Effects of Multivitamin Supplement on Cataract

and Age-Related Macular Degeneration in a

Randomized Trial of Male Physicians
William G. Christen, ScD,1 Robert J. Glynn, ScD,1,3,8 JoAnn E. Manson, MD, DrPH,1,7 Jean MacFadyen, BA,1
Vadim Bubes, PhD,1 Miriam Schvartz, MD,1 Julie E. Buring, ScD,1,2,6,7 Howard D. Sesso, ScD, MPH,1,2,7
J. Michael Gaziano, MD, MPH1,2,4,5

Purpose: To test whether long-term multivitamin supplementation affects the incidence of cataract or age-
related macular degeneration (AMD) in a large cohort of men.
Design: Randomized, double-blind, placebo-controlled trial.
Participants: A total of 14 641 US male physicians aged
50 years.
Intervention: Daily multivitamin or placebo.
Main Outcome Measures: Incident cataract and visually significant AMD responsible for a reduction in best-
corrected visual acuity to 20/30 or worse based on self-reports confirmed by medical record review.
Results: During an average of 11.2 years of treatment and follow-up, a total of 1817 cases of cataract and
281 cases of visually significant AMD were confirmed. There were 872 cataracts in the multivitamin group and 945
cataracts in the placebo group (hazard ratio [HR], 0.91; 95% confidence interval [CI], 0.83e0.99; P ¼ 0.04). For
visually significant AMD, there were 152 cases in the multivitamin group and 129 cases in the placebo group (HR,
1.19; 95% CI, 0.94e1.50; P ¼ 0.15).
Conclusions: These randomized trial data from a large cohort of middle-aged and older US male physicians
indicate that long-term daily multivitamin use modestly and significantly decreased the risk of cataract but had no
significant effect on visually significant AMD. Ophthalmology 2014;121:525-534 ª 2014 by the American
Academy of Ophthalmology.

Nutritional factors are postulated to play a causal role in the AMD in 1 eye.24 A second trial, in a population at high risk
development of cataract and age-related macular degenera- of cardiovascular disease (CVD), showed that daily folic acid,
tion (AMD), 2 leading causes of visual impairment in older vitamin B6, and vitamin B12 reduced AMD incidence by 35%
Americans.1 Considerable observational evidence suggests to 40%.25 Five other trials testing high-dose vitamin E,
that persons with higher dietary intake or blood levels of vitamin C, and beta-carotene, alone or in combination, re-
nutrients with antioxidant capabilities have lower rates of ported no benefit on AMD.26e30 We present the final results
cataract and AMD.2e7 Moreover, animal studies show that for cataract and AMD from the multivitamin component of
supplementation with antioxidants and other micronutrients the Physicians’ Health Study (PHS) II.31
can prevent or delay the formation of lens opacities3,8,9 and
can reduce the damaging effects of reactive oxygen species
on the retina.10e12 However, results of nutritional supple- Methods
mentation trials in humans have been inconsistent.
For cataract, randomized trials conducted among older, Study Design
generally well-nourished populations indicate that high-dose The PHS II was a randomized, double-blind, placebo-controlled,
supplements of selected nutrients (most commonly vitamin factorial trial evaluating a daily multivitamin (Centrum Silver;
E, vitamin C, and beta-carotene), alone or in combination, for Pfizer Inc., Madison, NJ), alternate-day vitamin E (400 IU syn-
up to 10 years have little material impact on cataract thetic a-tocopherol), and daily vitamin C (500 mg synthetic
occurrence.13e21 Conversely, 2 trials testing a multivitamin ascorbic acid) in the prevention of cancer and CVD among 14 641
supplement, one in a nutritionally deficient population in male physicians aged 50 years and older.31 A fourth randomized
China22 and the other in a well-nourished population in component, alternate day beta-carotene (50 mg Lurotin; BASF
Italy,23 reported significantly lower rates of cataract in the Corporation, Florham Park, NJ), was terminated in March 2003.
The vitamin E and vitamin C components of the trial ended as
multivitamin group. For AMD, the Age-Related Eye Dis-
scheduled on August 31, 2007.18,30,32,33 Cataract and AMD were
ease Study (AREDS) demonstrated that daily supplementa- prespecified secondary end points of PHS II.
tion with zinc and high-dose antioxidants vitamin E, vitamin The study design for PHS II has been described by Christen
C, and beta-carotene could reduce the risk of advanced AMD et al.31 Briefly, recruitment, enrollment, and randomization of men
by 25% in persons with intermediate AMD or advanced into PHS II occurred in 2 phases (Fig 1). Phase 1 began in 1997

 2014 by the American Academy of Ophthalmology ISSN 0161-6420/14/$ - see front matter 525
Published by Elsevier Inc. http://dx.doi.org/10.1016/j.ophtha.2013.09.038
 Ophthalmology Volume 121, Number 2, February 2014

Figure 1. Flow diagram of the multivitamin component of the Physicians’ Health Study II. A total of 3144 participants who reported a diagnosis of cataract
at baseline and 408 participants who reported a diagnosis of age-related macular degeneration (AMD) at baseline were excluded.

and included 7641 willing and eligible participants from PHS For new reports of cataract or AMD, participants were asked to
I20,29,34,35 who retained their original beta-carotene treatment provide written consent to obtain medical records. Ophthalmolo-
assignment and were newly randomized to a multivitamin, vitamin gists/optometrists were contacted by mail and asked to provide
C, and vitamin E. Phase 2 began in 1999 and consisted of 7000 information about the reported end point by completing a ques-
new physician participants who were independently randomized to tionnaire or supplying copies of relevant medical records.
each of a multivitamin, beta-carotene, vitamin C, and vitamin E, or
their matching placebos. All participants provided written informed Cataract
consent, and PHS II was approved by the institutional review board The cataract questionnaire asked about the presence of lens opac-
of the Brigham and Women’s Hospital, Boston, Massachusetts. ities, diagnosis date, best-corrected visual acuity, cataract extrac-
Participants completed annual questionnaires providing infor- tion, other ocular abnormalities, and cataract type and cause.
mation on compliance with pill taking, potential adverse events, The co-primary vision end point was incident cataract, defined
updated risk factors, and the occurrence of any new end points, as a confirmed lens opacity, diagnosed after randomization but
including cataract and AMD. Treatment and follow-up continued before June 1, 2011; age-related in cause; and responsible for
in blinded fashion through June 1, 2011, the scheduled end of the reduced best-corrected visual acuity to 20/30 or worse. Cataract
multivitamin component. Morbidity and mortality follow-up were extraction was a prespecified secondary end point and was defined
98.2% and 99.9%, respectively. as the surgical removal of an incident cataract.
Compliance with pill taking was based on self-report and defined A total of 11 497 participants did not report cataract at baseline
as taking at least two thirds of the study agents. Adherence at 6 years and are included in this analysis. Of these, 5736 men were in the
was 73.6% for active multivitamin and 73.3% for placebo (P ¼ 0.68). multivitamin group and 5761 men were in the placebo group (Fig 1).

