RESUMO
BACKGROUND: The aim of our study was to determine in a group of premenopausal (preM) and postmenopausal women with hormone replacement therapy (postM-HRT) or without HRT (postM-noHRT) whether the strength of the association of body mass index (BMI), waist circumference (WC), and waist-to-hip ratio (WHR) with hypertension is affected by the way in which these variables are used (i.e., continuous or dichotomous) in logistic regression modeling, where weight and age are set as continuous variables only. METHODS: We studied 639 preM, 341 postM-noHRT, and 77 postM-HRT healthy women age 20-69 years. The main outcome variable was hypertension: BP > or =140 or 90 mmHg on two occasions or hypertension treatment. Risk factors were assessed as continuous variables (age, postmenopausal years, weight, stature, BMI, WC, hip circumference, and WHR); and dichotomous variables (overweight: BMI > or =25; high WC: WC > or =88 cm; and high WHR: WHR > or =0.84). Stepwise regression models for hypertension were developed for continuous (model 1) and dichotomous (model 2) variables in each group. RESULTS: In preM, model 1 included BMI (OR=1.202, CI: 1.115-1.296) and age (OR=1.154, CI: 1.081-1.296); and model 2 included weight (OR=1.070, CI: 1.040-1.100) and age (OR=1.157, CI: 1.084-1.235). In postM-noHRT, both models included age (OR=1.074, CI: 1.035-1.113) and weight (OR=1.025, CI: 1.004-1.046). Overweight and age were associated with hypertension in preM and postM-noHRT; overweight being more relevant in preM, and age in postM-noHRT. No models were found for postM-HRT. CONCLUSIONS: The strength of the association of risk factors with hypertension depends on how the variables are analyzed. Acknowledging these differential effects and assigning differential risks by age could be useful in intervention programs.
