo controlled trial suggested calcium supplements might increase the conduct of the study; collection, management, analysis, and
risk of myocardial infarction and cardiovascular events interpretation of the data; and preparation, review, or approval of the
manuscript. The authors are independent from the funders.
WHAT THIS STUDY ADDS Competing interests: All authors have completed the unified competing
interest form at www.icmje.org/coi_disclosure.pdf (available on request
A meta-analysis of trials totalling 12 000 participants found that calcium supplements from the corresponding author) and declare that: (1) no author has
increase the risk of myocardial infarction by about 30% support from companies for the submitted work; (2) IR has received
research support from and acted as a consultant for Fonterra that might
Given the modest benefits of calcium supplements on bone density and fracture prevention, have an interest in the submitted work in the previous 3 years; JB, IR, AA
a reassessment of the role of calcium supplements in the management of osteoporosis is and GM had study drugs for clinical trials of calcium supplementation
warranted supplied by Wyeth; Mission Pharmacal; Shire Pharmaceuticals and
Nycomed; and Shire Pharmaceuticals and Nycomed, respectively, might
have an interest in the submitted work in the previous 3 years; (3) their
spouses, partners, or children have no financial relationships that may be
Calcium supplements modestly increase bone relevant to the submitted work; and (4) no author has non-financial
density3 15 33 and have marginal efficacy against interests that may be relevant to the submitted work.
fracture.2 3 In a pooled analysis of studies contributing Ethical approval: Not required.
patient level data,15 21 22 25 28 33 the hazard ratio for frac- Data sharing: No additional data available.
ture was 0.90 (95% confidence interval 0.80 to 1.01)
and NNT for five years to prevent one fracture was 39. 1 Sambrook P, Cooper C. Osteoporosis. Lancet 2006;367:2010-8.
2 Bischoff-Ferrari HA, Dawson-Hughes B, Baron JA, Burckhardt P, Li R,
A recent meta-analysis of the effect of calcium with or Spiegelman D, et al. Calcium intake and hip fracture risk in men and
without vitamin D on fracture reported a similar NNT women: a meta-analysis of prospective cohort studies and
randomized controlled trials. Am J Clin Nutr 2007;86:1780-90.
for five years of 48.3 Later meta-analyses reported that 3 Tang BMP, Eslick GD, Nowson C, Smith C, Bensoussan A. Use of
combined supplementation with calcium and vitamin D calcium or calcium in combination with vitamin D supplementation
reduced fractures, whereas vitamin D alone did not.56 57 to prevent fractures and bone loss in people aged 50 years and
older: a meta-analysis. Lancet 2007;370:657-66.
Incorporating the results from the current analysis of 4 American Association of Clinical Endocrinologists. Medical
studies contributing patient level data, treatment of guidelines for clinical practice for the prevention and treatment of
1000 people with calcium for five years would cause postmenopausal osteoporosis. Endo Pract 2003;9:545-64.
5 National Osteoporosis Foundation. Physician’s guide to prevention
an additional 14 myocardial infarctions, 10 strokes, and treatment of osteoporosis. National Osteoporosis Foundation,
and 13 deaths, and prevent 26 fractures. 2008.
6 Knox EG. Ischaemic-heart-disease mortality and dietary intake of
calcium. Lancet 1973;1:1465-7.
Conclusions 7 Bostick RM, Kushi LH, Wu Y, Meyer KA, Sellers TA, Folsom AR.
In summary, randomised studies suggest that calcium Relation of calcium, vitamin D, and dairy food intake to ischemic
heart disease mortality among postmenopausal women. Am J
supplements without coadministered vitamin D are Epidemiol 1999;149:151-61.
associated with an increased incidence of myocardial 8 Iso H, Stampfer MJ, Manson JE, Rexrode K, Hennekens CH,
infarction. The vascular effects of calcium supple- Colditz GA, et al. Prospective study of calcium, potassium, and
magnesium intake and risk of stroke in women. Stroke
ments, especially without vitamin D, should be studied 1999;30:1772-9.
further. Although the magnitude of the increase in risk 9 Griffith LE, Guyatt GH, Cook RJ, Bucher HC, Cook DJ. The influence of
is modest, the widespread use of calcium supplements dietary and nondietary calcium supplementation on blood pressure
—an updated metaanalysis of randomized controlled trials. Am J
means that even a small increase in incidence of cardio- Hypertens 1999;12:84-92.
