A statistician for a pharmaceutical company wants to estimate the proportion of patients who experience dizziness as a side effect by taking a sample of patients. How large of a sample size is needed to estimate this proportion to within 2% at the 99% confidence level? No prior estimate is available.