Volume 340 Number 2
·
101
CALCIUM SUPPLEMENTS FOR THE PREVENTION OF COLORECTAL ADENOMAS
CALCIUM SUPPLEMENTS FOR THE PREVENTION OF COLORECTAL ADENOMAS
J.A. B
ARON
, M.D., M. B
EACH
, M.D., P
H
.D., J.S. M
ANDEL
, P
H
.D., R.U.
VAN
S
TOLK
, M.D., R.W. H
AILE
, D
R
.P.H.,
R.S. S
ANDLER
, M.D., M.P.H., R. R
OTHSTEIN
, M.D., R.W. S
UMMERS
, M.D., D.C. S
NOVER
, M.D., G.J. B
ECK
, P
H
.D.,
J.H. B
OND
, M.D.,
AND
E.R. G
REENBERG
, M.D.,
FOR
THE
C
ALCIUM
P
OLYP
P
REVENTION
S
TUDY
G
ROUP
*
A
BSTRACT
Background and Methods
Laboratory, clinical, and
epidemiologic evidence suggests that calcium may
help prevent colorectal adenomas. We conducted a
randomized, double-blind trial of the effect of sup-
plementation with calcium carbonate on the recur-
rence of colorectal adenomas. We randomly as-
signed 930 subjects (mean age, 61 years; 72 percent
men) with a recent history of colorectal adenomas to
receive either calcium carbonate (3 g [1200 mg of el-
emental calcium] daily) or placebo, with follow-up
colonoscopies one and four years after the qualify-
ing examination. The primary end point was the pro-
portion of subjects in whom at least one adenoma
was detected after the first follow-up endoscopy but
up to (and including) the second follow-up examina-
tion. Risk ratios for the recurrence of adenomas were
adjusted for age, sex, lifetime number of adenomas
before the study, clinical center, and length of the
surveillance period.
Results
The subjects in the calcium group had a
lower risk of recurrent adenomas. Among the 913 sub-
jects who underwent at least one study colonoscopy,
the adjusted risk ratio for any recurrence of adenoma
with calcium as compared with placebo was 0.85 (95
percent confidence interval, 0.74 to 0.98; P=0.03).
The main analysis was based on the 832 subjects (409
in the calcium group and 423 in the placebo group)
who completed both follow-up examinations. At
least one adenoma was diagnosed between the first
and second follow-up endoscopies in 127 subjects in
the calcium group (31 percent) and 159 subjects in
the placebo group (38 percent); the adjusted risk ratio
was 0.81 (95 percent confidence interval, 0.67 to 0.99;
P=0.04). The adjusted ratio of the average number of
adenomas in the calcium group to that in the placebo
group was 0.76 (95 percent confidence interval, 0.60
to 0.96; P=0.02). The effect of calcium was independ-
ent of initial dietary fat and calcium intake.
Conclusions
Calcium supplementation is associat-
ed with a significant — though moderate — reduction
in the risk of recurrent colorectal adenomas. (N Engl
J Med 1999;340:101-7.)
©1999, Massachusetts Medical Society.
