The Women’s Antioxidant Cardiovascular Study (WACS) began in 1995 – 1996. WACS was a 2×2×2 randomized placebo-controlled trial of 3 antioxidants: 402 mg (600 IU) of vitamin E every other day, 500 mg of vitamin C daily, and 50 mg of β-carotene every other day for the secondary prevention of cardiovascular disease (CVD). Eligible participants were female health professionals, 40+ years, with at least three coronary risk factors or prevalent cardiovascular disease (CVD). The women were 94.0% Caucasian, 3.3% African-American, 0.9% Latino-American, 0.7% Asian-American and 1.1% of other / multiple race. Coronary risk factors included parental history of premature MI, diabetes, hypertension, high cholesterol, obesity (BMI ≥ 30 kg/m2). CVD included myocardial infarction, stroke, revascularization procedures (percutaneous transluminal angioplasty, coronary artery bypass graft, carotid endarterectomy, or peripheral artery surgery), and symptomatic angina pectoris or transient cerebral ischemia. In a three-month run-in phase to assess compliance, women received placebo caplets. Women (n= 8,171) who reported good compliance, had no history of cancer in the past 10 years, active liver disease, chronic kidney failure, or use of anticoagulants, and who expressed willingness to forgo the use of out-of-study vitamin supplements beyond the recommended daily allowance were randomized.
Every year during follow-up, the women were sent a 12-months’ supply of calendar packs containing active agents or placebo. Women completed annual mailed questionnaires on compliance, side effects, health and lifestyle characteristics and clinical endpoints. Participants were followed through the scheduled end (January 31, 2005).8
When assessed on annual questionnaires, participants’ compliance to assigned study agents was high and comparable between the active and placebo groups: average compliance (defined as taking at least two-thirds of assigned study medications) during follow-up was 83% and did not differ significantly between the two groups.8
Participants provided written informed consent; the trial was approved by the institutional review board of Brigham and Women’s Hospital, Boston and was monitored by an external data and safety monitoring board.
The results of the primary trial have been published;8
briefly, antioxidant supplementation did not protect against cardiovascular disease, and it did not cause any major adverse side effects.8
After a mean 3.5 years after randomization, from December 1998-July 2000, we initiated a substudy of cognitive function. The substudy was focused on the oldest women: all active participants aged 65 years or older (n=3170). Of these women, we could not contact 190 by telephone; of the 2980 women we contacted, 156 (5%) declined participation and 2824 (95%) completed the initial telephone cognitive assessment (). Participation in our initial cognitive interview was virtually identical across all the treatment and placebo groups (range 94–95%).
Flow chart of participation in the Cognitive Cohort of Women's Antioxidant Cardiovascular Study (WACS)
Participants received three follow-up cognitive assessments approximately every two years. High follow-up was maintained across the groups (): 93 % completed at least one follow-up assessment, and 81 % completed at least 3. In the fourth assessment, 24% of participants were not contacted for their interview, as only a short interval had passed between their third interview and the end of the trial in January 2005. Follow-up rates were nearly identical across treatment groups at each assessment.
Cognitive Function Assessment
Substantial research implicates vascular factors in cognitive health—including cognitive outcomes not traditionally associated with vascular health, such as general cognition, episodic memory, and Alzheimer dementia (AD).1
Thus, the emphasis of this study was not on executive function measures, but on general cognition. We hypothesized that if cardiovascular disease and cognitive decline share similar pathways of development, then antioxidants that may protect against the development of cardiovascular disease may also confer benefits for the maintenance of general cognitive function.
We assessed cognitive function by telephone and administered 5 tests measuring general cognition, verbal memory and category fluency. For general cognition, we used the Telephone Interview of Cognitive Status (TICS)9
a telephone adaptation of the Mini-Mental State Examination (MMSE). For verbal memory, we administered the delayed recall of the TICS 10-word list, and the immediate and delayed recalls of the East Boston Memory Test,10
in which a short paragraph is read and 12 key elements are repeated immediately and 15 minutes later. Finally, in a test of category fluency (used to measure executive retrieval functions),11
women were asked to name as many animals as possible in one minute.
