Hormone replacement therapy and acute coronary outcomes: methodological issues between randomized and observational studies.