Ascertainment and Confirmation of Study End Points Age-Related Macular Degeneration

Participants who reported cataract (n ¼ 3144) or AMD (n ¼ 408) at The AMD questionnaire requested information about diagnosis date,
baseline were excluded from analyses of the respective end point. best-corrected visual acuity at diagnosis, and date when best-corrected

526
 Christen et al 
Multivitamins and Cataract and AMD

Table 1. Baseline Characteristics by Multivitamin Randomized Treatment Assignment in Physicians’ Health Study II*

Cataract Visually Significant AMD

No. of Multivitamin Placebo No. of Multivitamin Placebo
Participantsy (n ¼ 5736) (n ¼ 5761) Participants (n ¼ 7111) (n ¼ 7122)
Age, yrs (mean [SD]) 61.9 (7.9) 62.0 (7.9) 63.9 (8.9) 64.0 (9.0)
Age, yrs
50e59 5606 49.1 48.5 5865 41.2 41.2
60e69 3914 33.8 34.3 4647 32.8 32.5

70 1977 17.2 17.2 3721 26.0 26.3
Cigarette smoking
Never 6824 60.0 58.8 8072 57.1 56.4
Former 4264 36.7 37.6 5642 39.4 39.9
Current 400 3.4 3.6 507 3.5 3.6
Alcohol use
Rarely/never 2108 18.9 18.0 2650 19.2 18.3

1 drink/mo 9317 81.1 82.0 11 492 80.8 81.7
Body mass index (kg/m2)
<25 4717 41.3 40.8 5929 41.8 41.5
25e<30 5551 47.8 48.7 6817 47.8 48.0

30 1228 10.9 10.5 1484 10.4 10.5
History of hypertensionz
Yes 4426 38.1 39.4 5896 41.0 42.3
No 6998 61.9 60.6 8258 59.0 57.7
History of high cholesterolx
Yes 3979 35.5 36.5 5048 36.1 37.3
No 7076 64.5 63.5 8706 63.9 62.7
History of diabetes
Yes 547 4.8 4.8 853 6.3 5.7
No 10 940 95.2 95.2 13 365 93.7 94.3
Aspirin use
Yes 8742 77.7 76.6 10 868 77.7 77.2
No 2589 22.3 23.4 3158 22.3 22.8
Exercise
1 time/wk
Yes 7023 63.3 61.9 8565 62.5 60.9
No 4194 36.7 38.1 5319 37.5 39.1
Self-reported history of
CVDk
Yes 454 3.7 4.2 717 5.0 5.0
No 11 043 96.3 95.8 13 516 95.0 95.0

AMD ¼ age-related macular degeneration; CVD ¼ cardiovascular disease; SD ¼ standard deviation.

*Data are given as percentage of participants unless otherwise noted.
y
For some variables, numbers do not total 11 497 for cataract and 14 233 for AMD because of missing data for that variable.
z
History of hypertension was defined as self-reported systolic blood pressure of at least 140 mmHg, diastolic blood pressure of at least 90 mmHg, or past or
current treatment for hypertension.
x
History of high cholesterol was defined as self-reported total cholesterol level of at least 240 mg/dl or past or current treatment for high cholesterol.
k
History of CVD included nonfatal myocardial infarction or nonfatal stroke.

visual acuity reached 20/30 or worse. Information also was requested A total of 14 233 participants did not report AMD at baseline and
about signs of AMD observed (drusen, retinal pigment epithelium are included in this analysis. Of these, 7111 men were in the
[RPE] hypo-/hyperpigmentation, geographic atrophy, RPE detach- multivitamin group and 7122 men were in the placebo group (Fig 1).
ment, subretinal neovascular membrane, or disciform scar) when vi-
sual acuity was first noted to be 20/30 or worse, and the date Statistical Analysis
exudative neovascular disease (defined by presence of RPE detach-
ment, subretinal neovascular membrane, or disciform scar) was first In separate analyses of cataract and AMD, participants were clas-
noted. The questionnaire also asked whether there were other ocular sified according to randomized multivitamin treatment assignment
abnormalities and, if so, whether the AMD, by itself, was significant and followed until the occurrence of that end point, death, or the
enough to reduce best-corrected visual acuity to 20/30 or worse. end of the multivitamin component of PHS II, whichever came
The co-primary vision end point was visually significant AMD first. Estimated power for the primary study end points of incident
defined as confirmed AMD diagnosed after randomization but cataract and visually significant AMD was based on event rates
before June 1, 2011, and best-corrected visual acuity loss to 20/30 observed in PHS I and estimated to be greater than 80% to detect a
or worse attributable to AMD. Two secondary end points were 10% reduction in cataract and a 20% reduction in visually signif-
total AMD, composed of all incident cases with or without vision icant AMD.
loss, and advanced AMD, composed of cases of exudative neo- The distributions of baseline characteristics in the multivi-
vascular AMD plus geographic atrophy. tamin and placebo groups were compared using 2-sample t tests,

527
 Ophthalmology Volume 121, Number 2, February 2014

Table 2. Confirmed Cases of Age-Related Cataract and Cataract Extraction According to Multivitamin Randomized Treatment
Assignment in Physicians’ Health Study II

Cataract Cataract Extraction

Cataract Active Placebo Active Placebo
by Age (n ¼ 5736) (n ¼ 5761) HR* 95% CI P Value P Trendy (n ¼ 5736) (n ¼ 5761) HR* 95% CI P Value P Trendy
Total cataract
50e59 yrs 203 209 0.97 0.80e1.17 0.74 0.29 151 145 1.04 0.83e1.31 0.73 0.18
60e69 yrs 407 436 0.94 0.82e1.07 0.33 291 329 0.88 0.75e1.03 0.12

