“While observational studies play a role in developing good science, they do not constitute good science in and of themselves. Because they rely on correlation without causation, they should not be used to make health claims.”