From the Departments of Medicine (J.A.B., R.R.), Community and Family
Medicine (J.A.B.), and Anesthesia (M.B.) and the Norris Cotton Cancer Cen-
ter (E.R.G.), Dartmouth–Hitchcock Medical Center, Lebanon, N.H.; the
Veterans Affairs Medical Center, White River Junction, Vt. (M.B.); the De-
partment of Environmental and Occupational Health, School of Public
Health and School of Medicine (J.S.M.), and the Departments of Pathology
(D.C.S.) and Medicine (J.H.B.), School of Medicine, University of Minne-
sota, and the Veterans Affairs Medical Center (J.H.B.), Minneapolis; the
Center for Colon Polyps and Colon Cancer, Department of Gastroenter-
ology (R.U.S.), and the Department of Biostatistics and Epidemiology
(G.J.B.), Cleveland Clinic Foundation, Cleveland; the Department of Pre-
ventive Medicine, University of Southern California School of Medicine,
Los Angeles (R.W.H.); the Department of Medicine, University of North
Carolina, Chapel Hill, N.C. (R.S.S.); the James A. Clifton Center for Di-
gestive Diseases, Department of Internal Medicine, University of Iowa Col-
lege of Medicine, Iowa City (R.W.S.); and the Department of Pathology,
Fairview Southdale Hospital, Minneapolis (D.C.S.). Address reprint requests
to Dr. Baron at 7927 Rubin Bldg., Dartmouth–Hitchcock Medical Center,
1 Medical Center Dr., Lebanon, NH 03756.
Other authors were H. Frankl, M.D., Department of Internal Medicine,
Southern California Permanente Group, Los Angeles; and L. Pearson,
M.Phil., Department of Community and Family Medicine, Dartmouth–
Hitchcock Medical Center, Lebanon, N.H.
*Additional study investigators are listed in the Appendix.
IETARY patterns have repeatedly been as-
sociated with the risk of colorectal neo-
plasia: a diet rich in vegetables and fruits is
associated with a lower risk, whereas intake
of animal fat and red meat seems to increase risk.
1
The
underlying mechanisms are not clear, but the chang-
es in risk may in part be due to alterations in bile acids,
which are carcinogenic in animal models.
2
D
Newmark and colleagues
3
proposed that calcium
binds bile acids in the bowel lumen, inhibiting their
proliferative and carcinogenic effects. In support of
this hypothesis, studies in animals have indicated a
protective effect of dietary calcium on bile-induced
mucosal damage and experimental bowel carcinogen-
esis.
4,5
However, the results of epidemiologic research
have been inconsistent; in some studies a decreased
risk of colorectal cancer was associated with calcium
intake, whereas in others no association was found.
6,7
Mixed results have also been reported regarding large-
bowel adenomas,
6,7
which are likely precursors of most
colorectal cancers.
8
To clarify the effect of calcium intake on colorectal
carcinogenesis, we conducted a clinical study of the
effect of supplementation with calcium carbonate on
the risk of recurrence of colorectal adenomas. We
hypothesized that subjects randomly assigned to re-
ceive calcium would have a reduced risk of recurrent
adenomas as well as reduced numbers of adenomas.
METHODS
The Calcium Polyp Prevention Study involved six clinical centers:
the Cleveland Clinic Foundation, Dartmouth–Hitchcock Medical
Center, the University of Southern California–Southern California
Permanente Medical Group, the University of Iowa, the Univer-
sity of Minnesota, and the University of North Carolina. Dart-
mouth was the coordinating center, and the University of Minne-
sota was the pathology center. Human-subjects committees at
each center approved the study protocol; an independent data and
safety monitoring committee reviewed the study twice a year.
The New England Journal of Medicine
Downloaded from nejm.org on February 5, 2021. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
102
·
January 14, 1999
The New England Journal of Medicine
Recruitment and Randomization
Staff at each clinical center monitored colonoscopy and pathology
records at associated endoscopy units to identify subjects who had
at least one histologically confirmed large-bowel adenoma removed
within three months before recruitment and whose entire large-
bowel mucosa was subsequently examined and judged free of pol-
yps. Eligible subjects were less than 80 years old, in good health, and
without a history of familial polyposis, invasive large-bowel cancer,
malabsorption syndromes, or any condition that might be worsened
by supplemental calcium. Our goal was for 860 subjects to undergo
randomization in order for the study to have 80 percent power to
detect a 25 percent reduction in the recurrence of adenomas.
We reviewed data on 2918 apparently eligible subjects. We were
unable to contact 223, 1066 declined to participate, 510 were
found to be ineligible, and 1 did not enroll for unknown reasons.
After written informed consent had been obtained, the remaining
1118 subjects began a three-month placebo run-in period to assess
their adherence to the study regimen of one tablet twice a day with
meals. At the end of the run-in period, 930 subjects had taken at
least 80 percent of their prescribed tablets, wished to continue the
study, and were considered appropriate for randomization. We
assigned these subjects to calcium or placebo using computer-
generated random numbers, blocked according to study center.