70 yrs 262 300 0.85 0.72e1.00 0.06 194 227 0.84 0.69e1.02 0.08
Subtotal 872 945 0.91 0.83e0.99 0.04 636 701 0.89 0.80e0.99 0.04
Any NS
50e59 yrs 175 187 0.93 0.76e1.15 0.51 0.46 132 129 1.02 0.80e1.30 0.85 0.29
60e69 yrs 374 425 0.88 0.77e1.01 0.08 272 319 0.85 0.72e1.00 0.05

70 yrs 251 288 0.85 0.72e1.01 0.06 186 215 0.85 0.70e1.04 0.11
Subtotal 800 900 0.87 0.79e0.96 0.005 590 663 0.87 0.78e0.98 0.018
Any cortical
50e59 yrs 75 79 0.94 0.69e1.30 0.72 0.27 56 54 1.03 0.71e1.50 0.86 0.11
60e69 yrs 169 169 1.00 0.81e1.24 0.98 123 130 0.95 0.74e1.21 0.66

70 yrs 112 139 0.79 0.61e1.01 0.06 78 107 0.72 0.54e0.96 0.027
Subtotal 356 387 0.90 0.78e1.04 0.17 257 291 0.87 0.73e1.03 0.10
Any PSC
50e59 yrs 84 69 1.21 0.88e1.66 0.24 0.18 75 55 1.36 0.96e1.92 0.08 0.13
60e69 yrs 102 112 0.91 0.70e1.20 0.51 85 96 0.89 0.66e1.19 0.42

70 yrs 61 67 0.89 0.63e1.26 0.50 53 56 0.93 0.64e1.35 0.69
Subtotal 247 248 0.98 0.82e1.17 0.85 213 207 1.02 0.84e1.23 0.86

CI ¼ confidence interval; HR ¼ hazard ratio; NS ¼ nuclear sclerosis; PSC ¼ posterior subcapsular.

*Adjusted for age; Physicians’ Health Study cohort; and vitamin C, vitamin E, and beta-carotene treatment assignment.
y
Test for trend of the effect of age on the association between randomized treatment assignment and cataract.

chi-square tests for proportions, and tests for trend for ordinal Individuals were considered the unit of analysis because eyes
categories. KaplaneMeier curves estimated cumulative incidence were not examined independently and were classified according to
over time by randomized group and were compared using a crude the status of the worse eye as defined by disease severity.36,37
log-rank test. Cox proportional-hazards models were used to es-
timate the hazard ratio (HR) of cataract and AMD among those in
the multivitamin group compared with placebo after adjustment Results
for age (years) at baseline, PHS cohort (original PHS I partici-
pant, new PHS II participant), and randomized beta-carotene, As expected in this large randomized trial, baseline characteristics
vitamin E, and vitamin C assignments. Models also were fit had comparable distributions between the multivitamin and placebo
separately within 3 baseline age groups: 50 to 59 years, 60 to 69 groups (Table 1). During a mean follow-up of 11.2 years (median
years, and
70 years. Tests of trend for the effect of age on the [interquartile range], 11.2 years [10.7e13.3 years]; maximum, 13.8
association between multivitamins and cataract or AMD were years), 1817 cataracts and 1337 cataract extractions were confirmed.
calculated by including a term for the interaction of multivitamins We also confirmed 538 cases of AMD, including 281 cases of
and age (with values 1 to 3 corresponding to the 3 age groups) in visually significant AMD and 144 cases of advanced AMD.
a proportional hazards model. The proportionality assumption
was not violated for any cataract (diagnosis, P ¼ 0.99; extraction, Cataract
P ¼ 0.75) or AMD end point (visually significant AMD, P ¼
0.40; total AMD, P ¼ 0.86; advanced AMD, P ¼ 0.98). Ninety- Overall, there was a significant 9% lower risk of cataract in the
five percent confidence intervals (CIs) and 2-sided P values were multivitamin group compared with placebo (872 vs. 945 cases;
calculated. HR, 0.91; 95% CI, 0.83e0.99; P ¼ 0.04) (Table 2). For subtypes,
We analyzed subgroup data by categories of baseline variables there was a significant 13% reduced risk of nuclear sclerosis in the
and by the other randomized treatment assignments. We explored multivitamin group (800 vs. 900 cases; HR, 0.87; 95% CI,
possible effect modification by using interaction terms between 0.79e0.96; P ¼ 0.005). There was a nonsignificant reduction in
subgroup indicators and multivitamin assignment. cortical cataract (356 vs. 387 cases; HR, 0.90; 95% CI,
We also considered the possibility that the apparent effect of 0.78e1.04; P ¼ 0.17). There were similar numbers of posterior
multivitamins on 1 end point (e.g., cataract) reflected, at least in subcapsular (PSC) cataracts in the multivitamin and placebo
part, the effect of the intervention on the second end point (e.g., groups (247 vs. 248 cases; HR, 0.98; 95% CI, 0.82e1.17; P ¼
AMD). To address this, 2 separate proportional hazards models 0.85). The findings were similar for extraction of cataract and
were fitted to estimate the effect of the intervention on 1 end point subtypes.
while adjusting for a diagnosis of the other as a time-varying The benefits of daily multivitamin use were greater in older
covariate. Models also were fitted to estimate the effect of the men, although no test of trend attained statistical significance for
intervention on 1 end point before and after diagnosis of the second cataract diagnosis or extraction (Table 2). For both end points, a
end point. beneficial effect of multivitamins began to emerge midway

528
 Christen et al 
Multivitamins and Cataract and AMD

For both cataract and AMD, HR estimates for multivitamin

treatment were not materially altered in analyses that accounted for
diagnosis of the second end point. For example, the HR for cataract
changed little (0.90; 95% CI, 0.82e0.99; P ¼ 0.03) after adjust-
ment for a diagnosis of visually significant AMD as a time-varying
covariate. Likewise, the HR for visually significant AMD changed
little (1.18; 95% CI, 0.94e1.50; P ¼ 0.16) after adjustment for
diagnosed cataract. We also compared HRs before and after
diagnosis of the second end point. The HR for cataract was 0.92
(95% CI, 0.84e1.01) with no prior diagnosis of visually significant
AMD and 0.69 (95% CI, 0.41e1.17) after a diagnosis of visually
significant AMD. For visually significant AMD, the HR was 1.31
(95% CI, 0.92e1.87) with no prior diagnosis of cataract and 1.08
(95% CI, 0.79e1.48) after a diagnosis of cataract.