The primary, pre-specified outcome of this trial was the change from baseline of the global composite score, which is an average of all five cognitive tests made into z-scores. In addition, because verbal memory is strongly associated with risk of Alzheimer disease,12
our key secondary outcome was the change from baseline of the verbal memory composite score; this composite score was calculated by averaging scores across four measures of verbal memory (the immediate and delayed recalls of both the East Boston Memory Test and 10 word list). To calculate the composite scores for participants who did not complete all tests (only 0.5% for both the global composite score and the verbal memory score), we used the mean of the z-scores of the tests that were completed.
The telephone cognitive interviews were administered by trained interviewers, who were masked to the participants’ randomized treatment assignment. There was high reliability and validity of our telephone cognitive test battery. In a test-retest reliability study of the TICS, administered twice 31 days apart, we found a correlation of 0.7 (p<0.001) among 35 high-functioning, educated women. In a validation study of our telephone instrument, 61 women who had completed an extensive in-person interview were administered our brief telephone-administered assessment; we found a correlation of 0.81 comparing the global composite scores on those two measures, demonstrating high validity of our telephone method. Importantly, among 88 older female health professionals, cognitive impairment as determined by our telephone assessment was strongly associated with dementia diagnosis after three years; poor performance in the TICS and in verbal memory were both associated with significant 8 and 12 fold increases, respectively, of dementia.
Characteristics at baseline between randomized groups were compared using Wilcoxon rank sum tests and chi-square tests for proportions. Mean performance at each assessment by treatment assignment was evaluated using repeated measures analysis of means, which takes into account correlations between assessments. The mean for each intervention group at each time was estimated, allowing for an interaction of group and time, and modeling the correlation of measures over time with an unstructured covariance matrix. Such general linear models of response profiles address the non-linearity of scores and impose minimal structure on outcome trends over time.13
Second, the primary analytic outcome was the mean difference in cognitive change from the initial to the second through fourth assessments. The mean difference in change was basically calculated by subtracting the baseline score from follow-up scores and then taking the difference of cognitive change between the treatment and placebo groups. Thus, a negative value for mean difference in cognitive change indicates an adverse effect of treatment. The mean differences in cognitive change were evaluated by treatment assignment in a repeated measures model. This included fixed effects for time and a common intervention effect over time for each group, reflecting the average difference between groups over time. All models were fitted by maximum likelihood, incorporating the longitudinal correlation within study subjects using unstructured covariance structures; for statistical testing, we used Wald tests.13
For statistical analyses, Proc Mixed in SAS (SAS release 9.1, SAS Institute Inc., Cary, NC) was used.
We also evaluated the differences in cognitive change between those assigned to any of the 3 antioxidants compared with those assigned to all placebos. We further evaluated taking various combinations of antioxidants (e.g. vitamin E and vitamin C versus placebos for both).
We examined effect modification by key risk factors for cognitive change at randomization as well as by incident cardiovascular disease during the trial. We also selected factors that may affect the metabolism of antioxidants (e.g., smoking). Tests of effect modification were performed by evaluating interaction terms in models of mean change in cognition.
In secondary analyses, we examined the influence of non-compliance by repeating the main analyses after excluding women who were taking less than two-thirds of their assigned study medications.
We also constructed models adjusting for assignment to other antioxidant agents or B vitamins, but results were essentially unchanged (data not shown); thus we did not include assignment to other supplements as covariates in models. Also, effect modification by assignment to other trial agents was not observed.
Finally, to assess the impact of antioxidant supplementation on the risk of substantial cognitive change, we fitted logistic regression models adjusting for follow-up time between the first and last assessments, defining the outcome as those in the worst 10% of the distribution of cognitive change from the initial to the final cognitive assessment.