vascular disease could translate into a large burden of 10 Reid IR, Mason B, Horne A, Ames R, Clearwater J, Bava U, et al. Effects
of calcium supplementation on serum lipid concentrations in normal
disease in the population. The likely adverse effect of older women: a randomized controlled trial. Am J Med
calcium supplements on cardiovascular events, taken 2002;112:343-7.
together with the possible adverse effect on incidence 11 Reid IR, Horne A, Mason B, Ames R, Bava U, Gamble GD. Effects of
calcium supplementation on body weight and blood pressure in
of hip fracture2 58 and its modest overall efficacy in normal older women: a randomized controlled trial. J Clin Endocrinol
reducing fracture (about 10% reduction in total Metab 2005;90:3824-9.
fractures)2 3 suggest that a reassessment of the role of 12 Goodman WG, Goldin J, Kuizon BD, Yoon C, Gales B, Sider D, et al.
Coronary-artery calcification in young adults with end-stage renal
calcium supplements in the prevention and treatment disease who are undergoing dialysis. N Engl J Med
of osteoporosis is warranted. 2000;342:1478-83.
13 Block GA, Raggi P, Bellasi A, Kooienga L, Spiegel DM. Mortality effect
We thank the investigators who provided unpublished data on of coronary calcification and phosphate binder choice in incident
cardiovascular events from their studies including Larry Riggs, Richard hemodialysis patients. Kidney Int 2007;71:438-41.
Prince, Claire Bonithon-Kopp, Joan Lappe, Boyd Scott, and Kathy Zhu, as 14 Russo D, Miranda I, Ruocco C, Battaglia Y, Buonanno E, Manzi S,
well as Adrian M Grant, M K Campbell, Alison M McDonald and Gladys C et al. The progression of coronary artery calcification in predialysis
McPherson from the Randomised Evaluation of Calcium or Vitamin D trial. patients on calcium carbonate or sevelamer. Kidney Int
Contributors: MJB, AG, and IRR drafted the protocol. AA and JAB critically 2007;72:1255-61.
revised the protocol. MJB and AG carried out the literature search and 15 Reid IR, Mason B, Horne A, Ames R, Reid HE, Bava U, et al.
Randomized controlled trial of calcium in healthy older women. Am J
event adjudication. All authors provided individual patient data from their
Med 2006;119:777-85.
studies. MJB and GDG did the analyses. MJB drafted the paper. All authors
16 Bolland MJ, Barber PA, Doughty RN, Mason B, Horne A, Ames R, et al.
critically reviewed the paper. MJB had full access to all the data in the Vascular events in healthy older women receiving calcium
study and takes responsibility for the integrity of the data and the supplementation: randomised controlled trial. BMJ 2008;336:262-6.
accuracy of the data analysis. MJB is the guarantor. 17 Autier P, Gandini S. Vitamin D supplementation and total mortality: a
Funding: This review was funded by the Health Research Council of New meta-analysis of randomized controlled trials. Arch Intern Med
Zealand and the University of Auckland School of Medicine Foundation. 2007;167:1730-7.
page 8 of 9 BMJ | ONLINE FIRST | bmj.com

9. RESEARCH
18 Smith EL, Gilligan C, Smith PE, Sempos CT. Calcium supplementation risk of ischemic heart disease among men. Am J Clin Nutr
and bone loss in middle-aged women. Am J Clin Nutr 2003;77:814-8.
1989;50:833-42. 39 Ascherio A, Rimm EB, Hernan MA, Giovannucci EL, Kawachi I,
19 Dawson-Hughes B, Dallal GE, Krall EA, Sadowski L, Sahyoun N, Stampfer MJ, et al. Intake of potassium, magnesium, calcium, and
Tannenbaum S. A controlled trial of the effect of calcium fiber and risk of stroke among US men. Circulation
supplementation on bone density in postmenopausal women. N 1998;98:1198-204.