Discussion
In this large-scale randomized trial of middle-aged and older
men, long-term daily multivitamin use modestly and
significantly reduced the co-primary vision end point of
cataract after more than 10 years of treatment and follow-up.
There was no significant benefit or risk of daily multivitamin
use on visually significant AMD, the second co-primary
vision end point, although the HRs tended to be modestly
elevated.
Our findings for cataract are consistent with results of 2
prior trials of multivitamin use in cataract prevention. In the
Linxian Cataract Study, conducted in a nutritionally deficient
population in China, persons aged 65 to 74 years randomized
Figure 2. Cumulative incidence rates of cataract in the multivitamin and to a daily supplement of 14 vitamins and 12 minerals at 2 to 3
placebo groups in the Physicians’ Health Study II. Y ¼ year. times the US Recommended Dietary Allowance (RDA),
compared with placebo, had a significant 36% lower preva-
lence of nuclear cataract after 6 years of treatment.22 There
through follow-up and persisted throughout the remainder of the was no difference in the prevalence of cortical or PSC
trial (crude log-rank P ¼ 0.05 for both end points) (Fig 2).
cataract in those aged 65 to 74 years, nor was there any
The effect of multivitamins on cataract did not differ markedly
within categories of baseline characteristics (Table 3). However, difference in the prevalence of any cataract type in those
the effect did appear to vary according to vitamin C treatment aged 54 to 64 years. In the Italian American Clinical Trial
assignment (P interaction ¼ 0.04), with a significant benefit of Nutritional Supplements and Age-related Cataract, per-
observed only among men in the vitamin C placebo group. sons aged 55 to 75 years randomized to a daily multivitamin
(Centrum), compared with placebo, had a significant 18%
Age-Related Macular Degeneration reduction in cataract development or progression after 9 years
of treatment and follow-up.23 Analyses of subtypes indicated
Men in the multivitamin group had a 19% increased risk of visually a significant 34% reduction in nuclear cataract (HR, 0.66;
significant AMD that was not statistically significant (152 vs. 129 95% CI, 0.50e0.88), a nonsignificant 22% reduction in
cases; HR, 1.19; 95% CI, 0.94e1.50; P ¼ 0.15) (Table 4). There
cortical cataract (HR, 0.78; 95% CI, 0.60e1.02), and a
was also a significant 22% increased risk of total AMD (with or
without vision loss) (294 vs. 244 cases; HR, 1.22; 95% CI, significant 2-fold increased risk of PSC cataract (HR, 2.00;
1.03e1.44; P ¼ 0.02) and a nonsignificant 22% increased risk of 95% CI, 1.35e2.98) in the multivitamin group. Neither the
advanced AMD (79 vs. 65 cases; HR, 1.22; 95% CI, 0.88e1.70; Linxian nor the Clinical Trial of Nutritional Supplements and
P ¼ 0.23) in the multivitamin group. Age-related Cataract studies examined the effect of the
Hazard ratios for all AMD end points tended to be highest in the intervention on AMD.
oldest age group, although no test of trend attained statistical sig- Taken together, our findings in PHS II and 2 prior trials
nificance. For the primary end point of visually significant AMD, the indicate that long-term daily multivitamin use may have a
curves appeared to diverge midway through follow-up but never small to moderate beneficial effect on risk of cataract,
attained statistical significance (crude log-rank P ¼ 0.18) (Fig 3). particularly nuclear cataract. Given that an estimated 10
Curves for total AMD began to diverge earlier in the trial and
million adults in the United States have impaired vision due
persisted throughout the trial (log-rank P ¼ 0.03). For advanced
AMD, there was no apparent effect of a daily multivitamin at any to cataract,38,39 even a modest reduction in risk of cataract
point during the trial (crude log-rank P ¼ 0.25). would have a large public health impact. Although the main
The effect of multivitamins on visually significant AMD did not trial results in AREDS indicated no benefit on lens opacity
differ appreciably within categories of baseline characteristics progression for daily treatment with high-dose vitamin E
(Table 3). (400 IU), vitamin C (500 mg), and beta-carotene (15 mg),

529
 Ophthalmology Volume 121, Number 2, February 2014

Table 3. Relative Rates of Cataract and Visually Significant Age-Related Macular Degeneration by Randomized Treatment Assignment
Within Subgroups in Physicians’ Health Study II*

Cataract Age-Related Macular Degeneration

No. of Cataract/Total No. of AMD/Total
Variables Active Placebo HR (95% CI)y P Interactionz Active Placebo HR (95% CI)y P Interactionz
Cigarette smoking
Never 477/3437 504/3387 0.93 (0.82e1.05) 0.45 64/4055 57/4017 1.08 (0.76e1.55) 0.75
Former 366/2100 401/2164 0.91 (0.79e1.05) 82/2799 67/2843 1.30 (0.94e1.79)
Current 29/192 40/208 0.70 (0.43e1.13) 6/249 5/258 1.16 (0.35e3.83)
Alcohol use
Rarely/never 180/1079 164/1029 0.99 (0.80e1.23) 0.43 37/1356 20/1294 1.62 (0.94e2.80) 0.17

1 drink/mo 687/4618 776/4699 0.89 (0.80e0.99) 115/5705 107/5787 1.11 (0.85e1.44)
Body mass index (kg/m2)
<25 374/2369 411/2348 0.89 (0.77e1.02) 0.71 66/2976 59/2953 1.11 (0.78e1.58) 0.87
25e<30 399/2744 443/2807 0.90 (0.79e1.03) 73/3396 60/3421 1.24 (0.88e1.75)