The study tablets contained a total of 3 g of calcium carbonate
(1200 mg of elemental calcium) or an identical-appearing cellu-
lose–sucrose placebo. The trial was double-blind: neither subjects
nor study staff were aware of the treatment assignments.
Study Protocol
The subjects underwent two follow-up colonoscopies as part of
their routine clinical care, usually by the same physician who had
conducted the initial examination. The first follow-up examina-
tion was planned for approximately 1 year after the qualifying
colonoscopy (about 9 months after randomization), and the sec-
ond follow-up examination was planned for 36 months after that.
Large-bowel endoscopy was otherwise discouraged unless clini-
cally indicated (e.g., for rectal bleeding). Follow-up examinations
were considered adequate if the entire large-bowel mucosa was
visualized and no polyps remained at the end of the procedure.
We designated the time from randomization to the first follow-
up examination as the first study interval, and the period follow-
ing the first follow-up examination and through the second as the
second study interval (the main risk period).
At each colonoscopy, the endoscopist recorded the size and loca-
tion of all mucosal lesions, using standard clinical technique. Accord-
ing to the protocol, all polyps were removed and examined histolog-
ically at the clinical center and by the study pathologist, who
classified the polyps as neoplastic (adenomas) or non-neoplastic
(e.g., hyperplastic polyps or lymphoid follicles). The study patholo-
gist also reviewed polyps detected by the qualifying endoscopic ex-
amination for a sample of 25 percent of the subjects. The study pa-
thologist and the clinical center agreed as to presence or absence of
neoplasia in 2349 of the 2541 specimens reviewed (92 percent). In
cases of disagreement, we accepted the study pathologist’s diagnosis.
At enrollment and at the time of each of the two follow-up co-
lonoscopies, we obtained specimens of venous blood in mineral-
free tubes. Serum was initially stored at ¡20°C or below, pending
shipment to Dartmouth for storage at ¡70°C until analysis. At
enrollment and at the end of the study, we also assessed the sub-
jects’ diet with a validated food-frequency questionnaire.
9
Every
six months, we sent questionnaires to the subjects regarding their
adherence to study treatment; their use of medications, over-the-
counter drugs, and nutritional supplements; and the occurrence
of symptoms, illnesses, and hospitalizations. Recruitment began
in November 1988 and ended in April 1992. Follow-up ended in
December 1996.
End Points
The primary outcome measure was the proportion of subjects
in whom at least one adenoma was detected during the second
study interval — that is, after the first follow-up colonoscopy, up
to and including the second follow-up examination (including
adenomas detected during interim endoscopies). This end point
provided for the removal of adenomas overlooked at the qualify-
ing colonoscopy (thus minimizing the numbers of polyps present
at the start of the main risk period) and allowed for a latent pe-
riod of calcium action. If a subject did not undergo the follow-
up examinations as planned, we used the two clinically indicated
colonoscopies at least one year apart that provided the longest
follow-up interval.
Statistical Analysis
For our statistical analyses, we compared proportions using
Fisher’s exact test and measured data using t-tests or rank tests.
10
Our main analysis considered two related outcomes: whether sub-
jects in the two treatment groups had different probabilities of
having at least one adenoma, and whether the average numbers
of adenomas in the two groups differed. To address the first ques-
tion, we used overdispersed log-linear quasi-likelihood models pro-
grammed in SPlus (MathSoft, Seattle) to provide unadjusted and
adjusted estimates and confidence intervals for the relative risk of
at least one recurrent adenoma.
11
Similar models (with variance
proportional to the mean) were used to analyze the ratios of the
average number of adenomas in the two treatment groups.
11
Covariates included age (as a linear term), sex, the lifetime
number of adenomas before study entry, clinical center, and the
length of the surveillance period. Possible interactions were con-
sidered with the use of product interaction terms. Subgroup anal-
yses included investigation of subjects whose diets were above and
below the median for the calorie-adjusted intake
12
of selected nu-
trients. To assess possible distortions introduced by subjects who
did not complete the study, we also performed sensitivity analyses
by imputing patterns of recurrence for these subjects to deter-
mine outcomes that would have altered our conclusions, had they
been observed. All P values were two-sided; P<0.05 was taken
to indicate significance.