Engl J Med 1990;323:878-83. 40 Van der Vijver LP, van der Waal MA, Weterings KG, Dekker JM,
20 Elders PJ, Netelenbos JC, Lips P, van Ginkel FC, Khoe E, Schouten EG, Kok FJ. Calcium intake and 28-year cardiovascular and
Leeuwenkamp OR, et al. Calcium supplementation reduces vertebral coronary heart disease mortality in Dutch civil servants. Int J
bone loss in perimenopausal women: a controlled trial in 248 Epidemiol 1992;21:36-9.
women between 46 and 55 years of age. J Clin Endocrinol Metab 41 Hsia J, Heiss G, Ren H, Allison M, Dolan NC, Greenland P, et al.
1991;73:533-40. Calcium/vitamin D supplementation and cardiovascular events.
21 Reid IR, Ames RW, Evans MC, Gamble GD, Sharpe SJ. Effect of calcium Circulation 2007;115:846-54.
supplementation on bone loss in postmenopausal women. N Engl J 42 Wang TJ, Pencina MJ, Booth SL, Jacques PF, Ingelsson E, Lanier K,
Med 1993;328:460-4. et al. Vitamin D deficiency and risk of cardiovascular disease.
22 Reid IR, Ames RW, Evans MC, Gamble GD, Sharpe SJ. Long-term Circulation 2008;117:503-11.
effects of calcium supplementation on bone loss and fractures in 43 Jackson RD, LaCroix AZ, Gass M, Wallace RB, Robbins J, Lewis CE,
postmenopausal women: a randomized controlled trial. Am J Med et al. Calcium plus vitamin D supplementation and the risk of
1995;98:331-5. fractures. N Engl J Med 2006;354:669-83.
23 Recker RR, Hinders S, Davies KM, Heaney RP, Stegman MR, Lappe JM, 44 Wactawski-Wende J, Kotchen JM, Anderson GL, Assaf AR, Brunner RL,
et al. Correcting calcium nutritional deficiency prevents spine O’Sullivan MJ, et al. Calcium plus vitamin D supplementation and the
fractures in elderly women. J Bone Miner Res 1996;11:1961-6. risk of colorectal cancer. N Engl J Med 2006;354:684-96.
24 Riggs BL, O’Fallon WM, Muhs J, O’Connor MK, Kumar R, Melton LJ 3rd. 45 Lappe JM, Davies KM, Travers-Gustafson D, Heaney RP. Vitamin D
Long-term effects of calcium supplementation on serum parathyroid status in a rural postmenopausal female population. J Am Coll Nutr
hormone level, bone turnover, and bone loss in elderly women. J 2006;25:395-402.
Bone Miner Res 1998;13:168-74.
46 Reid IR, Bolland MJ, Grey A. Does calcium supplementation increase
25 Baron JA, Beach M, Mandel JS, van Stolk RU, Haile RW, Sandler RS,
cardiovascular risk? Clin Endocrinol (Oxf)
et al. Calcium supplements for the prevention of colorectal
2010; published online 23 Feb.
adenomas. Calcium Polyp Prevention Study Group. N Engl J Med
47 Reid IR, Schooler BA, Hannan SF, Ibbertson HK. The acute
1999;340:101-7.
biochemical effects of four proprietary calcium preparations. Aust N Z
26 Peacock M, Liu G, Carey M, McClintock R, Ambrosius W, Hui S, et al.
J Med 1986;16:193-7.
Effect of calcium or 25OH vitamin D3 dietary supplementation on
bone loss at the hip in men and women over the age of 60. J Clin 48 Lind L, Skarfors E, Berglund L, Lithell H, Ljunghall S. Serum calcium: a
Endocrinol Metab 2000;85:3011-9. new, independent, prospective risk factor for myocardial infarction in
27 Bonithon-Kopp C, Kronborg O, Giacosa A, Rath U, Faivre J. Calcium middle-aged men followed for 18 years. J Clin Epidemiol
and fibre supplementation in prevention of colorectal adenoma 1997;50:967-73.
recurrence: a randomised intervention trial. European Cancer 49 Jorde R, Sundsfjord J, Fitzgerald P, Bonaa KH. Serum calcium and
Prevention Organisation Study Group. Lancet 2000;356:1300-6. cardiovascular risk factors and diseases: the Tromso study.