30 99/623 91/605 1.02 (0.77e1.35) 13/739 10/745 1.27 (0.56e2.91)
History of hypertension
Yes 399/2167 434/2259 0.94 (0.82e1.07) 0.63 75/2899 58/2997 1.33 (0.94e1.87) 0.35
No 470/3525 506/3473 0.89 (0.79e1.01) 77/4163 71/4095 1.07 (0.78e1.48)
History of high cholesterol
Yes 349/1958 361/2021 0.98 (0.84e1.13) 0.19 51/2485 45/2563 1.17 (0.78e1.75) 0.99
No 508/3560 572/3516 0.86 (0.76e0.97) 98/4395 83/4311 1.16 (0.87e1.56)
History of diabetes
Yes 56/273 55/274 0.98 (0.68e1.43) 0.41 8/446 9/407 0.85 (0.33e2.22) 0.56
No 816/5458 890/5482 0.90 (0.82e0.99) 144/6658 120/6707 1.21 (0.95e1.54)
Exercise
1 time/wk
Yes 503/3544 543/3479 0.91 (0.80e1.02) 0.98 83/4340 85/4225 0.97 (0.72e1.31) 0.055
No 361/2055 394/2139 0.91 (0.79e1.05) 67/2604 44/2715 1.56 (1.06e2.28)
Self-reported history of CVD
Yes 40/212 45/242 0.98 (0.64e1.50) 0.91 10/358 10/359 0.93 (0.39e2.24) 0.59
No 832/5524 900/5519 0.91 (0.82e1.00) 142/6753 119/6763 1.21 (0.95e1.54)
Randomized to vitamin C
Yes 455/2889 452/2879 1.00 (0.88e1.14) 0.04 75/3574 65/3575 1.15 (0.82e1.60) 0.79
No 417/2847 493/2882 0.82 (0.72e0.94) 77/3537 64/3547 1.22 (0.88e1.71)
Randomized to vitamin E
Yes 437/2869 479/2871 0.89 (0.78e1.01) 0.67 75/3537 69/3574 1.12 (0.81e1.55) 0.61
No 435/2867 466/2890 0.93 (0.81e1.06) 77/3574 60/3548 1.26 (0.90e1.77)
Randomized to beta-carotene
Yes 421/2885 439/2872 0.94 (0.82e1.07) 0.53 71/3591 60/3580 1.20 (0.85e1.69) 0.92
No 451/2851 506/2889 0.88 (0.78e1.00) 81/3520 69/3542 1.17 (0.85e1.62)
Aspirin use
Yes 708/4403 759/4339 0.89 (0.80e0.99) 0.50 121/5460 100/5408 1.22 (0.93e1.58) 0.61
No 162/1265 183/1324 0.97 (0.78e1.20) 30/1564 29/1594 1.03 (0.62e1.72)

AMD ¼ age-related macular degeneration; CI ¼ confidence interval; CVD ¼ cardiovascular disease; HR ¼ relative risk.
*Baseline factors are defined as in Table 1.
y
Adjusted for age; Physicians’ Health Study cohort; and vitamin C, vitamin E, and beta-carotene treatment assignment.
z
Test of the null hypothesis of no difference in treatment effect across risk factor subgroups.

a propensity score analysis showed that self-selection for levels (zinc [15 mg], vitamin E [45 IU], vitamin C
Centrum use (provided by AREDS) by approximately two [60 mg], beta-carotene [5000 IU vitamin A, 20% as beta-
thirds of AREDS participants was associated with a signif- carotene]) rather than the high-dose formulation tested in
icant reduction in lens opacity progression, particularly for AREDS. In addition, AREDS tested a higher-risk popula-
nuclear opacities.40 tion than PHS II (no reported diagnosis of AMD at base-
PHS II is the first large-scale randomized trial to test a line), and their primary end point was advanced AMD,
multivitamin supplement in AMD prevention. Our finding whereas in PHS II, the primary end point was visually
of no significant benefit appears to contrast with the benefits significant AMD, most commonly characterized by some
for other vitamin/mineral combinations in prior trials of combination of drusen and RPE changes. Thus, our end
AMD. For example, AREDS demonstrated that daily zinc point represented an earlier stage of disease development
(80 mg) and a high-dose antioxidant combination of vitamin than the advanced AMD end point in AREDS. It is worth
E (400 IU), vitamin C (500 mg), and beta-carotene (15 mg) noting that AREDS reported no benefit of the zinc and
significantly slowed the progression of AMD.24 The antioxidant combination in participants with early-stage
multivitamin in PHS II included these nutrients at RDA AMD at baseline.24

530
 Christen et al 
Multivitamins and Cataract and AMD

Table 4. Confirmed Cases of Age-Related Macular Degeneration According to Multivitamin Randomized Treatment Assignment in
Physicians’ Health Study II

AMD by Age Active (n [ 7111) Placebo (n [ 7122) HR* 95% CI P Value P Trendy
Visually significant AMD
50e59 yrs 6 8 0.76 0.26e2.18 0.61 0.43
60e69 yrs 41 36 1.14 0.73e1.78 0.58

70 yrs 105 85 1.24 0.93e1.65 0.14
Total 152 129 1.19 0.94e1.50 0.15
Total AMD
50e59 yrs 28 29 0.98 0.58e1.64 0.93 0.13
60e69 yrs 97 90 1.07 0.80e1.43 0.64

70 yrs 169 125 1.38 1.09e1.73 0.007
Total 294 244 1.22 1.03e1.44 0.02
Advanced AMD
50e59 yrs 3 7 0.43 0.11e1.67 0.22 0.14
60e69 yrs 21 18 1.15 0.61e2.17 0.66

70 yrs 55 40 1.38 0.92e2.08 0.12
Total 79 65 1.22 0.88e1.70 0.23

AMD ¼ age-related macular degeneration; CI ¼ confidence interval; HR ¼ hazard ratio.

*Adjusted for age; Physicians’ Health Study cohort; and vitamin C, vitamin E, and beta-carotene treatment assignment.
y
Test for trend of the effect of age on the association between randomized treatment assignment and AMD.