RESULTS
A total of 930 subjects, whose characteristics are
summarized in Table 1, underwent randomization;
there were no significant differences between the
two treatment groups in demographic characteris-
tics, dietary patterns, or history of adenomas. The
mean (±SD) age was 61±9 years, and 72 percent
were men. Most subjects had had only one or two
adenomas removed from the large bowel before en-
tering the study. The mean estimated diameter of
the largest qualifying adenoma was 0.7±0.6 cm; in
the sample sent for pathological review, 99 percent
of the specimens had mild or moderate atypia. The
mean estimated daily dietary intake of calcium at
study entry was similar in the two study groups and
was less than three quarters of the amount later pro-
vided in the form of supplements by the study inter-
vention. Fewer than 3 percent of the subjects were
taking calcium supplements at the start of the trial;
all agreed to discontinue them during the study.
Of the 930 subjects who underwent randomization,
832 (89 percent) completed two follow-up colon-
oscopies (Table 2). We could not include 98 subjects
(43 in the placebo group and 55 in the calcium
group) in the main analyses: 47 died, 25 no longer
wished to participate, 18 could not be examined be-
cause they were too ill or had moved, and 8 dropped
The New England Journal of Medicine
Downloaded from nejm.org on February 5, 2021. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
CALCIUM SUPPLEMENTS FOR THE PREVENTION OF COLORECTAL ADENOMAS
Volume 340 Number 2
·
103
out for unknown reasons. In addition to the study-
mandated colonoscopies, an interval colonoscopy or
sigmoidoscopy was performed during the main risk
period (second study interval) in 86 subjects. The
proportions of subjects with inadequate study colon-
oscopies or with interim endoscopies did not differ
significantly between the treatment groups (Table 2).
Self-reported adherence to the study regimen grad-
ually declined during the trial (Table 3). Neverthe-
less, even during the fourth year, over 80 percent of
the subjects took the study agents 90 to 100 percent
of the time, and a further 7 percent took them 50
to 89 percent of the time. Use of supplemental cal-
cium was reported at least once by only 19 subjects
(2 percent) during the study (9 in the placebo group
and 10 in the calcium group).
*Plus–minus values are means ±SD. None of the differences between groups were significant.
Because of rounding, percentages do not always total 100.
†The lifetime number of colorectal adenomas found and removed before randomization is given.
‡Dietary information was missing for 10 subjects in the placebo group and 13 in the calcium
group.
T
ABLE
1.
B
ASE
-L
INE
C
HARACTERISTICS
OF
THE
930 S
UBJECTS
.*
C
HARACTERISTIC
A
LL
R
ANDOMIZED
S
UBJECTS
S
UBJECTS
W
HO
C
OMPLETED
S
TUDY
PLACEBO
(
N
=
466)
CALCIUM
(
N
=
464)
PLACEBO
(
N
=
423)
CALCIUM
(
N
=
409)
Sex — no. (%)
Male
Female
327 (70)
139 (30)
345 (74)
119 (26)
296 (70)
127 (30)
302 (74)
107 (26)
Age — yr 61.0±9.1 61.0±9.1 60.9±9.0 60.7±8.8
Study center — no. (%)
Cleveland Clinic
Dartmouth–Hitchcock
University of Iowa
University of Minnesota
University of North Carolina
University of Southern California
72 (15)
85 (18)
87 (19)
81 (17)
64 (14)
77 (17)
71 (15)
72 (16)
86 (19)
85 (18)
59 (13)
91 (20)
70 (17)
76 (18)
79 (19)
75 (18)
56 (13)
67 (16)
67 (16)
62 (15)
74 (18)
72 (18)
51 (12)
83 (20)
No. of prior adenomas†
Daily dietary intake‡
Calories — kcal
Fat — g
Fiber — g
Calcium — mg
Taking supplemental calcium
— no. (%)
2.6±2.8
2010±756
88.1±42.9
16.2±7.8
865±423
13 (3)
2.4±2.5
2040±761
87.2±41.3
16.6±8.0
889±451
11 (2)
2.6±2.9
2011±742
87.9±42.4
16.4±8.0
866±421
12 (3)
2.5±2.6
2032±756
86.1±40.5
16.7±8.0
893±451
11 (3)
*Plus–minus values are means ±SE.
