Our analyses utilized baseline data from Women's Health Initiative (WHI) hormone trials cohort. The WHI is a large, multifaceted study focused on common causes of morbidity and mortality among postmenopausal women aged 50–79 years. The WHI Estrogen plus Progestin (E+P) and Estrogen Alone (E-alone) randomized, placebo-controlled trials examined the risks and benefits of hormone therapies in women without and with hysterectomy, respectively. Detailed descriptions of eligibility criteria, recruitment methods, and study design have been published previously.23–25
Briefly, 40 clinical centers throughout the United States enrolled participants between 1993–1998. Eligible women were postmenopausal, had no medical condition associated with a predicted survival of less than 3 years, and were likely to be residing in the same geographic area for at least 3 years. Additional exclusion criteria included attention to safety issues and adherence and retention concerns. Women who were taking hormones, but who were interested in participating in the hormone trials, were required to stop their hormones for at least three months prior to joining the study and had to be willing to be randomized to either the hormone or placebo groups.
Participants completed health questionnaires assessing previous hormone use, history of cardiovascular disease, diabetes mellitus, stroke, lifestyle factors and demographic data. WHI hormone study participants who were 65 years and older also participated in baseline cognitive evaluation using the Modified Mini-Mental State Examination (3MSE), a 100-point validated test of global cognitive functioning.26
The 3MSE's 46 items contribute to a total score from 0–100, with a higher score reflecting better cognitive function. The test items measure temporal and spatial orientation, immediate and delayed recall, executive function, naming, verbal fluency, abstract reasoning, praxis, writing, and visuo-constructional abilities. The 3MSE has demonstrated good internal consistency and temporal reliability, sensitivity, and specificity for detecting cognitive impairment and dementia.27–29
Standardized 3MSE assessments were conducted by trained technicians in a quiet private area.
The standardized blood pressures were measured in the right arm after subjects had been seated quietly for at least 5 minutes. Hypertension was defined as a blood pressure equal to or greater than 140 mmHg systolic or 90 mmHg diastolic or if the participant was taking antihypertensive agents. Weight and height measurements were obtained using a calibrated beam balance scaled and a stadiometer anchored to the wall, with subjects wearing light street clothing and without shoes. Waist and hip measurements were obtained to the nearest 0.5 cm over non-binding undergarments at the level of the umbilicus and the fullest hip circumference. Body mass index (BMI) was calculated by dividing the weight in kilograms by the height in meters squared. Women were classified into BMI categories that correspond to standard World Health Organization-designated categories for underweight, desirable weight, overweight and obesity I, II, and III. In this classification, overweight is defined as BMI≥25.0 and obesity as BMI ≥ 30.0 according to World Health Organization criteria. All data were entered electronically into a centralized data base.
A majority of the study subjects were white (86.61%). Since ethnic minority women were not equally distributed among BMI categories, our analyses were restricted to white women. Means and distributions for age, 3MSE scores, education, waist circumference, and other variables of interest (disease and lifestyle) were stratified by BMI category. Means for variables of interest were also calculated after adjusting for age and education. Spearman Rank Correlation procedures were used to determine correlations between 3MSE score and BMI, waist circumference, age, education, systolic blood pressure, and smoking history. Odds ratios were calculated for atrial fibrillation, heart disease, hypertension, stroke, diabetes mellitus and previous hormone use. Multivariate logistic regression analyses were used to determine if increasing BMI category predicts 3MSE score when controlled for age, education hypertension, stroke, smoking, diabetes mellitus, and heart disease. Because the 3MSE scores were skewed to the right in this cognitively intact cohort, each score was subtracted from 100, which skewed the data to the left, permitting the use of a generalized linear model (GLM) with discrete values. The Poisson distribution with an over-dispersion parameter gamma was used because the transformed 3MSE scores had more variability than a pure Poisson process. Coefficients of the model are expressed in terms of the odds ratio and a 95% confidence interval. A log link function was used to model the transformed 3MSE scores. An ordinal logistic regression model was also used to validate the model independence of the results. This model (data not presented) lead to the same conclusions. Linear regression contrasts within the GLM were used for trend tests, such as age, education, waist-hip ratio, exercise and 3MSE. Tests for trends in the continuous covariates across BMI categories used a linear contrast within an analysis of variance. Tests for trends on binary variables, such as hypertension, heart disease, stroke, transient ischemic attack and diabetes were based on linear contrasts within logistic regression. SAS Version 9.1.3 was used for analyses.