Our findings also appear to contrast with the findings in the vitamin B6, and vitamin B12 for 7.3 years reduced the risk of a
Women’s Antioxidant and Folic Acid Cardiovascular Study diagnosis of AMD by 35% to 40%.25 Both WAFACS and
(WAFACS), where combined treatment with folic acid, PHS II used the same method of case ascertainment and the
same diagnostic criteria for AMD but differed in other
important respects. The WAFACS tested pharmacologic
doses of folic acid (2.5 mg/d), vitamin B6 (50 mg/d), and
vitamin B12 (1 mg/d), whereas the multivitamin in PHS II
contained RDA levels of these nutrients (folic acid [400 mg],
vitamin B6 [3 mg], vitamin B12 [25 mg]). In addition,
WAFACS was composed of women with preexisting CVD
or 3 or more CVD risk factors, whereas PHS II was
composed primarily of apparently healthy men; therefore, it
seems possible that our finding of no benefit (and possible
harm) may reflect, at least in part, the lower dosage of the
PHS II multivitamin plus the generally lower risk profile of
the PHS II population.

Study Limitations
Several possible limitations of our study need to be
considered. At the initiation of PHS II in 1997, we selected a
commonly used multivitamin, Centrum Silver, to increase
the potential generalizability of study findings. The same
formulation was used throughout PHS II. However, after the
inception of PHS II, the doses of several nutrients in
Centrum Silver were changed and several nutrients were
added, including lutein (250 mg). Lutein has been shown to
be of possible benefit against cataract and AMD in obser-
vational studies41e49 and several small randomized tri-
als,50e55 although the recently completed AREDS 2
reported no overall benefit on either end point.56,57 How-
ever, the effect of adding lutein to the Centrum Silver
formulation could not be addressed in our study. Compli-
ance with study medication is a concern in any long-term
Figure 3. Cumulative incidence rates of age-related macular degeneration randomized trial; however, compliance with the daily
(AMD) in the multivitamin and placebo groups in the Physicians’ Health multivitamin remained high throughout follow-up. Finally,
Study II. Y ¼ year. the PHS II population is composed of generally well-

531
 Ophthalmology Volume 121, Number 2, February 2014

nourished male physicians, and our findings may not apply 12. Tso MO. Experiments on visual cells by nature and man:
to women or to less well-nourished populations. in search of treatment for photoreceptor degeneration. Frie-
Several aspects of our methodology also deserve consid- denwald lecture. Invest Ophthalmol Vis Sci 1989;30:2430–54.
eration. Case identification was based on participant reports, 13. Teikari JM, Rautalahti M, Haukka J, et al. Incidence of cata-
and thus some underascertainment of cataract and AMD is ract operations in Finnish male smokers unaffected by alpha
tocopherol or beta carotene supplements. J Epidemiol Com-
plausible. Such underascertainment would likely reduce study munity Health 1998;52:468–72.
power but is not associated with bias in randomized com- 14. Age-Related Eye Disease Study Research Group. A randomized,
parisons. Random misclassification was reduced by the use placebo-controlled, clinical trial of high-dose supplementation
of medical records to confirm the participant reports. Non- with vitamins C and E and beta carotene for age-related cataract
random misclassification was unlikely because medical re- and vision loss: AREDS report no. 9. Arch Ophthalmol
cords were reviewed by an investigator (W.G.C.) masked to 2001;119:1439–52.
treatment assignment, and study participants and treating 15. REACT Group, Chylack LT Jr, Brown NP, Bron A, et al. The
ophthalmologists and optometrists were similarly unaware of Roche European American Cataract Trial (REACT): a ran-
treatment assignment. Finally, the equal distribution of domized clinical trial to investigate the efficacy of an oral
baseline characteristics between the multivitamin and placebo antioxidant micronutrient mixture to slow progression of age-
related cataract. Ophthalmic Epidemiol 2002;9:49–80.
groups indicates that confounding by measured factors is 16. McNeil JJ, Robman L, Tikellis G, et al. Vitamin E supple-
unlikely and that other potential confounders, which were mentation and cataract: randomized controlled trial. Ophthal-
unmeasured or unknown, also were likely to be evenly mology 2004;111:75–84.
distributed between the 2 treatment groups. 17. Christen WG, Glynn RJ, Chew EY, Buring JE. Vitamin E and
In conclusion, the finding in this large-scale randomized age-related cataract in a randomized trial of women.
trial of middle-aged and older men that long-term daily Ophthalmology 2008;115:822–9.
multivitamin use is associated with a modest but significant 18. Christen WG, Glynn RJ, Sesso HD, et al. Age-related cataract
reduction in cataract, and in particular nuclear cataract, is in a randomized trial of vitamins E and C in men. Arch
consistent with results of previous trials of multivitamin use Ophthalmol 2010;128:1397–405.
in cataract prevention. The finding of no significant benefit 19. Christen W, Glynn R, Sperduto R, et al. Age-related cataract in
a randomized trial of beta-carotene in women. Ophthalmic
or harm for multivitamin use in the prevention of visually Epidemiol 2004;11:401–12.
significant AMD needs to be confirmed in other populations 20. Christen WG, Manson JE, Glynn RJ, et al. A randomized trial
of men and women. of beta carotene and age-related cataract in US physicians.
Arch Ophthalmol 2003;121:372–8.
References 21. Gritz DC, Srinivasan M, Smith SD, et al. The Antioxidants in
Prevention of Cataracts Study: effects of antioxidant supple-
ments on cataract progression in South India. Br J Ophthalmol
1. Congdon NG, West KP Jr. Nutrition and the eye. Curr Opin 2006;90:847–51.
Ophthalmol 1999;10:464–73. 22. Sperduto RD, Hu TS, Milton RC, et al. The Linxian cataract
2. Chiu CJ, Taylor A. Nutritional antioxidants and age-related studies. Two nutrition intervention trials. Arch Ophthalmol
cataract and maculopathy. Exp Eye Res 2007;84:229–45. 1993;111:1246–53.
3. Agte V, Tarwadi K. The importance of nutrition in the pre- 23. Clinical Trial of Nutritional Supplements and Age-Related
vention of ocular disease with special reference to cataract. Cataract Study Group. A randomized, double-masked,
Ophthalmic Res 2010;44:166–72. placebo-controlled clinical trial of multivitamin supplementa-
4. Fletcher AE. Free radicals, antioxidants and eye diseases: evi- tion for age-related lens opacities: Clinical Trial of Nutritional
dence from epidemiological studies on cataract and age-related Supplements and Age-Related Cataract report no. 3.
macular degeneration. Ophthalmic Res 2010;44:191–8. Ophthalmology 2008;115:599–607.
5. Krishnadev N, Meleth AD, Chew EY. Nutritional supplements 24. Age-Related Eye Disease Study Research Group.
for age-related macular degeneration. Curr Opin Ophthalmol A randomized, placebo-controlled, clinical trial of high-dose
2010;21:184–9. supplementation with vitamins C and E, beta carotene, and
6. Seddon JM. Multivitamin-multimineral supplements and eye zinc for age-related macular degeneration and vision loss:
disease: age-related macular degeneration and cataract. Am J AREDS report no. 8. Arch Ophthalmol 2001;119:1417–36.
Clin Nutr 2007;85:304S–7S. 25. Christen WG, Glynn RJ, Chew EY, et al. Folic acid, pyridoxine,
7. Cui YH, Jing CX, Pan HW. Association of blood antioxi- and cyanocobalamin combination treatment and age-related
dants and vitamins with risk of age-related cataract: a meta- macular degeneration in women: the Women’s Antioxidant
analysis of observational studies. Am J Clin Nutr 2013;98: and Folic Acid Cardiovascular Study. Arch Intern Med
778–86. 2009;169:335–41.
8. Gerster H. Antioxidant vitamins in cataract prevention. 26. Teikari JM, Laatikainen L, Virtamo J, et al. Six-year sup-
Z Ernahrungswiss 1989;28:56–75. plementation with alpha-tocopherol and beta-carotene and
9. Taylor A. Role of nutrients in delaying cataracts. Ann N Y age-related maculopathy. Acta Ophthalmol Scand 1998;76:
Acad Sci 1992;669:111–24. 224–9.
10. Organisciak DT, Wang HM, Li ZY, Tso MO. The protective 27. Taylor HR, Tikellis G, Robman LD, et al. Vitamin E supple-
effect of ascorbate in retinal light damage of rats. Invest mentation and macular degeneration: randomised controlled
Ophthalmol Vis Sci 1985;26:1580–8. trial. BMJ 2002;325:11.
11. Ham WT Jr, Mueller HA, Ruffolo JJ Jr, et al. Basic mecha- 28. Christen WG, Glynn RJ, Chew EY, Buring JE. Vitamin E and
nisms underlying the production of photochemical lesions in age-related macular degeneration in a randomized trial of
the mammalian retina. Curr Eye Res 1984;3:165–74. women. Ophthalmology 2010;117:1163–8.