†P=0.04 for the difference between groups.
T
ABLE
2.
N
UMBERS
OF
S
TUDY
S
UBJECTS
W
HO
C
OMPLETED
THE
S
TUDY
E
XAMINATIONS
.*
S
UBJECTS
P
LACEBO
(N=466) C
ALCIUM
(N=464)
Died — no. (%) 22 (5) 25 (5)
Dropped out — no. (%)
Lost interest
Ill or moved
Other or unknown reasons
11 (2)
8 (2)
2 (<1)
14 (3)
10 (2)
6 (1)
Received first follow-up colonoscopy — no. (%)
First surveillance interval — mo
Interim endoscopy during first surveillance
interval — no. (%)
Inadequate first follow-up colonoscopy — no. (%)
459 (98)
13±0.2
5 (1)
21 (5)
454 (98)
14±0.3
14 (3)†
22 (5)
Received second follow-up colonoscopy — no. (%)
Second surveillance interval — mo
Interim endoscopy during second surveillance
interval — no. (%)
Inadequate second follow-up colonoscopy — no. (%)
423 (91)
37±0.2
51 (12)
29 (7)
409 (88)
37±0.2
35 (9)
35 (9)
The New England Journal of Medicine
Downloaded from nejm.org on February 5, 2021. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
104
·
January 14, 1999
The New England Journal of Medicine
Among the 832 subjects who completed the study,
at least one colorectal adenoma was diagnosed dur-
ing the main risk period (the second study interval)
in 127 subjects in the calcium group (31 percent)
and 159 subjects in the placebo group (38 percent)
(Table 4). The mean size of the largest adenoma was
the same in the two groups (0.4 cm; P=0.43), but
more adenomas were found in the placebo group
(mean number per patient, 0.73 vs. 0.55; P=0.03).
The unadjusted risk ratio for having at least one ad-
enoma in the calcium group as compared with the
placebo group was 0.83 (95 percent confidence in-
terval, 0.68 to 1.00; P=0.05); after adjustment the
risk ratio was 0.81 (95 percent confidence interval,
0.67 to 0.99; P=0.04). The unadjusted ratio of the
average number of adenomas in the calcium group
to that in the placebo group was 0.75 (95 percent
confidence interval, 0.58 to 0.97; P=0.03); after ad-
justment it was 0.76 (95 percent confidence interval,
0.60 to 0.96; P=0.02). During the main risk period,
invasive large-bowel cancer was found in four sub-
jects (three in the placebo group and one in the cal-
cium group), but no adenomas with severe atypia
were found (P=0.62 for the difference in the pro-
portions with severe atypia or cancer). Analysis of
adenomas detected at the second follow-up exami-
nation (excluding findings on interval endoscopies)
showed similar results (Table 4).
A similar effect of calcium was found during the
first study interval. Among the subjects who com-
pleted the trial, at least one adenoma was found in the
period up to and including the first follow-up exam-
ination in 103 subjects in the calcium group (25 per-
cent) and 138 subjects in the placebo group (33 per-
cent) (Table 4). The unadjusted risk ratio for at least
one adenoma in this early interval was 0.77 (95 per-
cent confidence interval, 0.62 to 0.96; P=0.02); the
unadjusted ratio of the average numbers of adenomas
was 0.73 (95 percent confidence interval, 0.54 to
0.97; P=0.03). These estimates were virtually un-
changed after multivariate adjustment. Analysis of ad-
enomas detected at the first follow-up examination
*Numbers of subjects do not sum to 930 because
of dropouts, deaths, and missing data. Because of
rounding, percentages do not always total 100.
T
ABLE
3.