28 Grant AM, Avenell A, Campbell MK, McDonald AM, MacLennan GS, Hypertension 1999;34:484-90.
McPherson GC, et al. Oral vitamin D3 and calcium for secondary 50 Foley RN, Collins AJ, Ishani A, Kalra PA. Calcium-phosphate levels and
prevention of low-trauma fractures in elderly people (Randomised cardiovascular disease in community-dwelling adults: the
Evaluation of Calcium Or vitamin D, RECORD): a randomised placebo- Atherosclerosis Risk in Communities (ARIC) Study. Am Heart J
controlled trial. Lancet 2005;365:1621-8. 2008;156:556-63.
29 Prince RL, Devine A, Dhaliwal SS, Dick IM. Effects of calcium 51 Nilsson IL, Yin L, Lundgren E, Rastad J, Ekbom A. Clinical presentation
supplementation on clinical fracture and bone structure: results of a of primary hyperparathyroidism in Europe—nationwide cohort
5-year, double-blind, placebo-controlled trial in elderly women. Arch analysis on mortality from nonmalignant causes. J Bone Miner Res
Intern Med 2006;166:869-75. 2002;17(suppl 2):N68-74S.
30 Bonnick S, Broy S, Kaiser F, Teutsch C, Rosenberg E, DeLucca P, et al. 52 Vestergaard P, Mollerup CL, Frokjaer VG, Christiansen P,
Treatment with alendronate plus calcium, alendronate alone, or Blichert-Toft M, Mosekilde L. Cardiovascular events before and after
calcium alone for postmenopausal low bone mineral density. Curr surgery for primary hyperparathyroidism. World J Surg
Med Res Opin 2007;23:1341-9. 2003;27:216-22.
31 Lappe JM, Travers-Gustafson D, Davies KM, Recker RR, Heaney RP. 53 Green JH, Booth C, Bunning R. Postprandial metabolic responses to
Vitamin D and calcium supplementation reduces cancer risk: results milk enriched with milk calcium are different from responses to milk
of a randomized trial. Am J Clin Nutr 2007;85:1586-91. enriched with calcium carbonate. Asia Pac J Clin Nutr
32 Lappe JM, Heaney RP. Calcium supplementation: results may not be 2003;12:109-19.
generalisable. BMJ 2008;336:403. 54 Pletcher MJ, Tice JA, Pignone M, Browner WS. Using the coronary
33 Reid IR, Ames R, Mason B, Reid HE, Bacon CJ, Bolland MJ, et al. artery calcium score to predict coronary heart disease events: a
Randomized controlled trial of calcium supplementation in healthy, systematic review and meta-analysis. Arch Int Med
nonosteoporotic, older men. Arch Intern Med 2008;168:2276-82. 2004;164:1285-92.
34 Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta- 55 Demer LL. A skeleton in the atherosclerosis closet. Circulation
analysis detected by a simple, graphical test. BMJ 1997;315:629-34. 1995;92:2029-32.
35 Chung M, Balk EM, Brendel M, Ip S, Lau J, Lee J, et al. Vitamin D and 56 Avenell A, Gillespie WJ, Gillespie LD, O’Connell D. Vitamin D and
calcium: a systematic review of health outcomes. Evidence report No vitamin D analogues for preventing fractures associated with
183. Agency for Healthcare Research and Quality, 2009. involutional and post-menopausal osteoporosis. Cochrane
36 Weingarten MA, Zalmanovici A, Yaphe J. Dietary calcium Database Syst Rev 2009;2:CD000227.
supplementation for preventing colorectal cancer and adenomatous 57 DIPART (Vitamin D Individual Patient Analysis of Randomized Trials)
polyps. Cochrane Database Syst Rev 2008;1:CD003548. Group. Patient level pooled analysis of 68 500 patients from seven
37 Pentti K, Tuppurainen MT, Honkanen R, Sandini L, Kroger H, Alhava E, major vitamin D fracture trials in US and Europe. BMJ
et al. Use of calcium supplements and the risk of coronary heart 2010;340:b5463.
disease in 52-62-year-old women: the Kuopio Osteoporosis Risk 58 Reid IR, Bolland MJ, Grey A. Effect of calcium supplementation on hip
Factor and Prevention Study. Maturitas 2009;63:73-8. fractures. Osteoporos Int 2008;19:1119-23.
38 Al-Delaimy WK, Rimm E, Willett WC, Stampfer MJ, Hu FB. A
prospective study of calcium intake from diet and supplements and Accepted: 21 May 2010
BMJ | ONLINE FIRST | bmj.com page 9 of 9