532
 Christen et al 
Multivitamins and Cataract and AMD

29. Christen WG, Manson JE, Glynn RJ, et al. Beta carotene 44. Brown L, Rimm EB, Seddon JM, et al. A prospective study of
supplementation and age-related maculopathy in a random- carotenoid intake and risk of cataract extraction in US men.
ized trial of US physicians. Arch Ophthalmol 2007;125: Am J Clin Nutr 1999;70:517–24.
333–9. 45. Vu HT, Robman L, Hodge A, et al. Lutein and zeaxanthin and
30. Christen WG, Glynn RJ, Sesso HD, et al. Vitamins E and C the risk of cataract: the Melbourne visual impairment project.
and medical record-confirmed age-related macular degenera- Invest Ophthalmol Vis Sci 2006;47:3783–6.
tion in a randomized trial of male physicians. Ophthalmology 46. Lyle BJ, Mares-Perlman JA, Klein BE, et al. Antioxidant
2012;119:1642–9. intake and risk of incident age-related nuclear cataracts in the
31. Christen WG, Gaziano JM, Hennekens CH. Design of Physi- Beaver Dam Eye Study. Am J Epidemiol 1999;149:801–9.
cians’ Health Study IIea randomized trial of beta-carotene, 47. Christen WG, Liu S, Glynn RJ, et al. Dietary carotenoids,
vitamins E and C, and multivitamins, in prevention of can- vitamins C and E, and risk of cataract in women: a prospective
cer, cardiovascular disease, and eye disease, and review of study. Arch Ophthalmol 2008;126:102–9.
results of completed trials. Ann Epidemiol 2000;10:125–34. 48. Bone RA, Landrum JT, Mayne ST, et al. Macular pigment in
32. Gaziano JM, Glynn RJ, Christen WG, et al. Vitamins E and C donor eyes with and without AMD: a case-control study.
in the prevention of prostate and total cancer in men: the Invest Ophthalmol Vis Sci 2001;42:235–40.
Physicians’ Health Study II randomized controlled trial. 49. Moeller SM, Parekh N, Tinker L, et al; CAREDS Research
JAMA 2009;301:52–62. Study Group. Associations between intermediate age-related
33. Sesso HD, Buring JE, Christen WG, et al. Vitamins E and C in macular degeneration and lutein and zeaxanthin in the Carot-
the prevention of cardiovascular disease in men: the Physi- enoids in Age-related Eye Disease Study (CAREDS): ancillary
cians’ Health Study II randomized controlled trial. JAMA study of the Women’s Health Initiative. Arch Ophthalmol
2008;300:2123–33. 2006;124:1151–62.
34. Hennekens CH, Buring JE, Manson JE, et al. Lack of effect of 50. Olmedilla B, Granado F, Blanco I, Vaquero M. Lutein, but not
long-term supplementation with beta carotene on the incidence alpha-tocopherol, supplementation improves visual function
of malignant neoplasms and cardiovascular disease. N Engl J in patients with age-related cataracts: a 2-y double-blind,
Med 1996;334:1145–9. placebo-controlled pilot study. Nutrition 2003;19:21–4.
35. Steering Committee of the Physicians’ Health Study Research 51. Arnold C, Winter L, Frohlich K, et al. Macular xanthophylls
Group. Final report on the aspirin component of the ongoing and omega-3 long-chain polyunsaturated fatty acids in age-
Physicians’ Health Study. N Engl J Med 1989;321:129–35. related macular degeneration: a randomized trial. JAMA
36. Ederer F. Shall we count numbers of eyes or numbers of Ophthalmol 2013;131:564–72.
subjects? Arch Ophthalmol 1973;89:1–2. 52. Piermarocchi S, Saviano S, Parisi V, et al; CARMIS Study
37. Glynn RJ, Rosner B. Accounting for the correlation between Group. Carotenoids in Age-related Maculopathy Italian Study
fellow eyes in regression analysis. Arch Ophthalmol 1992;110: (CARMIS): two-year results of a randomized study. Eur J
381–7. Ophthalmol 2012;22:216–25.
38. Eye Diseases Prevalence Research Group. Prevalence of 53. Ma L, Yan SF, Huang YM, et al. Effect of lutein and zeaxanthin
cataract and pseudophakia/aphakia among adults in the United on macular pigment and visual function in patients with early
States. Arch Ophthalmol 2004;122:487–94. age-related macular degeneration. Ophthalmology 2012;119:
39. National Eye Institute. National Plan for Eye and Vision 2290–7.
Research: Lens and Cataract Program. Available at: http:// 54. Beatty S, Chakravarthy U, Nolan JM, et al. Secondary out-
www.nei.nih.gov/strategicplanning/np_lens.asp. Accessed comes in a clinical trial of carotenoids with coantioxidants
September 15, 2013. versus placebo in early age-related macular degeneration.
40. Age-Related Eye Disease Study Research Group. Centrum use Ophthalmology 2013;120:600–6.
and progression of age-related cataract in the Age-Related Eye 55. Berrow EJ, Bartlett HE, Eperjesi F, Gibson JM. The effects of a
Disease Study: a propensity score approach. AREDS report lutein-based supplement on objective and subjective measures
no. 21. Ophthalmology 2006;113:1264–70. of retinal and visual function in eyes with age-related
41. Chasan-Taber L, Willett WC, Seddon JM, et al. A prospective maculopathyea randomised controlled trial. Br J Nutr
study of carotenoid and vitamin A intakes and risk of cataract 2013;109:2008–14.
extraction in US women. Am J Clin Nutr 1999;70:509–16. 56. Age-Related Eye Disease Study 2 (AREDS2) Research
42. Moeller SM, Voland R, Tinker L, et al; CAREDS Study Group. Group, Chew EY, SanGiovanni JP, Ferris FL, et al. Lutein/
Associations between age-related nuclear cataract and lutein and zeaxanthin for the treatment of age-related cataract: AREDS2
zeaxanthin in the diet and serum in the Carotenoids in the Age- randomized trial report no. 4. JAMA Ophthalmol 2013;131:
Related Eye Disease Study, an ancillary study of the Women’s 843–50.
Health Initiative. Arch Ophthalmol 2008;126:354–64. 57. Age-Related Eye Disease Study 2 Research Group. Lutein þ
43. Mares-Perlman JA, Brady WE, Klein BE, et al. Serum carot- zeaxanthin and omega-3 fatty acids for age-related macular
enoids and tocopherols and severity of nuclear and cortical degeneration: the Age-Related Eye Disease Study 2 (AREDS2)
opacities. Invest Ophthalmol Vis Sci 1995;36:276–88. randomized clinical trial. JAMA 2013;309:2005–15.