S
ELF
-R
EPORTED
A
DHERENCE
TO
S
TUDY
T
REATMENT
, A
CCORDING
TO
T
REATMENT
A
SSIGNMENT
AND
S
TUDY
Y
EAR
.*
Y
EAR
AND
P
ERCENTAGE
OF
T
ABLETS
T
AKEN
P
LACEBO
(N=466)
C
ALCIUM
(N=464)
number (percent)
Year 1
90–100
50–89
<50
409 (88)
42 (9)
14 (3)
393 (85)
50 (11)
18 (4)
Year 2
90–100
50–89
<50
373 (81)
52 (11)
38 (8)
371 (82)
44 (10)
40 (9)
Year 3
90–100
50–89
<50
377 (83)
33 (7)
44 (10)
358 (80)
39 (9)
52 (12)
Year 4
90–100
50–89
<50
358 (82)
31 (7)
49 (11)
346 (79)
34 (8)
58 (13)
*The first study interval was from randomization to the first follow-up colonoscopy; the second study interval (the main risk period) was
after the first follow-up colonoscopy and up to and including the second follow-up colonoscopy. Four hundred twenty-three subjects in the
placebo group and 409 in the calcium group completed the study; 459 and 454, respectively, had at least one endoscopy.
†The risk ratio for at least one adenoma and the ratio of the mean numbers of adenomas in the calcium group as compared with the
placebo group are given. Both estimates have been adjusted for age, sex, clinical center, number of previous adenomas, and length of follow-up.
CI denotes confidence interval.
T
ABLE
4.
O
UTCOMES
WITH
R
ESPECT
TO
R
ECURRENCE
OF
A
DENOMAS
.
S
UBJECTS
*P
LACEBO
C
ALCIUM
A
DJUSTED
R
ELATIVE
R
ISK
OF
»1 A
DENOMA
(95% CI)†
A
DJUSTED
R
ATIO
OF
M
EAN NO. OF
A
DENOMAS
(95% CI)†
PERCENTAGE WITH
»1
ADENOMA
MEAN
NO. OF
ADENOMAS
PERCENTAGE
WITH
»1
ADENOMA
MEAN
NO. OF
ADENOMAS
Completed study
First study interval
First study examination
Second study interval
Second study examination
First or second study interval
33
33
38
36
52
0.60
0.59
0.73
0.62
1.32
25
24
31
30
45
0.43
0.40
0.55
0.51
0.98
0.78 (0.63–0.96)
0.75 (0.61–0.94)
0.81 (0.67–0.99)
0.83 (0.68–1.01)
0.85 (0.74–0.98)
0.75 (0.58–0.96)
0.70 (0.54–0.89)
0.76 (0.60–0.96)
0.83 (0.65–1.05)
0.75 (0.62–0.90)
Had at least one endoscopy
First or second study interval
Study examinations
51
50
1.26
1.15
43
42
0.92
0.86
0.85 (0.74–0.98)
0.84 (0.73–0.97)
0.75 (0.63–0.90)
0.77 (0.64–0.91)
The New England Journal of Medicine
Downloaded from nejm.org on February 5, 2021. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
CALCIUM SUPPLEMENTS FOR THE PREVENTION OF COLORECTAL ADENOMAS
Volume 340 Number 2 · 105
yielded similar findings. At or before the first follow-
up examination, invasive cancer was found in four
subjects (two in the calcium group and two in the
placebo group), and an adenoma with severe atypia
was removed from one subject in each group.
A total of 913 subjects underwent at least one study
colonoscopy. The unadjusted risk ratio for having at
least one adenoma after randomization was 0.85 (95
percent confidence interval, 0.74 to 0.98; P=0.03);
the corresponding ratio of the average numbers of
adenomas was 0.74 (95 percent confidence interval,
0.59 to 0.92; P<0.001). Restriction of the analysis
to adenomas detected at study follow-up examina-
tions and adjustment for age, clinical center, sex,
length of the surveillance period, and number of
previous adenomas left these estimates substantially
unchanged (Table 4).