Footnotes and Financial Disclosures

2
Originally received: July 18, 2013. Division of Aging, Department of Medicine, Brigham and Women’s
Final revision: September 20, 2013. Hospital and Harvard Medical School, Boston, Massachusetts.
Accepted: September 24, 2013. 3
Division of Pharmacoepidemiology and Pharmacoeconomics, Department
Available online: November 21, 2013. Manuscript no. 2013-1171.
of Medicine, Brigham and Women’s Hospital and Harvard Medical School,
1
Division of Preventive Medicine, Department of Medicine, Brigham and Boston, Massachusetts.
Women’s Hospital and Harvard Medical School, Boston, Massachusetts.

533
 Ophthalmology Volume 121, Number 2, February 2014
4
Division of Cardiovascular Disease, Department of Medicine, Brigham initiated research funding from the National Institutes of Health, the To-
and Women’s Hospital and Harvard Medical School, Boston, mato Products Wellness Council, and Cambridge Theranostics, Ltd. J.M.G.
Massachusetts. received investigator-initiated research funding from the National Institutes
5
VA Boston Healthcare System, Boston, Massachusetts. of Health, the Veterans Administration, and the BASF Corporation; assis-
6
Department of Population Medicine, Harvard Medical School, Boston, tance with study agents and packaging from BASF Corporation and Pfizer
Massachusetts. Inc. (formerly Wyeth, American Home Products, and Lederle); and assis-
7 tance with study packaging provided by D.S.M. Nutritional Products Inc.
Department of Epidemiology, Harvard School of Public Health, Boston, (formerly Roche Vitamins). No other authors reported financial disclosures.
Massachusetts.
8 Supported by Grants CA 097193 (which included funding from the
Department of Biostatistics, Harvard School of Public Health, Boston,
National Eye Institute and the National Institute on Aging), CA 34944, CA
Massachusetts. 40360, HL 26490, and HL 34595 from the National Institutes of Health
Financial Disclosure(s): (Bethesda, MD), and an investigator-initiated grant from BASF Corporation
The author(s) have made the following disclosure(s): W.G.C. received (Florham Park, NJ). Study agents and packaging were provided by BASF
research funding support from the National Institutes of Health and DSM Corporation and Pfizer Inc. (formerly Wyeth, American Home Products,
Nutritional Products Inc. (formerly Roche Vitamins). R.J.G. received and Lederle) (New York, NY), and study packaging was provided by DSM
investigator-initiated research funding from the National Institutes of Nutritional Products, Inc. (formerly Roche Vitamins) (Parsippany, NJ). The
Health, Bristol-Meyers Squibb, AstraZeneca, and Novartis, and signed a National Institutes of Health, BASF, Pfizer Inc., and DSM Nutritional
consulting agreement with Merck to give an invited talk. J.E.M. received Products Inc., had no role in the study design; conduct of the study;
investigator-initiated research funding from the National Institutes of Health collection, management, analysis, and interpretation of the data; or prepa-
and assistance with study pills and packaging from BASF and Cognis ration, review, or approval of the manuscript.
Corporations for the WAFACS and from Pronova BioPharma and Phar- Trial registration: clinicaltrials.gov identifier: NCT00270647
mavite for the VITamin D and OmegA-3 TriaL, and funding from the
nonprofit Aurora Foundation. J.E.B. received investigator-initiated research Correspondence:
funding from the National Institutes of Health and assistance with study William G. Christen, ScD, 900 Commonwealth Ave. East, Boston, MA
pills and packaging from Natural Source Vitamin E Association and Bayer 02215-1204. E-mail: wchristen@rics.bwh.harvard.edu.
Healthcare for the Women’s Health Study. H.D.S. received investigator-

534