We also assessed whether the effect of calcium sup-
plementation differed according to the size or location
of the adenomas. During the second study interval,
an adenoma 0.5 cm or greater in diameter was found
in 120 subjects (63 in the placebo group and 57 in
the calcium group); the unadjusted risk ratio for hav-
ing at least one adenoma of this size was 0.87 (95
percent confidence interval, 0.63 to 1.21; P=0.70).
In 166 subjects, the largest adenoma was less than
0.5 cm in diameter (96 in the placebo group and 70
in the calcium group); the corresponding unadjusted
risk ratio was 0.75 (95 percent confidence interval,
0.57 to 0.98; P=0.03). During the second interval,
144 subjects had at least one adenoma in the splenic
flexure or more distally, and 200 had at least one ad-
enoma proximal to the splenic flexure. Calcium had a
similar effect on the recurrence of adenoma in both
regions of the bowel (data not shown).
The sensitivity analysis suggested that it is ex-
tremely unlikely that the outcomes of the 98 subjects
who did not complete the study would have nullified
our findings had they been able to be included.
Among these subjects, recurrent adenomas would
have had to be at least twice as frequent in the cal-
cium group as in the placebo group to eliminate the
statistical significance of the overall effect of calcium.
There was no evidence of modification of the ef-
fect of calcium by age, sex, or base-line dietary intake
of calcium, fat, or fiber (data not shown). The effect
of calcium was nonsignificantly stronger among sub-
jects who reported taking all their study agents and
among those who did not report any use of aspirin
or other nonsteroidal antiinflammatory drugs (data
not shown).
Medical symptoms and complications were not as-
sociated with treatment assignment. Similar propor-
tions of subjects in the calcium and placebo groups
were hospitalized for any reason, were hospitalized
with cancer, or stopped treatment because of per-
ceived side effects (Table 5). The frequency of diges-
tive symptoms (including constipation) did not dif-
fer substantially between the two treatment groups.
Two subjects assigned to calcium and one assigned
to placebo were found to have definite or probable
urinary stones during the study.
DISCUSSION
In this randomized, clinical trial, assignment to
calcium supplementation was associated with a sig-
nificant — though moderate — reduction in the risk
of recurrent adenomas. The reduced risk became ap-
parent as early as the first colonoscopic follow-up,
after approximately nine months of treatment. There
was no indication of a greater effect among subjects
with a low base-line dietary intake of calcium or a
high intake of fat. The intervention was well accept-
ed and without major toxicity.
Epidemiologic data regarding the association be-
tween dietary calcium and the risk of colorectal can-
cer have varied considerably but in the aggregate are
consistent with the effect we observed.
6,7
Many
studies
13-18
found at least suggestions of an inverse
association, but others found no relation
19,20
or even
the possibility of an increased risk with higher in-
take.
21,22
The results of investigations of calcium in-
take and the risk of colorectal adenomas have also
been conflicting,
20,23-25
as have those of studies that
considered calcium supplementation separately.
16,21,26
These mixed findings may reflect the difficulties of
dietary epidemiology. The effects of calcium intake
are likely to be confounded by factors such as intake
of calories, dietary fat, and phosphate and perhaps use
of vitamin and mineral supplements, aspirin, or other
agents with anticarcinogenic effects. Moreover, the
measurement error inherent in dietary assessment
would tend to obscure any association between cal-
cium intake and the risk of neoplasia.
12
*There were no significant differences between
the groups.
†Some patients were hospitalized for more than
one disorder.
TABLE 5. MEDICAL EVENTS
AFTER RANDOMIZATION.*
EVENT
PLACEBO
(N=466)
CALCIUM
(N=464)
no. (%)
Deaths 22 (5) 25 (5)
Subjects hospitalized†
All cancers
Cardiac disease
Stroke
Gastrointestinal disease
Other
164 (35)
21 (5)
46 (10)
11 (2)
32 (7)
114 (24)
172 (37)
15 (3)
50 (11)
12 (3)
38 (8)
126 (27)
Stopped treatment because
of perceived toxicity
13 (3) 12 (3)
The New England Journal of Medicine
Downloaded from nejm.org on February 5, 2021. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.